Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.
6 HEALTH EFFECTS The studies discussed in this chapter demonstrate that veteransâ reactions to deployment to a war zone and its inherent stressors are varied. Some veterans suffer no ill consequences and may even benefit from deployment, and some experience initial distress but become acclimated to some of the stressors; but the body of evidence suggests that most veterans perceive deployment to a war zone as stressful (Gifford et al. 2006; MHAT 2006b). Given that few studies provide a good indication of the deployment-related stressors to which veterans were exposed and the amount of stress experienced during those exposures, the committee decided that deployment to a war zone itself was sufficient to elicit a stress response and that deployment would serve as a surrogate for deployment-related stress (the stress response is discussed in detail in Chapter 4). Therefore, military personnel deployed to a war zone would constitute the study group of interest and military personnel serving at the same time but not deployed to a war zone would be the appropriate comparison group. The studies reviewed in this chapter generally compared Gulf War veterans or veterans of other deployments, such as to Vietnam, with veterans who were deployed during the same period to noncombat areas within the United States or elsewhere (for example, Germany). In some cases, the committee included studies of other military personnel, such as peacekeepers, or civilians that had similar exposures if the studies appeared to be relevant to the discussion. ORGANIZATION OF THIS CHAPTER The committee discusses health effects in the order they occur in the International Statistical Classification of Diseases and Related Health Problems, 10th Edition (ICD-10,)1 except for the last section, which examines the many reports of increased health symptoms in deployed veterans. The committee did not examine health effects related to or resulting from infectious and parasitic diseases as the Institute of Medicine (IOM) Committee on Gulf War and Health: Infectious Diseases has examined those outcomes and released its report (IOM 2007). The committee also did not consider health effects that are most likely to be associated with exposures to environmental agentsâsuch as sarin nerve gas, the defoliant Agent Orange, or combustion products from oil-well firesâor with the use of various vaccines and other 1 The International Statistical Classification of Diseases and Related Health Problems (ICD) provides a detailed description of known diseases and injuries. Every disease (or group of related diseases) is given a unique code. ICD is periodically revised and is currently in its 10th edition (ICD-10) and available at http://www.who.int/classifications/apps/icd/icd10online/. 115
116 GULF WAR AND HEALTH prophylactic measures, although anticipation of exposure to such agents may be a deployment- related stressor, as indicated in Chapter 3. Those stressors have already been thoroughly evaluated by other IOM committees. In general, for each health effect presented in this chapter, the committee identifies first the primary studies and then the secondary or supporting studies, as defined by the criteria in Chapter 2. A primary study had to include information about the putative exposure (deployment) and specific health effects, demonstrate rigorous methods, include details of its methods, include an appropriate control or reference group, have adequate statistical power to detect effects, and provide appropriate adjustment for confounders. Many of the large cohort studies examined multiple effects and so might be referred to in more than one place. A given study might be deemed a primary study for one or more health effects and be a secondary study for another effect, as determined by how the particular health effects were defined and measured. For example, a study that was well designed for assessing a neurobehavioral effect might not be well designed for assessing a psychiatric disorder. In general, only primary studies appear in the evidence tables that accompany the discussions of health effects. A secondary study was typically a study that had methodologic limitations, such as not including a rigorous or well-defined measure of exposure, that is, deployment to a war zone or posttraumatic stress disorder (PTSD) as a marker of trauma (see Chapter 1). Some studies assessed past trauma that was not necessarily peculiar to war-zone deployment, so they were included as secondary studies. The secondary studies were reviewed and included in the discussion because they evaluated the same health effects and in some cases provided useful information on veteran populations from the same conflicts as the primary studies; they add information that might increase or decrease confidence in the conclusions based on the primary studies. Confidence in a secondary study is substantially reduced if the statistical analysis did not adjust for confounders, if the data were obtained from self-reported cross-sectional surveys or from screening instruments that relied solely on self-reported answers, or if response rates were unacceptably low. Without evidence from primary studies, the potential for unreliable findings due to bias, chance, or multiple comparisons may outweigh the extent to which secondary studies may contribute, even collectively, to the overall conclusion of the committee about an association between deployment-related stress and any specific health effect. Understanding the relationship between a health effect and deployment-related stress may be hampered by the exposure not being explicitly defined as deployment or, more commonly, by many potentially harmful exposures being compared with multiple effects. With rare exceptions, the chapter excludes studies of participants in Gulf War registries established by the Department of Veterans Affairs (VA) or the Department of Defense (DoD), which were not intended to be representative of the entire group of Gulf War veterans. Registry participants can not be considered representative of all Gulf War veterans in that they are self- selected subjects, many of whom have joined the registries because they believe that they have symptoms of a new medical syndrome; they were not randomly selected from all Gulf War military personnel, and there is no nondeployed control group. One main exception to the use of registries occurs for studies in which the groups of interest are veterans with and without PTSD. As discussed in Chapter 5, PTSD can be diagnosed only after exposure to a traumatic event, thus, the committee agreed that studies of veterans with deployment-related PTSD compared with deployed veterans who had not developed PTSD were appropriate comparison populations when determining whether health effects were associated with deployment-related PTSD.
HEALTH EFFECTS 117 Many of the studies had other limitations. Few studies measured the stress that troops experienced during deployment; rather, most asked veterans about exposure to possible stressors after their return from deployment. Some studiesâsuch as the congressionally mandated National Vietnam Veterans Readjustment Study (NVVRS), the Vietnam Experience Study (VES), and some Gulf War studiesâassessed veteransâ war-zone exposures many years after they had returned home. Furthermore, many studies did not verify veteransâ reported exposures against military records. Finally, although many studies used various scales, such as the Combat Exposure Scale, to determine possible exposures during deployment, the scales did not ascertain the emotional response of the veterans to the exposures; that is, the studies asked only whether exposure to a stressor occurred and not about the degree to which the veterans found the experience to be stressful. Studies that did ascertain veteransâ reactions to stressors, such as seeing a comrade wounded or firing a gun at the enemy, and that asked veterans to rate their responses on a scale, such as âneverâ to âalways,â are rare. The committee acknowledges that many of the health effects associated with deployment were discussed in a previous volume of the Gulf War and Health series, Volume 4: Health Effects of Serving in the Gulf War. In that volume, the health effects that had been found in deployed Gulf War veterans were identified, and their prevalence compared with the prevalence of the same effects in nondeployed Gulf War veterans. However, that review was restricted to Gulf War veterans; veterans of other conflictsâsuch as the Vietnam War, World War II, the Korean War, and Operation Enduring Freedom (OEF) and Operation Iraqi Freedom (OIF)â were not considered. Furthermore, that report did not seek to establish whether there was any association between being deployed and specific health effects. As a consequence, although the present committee reviewed many of the same studies as the committee that prepared Volume 4, the assessment process used here is different, as are some of the conclusions because a broader array of studies were considered using defined categories of association. CANCER Each year over a million people receive a diagnosis of cancer in the United States. About one of two American men and one of three American women will have cancer at some point in their lives. Cancer can develop at any age, but about 77% of all cancers are diagnosed in people 55 years old or older. Military personnel during the Gulf War had a mean age of 28 years and therefore are now in their mid-40s. Insufficient time had elapsed for most forms of cancer to be detected among Gulf War veterans by surveys conducted in the 1990s and early 2000s. That is not the case, however, for veterans of the Vietnam War, many of whom are now at an age when most cancers are likely to be diagnosed. Therefore, given the substantial differences in ages between veterans of the Vietnam War and the Gulf War, the studies are discussed separately. Primary studies for this health effect were those that compared deployed vs nondeployed veteran populations from either the Vietnam War or the Gulf War. A primary study must have indicated that the presence of cancer or death from cancer was confirmed, as by physical examination, medical record review, or death certificates. Of particular concern in studies of Vietnam veterans is possible confounding from exposure to Agent Orange, a toxic herbicide sprayed on foliage in Vietnam. Most of the studies reviewed in this section did not distinguish between the type of cancers that were seen in or reported by veterans, so the occurrence of cancer is addressed as a specific endpoint. The few studies reporting specific results on testicular cancer, which occurs
118 GULF WAR AND HEALTH predominantly in younger men, and on skin cancer, which is relatively common, are discussed in separate sections. The occurrence of cancer in veterans with PTSD is discussed at the end of the section. The primary studies for cancer are summarized in Table 6-1 at the end of the section. All Cancers Vietnam War The committee identified three primary studies that examined the effect of deployment to Vietnam on the development of cancer in veterans of the Vietnam War (CDC 1988b; Selected Cancers Cooperative Study Group 1990a,b,c; Watanabe and Kang 1995). Two other primary studies (Dalager et al. 1995; Kang et al. 2000c) focused exclusively on cancers in female Vietnam veterans. In response to a congressional mandate, Centers for Disease Control and Prevention (CDC) undertook the VES to assess the health status of the 5 million Vietnam-theater and Vietnam-era veterans who served in the U.S. Army during 1965-1971; the study was completed in 1988, about 15-20 years after the war (CDC 1988b). It consisted of a nationally representative random sample of 7924 theater veterans and 7364 era veterans who completed a phase 1 telephone interview. In phase 2, a random subsample of 2490 theater veterans and 1972 era veterans also completed physical- and psychologic-health screening examinations in 1985-1986 at a medical facility. On examination, 1.9% of theater veterans and 1.3% of era veterans had cancers (unspecified); the study findings were not significant (odds ratio [OR] 1.4, 95% confidence interval [CI] includes 1.0, p > 0.05). The OR was adjusted for age at enlistment, race, year of enlistment, enlistment status, score on a general technical test, and primary military occupational specialty. The study had the advantage of including a physical examination and a large study population, but it is limited in that information on exposure to herbicides, particularly Agent Orange, was not provided, the types of cancer that were screened for were not indicated, the study was not designed to assess the presence of relatively rare cancers, and the participation rate of 75% and 63% for the theater and era veterans, respectively. As a followup to the VES, CDC conducted a further population-based case-control assessment for six cancers in Vietnam-theater and Vietnam-era veterans: non-Hodgkinâs lymphoma (NHL), Hodgkinâs lymphoma, soft-tissue and other sarcomas, nasal cancer, nasopharyngeal cancer, and primary liver cancer. Those cancers were chosen because cancers of the liver, nasal cavity, and nasopharnyx, and Hodgkinâs disease have been associated with exposure to phenoxy herbicides, such as Agent Orange, in some animal studies and a few human studies. Study participants were all men born in 1929-1953 who were first diagnosed as having cancer in 1984-1988 and were listed in any of eight city or state cancer registriesâthose in Atlanta, Detroit, San Francisco, Seattle, Miami, Connecticut, Iowa, and Kansas. Controls were selected by random-digit telephone dialing in the relevant locations and frequency-matched by age to the men with cancer. All cancers were confirmed pathologically. An analysis of 1157 men with NHL and 1776 controls found that the risk of NHL in men who served in Vietnam compared with those who served in the military in 1964-1972 but not in Vietnam was 1.52 (95% CI 1.00-2.32, p = 0.01) when adjusted for registry, age group in 1968, ethnicity, education, and other covariates, including reported exposures to pesticides, reported medical history and medication use, smoking, marital status, and being raised as Jewish (Selected Cancers Cooperative Study Group 1990a). There was no suggestion of an increasing trend in risk of NHL
HEALTH EFFECTS 119 corresponding to service in support units (OR 1.50, 95% CI 1.02-2.21), combat-support units (OR 1.18, 95% CI 0.65-2.15), or combat units (OR 1.25, 95% CI 0.63-2.45). With respect to soft-tissue and other sarcomas, the same eight cancer registries showed that 342 men had confirmed cancers compared with the 1776 controls. The OR for soft-tissue and other sarcomas in Vietnam-theater veterans compared with Vietnam-era veterans is a nonsignificant 0.74 (95% CI 0.39-1.41) with the same adjustment factors as for NHL (Selected Cancers Cooperative Study Group 1990b). Again, there was no trend in cancer risk in men who served in support units (OR 0.76, 95% CI 0.37-1.53), combat-support units (OR 1.03, 95% CI 0.44-2.41), or combat units (OR 0.76, 95% CI 0.25-2.28). Combining data from the eight cancer registries shows that 28 of the 310 men with Hodgkinâs disease had served in Vietnam, two of the 48 with nasal carcinoma, three of the 80 with nasopharyngeal carcinoma, and eight of the 130 with primary liver cancer (Selected Cancers Cooperative Study Group 1990c). Although the power of the study was good for Hodgkinâs disease (96%), it was less robust for the other cancers. In comparisons of Vietnam- theater veterans with Vietnam-era veterans with the same adjustments as for NHL, the OR was 1.23 for Hodgkinâs disease (95% CI 0.65-2.32), 0.31 for nasal carcinoma (95% CI 0.04-2.20, additionally adjusted for occupational exposures), and 0.53 for primary liver cancer (95% CI 0.14-1.94, additionally adjusted for occupational exposures); the OR for nasopharyngeal carcinoma could not be calculated with conditional logistic regression. There was no trend (p = 0.17) in risk of Hodgkinâs disease related to serving in Vietnam in a support unit (OR 1.58, 95% CI 0.90-2.77), a combat-support unit (OR 0.50, 95% CI 0.14-1.76), or a combat unit (OR 0.94, 95% CI 0.34-2.59). The relationship of combat to the other three cancers was not assessed. Although the authors did not sample blood for possibly elevated dioxin levels, they asserted that, based on their locations and occupations during the war, the Vietnam-veteran study participants were not at greater risk of exposure to Agent Orange than the nondeployed veterans. Those studies are limited by the small sample sizes for some cancers. Postservice mortality was assessed in 10,716 Marines who served in Vietnam and 9346 Marines who were not deployed to Vietnam (Watanabe and Kang 1995). An analysis of death certificates showed that as of 1991, for Marines who had served in Vietnam, there were no statistically significant increases in all-cancer mortality (rate ratio 1.08, 95% CI 0.84-1.39), cancer of the larynx (rate ratio 2.60, 95% CI 0.27-25.0), lung cancer (rate ratio 1.12, 95% CI 0.71-1.76), or lymphosarcoma and reticulosarcoma (rate ratio 1.21, 95% CI 0.27-5.41). No statistically significant increase was seen for all cancers or for lung cancer. The relative-risk (RR) estimates based on the Cox proportional-hazards model were 1.20 (95% CI 0.93-1.55) for all cancers and 1.33 (95% CI 0.84-2.10) for lung cancer, adjusted for year of birth and military rank. Death certificates were obtained from VA regional offices, federal records centers, and state vital-statistics offices. Comparisons with the U.S. male population showed slightly increased standardized mortality ratios for laryngeal cancer and for lymphosarcoma and reticulosarcoma, but the increases were not statistically significant. The authors considered the latent period for many cancers to be 15-20 years but found no difference in RRs in cancer deaths whether veterans were followed for less than or more than 16 years. The authors suggested that there may have been insufficient observation time since the war to detect excess deaths from cancer and that there was insufficient statistical power to detect rare causes of death. They also noted that many of the Marines were posted to areas in Vietnam where Agent Orange was used. A similar study of cancer mortality was conducted in female Vietnam veterans (Dalager et al. 1995). The vital status of 4586 female Vietnam veterans and 5325 female Vietnam-era
120 GULF WAR AND HEALTH veterans was compared more than 20 years after the war. Vital status as of 1991 was determined from VA beneficiary records, information from the Social Security Administration and the Internal Revenue Service, the National Death Index, and military personnel records; copies of official death certificates were used to determine cause of death. The Cox proportional-hazard multivariate regression model was used to derive cause-specific mortality risks for female Vietnam theater veterans vs female Vietnam-era veterans adjusted for rank, nursing status, duration of military service, and age at entry into followup. The RR for all cancer mortality was 1.00 (95% CI 0.75-1.34). The authors noted that there were increased risks of cancer of the pancreas, uterine corpus, brain, and other parts of the central nervous system in female Vietnam veterans, but the number of deaths was small and did not reach statistical significance, and adjustment for confounders was not possible. Female theater veterans had a lower mortality than female era veterans from lymphopoietic cancer and lung cancer. This study and that by Watanabe and Kang (1995) discussed above had the advantage of being conducted after a sufficient latent period for many cancers to be evident, however the number of cancer cases was small in each study, and their study populations were to small to have much potential to identify changes in the incidence of relatively rare cancers. Not all cancers result in death, so a second study of the prevalence of gynecologic cancers among female Vietnam veterans was undertaken (Kang et al. 2000c). The prevalence of malignant tumors was ascertained in 4140 female Vietnam-theater veterans and 4140 female Vietnam-era veterans. Participants were given a structured health questionnaire via telephone interview, and self-reported gynecologic (breast, ovarian, uterine, or cervical) cancer was confirmed by review of medical or hospital records. The response rate of the 6430 women interviewed was 83-87%. A multivariate logistic model was used to calculate the ORs and 95% CIs. The adjusted OR was 1.14 (95% CI 0.94-1.40) for any gynecologic cancer; 1.18 (95% CI 0.91-1.51) for breast cancer, 1.83 (95% CI 0.72-4.61) for ovarian cancer, 1.00 (95% CI 0.61- 1.61) for uterine cancer, and 1.11 (95% CI 0.74-1.66) for cervical cancer. Adjustments included age, race, branch of service, pay grade, marital status, nursing occupation, smoking, drinking, family history of cancer, use of birth-control pills, and postmenopausal estrogen and progestin use. Medical records were used to confirm self-reported cancer; 222 records were reviewed, and 99% of self-reported breast cancers, 76% of cervical cancers, 78% of ovarian cancers, and 61% of uterine cancers were confirmed. However, 10% of the self-reported cancers were not confirmed, and 5% of the medical records had no information on cancer. This study indicates that service in Vietnam did not increase the risk of gynecologic cancer in female Vietnam veterans. The study had a high response rate and relatively high validation rate for the diagnosis of gynecologic cancers that had a sufficient latent period for detection. It is limited by lack of information on possible exposures during service in Vietnam, including exposure to Agent Orange. Gulf War There was only one primary study in Gulf War-deployed veterans compared with their nondeployed counterparts, which indicated an increased risk of cancer. McCauley et al. (2002b) looked at cancer rates in Gulf War veterans residing in Oregon, Washington, California, Georgia, and North Carolina in 1999 as part of a larger study to assess neurologic and neurophysiologic signs and symptoms in veterans who may have been exposed to chemical-warfare agents as a result of the destruction of munitions at Khamisiyah, Iraq. Names of possible participants were obtained from the Defense Manpower Data Center (DMDC) maintained by the DoD. Gulf War-
HEALTH EFFECTS 121 deployed veterans were categorized as those who were within a 50-km radius of Khamisiyah between March 4 and March 13, 1991, and were therefore potentially exposed to chemical agents (n = 653), and Gulf War veterans who were deployed to Southwest Asia but outside of the 50-km radius (n = 610). A control group consisted of nondeployed veterans in the military at that time but not deployed to Southwest Asia (n = 516). A structured telephone interview asked veterans about physician-diagnosed medical conditions, hospitalizations, and disability during and after return from the Gulf War. The frequency of any cancer was 1.2% for veterans deployed in the Khamisiyah region, and 2.1% for those deployed outside the Khamisiyah region, for an OR of 0.4 (95% CI 0.1-1.4); the frequency of any cancer for nondeployed veterans was 0.6%. When the 21 cases of cancer among all deployed veterans (Khamisiyah region or elsewhere in the gulf combined) were compared with the 3 among nondeployed veterans, there was a nonsignificant increase in cancer frequency among the deployed troops (OR 3.0, 95% CI 1.0-13.1). All statistics were adjusted for age, gender, race, and region of residence. Twenty of the 24 deployed and nondeployed veterans reporting a diagnosis of cancer were asked for details on year of diagnosis and cancer type. When the 7 cases of skin cancer and the 4 cases of cancer for which details were not obtained were removed from the analysis, the resulting OR for non-skin cancers in deployed vs nondeployed was 4.94 (95% CI 0.6-38.1); there was no apparent trend for any specific type of cancer among the remaining 9 cancer cases. This study is limited by the small sample size and few reports of cancer and by the incomplete verification of the diagnosis with medical records or examination. A potential confounder of stress as a risk factor, exposure to chemical-warfare agents, was the target of the researchersâ investigation: although this exposure is subject to great uncertainty, the primary result for proximity to Khamisiyah (OR = 0.4) argues against its importance in the development of cancer. Macfarlane et al. (2003) assessed all first diagnoses of malignant cancer in a cohort of UK military personnel. The deployed group consisted of all military personnel who served in the Persian Gulf during September 1990-June 1991 (n = 51,721). The comparison group was randomly selected from members of the armed services who were in service on January 1, 1991, but not deployed in the Persian Gulf and was stratified to match the Gulf War cohort on age, sex, service branch, rank, and level of fitness for active service (n = 50,755). Followup was from April 1, 1991, until diagnosis of cancer, emigration, death, or July 31, 2002, whichever was earlier. Cancers were identified through the National Health Service Central Register. During followup, 270 incident cases of cancer were identified among the Gulf War veterans and 269 cases among the nondeployed group; the RRâafter adjustment for sex, age group, service branch, and rankâwas 0.99 (95% CI 0.83-1.17). Thus, there was no evidence of an association of Gulf War service with site-specific cancers. In subgroups of cohort members who participated in morbidity surveys that yielded more information on potential risk factors (28,518 deployed veterans and 20,829 nondeployed veterans), the RR was 1.11 (95% CI 0.86-1.44). That result did not change after adjustment for smoking or alcohol use, and there was no evidence of associations with exposure to pesticides; multiple vaccinations against anthrax, plague, and pertussis; or reported exposure to depleted uranium. Two studies by Gray et al. (1996, 2000) attempted to determine whether Gulf War veterans were at increased risk for hospitalization after the war compared with nondeployed veterans. DoD hospital records on 547,076 active-duty Gulf War veterans and 618,335 nondeployed active-duty veterans in all service branches were examined with demographic information from military records. The multiple logistic-regression models were adjusted for sex, age, race or ethnic group, marital status, branch of service, rank, salary, occupation, and prewar
122 GULF WAR AND HEALTH hospitalizations. Using August 1991 as the end of the war, Gray et al. (1996) found a slightly increased risk of hospitalization for any neoplasm in the last 5 months of 1991 in deployed vs nondeployed veterans (standardized rate ratio about 1.10, 95% CI 1.0-1.2), but this dropped to about 0.92 (95% CI about 0.87-1.0) in 1992 and about 0.94 (95% CI about 0.87-1.1) in 1993. Most of the neoplasms diagnosed in 1991 were benign (for example, benign neoplasms of bone, articular cartilage, connective tissue and other soft tissue, skin, breast, and digestive system). In 1991, hospitalizations for malignant neoplasm of the testis were slightly higher in deployed men (rate ratio 2.12, 95% CI 1.11-4.02), but no difference was seen in 1992 or 1993. Expanding this study to include National Guard and reserve veterans, Gray et al. (2000) examined hospital records from the DoD, VA, and California Office of Statewide Health Planning and Development hospital systems for 1991-1994. The authors could not directly compare rates of hospitalization in the three hospital systems, so they compared proportional morbidity ratios (PMRs) of hospital-discharge diagnoses (14 diagnostic categories from ICD-9) in Gulf War- deployed and nondeployed veterans. PMRs for neoplasms were not higher in deployed veterans in any of the hospital systems. Several other large secondary studies looked at cancer in large populations of Gulf War veterans and found that there was no significant increase in any cancer in deployed veterans compared with nondeployed veterans. Although the studies below used self-reports of medical conditions, they used structured telephone interviews by trained interviewers or in-person interviews of the veterans. The Iowa Persian Gulf Study Group (1997) used a cross-sectional telephone-interview survey of Gulf War-deployed and nondeployed veterans who claimed Iowa as their place of residence in their military records. Randomly selected subjects were stratified on age, sex, rank, race, and branch of military service. Of the 4886 eligible subjects, 3695 (76%) completed the interview (1896 deployed and 1799 nondeployed). Deployed and nondeployed veterans had a similar prevalence of any cancer (prevalence 1.0% and 1.9% in deployed regular military and reserves, respectively, vs 1.0% and 0.6% in nondeployed regular military and reserves; Cochran-Mantel-Haenszel rate difference 0.8, 95% CI 0.2-1.4, p â¤ 0.05) and aplastic anemia (prevalence 0.1% and 0.0%, respectively, vs 0.0% for all nondeployed veterans; rate difference 0.1, 95% CI -0.1-0.2). Although deployed veterans were asked about exposure to unspecified psychologic stressors, similar data on nondeployed veterans are not provided. This study was limited by restriction of participants to Iowans, a sample size that may not be sufficient to detect cancers, lack of specificity as to reported cancers, and the use of self-reports of health effects. Kang et al. (2000b) used a nationally representative stratified random sample of Gulf War veterans and nondeployed veterans to assess the health of veterans with telephone interviews. On the basis of responses from 11,441 deployed veterans and 9476 nondeployed veterans, the authors estimated the population prevalence of self-reported cancers other than skin cancer during the preceding 12 months to be 0.7% in deployed and 0.6% in nondeployed for a statistically significant increased rate difference of 0.18 (95% CI 0.15-0.21, p â¤ 0.05). Prevalence was controlled for sex, age, race, marital status, rank, branch, unit component, active-duty status, and exposures. This study was limited by the use of self-reports and the lack of information on the number of cancer cases reported in the surveyed veterans. In 1998, Steele (2000) assessed the prevalence of cancer and other medical conditions in a stratified random sample of Kansas Gulf War veterans (n = 1545) and veterans who were not deployed to the gulf (n = 435), using names supplied by the DMDC. The study was conducted to identify cases of Gulf War illness. In telephone interviews, participants were asked whether they
HEALTH EFFECTS 123 had ever had a diagnosis or treatment by a physician for any of 16 specific medical conditions or for any medical conditions in five general categories and, if so, when the conditions had developed. Deployed veterans were also asked whether they had been notified by DoD about being in the area of the Khamisiyah munitions demolition. Cancers, other than skin cancer, were diagnosed after 1990 in 1% each of deployed and nondeployed veterans for a nonsignificant risk of 1.21 (95% CI 0.40-3.69), adjusted for sex, age, income, and education level. Kang and Bullman (2001) studied causes of postwar mortality in Gulf War veterans through 1997. They ascertained vital status of a stratified random sample of Gulf War-deployed veterans and nondeployed veterans from VA and Social Security Administration databases, and death certificates were obtained from VA and the National Death Index database. From 1990 to 1997, there were 4506 deaths among 621,902 Gulf War veterans, and causes of death were obtained for 94.7% of them. The OR for mortality from any cancer during this period was 0.90 (95% CI 0.81-1.01) in men and 1.11 (95% CI 0.78-1.57) in women adjusted for age, race, branch of service, unit component, and marital status. The main strength of this study was the large sample. As in previous studies of Gulf War veterans, the latent period for most cancers is too long for their manifestation during the study period, and there was also a lack of information on risk factors. Several secondary studies also looked at the prevalence of any cancer in Gulf War veterans. Eisen et al. (1991) and Proctor et al. (2001) examined U.S. veterans, Simmons et al. (2004) surveyed a large cohort of UK veterans, Goss Gilroy Inc. (1998) surveyed all Canadian Gulf War veterans, and OâToole et al. (1996b) surveyed the entire Australian cohort of Gulf War veterans. Each of those studies found no significant increases in any cancer, on the basis of self- reports, in veterans deployed to the gulf compared with nondeployed veterans. Testicular Cancer Two studies have focused specifically on testicular cancer. Testicular cancer is relatively uncommon in the United States. The annual age-adjusted incidence is 5.3 cases per 100,000 men. However, it is one of the few cancers whose usual age of onset is in the same range as the age of the Gulf War veterans, about 20-44 (Ries et al. 2005). Knoke et al. (1998) examined testicular cancer in 517,223 deployed and 1,291,323 nondeployed male veterans on active duty during the time of the Gulf War. The authors identified cases of all first hospital admissions, in U.S. military hospitals worldwide, for a principal diagnosis of testicular cancer. Cases were identified by examining the DoD hospitalization database through April 1, 1997. A total of 505 cases were ascertained: 134 in the deployed and 371 in the nondeployed. In Cox proportional-hazards models adjusted for race and ethnicity, age, and occupation, no association with deployment status was observed (RR 1.05, 95% CI 0.86-1.29). The deployed did have an increased risk in the early months after the end of the deployment period. The initial increase in risk was originally reported in a study of all hospitalizations in the cohort by Gray et al. (1996) discussed above. However, by the end of 1996, the cumulative probability of hospitalization of the two groups was the same (0.034% for deployed and 0.035% for nondeployed). There was no interaction between covariates and deployment status. The authors also assessed the association of testicular cancer with specific occupations for both deployed and nondeployed veterans. The highest RRs were observed for men engaged in electronic-equipment repair (RR 1.56, 95% CI 1.23-2.00), construction-related trades (RR 1.42, 95% CI 0.93-2.17), and electric or mechanical repair (RR 1.26, 95% CI 1.01- 1.58). The followup period was short for a cancer assessment, but it did include the age range
124 GULF WAR AND HEALTH (22-31 years) when the disease might appear. No specific Gulf War exposures were assessed, although risk by occupational group was calculated. There was some evidence of an association of testicular cancer with Gulf War deployment in a pilot cancer-registry-based study. Levine et al. (2005) matched a stratified random sample of 621,902 Gulf War deployed active-duty, reserve, and National Guard veterans and 746,248 nondeployed veterans with the central cancer registries of New Jersey and the District of Columbia. From 1991 to 1999, 17 deployed and 11 nondeployed veterans were identified with testicular cancer for a proportional incidence rate of 2.33 (95% CI 0.95-5.70) adjusted for state of residence, deployment status, race, and age. The greatest proportions of testicular cancer were in deployed men in the age groups of 25-29 and 30-34 years (standardized incidence ratio 1.42) and in nondeployed men in age groups of 30-34 and 35-39 years (standardized incidence ratio 0.94). The number of excess cases peaked 4-5 years after deployment, as opposed to the findings in the Knoke et al. study, which found the excess in the first few months after the soldiers returned home. Gray et al. (1996) analyzed hospital records from DoD facilities for the last 5 months of 1991 and all of 1992 and 1993. In 1991, hospitalizations for malignant neoplasm of the testis were slightly higher in deployed men (rate ratio 2.12, 95% CI 1.11-4.02), but no difference was seen in 1992 or 1993. Skin Cancer There are two types of skin cancer: melanoma, which forms in the skin cells that make pigment and is less common, and the more common nonmelanoma skin cancer, which typically begins in cells that do not make pigmentsâbasal cells (small round cells in the base of the outer layer of skin) or squamous cells (flat cells that form the surface of the skin). The annual age- adjusted incidence of melanoma of the skin is about 18.5 per 100,000 people. Skin cancer was assessed separately by the committee because of the potential for military personnel to be exposed to ultraviolet radiation and environmental toxicants and because several of the studies considered by the committee provided an analysis of skin cancer as distinct from other cancers. The committee identified only one primary study that assessed the prevalence of skin cancer in Vietnam veterans: CDC (1988b) looked for skin cancer as part of the VES. The prevalence of skin cancer on dermatologic examination was 0.6% in Vietnam-theater veterans and 0.7% in Vietnam-era veterans for a nonsignificant OR of 0.8 (95% CI 0.4-1.7) adjusted for age at enlistment, race, year of enlistment, enlistment status, score on general technical test, and primary military occupation. One primary study that assessed the risk of skin cancer in Gulf War veterans was identified. As of 2002, when Australian Gulf War veterans were compared with their nondeployed counterparts, they had no increase in prevalence of probable or possible skin cancers diagnosed after 1991 (Kelsall et al. 2004a). The entire Australian cohort of 1871 veterans who were deployed to Southeast Asia was compared with nondeployed veterans frequency matched for service type, sex, and age. Participants completed a self-report questionnaire about medical conditions that had been diagnosed or treated by a medical doctor and about when the conditions had been diagnosed. Participants also underwent a comprehensive health assessment by specially trained health professionals who were blinded to the deployment status of the participants and asked further questions about the diagnoses and determined whether the self- reports were unlikely, possible, or probable according to pre-established criteria. The participation rate was 80.5% for Gulf War veterans and 50.5% for the control group. This study
HEALTH EFFECTS 125 has the advantage of a good participation rate for the deployed veterans and use of criteria for assessing self-reports of health but is limited in that self-reports were not verified by a physical examination and the participation rate for the control group was poor. Although the veterans were asked about psychologic stressors experienced during their deployment, the results of this questionnaire were not correlated with specific health outcomes. Three secondary studies also estimated the risk of skin cancer in Gulf War veterans on the basis of self-reports. As discussed above, the Iowa Persian Gulf Study Group (1997) also surveyed deployed and nondeployed Gulf War veterans from a restricted geographic area to ascertain the prevalence of skin cancer. Compared with nondeployed veterans, Gulf War veterans had a significantly increased prevalence rate difference for skin cancer (Cochran- Mantel-Haenszel rate difference 0.8, 95% CI 0.4-1.3, p â¤ 0.05). In the Kang et al. (2000b) study, the population prevalence of self-reported skin cancer was estimated to be 1.5% in Gulf War veterans and 1.4% in Gulf War-era veterans for a statistically significant increased rate difference of 0.15 (95% CI 0.11-0.19, p â¤ 0.05). Steele (2000) looked at skin cancer in Gulf War-deployed and nondeployed Kansas veterans in 1998. On the basis of a structured telephone interview in which veterans were asked whether they had ever received a physicianâs diagnosis of or treatment for a medical condition and when it had developed, the OR for skin cancer occurring after 1990 in deployed veterans was 1.17 (95% CI 0.47-2.90). As noted in the discussion of this study above, its limitations include a restricted geographic area, lack of verification of medical conditions by medical record or examination, and poor exposure data. PTSD and Cancer Few studies have assessed cancer in Vietnam War and Gulf War veterans who have PTSD. The committee identified only one primary study: Boscarino (2005) examined excess postservice mortality from cancer in Vietnam veterans by using data from the VES 16 years after the war. Mortality was assessed in Vietnam veterans who were known to be alive in 1983 and who completed a telephone interview at that time on PTSD symptoms and health status; 7924 Vietnam-theater and 7364 Vietnam-era veterans completed the telephone interview. Vital status was assessed for the period January 1985-December 2000 with the VA Beneficiary Identification Record Locator Subsystem (BIRLS) Death File, the Social Security Administration Death Master File, and the National Death Index Plus. Cause of death was coded according to ICD. During the telephone interview, veterans were asked about 15 PTSD-related symptoms and their frequency; in 1985-1986, PTSD was diagnosed in a subsample of the veterans (2490 theater veterans and 1972 era veterans) on the basis of personal interviews with the Diagnostic Interview Schedule Version III (DIS-III). With DIS-III, 377 veterans were diagnosed with lifetime PTSD on the basis of combat exposure, which was assessed with the Combat Exposure Scale. Boscarino reported that in the telephone interview there was a clear dose-response relationship between low, moderate, high, and very high combat exposure and whether the criteria for PTSD were met. The Cox proportional-hazards ratio for cancer mortality (188 total deaths) was 1.9 for PTSD-positive Vietnam-theater veterans (95% CI 1.1-3.3, p = 0.018) and 0.9 (95% CI 0.3-3.1) for Vietnam era veterans with the model adjusted for race, Army volunteer status, Army entry age, Army discharge status, Army illicit drug use, age at interview, intelligence, and pack-years of cigarette-smoking. Strengths of this study include a large sample, a sufficient latent period (17 years) for death from cancer, the use of an in-person structured interview to diagnose PTSD, and an assessment of combat exposure; limitations include lack of specification as to cancer type.
126 GULF WAR AND HEALTH The committee identified several secondary studies that assessed the risk of cancer in veterans of Vietnam (Boscarino 1997) or veterans of any war (Schnurr et al. 2000; Spiro et al. 2006) who had PTSD. Boscarino (1997) found no increase in the presence of cancer in Vietnam veterans who had combat-related lifetime PTSD (n = 1067) compared with those without PTSD (n = 332) (OR 0.87, 95% CI 0.25-2.96, p = 0.817) after adjustment for a variety of demographic, social, and Army characteristics. Schnurr et al. (2000) assessed the prevalence of physician- diagnosed medical disorders and combat-related PTSD in 605 veterans of World War II and the Korean War. Veterans were screened for PTSD with the Mississippi Scale for Combat-Related PTSD. The hazard ratio for cancer was a nonsignificant 1.05 (95% CI 0.90-1.21) adjusted for age, smoking, alcohol consumption, and body-mass index (BMI) at study entry. Similar results were seen by Spiro et al. (2006), who examined the association between self-reports of PTSD and health conditions in 2425 veterans who participated in the Veterans Health Study conducted on outpatients at VA ambulatory-care clinics in 1993-1995. PTSD was assessed with the PTSD Checklist, measures of exposure to traumatic events were obtained, and a medical-history interview was conducted that asked about 22 conditions. An OR for cancer of 1.16 (95% CI 0.82-1.65) was seen in veterans with PTSD compared with veterans with depression and veterans with neither depression nor PTSD adjusted for age and for depression. The OR for skin cancer was an insignificant 1.24 (95% CI 0.85-1.79) for veterans with PTSD vs those without PTSD with or without comorbid depression. Summary and Conclusions In general, the three primary studies of male Vietnam veterans considered by this committee did not find statistically significant increases in cancer associated with deployment to Vietnam. The exception is the 50% increase in the risk of NHL in Vietnam veterans compared with era veterans; however, this risk does not appear to correlate with exposure to combat (Selected Cancers Cooperative Study Group 1990a). Two studies of cancer in female Vietnam veterans also failed to find an increased risk of all cancers or gynecologic cancers associated with serving in Vietnam. With respect to the specific question of concern in this review, studies of Vietnam veterans found no evidence that serving in a combat unit increases cancer risk. The studies included a sufficient observation period to detect most cancers that might result from deployment-related stressors experienced in Vietnam. The committee did not identify any secondary studies of cancer in Vietnam veterans. There is no consistent evidence of a higher overall incidence of cancer in Gulf War veterans than in nondeployed veterans based on the four primary studies and nine secondary studies considered by the committee. Only one study found an increased (but nonsignificant) risk of any cancer in Gulf War veterans (McCauley et al. 2002b), but the sample was small, and there was no verification of the self-reported diagnoses. The other three primary studies and the secondary studies were all negative for an increased risk of cancer in Gulf War veterans. However, many of the Gulf War veterans are young for cancer diagnosis, and for most cancers the followup period after the Gulf War has probably been too short to expect their onset. The incidence of and mortality from cancer in general and testicular cancer in particular have been assessed in cohort studies. Results regarding testicular cancer from three primary studies of Gulf War veterans were mixed: one study concluded that there was no evidence of an excess risk, a small registry-based study suggested that there may be an increased risk, and a third study of DoD hospitalization records found a slight increase in the first 5 months immediately after the war, but not in the following 2 years. Although the results are inconsistent,
HEALTH EFFECTS 127 the committee believes that followup is warranted to see whether such an association exists when more time has passed, inasmuch as it is still early for the development of most cancers in Gulf War veterans. In the three primary and two secondary studies considered by the committee, the prevalence of skin cancer does not appear to be increased in deployed veterans of either the Vietnam War or the Gulf War when skin cancer is assessed by dermatologic examination. Boscarino (2005) found an increase in the risk of death from cancer in veterans with PTSD, specifically Vietnam veterans who had participated in the CDC VES. Three secondary studies that assessed the risk of cancer in veterans of other wars, including World War II and the Gulf War, found no increase in the prevalence of any cancer in veterans with combat-related PTSD compared with veterans without PTSD. In general, the studies reviewed by the committee for cancer did not indicate an increased risk of any cancer in Vietnam or Gulf War veterans with the possible exceptions of testicular cancer in Gulf War veterans and NHL in Vietnam veterans. However, the committee emphasizes that given the latent period of most cancers, a sufficient amount of time has not elapsed since deployment for cancer to develop. Furthermore, many of the studies of Gulf War veterans, but not of Vietnam veterans, did not look for cancer as a health effect. The assessment of cancer as a result of deployment is also complicated by other potential exposures, such as to Agent Orange and other environmental contaminants, during both the Vietnam War and the Gulf War. The committee concludes that there is inadequate/insufficient evidence to determine whether an association exists between deployment to a war zone and cancer.
128 TABLE 6-1 Cancer Reference Study Design Population Outcomes Results Adjustments Comments All cancers CDC 1988b Retrospective 2490 Vietnam- Cancer (unspecified) OR 1.4, 95% CI includes Age at enlistment, race, Low participation rate cohort, theater veterans, diagnosed with 1.0, NS year of enlistment, in control group, CI VES prevalence, 1972 Vietnam-era medical examination enlistment status not given population-based, veterans randomly (volunteer vs draftee), telephone selected from 7924 score on general interview theater veterans technical test, primary followed by and 7364 era military occupational screening veterans who had specialty medical entered Army in examination 1965-1971 SCCSG 1990a Case-control 1157 male Cancer diagnoses NHL OR 1.52, 95% CI Registry, age group in Possible exposure to Vietnam-theater from eight city or 1.00-2.32 1968, ethnicity, Agent Orange veterans with state cancer registries education, other considered although NHL, 1776 covariates, including dioxin in blood not Vietnam-era reported exposures to measured controls frequency- pesticides, reported matched by age medical history and drugs, smoking, marital status, being raised Jewish SCCSG 1990b Case-control 342 male Vietnam- Cancer diagnoses Soft-tissue or other Registry, age group in Limited power to theater veterans from eight city or sarcomas OR 0.74, 95% 1968, ethnicity, detect rare outcome with soft-tissue or state cancer registries CI 0.39-1.41, NS education, other other sarcomas, covariates, including 1776 Vietnam-era reported exposures to controls frequency- pesticides, reported matched by age medical history and drugs, smoking, marital status, being raised Jewish
TABLE 6-1 Cancer Reference Study Design Population Outcomes Results Adjustments Comments SCCSG 1990c Case-control 310 male Vietnam- Cancer diagnoses Hodgkinâs disease OR Registry, age group in Limited power to theater veterans from eight city or 1.23, 95% CI 0.65-2.32; 1968, ethnicity, detect differences with Hodgkinâs state cancer registries nasal carcinoma OR 0.31, education, other disease, 48 with 95% CI 0.04-2.20; covariates, including nasal carcinoma, primary liver cancer OR reported exposures to 80 with 0.53, 95% CI 0.14-1.94, pesticides, reported nasopharyngeal all NS medical history and carcinoma, 130 drugs, smoking, marital with primary liver status, being raised cancer; 1776 Jewish, occupational Vietnam-era exposure controls frequency- matched by age Watanabe and Cross-sectional 10,716 Vietnam Cancer mortality All cancers RR 1.20, 95% Race, education, year of Did not adjust for Kang 1995 Marine veterans, CI 0.93-1.55, NS; lung birth, adjusted for latent other risk factors for 9346 non-Vietnam cancer RR 1.33, 95% CI period less than or more cancer, possibly Marine veterans 0.84-2.10; all cancers rate than 16 years of insufficient followup ratio 1.08, 95% CI 0.84- followup period for some 1.39; larynx rate ratio cancers, limited 2.60, 95% CI 0.27-25.0; statistical power for lung rate ratio 1.12, 95% rare causes of death CI 0.71-1.76; lymphosarcoma and reticulosarcoma rate ratio 1.21, 95% CI 0.27-5.41; all cancer rate ratios NS Kang et al. Cohort 4140 female Gynecologic cancers Any gynecologic organ Age, race, branch of 85% statistical power 2000c Vietnam-theater (breast, ovary, uterus, OR 1.14, 95% CI 0.94- service, pay grade, if RR had been at least veterans, 4140 cervix) 1.40; breast OR 1.18, 95% marital status, nursing 1.30 for gynecologic Vietnam-era CI 0.91-1.51; ovary OR occupation, smoking, cancers veteran controls 1.83, 95% CI 0.72-4.61; drinking, family history uterus OR 1.00, 95% CI of cancer, use of birth- 0.61-1.61; cervix OR 1.11, control pills, 95% CI 0.74-1.66; all NS postmenopausal estrogen, progestin 129
TABLE 6-1 Cancer 130 Reference Study Design Population Outcomes Results Adjustments Comments Dalager et al. Cross-sectional 4586 female Cancer mortality All cancers RR 1.00, 95% Rank, nursing status, Limited statistical 1995 Vietnam-theater CI 0.75-1.34, NS duration of military power, small sample, veterans, 5325 service, age at entry into insufficient latent female Vietnam- followup period for outcome, era veterans did not adjust for other risk factors for cancer McCauley et al. Cross-sectional GW veterans: 653 Cancer; all deployed OR 3.0, 95% CI 1.0-13.1 Age, sex, race, region of Possible recall and 2002b Khamisiyah vs nondeployed residence selection bias, limited veterans and 610 generalizability, did non-Khamisiyah not adjust for other veterans, 516 risk factors for cancer nondeployed veterans Macfarlane et Cohort 51,721 UK GW- Cancers identified Any cancer RR 0.99, 95% Sex, age, service branch, Followup was from al. 2003 deployed veterans, from National Health CI 0.83-1.17 rank April 1991 to 50,755 Service Central diagnosis of cancer, nondeployed Register emigration, death, or veterans; random July 2002, too short samples for most cancers to develop Gray et al. 1996 Retrospective 547,076 active- Hospital-discharge Standardized rate ratio Prewar hospitalization, ORs are statistically cohort, duty GW veterans, diagnoses for ~1.10 (exact value not sex, age, race, branch of significantly below 1, hospitalization 618,335 non-GW neoplasms in DoD given), 95% CI 1.0-1.2 for service, marital status, but no values given; veterans facilities last 5 months of 1991; rate rank, length of service, no separation of ratio 0.92, 95% CI ~0.87- salary, occupation specific illnesses 1.0 in 1992; rate ratio 0.94, 95% CI ~0.87-1.1); testicular cancer rate ratio 2.12, 95% CI 1.11-4.02 in 1991, no difference in 1992 and 1993
TABLE 6-1 Cancer Reference Study Design Population Outcomes Results Adjustments Comments Gray et al. 2000 Retrospective 652,979 GW Hospital-discharge DoD PMR 0.98, 95% CI Age, sex, race Able to assess only cohort, veterans, diagnoses for 0.94-1.01; illnesses that resulted hospitalization 652,922 randomly neoplasms in three VA PMR 0.88, 95% CI in hospitalization; selected non-GW- hospital systems: 0.78-0.98; COSHPD PMR possible undetected deployed veterans DoD, VA, COSHPD 0.86, 95% CI 0.61-1.1 confounders Testicular cancer Knoke et al. Cohort, 517,223 active- First diagnosis of Rate ratio 1.05, 95% CI Race or ethnicity, age, Short followup time 1998 hospitalization duty male GW- testicular cancer at 0.86-1.29 occupation, rank, salary, but right age range, deployed veterans, U.S. military length of service, branch no specific exposures (Followup of 1,291,323 hospitals of service, marital status evaluated, Gray et al. nondeployed worldwide July 31, military hospitals only, 1996) active-duty veteran 1991-April 1, 1996 only active-duty controls personnel evaluated Levine et al. Cohort, cancer 621,902 GW Diagnosis of Proportional incidence ratio Age (and age squared), Possible confounding 2005 registry veterans vs testicular cancer 2.33, 95% CI 0.95-5.70 state of residence, by occupational 746,248 from cancer deployment status, race exposures to chemicals nondeployed registries of New in gulf veteran controls Jersey and District of Columbia Skin cancer CDC 1988b Retrospective 2490 Vietnam- Skin cancer OR 0.8, 95% CI 0.4-1.7, NS Age at enlistment, race, Low participation rate cohort, theater veterans, diagnosed via year of enlistment, in control group, CI VES prevalence, 1972 Vietnam-era screening medical enlistment status not given population-based veterans examination (volunteer vs draftee), score on general technical test, primary military occupational specialty Kelsall et al. Cross-sectional, 1456 Australian Skin cancer OR 1.0, 95% CI 0.7-1.3 Service type, rank, age, Low participation in 2004a mailed GW veterans, 1588 determined with education, marital status control group possibly questionnaire nondeployed self-reported leading to participation veterans diagnosis assessed bias by medical practitioner 131
TABLE 6-1 Cancer 132 Reference Study Design Population Outcomes Results Adjustments Comments PTSD and cancer Boscarino 2005 Cross-sectional, 7924 Vietnam- Cancer mortality about Vietnam-theater veterans Race, Army volunteer Risk factors based on mortality in theater veterans 30 years after war; with PTSD status, Army entry age, self-reports, PTSD VES subset of VES (836 with PTSD), mortality and vital HR 1.9, 95% CI 1.1-3.3; Army discharge status, status determined with 7364 Vietnam-era status determined with Vietnam-era veterans Army illicit-drug abuse, nonvalidated veterans (214 withVA BIRLS database, with PTSD HR 0.9, 95% intelligence, age, pack- questionnaire PTSD) SSA Master Beneficiary CI 0.3-3.1 years of cigarette- Record database, NDI smoking Plus NOTE: BIRLS = Beneficiary Identification Record Locator Subsystem Death File, CI = confidence interval, COSHPD = California Office of Statewide Health Planning and Development, GW = Gulf War, HR = hazard ratio, IPSG = Iowa Persian Study Group, NDI = National Death Index, NHL = non- Hodgkinâs lymphoma, NS = not significant, OR = odds ratio, PMR = proportional morbidity ratio, PTSD = posttraumatic stress disorder, RR = relative risk, SCCSG = Selected Cancers Cooperative Study Group, SSA = Social Security Administration, VA = Department of Veterans Affairs, VES = Vietnam Experience Study.
HEALTH EFFECTS 133 ENDOCRINE DISEASES Endocrine diseases include disorders of the adrenals, pituitary, thyroid, parathyroids, pancreas, gonads, and bone. The most common endocrine disorders are diabetes mellitus and disorders of the thyroid, such as hypothyroidism and hyperthyroidism. Hypothyroidism is characterized by deficient secretion of thyroid hormones either primarily because of a defect in the thyroid or secondarily because of a defect in the pituitaryâs production of thyroid-stimulating hormone (TSH, also called thyrotropin). There are two types of diabetes mellitus: type 1, a marked deficiency of pancreatic insulin secretion, and type 2, a combination of insulin resistance and decreased insulin secretion; both types lead to increased serum glucose concentrations. In primary papers that evaluated those health effects in veterans, the committee required that an appropriate laboratory test be used to confirm the presence of diabetes or any thyroid disease or that veterans be receiving medication for the conditions. Papers that relied solely on self-reports of the conditions, even if they were self-reports of a physicianâs diagnosis, were considered to be secondary papers. None of the primary or secondary studies of deployed veterans considered below described diseases of the adrenals, pituitary, or parathyroids. Therefore, the committee assumed that those diseases were not looked for in the population or were so uncommon, even in the studies of large populations of veterans, that they could not be evaluated. The primary studies of diabetes and thyroid disease are summarized in Table 6-2. Diabetes Primary Studies CDC (1988b) completed the VES in 1988, about 15-20 years after the Vietnam War. From a nationally representative random sample of 7924 theater veterans and 7364 era veterans who completed a telephone interview, 2490 of the theater veterans and 1972 of the era veterans also completed physical- and psychologic-health screening examinations in 1985-1986. The prevalence of diabetes in the theater and era veterans was 1.7% and 1.5%, respectively, on the basis of examination by a physician for a nonsignificant OR of 1.1 (p > 0.05) adjusted for age at enlistment, race, year of enlistment, enlistment status, score on general technical tests, and primary military occupational specialty. The committee identified only one primary study that looked at diabetes among U.S. Gulf War veterans (Eisen et al. 2005). In a cross-sectional prevalence study conducted in 2001, 1061 Gulf War-deployed veterans and 1128 nondeployed veterans received physical examinations at 16 VA medical centers. The study was conducted as part of the National Health Survey of Gulf War Era Veterans and Their Families. Study participants were randomly selected for a physical examination from among the 11,441 deployed and 9476 nondeployed veterans who had participated in the 1995 phase of the survey by completing a mail or telephone questionnaire about their health. Diabetes mellitus, defined as a fasting glucose concentration of at least 7.0 mmol/L at examination or as taking a hypoglycemic medication, was found in 4.2% of the deployed and 3.5% of the nondeployed veterans for a nonsignificant OR of 1.52 (95% CI 0.81- 2.85, p = 0.19) adjusted for age, sex, race, cigarette-smoking, duty type, service branch, years of education, and rank. The study is limited by the relatively low participation rate (53% deployed and 39% nondeployed), which may have introduced participation bias.
134 GULF WAR AND HEALTH Secondary Studies Two secondary studies that assessed the association of diabetes with having served in Vietnam were identified. Eisen et al. (1991) used the Vietnam Era [1965-1975] Twin Registry of male-male (monozygotic and dizygotic) twins born in 1939-1957 to assess the prevalence of self-reported diabetes in 2260 pairs of male monozygotic twins who were discordant for the disease. Twins were surveyed by mail or telephone interview in 1987 and asked whether they had ever had diabetes (or 12 other health conditions). Of the twins with and without current diabetes, 50% and 55%, respectively, had served in Southeast Asia for an insignificant OR of 0.8 (95% CI 0.2-3.4). When twins were surveyed as to whether they had ever had diabetes since service in Southeast Asia (as opposed to currently having diabetes), there was still an insignificant OR of 3.0 (95% CI 0.6-14.9), indicating that service in Vietnam was not associated with an increased frequency of diabetes. The study had a high response rate, 74.4%, but is limited by the lack of a physical examination to confirm the presence of diabetes. In the only study that attempted to associate self-reports of diabetes with combat stress, OâToole et al. (1996b) found that the risk of diabetes did not appear to be linked to increasing combat exposure in Australian Vietnam veterans. A random sample of 641 Australian Army veterans posted to Vietnam in 1964-1972 was interviewed with a health survey questionnaire and a combat-exposure index 20-25 years after the war. Combat-exposure responses were divided into severity quartiles. The ORs for diabetes compared with combat exposure were 1.00, 0.56, 0.92, and 0.61 for each quartile of increase in combat score (p = 0.557); the lowest score quartile was used as the referent. Five secondary studies looked at the presence of diabetes among Gulf War veterans, on the basis of self-reports, and found no significant increase in the prevalence of diabetes in Gulf War veterans compared with their nondeployed counterparts. Gray et al. (2002) surveyed members of the U.S. Naval Mobile Construction Battalions (Seabees) in 1997-1999 by mail questionnaire about medical conditions for which they had received a physicianâs diagnosis. Seabees deployed to the gulf region (n = 3831) were compared with Seabees deployed elsewhere (n = 4933) and not deployed (n = 3104). Diabetes mellitus was reported by 1.04% of Gulf War- deployed Seabees, 0.91% of those deployed elsewhere, and 1.61% of nondeployed Seabees. The OR for having a diagnosis of diabetes among Gulf War Seabees compared with nondeployed Seabees was 0.77 (95% CI 0.49-1.23) adjusted for age, sex, active-duty or reserve status, race and ethnicity, current smoking, and current alcohol drinking. Similar results were seen by McCauley et al. (2002b), who conducted a telephone survey of Gulf War veterans from Oregon, Washington, California, North Carolina, and Georgia to assess the prevalence of self-reported medical conditions or hospitalizations 9 years after the war. Veterans were categorized as having been within 50 miles of the Khamisiyah munitions bunker in Iraq (n = 653) with possible exposure to chemical-warfare agents, in the Gulf War region but not in the Khamisiyah area (n = 610), and not deployed to the gulf region (n = 516). The frequency of self-reported diagnoses of diabetes with onset after Gulf War service was 2.3% in all deployed veterans, whether or not they were near Khamisiyah, and 2.5% in nondeployed veterans for a nonsignificant OR of 1.0 (95% CI 0.5-2.4) adjusted for age, sex, race, and region of residence. In 1998, Steele (2000) assessed the prevalence of self-reports of physician-diagnosed or treated diabetes in 1548 Gulf War-deployed veterans and 482 veterans who served elsewhere. All veterans were living in Kansas at the time of the telephone survey; disease onset must have occurred during 1990-1998. The incidence of diagnosis in this time period was 1% in both
HEALTH EFFECTS 135 groups for a nonsignificant OR of 1.22 (95% CI 0.45-3.30) adjusted for sex, age, income, and education level. Simmons et al. (2004) surveyed all 51,581 male UK veterans who served in the Gulf War and a demographically similar comparison cohort of 51,688 UK male veterans who were not deployed to the gulf. Among the 23,358 deployed men who responded, 0.2% reported having diabetes with onset after 1990, as did 0.4% of 17,730 nondeployed men who responded, for a nonsignificant OR of 0.7 (95% CI 0.5-1.0) adjusted for age at the time of the survey, service and rank at the time of the Gulf War, serving status at the time of the survey, alcohol consumption, and smoking. Kang et al. (2000b) conducted a mail and telephone survey to assess the health status of a stratified random sample of Gulf War-deployed veterans compared with veterans who were in the military at the time of the Gulf War but were not deployed to the region. On the basis of 15,817 mail responses and 5,100 telephone responses of self-reports of medical conditions during the preceding 12 months, the authors estimated the population prevalence of diabetes to be 0.8% in Gulf War veterans and 0.9% in Gulf War-era veterans (significant decreased rate difference -0.08, 95% CI -0.11 to -0.05). The authors further estimated that 1.7% of the deployed veterans and 1.5% of the nondeployed veterans had some other endocrine disorder (statistically significant increased rate difference 0.2, 95% CI 0.16-0.24). The response rate in the study was 70%. Thyroid Disease Primary Studies Only one primary study looked at endocrine function in Vietnam veterans. The VES (CDC 1988b) discussed above also assessed thyroid-hormone concentrations in deployed and nondeployed Vietnam veterans 15-20 years after the war. Deployed and nondeployed veterans had the same mean TSH concentration (1.6 mU/L) and the same mean free-thyroxine (free-T4) index (2.2). Of the deployed veterans, 1% had TSH outside the reference range (reference range defined as â¤ 10 mU/L for both cohorts combined) compared with 0.6% of the nondeployed veterans for an OR of 2.0 (95% CI 0.9-4.3); 5.4% of the deployed veterans were outside the reference range of the free-triiodothyronine (T3) index (â¥ the fifth percentile for the combined cohorts) compared with 4.6% of the nondeployed vets for an OR of 1.2 (95% CI 0.9-1.5). The ORs were adjusted for age at enlistment, race, year of enlistment, enlistment status, score on general technical test, and primary military occupational specialty. In the Eisen et al. study (2005) discussed above, thyroid function was assessed in the Gulf War-deployed and nondeployed veterans. Hypothyroidism was defined as an untreated TSH concentration of 10.0 mU/mL or greater or as taking medication for hypothyroidism. Hyperthyroidism required an untreated TSH concentration of less than 0.1 mU/mL or as taking medication for hyperthyroidism. Of the deployed Gulf War veterans, 1.6% had hypothyroidism compared with 1.2% of nondeployed veterans for a nonsignificant OR of 1.70 (95% CI 0.75- 3.87, p = 0.2) adjusted for age, sex, race, cigarette-smoking, duty type, service branch, and rank. Of the deployed veterans, 0.3% had hyperthyroidism compared with 0.1% of the nondeployed veterans for a nonsignificant OR of 4.86 (95% CI 0.68-34.58, p = 0.11) adjusted for age, sex, race, cigarette-smoking, years of education, and duty type. Two retrospective cohort studies by Gray et al. (1996, 2000) looked at hospitalizations of Gulf War veterans for âendocrine, nutritional, and metabolic diseasesâ (ICD-10 E00-E99). In the
136 GULF WAR AND HEALTH first study, DoD hospital-discharge diagnoses by ICD category were determined for 1991, 1992, and 1993, and Gulf War veterans (n = 547,076) were compared with other veterans from the same period (n = 618,335); all were regular active-duty personnel in the Army, Navy (including Marines), and Air Force during the war. All data were obtained from the DMDC. The multivariate ORs for each year were all about 0.9 (exact values not given) adjusted for prewar hospitalizations, sex, age, race or ethnic group, branch of service, marital status, rank, length of service, salary, and occupation; thus deployed Gulf War veterans were not at increased risk for hospitalizations for endocrine, nutrition, or metabolic disorders. In the second study, hospital- discharge records of three hospital systemsâDoD (n = 182,164), VA (n = 16,030), and the California Office of Statewide Health Planning and Development (n = 5185)âfor 1991-1994 were examined. PMRs of hospital-discharge diagnoses by ICD-9 category were compared for Gulf War veterans and veterans not deployed to the gulf. The PMRs were 0.99 (95% CI 0.93- 1.06) for DoD, 1.08 (95% CI 0.92-1.24) for VA, and 0.81 (95% CI 0.48-1.14) for California Office of Statewide Health Planning and Development; all were nonsignificant. All PMRs were adjusted for age and sex, and the DoD PMR was also adjusted for race. Limitations of the studies include the use of hospital-discharge diagnoses and the fact that most endocrine diseases do not require hospitalization. Secondary Studies Several secondary studies assessed thyroid function in Gulf War veterans on the basis of self-reports. A mail survey of the entire Canadian military contingent of 2924 male veterans who served in the Gulf War and 3241 Canadian veterans who were in the military but had not been posted to the gulf region asked about the presence of goiter (a form of thyroid disease) or thyroid trouble. Positive responses were reported by 0.9% of the Gulf War veterans and 0.7% of the nondeployed veterans 20-44 years old and by 2.0% of deployed and 1.3% of nondeployed veterans 45-64 years old; the median age of the deployed was 36 years, and that of the nondeployed was 37 years (Goss Gilroy Inc. 1998). In the Gray et al. (2002) study discussed above for diabetes, 1.15% of the deployed Seabee veterans reported having a thyroid condition, as did 0.69% of the Seabees deployed elsewhere and 0.97% of the nondeployed Seabees. Comparing thyroid conditions in deployed Seabees and nondeployed Seabees yielded a nonsignificant OR of 1.49 (95% CI 0.89-2.5) adjusted for age, sex, active-duty or reserve status, race or ethnicity, current smoking, and current alcohol-drinking. As with diabetes, Steele (2000) conducted telephone interviews of 1545 Gulf War- deployed and 435 nondeployed veterans living in Kansas that elicited self-reports of physician- diagnosed or treated thyroid conditions; the condition must have first occurred in 1990-1998. The prevalence of a new thyroid condition was 2% in the deployed group and 1% in the nondeployed group for a nonsignificant OR of 2.32 (95% CI 0.81-6.67) adjusted for sex, age, income, and education level. Morgan et al. (2000) attempted to determine the effects of military stress on a small group of 72 soldiers undergoing survival training, a realistic simulation of combat. Subjects provided serum samples before the field phase of their training, immediately after a highly intense interrogation phase, and at recovery a day after the end of the experience. During the stress of interrogation, total T3 and free T3 were suppressed. Total and free T4 were slightly increased during interrogation but were decreased at recovery, possibly because of conversion to T3. The authors found an unexplained increase in TSH from baseline to recovery with an
HEALTH EFFECTS 137 intermediate increase at interrogation. The findings do not represent an endocrine disease entity and are similar to other changes in the endocrine system that may occur after acute stress. Posttraumatic Stress Disorder and Endocrine Diseases Several secondary studies of veterans have examined the association between PTSD and endocrine diseases, such as diabetes or thyroid conditions, but the populations have generally been small and the conclusions based on self-reports of PTSD or medical conditions. In an analysis of VES data collected in 1985-1986 on a random sample of 2490 Vietnam-theater veterans, Boscarino (1997) found that 9.0% of 332 veterans with lifetime PTSD and high combat exposure also had an endocrine-nutritional-metabolic disease compared with 4.0% of 1067 veterans without PTSD and low or no combat exposure. Comparison of the two groups for endocrine-nutritional-metabolic disease yielded a nonsignificant OR of 1.58 (95% CI 0.92-2.73, p = 0.100) adjusted for numerous military, demographic, and socioeconomic variables. Boscarino (2004) further assessed the prevalence of autoimmune diseases in the VES cohort of 2490 Vietnam-theater veterans. The author assessed 178 veterans with any of 9 autoimmune diseases for the presence of PTSD: 54 veterans were characterized as having PTSD within the last 30 days on the basis of a DIS and meeting the full Diagnostic and Statistical Manual of Mental Disorders-III (DSM-III) criteria and 124 veterans were characterized as having comorbid PTSD (PTSD concurrent with psychopathology on the basis of the main Minnesota Multiphasic Personality Inventory scales and considered by the author to be a more severe type of PTSD). Of the 178 veterans with an autoimmune disease and PTSD, six had type 1 diabetes mellitus, four had hypothyroidism, and two had Graves disease (the most common form of hyperthyroidism). The OR for veterans with PTSD compared with veterans without PTSD was 2.9 (95% CI 0.9-8.9, p = 0.066) for diabetes, 8.5 (95% CI 1.9-37.9, p = 0.005) for hypothyroidism, and 3.2 (95% CI 0.6-16.5, p = 0.157) for Graves disease. The ORs were adjusted for age, education, race, intelligence, income, geographic region, Army volunteer status, number of times married, and history of antisocial personality, alcohol abuse, drug abuse, and cigarette-smoking. The study had the advantage of using laboratory tests to confirm the presence of disease, but the number of cases was small. The Veterans Health Study assessed the prevalence of PTSD, depression, and several medical conditions in a sample of 2425 male ambulatory-care patients at a VA medical facility in the Boston area (Spiro et al. 2006). PTSD was screened for based on responses to the PTSD Checklist for Civilians, reported exposure to traumatic events was assessed with the Laufer Combat Exposure Scale, and health status was based on the self-administered 36-item Medical Outcomes Study Short-Form (SF-36), an instrument to measure physical and mental health, physical and social functioning, and general well-being. Data were collected in 1993-1995; the response rate was 57.2%. The screening criteria for PTSD were met by 20.2% of the patients; when they were compared with those who did not meet the criteria for either PTSD or depression, the OR for type 2 diabetes was a nonsignificant 0.81 (95% CI 0.59-1.11). For thyroid disease, the risk associated with having PTSD was again nonsignificant (OR 1.27, 95% CI 0.58- 2.79). ORs were adjusted for age and depression. Schnurr et al. (2000) failed to find an association in veterans between combat-related PTSD and physician-diagnosed medical conditions. Of 605 combat veterans of World War II and the Korean War assessed for PTSD with the Mississippi Scale for Combat-Related PTSD in 1990, only six were positive for PTSD. Endocrine disease was found in 92 (15%) of the veterans on medical examination. The hazard ratio for having both PTSD and an endocrine disorder was a
138 GULF WAR AND HEALTH nonsignificant 1.06 (95% CI 0.88-1.27) adjusted for age, smoking, alcohol consumption, and BMI at study entry. Female veterans who attended a VA Puget Sound medical facility in 1996-1998 participated in a cross-sectional survey to determine any association between PTSD (screened for with the PTSD Checklist-Civilian Version) and medical conditions as reported on the veteran version of SF-36 (Dobie et al. 2004). Of the 1259 eligible veterans, 266 were positive for PTSD, and 8.9% of the latter had diabetes and 7.7% of those who were negative for PTSD had diabetes, for a nonsignificant OR of 1.51 (95% CI 0.90-2.52) adjusted for age. Several small studies have examined the association between thyroid hormones and combat-related PTSD in veterans of various wars. Mason et al. (1994) conducted serum radioimmunoassays for total T4, free T4, total T3, free T3, T4-binding globulin, and TSH in 96 male Vietnam veterans who had PTSD and were patients at VA inpatient treatment groups. Veterans with PTSD were compared with 24 demographically similar community controls who did not have PTSD (11 with combat exposure). The PTSD group had moderately higher total T4, total T3, free T3, and T3:T4 ratios, but not free T4. There was also an increase in T4-binding globulin in the PTSD group but no difference between the groups in TSH. Similar TSH in the two groups suggests that there was no significant difference in thyroid metabolic disease between them. A second study to investigate serum thyroid-hormone concentrations in 65 male Vietnam veterans with PTSD was conducted by Wang et al. (1995). PTSD symptom severity was rated with the Clinician-Administered PTSD Scale (CAPS). Increased CAPS scores correlated positively with free T3, total T3, and total T4, but not free T4 or T4-binding globulin. When serum thyroid hormones were studied in 12 World War II veterans with PTSD and 18 without PTSD, similar correlations were seen between total and free T3 and PTSD severity (particularly hyperarousal symptoms), but significant differences were not found in total T4, free T4, T4-binding globulin, or TSH between the groups, although there was a slight trend toward higher TSH in the PTSD group (Wang and Mason 1999). Obesity Exposure to stressors may affect eating behavior and theoretically might predispose to obesity or eating problems (see Chapter 4 for a discussion of the effects of stress on obesity). No primary studies of obesity in veterans were identified. A few secondary studies that examined small, selected samples, such as David et al. (2004), or used only self-report measures suggest that PTSD may be associated with an increased frequency of obesity and eating disorders. Dobie et al. (2004) screened female veterans who received care at a VA medical facility in 1996-1998 with a mail questionnaire that included the PTSD Checklist-Civilian Version and SF-36. A comparison of 940 veterans who were negative for PTSD and 266 who were positive for PTSD found that 60 and 66 of them, respectively, were also positive for an eating disorder (OR 5.00, 95% CI 3.37-7.38). Obesity (defined as a BMI over 30) was reported by 35.3% of those without PTSD and 47.0% of those with PTSD, respectively (OR 1.78, 95% CI 1.34-2.35). Vieweg et al. (2006) examined 144 male Vietnam veterans with PTSD who were 50-59 years old and were patients at a VA medical center. Their mean BMI was 30.2, and 82.8% of them were overweight or obese, far exceeding the current general U.S. population rate of 64.5%. Other studies, such as Stretch et al. (1995), which had substantial limitations, did not report such an association. Fielder et al. (2006) looked at the 12-month prevalence of bulimia
HEALTH EFFECTS 139 and anorexia nervosa in 892 male and 75 female Gulf War-deployed veterans 10 years after the war and found no evidence of either eating disorder. Summary and Conclusions Acute stress and chronic stress activate the endocrine system and thereby influence the immune system, as described in Chapter 4, but it is unclear whether these interactions produce endocrine diseases. Stress also increases caloric intake, and the hormones released by acute and chronic stress, such as cortisol, can accentuate obesity and lead to insulin resistance, a central feature of type 2 diabetes (see Chapter 4). There are several reports that stress with or without comorbid depression has increased the incidence of type 2 diabetes in nonmilitary populations (Kawakami et al. 1999; Mooy et al. 2000; Eaton et al. 1996). Two primary studies examined the association between deployment and diabetes in Vietnam War veterans (CDC 1988b) and Gulf War veterans (Eisen et al. 2005) and found no increase in the risk of diabetes in deployed veterans of either war. Several secondary studies, one of Vietnam veterans and five of Gulf War veterans, supported the lack of association between deployment to either the Vietnam War or the Gulf War and the presence of diabetes, and one study (OâToole et al. 1996b) in Vietnam veterans also showed no increase in diabetes with increasing combat exposure. Furthermore, there were no significant associations between the presence of thyroid diseases and deployment to a war zone, whether in Vietnam or in the Persian Gulf. The four primary studies, one of Vietnam veterans and three of Gulf War veterans, showed no evidence of increased or decreased thyroid function in deployed veterans compared with their nondeployed counterparts. The four secondary studies, all of Gulf War veterans, also showed no association between deployment and thyroid disease. The prevalence of thyroid disease of about 1-2% was slightly lower in both deployed and nondeployed veterans than in the general populationâ5% for hypothyroidism and 1% for hyperthyroidism (Hollowell et al. 2002). That finding may reflect a male-dominated military inasmuch as thyroid disease is much more common in women than in men. Five secondary studies that compared endocrine diseases in veterans with PTSD and those without PTSD did not show any significant changes in diabetes or thyroid disease in the two groups, except that one study, by Boscarino (2004), found that Vietnam veterans with PTSD had a greater risk of hypothyroidism. Two studies were of Vietnam veterans, one of veterans of World War II and the Korean War, one of a general veteran population in Boston, and one of a general female veteran population. Many case-control studies of nonmilitary populations have reported a significant relationship between stressful events and the onset of Graves disease, the most common form of hyperthyroidism (Mizokami et al. 2004). In the three small thyroid-hormone studies of veterans, changes in serum thyroid- hormone concentrations were found in veterans with PTSD compared with veterans without PTSD (Mason et al. 1994; Wang and Mason 1999; Wang et al. 1995) and in men experiencing military survival training (Morgan et al. 2000). However, that no differences in TSH concentrations were seen in those groups suggests that they had no hyperthyroidism or hypothyroidism. No primary studies of obesity in veterans were identified. One secondary study of male Vietnam veterans found that veterans with PTSD were more overweight than the general U.S. population, a second found that female veterans with PTSD were more likely to have an eating
140 GULF WAR AND HEALTH disorder than those without PTSD, and a third found no evidence of eating disorders in Gulf War veterans. The committee concludes that there is inadequate/insufficient evidence to determine whether an association exists between deployment to a war zone and diabetes mellitus. The committee also concludes that there is inadequate/insufficient evidence to determine whether an association exists between deployment to a war zone and thyroid disease.
TABLE 6-2 Endocrine Diseases Study Study Design Population Outcomes Results Adjustments Comments CDC 1988b Retrospective 2490 Vietnam- Screening medical Diabetes OR 1.1, 95% CI Age at enlistment, race, Low participation cohort, prevalence, theater veterans, examination, includes 1.0; out of TSH year of enlistment, rate in control VES population-based, 1972 Vietnam-era standard reference range OR 2.0, enlistment status group telephone interview veterans randomly hematologic assays, 95% CI 0.9-4.3; out of (volunteer vs draftee), with screening selected serum chemistry reference range for free-T3 score on general medical examination assays index OR 1.2, 95% CI 0.9- technical test, primary at followup 1.5 military occupational specialty Eisen et al. 2005 Cross-sectional, 1061 GW-deployed Diabetes, Diabetes OR 1.52, 95% CI Age, sex, race, cigarette- Low participation prevalence, veterans, 1128 hypothyroidism, 0.81-2.85; smoking, duty type, rates, deployed NHSGWEVTF population-based nondeployed hyperthyroidism hypothyroidism OR 1.70, service branch, (53%), veterans 95% CI 0.75-3.87; education, rank nondeployed (Derived from hyperthyroidism OR 4.86, (hyperthyroidism not (39%) Kang et al. 2000b) 95% CI 0.68-34.58; all NS adjusted for service branch or rank) Gray et al. 1996 Retrospective 547,076 active-duty Hospital-discharge OR about 0.9 (exact value Prewar hospitalization, ORs are cohort, GW veterans, diagnoses for not given), 95% CI < 1.0 sex, age, race, branch of statistically hospitalization 618,335 non-GW endocrine- for all 3 years service, marital status, significantly veterans nutritional-metabolic rank, length of service, below 1, but no disease salary, occupation values given; no separation of specific illnesses Gray et al. 2000 Retrospective 652,979 GW- Hospital-discharge DoD PMR 0.99, 95% CI Age, sex, race Able to assess cohort, deployed veterans, diagnoses for 0.93-1.06); only illnesses that hospitalization 652,922 randomly endocrine, VA PMR 1.08, 95% CI resulted in selected nutritional, and 0.92-1.24 ; COSHPD PMR hospitalization; nondeployed metabolic disease in 0.81, 95% CI 0.48-1.14 possible veterans three hospital undetected systems: DoD, VA, confounders COSHPD NOTE: CI = confidence interval, COSHPD = California Office of Statewide Health Planning and Development, DoD = Department of Defense, GW = Gulf War, NHSGWEVTF = National Health Survey of Gulf War Era Veterans and Their Families, NS = not significant, OR = odds ratio, PMR = proportional morbidity ratio, PTSD = posttraumatic stress disorder, T3 = triiodothyronine, TSH = thyroid-stimulating hormone, VA = Department of Veterans Affairs, VES = Vietnam Experience Study. 141
142 GULF WAR AND HEALTH PSYCHIATRIC DISORDERS Numerous investigators have suggested that psychiatric disorders may occur as a consequence of serving in the military during wartime. There are reports associating war-zone trauma with mental illnesses as far back as the Civil War (Pizarro et al. 2006; Wessely 2005), even though the terminology used to classify war-related psychiatric traumatic events has changed considerably. Despite these changes in terminology, there is a substantial literature on the psychiatric effects of war in general, mainly in the context of trauma. There are reports on the prevalence of a wide array of psychiatric disorders and their association with being a veteran (for example, Beebe 1975; Robins et al. 1974; Snow et al. 1988) and with specific combat experiences, including witnessing death and atrocities (Archibald and Tuddenham 1965; Grinker and Spiegel 1945; Kardiner and Spiegel 1947). This section focuses on assessing the evidence of a relationship between deployment-related stress and psychiatric disorders. Substance-use disorders are discussed in the next section. The challenge for the committee in understanding the plethora of studies available on several potentially diverse cohorts of veterans was first to distinguish clearly between the primary and secondary studies and then to identify studies that explicitly used deployment to a war zone as the exposure of interest in contrast with studies that used serving in the military during wartime, but not necessarily deployment to a war zone, as an overall indicator of risk of development of any psychiatric disorder. As described in Chapter 1, the committee agreed that studies that used PTSD as a presumed marker for deployment-related trauma would also be included in this analysis. Primary studies provided the basis of the committeeâs findings on the relationship between deployment-related stress and psychiatric effects. Primary studies were those in which exposure status was determined according to whether subjects had been deployed, not deployed, or deployed to a nonwar zone (for example, Germany) or that used a diagnosis of PTSD as a marker of war-zone trauma. To diagnosis psychiatric disorders and PTSD, primary studies also included an in-person interview and used either the Structured Clinical Interview for DSM-III-R (SCID); the Composite International Diagnostic Interview (CIDI), a comprehensive and standardized diagnostic interview designed for assessing mental disorders according to the definitions of the Diagnostic Criteria for Research of ICD-10 and DSM-III-R; or the DIS, a research diagnostic interview that assesses psychiatric disorders according to DSM-III-R criteria but is designed to be administered by trained lay interviewers; or the CAPS. Primary studies are summarized in Table 6-3. The secondary studies reviewed and evaluated the same functional domains of interestâ such as depression, anxiety, and substance useâand add supplementary information that may increase or decrease confidence in the conclusions drawn from the primary studies. Primary Studies The committee identified 11 citations for 7 primary studies (CDC 1988a; Dohrenwend et al. 2006; Fiedler et al. 2006; Ikin et al. 2004; Jordan et al. 1991; Koenen et al. 2003a,b; Kulka et al. 1990; Proctor et al. 2001; Toomey et al. 2007; Wolfe et al. 1999). The VES was a major study of U.S. Army Vietnam veterans that began with a postservice mortality study (CDC 1987). CDC (1988a) used the cohort to identify and recruit a
HEALTH EFFECTS 143 subsample for phase I of 7924 U.S. Army Vietnam-theater veterans and 7364 Vietnam-era veterans. In phase I, the veterans participated in telephone interviews about their psychosocial health, physical health, and reproductive outcomes. In phase II conducted in 1985-1986, 3317 of the theater veterans and 3126 of the era veterans were invited to participate in a comprehensive health examination to determine the effects of war-zone deployment on psychosocial characteristics (CDC 1988a). Of the theater and era veterans, 2490 (75%) and 1972 (63%), respectively, were enrolled in phase II, which included an examination of psychologic health. The investigators used the DIS and the Minnesota Multiphasic Personality Inventory to assess the veterans for PTSD, alcohol abuse, drug abuse, generalized anxiety, and depression. Six characteristics were controlled for: age at entry into the Army, race, score on the enlistment general technical test (a measure of mental aptitude), enlistment status (drafted or volunteered), year of entry into the Army, and primary military occupational specialty (tactical or nontactical). Vietnam veterans were significantly more likely than era veterans to meet the DIS criteria for generalized anxiety (OR 1.5, 95% CI 1.1-2.1) and depression (OR 2.0, 95% CI 1.4-2.9) in the month prior to examination. Only 2.2% of the theater veterans met the full DIS criteria for combat-related PTSD in the previous month (current), although 14.7% had ever met the criteria (lifetime); results for era veterans were not given. The congressionally mandated NVVRS constituted nationally representative cohorts of male and female Vietnam theater veterans, era veterans, and civilians; the veteran cohorts were randomly chosen from among all military personnel who had served on active duty during the Vietnam War and who had left service as of September 1987 (Kulka et al. 1990; Weiss et al. 1992). All study participants were interviewed in November 1986-February 1988. The NVVRS was created to determine the prevalence of specific psychiatric disorders and other adjustment problems, both during the course of life (as a measure of predisposing factors) and at the time of the interview. It also included self-reports of war-zone stressor exposures used to create a war- zone stress index to characterize theater veterans as having been subjected to âhigh exposureâ or âlow or moderate exposure.â A sample of veterans and civilians was obtained from the National Personnel Records Center, the DMDC, and a list developed by VA of all female theater veterans. The NVVRS conducted a total of 3016 interviews, in what was called the National Survey of the Vietnam Generation, with 1200 male theater veterans, 412 male era veterans, 448 male civilians, 432 female theater veterans, 304 female era veterans, and 218 female civilians. The NVVRS also included a preliminary validation component for the survey-based PTSD measures and a clinical- interview component. Response rates were over 83% for Vietnam-theater veterans, 76% for Vietnam-era veterans, and 70% for civilians of those sampled and eligible. Detailed reviews and analyses of military data records were conducted to identify potential differences between veteran respondents and nonrespondents. Although no significant differences were identified, the investigators weighted the data for all the analyses to account for interview-level nonresponse and different probabilities of selection in the civilian cohort. The investigators also controlled for age, race or ethnicity, and female veteran occupation. The occupation of the female veterans was important because many of them were likely to be nurses and therefore to have different exposures to war-zone stressors than their male counterparts (Kulka et al. 1990). Although the DIS was used to interview both veterans and civilians, only a subsample of 260 veterans underwent a semistructured clinical interview by experienced mental-health professionals to also diagnose PTSD and to validate the use of the interview instrument for PTSD used by the researchers. A composite PTSD diagnosis for the subsample was made based on the interviews,
144 GULF WAR AND HEALTH the Mississippi Scale for Combat-Related PTSD, the SCID, and the PTSD Scale of the Minnesota Multiphasic Personality Inventory. The estimated prevalence of current PTSD was 15.2% in all male theater veterans and 8.5% in female theater veterans (Kulka et al. 1990). The prevalence of current (6-month) PTSD was 2.5% in male and 1.1% in female era veterans and 1.2% and 0.3%, respectively, in civilians. The prevalence of lifetime PTSD was 30.6% in male theater veterans and 26.9% in female theater veterans. High levels of war-zone stress, including exposure to at least one traumatic event, were reported by 75.2% of the male theater veterans and 9.2% of the era veterans, and 32.9% of the veterans reporting low or moderate war-zone stress also reported definite traumatic experiences. The prevalence estimate of PTSD among veterans with high war-zone exposure was 35.8% in men and 17.5% in women. The likelihood of having PTSD was increased in men who were in the Army or Marine Corps, were in junior enlisted pay grades, were 17-19 years old when they entered Vietnam, had served on active duty for more than 4 but less than 20 years, had ever been wounded or injured in combat, or had received a combat medal. Women who had more war-zone stress, were younger, or had been on active duty for more than 4 but less than 20 years had increased prevalence of PTSD. PTSD was highly comorbid with other psychiatric disorders: 75% of men with PTSD had a lifetime diagnosis of alcohol abuse or dependence, 44% had generalized anxiety disorder (GAD), and more than 20% had another psychiatric disorder. The NVVRS has been criticized for overestimating the lifetime rate of PTSD. To address that issue, Dohrenwend et al. (2006) reanalyzed the NVVRS data from the military records of 1200 Vietnam theater veterans with a records-based military-history measure (MHM) of exposure to war-zone stressors. They found that 96.5% of veterans categorized as probably having had low exposure with the MHM had reported their exposure as low or moderate in the NVVRS, and 72.1% of those categorized as having had very high exposure with the MHM had reported their exposure as high. Over 86% of the veterans with war-related PTSD described events judged by the blind raters to have been personally life-threatening or to have involved witnessing death of or physical harm to others. A diagnosis of PTSD was based on the SCID and included a functional assessment with the Global Assessment of Functioning scale. The military records of the clinical subset of 260 veterans examined for the NVVRS were re-examined for this study. A dose-response relationship between PTSD and exposure to war-zone stressors was established. Current (as of 1988) war-related PTSD was diagnosed in 0.3% of low-exposure veterans, 14.4% of moderate-exposure, 27.0% of high-exposure, and 28.1% of very high- exposure. Lifetime war-related PTSD prevalence was 22.5%, and current war-related PTSD prevalence 12.2%. When Dohrenwend et al. adjusted the diagnoses for impairment of functioning and documentation of exposure to war-related traumatic events, the prevalence of lifetime and current war-related PTSD dropped to 18.7% and 9.1%, respectively, in the veteran subset. Jordan et al. (1991) used the NVVRS data to determine that the most prevalent current disorders among male theater veterans were alcohol abuse or dependence (11%) and GAD (5%); these rates were not different from those of Vietnam-era veterans or civilians. The most prevalent current disorders in female Vietnam-theater veterans were depression and GAD (both 4%). Although the rate of depression was higher in female theater veterans than in female era veterans or civilians, that was not the case for GAD, which did not differ significantly among the groups. Overall, Jordan et al. (1991) found that the NVVRS showed few differences in the prevalence and distribution of psychiatric disorders among veterans who saw combat in Vietnam
HEALTH EFFECTS 145 and veterans who did not. The major differences appear to be related to the level of combat stress. Of male theater veterans exposed to high war-zone stress, 43% met the criteria for a specific psychiatric disorder other than alcohol abuse or dependence. That was significantly higher than rates found for either low or moderate war-zone stress in male theater veterans (21.1%), era veterans (25.6%), or civilians (18.1%). Because the use of self-reported war-zone stress-exposure information was found to correspond well with the information found in the military records, such as receipt of combat medals, this is one of the few early studies that differentiated between the types of stress encountered in a war zone. Among male veterans with high war-zone stress exposure, 63.1% had at least one lifetime psychiatric disorder and 29.8% had a current diagnosis. Compared with the 783 male veterans with low war-zone stress, the 406 with high war-zone stress had more GAD (10.8% vs 24.4%, p < 0.05), dysthymia (1.4% vs 12.3%, p < 0.001), major depression disorder (MDD) (3.1% vs 11.1%, p < 0.01), obsessive- compulsive disorder (0.5% vs 5.5%, p < 0.01), and antisocial personality disorder (7.1% vs 16.6%, p < 0.01). Female veterans had fewer psychiatric disorders associated with high war-zone stress, although 22.3% had lifetime major depression and 9.9% had lifetime dysthymia, and these rates were significantly higher than in female veterans exposed to low war-zone stress. Two primary studies were included that used data from the Vietnam Era Twin Registry study, a longitudinal twin study. Koenen et al. (2003b) used the Combat Exposure Index and the DIS-III-R to measure PTSD and other psychiatric outcomes in monozygotic twin pairs. The cohort consists of male-male twin pairs born in 1937-1957 in which both members of the pair served in the military during the Vietnam era (Eisen et al. 1991). Zygosity was determined by a questionnaire and blood-group typing methods that achieved 95% accuracy. Thirty-seven male twin pairs were identified in which one twin was a Vietnam-theater veteran with combat-related PTSD and a score of at least 7 on the Combat Exposure Index, and 76 male twin pairs were identified in which one twin was a Vietnam-theater veteran without lifetime PTSD and score of at least 7 on the Combat Exposure Index; all co-twins were not deployed to Vietnam. Deployed twins with PTSD were more likely than twins without PTSD to have any mood disorder (OR 8.89, 90% CI 3.66-21.59), MDD (OR 6.32, 90% CI 2.57-15.55), dysthymia (OR 3.80, 90% CI 1.09-13.28), any anxiety disorder (OR 8.63, 90% CI 2.20-33.83), or panic disorder (OR 4.48, 90% CI 1.04-19.41). Twins with PTSD were also more likely than their co-twins to have any mood disorder (OR 4.00, p < 0.05) or MDD (OR 2.50, p < 0.05) but not other comorbid disorders. Combat-exposed twins that had diagnoses of any of the comorbid psychiatric disorder above also reported more PTSD symptoms in all three symptom clusters (see Chapter 5), including symptoms that had no diagnostic overlap with the comorbid disorders, than did veterans without these disorders. In another study of the same Vietnam Era Twin Registry cohort, Koenen et al. (2003a) used the entire sample of all 1874 monozygotic twin pairs from the Vietnam Era Twin Registry on whom complete DIS-III-R diagnostic information was available. The study focused on five lifetime diagnoses: MDD, alcohol dependence, drug dependence, cannabis dependence, and tobacco dependence. (Information on substance dependence from the study is discussed in the next section of this chapter.) Measures used were a four-level index of combat exposure (Janes et al. 1991) and a structured psychiatric interview administered to the Vietnam Era Twin Registry twins by telephone with DIS-III-R. Participants were given a diagnosis of combat PTSD (C- PTSD) if they reported combat as a traumatic event and met DSM-III-R criteria for PTSD in relation to the event. Conditional logistic regression was used to account for the paired structure of the data and to calculate the matched pairsâ ORs and 95% CIs. Premilitary trauma history, age
146 GULF WAR AND HEALTH at entry into the military, and education level at the time of entry into the military were controlled for in three models for each psychiatric diagnosis. Combat exposure unadjusted for C- PTSD was found to be significantly associated with MDD (OR 1.22, 95% CI 1.03-1.44) but it was insignificant when adjusted for C-PTSD (OR 1.10, 95% CI 0.92-1.33). C-PTSD unadjusted for combat exposure was also significantly associated with MDD (OR 3.04, 95% CI 1.48-6.24); however, after adjustment for combat exposure, the OR for the association between C-PTSD and MDD was substantially reduced (OR 2.55, 95% CI 1.16-5.61). The investigators suggest that C- PTSD is a mediator of the relationship between combat exposure and MDD. Although the study design using monozygotic twins eliminates the confounding effects of genetic and shared environmental risk factors in the relationships examined, it remains unclear whether other, unmeasured factors might increase the risk of C-PTSD and other mental disorders. Ikin et al. (2004) conducted in-person interviews of 1381 Australian Gulf War veterans using the CIDI for psychiatric disorders as part of a cross-sectional survey of all Australian veterans deployed to the Gulf War and 1377 comparison veterans not deployed to the gulf. They found that the prevalence of any psychiatric disorder (first present after the Gulf War) was 30.8% in Gulf War veterans and 21.1% in the comparison group (OR 1.6, 95% CI 1.3-1.9) controlled for service type, rank, age, education, and marital status. Deployed veterans were almost 4 times as likely to meet criteria for PTSD (OR 3.9, 95% CI 2.3-6.5) and also were at increased risk for MDD (OR 1.6, 95% CI 1.3-2.0). Proctor et al. (1998) selected stratified random samples of two demographically heterogeneous Gulf War cohorts from Fort Devens, Massachusetts (n = 2313) and New Orleans (n = 928) that were being studied longitudinally. The resulting subcohorts (Fort Devens, n = 186; New Orleans, n = 66) were compared to a small comparison population of Germany-deployed veterans from the Maine National Guard (n = 48). This study met the criteria for a primary study in that the in-person CAPS was used for PTSD and the SCID was used for other psychiatric disorders. Data were collected in 1994-1996. The samples were selected to produce equal representation of high and low symptom veterans. After adjustment for oversampling of women, participation bias, and age, sex, and education differences, both gulf-deployed groups had higher symptom prevalence of the items that make up the body-system symptom scores in the SF-36 than the Germany-deployed group. Approximately 5% of the Fort Devens cohort, 7% of the New Orleans cohort, and none of the Germany cohort were diagnosed with current PTSD and those with PTSD scored higher on the Expanded Combat Exposure Scale. Compared with the Germany-deployed group, significant differences were found for both gulf-deployed groups in reporting of frequent periods of anxiety or nervousness: OR 7.1 for the Fort Devens cohort and 5.3 for the New Orleans cohort (95% CI excludes 1.0 for both ORs). Self-reported symptoms of frequent periods of feeling depressed were significant only for the Fort Devens cohort (OR 6.0, 95% CI excludes 1.0). Using the same well-characterized Fort Devens, New Orleans, and Germany-deployed subcohorts as Proctor et al., Wolfe et al. (1999) used the SCID to examine the relationship of psychiatric disorders to health problems. The prevalences of panic disorder, agoraphobia, social phobia, simple phobia, obsessive-compulsive disorder, GAD, and somatoform disorder did not differ between the gulf-deployed and Germany-deployed cohorts. There were slight but significant differences in the rates of current and lifetime PTSD between each of the gulf- deployed groups (5.4% and 6.5%, respectively, in the Fort Devens group and 7.2% and 8.2% in the New Orleans group) and the Germany-deployed group (no PTSD). The largest differences were seen in MDD: all three groups had 2.5-3 times more lifetime MDD than current MDD, and
HEALTH EFFECTS 147 the Fort Devens group had twice the prevalence of lifetime MDD as the New Orleans group (22.5% Fort Devens, 10.2% New Orleans, and 4.2% Germany-deployed). The investigators found a small but significant association between increased health-symptom reports and the diagnosis of PTSD or MDD for the combined deployed groups. However, almost two-thirds of gulf participants reporting moderate to high health symptoms were found to have no axis I (major psychiatric) disorders. The rates of most psychiatric disorders were lower than national comorbidity estimates except PTSD, MDD, and dysthymia, which had rates similar to those in the National Comorbidity Survey (NCS) (Kessler et al. 1995). The Wolfe et al. study had a participation rate of 62% for the Fort Devens cohort and 85% for the Germany-deployed unit, but budget constraints limited participation by the New Orleans cohort to 37%. Fiedler et al. (2006) compared the prevalence of MDD, panic attacks, social phobia, obsessive-compulsive disorder, GAD, and any psychiatric disorder in random samples of 967 Gulf War-deployed veterans and 784 era veterans. The sample was obtained from the DMDC. Study participants were administered the 12-month version of the CIDI using the DSM-IV criteria to assess PTSD and the CIDI-Short Form/DSM-IV for the remaining psychiatric disorders by telephone interview with computer-assisted technology. Deployed veterans had a significantly greater prevalence of any psychiatric disorder than nondeployed veterans (26.1% vs 16.1%, p < 0.05) including MDD (15.1% vs 7.8%), panic attack (1.6% vs 0.5%), social phobia (3.6% vs 1.7%), obsessive-compulsive disorder (2.8% vs 1.1%), PTSD (3.4% vs 0.9%), GAD (6.0% vs 2.7%) and any anxiety disorder (16.0% vs 9.7%). After control for demographic variables and other psychiatric disorders, deployed veterans were found to have higher risks of MDD (OR 2.07, 95% CI 1.50-2.85) and any anxiety disorder (OR 1.81, 95% CI 1.34-2.45). Lower rank, female sex, and divorced-single status all were significant independent predictors of psychiatric disorder. The study is limited in that the response rate was suboptimal (55%) and responders differed from nonresponders in that a greater proportion of whites and officers volunteered for the study. In a recent study by Toomey et al. (2007), 1061 veterans deployed to the Gulf War were compared with 1128 Gulf War-era veterans. They used the randomly selected subset of 11,441 deployed and 9476 nondeployed veterans who have been followed since 1995 for the National Health Survey of Gulf War Era Veterans and Their Families (Eisen et al. 2005). The study used structured clinical interviews for all psychologic examinations conducted in 1998-2001; PTSD was diagnosed with the CAPS, and the CIDI was used to diagnose the other axis I psychiatric disorders. The investigators calculated the prevalence of mental disorders beginning during the deployment period and evaluated their prevalence 10 years later. They found that 10 years after the end of the Gulf War, those deployed to that region had a persistent increased prevalence of mental disorders, especially PTSD (OR 5.78, 95% CI 2.62-12.74), all anxiety disorders (OR 4.43, 95% CI 2.49-7.88), and MDD (OR 1.81, 95% CI 1.03-3.32). The multiple-regression model was adjusted for age, sex, ethnicity (white vs other), years of education (less than 12 vs 12 or more), duty type (active vs reserve or Guard), service branch (Army or Marines vs Navy or Air Force) and rank (enlisted vs officer) unless otherwise noted. The participation rates in the study were 53% for the deployed veterans and 39% for the nondeployed; the authors examined possible participation bias and determined that, overall, participation bias was independent of deployment status.
148 GULF WAR AND HEALTH Secondary Studies Secondary studies were based primarily on retrospective or cross-sectional study designs with self-reported information whose major limitations were poor response rates and the potential for recall bias. The number and types of psychiatric disorders reported varied widely from study to study and included current (those occurring in the preceding month or week), 12- month, and lifetime reports mainly of MDD, GAD, alcohol abuse or dependence, specific phobias, and PTSD. Although many of the studies used different case definitions or different screening instruments, one (Eisen et al. 2004) assessed the test-retest reliability of the DIS-III-R interviews used. The investigators found that in general test-retest reliability of the lifetime prevalence of most of the DSM-III-R psychiatric diagnoses was acceptable (Kappa statistics 0.54-0.76); however, reliability was poor (Kappa ï statistic < 0.40) for panic disorder and GAD. Eleven secondary studies on nine veteran populations used self-reported data to consider whether deployed veterans differed from nondeployed veterans for a wide variety of psychiatric disorders (Black et al. 2004a; Dlugosz et al. 1999; Erickson et al. 2001; Forman-Hoffman et al. 2005; Goss Gilroy Inc. 1998; Gray et al. 2002; Grieger et al. 2006; Hoge et al. 2004, 2006; OâToole et al. 1996a; Simmons et al. 2004). OâToole et al. (1996a) studied Australian Vietnam veterans, and Hoge et al. (2004, 2006) studied U.S. troops returning from OEF and OIF; the other eight studies studied Gulf War veterans. In 1997, Goss Gilroy Inc. (1998) mailed a health-status questionnaire to all Canadian military personnel who had deployed to the Gulf War (n = 4262) and a comparison group of Canadian military personnel who had served elsewhere (n = 5699), matched on age and gender. The response rates were 73% (n = 3113) and 60.3% (n = 3439), respectively. The major self- reported health outcomes found to be significantly increased in deployed Gulf War veterans, compared with nondeployed veterans, were health provider-diagnosed PTSD (OR 3.34, 95% CI 2.13-5.26), anxiety (OR 2.20, 95% CI 1.55-3.12), MDD (OR 3.67, 95% CI 3.04-4.44), and symptoms of PTSD (OR 2.69, 95% CI 1.69-4.26). Gray et al. (2002) conducted a cross-sectional survey in 1997-1999 of 3831 active-duty and reserve Navy Seabees who had served in the Gulf War and 3104 Seabees who had remained in the United States. The mailed survey gathered self-reports on prewar medical history, war exposure, symptoms, geographic service during the war, PTSD, depression, and anxiety. For deployed vs nondeployed Seabees, risks were significantly increased for depression (OR 1.77, 95% CI 1.41-2.27), PTSD (OR 4.23, 95% CI 2.59-6.92), and suicidal thoughts (OR 2.16, 95% CI 1.64-2.84) when adjusted for age, sex, active-duty or reserve status, race or ethnicity, current smoking, and current alcohol drinking. Using the Iowa Persian Gulf cohort, Black et al. (2004b) administered a telephone survey to a population-based sample of 4886 members of the military in Iowa who were enlisted at the time of the Gulf War. Subjects were randomly selected from Gulf War regular military and National Guard or reserves and non-Gulf War regular military and National Guard or reserves. Medical and psychiatric disorders were assessed in 3695 of those selected, with a final participation rate of 76%. Deployed military personnel (n = 1896) were twice as likely as nondeployed military personnel (n = 1799) to report having a current anxiety disorder (estimated prevalence 4.0% [SE 1.0] vs 1.8% [SE 0.4]) or any anxiety disorder (estimated prevalence 5.9% [SE 0.6] vs 2.8% [SE 0.5]). In a multivariate model, predeployment psychiatric difficultiesâ which included predeployment psychiatric treatment and predeployment diagnosis of PTSD, depression, or anxietyâwere independently associated with current anxiety disorder; this suggested that predeployment precursors were associated with development of anxiety (OR 4.4,
HEALTH EFFECTS 149 95% CI 2.2-8.9). Participation in combat was strongly associated with the development of various manifestations of anxiety: PTSD (OR 2.1, 95% CI 1.7-4.2), panic disorder (OR 6.2, 95% CI 3.0-13.0), and GAD (OR 2.1, 95% CI 1.2-3.5). Using self-reports of symptoms occurring after 1990 from 1296 male UK veterans deployed to the Gulf War compared with 12,364 male UK veterans who were fit for duty but not deployed to the gulf, Simmons et al. (2004) found significantly (p < 0.001) increased risks of depression (OR 16.1, 95% CI 12.7-20.4); anxiety, stress, or sleep disturbance (OR 10.8, 95% CI 8.7-13.5); mood swings, aggression, or irritability (OR 16.1, 95% CI 13.2-19.7); and PTSD and associated symptoms (OR 34.9, 95% CI 20.8-58.7). ORs were adjusted for age at the time of the survey, service and rank at the time of the war, serving status at the time of the survey, alcohol consumption, and smoking. The questionnaire for the study focused on reproduction and child health but included open-ended questions about new medical problems or changes in health experienced by the veterans since 1990. However, confidence in the findings is weakened by the retrospective nature of the study and the relatively poor response rate (53% of Gulf War veterans and 42% of non-Gulf War veterans). Dlugosz et al. (1999) examined risk factors for hospitalization for a mental disorder after service in the Gulf War. In active-duty men (n = 1,775,236) and women (n = 209,760) in the U.S. Army, Air Force, Navy, and Marine Corps, the investigators identified 30,539 initial postwar hospitalizations with a principal discharge diagnosis of a mental disorder by using the DMDC hospitalization database grouped into 10 ICD-9 codes. A sample of the military hospital charts was reviewed to assess reliability of the diagnoses in the large database files. Using Cox proportional-hazards regression models, the investigators examined the association between Gulf War deployment and hospitalization for mental disorders. Adjusted incidence risk ratios showed that being deployed in combat or combat support units but not being in the ground war in Iraq or Kuwait in February 1991 was associated with an increased risk of hospitalization because of acute reactions to stress (risk ratio 1.57, 95% CI 1.11-2.22). Veterans who had served in combat troops or combat support troops during the ground war were not at increased risk for hospitalizations for mood, neurotic disorders, personality disorders, adjustment disorders, or acute stress reactions. Moreover, Gulf War veterans were not significantly different from their nondeployed counterparts with regard to psychiatric comorbidities at the time of initial hospitalization. Patterns of comorbid psychiatric disorders in veterans of the Gulf War were also reported by Forman-Hoffman et al. (2005). In Phase I, data were obtained from the Iowa Gulf War Study conducted in 1995-1996 on a stratified random sample of 3695 Gulf War-deployed and elsewhere-deployed military personnel who resided in Iowa and participated in a telephone survey. In Phase II, 374 veterans from Phase I who had symptoms of cognitive dysfunction, depression or chronic widespread pain and 228 veterans who did not have any of these conditions, received in-person assessments in 1999-2002. Assessment included administration of the Anxiety Sensitivity Index for depressive and anxious symptoms, the Mini-Mississippi Index for PTSD, the Barsky Amplification Scale for social support and somatization symptoms, and the SF-36 for personality traits, pain, and health-related quality of life and the SCID-IV. Mental- health comorbidity was based on a diagnosis of at least two current mental disorders from the SCID-IV and independent psychiatrist review in which the psychiatrist was blinded to case and deployment status. Of the 602 surveyed veterans, 32% had a current mental disorder, primarily anxiety disorders (22.4%), depressive disorders (14.2%), and substance-use disorders (5.9%). Over 35% of veterans with a current mental disorder had at least one other co-occurring mental
150 GULF WAR AND HEALTH disorder, mostly depressive and anxiety disorders; and this comorbidity was associated with significant impairment of health-related quality of life in this population of veterans. OâToole et al. (1996a) studied a random sample of 641 Australian Army Vietnam veterans using a battery of self-reporting instruments to assess psychologic status. Vietnam- related PTSD was assessed by the PTSD module of the SCID and the DIS. The most prevalent lifetime psychiatric conditions in Australian Vietnam veterans were alcohol abuse or dependence (42.6%), PTSD (18.7%), somatoform pain disorder (16.5%), social phobias (14.8%), and simple phobias (10.2%); somatoform pain disorder emerged as one of the most prevalent and enduring disorders. The most prevalent 6-month psychiatric conditions were also alcohol abuse or dependence (20.1%), somatoform pain disorder (12.6%), social phobias (11.3%), and simple phobias (7.8%); PTSD was not diagnosed. Furthermore, PTSD, alcohol abuse or dependence, and phobias were all related to combat exposure (a combat self-report scale was contained in the SCID) but not to being posted to a combat unit. The risk of a diagnosis of each of those and other psychiatric disorders increased with increasing combat exposure. When the prevalence of lifetime or current PTSD was compared with responses to a 21-item combat index, there was a linear dose-response relationship with increasing combat exposure. The OR for each combat- score quartile for lifetime PTSD was 1.00, 3.03, 5.36, and 9.18; for current PTSD, the ORs were 1.00, 2.11, 6.97, and 10.33 for each quartile increase in combat exposure. For lifetime alcohol abuse or dependence, the ORs were 1.00, 1.07, 1.56, and 1.86 (p = 0.002); for GAD, the ORs were 1.00, 0.91, 1.81, and 3.14 (p = 0.003); for social phobia, the ORs were 1.00, 1.24, 1.80, and 2.18 (p = 0.012); for panic disorder, the ORs were 1.00, 3.20, 3.27, and 10.10 (p = 0.001); and for dysthemia, the ORs were 1.00, 0.98, 1.67, and 2.58 (p = 0.009). A number of secondary studies were cross-sectional surveys of diverse populations of veterans, including two of veterans returning from deployments to OEF and OIF (Hoge et al. 2004, 2006). In January 2003, Hoge et al. (2004) surveyed 2530 soldiers from one Army infantry brigade for mental-health problems before a year-long deployment to Iraq. In 2003, they also surveyed 1962 soldiers returning from a 6-month deployment in Afghanistan, 894 soldiers returning from an 8-month deployment to Iraq, and 815 Marines returning from a 6-month deployment to Iraq; the second group of soldiers and the Marines had been in the forefront of ground-combat operations in Iraq. PTSD was assessed with the 17-item PTSD Checklist from VA; MDD and GAD were assessed with a questionnaire that included questions about functional impairment. Compared with the prevalence of mental-health problems in the Army group before deployment, MDD was significantly increased in the Army groups after deployment to Afghanistan (OR 1.33, 95% CI 1.03-1.71, p < 0.05) or to Iraq (OR 1.53, 95% CI 1.12-2.08, p < 0.01), but not in the Marines. PTSD was not significantly increased in the soldiers deployed to Afghanistan but was in those deployed to Iraq (OR 2.84, 95% CI 2.17-3.72, p < 0.01) and in the Marines (OR 2.66, 95% CI 2.01-3.51, p < 0.01). GAD was not increased in any of the groups after deployment. This study is limited because different groups of soldiers were surveyed before and after deployment, no Marines were surveyed before deployment, and the mental-health assessments were based on screening instruments. Similar results were seen in a later study by Hoge et al. (2006), who surveyed 303,905 Army soldiers and Marines returning from deployments mainly in Iraq and Afghanistan. All returning veterans were required to complete a brief postdeployment health assessment (PDHA). The PDHA includes two questions for depression modified from the two-item patient health questionnaire and the four-item screen for PTSD developed by the National Center for PTSD, both of which are intended for use in primary care settings; and four questions related to suicide,
HEALTH EFFECTS 151 interpersonal relationships, and interest in receiving care. The PDHA also included three questions about combat experiences, including whether the solider had seen anyone wounded, killed, or dead; had engaged in direct combat and discharged a weapon; or felt in great danger of being killed. The investigators did not assess for prior trauma. Of the veterans returning from Iraq (n = 222,620), Afghanistan (n = 16,318), or deployment elsewhere (Bosnia, Turkey, Uzbekistan, Kosovo, on a ship, or other; n = 64,967), 19.1%, 11.3%, and 8.5%, respectively, met the risk criteria for a mental-health problem. The 8.5% for the deployed-elsewhere soldiers was similar to the prevalence of mental-health problems reported by soldiers prior to their first deployment to OIF or OEF. The OR for OIF veterans compared to those deployed elsewhere was 2.72 (95% CI 2.63-2.80, p < 0.001). Mental-health problems were associated with combat exposure. Almost 10% of the OIF veterans, 4.7% of the OEF veterans, and 2.1% of those deployed elsewhere screened positive for PTSD. Erickson et al. (2001) surveyed a cohort of 2949 Army veterans from Fort Devens, Massachusetts, immediately after their return from the Gulf War (time 1) and then 18-24 months later (time 2) to examine the temporal relationship between depression and PTSD. The Mississippi Scale for Combat-Related PTSD and the Brief Symptom Inventory were administered to all participants, and results were analyzed at time 2. A latent-variable, cross-lag panel model found evidence of a reciprocal relationship between PTSD and depression, which means the best-fit model includes both the progression from PTSD to depression and from depression to PTSD. Essentially, it provides some support for Kesslerâs notion that each outcome could be antecedent to the other (Kessler et al. 1995). The committee notes that the use of latent variables required that an index be developed both for the measurement of PTSD and for the measurement of depression; thus, the results are not tied strictly to diagnosis, but rather to separate symptom clusters. Moreover, after adjustment for military and demographic characteristics, the results suggest that the reciprocal relationship only held up on followup at time 2 for re-experiencing and avoidance-numbing symptoms. Grieger et al. (2006) examined the influence of battle injury on combat-related PTSD or of depression on PTSD in 243 soldiers hospitalized after serious combat injury, including life- threatening or seriously disfiguring injuries (n = 613). The longitudinal cohort was evaluated at 1, 4, and 7 months with the PTSD Checklist; depression was assessed with the Patient Health Questionnaire. Soldiers who did not meet the criteria for PTSD or depression at 1 month but who did at 7 months were compared with soldiers who remained below the diagnostic threshold for PTSD or depression at 7 months. The investigators noted that most of the soldiers with PTSD or depression at 7 months did not meet criteria for either condition at 1 month. Moreover, at 1 month, 4.2% of the soldiers had probable PTSD and 4.4% had depression; at 4 months, 12.2% had PTSD and 8.9% had depression; and at 7 months, 12.0% had PTSD and 9.3% had depression. After controlling for demographic characteristics (age, marital status, and sex), early severity of physical problems was strongly associated with later PTSD or depression. Of those contacted, fewer than 0.5% refused to participate. The investigators note that there were no differences at 1 month in the rates of probable depression, probable PTSD, or either condition between soldiers who completed all three assessments and those who did not complete the followup assessments. Suicide ideation was not included in this section as a psychiatric outcome itself, but, as in depression, there are consistent reports that ideation is greater in deployed veterans than in nondeployed veterans (Fontana and Rosenheck 1995a; Fu et al. 2002; Gray et al. 2002). Suicide ideation is also briefly discussed in the section âSuicide and Accidental Death.â
152 GULF WAR AND HEALTH Summary and Conclusions The committee considered 11 citations on seven primary studies: three on veterans of the Vietnam War and four on veterans of the Gulf War. Three of the publications on Vietnam veterans were based on the NVVRS, one on the VES, and two on the Vietnam Era Twin Registry. Of the publications on Gulf War veterans, two studies analyzed datasets from well- defined cohorts from Louisiana, Massachusetts, and Maine; one used a subset of veterans from the National Health Survey of Gulf War Era Veterans and Their Families; one used a random sample of veterans; and one study looked at Australian Gulf War veterans. All the studies, regardless of the veteran population or the techniques of ascertainment, found that veterans who were deployed to war zones had a greater prevalence of psychiatric disordersâparticularly PTSD, other anxiety disorders, and MDDâthan did veterans who served in the military at the same time but were not deployed to a war zone. PTSD was also found to be highly comorbid with other psychiatric disorders, particularly GAD and MDD. Furthermore, both the prevalence and the severity of those disorders were associated with the level of combat experienced. The 11 secondary studies, most of them of Gulf War veterans, also showed an association between deployment and PTSD, other anxiety disorders, and MDD, as well as other psychiatric disorders. The committee concludes that there is sufficient evidence of an association between deployment to a war zone and the development of psychiatric disorders, including PTSD, other anxiety disorders, and depressive disorders.
TABLE 6-3 Psychiatric Disorders Study Study Design Population Outcomes Results Adjustments Comments CDC 1988a Retrospective 2490 Vietnam- Screening medical- Generalized anxiety OR 1.5, Age at entry into Army, Low participation cohort, theater veterans, health examination and 95% CI 1.1-2.1; depression race, score on of control group VES prevalence, 1972 Vietnam-era psychologic evaluation OR 2.0, 95% CI 1.4-2.9 enlistment general population- veterans with DIS and MMPI technical test, based, telephone enlistment status, year interview with of entry, primary screening military occupational medical specialty examination at followup Kulka et al. 1990 Cross-sectional, 1200 Vietnam- DIS for psychiatric Current combat-related Weighting of data for Response rates NSVG theater men and disorders, self-reported PTSD: 15.2% men, 8.5% all analyses to account over 83% of NVVRS 432 women, 412 levels of exposure to women for interview-level Vietnam-theater Vietnam-era men war-zone stress; Lifetime combat-related nonresponse and veterans, 76% of and 304 women, composite diagnoses of PTSD: 30.6% men, 26.9% different probabilities Vietnam-era 668 nationally 260 theater veterans women of selection in civilian veterans, 70% of representative with MCRPS, SCID, cohort; control for age, civilians civilians; subset of and PTSD Scale of the race or ethnicity, 344 theater MMPI female veteran Because use of veterans also occupation self-reported war- underwent clinical Detailed reviews and zone stress interview with analyses of military exposure mental-health data records to identify information professional potential differences corresponded well between veteran with information in respondents and military records, nonrespondents such as receipt of combat metals, this is one of few early studies that differentiated between types of stress encountered in a war zone 153
TABLE 6-3 Psychiatric Disorders 154 Study Study Design Population Outcomes Results Adjustments Comments Dohrenwend et al. Cross-sectional, 1200 Vietnam Reanalyzed data from Current: 12.2 (9.1% Global Assessment of PTSD prevalence 2006 NSVG theater men; subset NSVG and compared adjusted for both Functioning, rates were of 260 who self-reports of war- impairment and documentation of determined only NVVRS underwent clinical zone stress with documentation) exposure to traumatic for subset of 260 interview military and other Lifetime: 22.5% (18.7% experiences in war theater veterans records; used SCID to adjusted for both zone diagnose PTSD impairment and Good verification documentation) of reported level of war-zone stress exposure with military and other records Jordan et al. 1991 Cross-sectional, 1200 Vietnam- DIS for psychiatric Prevalence rates for any Weighting of data for Response rates NSVG theater men and disorders, self-reported lifetime psychiatric disorder all analyses to account over 83% of NVVRS 432 women, 412 levels of exposure to other than alcohol abuse or for interview-level Vietnam-theater Vietnam-era men war-zone stress dependence in male veterans nonresponse and veterans, 76% of and 304 women, 43% for high war-zone different probabilities Vietnam-era 668 nationally Detailed reviews and stress, 25.6% for moderate of selection in civilian veterans, 70% of representative analyses of military war-zone stress, 21% for cohort; control for age, civilians civilians data records to identify low war-zone stress, and race or ethnicity, potential differences 18.1% for civilians female veteran Because use of between veteran occupation self-reported war- respondents and Prevalence of low-war- zone stress nonrespondents zone-stress veterans vs high- exposure war-zone stress-veteransâ information major depression 3.1% vs corresponded well 11.1%, p < 0.01; dysthymia with information in 1.4% vs 12.3%, p < 0.001; military records, obsessive-compulsive such as receipt of disorder 0.5% vs 5.5%, p < combat metals, this 0.01; GAD 10.8% vs 24.4%, is one of few early p < 0.05; alcohol abuse and studies that dependence 37.2% v 45.6%, differentiated p < 0.05; antisocial between types of personality disorder 7.1% vs stress encountered 16.6%, p < 0.01 in a war zone
TABLE 6-3 Psychiatric Disorders Study Study Design Population Outcomes Results Adjustments Comments Koenen et al. Co-twins control 1874 male CEI, DIS-III-R Major depression Conditional logistic Study eliminates 2003a study monozygotic twin comorbidity: regression; premilitary, genetic and shared pairs; both twins combat adjusted for PTSD trauma history, age at environmental VET Registry served in military OR 1.10, 95% CI 0.92-1.33; entry into military, factors as during Vietnam PTSD adjusted for combat level of education confounders of War OR 2.55, 95% CI 1.16-5.61 comorbidity of combat exposure and combat-related PTSD and other psychiatric disorders Koenen et al. Co-twins control Twins in VET CEI, DIS-III-R Lifetime PTSD vs non- 2003b study Registry; PTSD for: any mood 37 male twin pairs disorder OR 8.89, 90% CI VET Registry with one twin 3.66-21.59; major being a Vietnam depression OR 6.32, 90% CI veteran with PTSD, 2.57-15.55; dysthymia OR 76 male twin pairs 3.80, 90% CI 1.09-13.28; with one twin any anxiety disorder OR being a Vietnam 8.63, 90% CI 2.20-33.83; veteran without panic disorder OR 4.48, PTSD; all co-twins 90% CI 1.04-19.41; served in the ORs also increased for all military but were symptoms comparing PTSD not deployed to with co-twins or PTSD co- Vietnam twins with non-PTSD co- twins Ikin et al. 2004 Cross-sectional 1381 GW-deployed CIDI PTSD OR 3.9, 95% CI 2.3- Service type, rank, age, GW veterans were survey of all veterans, 1377 6.5; major depression OR education, marital younger, more Australian GW- nondeployed 1.6, 95% CI 1.3-2.0 status likely in Navy, and deployed comparison lower-ranked than veterans veterans comparison group 155
TABLE 6-3 Psychiatric Disorders 156 Study Study Design Population Outcomes Results Adjustments Comments Proctor et al. 2001 Cohort study, Fort Devens cohort CAPS used for PTSD, Both gulf-deployed groups Adjustment for Health symptoms part of larger (n = 186) and New SCID used for other reported higher individual oversampling of are all self-reports longitudinal GW Orleans cohort (n = psychiatric disorders symptom prevalence on women, participation cohort study 66) of GW- items that make up body bias, age, sex, deployed veterans, system symptom scores education small comparison compared with Germany- population of deployed group; report of Germany-deployed frequent periods of anxiety veterans from or nervousness OR 7.1 (Fort Maine National Devens) and OR 5.3 (New Guard (n = 48) Orleans); self-reported symptoms of frequent periods of feeling depressed significant only for Fort Devens cohort (OR 6.0); 95% CI excludes 1.0 for all comparisons Wolfe et al. 1999 Cohort study as Fort Devens cohort SCID to examine Rates of current and lifetime Adjustment for over- part of larger (n = 148) and New relationship of PTSD: 5.4% and 6.5% (Fort sampling of females, longitudinal GW Orleans cohort (n = psychiatric disorders Devens), 7.2% and 8.2% participation bias, age, cohort study 56) of GW- with health problems (New Orleans), 0% and 0% sex, education (same cohorts as deployed veterans, (Germany-deployed) used by Proctor small comparison et al. 2001) population of Rates of current and lifetime Germany-deployed MDD: 6.6% and 22.5% veterans from (Fort Devens), 4.5% and Maine National 10.2% (New Orleans), 0% Guard (n = 48) and 4.2% (Germany- deployed) Fiedler et al. 2006 Cross-sectional 967 GW-deployed Telephone interview Deployed vs nondeployed Self-reports 10 years Response rate and 784 era using CIDI based on had significantly (p < 0.05) after conflict 55%, differences in veterans (67% the DSM-III-R greater prevalence in demographics active duty, 15% preceding 12 months: between National Guard, major depression 15.1% vs respondents and 18% reserve); 1765 7.8%, panic attack 1.6% vs nonrespondents
TABLE 6-3 Psychiatric Disorders Study Study Design Population Outcomes Results Adjustments Comments deployed and 1832 0.5%, social phobia 3.6% vs relatively small GW-era veterans 1.7%, obsessive-compulsive disorder 2.8% vs 1.1%, PTSD 3.4% vs 0.9%, GAD 6.0% vs 2.7%, any anxiety disorder 16.0% vs 9.7%, any psychiatric disorder 26.1% vs 16.1% Depression OR 2.07, 95% CI 1.50-2.85 Any anxiety disorder OR 1.81, 95% CI 1.34-2.45 Toomey et al. 2007 Longitudinal, 1061 GW- CIDI used to diagnose 10 years after GW, Age, sex, ethnicity, Participation rate cross-sectional deployed, 1128 axis I mental disorders,persistent increased years of education, duty 53% of deployed NHSGWEVTF GW-era veterans; CAPS used to diagnose prevalence of mental type (active-duty vs veterans, 39% of used subset of PTSD, CES disorders: war-onset current reserve or National nondeployed; (Derived from veterans followed PTSD OR 5.78, 95% CI Guard), service branch, possible Kang et al. 2000b) since 1995 2.62-12.74; all anxiety rank participation bias disorders OR 4.43, 95% CI although authors 2.49-7.88; anxiety disorders report that degree other than PTSD OR 3.79, of bias was 95% CI 1.80-7.99; MDD independent of OR 1.81, 95% CI 1.03-3.32; deployment status pain disorder OR 91.66, 95% CI 10.52-798.21; at least one mental disorder OR 2.12, 95% CI 1.44-3.11 NOTE: CAPS = Clinician-Administered PTSD Scale, CEI = Combat Exposure Index, CI = confidence interval, CIDI = Composite International Diagnostic Interview, DIS = Diagnostic Interview Schedule, DIS-III-R = Diagnostic Interview Schedule for DSM-III, DSM-III = Diagnostic and Statistical Manual for Mental Health Disorders, third edition, GAD = generalized anxiety disorder, GW = Gulf War, MDD = major depressive disorder, NHSGWEVTF = National Health Survey of Gulf War Era Veterans and Their Families, NSVG = National Survey of the Vietnam Generation, NVVRS = National Vietnam Veterans Readjustment Study, OR = odds ratio, PTSD = posttraumatic stress disorder, SCID = Structured Clinical Interview for DSM-III-R, VES = Vietnam Experience Study, VET Registry = Vietnam Era Twin Registry. 157
158 GULF WAR AND HEALTH SUBSTANCE-USE DISORDERS DSM-IV defines substance-use disorders as dependence (characterized by tolerance, withdrawal, needing increasing amounts, persistent desire, and unsuccessful efforts to cut down) or abuse (characterized by recurrent use causing domestic, work, interpersonal, or legal problems, or use in physically hazardous situations) of drugs or alcohol. The lifetime and 12- month prevalences of substance-use disorders in the NCS were about 15% and about 4%, respectively (Kessler et al. 2005a,b). Those rates can be 2-3 times higher in men than in women, depending on the substance. The most reliable method for determining a history of substance-use disorders is the diagnostic interview. In treated populations, current drug use is validated with urine screens. In community and military populations in general, current alcohol problems are often assessed with a screening questionnaire. Two well-validated screening tools that were used in several military studies are the CAGEâa four-item scale to assess cutting down [C], feeling annoyed by people criticizing your drinking [A], feeling guilty about drinking [G], and using alcohol as an eye-opener in the morning [E])âand the Alcohol Use Disorders Identification Test (AUDIT)âa 10-item scale developed by the World Health Organization. A primary study for substance-use disorders is defined as one that had a generalizable sample of deployed and nondeployed veterans, an indicator variable for combat stress, and reliable ascertainment of a substance-use disorder or a problem with alcohol or drugs. Secondary studies failed to fulfill all of those criteria or focused on other populations of interest, such as peacekeepers. Primary studies of substance-use disorders are summarized in Table 6-4. Primary Studies The committee identified six primary studies on the relationship of deployment stress to substance-use disorders: three of veterans of the Vietnam War and three of veterans of the 1991 Gulf War. The two studies with the richest sources of evidence were the VES conducted by CDC (1988a) and the congressionally mandated NVVRS (Jordan et al. 1991). Designs characteristics of those studies were provided in the preceding section on psychiatric disorders; information specifically related to substance-use disorders is discussed below. The VES involved a cohort of 2490 Vietnam-theater veterans and a comparison group of 1972 Vietnam-era veterans, who participated in a face-to-face structured diagnostic interview, the DIS, to assess substance-use disorders in 1985-1986. After adjustment for the six baseline characteristics of age at entry into the Army, race, score on an enlistment general technical test, enlistment status (drafted or volunteer), year of entry into the Army, and primary military occupational specialty (tactical or nontactical), CDC (1988a) found that the current (1-month) prevalence of alcohol disorders was significantly higher in the theater veterans (13.7%) than in the era controls (9.2%) (OR 1.5, 95% CI 1.2-1.8). The rates of drug disorders were not significantly different between the groups (0.4% for theater veterans and 0.5% for era veterans; OR 0.9, 95% CI 0.4-2.0). Analysis of the 1-year prevalence data from the VES, however, did not find significant differences in rates of alcohol or drug-use disorders between the theater veterans (14% and 4%, respectively) and the era veterans (16% and 4%, respectively) (Boscarino 1995). There was no significant relationship between severity of combat exposure and either alcohol or drug-use disorders.
HEALTH EFFECTS 159 Examining lifetime and current (6-month) prevalence of several psychiatric disorders in the NVVRS, Jordan et al. (1991) found significantly higher rates of alcohol and drug abuse or dependence in Vietnam-theater veterans (1200 men and 432 women) than in Vietnam-era veterans (412 men and 304 women) or civilian controls (450 men and 218 women). The most prevalent disorder among male veterans was alcohol abuse or dependence: 39.2% lifetime and 11.2% current for theater veterans and 37.9% lifetime and 9.2% current for era veterans. The prevalence of lifetime and current alcohol abuse or dependence in female theater veterans was 9.1% and 4.9%, respectively, and in female era veterans 4.2% and 1.1%, respectively. The prevalence of lifetime and current drug abuse or dependence was 5.7% and 1.8% in male theater veterans, respectively, and 5.3% and 1.0% in male era veterans. For women, the prevalence of lifetime and current drug abuse or dependence was 1.0% or less for all groups. However when the theater group was dichotomized by level of war-zone stressâhigh (406 men and 170 women) and low or moderate (783 men and 262 women)âthe investigators found that the men and women exposed to high war-zone stress did have significantly higher rates of both lifetime and current alcohol and drug abuse or dependence than the other groups. The 6-month prevalence of alcohol disorders among men was 17.2% in the high group, 8.8% in the low or moderate group, 9.2% in the era controls, and 7.8% in the civilian sample, and 3.4%, 1.6%, 1.0%, and 0.9%, respectively, for women. The lifetime prevalence was 45.6%, 37.2%, 37.9%, and 27.1%, respectively, for men and 10.6%, 8.2%, 4.6%, and 1.5%, respectively, for women. For drug disorders, the current prevalence among men were 3.9%, 1.1%, 1.0%, 0.9%, respectively; the prevalence among female veterans was 0.0% for all groups. The lifetime prevalence of drug disorders was 8.4%, 4.9%, 6.0%, and 3.4%, respectively, for men, and 2.57%, 0.0%, 0.6%, and 0.9%, respectively, for women. Thus, an association of problems with substance use was demonstrated with exposure to combat stress, but not with deployment to a war zone. Goldberg et al. (1990) used the Vietnam Era Twin Registry to assess alcohol-drinking patterns in Vietnam-theater veterans. They analyzed 2169 monozygotic twin pairs. Veterans who served in Vietnam had a higher consumption of alcohol than those who did not serve (17.0% vs 21.4%), and the degree of combat exposure was related to higher daily alcohol consumption. For example, the prevalence of high average daily consumption increased from 15.3% in theater veterans with no combat exposure, to 20.7%, 24.5% and 24.7% in veterans with low, medium, and high combat exposure, respectively (Ï2 trend = 6.625, p = 0.010). The authors noted that they did not control for alcohol consumption before military service because of the potential for recall bias. McLeod et al. (2001) also used data from the Vietnam Era Twin Registry and found that combat exposure was associated with alcohol consumption. Koenen et al. (2003a) also used telephone interview data from the 1993 Vietnam Era Twin Registry study to assess severity of combat exposure, alcohol consumption, and substance- use disorders in 1874 male-male monozygotic twin pairs. As in the NVVRS, severity of combat exposure was significantly associated with alcohol dependence (OR 1.16, 95% CI 1.02-1.31), drug dependence (OR 1.34, 95% CI 1.03-1.48), and cannabis dependence (OR 1.36, 95% CI 1.09-1.71). After adjustment for combat-related PTSD (diagnosed with the DIS-III-R), combat exposure was significantly associated only with alcohol dependence (OR 1.15, 95% CI 1.01- 1.30) and cannabis dependence (OR 1.31, 95% CI 1.04-1.66). In the Gulf War, stress appears to have increased the risk of substance-related conditions. In the large Iowa cohort of regular military, National Guard, and reservists surveyed by telephone in 1995-1996 (1896 deployed and 1799 nondeployed military personnel), the prevalence of symptoms of alcohol abuse as determined with CAGE was 17.0% and 19.4% in
160 GULF WAR AND HEALTH Gulf War-deployed regular military and reserve/National Guard veterans, respectively, and 12.2% and 16.8% in nondeployed regular military and reserve controls, respectively, for a statistically significant prevalence difference of 2.4 (95% CI 0.4-4.5, p â¤ 0.05) (Iowa Persian Gulf Study Group 1997). The response rate for the telephone interview (76%) was relatively high. The CIDI was administered in Australia as part of a two-phase study of 1381 Gulf War veterans and 1377 nondeployed controls (Ikin et al. 2004). Health professionals conducted interviews in clinics throughout the country. The prevalence of alcohol and drug abuse or dependence before the Gulf War was similar in the deployed and nondeployed groups (23.7% and 27.9%, respectively, for alcohol dependence or abuse and 2.8% and 2.3%, respectively, for drug dependence or abuse). The incidence rates for the disorder being first present after the war were significantly higher in deployed veterans than nondeployed veterans for both alcohol disorders (19.8% vs 12.6%, p = 0.001) and drug disorders (3.7% vs 1.8%, p = 0.015) for ORs of 1.5 (95% CI 1.2-2.0) for alcohol and 1.9 (95% CI 1.1-3.2) for drugs adjusted for service type, rank, age, education, and marital status. Significantly more deployed than nondeployed veterans also met criteria for alcoholism in the year before the interview (4.3% vs 2.5%, p = 0.011) for an adjusted OR of 1.8 (95% CI 1.1-2.8) but not the criteria for drug dependence or abuse (0.7% vs 0.6%, p = 0.863, OR 0.8, 95% CI 0.3-2.5). Fiedler et al. (2006) conducted telephone interviews 10 years after the Gulf War with a national sample of 967 U.S. military personnel deployed to the Gulf and 784 nondeployed controls using the 12-month version of the CIDI. They found similar rates of alcohol disorder in the two groups (4.6% in deployed and 3.1% in nondeployed veterans), but the deployed had significantly higher rates of drug dependence (1.2% vs 0.1%, p < 0.05) and âany dependenceâ (5.1% vs 3.2%, p < 0.05) than the nondeployed. Thus, the preponderance of the evidence from the six primary studies suggests that deployed Vietnam and Gulf War veterans, particularly personnel exposed to greater war-zone stress, have higher rates of alcohol or drug disorders than nondeployed controls. Secondary Studies Secondary studies are those with inadequate or no comparison groups, studies of treated populations or nonmilitary populations, and studies that used alcohol or drug measures of unknown reliability and validity. Four studies of Vietnam veterans are in this category. Helzer (1984) interviewed 943 Vietnam veterans in 1972, of whom 605 were eligible for a 2-year followup; at the time of followup, 571 were reinterviewed (94% response rate). The original sample included 470 respondents chosen at random from all enlisted personnel who returned from Vietnam in September 1971 and 495 others identified as illicit drug users on the basis of a urine screening test at the time of departure from Vietnam. The interview schedule was not described in detail, but problem use of alcohol was defined as having regular use of alcohol plus one or two alcohol-abuse symptoms; alcoholism was defined having regular use of alcohol plus three or more alcohol-abuse symptoms or at least one hospitalization because of drinking. At the first interview, 42% of combatants (those who were on combat patrols or dangerous duties, were under enemy fire, or were surrounded by the enemy) were problem drinkers or alcoholics compared with 28% of noncombatants (p < 0.001); a similar pattern was found at followup (45% vs 29%, p < 0.001). In the subsample of 297 veterans with no preservice alcohol problems, 24% of combat veterans and 19% of noncombat veterans were problem drinkers at the first interview, and 31% and 21% at the followup (p < 0.05 for both comparisons). After additional analyses,
HEALTH EFFECTS 161 however, the author concluded that the effects of combat were small and that preservice alcohol and antisocial problems were stronger predictors. Reifman and Windle (1996) conducted a secondary analysis of VES data on illicit drug use on the basis of self-reports from 2490 Vietnam veterans. The VES was conducted in 1985- 1986. They found that combat exposure severity was significantly related to drug use in the preceding year (OR 1.01, 95% CI 1.01-1.02); the relationship remained even after controlling for PTSD. Prigerson et al. (2002) conducted a secondary analysis of data from the 1990-1992 NCS, a national survey of mental health and substance abuse in the general population (Kessler et al. 1995). They found that combat exposure, mostly in the Vietnam war, contributed significantly and directly to 12-month substance abuse with an estimated relative risk of 2.22 (8.0% attributable to combat exposure; 95% CI 1.06-4.15) adjusted for age, race, urbanicity, and low socioeconomic status in the family of origin. An analysis of 641 Australian Vietnam veterans also found that severity of combat exposure based on a combat exposure scale developed for the study was significantly associated (p = 0.007) with current (1-month) alcohol abuse or dependence with ORs of 1.00, 1.21, 1.42, and 2.21 for increasing quartiles of combat exposure; no association with drug abuse or dependence was seen with ORs of 1.00, 2.83, 1.21, 1.74, respectively (OâToole et al. 1996a). Four surveys of Gulf War veterans also included items on alcohol and drug-abuse symptoms. In telephone interviews with 1545 Gulf War-deployed and 435 nondeployed veterans living in Kansas in 1998, Steele (2000) included âalcohol or drug dependenceâ as one of 37 conditions on a checklist. Veterans indicated whether they had been diagnosed or treated by a physician for any condition with new onset during 1990-1998. Prevalence was similar in the deployed (3%) and nondeployed (2%) groups for a nonsignificant OR of 1.47 (95% CI 0.65- 3.31) adjusted for sex, age, income, and education level. In 2002, Jones et al. (2006) mailed questionnaires to 1382 UK armed forces personnel that included three items from AUDIT (how often the person drank in the preceding 12 months, how many drinks are usually consumed when the person drinks, and whether a relative, friend, or doctor was ever concerned or suggested that the person cut down). They found that being deployed to more than one country (as opposed to never or to one country) within the preceding 5 years was significantly related to excessive alcohol intake (OR 2.3, 95% CI 1.5-3.6, p < 0.001). McCauley et al. (2002b) conducted a telephone survey of 653 Gulf War-deployed troops who had been near the Khamisiyah munitions site in Iraq in March 1991, 610 troops deployed elsewhere in the gulf, and 516 troops not deployed to Southwest Asia. The questionnaire focused on medical conditions that had been diagnosed by a physician since the Gulf War. The comparison of deployed with nondeployed veterans was not significant for combined alcohol or substance abuse (OR 1.7, 95% CI 0.9-3.4). Dlugosz et al. (1999) examined risk factors for hospitalization for a mental disorder after service in the Gulf War. In active-duty men (n = 1,775,236) and women (n = 209,760) in the U.S. Army, Air Force, Navy, and Marine Corps, the investigators identified 30,539 initial postwar hospitalizations. Adjusted incidence risk ratios showed that service in the Gulf War in a combat occupation was associated with an increased risk of hospitalization for alcohol-related disorders (risk ratio 1.13, 95% CI 1.04-1.23) although being in a combat support occupation was not (risk ratio 1.00, 95% CI 0.91-1.11). However, being in the war in a support occupation was associated with an increased risk of drug-related hospitalization (risk ratio 1.42, 95% CI 1.03- 1.96), although being in a combat occupation was not (risk ratio 1.16, 95% CI 0.82-1.65).
162 GULF WAR AND HEALTH Moreover, Gulf War veterans were not significantly different from their nondeployed counterparts with regard to psychiatric comorbidities at the time of initial hospitalization. In 2003, Hoge et al. (2004) administered an anonymous survey to Army and Marine troops 1 week before a year-long (Army) or 6-month (Marines) deployment to Iraq or Afghanistan (n = 2530) and 3-4 months after return from combat duty (n = 3671). The survey asked two questions about problems related to the use of alcohol. The rates of alcohol misuse were lower before deployment than after. Specifically, before deployment to Iraq, 17.2% of the soldiers indicated using alcohol more than they meant to, after deployment to Iraq 24.2% did (OR 1.5, 95% CI 1.3-1.9) as did 24.5% after deployment to Afghanistan (OR 1.6, 95% CI 1.4- 1.8). In addition, before deployment to Iraq, 12.5% of the soldiers said that they felt a need to cut down on their drinking, after deployment to Iraq, 20.6% indicated a need to cut down (OR 1.8, 95% CI 1.5-2.2) as did 18.2% after deployment to Afghanistan (OR 1.6, 95% CI 1.3-1.9). Although the before and after findings were not matched by individual, these results are consistent with greater problem drinking being associated with deployment to a war zone. Thus, the findings from the Vietnam War, Gulf War, and OEF and OIF surveys that included screening measures of alcohol or drug problems are somewhat mixed, but the majority of studies point in the direction of increased substance problems as a function of deployment. Posttraumatic Stress Disorder and Substance-Use Disorders It is well established that alcohol use and drug use are comorbid with PTSD and other psychiatric conditions in clinical and nonclinical populations of veterans and nonveterans (Jacobsen et al. 2001; Kessler et al. 1995; Mellman et al. 1992; Sutker et al. 1993a). It has been suggested that the high rates of comorbidity between PTSD and substance-use disorders show that they may be functionally related to each other (Jacobsen et al. 2001). OâToole et al. (1998) found that Australian Vietnam veterans with PTSD were at higher risk for alcohol abuse or dependence (OR 1.6, 95% CI 1.2-2.1) and for drug abuse and dependence (OR 5.4, 95% CI 1.9- 15.5) than era veterans. Koenen et al. (2003a), in a study of Vietnam veterans from the NVVRS, found that combat-related PTSD unadjusted for combat exposure was significant only for drug dependence (OR 2.26, 95% CI 1.05-4.88) but not for alcohol or cannabis dependence (OR 1.39, 95% CI 0.76-2.56 and OR 2.24, 95% CI 0.95-5.31, respectively). An analysis of data from the Fort Devens Gulf War cohort (n = 1006) indicated that alcohol and drug use were significantly associated with the cardinal PTSD symptoms of avoidance, re-experiencing, and hyperarousal (Shipherd et al. 2005). A 1999 study of 1101 Canadian male peacekeepers using AUDIT and the PTSD Checklist-Military Version found significantly more alcohol problems in those with PTSD and a significant trend for AUDIT scores to increase with level of PTSD (none, subthreshold, and full) (Yarvis et al. 2005). Furthermore, Ouimette et al. (1996) found a positive correlation between severity of PTSD symptoms and severity of substance-abuse symptoms in 52 women who served overseas during the Vietnam era. Summary and Conclusions Studies of troops deployed to Vietnam and the Persian Gulf have consistently found higher rates of substance-use problems than in nondeployed controls. Of the six primary studies, three are of Vietnam veterans and three are of Gulf War veterans. Data from the VES and the NVVRS showed that deployment was associated with alcohol use, although only the NVVRS
HEALTH EFFECTS 163 found a significant association of drug abuse with deployment. The Vietnam Twin Registry study also found an association of deployment with alcohol abuse and with drug abuse. Two of the three studies of Gulf War veterans, one in an Iowa population and one in Australian veterans, found a higher prevalence of alcohol-use disorders in deployed veteran; a third conducted 10 years after the war, did not. The two studies that assessed drug-use disorders in Gulf War veterans both found an increased prevalence of such disorders in deployed veterans. Results from the secondary studies were also mixed: five of the seven studies found a positive association between alcohol abuse or dependence and deployment, but two studies did not. For drug-use disorders, the results were similarly mixed: two studies showed a positive association, but three did not. Combat exposure has also been associated with an increased likelihood of substance- related problems. The evidence is reasonably consistent for an association with alcohol abuse/dependence but was considered limited but suggestive for drug abuse for two reasons. First, studies of veterans did not adjust for predeployment substance abuse when analyzing postdeployment prevalences. Second, unlike alcohol, drug use is an illegal behavior, and studies of combat soldiers did not screen for drug abuse. The determinations of drug abuse were based on self-reports, without independent corroboration. Most of the recent studies of Gulf War veterans relied on screening measures or telephone interviews. The studies were cross-sectional, and the findings are based entirely on self-reports. Because substance-use disorders involve externalizing behaviors, corrobation by other sources would have enhanced the validity of information. An important limitation of those studies that might be corrected in the future is the lack of comprehensive, direct assessments of substance disorders that include age-of-onset information. The committee concludes that there is sufficient evidence of an association between deployment to a war zone and alcohol abuse. The committee also concludes that there is limited but suggestive evidence of an association between deployment to a war zone and drug abuse.
164 TABLE 6-4 Substance-Use Disorders Study Study Design Population Outcomes Results Adjustments Comments CDC 1988b Retrospective cohort, 2490 Vietnam-theater DIS-III-R, Alcohol prevalence Age at enlistment, Low participation prevalence, veterans, 1972 preceding month 13.7% vs 9.2%, OR 1.5, race, year of rate in control group; VES population-based, Vietnam-era veterans alcohol and drug 95% CI 1.2-1.8; drugs enlistment, actual confidence telephone interview randomly selected abuse or dependence prevalence 0.4% vs enlistment status interval not given with screening medical from 7924 theater 0.5%, OR 0.9, 95% CI (volunteer vs examination at veterans and 7364 era 0.4-2.0, NS draftee), score on followup veterans who had general technical entered Army in test, primary military 1965-1971 occupational specialty Boscarino 1995 Cross-sectional, used 2490 Vietnam-theater DIS-III-R, Alcohol: prevalence Age at Army entry, Predictors of alcohol subset of VES veterans, 1972 preceding year 16% vs 14% era (NS); enlistment status, abuse: childhood VES Vietnam-era veterans substance abuse or drugs: 4% in both Vietnam volunteer delinquency, illicit randomly selected dependence groups status Army drug use, not from 7924 theater being married, low veterans and 7364 era social support; veterans who had predictors of drug entered Army in abuse: delinquency, 1965-1971 Army drug; combat exposure in Vietnam not associated with alcohol or drug abuse Jordan et al. Cross-sectional, 1200 Vietnam-theater DIS for lifetime and Prevalence of alcohol or Data weighted to Response rates were 1991 NSVG men and 432 women 6-month (current) drug abuse or account for greater than 83% for 412 Vietnam-era men psychiatric dependence for high interview-level Vietnam-theatre NVVRS and 304 women, 688 disorders, self- war-zone stress, low nonresponse and veterans, 76% for nationally reported levels of war-zone stress, era, different Vietnam-era veterans representative exposure to war- and civilians: probabilities of and 70% for civilians zone stress Alcohol (men) lifetime selection in civilian civilians. Because the 46%, 37%, 38%, 27%; cohort; control for use of self reported Detailed reviews 6 months 17%, 9%, 9%, age, race or war zone stress and analyses of 8%; ethnicity, female exposure information military data records Drugs (men) lifetime veteran occupation was found to to identify potential 8%, 5%, 6%, 3%; correspond well with differences between 6 months 4%, 1%, 1%, military records, such
TABLE 6-4 Substance-Use Disorders Study Study Design Population Outcomes Results Adjustments Comments veteran respondents 1%; as receipt of combat and nonrespondents Alcohol (women) metals, this is one of lifetime 11%, 8%, 5%, the few early studies 2%; 6 month 3%, 2%, that differentiated 1%, 1% between the types of Drugs (women) lifetime stress encountered 3%, 0%, 1%, 1%; 6 while in a war zone months all 0% Goldberg et al. Vietnam Era Twin 2169 male 5-level alcohol Average daily alcohol Co-twin alcohol Response rate 74%; 1990 registry; telephone monozygotic twin index; high alcohol consumption greater in drinking measures no control for alcohol survey in early 1987 pairs both serving consumption = ~5 twins who served in SE Demographic and consumption prior to VET Registry active duty military drinks/day; 5-level Asia (21.4%) vs no SE service variables military service during Vietnam War, combat exposure Asia service (17%). tested and found not 829 theater veterans, index Prevalence of heavy to be confounders 1340 era veterans alcohol consumption increased with combat exposure: no 15.3%, low 20.7%, medium 24.5%, and high 24.7% (Ï2 = 6.625) Koenen et al. Co-twins control study 1874 male Combat Exposure Combat exposure Education, age at Temporality not 2003a monozygotic twin Index, associated with alcohol entry to military, considered for pairs in which both DIS-III-R dependence OR 1.16, premilitary trauma, association of VET Registry twins served in 95% CI 1.02-1.31; drug co-twin, PTSD combat-related PTSD military during dependence OR 1.34, with alcohol and Vietnam War 95% CI 1.03-1.48; substance use cannabis dependence outcomes OR 1.36, 95% CI 1.09- 1.71; adjusted for PTSD, combat exposure associated with alcohol dependence OR 1.15, 95% CI 1.01-1.30 and cannabis dependence OR 1.31, 95% CI 1.04- 1.66 165
TABLE 6-4 Substance-Use Disorders 166 Study Study Design Population Outcomes Results Adjustments Comments Iowa Persian Cross-sectional, 1896 GW veterans, CAGE Prevalence of alcohol Age at enlistment, No information on Gulf Study prevalence 1799 nondeployed abuse 17.4% in race, year of CAGE cut-point; Group 1997 veterans deployed vs 12.6% in enlistment, 76% response rate nondeployed (p = 0.02); enlistment status prevalence difference (volunteer vs 2.4, 95% CI 0.4-4.5 draftee), score on general technical test, primary military occupational specialty Ikin et al. 2004 Cross-sectional, 1381 GW-deployed CIDI for Alcohol dependence or Service type, rank, GW veterans were mailed questionnaire Australian veterans, disorders pre-Gulf abuse OR 1.5, 95% CI age, education, younger, more likely followed by health 1377 nondeployed War, post-Gulf War, 1.2-2.0; drug marital status to be in Navy, and assessment at clinic comparison veterans preceding 12 months dependence or abuse lower ranked; OR 1.9, 95% CI 1.1-3.2 response rate was 50% Fiedler et al. Cross-sectional, 967 GW-deployed vs Telephone interview Deployed vs Response rate was 2006 prevalence study, 784 era veterans (67% with CIDI based on nondeployed had 59% deployed and telephone interview, active duty, 15% the DSM-III-R greater 12-month 51% nondeployed; 10 years after conflict National Guard, 18% prevalence of following differences in reserve); random psychiatric disorders in demographics sample from 1765 previous 12 months: between respondents deployed and 1832 alcohol dependence and nonrespondents GW-era veterans 4.6% vs 3.1% (NS); was relatively small; drug dependence 1.2% being deployed to vs 0.1% (p < 0.05) ; any places other than the dependence 5.1% vs Gulf was a risk factor 3.2% (p < 0.05) for alcohol/drug dependence NOTE: CAGE = cutting down, feeling annoyed by people criticizing your drinking, feeling guilty about drinking, and using alcohol as an eye-opener in the morning, CI = confidence interval, CIDI = Composite International Diagnostic Interview, DIS = Diagnostic Interview Schedule, DIS-III-R = Diagnostic Interview Schedule for DSM-III, DSM-III = Diagnostic and Statistical Manual for Mental Health Disorders, third edition, GW = Gulf War, NS = not significant, NSVG = National Survey of the Vietnam Generation, NVVRS = National Vietnam Veterans Readjustment Study, OR = odds ratio, PTSD = posttraumatic stress disorder, VES = Vietnam Experience Study, VET Registry = Vietnam Era Twin Registry.
HEALTH EFFECTS 167 NEUROBEHAVIORAL AND NEUROCOGNITIVE EFFECTS This section focuses on neurobehavioral performance as measured by tests of cognition and in some cases sensory integrity or motor speed and coordination. For the purposes of this section, in addition to the criteria for primary and secondary studies established in Chapter 2, the committee required that primary studies of neurobehavioral effects have used data derived from neurobehavioral tests rather than relying on self-reports of neurobehavioral deficits. Secondary studies had additional methodologic limitations or did not include combat veterans (see Chapter 2). The secondary studies were reviewed and included in the discussion because they evaluated the same functional domains, such as attention and memory, and in some cases used the same neurobehavioral tests as did primary studies; they therefore provide valuable supplementary information that helps to increase or decrease confidence in the conclusions drawn from the primary studies. Confidence in a study is substantially reduced if its statistical analysis did not adjust for confounders or if individually administered neurobehavioral tests were given by examiners not blinded to the status of cases and controls; blinding is of less concern if tests were administered on a computer. Summaries of the primary studies are given in Table 6-5. Primary Studies The committee identified three primary studies that compared deployed veterans with those deployed elsewhere or not deployed. David et al. (2002) compared the neurobehavioral-test performance of 209 UK soldiers deployed to the Persian Gulf, 54 UK Bosnia peacekeeping soldiers, and 78 UK Gulf War-era nondeployed soldiers. Study participants were a random sample of a larger cohort that had responded to an earlier mailed survey about symptoms, illnesses, and exposures (Unwin et al. 1999). A broad array of neurobehavioral tests were administered to all participants. The results of the data analysis were incompletely reported, so evaluation was limited by the lack of standard deviations of the mean test scores. No differences were reported among the groups after correction for age, education, intelligence (according to the National Adult Reading Test), and Beck Depression Inventory score. Proctor et al. (2003) studied 143 Gulf War veterans and 72 nondeployed veterans of the Danish military; participants were randomly selected from among the 916 Gulf War deployed veterans and 236 nondeployed veterans studied by Ishoy et al. (1999). Deployed veterans served in the gulf region during August 1990-December 1997 as peacekeepers and thus had no direct combat exposure. Neuropsychologic tests addressing mood, attention, executive function, motor skills, visuospatial abilities, verbal memory, and visual memory were administered. The self- reports of Danish Gulf War veterans suggested a significantly higher prevalence of eight of 16 neuropsychologic symptoms than their nondeployed counterparts, but no significant differences between the deployed and the nondeployed were found by thorough analyses of the data from the neurobehavioral tests. In the Neurocognition Deployment Health Study, Vasterling et al. (2006) determined the effects of war-zone deployment on neuropsychologic health. A cohort of 654 active-duty Army soldiers were examined before deployment to Iraq in 2003 as part of OIF and on return from Iraq deployment in 2005 and were compared with 307 soldiers who were similar in military characteristics but not deployed overseas. After adjustment for deployment-related head injury, stress, and depression, deployment to Iraq was associated with statistically significant
168 GULF WAR AND HEALTH neuropsychologic effects: lower scores on tests of attention, working memory, executive function, verbal learning, and visual memory retention; more distress from confusion and tension, but higher proficiency in simple reaction time. Secondary Studies Four secondary studies addressed whether Gulf War-deployed veterans differed from nondeployed veterans (Axelrod and Milner 1997; Lindem et al. 2003; Vasterling et al. 2003; White et al. 2001), and a further study examined cognitive function in exercises designed to simulate the stress of combat (Lieberman et al. 2005b). Only Axelrod and Milner (1997) found reliable differences in neurobehavioral test performance between the groups after correction for age and education. Lindem et al. (2003) found that Gulf War-deployed veterans who scored poorly on the Test of Memory Malingering (TOMM), which assesses motivation to perform well, had lower scores on neuropsychologic tests of attention, executive function, and memory. Their study is part of a larger study begun by McEwen (2004) to compare veterans recruited from three cohorts: Fort Devens and New Orleans (n = 58), and 19 Germany-deployed veterans from a Maine National Guard unit. There were no significant differences in mean TOMM scores between the Gulf War-deployed and the Germany-deployed groups. Those who scored lower on the TOMM had a greater prevalence of lifetime PTSD (13.6%) than those with higher scores (1.8%). White et al. (2001) used a larger sample of the McEwen study cohorts from Fort Devens and from New Orleans (total n = 193) and compared them to 47 Germany-deployed veterans. No differences in neuropsychologic test performance were seen between the Gulf War-deployed and Germany-deployed troops, but poorer performance on cognitive tests in Gulf War-deployed veterans was associated with self-reports of exposure to chemical-warfare agents. Similar results were seen in a small study by Vasterling et al. (2003), who found that Gulf War-deployed veterans reported more concerns about cognitive functioning than nondeployed veterans, but those concerns were not confirmed by neurocognitive measures. A study by Lieberman et al. (2005a) found significant (p < 0.001) decrements in cognitive function, vigilance, reaction time, attention, memory, and reasoning during stressful combat-like training compared with baseline prestress performance in 31 U.S. Army officers who had volunteered for an intense training exercise. Posttraumatic Stress Disorder and Neurocognitive and Neurobehavioral Effects The committee identified three analyses of Vietnam veterans drawn from the CDC VES and seven secondary studies that examined the relationship between PTSD and neurocognitive and neurobehavioral outcomes. Barrett et al. (1996) used information from the VES to compare cognitive impairment from PTSD and other psychiatric diagnoses in Vietnam veterans. The 2441 veterans were categorized as to whether they had lifetime PTSD and whether the PTSD was comorbid with a current diagnosis of another psychiatric disorder, using the DIS. Several neurobehavioral testsâthe California Verbal Learning Test (CVLT), the Rey-Osterrieth Complex Figure Drawing Test, the Wisconsin Card Sorting Test (WCST), and the Wechsler Adult Intelligence Scale-Revised (WAIS-R)âwere administered to all participants. Results indicate that PTSD alone was not associated with impairment in cognition; however, veterans with PTSD and current depression, anxiety, or substance use (n = 128) had lower scores on all tests of cognitive functioning than veterans with PTSD alone (n = 236), veterans with only
HEALTH EFFECTS 169 another psychiatric diagnosis (n = 242), and veterans with no psychiatric diagnoses (n = 1835). After adjustment for military and demographic characteristics, results suggest that cognitive deficits might be associated with concomitant diagnoses of PTSD and another psychiatric disorder. Crowell et al. (2002) examined the influence of combat-related PTSD on neurocognitive functioning in a randomly selected subsample of middle-aged Vietnam Army veterans from the VES. Veterans were categorized as having current PTSD (n = 80), having PTSD in the preceding year but without current symptoms (n = 80), psychiatric controls with a DSM-III diagnosis but not PTSD (n = 80), and normal controls (n = 80). PTSD was diagnosed with the DIS-III-A. All veterans completed the WAIS-R, CVLT, Rey-Osterrieth Complex Figure Drawing Test, the Paced Auditory Serial Addition Test (PASAT), the Word List Generation Tasks, the WCST, and the Grooved Pegboard Test. After control for demographic characteristics and comorbid psychiatric conditions, the four groups showed no appreciable differences in cognitive functioning. Zalewski et al. (1994) also used data from the VES to compare neuropsychologic performance in 241 Vietnam veterans with PTSD, those with GAD (n = 241), those with no history of psychiatric illness (n = 241). Comorbid psychiatric disordersâprimarily substance abuse or dependenceâwere present in 80% of the PTSD group and 72% of the GAD group. Cognitive functioning was measured with the WAIS-R block design subtest, the CVLT, the Rey- Osterrieth Complex Figure Drawing Test, and the PASAT. A one-way multivariate analysis of variance revealed no significant differences among the three groups on independent measures of cognitive function. The secondary studies addressed whether veterans with PTSD differed from veterans without a PTSD diagnosis. Findings were inconsistent; when results were positive in the numerous studies described below, the domains that were most often affected pertained to attention and memory. In a small study of 32 Vietnam combat veterans, Gilbertson et al. (2001) found that although veterans with PTSD demonstrated poorer performance on most of the neurocognitive tests, only attention and memory were significant (p = 0.003) predictors of PTSD status in combat veterans. Similarly, Koso and Hansen (2006) found cognitive impairments with large effect sizes pertaining to attention, working memory, executive function, and memory in 20 Bosnian male combat veterans with PTSD, age- and intelligence-matched to veterans without PTSD. Uddo et al. (1993) found that Vietnam veterans with PTSD had memory and attention deficits, and two studies by Vasterling et al. found attention and memory deficits to be associated with PTSD in Gulf War veterans (Vasterling et al. 1998) and in Vietnam veterans (Vasterling et al. 2002) even when they controlled for combat exposure. Yehuda et al. (1995) demonstrated that combat veterans with PTSD have specific deficits in memory related to retroactive interference and a decrement in retention although they have normal abilities in initial attention, immediate memory, and cumulative learning and active inference from previous learning. Finally, Vasterling et al. (2000) sought to determine whether dysfunction of the frontolimbic system of the brain was implicated in PTSD. They compared 51 Vietnam-combat veterans with and without PTSD (n = 26 and n = 25, respectively) with 17 Vietnam-era veterans without a psychiatric disorder. The University of Pennsylvania Smell Identification Test, the Continuous Performance Test, the Auditory-Verbal Learning Test, and the WCST were used to determine cognitive functioning. Olfactory identification was used to assess orbitofrontal
170 GULF WAR AND HEALTH integrity. PTSD-diagnosed veterans had more performance deficits in odor identification and verbal learning, but not on the other cognitive tests, than veterans without PTSD; this suggested frontolimbic dysfunction in PTSD (see Chapter 5 for further discussion of the neurobiology of PTSD). Summary and Conclusions Three primary studies of the association of deployment to a war zone, whether OIF or the Persian Gulf, with neurobehavioral performance had mixed results. David et al. (2002) and Proctor et al. (2003) found only slight differences in cognitive functioning in UK troops deployed to the Gulf War and in Danish peacekeepers deployed to the Gulf after the conflict ended, respectively. Vasterling et al. (2006) found significant neurocognitive and neurobehavioral impairments in veterans deployed to OIF. In conclusion, primary studies of deployed veterans and veterans not deployed to a war zone do not clearly demonstrate differences in cognitive and motor measures as determined with neurobehavioral testing. Results of secondary studies also are mixed. Primary studies specifically of PTSD and neurocognitive performance found that PTSD alone was not associated with deficits on tests of cognition. However, the studies do not appear to have included specific tests for memory, such as explicit memory. Secondary studies of PTSD were not uniform in their findings; however, when test results were positive, the most common findings were deficits in attention and memory. The committee concludes that there is inadequate/insufficient evidence to determine whether an association exists between deployment to a war zone and neurocognitive and neurobehavioral effects.
TABLE 6-5 Neurobehavioral and Neurocognitive Effects Reference Study Design Population Outcomes Results Adjustments Comments David et al. Case-control, 209 male UK GW- WAIS-R scaled scores: Only difference between ANCOVA adjusted Careful treatment of 2002 clinical deployed, 54 vocabulary, groups was lower score in for education, age, potential evaluations Bosnia-deployed, 78 digit span, arithmetic, Purdue pegboard for GW- NART, BDI; confounders, such as era nondeployed similarities, picture deployed group multiple depression, mood, soldiers randomly arrangement, block design, comparison intelligence, selected from larger object assembly, digit adjustment for least education cohort of UK symbol, PASAT, significant veterans who sustained attention to difference participated in response task, Stroop procedure and earlier mailed survey Trailmaking A and B Bonferroni (see Unwin et al. 1999) WMS: logical memory (immediate and delayed recall), verbal paired associates (immediate and delayed recall), Camden recognition memory test, Purdue pegboard; individually administered tests, blinded examiners Proctor et al. Cross-sectional 143 male Danish WAIS-R Information No overall differences in MANCOVA by Response rate 75% 2003 GW-deployed subscale, continuous neuropsychologic neuropsychologic veterans, 72 male performance test, trail- domains, domain, adjusted nondeployed troops making, significant test differences for age randomly selected WCST, Purdue pegboard, in domains (p â¤ 0.05) for from 84% and 58% WAIS-R block design, CVLT and WCST of total Danish CVLT, armed forces WMS visual reproductions, deployed and TOMM; individually nondeployed, administered tests except in respectively, at time computer-based NES; of GW blinded examiners 171
TABLE 6-5 Neurobehavioral and Neurocognitive Effects 172 Reference Study Design Population Outcomes Results Adjustments Comments Vasterling et Cohort-controlled, 961 male and female MOS-CF, POMS, trail- Deployment associated Multiple linear Response rate 94% al. 2006 prospective active-duty Army making Parts B and A, with compromise on regression analyses predeployment, 75% soldiers: NES3, CPT, WMS3, Verbal sustained attention (Ã = adjusted for postdeployment 654 Iraq-deployed, Paired Associates I and II, 0.11, p < 0.001), verbal battalion 307 nondeployed WMS Visual Reproductions learning (Ã = -1.51, p = membership, I and II, ANAM simple 0.003), visual spatial deployment-related reaction time, code memory (Ã = head injury, stress, substitution and learning, -3.82, p < 0.001); depression code substitution and delay, deployment associated matching to sample, logical with negative state affect relations, mathematical on measures of confusion processing, running (Ã = 1.40, p < 0.001), memory, tapping (right and tension (Ã = 1.24, p < left) 0.001); deployment associated with improved simple reaction time (Ã = 4.30, p = 0.003) Barrett et al. Cross-sectional, 2,441 Vietnam CVLT, Rey-Osterrieth Veterans with PTSD and ANOVA 1996 neuropsychologic veterans: 236 with complex figure test, concurrent psychiatric examination PTSD, 242 with WCST, WAIS-R disorder demonstrated (Derived other psychiatric (information and block impairment in cognitive from VES) disorders, 128 with design); PTSD diagnosed functioning compared PTSD and another with DIS with other groups psychiatric diagnosis, 1835 with no psychiatric diagnosis Crowell et Cross-sectional Vietnam veterans DIS-III-A and MMPI used No statistically significant ANOVA, Participants include al. 2002 with current PTSD to determine presence of differences among the MANOVA, only male Vietnam- (n = 80), PTSD in combat stressors and four groups on matched with era Army veterans (Derived preceding year but diagnosis of PTSD neuropsychologic respect to age, exposed to combat- from VES) no active symptoms measures education, race, related stressors (n = 80), WAIS-R (information and premilitary psychiatrically block design), measures of
TABLE 6-5 Neurobehavioral and Neurocognitive Effects Reference Study Design Population Outcomes Results Adjustments Comments matched control CVLT, Rey-Osterreith cognitive group without PTSD Complex Figure, functioning, income (n = 80), normal PASAT, Word List level control group (n = Generation Task, 80) WCST, Grooved Pegboard Test Zalewski et Cross-sectional 723 Vietnam WAIS-R block design, No differences in MANOVA Data collected in al. 1994 veterans: 241 with CVLT, cognitive functioning 1985-1986, high PTSD, 241 with Rey-Osterreith Complex among groups levels of comorbid (Derived GAD, 241 with no Figure, PASAT psychiatric disorders from VES) history of in groups with PTSD psychiatric diagnosis and GAD NOTE: ANAM = Automated Neuropsychological Assessment Matrices, ANCOVA = analysis of covariance, ANOVA = analysis of variance, BDI = Beck Depression Inventory, CPT = Continuous Performance Test, CVLT = California Verbal Learning Test, DIS = Diagnostic Interview Schedule, GAD = generalized anxiety disorder, GW = Gulf War, MANOVA = multivariate analysis of variance, MDD = major depressive disorder, MMPI = Minnesota Multiphasic Personality Inventory, MOS-CF = Medical Outcomes Study Cognitive Functioning Scale, NART = National Adult Reading Test, NES = Neurobehavioral Evaluation System, PASAT = Paced Auditory Serial Addition Test, PTSD = posttraumatic stress disorder, R = revised, POMS = Profile of Mood States, TOMM = Test of Memory Malingering, VES = Vietnam Experience Study, WAIS = Wechsler Adult Intelligence Scale, WCST = Wisconsin Card Sorting Test, WMS = Wechsler Memory Scale. 173
174 GULF WAR AND HEALTH CHRONIC FATIGUE SYNDROME Many veterans returning from the Gulf War and other wars have reported experiencing chronic fatigue (McCauley et al. 2002a). Unexplained chronic fatigue, experienced by both the general public and veterans, has been the subject of much discussion by clinicians and researchers alike (Buskila 2000; Straus 1991), but its etiology and course are still unclear. The prevalence of chronic fatigue in the general adult population appears to be less than 2% (Buskila 2000). CDC developed a case definition of chronic fatigue syndrome (CFS), first published in 1988 and revised in 1994 (Box 6-1). It is characterized by the presence of severe fatigue with related functional impairment and the occurrence of at least four of eight other defining symptoms over at least 6 months (Freeman et al. 2005; Fukuda et al. 1994). The most commonly reported symptoms are headache, postexertional malaise, impaired cognition, and muscle pain (Wills et al. 2003). McCauley et al. (2002a) found that 103 of 799 veterans deployed to the Gulf War in 1990-1991 fulfilled the case definition in 1998. The committee notes that fatigue, but not symptoms of fatigue sufficient to meet the CDC definition, is one of the most widely reported symptoms in surveys of Gulf War veterans (Cherry et al. 2001a; Engel et al. 2000; Gray et al. 1999; Ishoy et al. 1999; Kang et al. 2000b; Kelsall et al. 2004a; Simmons et al. 2004; Steele 2000; Unwin et al. 1999). As in fibromyalgia and chronic pain (discussed later in this chapter), no laboratory tests or pathologic physical signs are widely accepted or provide a definitive diagnosis. The CDC criteria require that three elements be completed as part of a comprehensive evaluation. The first element, determining whether the symptom criteria for CFS are present, requires that a person be queried specifically about length and severity of fatigue and about eight ancillary symptoms. The second, determining whether other medical conditions are present, mandates a complete physical examination, a battery of specified laboratory tests, and a medical history. The third is an assessment of exclusionary conditions (such as lupus, mononucleosis, depression, and multiple sclerosis) (Fukuda et al. 1994; McEwen 2002). BOX 6-1 Case Definition of Chronic Fatigue Syndrome CDC criteria for CFS require the presence of both the following: Clinically evaluated, unexplained persistent or relapsing chronic fatigue that is of new or definite onset (that is, not lifelong), is not the result of ongoing exertion, is not substantially alleviated by rest, and results in substantial reduction in previous levels of occupational, educational, social, or personal activities. The concurrent occurrence of four or more of the following symptoms: substantial impairment in short- term memory or concentration; sore throat; tender lymph nodes; muscle pain; multi-joint pain without swelling or redness; headaches of a new type, pattern, or severity; unrefreshing sleep; and post-exertional malaise lasting more than 24 hours. These symptoms must have persisted or recurred during 6 or more consecutive months of illness and must not have predated the fatigue. SOURCE: CDC (2007a).
HEALTH EFFECTS 175 A primary study for CFS requires that it be diagnosed by a health professional. A secondary study is one in which a CFS-like condition has been documented and a comparison is made between deployed and nondeployed veteran populations. Self-reports of CFS and self- reports of a physician diagnosis of CFS were included in the secondary studies. The primary studies of CFS are summarized in Table 6-6. Primary Studies The only primary study identified by the committee is that of Eisen et al. (2005), a cross- sectional prevalence study of 12 health measures in 1061 Gulf War-deployed and 1128 nondeployed veterans conducted in 2001 as part of the National Health Survey of Gulf War Era Veterans and Their Families. All randomly selected study participants had participated in the 1995 phase of the survey by completing a mail or telephone questionnaire about their health. CFS was diagnosed by clinical examination on the basis of the International Chronic Fatigue Syndrome Study Group case definition (Fukuda et al. 1994) by VA clinicians who were blind to the deployment status of the veterans. Veterans with psychiatric disorders were excluded from the sample. Eisen et al. (2005) found that only three of the 38 deployed veterans who self- reported CFS met the criteria on examination and only two of the eight nondeployed veterans who self-reported CFS received this diagnosis, so the authors concluded that self-reports of CFS in both deployed and nondeployed veterans are unreliable. Clinically diagnosed CFS, however, had the largest OR of the 12 medical illnesses or symptoms addressed. CFS was more prevalent in deployed veterans (1.6%) than in nondeployed veterans (0.1%) giving an OR of 40.6 (95% CI 10.2-161.15, p < 0.001) after adjustment for age, sex, race, cigarette-smoking, duty type, service branch, and rank. The strengths of this large study are its population-based design, stratified sampling method, analysis of participation bias, comprehensive examination, and use of computer-based algorithms by researchers who were blinded to deployment status. One limitation is the low response rates: 53% of the eligible deployed veterans and 39% of the eligible nondeployed veterans. Secondary Studies The committee identified six secondary studies that explored the relationship between deployment to a war zone and CFS. In all six, a diagnosis of CFS was self-reported, CFS was determined on the basis of self-reports of symptoms similar to those of CFS, or the presence of CFS was based on criteria other than those of CDC. Gray et al. (2002) found an OR of 7.60 (95% CI 4.76-12.13) for self-reports of physician-diagnosed CFS in Gulf War-deployed Seabees vs nondeployed Seabees; the OR was adjusted for age, sex, active-duty or reserve status, race or ethnicity, current smoking, and current alcohol-drinking. The prevalence of CFS was 5.17% in the Gulf War-deployed Seabees (n = 3831) who reported such a diagnosis, 0.79% in Seabees deployed elsewhere (n = 4933), and 0.68% in nondeployed Seabees (n = 3104). In a survey of UK military personnel deployed to the Gulf War or to Bosnia or on active duty but not deployed, Reid et al. (2001) found that the prevalence of CFS was not statistically different between the Gulf War-deployed and nondeployed troops (2.1% vs 1.8%), but both groups had a greater prevalence of CFS than did the group deployed to Bosnia (0.7%). CFS was determined by the researchers on the basis of responses to a fatigue questionnaire combined with the SF-36 questionnaire for functional disability to meet the CDC criteria for CFS. The OR for CFS in the gulf vs Bosnia groups was 2.3 (95% CI 1.2-4.3) and in the gulf vs era veterans 1.2
176 GULF WAR AND HEALTH (95% CI 0.8-1.8) adjusted for sex, age, marital status, education, rank, and employment status. CFS was associated with reported exposures to combat-related injury (OR 4.1, 95% CI 2.2-7.7), explosions of Scuds (OR 2.6, 95% CI 1.5-4.6) or artillery (OR 2.4, 95% CI 1.4-4.1), hearing chemical alarms (OR 2.5, 95% CI 1.2-5.3), witnessing a personâs death (OR 2.2, 95% CI 1.3- 3.8), seeing maimed soldiers (OR 2.0, 95% CI 1.2-3.6), and burning rubbish or feces (OR 2.0, 95% CI 1.0-3.4). In a telephone survey of Iowa veterans of the Gulf War, a greater prevalence of chronic fatigue symptoms was reported by regular military (1.0%, n = 985) and National Guard/reserve (2.9%, n = 911) veterans than by nondeployed regular (0.2%, n = 968) and National Guard/reserve veterans (1.1%, n = 831) (Iowa Persian Gulf Study Group 1997). The statistically significant increased prevalence rate difference between deployed and nondeployed veterans was 1.4 (95% CI 0.9-2.0) after adjustment for age, sex, race, branch of military, and rank. The presence of a CFS-like condition was based on a combination of symptoms given in the CDC criteria (Freeman et al. 2005) and scores on the Chalder fatigue scale, a widely used scale to measure physical and mental fatigue in CFS patients (Putnam et al. 2006). The overall response rate was high (76%). Canada deployed about 4500 troops to the gulf region in 1991 to provide naval operations and medical support. In 1997, a survey of the health status of the entire Canadian Gulf War forces was undertaken, comparing Gulf War-deployed troops with Canadian forces deployed elsewhere at the time (Goss Gilroy Inc. 1998). The assessment was based on a mailed survey that used the same questions as were used by the Iowa Persian Gulf Study Group (1998). Deployed veterans were more than 5 times as likely as nondeployed veterans to report symptoms of CFS (OR 5.27, 95% CI 3.95-7.03). Veterans who had served with land units had a slightly higher risk of CFS than those who served in air or sea units (OR 1.56, 95% CI 1.15-2.12). The response rate of the Gulf War veterans was 73%. A cross-sectional mailed survey by Unwin et al. (1999) of three cohorts of UK veteransâ3284 deployed to the gulf, 1815 deployed to Bosnia, and 2408 nondeployedâfound that self-reported CFS (or myalgic encephalitis) was more strongly associated with deployment to the gulf than with deployment to Bosnia (OR 2.1, 95% CI 1.1-4.0) or no deployment (OR 2.7, 95% CI 1.6-4.6). ORs were adjusted for age, smoking, alcohol consumption, marital status, education, rank, employment, military or civilian status on followup, and a general health questionnaire. The prevalence of CFS (or myalgic encephalitis) was low in all the cohortsâ3.3% in the gulf, 0.8% in the Bosnia, and 0.8% in the era cohortâalthough reports of symptoms of fatigue were relatively common: 50.7%, 26.3%, and 27.7%, respectively. Fatigue was associated with belief of exposure to chemical attack. Zhang et al. (1999) reported that Gulf War veterans with CFS have several immune- function abnormalities in lymphocyte subpopulations not seen in healthy veterans or in civilians with CFS. Posttraumatic Stress Disorder and Chronic Fatigue Syndrome Only one secondary study examined the relationship between CFS and PTSD in Gulf War veterans. Kang et al. (2003) sent a self-administered 48-item questionnaire in 1995 to a large sample of active-duty, reserve, or National Guard veterans deployed during the Gulf War (n = 11,441) or not deployed (n = 9476). Current CFS-like illness was assessed according to responses to questions based on the CDC criteria for CFS (Fukuda et al. 1994); veterans had to have at least four of eight symptoms for 6 months or longer. The risk of current PTSD was
HEALTH EFFECTS 177 estimated on the basis of responses to the 17-item PTSD Checklist that corresponded to the symptoms in the DSM-III-R. No clinical examinations were conducted for either CFS or PTSD. Compared with nondeployed veterans, Gulf War veterans had a greater risk for both PTSD (12.1% vs 4.3%) and CFS (5.6% vs 1.2%). The OR for current PTSD in deployed vs nondeployed veterans was 3.1 (95% CI 2.7-3.4) and 4.8 (95% CI 3.9-5.9) for CFS; ORs were adjusted for age, marital status, rank, and unit component and PTSD was also adjusted for sex. The authors also attempted to determine the effects of deployment stressors on the risk of PTSD and CFS in the reserve and National Guard units that were activated but not deployed and were deployed to the gulf. Although PTSD increased with intensity of stress, a risk of a CFS-like illness did not show a similar relationship and increased from 0.8% of members who were not activated to 1.7% of members who were activated and deployed but not to the gulf. The risk of having CFS was higher in members deployed to the gulf than deployed elsewhere but did not vary significantly with increasing combat stress, ranging from 5.4% to 7.3%. This study did not identify members who had comorbid PTSD and CFS, nor did it attempt to determine how soon after deployment the symptoms of each disorder were apparent. The study had the advantage of a population-based cohort with a relatively high participation rate of 70%, but it is limited by the lack of a physical examination for CFS or a psychiatric interview for PTSD. Summary and Conclusions Because the diagnosis of CFS depends entirely on symptoms, not on physical or laboratory findings, its prevalence varies widely from study to study. The only primary study demonstrated a higher prevalence of CFS in Gulf War-deployed veterans than in nondeployed veterans, although the absolute difference in prevalence was very small (1.6% vs 0.1%). Of the five secondary studies comparing Gulf War-deployed veterans with nondeployed veterans or veterans deployed elsewhere, three showed a higher prevalence of self-reports of CFS, CFS symptoms, or CFS-like illnesses; the other two studies did not see a difference in the prevalence of CFS between the two groups. Of the three secondary studies using the CDC definition of CFS, two had a positive association, and one no association. In addition, some of the secondary studies reviewed were not limited to CFS but included fatigue or CFS-like illnesses. The committee concludes that there is limited but suggestive evidence of an association between deployment to a war zone and chronic fatigue syndrome.
TABLE 6-6 Chronic Fatigue Syndrome 178 Study Study Design Population Outcomes Results Adjustments Comments Eisen et al. 2005 Population-based, 1061 GW-deployed vs CFS based on in- OR 40.6, 95% CI Age, sex, race, Low participation cross-sectional, 1128 nondeployed person interviews 10.2-161.15 cigarette-smoking,rates (53% of NHSGWEVTF prevalence, veterans, veterans selected according to CDC duty type, servicedeployed and 39% of in-person medical from among those who CFS criteria and branch, rank nondeployed), but (Derived from and psychiatric had participated in 1995 exclusionary analysis of Kang et al. 2000b) evaluations mail and telephone survey diagnoses from nonparticipants and history, interviews, participants reveals examinations, that participants, both laboratory testing deployed and nondeployed, are more likely to report symptoms of CFS NOTE: CDC = Centers for Disease Control and Prevention, CFS = chronic fatigue syndrome, CI = confidence interval, GW = Gulf War, NHSGWEVTF = National Health Survey of Gulf War Era Veterans and Their Families, OR = odds ratio.
HEALTH EFFECTS 179 SLEEP DISTURBANCES Complaints of disturbed sleep are common symptoms of many disorders and have been frequently reported by veterans of the Gulf War (Barrett et al. 2002b) and the Vietnam War (Neylan et al. 1998). Sleep-related problems affect 50-70 million Americans of all ages (National Heart Lung and Blood Institute 2007). Disturbances of sleep are generally assessed in two ways. Most commonly, people are simply asked to report on characteristics of their sleep, such as how long they sleep, whether their sleep is disturbed by nightmares, and whether their sleep is restful. Sleep disturbances and sleep disorders, such as sleep apnea, can also be assessed objectively in a laboratory, usually by recording the electroencephalogram and determining how much time is spent in one of the defined sleep stages and how much time is spent awake. Such laboratory assessments are time-consuming and expensive and therefore are not often used. This section considers the types of sleep disturbances reported in large surveys of deployed and nondeployed Gulf War veterans. No studies that included objective measures of sleep disturbance or a physicianâs diagnosis of a sleep problem or disorder in veterans of the Gulf War or any other war, except for those with PTSD as discussed below, were identified. None of the studies reviewed in this section met the committeeâs criteria for a primary study, so no summary table is included. Self-Reports of Sleep Disturbance The committee identified several studies that assessed self-reported information on sleep from deployed and nondeployed veterans of the 1991 Gulf War (Cherry et al. 2001a; Engel et al. 2000; Gray et al. 2002; Ishoy et al. 1999; Kang et al. 2000b; Kelsall et al. 2004a; Proctor et al. 1998; Steele 2000; Unwin et al. 1999). However, none of them supplemented the self-reported data with systematically collected information from a clinical interview or other objective measures. Several researchers have found increased reports of disturbed sleep among U.S. veterans deployed to the Gulf War. In 1995, the VA used a mailed questionnaire to survey 11,441 U.S. Gulf War-deployed veterans and 9476 nondeployed controls; Kang et al. (2000b) found a higher frequency of sleep problems in the deployed veterans, with several sleep symptoms being significantly more prevalent among the Gulf War veterans than the nondeployed veterans: unrefreshing sleep (47% vs 24%, rate difference 23, 95% CI 22.8-23.2), difficulty in sleeping (37% vs 21%, rate difference 16, 95% CI 15.9-16.1), and excessive sleepiness (30% vs 14%, rate difference 16, 95% CI 15.9-16.1). Using data from a self-report questionnaire, Proctor et al. (1998) compared the sleep problems of 252 Gulf War-deployed veterans with those of 48 era veterans who had been deployed to Germany; about 30% of the Gulf War veterans and 11% of the comparison group reported an inability to fall asleep (OR 3.4-3.6, 95% CI excludes 1.0). Trouble in sleeping was reported by 14.6% of 527 Gulf War-deployed U.S. Navy Seabees and 4.9% of their nondeployed counterparts for an OR of 3.4 (95% CI 2.3-5.0) 3-4 years after the war (Gray et al. 1999). In a study of Navy Seabees, Gray et al. (2002) found that the 3832 Seabees deployed to the Persian Gulf had more nightmares and flashbacks (OR 4.58, 95% CI 3.50-6.00) and more trouble in sleeping (OR 3.08, 95% CI 2.71-3.50) than 3104 nondeployed Seabees. In 1998, Steele (2000) assessed symptoms suggestive of Gulf War illness in 1435 deployed and 409
180 GULF WAR AND HEALTH nondeployed veterans, all from Kansas. Deployed veterans reported not feeling rested after sleep and problems in falling or staying asleep significantly more often than nondeployed veterans (OR 2.69, 95% CI 2.04-3.54 and OR 2.98, 95% CI 2.18-4.08, respectively). A similar pattern was observed among Gulf War veterans from other countries. Unwin et al. (1999) conducted a cross-sectional mailed survey of a random sample of UK veterans: 3284 Gulf War-deployed, 1815 deployed to the Bosnia conflict, and 2408 era veterans. The Gulf War veterans reported a higher frequency of sleeping difficulties than the Bosnia cohort (OR 1.7, 95% CI 1.5-2.0) or the era cohort (OR 1.9, 95% CI 1.7-2.2) adjusted for age, smoking, alcohol consumption, marital status, educational attainment, rank, employment status, civilian or military status on followup, and score on a general health questionnaire. More UK veterans deployed to the Gulf War than nondeployed veterans reported waking up feeling tired and worn out, losing sleep because of worry, and having nightmares (Cherry et al. 2001a). Australian Gulf War veterans also reported more sleep problems than their nondeployed counterparts (Kelsall et al. 2004a). The most frequently reported symptom among the deployed veterans was feeling unrefreshed after sleep (OR 1.7, 95% CI 1.4-2.1). More deployed veterans also reported sleep difficulties (OR 1.8, 95% CI 1.5-2.2). ORs were adjusted for service type, rank, age, education, and marital status. Ishoy et al. (1999) compared the frequency of self-reported sleep difficulties in Danish veterans who had been deployed to the Persian Gulf during 1990-1997 as peacekeepers and age- and sex-matched controls. They found a higher frequency of sleep problems in the deployed subjects. For example, 19.4% of the deployed and 9.1% of the nondeployed controls reported problems in falling asleep. Posttraumatic Stress Disorder and Sleep Because difficulty in falling asleep or staying asleep and nightmares are two of the diagnostic criteria for PTSD, it is difficult to distinguish between PTSD and comorbid psychiatric disorders, such as MDD and GAD, many of which are also characterized by sleep disturbances. Nevertheless, several studies have looked at sleep disturbances in veterans with and without PTSD. Using questionnaire data from the NVVRS, Neylan et al. (1998) obtained self- reported information on five items in the Mississippi Scale for Combat-Related PTSD that assess sleep. Three domains of sleep were addressed: difficulties in sleep onset, nightmares, and sleep maintenance disturbance. Subjects were asked about the frequency with which the sleep problems occurred. Vietnam-theater veterans who met the case definition of PTSD at the time of the survey reported more disturbances in all three domains. For example, difficulties in falling asleep at least sometimes were reported by 44.0% of combat veterans with PTSD, 5.5% of combat veterans without PTSD, 9.4% of era veterans, and 5.0% of civilian comparison subjects (p < 0.0001). Difficulties in staying asleep occurred in 90.7% of combat veterans with PTSD, 62.5% of combat veterans without PTSD, and 63.1% of era veterans (p < 0.0001). Nightmares were reported by 52.4% of the combat veterans with PTSD, 4.8% of those without PTSD, and 5.7% of era veterans. Frequent or very frequent nightmares and difficulties in falling asleep were reported only by subjects with PTSD. Using hierarchic multiple regression, the investigators found that 48% of the variance in frequency of nightmares was accounted for by non-sleep- related PTSD symptoms. Combat exposure accounted for an additional 9% of the variance for nightmares. The weighted Pearson correlation of combat exposure and nightmares was 0.63 (p < 0.001). Limitations of this study include its retrospective assessment of combat exposure and that sleep measures were subjective and not standardized.
HEALTH EFFECTS 181 Engel et al. (2000) assessed physical symptoms in 21,244 Gulf War veterans who enrolled themselves in DoDâs Comprehensive Clinical Evaluation Program (CCEP) in 1994- 1996. Of the total sample, 23.2% of those with no psychologic condition reported having sleep disturbance compared with 64.5% of those with PTSD and 45.3% of those with a psychologic condition but not PTSD. Only 8.7% of those in the CCEP who considered themselves healthy reported sleep disturbance compared with 34% of those with a medical illness. Inman et al. (1990) compared 35 Vietnam veterans with PTSD and 37 veterans with insomnia but without PTSD. The researchers found no differences in the severity of the insomnia between the two groups, but the veterans with PTSD had more sleep-related anxiety symptoms, such as talking or shouting during sleep, fear of going to sleep, having disturbing thoughts while lying in bed, waking from a frightening dream, and finding it hard to return to sleep. Several small studies have used objective sleep measures to assess sleep quality in veterans with PTSD. In 1991-1994, Engdahl et al. (2000) studied 59 combat veterans of World War II and the Korean War (including 30 prisoners of war), 30 of whom had current PTSD (as diagnosed with the SCID). There was no difference between veterans with and without PTSD in any of 18 sleep measures gathered over three nights by polysomnography, except for a small increase in rapid-eye-movement (REM) sleep and fewer arousals per minute from non-REM sleep in the PTSD veterans; the amount of REM-sleep correlated with PTSD severity. Other studies identified sleep abnormalities in the electrophysiologically monitored sleep of small samples of veterans with PTSD and other psychiatric disorders. Mellman et al. (1995a) found recurrent awakening, increased movement during sleep, and threatening dreams to be more prevalent in 20 Vietnam veterans with combat-related PTSD than in eight noncombat veterans without PTSD; these sleep abnormalities may be associated with nondiminished central noradrenergic activity at night (Mellman et al. 1995b). Dow et al. (1996) compared Vietnam veterans who had PTSD and depression with those who had depression alone; the only significant difference was prolonged sleep latency in the depressed group. Vietnam veterans with combat-related PTSD had less REM sleep and less sleep efficiency than a veteran group with depression; this suggests that sleep maintenance is impaired in PTSD (Mellman et al. 1997; Woodward et al. 2000; see also Chapter 5). Dagan et al. (1991) found that although veterans with PTSD complained of sleep disturbances, objective measures of sleep obtained using an actigraph to differentiate between sleep and waking were not impaired in those with PTSD. In a small study of 12 veterans with war-related PTSD and 12 controls without PTSD, there were no significant differences in polysomnographic recording between the two groups; however, those with PTSD had significantly higher auditory awaking thresholds and more hostile and aggressive dreams (Lavie et al. 1998). Increased sleep problems in veterans with combat neuroses or PTSD have been reported in veterans of other conflicts (Lavie et al. 1979). One study of over 4 million veterans in the Veterans Health Administration health-care database estimated that 2.9% of the veterans had sleep apnea and that almost 12% of those also had PTSD (Sharafkhaneh et al. 2005). Nevertheless, as concluded by reviews that were not limited to consideration of PTSD only in veterans (Caldwell and Redeker 2005; Harvey et al. 2003), objective findings regarding sleep disturbance even among people with PTSD are inconsistent. Harvey et al. (2003) reviewed over 25 studies of sleep in combat veterans, primarily from the Vietnam War, and other groups with PTSD and found mixed results. Some of the studies showed that those with PTSD had more disturbed sleep and lower sleep efficiency, but other studies showed that those with PTSD are
182 GULF WAR AND HEALTH not outside the normal range of objective sleep measures. Harvey et al. (2003) suggest that people with PTSD may perceive their sleep as more disturbed than it actually is. Summary and Conclusions There is consistent evidence that veterans deployed to war zones have a higher frequency of complaints of disturbed sleep than do comparable control groups; however, objective measures of sleep disturbances or diagnoses of sleep disorders, such as narcolepsy and sleep apnea, are lacking in veteran studies. None of the studies comparing deployed with nondeployed veterans of the Gulf War or other wars included objective measures of sleep problems. Because the studies relied solely on self-reports of sleep problems, the committee concludes that there is inadequate evidence to determine whether veterans deployed to a war zone are at increased risk for sleep disturbances. More research has been conducted on sleep problems in veterans with war-related PTSD. One study examined the relationship between complaints of sleep disturbances and an indicator of exposure to deployment-related stressors. In an examination of data from the NVVRS, Neylan et al. (1998) found a relationship between PTSD and complaints of sleep disturbance, especially nightmares, even after adjusting for the presence of symptoms of sleep disturbance in the diagnostic criteria for PTSD. Those data are consistent in suggesting that deployment to a war zone is associated with increased complaints of disturbed sleep. The committee notes that PTSD is highly comorbid with other psychiatric disorders, such as MDD, that are also characterized by sleep problems. The comorbidity adds to the complexity of assessing sleep disturbance in veterans and others with PTSD. There are several problems in drawing conclusions about the more general question of the role of deployment-related stress in producing sleep disturbances. The studies of self-reports of sleep disturbances cited in this section did not include information about the relationship between deployment-related stressors and the development of complaints of disturbed sleep, nor did they address the degree to which the sleep disturbances might be part of a syndrome, such as depression or PTSD. An additional limitation is the low response rates that characterize such studies and increase the likelihood of bias in the results. Objective sleep measures have not been studied in sufficiently large veteran populations to permit generalization about the association between deployment to a war zone and sleep disturbance or disorders or between PTSD and sleep disturbances. The committee concludes that there is inadequate/insufficient evidence to determine whether an association exists between deployment to a war zone and sleep disorders or objective measures of sleep disturbances.
HEALTH EFFECTS 183 CARDIOVASCULAR DISEASES The term cardiovascular disease encompasses a wide variety of conditions, the most important of which are related to the development of atherosclerosis in the arteries and high blood pressure; these can lead to coronary heart disease (CHD)âwhich may be manifested clinically as myocardial infarction (MI), angina, or sudden cardiac deathâand to cerebrovascular disease, which may present clinically as a stroke or transient ischemic attack. The three most important risk factors for cardiovascular disease are blood pressure, blood cholesterol, and smoking. All are affected by stress and lifestyle and so could be affected by deployment, particularly to a war zone (in the case of hypertension). Hypertension can be regarded both as a type of cardiovascular disease and as a risk factor for CHD and stroke. This section considers those medical conditions and also considered lesser manifestations of cardiovascular disease, including chest pain and arrhythmia, both of which may be manifestations of CHD but also occur commonly in the absence of any structural disease. Thus, chest pain that occurs in patients without CHD (that is, people who have normal coronary angiograms) tends to occur in younger people who have psychiatric conditions, such as anxiety and depression, which can occur in veterans deployed to a war zone. That war is stressful and may result in symptoms characteristic of heart disease is documented in the medical literature as Da Costa syndrome after the U.S. Civil War and as soldierâs heart in World War I (Jones et al. 2002). The scientific literature on the cardiovascular effects of exposure to wartime stress can be divided into three general categories: first, studies of the acute effects of war, including hospitalization for or death from myocardial infarction in the civilian population; second, studies of the delayed effects, including symptoms and objective findings of cardiovascular disease that persist or emerge after the war is over; and third, studies of changes in acute physiologic reactivity during simulated sounds or sights of war in veterans who have developed PTSD (see Chapter 5). The following sections review the epidemiologic information available on the cardiovascular effects seen in deployed and nondeployed veterans. No studies reported on cardiovascular effects in terms of quantified or categorized levels of stress experienced by veterans deployed to a war zone, so deployment itself is used as a surrogate for deployment- related stress, as explained in Chapter 1. For cardiovascular effects, primary studies were ones that met the committeeâs criteria, as given in Chapter 2; however, self-reports of cardiovascular disease and of hypertension were considered to be primary reports if a self-report was part of a thorough interview and specified that the veteran was reporting a previous physician diagnosis or treatment for the condition. Self-reports of cardiovascular disease based on a mail questionnaire or other instrument, such as the SF-36, were considered to be secondary studies if they were otherwise well conducted. Primary studies are summarized in Table 6-7. As in previous sections, the committee has also looked at the effects of the Gulf War, the Vietnam War, and World War II. PTSD is a well-described outcome of exposure to war and one of PTSDâs components is heightened autonomic arousal (for example increased blood pressure), so we also examine the cardiovascular consequences of PTSD. Cardiovascular Symptoms A number of studies have examined the array of self-reported cardiovascular symptoms and clinical disease in veterans, primarily after the Gulf War. Gray et al. (2002) administered
184 GULF WAR AND HEALTH mailed questionnaires to 3831 Gulf War veterans (Navy Seabees) and 3104 nondeployed Seabees 7 years after the war. The deployed veterans self-reported a significantly higher prevalence of all 33 medical problems, including chest pains, than the nondeployed veterans (OR 3.06, 95% CI 2.52-3.71) in the 12 months before the survey. Proctor et al. (1998) surveyed two stratified random samples of Gulf War veterans (220 from Fort Devens and 71 from New Orleans) and a control group of 50 Gulf War-era veterans who were deployed in Germany from a larger cohort of active-duty, reserve, and National Guard members. Participants completed the Health Symptom Checklist in 1992-1993 and an expanded version in 1994-1996. The authors found nonsignificantly higher prevalences of a wide variety of self-reported symptomsâ including chest pain, irregular heart beats, and racing heartâin the Gulf War veterans than in the nondeployed veterans. Another study by McCauley et al. (2002b) consisted of a telephone survey of 1779 Gulf War veterans in three groups: those who were possibly exposed to chemical-warfare agents from the destruction of the Khamisiyah munitions bunker, those who were deployed to Iraq but not exposed, and those who were not sent to Iraq. The deployed veterans reported more physician diagnoses of high blood pressure (OR 1.7, 95% CI 1.3-2.4) and heart disease (OR 2.5, 95% CI 1.1-6.6), but there was no evidence that those who had possibly been exposed to chemical-warfare agents had a higher prevalence of any cardiovascular events. Stretch et al. (1995) conducted a survey of Gulf War veterans from Hawaii and Pennsylvania 2-3 years after the war and found that significantly (p < 0.05) more of the 1524 deployed veterans than the 2524 nondeployed veterans (6.7% vs 2.4%) reported âheart problemsâ among other symptoms. UK veterans who had been deployed to both Bosnia and the Gulf War (n = 570) complained of more chest pain (25.2%) and rapid heartbeat (15.1%), as 1785 veterans who had not been deployed to either war (13.2% and 8.0%, respectively) or 2049 veterans who had served only in Bosnia (12.4% and 7.5%, respectively) (Hotopf et al. 2003a). The ORs for deployed to both conflicts vs deployed to neither conflict were 2.2 (95% CI 1.7-2.8) and 2.2 (95% CI 1.6-2.9) for chest pain and rapid heartbeat, respectively. Thus, there are consistent findings that deployment to a war zone, particularly the Gulf War, is associated with an increase in self-reports of many physical symptoms, including chest pain and increased heart rate, but these symptoms do not necessarily imply any structural heart disease. Hypertension Primary Studies In epidemiologic studies, hypertension is generally defined on the basis of measuring blood pressure (ideally more than once) or patients reporting that they have received a diagnosis of hypertension and are taking antihypertensive medication. Two primary studies compared deployed with nondeployed veterans for hypertension. In a cross-sectional prevalence study, Eisen et al. (2005) conducted medical examinations of 1061 deployed and 1128 nondeployed Gulf War veterans 10 years after the conflict. Veterans who participated in the study had been part of the larger National Health Survey of Gulf War Era Veterans and Their Families conducted in 1995. Hypertension was defined as a blood pressure greater than 140/90 mm Hg or a history of hypertension and taking antihypertensive medication. The difference in the prevalence of hypertension on clinical evaluation at a VA medical facility between deployed (9.1%) and nondeployed veterans (12.6%) was not significant (OR 0.90, 95% CI 0.60-1.33). ORs were adjusted for differences in age, sex, race, years of education, cigarette-smoking, duty type, service branch, and rank. The study was limited by low participation rates (53% of eligible
HEALTH EFFECTS 185 deployed and 39% of eligible nondeployed) and the number of years that had passed since the war. Blanchard et al. (2006) re-examined the same data as those of Eisen et al. to determine the prevalence of chronic multisymptom illness and came to a similar conclusion: hypertension was not higher in Gulf War-deployed veterans than in their nondeployed counterparts, whether or not the veterans had multisymptom illness. The VES, conducted by CDC in 1985-1986, looked at cardiorespiratory conditions in 2490 Vietnam-theater Army veterans and 1972 Vietnam-era veterans. In Phase 1 conducted about 15-20 years after the war, a telephone interview was used to assess whether the veterans had any one of a number of health effects; in Phase 2, participants received physical examinations to screen for health status. A diagnosis of hypertension was based on a measured blood pressure above 140/90 mmHg or by patients reporting that they were taking antihypertensive medication. The prevalence of hypertension was not significantly higher in the theater veterans (33.5%) than in the era veterans (31.4%) (OR 1.1, 95% CI 0.9-1.2) after adjustment for age at enlistment, race, year of enlistment, enlistment status, score on a general technical test, and primary military occupation. About 5% of the veterans in each group reported using antihypertensive medication at the time of the study (CDC 1988b). Secondary Studies Secondary studies are defined as those in which the diagnosis of hypertension was based only on self-reports. A cross-sectional study of 1456 Australian Gulf War veterans and 1588 nondeployed veterans, surveyed 10 years after the war, found no increase in the prevalence of self-reports of physician diagnoses or of treatment of hypertension after the war in the deployed veterans (OR 1.2, 95% CI 0.9-1.6) after adjustment for service type, rank, age, education, and marital status (Kelsall et al. 2004a). The strength of this study is that a physician assessed and rated the likelihood of each self-reported medical condition during a followup face-to-face interview although a physical examination was not conducted. Several other cross-sectional studies relied exclusively on self-reports. The largest, conducted by VA (Kang et al. 2000b), was a mailed survey with followup telephone interviews of a population-based sample of 11,441 Gulf War veterans and a stratified random sample of 9476 nondeployed veterans. The study was the first phase of the National Health Survey of Gulf War Era Veterans and Their Families conducted in 1995, which surveyed 30,000 deployed and nondeployed Gulf War veterans for health problems, and was used by Eisen et al. (2005). A slightly higher (but significant because of the large numbers) prevalence of hypertension was reported in the Gulf War veterans than in the nondeployed veterans (11.4% vs 7.6%, rate difference 3.84, 95% CI 3.75-3.93) in the preceding 12 months. In a cross-sectional study by Hotopf et al. (2003a), a stratified random sample of 2049 UK veterans of the Bosnia peacekeeping mission, 570 veterans who had served in both the Gulf War and Bosnia, and 1785 nondeployed veterans completed a mail questionnaire that asked about a variety of health effects. No difference in the frequency of self-reported hypertension was seen between the groups (6.1% in both the deployed and the nondeployed and 4.2% in those deployed only to Bosnia). Three other studies reported a statistically significant increase in self-reported hypertension. The Seabee Health Study (Gray et al. 2002), a survey of 3831 Gulf War-deployed Navy Seabees and 3104 nondeployed Seabees in which questionnaires were mailed to the participants 7 years after the Gulf War (1997-1999), found a significantly higher rate of self- reported hypertension in the Seabees who had been deployed than in those who had not (OR
186 GULF WAR AND HEALTH 1.82, 95% CI 1.48-2.26) after adjustment for age, sex, active-duty or reserve status, race or ethnicity, current smoking, and current alcohol-drinking. McCauley et al. (2002b) compared the self-reported rates of hypertension in 1263 veterans who were deployed to Iraq (653 within 50 km of Khamisiyah and 610 who were deployed to Southwest Asia but outside a 50-km radius) with rates in 516 nondeployed veterans. Eight years after the war, the incidence of hypertension was higher in the deployed veterans (OR 1.7, 95% CI 1.3-2.4). Finally, in a small study of 141 Gulf War-deployed veterans and 46 veterans deployed to Germany evaluated 4 years after the war, Proctor et al. (2001) found more hypertension in the deployed veterans (13.9% vs 4.4%, p < 0.05). Haley et al. (2004) used 24-hour blood-pressure monitoring, the gold standard for defining hypertension, to determine whether there were abnormalities of the autonomic nervous system that might explain the chronic symptoms experienced by many Gulf War veterans. They found no difference in blood pressure between 22 deployed and 18 nondeployed male Gulf War veterans. The deployed veterans had become ill during or shortly after deployment. Although the study used an appropriate approach to determine hypertension, it is limited by the small number of subjects and the inclusion of veterans who were reported to be ill. Heart Disease Primary Studies The most important form of heart disease that might be related to deployment stress is CHD. It is, however, relatively unlikely to be clinically manifested before the age of 45-50 years; thus, even if deployment did accelerate the development of CHD, it would be difficult to detect such an effect in Gulf War veterans, although not in Vietnam or World War II veterans. Because the average age of military personnel who were deployed in the Gulf War was around 28 years (Eisen et al. 2005), it is not surprising that no studies have reported CHD effects. For the present analysis, primary studies are defined as those in which the diagnoses of CHD were made by physical examination or CHD events (myocardial infarction and cardiac death) were based on hospitalizations or death certificates. A survey by Kang and Bullman (2001) performed 7 years after deployment found that cardiovascular mortality in male Gulf War veterans was lower (1.30 per 10,000 person-years) than in non-Gulf War veterans (2.05 per 10,000 person-years). The adjusted rate ratio derived from a Cox proportional-hazards model (controlled for age, race, branch of service, unit component, and marital status) was 0.90 for men (95% CI 0.81-1.01) and 0.96 for women (95% CI 0.55-1.69). The vital status of 621,902 Gulf War veterans and 746,248 other veterans was determined from VA and Social Security Administration databases; death certificates were used to establish cause of death. Both veteran groups had significantly lower mortality from cardiovascular disease than the general population. Two studies by Gray et al. (1996, 2000) looked at hospitalization records of Gulf War era veterans for discharge diagnoses of the ICD-9-CM category circulatory system diseases. In the first study, the researchers determined DoD hospital-discharge diagnoses for 1991, 1992, and 1993, comparing 547,076 Gulf War veterans with 618,335 other veterans from the same period. The multivariate ORs for each year were all 0.9-1.1 (exact values not given) adjusted for prewar hospitalizations, sex, age, race or ethnic group, branch of service, marital status, rank, length of service, salary, and occupation. In the second study, they examined hospital-discharge records for 1991-1994 for three hospital systems: 182,164 records from the DoD, 16,030 records from
HEALTH EFFECTS 187 the VA, and 5,185 records from the California Office of Statewide Health Planning and Development. PMRs of hospital-discharge diagnoses of Gulf War veterans and veterans not deployed to the gulf were compared. The PMRs were 0.94 (95% CI 0.91-0.98) for DoD, 0.85 (95% CI 0.76-0.93) for VA, and 0.98 (95% CI 0.82-1.14) for California Office of Statewide Health Planning and Development. All PMRs were adjusted for age and sex, and the DoD PMR was also adjusted for race. In the CDC VES study, discussed above, the researchers conducted physical examinations for cardiorespiratory conditions (CDC 1988b). Specifically, the 2490 Vietnam- theater and 1972 Vietnam-era veterans were evaluated for problems of the peripheral arterial system with a Doppler instrument, electrocardiograms that were read by cardiologists, and chest x-ray pictures. No significant differences were found between the theater and era veterans in the prevalences of altered peripheral arterial hemodynamics (4.7% vs 3.6%, OR 1.2, 95% CI 0.9- 1.7), ischemia (1.9% vs 1.8%, OR 1.1, 95% CI 0.7-1.7), or any electrocardiographic findings (14.3% vs 13.9%, OR 1.1, 95% CI 0.9-1.3), which included bradycardia, tachycardia, extrasystoles, and nonspecific ST- and T-wave changes. A slight increase in risk of left ventricular hypertrophy (1.6% vs 1.0%, OR 1.8, 95% CI 1.0-3.3) was seen in theater veterans. The prevalence of cardiac findings in the chest x-ray pictures did not differ significantly between theater and era veterans (1.0% vs 0.9%, OR 1.3, 95% CI 0.7-2.5). All ORs were adjusted for age at enlistment, race, year of enlistment, enlistment status (volunteer vs draftee), score on a general technical test, and primary military occupational specialty. As part of the VES, postservice mortality in a cohort of 9324 U.S. Army Vietnam veterans was compared with that in 8989 Army Vietnam-era veterans who served in Korea, Germany, or the United States (CDC 1987, 1988b). Veterans were assessed from discharge through 1983, and vital status was ascertained from files of the Army, VA, Social Security Administration, the Internal Revenue Service, and the National Center for Health Statistics. Death certificates were used to establish cause of death, which was categorized by an experienced nosologist who was blind as to veteran deployment status. During the roughly 12-18 years of followup, the crude mortality from diseases of the circulatory system was 9.4 per 100,000 person-years in Vietnam veterans and 19.0 per 100,000 person-years in era veterans, for a rate ratio of 0.49 (95% CI 0.25-0.99). The study results indicate that Vietnam veterans had a 51% lower mortality from cardiovascular diseases than era veterans; the difference was evident over the entire followup period and for all types of circulatory diseases. Secondary Studies Studies that relied on self-reports of heart disease or that combined âhardâ measures (for example, myocardial infarction) and âsoftâ measures (for example, angina) of heart disease were classified as secondary studies. A survey of 11,441 Gulf War veterans and 9476 nondeployed veterans found no difference in the rates of self-reported CHD or stroke (Kang et al. 2000b). Only one study reported any increase in heart disease. McCauley et al. (2002b) conducted telephone interviews with 653 veterans who were deployed near Khamisiyah in Iraq and compared them with 610 who were deployed to Southwest Asia but not near Khamisiyah and 516 who were not deployed. The interviews took place 8 years after deployment. The deployed veterans were more likely to report having a postwar diagnosis of heart disease (OR 2.5, 95% CI 1.1-6.6) although the disease was not specified. This studyâs positive findings based self-reports may merely reflect increased reporting of all medical conditions or more nonspecific symptoms among Gulf War veterans.
188 GULF WAR AND HEALTH Blood Lipid Concentrations The Danish Gulf War Study was conducted on 686 of the 821 Danish troops deployed to the Persian Gulf region as UN peacekeepers during 1990-1997 and 231 age- and sex-matched control veterans. Health examinations included blood tests of the veterans 7 years after deployment (Ishoy et al. 1999). Total cholesterol, high-density lipoprotein (HDL), and triglycerides were the same in the two groups. Other studies have examined lipid concentrations as a function of PTSD. Karlovic et al. (2004) compared blood lipids in 53 Croatian War veterans with PTSD and 49 with combat experience but no PTSD and found that those with PTSD had significantly higher total cholesterol (264 vs 226 mg/dL, p = 0.001), low-density lipoprotein (LDL) (169 vs 137 mg/dL, p = 0.002), and triglycerides (196 vs 138 mg/dL, p = 0.001) and had lower HDL (43 vs 62 mg/dL, p < 0.001). There was no difference in BMI between the two groups. Kagan et al. (1999) compared lipids in 73 Vietnam veterans with PTSD and 113 male volunteers admitted into a substance-abuse program who were matched for demographic factors, such as age. The lipid concentrations were compared with those of the general male population in the National Health and Nutrition Examination Survey (NHANES) and with averages of male veterans. Total cholesterol, LDL, and triglycerides were highest in the veterans with PTSD, and HDL was marginally lower. Solter et al. (2002) compared blood lipids in 103 Croatian veterans with combat-related PTSD and a control group of 92 patients with MDD. Veterans with combat-related PTSD had higher mean concentrations of cholesterol (6.2 vs 5.3 mmol/L, p < 0.001), LDL cholesterol (3.9 vs 3.5 mmol/L, p = 0.005), and triglycerides (2.9 vs 1.5 mmol/L, p < 0.001) and had lower HDL cholesterol (1.0 vs 1.3 mmol/L, p < 0.001) than the control group. Posttraumatic Stress Disorder and Cardiovascular Effects As discussed in Chapter 5, PTSD is one of the best-established health effects resulting from exposure to traumatic events during deployment to a war zone. It also has profound effects on the autonomic nervous system and other systems that mediate the development of cardiovascular disease, so it is important to examine the state of knowledge of these relationships. Excessive autonomic nervous system arousal in response to trauma-related cues is one of the diagnostic features of PTSD. Two general types of study are relevant here: those of the long-term effects of PTSD on cardiovascular variables and those of the psychophysiologic effects of simulated combat in subjects with and without PTSD. Heart Rate A number of studies have performed laboratory testing of veterans with and without PTSD. They have generally involved the measurement of heart rate, blood pressure, and other cardiovascular variables while the subjects were at rest and then compared the changes that occurred during mental-challenge tests. There have been many studies of a generally similar design, and the committee notes that in several of them the subjects with PTSD showed an increase in resting heart rate, which led to the suggestion that PTSD might be a risk factor for hypertension. Buckley and Kaloupek (2001) performed a meta-analysis of 34 studies that gathered indicators of basal cardiovascular activity, including heart rate, systolic blood pressure, and diastolic blood pressure of subjects with diagnosed PTSD and two types of comparison groups: subjects who had been exposed to trauma but did not have PTSD and subjects with no
HEALTH EFFECTS 189 history of exposure. Twenty-five of the studies included veterans whose original traumatic stress was combat-related; the time between the traumatic event and the measurement of cardiovascular activity ranged from 2 months to 29 years. In total, cardiovascular measures of 2670 subjects were analyzed in all the studies. The main finding was that people with a current PTSD diagnosis had a resting heart rate that was 5 beats/minute faster than subjects in the control groups. It has also been observed that an increase in heart rate immediately after a trauma is a predictor of PTSD (Shalev et al. 1998). Hypertension Findings that people with PTSD have an increased heart rate are fairly consistent, but results related to blood pressure have been less clear. Some of the epidemiologic studies described above that relied mostly on self-reports for the identification of hypertension also compared its prevalence in veterans with and without PTSD. However, one study that used a more thorough method to assess the prevalence of hypertension in veteran populations with possible PTSD found no increase in hypertension in aging veterans with PTSD. The VA Normative Aging Study was established in 1961 to follow 2280 community-dwelling men in the greater Boston area; more than 90% of the men were veterans at entry in the study. The cohort included 1002 veterans who completed a mailed Mississippi Scale for Combat-Related PTSD in 1990 and 944 veterans who completed the Keane PTSD scale of the Minnesota Multiphasic Personality Inventory in 1986 (Kubzansky et al. 2007). Men with pre-existing angina pectoris, a history of MI, or diabetes were excluded from the study; the mean age of the 1990 study population was 63 years. Study participants receive physical examinations every 3-5 years. A correlation analysis of scores from the Mississippi Scale for Combat-Related PTSD with systolic blood pressure showed no relationship; however, there was a slight correlation (r = -0.06, p = 0.04) between lower diastolic blood pressure and the score on the Mississippi Scale for Combat- Related PTSD but not on the Minnesota Multiphasic Personality Inventory-2. Veterans with higher PTSD scores were at slightly greater risk for total CHD (RR 1.21, 95% CI 0.93-1.57, adjusted for age, smoking, blood pressures, serum total cholesterol, BMI, family history of CHD, education, and alcohol intake), particularly nonfatal MI (RR 1.30, 95% CI 0.92-1.84, adjusted for coronary risk factors). The associations were slightly strengthened when depression was also controlled for. Several of the studies of deployed veterans described above observed a marked increase of PTSD as a result of deployment without any increase in the prevalence of hypertension (Kelsall et al. 2004a; Kubzansky et al. 2007; McCauley et al. 2002b; Schnurr et al. 2000; Spiro et al. 2006). A study of 147 Dutch resistance fighters in World War II (82 of whom had PTSD) found that the prevalence of hypertension in those with PTSD was not higher than in those without PTSD (32% vs 31%) (Falger et al. 1992). The authors noted that 56% of the veterans, 60-65 years old at the time of the study, were currently suffering from PTSD diagnosed with the SCID according to DSM-III-R criteria. This study is limited by the lack of confirmation by physical examination of the cardiovascular factors reported by participants. A study of Croatian veterans of the Balkan wars (Karlovic et al. 2004) included 43 with PTSD, 37 with PTSD and comorbid MDD, 38 with MDD alone, and 39 healthy controls. Veterans with PTSD (with or without MDD) showed no differences from the other veterans in blood pressure. One nationally representative study that reported an association between PTSD and hypertension in a civilian population is the NCS (Lauterbach et al. 2005). It compared 429 subjects with PTSD with 5448 subjects without PTSD. Hypertension was reported twice as
190 GULF WAR AND HEALTH frequently (14.2%) in those with PTSD as in those without (7.7%). There was no control for other predictors of hypertension, however, such as obesity. The meta-analysis of Buckley and Kaloupek (2001), discussed in the section above on heart rate, did not find any consistent differences in blood pressure between people with and without PTSD. Although blood pressure was measured much less often as an end point than was heart rate, some of the individual studies found a slight increase of hypertension in people with PTSD. Interpretation of the results is difficult because subjects who had been treated for hypertension were generally excluded, and it is not clear to what extent their hypertension reflected sustained differences, as opposed to an enhanced anticipatory response. A study by Hughes et al. (2006) that was published after the meta-analysis by Buckley and Kaloupek measured heart rate and blood pressure with a beat-to-beat monitor in 80 subjects with PTSD (about one-fourth of whom were combat veterans) and 50 healthy control subjects. They found no difference in resting heart rate or blood pressure between the groups. One of the most informative methods for studying psychosocial influences on cardiovascular function is 24-hour ambulatory blood-pressure and heart-rate monitoring, which was performed in several small studies. In a study of 11 Vietnam veterans with PTSD and seven without, Muraoka et al. (1998) found that the PTSD group had a significantly higher heart rate (by 9 beats/minute) both when they were awake and when they were asleep and a slightly higher systolic pressure while they were awake, but the difference was not significant. In a study of 117 Vietnam veterans, of whom 61 had PTSD and 56 did not, ambulatory monitoring of heart rate and blood pressure was carried out for 12 hours during the daytime (Beckham et al. 2000, 2003). The blood pressure was the same in the two groups, but there was slightly greater variability in the PTSD group. There was a nonsignificant increase in heart rate (78 vs 75 beats/minute) in the PTSD group. The variability of both blood pressure and heart rate was slightly but significantly higher in the PTSD group. However, there was evidence of an interaction between PTSD and smoking in that smokers who had PTSD had the highest blood pressure (Beckham et al. 2004). Haley et al. (2004) performed 24-hour monitoring of heart rate, polysomnography, and blood pressure in 22 Gulf War-deployed veterans and 18 nondeployed and found that heart rate declined less during the night in the deployed veterans independent of sleep measures. The heart- rate differences were attributed to differences in vagal control. The nocturnal fall in blood pressure was the same in the two groups. A more direct way to evaluate resting sympathetic nervous activity is to measure the kinetics of norepinephrine with a radioisotope-dilution technique. In a small study of 12 Vietnam veterans with PTSD and six healthy control subjects, that method failed to show any evidence of increased norepinephrine production (Murburg et al. 1995). It has been reported that 24-hour excretion of urinary catecholamines is increased in patients with PTSD (Yehuda et al. 1992). Plasma catecholamines measured over 24 hours may also be high in PTSD (Yehuda et al. 1998). The combination of all those findings constitutes fairly strong evidence that resting heart rate is slightly increased in patients with PTSD. In contrast, there is no clear indication that PTSD leads to hypertension. Such changes are of potential significance for the long-term development of heart disease because there is extensive evidence that increased heart rate is a risk factor for both hypertension and cardiovascular events (Palatini and Julius 1997). Cardiovascular Reactivity Several studies have established that there is sympathetic hyperreactivity in Vietnam veterans with PTSD as manifested by increases in heart rate, blood pressure, and plasma
HEALTH EFFECTS 191 catecholamines in response to simulated combat (Blanchard et al. 1991; Lindauer et al. 2006; Orr et al. 1998, 2003; Pallmeyer et al. 1986). Beckham et al. (2002) investigated the cardiovascular responses to a relived anger task in 118 male Vietnam combat veteransâ62 with PTSD and 56 without itâwho were outpatients at a VA medical facility. Current PTSD was diagnosed with the SCID for axis I disorders and the Mississippi Scale for PTSD. Veterans with PTSD reported more combat exposure as measured with the Combat Exposure Scale and had lower socioeconomic status than veterans without PTSD. Participants completed standardized diagnostic measures, hostility measures, and a laboratory session in which they relived a self- chosen anger memory while their heart rate and blood pressure were measured continuously with an Ohmeda Finapres blood-pressure monitor. Baseline diastolic and systolic blood pressure and heart rates did not differ between the two groups; however, veterans with PTSD took less time to feel anger, had greater mean heart rate and greater blood-pressure response during relived anger, and reported greater anger and anxiety during the task. The increased cardiovascular reactivity observed in those studies may be specific to simulated combat. Orr et al. (1998) also investigated the effects of three stressors on resting heart rate and systolic and diastolic blood pressure in 20 Vietnam veterans with and 15 without combat-related PTSD as diagnosed by the SCID according to criteria of the DSM-III-R (all participants were free of drug and alcohol use). Stressors were mental arithmetic and immersion of a hand in icy water for 1 minute. The only difference between the two groups in cardiovascular response was increased diastolic blood pressure in veterans with PTSD, but not those without PTSD; this suggested a reduced autonomic response in PTSD. The findings on mental arithmetic were confirmed by Pallmeyer et al. (1986), who studied heart rate and diastolic and systolic blood pressure in five groups: Vietnam veterans with PTSD (n = 12), Vietnam veterans without PTSD but with a comparable level of combat exposure (n = 10), Vietnam veterans with other psychiatric disorders (n = 5), era veterans without any psychiatric disorders (n = 5), and nonveterans with anxiety disorder (n = 8); all patients were medication free for the 72-hours prior to testing. Cardiovascular responses were measured during the performance of mental- arithmetic tasks and during repeated exposure to emotionally meaningful combat sounds and music. Veterans with PTSD had higher basal heart rates under all experimental conditions, and they were the only group to show a significant increase in heart rate in response to combat sounds, even at low volume. Orr et al. (2003) conducted a study of 130 monozygotic twin pairs recruited from the Vietnam Era Twin Registry; in each pair, one twin had served in combat in Vietnam and the other had served in the military but not in Vietnam. PTSD was diagnosed with the CAPS, and each combat-exposed twin completed the Combat Severity Scale and the SCID with DSM-IV criteria for other psychiatric disorders. Study participants were exposed to auditory startle stimuli and assessed for heart rate and other skin-conductance and orbicularis oculi electromyographic responses. Twins with PTSD had greater combat severity and greater heart-rate response than twins without PTSD. Cardiovascular Symptoms As in studies comparing deployment and nondeployment, PTSD in Gulf War veterans has been found to be associated with an increased prevalence of multiple physical symptoms. Hoge et al. (2007) studied 2863 OIF veterans 1 year after their return from combat duty in Iraq and found that chest pain was reported by 15.1% of veterans who had PTSD but by only 3.5% of those who did not (OR 4.98, 95% CI 3.51-7.05, p < 0.0001).
192 GULF WAR AND HEALTH The Veterans Health Study, conducted by the VA at four ambulatory-care medical clinics in the Boston area in 1993-1995, found that veterans with PTSD (n = 351) screened with the PTSD Checklist-Civilian Version were significantly more likely (p < 0.05) than veterans without PTSD (n = 1455) to complain of angina (OR 2.09, 95% CI 1.64-2.66) and congestive heart failure (OR 1.64, 95% CI 1.03-2.61) but not high blood pressure (OR 1.24, 95% CI 0.98-1.57) or transient ischemic attacks (OR 1.55, 95% CI 0.88-2.70); ORs were adjusted for age and depression. Health status was assessed with the SF-36 self-administered questionnaire (Spiro et al. 2006). Falger et al. (1992) found that angina was more common in Dutch World War II resistance fighters than in a control group of men of similar age who had not been in the resistance. About half the resistance fighters had PTSD in 1986, and those with it were more likely to complain of angina (31% vs 14%). An analysis of 605 male veterans of World War II and the Korean War with a followup of about 25 years found no increase in hypertension or CHD in relation to PTSD (Schnurr et al. 2000). Cardiovascular Disease Two primary studies assessed the effect of having PTSD on cardiovascular disease, one in Vietnam veterans (Boscarino 2005) and one in World War II and Korean War veterans (Kubzansky et al. 2007). Boscarino (2005) examined the causes of death among male Vietnam Army veterans about 30 years after their military service and 16 years after they had completed a telephone survey to ascertain their health status. The men were included in a national random sample of veterans from the VES. Cardiovascular mortality was increased in the 836 theater veterans with PTSD but not in the 214 era veterans with PTSD (hazard ratio 1.7, 95% CI 1.0-2.7, p = 0.034). Kubzansky et al. (2007) analyzed data on two cohorts of 1946 male veterans (average age, 60 and 63 years) who were evaluated for PTSD in 1986 and 1990 and followed for 11-15 years for the Normative Aging Study. Two measures were used to screen for PTSD: the Mississippi Scale for Combat-Related PTSD was administered to 1002 men in 1990 and the Keane PTSD scale was administered to 944 men in 1986. The end points included angina, MI, and death from CHD. Although there was a linear relationship between the severity of PTSD and the incidence of cardiovascular events in both cohorts, it lost significance if there was adjustment for potential confounders (including blood pressure and cholesterol) in both cohorts. Thus, for total CHD events, the RR was 1.19 (95% CI 0.98-1.43) in the Mississippi Scale for Combat- Related PTSD cohort and 1.21 (95% CI 0.99-1.48) in the Keane scale cohort. Adjusting for depression as an additional confounder was attempted in both cohorts. In the Mississippi Scale for Combat-Related PTSD cohort, this was done by using the Center for Epidemiological Studies Depression Scale score, but the relationship between PTSD and total CHD events was still not significant (RR 1.21, 95% CI 0.93-1.57), although it did achieve significance in the Keane scale cohort (RR 1.35, 95% CI 1.03-1.78). However, in the latter case, depression was controlled for by using the Symptom Checklist-90 scale, which is not generally regarded as a robust measure of depression. There was no evidence of any effect of PTSD on mortality. In a secondary study, Boscarino and Chang (1999) examined the electrocardiograms (ECGs) of 4462 randomly selected male Army veterans: 2490 who were deployed to Vietnam and 1972 who were not. The average interval between deployment and the analysis was 17 years. ECGs, medical examinations, and psychiatric evaluations with the DIS (based on DSM-III) were conducted in 1985-1986 at one medical facility. PTSD was present in 54 veterans, anxiety in
HEALTH EFFECTS 193 186, and depression in 157. All those with PTSD, 66% of those with anxiety, and 71% of those with depression had served in Vietnam. PTSD was associated with increased atrioventricular conduction defects (OR 2.81, 95% CI 1.03-7.66, p < 0.05) and MI (OR 4.44, 95% CI 1.20-16.43, p < 0.05) adjusted for numerous socioeconomic and health risk factors. Depression was associated with increased arrhythmia (OR 1.98, 95% CI 1.22-3.23, p < 0.01). Anxiety disorder was not associated with any ECG abnormalities. The study is limited by the small number of subjects in each psychiatric group. In another analysis, Boscarino (1997) examined the medical records of 1399 Vietnam veterans from the CDC VES, of whom 332 had PTSD, 17 years after exposure. Circulatory diseases (defined by ICD-9 codes 401-459) were more common in those with lifetime PTSD than in those without PTSD (25% vs 12.9%, OR 1.62, 95% CI 1.14-2.30) after adjustment for general technical test results at Army induction, race, region of birth, type of enlistment, Vietnam volunteer status, Army marital status, current age, hypochondriasis, physical limitations, psychiatric limitations, postinduction alcohol and drug dependence, cigarette pack-years, education level, and current household income. The study rates as a secondary study because the ICD-9 codes include a very broad range of conditions, including hypertension and angina. A meta-analysis of 11 studies reporting on the association between PTSD and cardiovascular end points (either physician-diagnosed or self-reported), which included the secondary studies discussed above (but not the Kubzansky study), found that there was more angina but not more MI in war veterans with PTSD than in those without it (Gander and von Kanel 2006). Summary and Conclusions Clinical manifestations of cardiovascular disease take many years to develop, particularly in relatively young people like soldiers at war, so the original source of stress, in this context combat exposure and other forms of deployment stress, will have ended many years before the stress response, such as high blood pressure or MI, becomes apparent. The committee found that self-reports of many cardiovascular symptoms, such as increased heart rate and chest pains, are increased in deployed veterans, particularly those from the Gulf War, compared with their nondeployed counterparts. Some of the studies reported an increase in hypertension related to deployment, but they are almost all based on self-reports, and others found no effect of deployment. A small increase in the prevalence of hypertension in deployed Gulf War veterans cannot be excluded, but the data cited above are inconsistent, and, because most are based on self-reports, not much reliance can be placed on them. The deployed veterans, particularly those who had unexplained symptoms after the Gulf War, probably sought more medical care, and this alone might increase the likelihood of receiving a diagnosis of hypertension. There are consistent findings that deployment to a war zone, particularly the Gulf War, is associated with an increase in self- reports of many physical symptoms, including chest pain and increased heart rate, but these symptoms do not necessarily imply any structural heart disease. The two primary studies, one on Gulf War veterans and the other on Vietnam veterans, that used physical examinations for hypertension were both negative. Of the six secondary studies for hypertension, two were negative and four were positive. Blood lipids, another important risk factor for CHD, do not appear to be affected by deployment, although PTSD may raise them.
194 GULF WAR AND HEALTH Because the potential followup period after the Gulf War is still relatively short (less than 20 years) and the deployed veterans are still relatively young, it is not surprising that there has been no suggestion that Gulf War veterans are at increased risk for CHD as a result of deployment. Symptoms of chest pain are common, but they appear to be part of a nonspecific increase in general symptomatology (see âSymptom Reportingâ later in this chapter) and do not themselves imply organic heart disease. Veterans of the Vietnam War are now at an age at which heart disease is prevalent, but again there is no consistent evidence that they are at increased risk as a result of their deployment. All five of the primary studies that assessed CHD in deployed and nondeployed veterans of the Gulf War and the Vietnam War showed no association; the two secondary studies were mixed. Apart from nonspecific symptoms, the one long-term medical consequence of deployment in the Gulf War and other wars is a marked increase in the rate of PTSD. This appears to involve sensitization of the sympathetic nervous system, but the most consistent finding has been an increase in cardiovascular reactivity to simulated trauma manifested by exaggerated heart-rate and blood-pressure responses. This pattern was observed in three ambulatory studies: one found a significant increase (Muraoka et al. 1998), a second had an increase that did not quite achieve statistical significance (Beckham et al. 2004), and in the third only nighttime heart rate was increased (Haley et al. 2004). The studies that showed increased catecholamine production suggest that it may be explained by increased sympathetic nerve activity although the spectral analysis done by Haley et al. (2004) indicates that altered parasympathetic tone also contributes. Although there may be an increase in resting heart rate, which is a risk factor for both hypertension and cardiovascular events, PTSD does not appear to lead to hypertension. The results for the association of PTSD and cardiovascular disease are mixed: one primary study and two secondary studies, all from the VES were positive, but one primary study on veterans of World War II and the Korean War found no association. The one primary study showed slight but not significant increase in total CHD in veterans with PTSD. Thus, there is suggestive, but not conclusive, evidence that PTSD increases the risk of CHD. The committee concludes that there is inadequate/insufficient evidence of an association between deployment to a war zone and hypertension. The committee also concludes that there is inadequate/insufficient evidence of an association between deployment to a war zone and coronary heart disease.
TABLE 6-7 Cardiovascular Diseases Study Study Design Population Outcomes Results Adjustments Comments Eisen et al. 2005 Cross-sectional, 1061 GW-deployed Hypertension Hypertension OR 0.90, Age, sex, race, cigarette- Study limited by prevalence veterans vs 1128 95% CI 0.60-1.33 smoking, duty type, low participation NHSGWEVTF nondeployed veterans service branch, rank, rate, number of education years since war (Derived from Kang et al. 2000b) CDC 1988b Retrospective 2490 Vietnam-theater Electrocardiograms, Hypertension OR 1.1, Age at enlistment, race, Low participation cohort, veterans, 1972 Doppler instrument 95% CI 0.9-1.2; year of enlistment, rate in control VES prevalence, Vietnam-era veterans for peripheral arterial ischemia OR 1.1, 95% enlistment status group; CI not given population- randomly selected from system, physical CI 0.7-1.7; (volunteer vs draftee), based, telephone 7924 theater veterans, examination left ventricular score on general technical interview with 7364 era veterans who hypertrophy OR 1.8, test, primary military screening had entered Army in 95% CI 1.0-3.3; any occupational specialty medical 1965-1971 electrocardiographic examination at finding OR 1.1, 95% CI followup 0.9-1.3; altered peripheral arterial hemodynamic finding OR 1.2, 95% CI 0.9-1.7 Kang and Cross-sectional, 621,902 GW-deployed Mortality and vital Cardiovascular Age, race, service branch, Study had good Bullman 2001 mortality veterans, status determined mortality: men rate ratio unit component, marital power to detect 746,248 nondeployed with VA BIRLS 0.90, 95% CI 0.81-1.01; status small increases in veterans database and SSA women rate ratio 0.96, risk; limited by Master Beneficiary 95% CI 0.55-1.69 relying on death Record database certificates rather than medical records Gray et al. 1996 Retrospective 547,076 active-duty Hospital-discharge OR about 0.9-1.1 (exact Prewar hospitalization, ORs statistically cohort, GW veterans, 618,335 diagnoses of value not given), 95% sex, age, race, service significantly below hospitalization non-GW veterans circulatory system CI < 1.0 for all 3 years branch, marital status, 1, but no values disease rank, length of service, given; no salary, occupation separation of specific illnesses 195
TABLE 6-7 Cardiovascular Diseases 196 Study Study Design Population Outcomes Results Adjustments Comments Gray et al. 2000 Retrospective 652,979 GW veterans, Hospital-discharge Circulatory system Age, sex, race (only for Able to assess only cohort, 652,922 randomly diagnoses of disease: DoD PMR DoD PMR) illnesses that hospitalization selected non-GW- circulatory system 0.94, 95% CI 0.91-0.98; resulted in deployed veterans disease at DoD, VA, VA PMR 0.85, 95% CI hospitalization; COSHPD hospital 0.76-0.93; COSHPD possible undetected systems PMR 0.98, 95% CI confounders 0.82-1.14 Boscarino 2005 Cross-sectional, 7924 Vietnam-theater Mortality determined Vietnam-theater Race, Army volunteer Risk factors based mortality in veterans (836 with 30 years after war, veterans with PTSD HR status, age at Army entry, on self-reports; (Derived from subset of VES PTSD), 7364 Vietnam- based on VA BIRLS 1.7, 95% CI 1.0-2.7, p = Army discharge status, PTSD status VES) era veterans (214 with and SSA Master 0.034; Vietnam-era Army illicit-drug abuse, determined with PTSD) Beneficiary Record veterans with PTSD HR intelligence, age nonvalidated databases, and 1.2, 95% CI 0.4-3.4, p = questionnaire National Death Index 0.69 Plus Kubzanky et al. Cohort study, 1946 veterans from Mailed questionnaireDiastolic blood pressure Age, smoking, blood Response rate 2007 prospective among 2280 containing MSCRP slight correlation (r = pressure, serum total 81.2%; study community-dwelling in 1986 or Keane -0.06, p = 0.04) cholesterol, BMI, family participants Normative Aging men in Greater Boston, PTSD scale from between lower diastolic history of CHD, assessed with Study MA, area 21-80 years MMPI in 1990 used blood pressure and education, alcohol intake physical old in 1961 to assess PTSD, score on MSCRP but examination every physical examinationnot on MMPI-2; higher 3-5 years for all cardiovascular PTSD scores had risk factors, slightly greater risk for including total CHD (RR 1.21, biochemical values 95% CI 0.93-1.57); nonfatal MI RR 1.30, 95% CI 0.92-1.84; angina pectoris RR 1.01, 95% CI 0.74-1.37 NOTE: BIRLS = Beneficiary Identification and Records Locator Subsystem, BMI = body-mass index, CHD = coronary heart disease, CI = confidence interval, COSHPD = California Office of Statewide Health Planning and Development, DoD = Department of Defense, GW = Gulf War, HR = hazard ratio, MI= myocardial infarction, MMPI = Minnesota Multiphasic Personality Inventory, MSCRP = Mississippi Scale for Combat-Related PTSD, NHSGWEVTF = National Health Survey of Gulf War Era Veterans and Their Families, OR = odds ratio, PMR = proportional morbidity ratio, PTSD = posttraumatic stress disorder, RR = relative risk, SSA = Social Security Administration, VA = Department of Veterans Affairs, VES = Vietnam Experience Study.
HEALTH EFFECTS 197 RESPIRATORY SYSTEM DISEASES The term respiratory disease encompasses a wide variety of conditions, of which the most prominent is chronic obstructive pulmonary disease (COPD). Asthma and emphysema are common COPDs. It is estimated that every year over 340,000 Americans die of respiratory disease and that more than 35 million Americans live with a chronic respiratory disease (American Lung Association 2007). The well-known risk factors for respiratory disease include smoking and exposure to some environmental contaminants, such as ozone. It is important to note that many factors associated with deployment, such as chemical or particulate air-pollutant exposure on the battlefield, could directly affect respiratory-disease risk. A few studies of deployed veterans collected focused objective data on respiratory effects (for example, pulmonary-function tests, validated pulmonary outcomes, and physical examination); a larger number of studies did not have a primary focus on respiratory disease and collected more-limited information on respiratory effects and symptoms and often included a respiratory-disease assessment only as part of a larger multisymptom checklist. Although they were not the focus of this committee, a much smaller number of studies evaluated other respiratory risk factors associated with deployment, such as exposure to oil-field smoke, various chemicals, or particles. For the purposes of this section, the committee defined a primary study according to its methodologic rigor and its use of pulmonary-function testing or a hospital record to identify respiratory effects. A secondary study used self-reported respiratory symptoms or self-reported physician-diagnosed conditions or lacked methodologic rigor. The primary studies of respiratory effects are summarized in Table 6-8. Primary Studies The committee identified seven primary studies and numerous secondary studies relevant to its determination of whether there is a link between deployment-related stressors and respiratory diseases. Most of the primary studies assessed veterans of the 1990-1991 Gulf War (Eisen et al. 2005; Gray et al. 1999, 2000; Ishoy et al. 1999; Karlinsky et al. 2004; Kelsall et al. 2004b); one was conducted on Vietnam veterans (CDC 1988b). The VES was a major study of all U.S. Army Vietnam veterans conducted 15-20 years after the war (CDC 1988b). It surveyed a random sample of all 5 million Army veterans who served during the Vietnam era. The study gathered information in 1985-1986 by telephone interview from 7924 U.S. Army Vietnam veterans and 7364 non-Vietnam veterans; medical examinations were conducted for a random subset of participants (2490 Vietnam and 1972 non- Vietnam veterans). The medical evaluation included pulmonary-function testing using a wedge spirometer and chest x-ray pictures. After adjustment for age at enlistment, race, year of enlistment, and service characteristics, the prevalence of abnormal pulmonary x-ray findings was somewhat elevated (16.0% vs 14.1%, OR 1.1, 95% CI 1.0-1.4), but the frequency of adverse findings did not differ significantly between theater and era veterans for diminished forced expiratory volume in 1 second (FEV1, â¤ 80% of predicted) (10.2% vs 10.9%, OR 0.9, 95% CI 0.7-1.1), diminished forced vital capacity (FVC, â¤ 80% of predicted) (7.1% vs 6.8%, OR 1.0, 95% CI 0.8-1.3), or the FEV1/ FVC% ratio (< 70%) (6.1% vs 6.1%, OR 1.0, 95% CI 0.8-1.3).
198 GULF WAR AND HEALTH In the first phase of VAâs nationally representative study, the National Health Survey of Gulf War Era Veterans and Their Families, Kang et al. (2000b) used a stratified random- sampling method to compare 11,441 Gulf War-deployed veterans with 9476 non-Gulf War veterans from the DoD DMDC database. Deployed veterans self-reported more of all but one of the 48 symptoms evaluated for occurrence in the preceding 12 months than did the nondeployed veterans; several of them were respiratory in nature: runny nose, sore throat, coughing, shortness of breath, wheezing, tightness in the chest, sinusitis, bronchitis, and asthma. It is important to note that higher rates of exposure to respiratory irritants (such as exhaust from tent heaters and oil-well smoke) were also reported. The data were not adjusted for potential confounders. Eisen et al. (2005), using the National Health Survey of Gulf War Era Veterans and Their Families conducted by Kang et al. (2000b), performed medical evaluations, including pulmonary-function testing, of 1061 Gulf War-deployed veterans and 1128 nondeployed veterans at VA medical centers throughout the country. The data were gathered 10 years after the Gulf War. The investigators categorized respiratory conditions as self-reported asthma, bronchitis, emphysema, or obstructive lung disease (history of disease, symptoms plus use of bronchodilators, or 15% improvement in FEV1 after bronchodilator use). After adjustment for age, sex, race, years of education, and cigarette-smoking, no significant differences were observed between deployed and nondeployed veterans in the prevalence of self-reports of asthma, bronchitis, or emphysema (5.9% vs 6.3%, OR 1.07, 95% CI 0.65-1.77) or clinical evaluation for obstructive lung disease (4.5% vs 5.9%, OR 0.91, 95% CI 0.52-1.59). As phase III of the National Health Survey of Gulf War Era Veterans and Their Families, Karlinsky et al. (2004) conducted a cross-sectional survey of 1036 U.S. Gulf War-deployed veterans and 1103 nondeployed veterans 10 years after deployment. The goal of this phase was to assess late-onset respiratory symptoms and pulmonary-function abnormalities. The study subjects were queried about self-reported symptoms and physician-diagnosed illnesses and hospitalizations and received medical examinations in 1999-2001, including pulmonary-function tests with spirometers. The analyses did not adjust for smoking or other confounders, but slightly more deployed veterans (51%) than nondeployed veterans (44%) had a history of smoking. Deployed veterans had more self-reports of wheezing than did nondeployed veterans (OR 1.67, 95% CI 1.06-2.62), but there were no statistical differences between the two groups in cough (OR 1.12, 95% CI 0.80-1.57), physician visits for pulmonary complaints (OR 1.07, 95% CI 0.51- 2.24), pulmonary hospitalizations (OR 0.91, 95% CI 0.13-6.51), asthma (OR 0.90, 95% CI 0.50- 1.62), bronchitis (OR 1.08, 95% CI 0.50-2.34), or emphysema (OR 4.45, 95% CI 0.74-26.68). Ishoy et al. (1999) evaluated respiratory effects in 1997-1998 in a population-based cohort of Danish soldiers who participated in peacekeeping operations in the Persian Gulf region in 1990-1997. The study population consisted of 686 deployed veterans and 231 age- and sex- matched nondeployed veterans. Respiratory disease was assessed by pulmonary-function testing with a Vitalograph spirometer. No differences in smoking status were observed between the groups, and no significant differences were observed between deployed and nondeployed veterans in mean FVC (100.7 Â± 11.6 standard deviation [SD] vs 100.7 Â± 13.1 SD), mean FEV1 (95.6 Â± 11.8 SD vs 96.4 Â± 13.7 SD), and peak flow (94.0 Â± 19.0 SD 19.0 vs 92.8 Â± 19.1 SD). In 1994, Gray et al. (1999) surveyed U.S. Navy Seabees from 14 commands who were still on active duty for at least 3 years after the Gulf War (n = 527) and nondeployed Seabees (n = 970). The study subjects were queried about self-reported physician-diagnosed illnesses, symptoms, and exposures. They also underwent pulmonary-function testing with a spirometer and a pneumotach flow-measuring device; data were adjusted for age, height, race, and cigarette-
HEALTH EFFECTS 199 smoking. Active-duty Gulf War Seabees were significantly more likely than nondeployed Seabees to report either cough (OR 1.8, 95% CI 1.2-2.8) or shortness of breath (OR 4.0, 95% CI 2.2-7.3). Markers of pulmonary function did not differ between the deployed and nondeployed veterans: FVC 4.96 L vs 4.99 L, respectively (p = 0.77), and FEV1 4.05 L vs 4.04 L, respectively (p = 0.81). This study is limited in that it surveyed only active-duty Seabees. In a cross-sectional survey comparing 1456 Australian Gulf War veterans with 1588 nondeployed veterans, study subjects gave self-reports of their respiratory symptoms in the preceding 12 months and underwent spirometric pulmonary-function testing and a physical examination (Kelsall et al. 2004b). The data were gathered in 2000-2002. Deployed veterans reported significantly more respiratory symptoms (wheeze, nocturnal chest tightness, cough, and dyspnea) than did nondeployed veterans. They also reported more asthma (OR 1.4, 95% CI 1.1- 1.9) and chronic bronchitis (OR 1.9, 95% CI 1.2-3.1) but were not at increased risk for airflow limitations, defined as FEV1/FCV% < 70% (OR 0.8, 95% CI 0.5-1.1) or emphysema (OR 1.0, 95% CI 0.8-1.4). Furthermore, there were no statistically significant differences between deployed and nondeployed veterans in any lung-function indexes (FEV1, FCV, or FEV1/FVC%). ORs were adjusted for age, height, smoking, weight, atopy, rank, service, education, and marital status. The study used self-reports of a physicianâs diagnosis or treatment for a diagnosis to determine the prevalence of asthma and bronchitis. During the medical- examination phase of the study, a physician queried each veteran about responses to the questionnaire regarding asthma and bronchitis and classified the likelihood of the diagnosis on a scale from unlikely to probable. Possible exposures to dust and oil-well fires were considered as possible contributors to the respiratory effects seen in the deployed veterans. Based on the European Community Respiratory Health Survey definition of respiratory disease, the only significant risk factor for increased asthma in deployed veterans was having an asthma attack or being awakened by shortness of breath at any time in the preceding 12 months or current use of asthma medications. For bronchitis, the only significant definition was having a doctorâs first diagnosis of chronic bronchitis in 1991 or later. However, if other definitions of the diseases were used, the differences between the two groups were not significant. The study had a response rate of 80.5% for the deployed veterans and 56.8% for the nondeployed. Gray et al. (2000) conducted a study of hospital-discharge records of three hospital systems for 1991-1994: those of DoD, VA, and California Office of Statewide Health Planning and Development. PMRs of hospital-discharge diagnoses of 652,979 Gulf War veterans and 652,922 era veterans not deployed to the gulf were compared. With adjustment for age, sex, and race, the PMRs for respiratory-disease were 1.02 (95% CI 0.99-1.04) for the DoD hospitals, 1.19 (95% CI 1.10-1.29) for the VA hospitals, and 1.06 (95% CI 0.82-1.29) for the California hospitals. Secondary Studies The committee identified numerous large, well-designed studies of Gulf War veterans that it considered secondary mainly because the absence of examinations for respiratory effects led to imprecision regarding evaluation of respiratory disease or lack of adequate adjustment for major confounding variables. In addition to the study by Kang et al. (2000b) discussed above, other large, well- designed cohort studies from several countries show mixed findings regarding the presence of respiratory disease in veterans based on self-reporting via questionnaires. Proctor et al. (1998) looked at 252 U.S. Gulf War-deployed veteransâ186 from the Fort Devens cohort and 66 from
200 GULF WAR AND HEALTH the New Orleans cohortâand compared them with 48 veterans deployed to Germany during the Gulf War. They did not observe any significant differences between groups in chronic lung problems, chronic respiratory symptoms, or allergies. However, a number of other studiesâ including Gulf War veterans from several U.S. states (Iowa Persian Gulf Study Group 1997; McCauley et al. 2002b; Spiro et al. 2006; Steele 2000), Canada (Goss Gilroy Inc. 1998), Australia (OâToole et al. 1996b), and the United Kingdom (Hotopf et al. 2006; Simmons et al. 2004; Unwin et al. 1999)âasked veterans about respiratory symptoms and found that deployed Gulf War veterans had a higher prevalence of self-reports of a variety of respiratory symptoms (colds, asthma, emphysema, chronic bronchitis, persistent cough, and lung disease) than nondeployed controls. Because of their reliance on self-reports or questionnaires, all those studies are vulnerable to recall or reporting bias and a lack specificity regarding the outcome. Gray et al. (2002) found that Gulf War-deployed Seabees self-reported more asthma with onset after August 1991 (2.4%) than did Seabees deployed to areas other than the gulf (1.7%) or nondeployed Seabees (1.5%); that indicated that deployed Seabees were at increased risk for asthma compared with nondeployed Seabees (OR 1.82, 95% CI 1.23-2.69). However, although adjustments were made for age, sex, active-duty or reserve status, race or ethnicity, current smoking, and current alcohol drinking, exposures to a number of toxicants were possible confounders. Gray et al. (1996) used hospitalization records from DoD medical facilities to compare the prevalence of ICD-9 respiratory diseases in 547,076 veterans deployed to the Gulf War compared with 618,335 nondeployed veterans. The OR for respiratory diseases decreased from slightly greater than 1 in 1991 to slightly less than 1 in 1993 (exact values not given). Posttraumatic Stress Disorder and Respiratory Diseases Self-reported health status was assessed in a cross-sectional survey that used a mailed questionnaire sent to 1259 female U.S. veterans who received care at the VA Puget Sound Health Care System in 1996-1998 (Dobie et al. 2004). The women were screened for PTSD with the PTSD Checklist-Civilian Version. Compared with female veterans who did not have PTSD (n = 940), those with PTSD (n = 266) had a greater prevalence of emphysema (13.8% vs 10.7%, OR 1.88, 95% CI 1.21-2.92) and asthma (24.4% vs 17.3%, OR 1.64, 95% CI 1.17-2.31%) adjusted for age. It is unclear whether the excess of respiratory symptoms could be explained by higher rates of smoking in the veterans with PTSD (39.5% vs 22.9). Boscarino (1997), using data from the VES, found a modest but significantly higher prevalence of respiratory diseases in Vietnam veterans who had lifetime combat-related PTSD (n = 1067) than in those without PTSD (n = 332) (OR 1.54, 95% CI 1.02-2.35, p = 0.042) adjusted for a variety of demographic, social, and Army characteristics. A study of Gulf War veterans found no statistical difference in self-reporting of pulmonary symptoms (difficulty in breathing, shortness of breath, common cold or influenza, and rapid breathing) between deployed veterans with PTSD or MDD or both (n = 20) and deployed veterans with neither psychiatric condition (n = 178) or veterans who were deployed to Germany during the Gulf War (n = 48) (Wolfe et al. 1999). Summary and Conclusions The committee placed the greatest weight on studies that included a medical evaluation, identified specific respiratory diagnoses, and adjusted for potential confounding variables.
HEALTH EFFECTS 201 Although an association between deployment and respiratory effects was noted in some studies that assessed respiratory effects on the basis of self-reports of symptoms, studies that used more objective markers of respiratory disease (such as pulmonary-function tests and respiratory examinations) did not document a consistent relationship between deployment and respiratory disease. None of the primary studies considered for respiratory effects (one study in Vietnam veterans and six in Gulf War veterans) found a higher risk of chronic respiratory effects in deployed than in nondeployed veterans on the basis of objective measures of pulmonary function and disease. However, deployed veterans reported more symptoms of respiratory effects, such as cough and shortness of breath, in one Gulf War study and of asthma and chronic bronchitis in a second Gulf War study than did the nondeployed veterans. In both studies, on examination, pulmonary function was not compromised in the deployed groups. Of the 12 secondary studies that relied on self-reports of respiratory effects, 11 found that deployed Gulf War veterans reported more symptoms of respiratory effects. The one secondary study that assessed hospitalizations of Gulf War veterans for respiratory disease found no increase in risk. Results regarding the relationship between PTSD and respiratory effects are mixed. No primary studies were identified; in the secondary studies, although modest increases in respiratory symptoms were reported in Vietnam veterans and in female veterans with PTSD, no such increase was seen in a small study of Gulf War veterans. Respiratory conditions are common in veterans regardless of deployment status. The population burden of asthma and COPD in all U.S. adults is high, but COPD takes many years to develop and is uncommon in people under the age of 50 years. The prevalence of respiratory disease is even higher in adults with a history of cigarette-smoking. Thus, with additional followup of the veterans, future studies might show an effect of war-related stress. Many studies fail to make clear whether there is a direct relationship between deployment stress and increased pulmonary complaints as opposed to some other mediating or moderating factor. Veterans of the Gulf War were exposed to numerous air pollutants, including smoke from oil-well fires, pesticides, exhaust from tent heaters, and possibly chemical-warfare agents. Veterans of other conflicts may have similar exposures to toxicants and to other agents, such as Agent Orange in the case of Vietnam veterans. Many of the studies considered by the committee identified other possible exposures of veterans, particular those who were deployed to the Gulf War, but were unable to eliminate possible confounding when assessing risks. The association between increased Gulf War deployment and cigarette-smoking is unknown. Cigarette-smoking was not controlled in the analyses in a number of the studies so it is unclear to what degree smoking may account for any increase in respiratory symptoms, complaints, or disease. The few studies that collected more-objective data on respiratory outcomesâsuch as data on pulmonary-function testing, data from physical examinations, and data on verified pulmonary events from medical-record review or physician questionnairesâdid not find strong indications of increased risk among veterans with combat deployment. The vast majority of available studies of veterans with respiratory diseases were not specifically designed to focus on these outcomes as primary variables but derived information from a larger nonspecific self-administered symptom checklist. The committee concludes that there is inadequate/insufficient evidence to determine whether an association exists between deployment to a war zone and chronic respiratory effects.
TABLE 6-8 Respiratory System Diseases 202 Study Study Design Population Outcomes Results Adjustments Comments CDC 1988 Retrospective 2490 Vietnam- Chest x-ray findings, Chest x-ray OR 1.1, 95% CI Age at enlistment, Low participation cohort, theater veterans, pulmonary-function 1.0-1.4; FEV1 â¤ 80% of race, year of rate in control group VES prevalence, 1972 Vietnam-era tests predicted OR 0.9, 95% CI enlistment, population-based, veterans randomly 0.7-1.1; FVC â¤ 80% of enlistment status telephone selected from 7924 predicted OR 1.0, 95% CI (volunteer vs interview with theater veterans and 0.8-1.3; FEV1/ FVC < 70% draftee), score on screening medical 7364 era veterans OR 1.0, 95% CI 0.8-1.3 general technical examination at who entered Army in test, primary followup 1965-1971 military occupational specialty Eisen et al. 2005 Cross-sectional, 1061 GW-deployed Self-reported asthma, Asthma, bronchitis, or Age, sex, race, Study limited by low prevalence veterans, 1128 bronchitis, or emphysema OR 1.07, 95% cigarette-smoking, participation rate NHSGWEVTF nondeployed emphysema; CI 0.65-1.77; duty type, service and number of years veterans obstructive lung obstructive lung disease OR branch, rank since war (Derived from disease (history of 0.91, 95% CI 0.52-1.59 Kang et al. 2000b) disease or symptoms and use of bronchodilators or 15% improvement in FEV1 after bronchodilator use) Karlinsky et al. Cross-sectional 1036 GW-deployed, Pulmonary-function Cough OR 1.12, 95% CI Age, height, No adjustment for 2004 1103 nondeployed, tests with spirometer, 0.80-1.57; wheezing OR smoking status, smoking although assessed 10 years self-reports of 1.67, 95% CI 1.06-2.62; weight, education, smoking history NHSGWEVTF after war respiratory symptoms physician visits for service reported or use of medication pulmonary complaints OR characteristics (Derived from 1.07, 95% CI 0.51-2.24; Kang et al. 2000b) pulmonary hospitalization: OR 0.91, 95% CI 0.13-6.51; asthma OR 0.90, 95% CI 0.50-1.62; bronchitis OR 1.08, 95% CI 0.50-2.34; emphysema OR 4.45, 95% CI 0.74-26.68
TABLE 6-8 Respiratory System Diseases Study Study Design Population Outcomes Results Adjustments Comments Ishoy et al. 1999 Cross-sectional 686 Danish Health examination No significant differences in Participation rate peacekeepers by physician, lung-function values for 83.6% deployed, deployed to gulf in including lung FVC, FEV1, peak flow 57.8% nondeployed 1990-1997, 231 age- function and self- between groups and sex-matched report questionnaire armed forces nondeployed controls Gray et al. 1999 Cross-sectional, 527 GW veterans, Pulmonary-function Cough OR 1.8, 95% CI 1.2- Age, height, race, Only active-duty medical 970 nondeployed testing with 2.8; shortness of breath OR smoking status Seabees surveyed evaluation from 14 U.S. Navy spirometer with 4.0, 95% CI 2.2-7.3; FVC Seabees commands Fleisch-type 4.96 L deployed, 4.99 L pneumotach flow- nondeployed, p = 0.77; measuring device FEV1 4.05 L deployed, 4.04 L nondeployed, p = 0.81 Kelsall et al. 2004b Cross-sectional 1456 deployed Self reports of Self-reported asthma OR 1.4, Age, height,Although physician Australian GW symptoms, 95% CI 1.1-1.9; bronchitis smoking, weight, reviewed and veterans, 1588 pulmonary-function OR 1.9, 95% CI 1.2-3.1; air atopy, education, classified likelihood nondeployed testing with flow limitation OR 0.8, 95% marital status, of accuracy of self- veterans spirometer CI 0.5-1.1; emphysema OR service, rank report of diagnosis, 1.0, 95% CI 0.8-1.4; no physical lung-function indexes all examinations ORs -0.04-0.7 conducted except for lung function Gray et al. 2000 Retrospective 652,979 GW Hospital-discharge DoD PMR 1.02, 95% CI Age, sex, race Able to assess only cohort, veterans, 652,922 diagnoses of 0.99-1.04; VA PMR 1.19, illnesses that hospitalization randomly selected respiratory disease in 95% CI 1.10-1.29; resulted in nondeployed DoD, VA, COSHPD CA PMR 1.06, 95% CI 0.82- hospitalization; veterans hospital systems 1.29 possible undetected confounders NOTE: CDC = Centers for Disease Control and Prevention, CI = confidence interval, COSHPD = California Office of Statewide Health Planning and Development, DoD = Department of Defense, FEV1 = forced expiratory volume in 1 second, FVC = forced vital capacity, GW = Gulf War, OR = odds ratio, NHSGWEVTF = National Health Survey of Gulf War Era Veterans and Their Families, PMR = proportional morbidity ratio, VA = Department of Veterans Affairs, VES = Vietnam Experience Study. 203
204 GULF WAR AND HEALTH DIGESTIVE SYSTEM DISORDERS Soon after the Gulf War, veterans reported gastrointestinal (GI) symptoms more frequently than most other symptoms (Kang et al. 2000b). That is understandable because there is a common and scientifically recognized association between acute and chronic stress and GI dysfunction (Creed et al. 2006; Drossman and Chang 2003). This dysfunction can lead to changes in intestinal movements (motility) that affect gastric emptying rates and intestinal transit time which in turn cause nausea, vomiting, bloating, diarrhea, and constipation. Psychological distress can also affect sensitivity of the visceral nerves, thus producing abdominal discomfort and pain (Drossman 2002, 2006a,b; Kellow et al. 2006a,b) (see Chapter 4 for more discussion on GI dysfunction in response to stress). The functional GI disorders or syndromes, such as irritable bowel syndrome (IBS) and functional dyspepsia, are characterized by recurrent or prolonged clusters of symptoms that range in severity from occasional mild episodes to more persistent and disabling episodes with impaired health-related quality of life. So named because they are disturbances of GI functioning rather than diseases, these disorders are understood in a biopsychosocial construct (Drossman 1998); that is, genetic or early environmental predisposing factors, including family enablement of illness behaviors (Levy et al. 2000) and a history of early trauma or abuse (Drossman et al. 1995), coupled with acute or chronic exposure to stress or acute GI infection (Spiller and Campbell 2006) can precipitate or exacerbate the disorders. The disorders are then sustained or perpetuated in the presence of psychologic comorbidities, including PTSD, anxiety, depression, maladaptive coping style, and impaired social networks (Creed et al. 2006; Drossman et al. 2002; Levy et al. 2006). Of particular relevance here is the growing evidence of development of postinfectious IBS. In some cases, functional GI disorders are triggered by pathogens, which usually cause acute gastroenteritis, and the symptoms are then sustained by stressful conditions (Drossman 1999; Dunlop et al. 2003; McKeown et al. 2006; Spiller and Campbell 2006). From a biologic perspective, the functional GI syndromes are characterized by dysregulation of neural pathways between the brain and gut (that is, the brain-gut axis) that produces persistent motility and sensory disturbances (visceral hypersensitivity), dysregulation of the hypothalamic-pituitary- adrenal (HPA) axis (see Chapter 4), alteration in corticolimbic pain modulation, and inflammation of the bowel mucosa associated with altered bacterial flora (Drossman 2006a,b; Drossman et al. 2002). Diagnosis of a functional GI disorder is based on identification of specific bowel symptoms that fulfill the Rome Criteria1 and a minimal period of symptoms, usually 6 months (Drossman 2006a,b). The diagnostic criteria have rarely been included in the assessment of Gulf War veterans, so the veteransâ diagnoses have been presumptive, that is, based on sufficient clusters of symptoms that are consistent with the Rome criteria for diagnosis. However, diagnostic criteria were used in a few supportive studies of selected cohorts of Gulf War veterans that yielded relevant physiologic data (Dunphy et al. 2003). 1 The Rome Criteria, developed by the Rome Foundation, are symptom-based diagnostic criteria for functional GI disorders, including functional gastroduodenal disorders, functional bowel disorders, and the group of disorders formerly referred to as functional biliary disorders. IBS is defined by Rome III criteria as abdominal discomfort or pain of at least 3 monthsâ duration in which two of the following three are present: pain that is improved by defecation, association of the onset of pain with a change in the frequency of the stool, and association of the onset of pain with a change in the consistency (looser or harder) of the stool (http://www.romecriteria.org).
HEALTH EFFECTS 205 GI diseases, sometimes called organic or structural, such as peptic ulcer or inflammatory bowel disease (that is, ulcerative colitis or Crohnâs disease), are characterized by morphologic abnormalities seen on x-ray pictures, endoscopically, or through laboratory tests. For example, in inflammatory bowel disease, intestinal inflammation may lead to ulcerations, strictures, and bowel injury. Acute stress can produce symptoms and even activate symptoms in people with existing disease (see Chapter 4), but the relationship of chronic stress to the onset of disease is difficult to study because onset may take years. For the purposes of this section, the committee defined primary studies according to their methodologic rigor (see Chapter 2) and outcome assessment, requiring sufficiently valid symptom clusters consistent with a functional GI diagnosis or, in the case of structural diseases, physical examination. Although the committee identified several papers that assessed the development of GI effects in veterans with PTSD, none met the criteria for a primary study, so they are discussed only as secondary studies. The primary studies and several secondary studies are summarized in Table 6-9. Primary Studies To determine the link between deployment-related stressors and GI diseases and disorders, the committee identified three primary studies (Eisen et al. 2005; Gray et al. 2002; Sostek et al. 1996) of deployment and GI effects and several secondary studies. During the Gulf War, U.S. Navy Seabees were responsible for building airports, supply points, and roads. Gray et al. (2002) surveyed nearly 12,000 active-duty, deployed and nondeployed Seabees in 14 commands who were still on active duty at least 3 years after the Gulf War. The subjects gave self-reports of physician-diagnosed illnesses, symptoms, and exposures. IBS was one of the physician-diagnosed illnesses listed on the survey. Active-duty Gulf War Seabees were much more likely than nondeployed Seabees to report having a diagnosis of peptic ulcer disease (OR 3.11, 95% CI 1.67-5.78), any new GI disease diagnosed after September 1990 (OR 2.10, 95% CI 1.39-3.17), or IBS (2.48% vs 0.81%, OR 3.57, 95% CI 2.22-5.73). IBS was one of a cluster of four physician-diagnosed conditionsâPTSD, CFS, and multiple chemical sensitivity were the other threeâfound to be more prevalent among Gulf War-deployed Seabees, and the four conditions were highly associated with one another. Among the deployed Seabees, receiving a diagnosis of one of the four also was associated with reporting 13-18 other symptoms, whereas deployed Seabees who did not report any of the four had fewer (average, six) other symptoms. The clustering of the symptoms and conditions is consistent with emerging data that indicate that multiple symptom reporting is associated with psychologic trauma, which may disrupt central mechanisms for filtering incoming visceral and somatic neural signals (Drossman et al. 1996; Mayer 2006; Whitehead et al. 2002). Because the focus of the study was to cluster symptoms and conditions that might shed light on a unique Gulf War illness, the analysis undertook no further evaluation of GI conditions in isolation. It was not possible to discern how a diagnosis of IBS was made. Another limitation is that the study included only active-duty veterans and thus was not representative of all Gulf War veterans. A study by Sostek et al. (1996) focused on evaluating the prevalence and spectrum of GI complaints in a group of 57 Gulf War-deployed veterans and 44 veteran controls from the same National Guard who were not deployed to the gulf. A 68-item survey asked the veterans to rate the frequency of their GI symptoms and other symptoms before, during, and after the war. There were no differences between the groups in demographics and no differences in prewar smoking, alcohol or medication consumption, or psychologic problems. The reporting of medical,
206 GULF WAR AND HEALTH including GI, symptoms was low (0-9%) and not different between groups. During the war, Gulf War veterans reported markedly higher rates of GI symptoms (abdominal pain, 63%; gas, 60%; nausea and vomiting, 39%; diarrhea, 67%; and blood in the stool, 23%) than the nondeployed veterans, and about 80% remained symptomatic after the war. The high reported proportion with blood in the stool suggests that the symptoms may have been related in part to intestinal infections. Postwar comparisons showed significant differences between deployed veterans and controls in abdominal pain (70% vs 9%, p = 0.001), relief of pain with bowel movements (47% vs 16%, p = 0.05), postprandial pain (46% vs 14%, p = 0.05), pain associated with increased or watery bowel movements (44% vs 5%, p = 0.001), loose stools (74% vs 18%, p = 0.001), mucus in the stools (19% vs 0%, p = 0.05), and incomplete rectal evacuation (60% vs 7%, p = 0.001). The frequency of blood in the stool was low and not significantly different between groups. In addition, deployed veterans reported significantly higher frequencies of all 10 non-GI symptoms, including headache, fatigue, and joint pains. A strength of the study is that it asked specific GI questions that were more than sufficient to meet the Rome criteria for IBS, functional dyspepsia, and possibly other functional GI disorders (Drossman 2006; Longstreth et al. 2006a,b). Furthermore, the temporal association of symptom development in deployed veterans, possibly enabled by exposure to infectious bacteria, and the retention of symptoms to a significantly higher degree than the controls after the war, strongly supports an association of GI-symptom development with war-zone exposure. The study may be limited by the relatively small sample compared with samples in larger epidemiologic studies, although type I error given these findings is unlikely. Considering the study design, there is also a possibility of retrospective recall bias. Ten years after the Gulf War, in the National Health Survey of Gulf War Era Veterans and Their Families, a nationally representative population-based study, VA conducted medical evaluations that included dyspepsia to determine the prevalence of common diseases in deployed veterans (Eisen et al. 2005). In 1999-2001, 1061 deployed and 1128 nondeployed veterans were evaluated at several VA centers. The veterans had been randomly selected from 11,441 deployed and 9476 nondeployed veterans who had participated in a 1995 VA survey (Kang et al. 2000b) that used a self-report questionnaire. Dyspepsia was diagnosed through in-person interviews according to two criteria: a history or symptoms of dyspepsia (frequent heartburn and recurrent abdominal pain) and use of antacids, histamine-2 receptor blockers, or other medications to treat dyspepsia. The prevalence of dyspepsia was 9.1% and 6.0% in deployed and nondeployed veterans, respectively (OR 1.87, 95% CI 1.16-2.99); the prevalence of gastritis was 5.9% and 4.2%, respectively (OR 1.57, 95% CI 0.88-2.78). A study limitation is that dyspepsia was diagnosed crudely as recurrent abdominal pain or frequent heartburn, which is more commonly associated with gastroesophageal reflux disease. Furthermore, IBS, a more common functional GI disorder, was not evaluated. Finally, despite three recruitment waves, the participation rate in the 2005 study was low: only 53% of Gulf War veterans and 39% of nondeployed veterans invited to participate were examined during the second phase of the study. To determine nonparticipation bias, the authors gathered previously collected findings from participants and nonparticipants from the DMDC and sociodemographic and self-reported health findings from the earlier VA study (Kang et al. 2000b). Both deployed and nondeployed participants were more likely than nonparticipants to report heartburn or indigestion; although this could limit the generalizability of findings, the authors adjusted for the disparity in their analysis of population prevalence.
HEALTH EFFECTS 207 Secondary Studies Most large-scale Gulf War cohort studies conducted in several countries found a postwar excess of self-reported GI symptoms in deployed vs nondeployed military personnel (Kelsall et al. 2004a; Proctor et al. 1998; Simmons et al. 2004; Unwin et al. 1999). Some specifically reported increased rates of âbowel disorder, including diarrhea, constipation, or bleedingâ (Kelsall et al. 2004a). Those studies are limited by being based largely on self-reports. Furthermore, most did not address functional GI disorders, such as IBS, through questionnaires or medical evaluations. A final limitation of many studies is that the outcome being measured was too broad, such as the ICD category Diseases of the Digestive System. In the 1995 prevalence survey of Gulf War veterans by Kang et al. (2000b) discussed above, gastritis and frequent diarrhea were among the top five medical conditions self-reported by deployed (25.2% and 21.2%, respectively) vs non-deployed veterans (11% and 5.9%, respectively) during the preceding 12 months. Enteritis and colitis were reported less frequently but were also more prevalent in deployed than in nondeployed veterans (6.6% vs 2.8% and 4.6% vs 2.2%, respectively). When asked about individual symptoms on the same questionnaire, veterans often reported GI symptoms at high rates. Of more than 50 symptoms, heartburn and indigestion were among the 10 most frequently reported; they also were among the top 10 symptoms reported as severe. Gulf War veterans reported many other symptoms, and that is consistent with the possibility that the stress of war is associated physiologically with the setting of low central nervous system (CNS) thresholds for symptom experience and reporting (see Chapter 4). Because these are self-report data, it is likely that the reports of gastritis, enteritis, and colitis represent functional GI symptoms rather than organic diseases. Nevertheless, the findings support a greater prevalence of those GI and other symptoms in deployed veterans than in nondeployed controls. Eisen et al. (1991) assessed data from the Vietnam Era Twin Registry to determine whether service in the Vietnam theater increased the risk of physical health problems. Twin pairs (n = 2260) that were discordant for currently having, ever having, or having been hospitalized for stomach conditions were surveyed for self-reports of a variety of physical health problems. Deployed twins were at slightly greater risk for ever having a stomach condition, such as dyspepsia (OR 1.4, 95% CI 1.0-1.9), or having been hospitalized for a stomach condition (OR 2.6, 95% CI 1.4-4.9) than the nondeployed twins. Being hospitalized for a stomach condition was also associated with combat exposure although the relationship was not linear. Some studies that did not assess war-zone deployment but did assess surrogate traumatic exposures in other populations lend support to the primary data. One prospective cohort study of IBS patients in a gastroenterology practice in Romania (Dumitrascu and Baban 1991) had several evaluation times beginning with the death of Nikolai Ceausescu in December 1990. Sixty IBS patients were tracked with regard to their bowel symptoms after a political uprising in Romania in December 1989 and in January, March, and June 1990. The trauma was related not only to the political upheaval but to exposure to television images of street fighting when previously all television programs had been censored. Painful GI episodes were reported by 80% of the patients at the onset of the uprising but only 33% of them reported painful episodes 6 months later. The value of this study is in the certainty of the IBS diagnosis and in its prospective design. A case-control study of patients hospitalized in Israel for non-GI complaints (Stermer et al. 1991) looked at the prevalence of GI symptoms in 239 Eastern European Holocaust survivors and 384 Eastern Europeans not exposed to the Nazis. The study used a checklist of 13 GI
208 GULF WAR AND HEALTH symptomsâincluding abdominal pain, irregular bowels, diarrhea, constipation, distention, nausea, and vomitingâthat were consistent with criteria for IBS, dyspepsia, aerophagia, and other functional symptoms; the results showed that all the symptoms were significantly more common in the Holocaust survivors (p < 0.0001-0.02). The study concluded that functional GI symptoms were more severe and more protracted (often for 15-40 years) among those more heavily exposed to trauma. A study involving victims of a severe flash flood and mudslides in Puerto Rico, in which almost 200 people were killed, reported more new-onset GI symptoms than in unexposed controls (Escobar et al. 1992). Over 1550 people had been interviewed in 1984, one year before the disaster as part of an epidemiologic study; 375 of them were reinterviewed in 1987, a little more than a year after the disaster. This prospective study reported incident symptoms. Finally, physiologic studies have yielded strong evidence of an association of acute or chronic stress with altered GI functioning and symptoms. One study of 12 Gulf War veterans, who reported the development of chronic abdominal pain and diarrhea during their time in gulf, showed that those veterans had greater pain intensity, pain unpleasantness, visceral sensitivity, and anxiety in relation to rectal distention and cutaneous hand stimulation than had deployed veterans without reports of chronic abdominal pain or civilians (Dunphy et al. 2003). Other studies in civilians with IBS have shown altered autonomic activity and lower pain thresholds in response to acute physical and psychologic stress (Murray et al. 2004) and a greater association of altered GI motility and GI symptoms with corticotropin-releasing hormone, a stress hormone, compared with controls without IBS (Fukudo et al. 1998). A full review of this subject can be found in Kellow et al. (2006a,b). Studies of associations of war-zone deployment with structural GI diseases, such as peptic ulcer or inflammatory bowel disease, have been reported in some epidemiologic studies. However, none provides sufficient evidence to document a diagnosis. Furthermore, several studies have attributed such diagnoses as gastritis and colitis to functional GI symptoms in the absence of sufficient documentation that the criteria for the diseases have been met. Posttraumatic Stress Disorder and Gastrointestinal Effects An association between functional GI disorders and PTSD appears to be caused by more than chance; however, the committee was unable to identify any studies on PTSD and functional GI disorders that met its criteria for primary studies, mainly because PTSD was assessed only by screening and not by an actual diagnosis. The studies discussed below are all secondary studies. As many as 36% of patients with IBS have been found to have PTSD and other psychiatric comorbidities (Irwin et al. 1996). Several studies have dealt with the question of whether deployment-related PTSD is associated with GI symptoms. Studying OIF veterans a year after their return, Hoge et al. (2007) found that veterans who had PTSD (16.6% of them on the basis of the PTSD Symptom Checklist) were about 2-3 times more likely than veterans without PTSD to report being frequently bothered by stomach pain (OR 3.86, 95% CI 2.80- 5.33); constipation, loose bowels, or diarrhea (OR 3.02, 95% CI 2.28-4.00); and nausea, gas, or indigestion (OR 3.44, 95% CI 2.65-4.48). Gulf War veterans who screened positive for PTSD reported a higher prevalence of GI symptoms and conditions than those without PTSD although the symptoms were not specified (Barrett et al. 2002a). Using data from the Normative Aging Study of older male World War II and Korean War veterans, Schnurr et al. (2000) found that veterans who screened positive for PTSD (n > 590) were more likely also to have new-onset lower GI-tract symptoms (hazard rate
HEALTH EFFECTS 209 1.23, 95% CI 1.04-1.45) according to physical examinations conducted periodically since the 1960s. PTSD was assessed with the Mississippi Scale for Combat-Related TPSD. In the final multivariate model, lower GI disorders increased by 23% for every 10-point increment in PTSD symptom scores. A study of Vietnam veterans from the VES found GI disturbances of a nonspecific nature to be more common in those with a diagnosis of combat-related PTSD (n = 332) than in those without PTSD (n = 1067) (OR 1.47, 95% CI 1.02-2.10) (Boscarino 1997). Finally, Dobie et al. (2004), in a study of female veterans who had received treatment at the VA Puget Sound Health Care System, found that women who screened positive for current PTSD (n = 266) were more likely than those who screened negative (n = 940) to report IBS (37% vs 18%, OR 2.82, 95% CI 2.06-3.85). Overall, most of the secondary studies are limited by self-reporting and cross-sectional design, which enables them to assess only comorbidity. Furthermore, the reported GI symptoms are among many other somatic symptoms reported at high frequency, and this raises the likelihood that PTSD may be like other stress-related conditions in being associated with a general tendency to set lower sensation thresholds and thus to lead to multiple symptom reporting. A limitation of studies that aimed to link PTSD with GI illness is that they used screening criteria, rather than in-person psychiatric evaluation, to assess PTSD. Summary and Conclusions GI symptoms and functional GI disorders have been associated with war-zone deployment in two large-scale and well-designed primary studies of veteran populations and one smaller primary study that used adequate criteria to identify functional GI disorders in Gulf War veterans. The nationally representative study of Gulf War veterans, which relied on medical evaluations, found a relationship between deployment and symptoms of dyspepsia (Eisen et al. 2005); Gulf War veterans also reported a higher prevalence of colitis, enteritis, heartburn, and indigestion. A large questionnaire survey of U.S. Navy Seabees deployed during the Gulf War found more self-reporting of physician-diagnosed IBS compared with nondeployed Seabees (Gray et al. 2002). And a study of Gulf War veterans found that the prevalence of GI symptoms consistent with functional GI disorders that developed during war-zone exposure was higher than the prevalence before deployment and persisted after deployment. The prevalence of those GI symptoms with non-GI symptoms was significantly greater in deployed than in nondeployed veterans (Sostek et al. 1996). Results of six secondary studies are consistent with those of the primary studies in showing higher reporting frequencies of GI symptoms in deployed veterans than in control populations. Vietnam theater veterans reported higher rates of hospitalizations for stomach conditions than era veterans (Eisen et al. 1991). Large cohort studies of Gulf War and Vietnam veterans found excess reporting of GI symptoms (Boscarino 1997; Kelsall et al. 2004a; Proctor et al. 1998; Simmons et al. 2004; Sostek et al. 1996; Unwin et al. 1999). Like the primary Gulf War study by Sostek et al., the Boscarino study (Boscarino 1997) of the Vietnam War provided the appropriate timeline between exposure and onset of GI disturbances. Symptoms were reported to begin during deployment and to persist for years thereafter (Schnurr et al. 2000). There is also some evidence that the development of these syndromes were enabled by exposure to GI infections during the stress of deployment (Sostek et al. 1996; Spiller and Campbell 2006). Three civilian populations exposed to war or severe trauma found a greater prevalence of GI symptoms or disorders in those with greater trauma exposure (Dumitrascu and Baban 1991; Escobar et al. 1992; Stermer et al. 1991).
210 GULF WAR AND HEALTH PTSD was associated with increased GI symptoms in several studies of veterans. In the largest military study, which had the longest followup period (nearly 20 years), combat-related PTSD was associated with more frequent later development of GI diseases in Vietnam-theater veterans than in Vietnam-era veterans (Boscarino 1997). Other studies corroborated the relationship between GI disturbances and combat or PTSD (a surrogate for trauma exposure) in veterans of the Vietnam War (Eisen et al. 1991), OIF (Hoge et al. 2007), and of World War II and the Korean War (Schnurr et al. 2000). The committee notes that a substantial number of physiologic studies support associations of acute and chronic stress with lowered pain-sensation threshold and with visceral hypersensitivity, increased motility, altered brain circuitry involving pain regulation, and altered HPA-axis reactivity (Chang 2006; Dunphy et al. 2003; Kellow et al. 2006a,b; Murray et al. 2004). It is possible that combat stress leads to central amplification of visceral and somatic symptoms via a variety of CNS-peripheral mechanisms, such as the HPA axis, pain-modulating circuits, and autonomic function (see Chapter 4). There are some limitations in this body of evidence, mostly related to methods of effect assessment. One is that the self-reporting of GI symptoms in most cases did not fulfill the criteria for diagnosing a functional GI disorder. None of the studies used contemporary criteria to assess functional GI disorders although in some cases the diagnoses can be inferred. There was also a lack of adequate medical diagnostic testing to identify a GI structural disease. Nevertheless, taken together, the overall pattern of symptoms found in many primary and secondary studies confirms an association of deployment-related stress and functional GI symptoms, including abdominal pain, diarrhea, nausea, and vomiting. Changes in GI functioning mark the acute stress response, and this is well supported in the physiologic literature. Furthermore, the findings suggest the development of deployment-related IBS, functional dyspepsia, and some other functional GI syndromes. The relationship provides a strong biologic rationale, as explained in Chapter 4, for the epidemiologic relationship found here between severe trauma exposure and long-term GI illness in veterans. The methods of the veteran studies are insufficient to determine a clear association between acute or chronic stress and the onset of a structural disorder (Drossman and Ringel 2004). Furthermore, the diagnosis of these diseases should be validated by medical records because physicians not infrequently place an organic label on a patientâs symptoms (for example, gastritis or peptic ulcer) without performing diagnostic studies, and this will confound the diagnosis in a survey, particularly if the data are based on the subjectsâ recollections of physiciansâ diagnoses. Thus, the association of deployment-related stress with GI symptoms is accepted, the association with functional GI disorders is supported but not complete, and an association with structural GI diseases cannot be determined. The committee concludes that there is limited but suggestive evidence of an association between deployment to a war zone and gastrointestinal symptoms consistent with functional gastrointestinal disorders, such as irritable bowel syndrome and functional dyspepsia. The committee also concludes that there is inadequate/insufficient evidence to determine whether an association exists between deployment to a war zone and the development of structural gastrointestinal diseases.
TABLE 6-9 Digestive System Disorders Study Study Design Population Outcomes Results Adjustments Comments Gray et al. 2002 Retrospective, U.S. Navy Seabees: Self-reported Deployed vs nondeployed: Age, sex, active-duty Study limited by case-control 3831 GW-deployed physician diagnoses, peptic ulcer disease OR 3.11, or reserve status, recall bias, IBS not veterans, 4933 self-reported 95% CI 1.67-5.78; IBS OR 3.57, race or ethnicity, analyzed exclusively, veterans deployed symptoms from 95% CI 2.22-5.73; new GI current smoking, response rate 70%, elsewhere, 3104 mailed questionnaire disease OR 2.10, 95% CI 1.39- current alcohol large sample nondeployed 3.17; clustering of CFS, PTSD, drinking veterans MCS, IBS: Seabees who had one had 13-18 other symptoms, Seabees without any averaged 6 other symptoms Sosteck et al. Cross- 57 male GW- Questionnaire about Prevalence of GI symptoms: Response rate 74%; 1996 sectional, deployed veterans, GI and non-GI abdominal pain 70% vs 9%; limitations include prevalence 44 nondeployed symptoms with recall diarrhea 74% vs 18%; small sample and veterans of National before, during, after incomplete rectal evacuation self-report Guard unit GW period 60% vs 7%; gas 74% vs 23%; questionnaire for decreased appetite 42% vs 7% recall before, during, (all p < 0.001) after GW Eisen et al. Cross- 1061 GW-deployed Physician evaluation, Dyspepsia OR 1.87, 95% CI Age, sex, race, Study limited by low 2005 sectional, veterans vs 1128 questionnaire for 1.16-2.99; gastritis OR 1.57, cigarette-smoking, participation rate, prevalence non-GW-deployed dyspepsia; GI 95% CI 0.88-2.78 duty type, service number of years NHSGWEVTF veterans symptoms and branch, rank since war; weak medical conditions diagnostic criteria (Derived from reported from earlier Kang et al. survey 2000b) 211
TABLE 6-9 Digestive System Disorders 212 Study Study Design Population Outcomes Results Adjustments Comments Kang et al. Cohort, 11,441 GW-deployed Self-reports of 48 Estimated prevalence: heartburn Mailed 2000b prevalence, veterans vs 9476 medical conditions 37% vs 25%; diarrhea 31% vs questionnaires conducted in nondeployed during preceding 12 15%; abdominal pain 23% vs followed by NHSGWEVTF 1995 veterans months; self-reports of 12%; nausea 19% vs 11%; telephone interviews exposure of veterans constipation 13% vs 7%; trouble in health registry; swallowing 12% vs 5%; colitis medical-record review 4.6% vs 2.2%; vomiting 10% vs of subset of 4200 6%; gastritis 25.2% vs 11.7%; veterans for clinic enteritis 6.6% vs 2.8%; frequent records, hospital diarrhea 21.2% vs 5.9% records (all significant at p â¤ 0.05) Eisen et al. Case-control, 2260 male Mail or telephone Twins who served in Vietnam PTSD; did not alter Response rate 75%; 1991 mail or monozygotic twins interviews of whether and been hospitalized for results potential for telephone from twin registry veteran had at time of stomach conditions OR 2.6, reporting bias; self- VET Registry survey who were on active interview, had since 95% CI 1.4-4.9 reports; comparing duty during Vietnam service, or had been monozygotic twins war hospitalized for 13 naturally adjusts for categories of health genetic problems; index of susceptibilities combat exposure Dumitrascu and Cohort study 60 civilians with IBS In-person interview on Painful IBS episodes high in Baban 1991 done at 4 from medical clinic bowel symptoms after January 1990, decreased in times in Cluj, Romania Romanian uprising in March, June 1990 (80%, 38%, December 1989 and in 33%, respectively), but diarrheal January, March, June episodes did not change (13%, 1990 12%, 10%, respectively) Stermer et al. Case-control 239 holocaust In-person interviews Compared prevalence of GI 1991 survivors vs 384 on medical history symptoms: abdominal pain, Eastern Europeans with emphasis on GI irregular bowels, diarrhea, not exposed to Nazis; symptoms constipation, distention, all born before 1941, heartburn, flatulence, anorexia, hospitalized in Israel nausea, vomiting, mucus, in 1985-1986 for tenesmus, aerophagia; all non-GI complaints significant (p < 0.001-0.0001)
TABLE 6-9 Digestive System Disorders Study Study Design Population Outcomes Results Adjustments Comments Boscarino 1997 Nested case- 332 PTSD-positive Lifetime PTSD Digestive diseases OR 1.47, Preservice, in- Large sample, control study Army Vietnam War evaluation by in- 95% CI 1.02-2.10 service, postservice generalizability to (Derived from in veterans with high- person interview with factors, including Army Vietnam VES) retrospective combat exposure, DIS; postservice self- intelligence, race, veterans; in-person cohort study 1067 PTSD-negative reported physician- region of birth, diagnoses of PTSD; conducted 17 controls mostly with diagnosed digestive enlistment status, lack of specificity as years after war low-combat exposure system diseases at in- volunteer status, to particular disease; person interviews; Army marital status, no verification of Combat Exposure Army medical health reporting Index profile, against medical hypochondriasis records Dobie et al. Cross- 266 female veterans Mailed questionnaire IBS OR 2.8, 95% CI 2.1-3.9 Age; no adjustment Response rate 65%; 2004 sectional with PTSD, 940 for health history, SF- for other self-administered veterans without 36, PCL-C (screening confounders questionnaire; PTSD; women had instrument) without criteria, IBS all received care at symptom reporting one VA medical may be part of center in 1996-1998 generalized somatic reporting tendency NOTE: CFS = chronic fatigue syndrome, CI = confidence interval, DIS = Diagnostic Interview Schedule, GI = gastrointestinal, GW = Gulf War, IBS = irritable bowel syndrome, MCS = multiple chemical sensitivity, NHSGWEVTF = National Health Survey of Gulf War Era Veterans and Their Families, OR = odds ratio; PCL-C = PTSD Checklist-Civilian, PMR = proportional morbidity ratio, PTSD = posttraumatic stress disorder, SF-36 = Short form 36, VA = Department of Veterans Affairs, VES = Vietnam Experience Study, VET Registry = Vietnam Era Twin Registry. 213
214 GULF WAR AND HEALTH SKIN DISORDERS Rashes were among the most frequently reported health problem soon after the Gulf War (Murphy et al. 1999). Rash usually refers to dermatitis, an umbrella term covering several subtypes, including atopic dermatitis, contact dermatitis, seborrheic dermatitis, and psoriasis. In both the Gulf War and the Vietnam War, troops were exposed to several toxicants that could cause allergic skin reactions, including pesticides; Agent Orange has been associated with the chloracne seen in some Vietnam veterans. For the purposes of this section on deployment-related stress and dermatologic effects, the committee defined a primary study according to methodologic rigor (Chapter 2) and use of a dermatologic examination. In a secondary study, the determination of a dermatologic effect was based on veteransâ self-reports of symptoms or self-reported physician-diagnosed conditions. This section excludes one skin condition, chloracne, because its cause in veterans was herbicide exposure peculiar to the Vietnam War (IOM 2007). Primary studies are summarized in Table 6-10. Primary Studies To determine the link between deployment-related stressors and dermatologic diseases, the committee identified four primary studies (CDC 1988b; Eisen et al. 2005; Higgins et al. 2002; Ishoy et al. 1999) and numerous secondary studies. In the large, nationally representative study of U.S. Gulf War veterans, Eisen et al. (2005) performed medical evaluations of more than 2,000 veterans. The study was performed 10 years after the Gulf War and used data derived from the 1995 VA National Health Survey of Gulf War Era Veterans and Their Families (Kang et al. 2000b). The investigators searched for dermatologic conditions by dividing them into two categories: group 1 consisted of freckles, seborrheic keratoses, moles, cherry hemangiomas, skin tags, and scars; group 2 consisted of dermatologic diagnoses not included in group 1. Diagnoses were made by a board-certified dermatologist, who evaluated group 2 conditions through teledermatology by using at least two digital photographs and results of a standardized history and physical examination. The prevalence of group 1 diagnoses did not differ between deployed and nondeployed veterans (OR 0.87, 95% CI 0.68-1.12). After adjustment for age, sex, race, years of education, cigarette-smoking, duty type, service branch, and rank, the prevalence of a diagnosis of one or more group 2 skin conditions was 34.6% in deployed veterans and 26.8% in nondeployed veterans (OR 1.38, 95% CI 1.06-1.80). Two skin conditions in group 2 were diagnosed more frequently (p = 0.02) in deployed than in nondeployed veterans: verruca vulgaris (warts) (1.6% vs 0.6%, OR 4.02, 95% CI 1.28-12.6) and atopic dermatitis (1.2% vs 0.3%, OR 8.1, 95% CI 2.4-27.7). Atopic dermatitis (a form of eczema) is an inflammatory condition manifested on the skin by dry, eczematous skin and papules. It tends to run in families (with allergic rhinitis and other allergies), and its course is often chronically relapsing (Leung 2000). Atopic dermatitis is not to be confused with contact dermatitis, a delayed hypersensitivity response to a specific agent that directly or indirectly injures the skin. Researchers in the UK conducted in-person dermatologic evaluations of UK Gulf War- deployed veterans (111 disabled and 98 nondisabled) and 133 disabled veterans not deployed to the Gulf War (Higgins et al. 2002). The cross-sectional study was conducted in 1999-2000; participants were randomly selected from three representative cohorts that had served in the gulf
HEALTH EFFECTS 215 or in Bosnia or had been on active duty but not deployed to either location. Disability was defined as reduced physical functioning according to the SF-36. All participants were examined by a dermatologist who was blinded as to deployment and health status. The prevalence of any skin condition was 47.7% in disabled Gulf War veterans, 36.7% in nondisabled Gulf War veterans, and 42.8% in disabled non-Gulf War veterans. The investigators found no differences among groups in any dermatologic conditions other than seborrheic dermatitis (8.1% in deployed vs 2.3% in nondeployed), which was more common in Gulf War veterans than in the two comparison groups. The prevalence of seborrheic dermatitis in the general UK population is estimated to be about 3%. The excess in deployed veterans was irrespective of disability status. Ishoy et al. (1999) reported on Danish peacekeepers deployed to the gulf during 1990- 1997. The 686 deployed veterans and 231 nondeployed age-, sex-, and profession-matched veterans each received a medical examination and were interviewed for a full medical history by a physician. Veterans indicated whether any condition had its onset before or after deployment to the gulf. The examinations found that the prevalence of the following conditions with onset during or after deployment or August 1990 was higher in deployed veterans than in nondeployed veterans: eczema (15.0% vs 3.0%, p < 0.001), retarded wound healing (6.0% vs 1.7%, p < 0.01), other skin problems (17.1% vs 5.2%, p < 0.001), hair loss or hair disease (4.2% vs 0.9%, p < 0.01), and sweaty, clammy, or damp hands (7.9% vs 3.9%, p < 0.05). The study limitations included performance of only multiple univariate observations and a lack of information on possible confounders, although such lifestyle factors as smoking, alcohol use, and physical activity were reported. The VES was a major study of all U.S. Army veterans conducted 17 years after the Vietnam War (CDC 1988b). It surveyed a random sample of the 5 million Army veterans who served during the Vietnam era: 2490 Vietnam-theater veterans and 1972 era veterans were interviewed and examined in 1985-1986. During the medical-examination phase of the study, a significantly higher prevalence of skin conditions as a global category (excluding chloracne) was found in theater veterans than in era veterans (33.0% vs 21.8%, OR 1.7, 95% CI excludes 1.0, p < 0.05). Upon dermatologic examination, none of the selected dermatologic conditions were found to be more prevalent in theater veterans than in era veterans; ORs for hyperpigmentation, hirsutism, folliculitis, tinea, any skin infection, and postinflammatory scars were all 0.9-1.2 (95% CIs 0.8-1.7). Secondary Studies The committee identified numerous large well-designed studies of Gulf War veterans that it considered secondary studies primarily because they lacked a dermatologic examination or were imprecise regarding specific dermatologic disorders. Most secondary studies found higher prevalence of dermatologic conditions in deployed vs nondeployed veterans. In the first phase of the VA National Health Survey of Gulf War Era Veterans and Their Families, Kang et al. (2000b) used a stratified random-sampling method to compare 11,441 Gulf War-deployed veterans with 9,476 nondeployed veterans on the basis of the DMDC. The study found that dermatologic conditions were in the top five self-reported medical conditions diagnosed by physicians in the preceding 12 months. The skin conditions listed were eczema or psoriasis (7.7% vs 4.4%, rate difference 3.34, 95% CI 3.26-3.42), other dermatitis (25.1% vs 12.0%, rate difference 13.16, 95% CI 13.04-13.28), and diseases of the hair or scalp or hair loss (16.9% vs 7.2, rate difference 9.65, 95% CI 9.55-9.75). A sample of participants were later evaluated by clinical examination in the Eisen et al. study (2005), which was the studyâs final phase.
216 GULF WAR AND HEALTH Other large cohort studies conducted in several countries reported similar findings in Gulf War veterans based on self-reports via questionnaires. The population-based survey of UK Gulf War-deployed veterans found the prevalence of dermatitis to be 21%, a rate higher than that in two control groups: one dispatched to Bosnia (OR 1.6, 95% CI 1.3-2.0) and the other era controls (OR 1.6, 95% CI 1.4-1.9) (Unwin et al. 1999). Simmons et al. (2004) found that male UK Gulf War veterans reported more skin allergies with onset after 1991 than their nondeployed counterparts (OR 3.3, 95% CI 3.0-3.7). Proctor et al. (1998) compared the prevalence of dermatologic conditionsâsuch as rashes, eczema, and skin allergiesâin 252 U.S. Gulf War- deployed veterans. The estimated prevalence was 15.5% for the 186 veterans from the Fort Devens cohort, 11.7% for the 66 veterans from the New Orleans cohort, and 1.9% for the 48 veterans deployed to Germany during the Gulf War. A higher prevalence of skin conditions was also reported in Australian population-based studies (Kelsall et al. 2004a). Moderate to severe rash and skin irritation was reported at a higher rate in 1456 deployed veterans than in 1588 nondeployed veterans (OR 2.0, 95% CI 1.6-2.5). On the basis of a diagnosis by a physician after 1991 (deemed by the investigators to be a probable diagnosis), deployed veterans reported more dermatitis (OR 1.8, 95% CI 1.3-2.6) and skin diseases other than dermatitis, skin cancer, eczema, or psoriasis (OR 1.3, 95% CI 1.1-1.7) compared with nondeployed veterans. In light of their reliance on self-reports or questionnaires, all those studies are susceptible to recall or reporting bias. They also lack specificity regarding specific skin effects. Eisen et al. (1991) performed a co-twin study in 1987 to determine the health effects of the Vietnam War. One member of each of 2260 male monozygotic twin pairs reported having a health problem and the other did not; it was then determined how many of the affected twins and unaffected twins had served in Vietnam. The study established the validity of self-reported health responses against medical records for a subset of the twins. It also searched for relationships between health problems and levels of combat exposure. The investigators constructed a combat exposure index by using responses to a separate questionnaire. Twins with persistent skin conditions (such as severe acne and rashes) that were present at any time since service were more likely to have served in Vietnam than were unaffected twins (OR 2.1, 95% CI 1.5-3.0). The prevalence of skin problems (either currently or ever) was significantly associated with increasing levels of combat exposure. At the highest level of combat exposure, skin problems were 3 times more likely than at the lowest combat level. The study is limited in that skin problems included âsevere acne,â which may have included chloracne associated with Agent Orange, although the VES showed its prevalence to be lower (1.9% in theater veterans) than that of other skin conditions (33% in theater veterans) upon examination (CDC 1988b). A study of Australian veterans deployed to the Vietnam War (a simple random sample of Army veterans, n = 641) found that combat exposure, as determined by Army records and a self- reported combat exposure scale, was related to self-reported chronic ârashâ in an increasing linear relationship (p = 0.041) (OâToole et al. 1996b). There was nearly a linear dose-response relationship for eczema, but it did not reach significance (p = 0.062). The study did not have dermatologic examinations, but trained interviewers did ask questions about chronic health conditions in person through the standardized Australian Bureau of Statistics Health Interview Survey 1989-1990. Posttraumatic Stress Disorder and Skin Disorders Gulf War veterans with PTSD display higher rates of dermatologic symptoms or conditions, according to two studies, one of Gulf War veterans and one of Vietnam veterans. The
HEALTH EFFECTS 217 studies were questionnaire-based, used self-reports of symptoms or physician-diagnosed conditions, and used veterans without PTSD as the comparison population. The assessment of PTSD, which was similar in the two studies, was based on a screening questionnaire, the PTSD Checklist, with standard cutoffs. In a nested case-control study drawn from the population-based study of Iowa veterans of the Gulf War, Barrett et al. (2002a) found the prevalence of current dermatologic conditions and dermatologic symptoms to be higher in 53 veterans with current PTSD (over 90%) than in 3629 veterans without PTSD (about 25%). The study was conducted by telephone interview 5 years after the Gulf War. This nested case-control component of the larger cohort study conducted by the Iowa Persian Gulf Study Group (1997) is limited by lack of specification in the analysis of which particular dermatologic symptoms or conditions might be more prevalent. The Veterans Health Study of 2425 consecutive male ambulatory-care patients seen at four VA medical centers screened veterans for PTSD with the PTSD Checklist. Patients who screened positive (n = 456) reported higher rates of dermatitis than those without PTSD (OR 2.37, 95% CI 1.88-3.0) (Spiro et al. 2006). The study is limited, however, by its lack of a control population, lack of reporting of veteransâ previous deployment status, and absence of identification as to whether PTSD was related to a deployment-related trauma. Another case-control study was nested within the VES. It sought to determine whether combat-related PTSD was associated with development of a wide variety of medical disorders (Boscarino 1997). Case subjects were PTSD-positive Vietnam veterans with high combat exposure, and controls were PTSD-negative veterans with low combat exposure. Lifetime PTSD status was determined with in-person interviews that used the DIS, and combat exposure was determined with the Laufer Combat Exposure Scale. Medical disorders were ascertained during the in-person interviews by asking veterans to respond affirmatively or negatively to a list of specific medical conditions as to whether they had been diagnosed by a physician and whether the condition occurred before or after discharge. The conditions were later collapsed into general medical categories to minimize the likelihood of reporting bias. The study found no excess of dermatologic diseases occurring after discharge in veterans with PTSD vs those without PTSD (OR 1.09, 95% CI 0.59-1.04) after adjustment for general technical test results, race, region of birth, type of enlistment, Vietnam volunteer status, marital status, age, hypochondriasis, physical limitations in the military record, and psychiatric limitations in the military record. Key advantages of the study are its focus on combat-related PTSD, its ascertainment of PTSD with the DIS, a well-validated structured psychiatric interview based on DSM criteria, and use of a combat exposure scale. Another study strength is its adjustment for a host of preservice and postservice behavioral risk factors and physical or psychiatric limitations reported in the military record at the time of service. The major study limitation is the lack of specificity regarding particular dermatologic diseases. In a later case-control analysis of the VES, Boscarino (2004) tested the hypothesis that combat-related PTSD increases the risk of autoimmune disease after discharge. The prevalence of 20 autoimmune diseases was examined in 124 Vietnam-theater veterans with comorbid PTSD and 54 veterans with current but noncomorbid PTSD (less severe PTSD according to the author). Psoriasis was the one of nine autoimmune diseases with dermatologic manifestations. After multivariate analysisâadjusted for age, education, race, intelligence, geographic region, Army volunteer status, number of times married, smoking, and history of psychiatric disorderâ psoriasis was found to be nearly 5 times more prevalent in veterans with comorbid PTSD than in veterans without PTSD (OR 4.7, 95% CI 1.9-11.7).
218 GULF WAR AND HEALTH Finally, older male World War II and Korean War veterans participating in the Normative Aging Study who screened positive for lifetime PTSD (n > 580) using the Mississippi Scale for Combat-Related PTSD were more likely to have new onset of dermatologic disorders (hazard rate 1.18, 95% CI 1.05-1.34) according to physical examinations conducted periodically beginning in the 1960s, than veterans without PTSD (Schnurr et al. 2000). Dermatologic diseases (ICD 690-698) included eczema, dermatitis, and psoriasis. Dermatologic disorders increased by 18% for every 10-point increment in PTSD symptom scores. Summary and Conclusions The committee placed the greatest weight on studies that included medical evaluation and identification of specific dermatologic diagnoses. All four primary studies showed a higher prevalence of some skin diseases or conditions in deployed than in nondeployed veterans. A nationally representative study of Gulf War veterans found a relationship between deployment and atopic dermatitis and verruca vulgaris (warts), but not other skin conditions (Eisen et al. 2005). A UK study of Gulf War veterans found a relationship between deployment and seborrheic dermatitis (Higgins et al. 2002). Each of those dermatologic associations is supported by a single primary study, although Ishoy et al. (1999) also found an increased prevalence of eczema in deployed Gulf War veterans. The VES showed no increase in selected skin conditions not related to exposure to Agent Orange in Vietnam-theater veterans (CDC 1988b). Secondary studies are largely consistent with the primary studies but lack specificity regarding dermatologic outcomes. Many report higher prevalences of self-reported symptoms or physician-diagnosed dermatologic conditions in deployed than in nondeployed veterans. Some secondary studies are somewhat more specific in reporting a greater prevalence of eczema or psoriasis (Kang et al. 2000b) or eczema alone (Wolfe et al. 1998). Dermatologic conditions are highly common in veterans regardless of deployment status. Verruca vulgaris is known to be caused by a virus, and seborrheic dermatitis is thought to be caused by a combination of sebaceous gland secretions, microfloral metabolism, and individual susceptibility (Ro and Dawson 2005). Therefore, the higher prevalence of those two conditions in deployed than in nondeployed veterans are more likely related to environmental conditions in the Gulf War or may simply be due to chance. The other two conditionsâatopic dermatitis and psoriasisâhave more plausible relationships with deployment stressors. Both are mediated by the immune system, which has strong interrelationships with the stress response. In the case of psoriasis, the epidemiologic evidence comes directly from one of the primary studies implicating combat-related PTSD of high severity in the onset of psoriasis (Boscarino 2004). Case subjects not only had a higher prevalence of other autoimmune diseases but displayed five biologic markers of autoimmune and other inflammatory diseases. Other controlled studies of psoriasis patients found that introduction of psychologic stressors is associated with biologic alterations in the stress response (Buske-Kirschbaum et al. 2007). In the case of atopic dermatitis, a large body of research over the last decade suggests immune dysregulation in its pathophysiology and psychologic stress as a key trigger in its maintenance or exacerbation (Leung and Soter 2001). Controlled clinical trials found that patients with atopic dermatitis show attenuated activity of the HPA axis and overreactivity of the sympathetic adrenomedullary system in response to psychologic stressors (Buske-Kirschbaum et al. 2002). Those findings are consistent with the findings from veteran studies. Skin conditions (as a broad category) were directly linked to deployment stressors via a combat exposure index
HEALTH EFFECTS 219 in two secondary studies (Eisen et al. 1991; O'Toole et al. 1996b). In addition, the rate of incident skin conditionsâincluding dermatitis, psoriasis, and eczemaâhad a dose-response relationship with PTSD symptom levels in a secondary study of Korean War and World War II veterans (Schnurr et al. 2000). In summary, there is a high frequency of self-reports of various types of rash and other skin conditions among deployed vs nondeployed veterans, and, in general, these reports are confirmed by dermatologic examination, particularly for eczema. Overall, very few studies have rigorously assessed the prevalence of skin conditions in Vietnam War and Gulf War veterans and results are mixed with increases for some skin conditions but not for others. Most are weak in design and limited by self-selection and possible reporting bias. (Skin cancer was discussed earlier in the section âCancer.â) The committee concludes that there is limited but suggestive evidence of an association between deployment to a war zone and skin disorders.
220 TABLE 6-10 Skin Disorders Reference Study Design Population Outcomes Results Adjustments Limitations/Comments Eisen et al. Population-based, 1061 GW- Full-body skin Atopic dermatitis OR Age, sex, race, years of Low participation rates 2005 cross-sectional, deployed vs 1128 examination by 8.1, 95% CI 2.4-27.7; education, smoking, duty 53% deployed, 39% prevalence, nondeployed physician; verruca vulgaris OR type, service branch, rank nondeployed; study NHSGWEVTF medical evaluation dermatologist used 4.02, 95% CI 1.28-12.6; conducted 10 years after teledermatology any group 2 war (Derived from for group 2 diagnosis OR 1.38, 95% Kang et al. diagnoses CI 1.06-1.80; any group 2000b) 1 diagnosis OR 0.87, 95% CI 1.06-1.80 Higgins et al. Prospective case- 111 UK disabled Skin examination Any skin condition Age, sex, rank, smoking, Skin disorders are 2002 comparison, cross- GW veterans, 98 by dermatologist 47.7% vs 36.7% vs alcohol common in veterans; sectional nondisabled GW blind to service 42.8%; seborrheic finding could have been prevalence veterans, 133 history, health dermatitis 8.1% by chance; study disabled non-GW status deployed vs 2.3% conducted 9-10 years veterans; samples nondeployed (disabled after war derived from larger and nondisabled) cohort study of UK veterans of GW, Bosnia peacekeepers, active-duty but nondeployed veterans Ishoy et al. Cross-sectional, 686 Danish Health Prevalence of skin Participation rate 83.6% 1999 prevalence peacekeepers examination by conditions with onset deployed, 57.7% deployed to gulf in physician, self- after gulf: nondeployed; lack of 1990-1997 vs 231 report eczema 15.0% vs 3.0%, information on age- and sex- questionnaire p < 0.001; retarded adjustment for matched armed wound healing 6.0% vs confounders in forces nondeployed 1.7%, p < 0.01; multivariate analysis controls other forms of skin problems 17.1% vs 5.2%, p < 0.001; hair loss or hair disease
TABLE 6-10 Skin Disorders Reference Study Design Population Outcomes Results Adjustments Limitations/Comments 4.2% vs 0.9%, p < 0.01; sweaty, clammy, or damp hands 7.9% vs 3.9%, p < 0.05 CDC 1988b Retrospective 2490 Vietnam- Dermatologic Other skin conditions Age at enlistment, race, Low participation rate in cohort, prevalence, theater veterans, examinations, OR 1.7, 95% CI year of enlistment, control group; conducted VES population-based, 1972 Vietnam-era laboratory testing excludes 1.0, p < 0.05; enlistment status 17 years after war telephone veterans randomly at medical hyperpigmentation OR (volunteer vs draftee), interview with selected from 7924 examination 1.2, 95% CI 0.9-1.7; score on general technical screening medical theater veterans, folliculitis OR 0.9, 95% test, primary military examination at 7364 era veterans CI 0.8-1.1; tinea OR occupational specialty followup who had entered 1.0, 95% CI 0.8-1.1; Army in 1965- any skin infection OR 1971 1.0, 95% CI 0.8-1.1; postinflammatory scars OR 1.0, 95% CI 0.9-1.2 NOTE: CI = confidence interval, GW = Gulf War, NHSGWEVTF = National Health Survey of Gulf War Era Veterans and Their Families, OR = odds ratio, VES = Vietnam Experience Study. 221
222 GULF WAR AND HEALTH FIBROMYALGIA AND CHRONIC WIDESPREAD PAIN Fibromyalgia, also called fibrositis, is a chronic and variable rheumatic condition characterized by widespread muscle and skeletal tenderness and fatigue. Diagnosis is based on two criteria of the American College of Rheumatology (ACR): a history of widespread pain lasting at least 3 months and the presence of 11 or more of 18 designated tender points on the body. Other nonspecific symptoms of fibromyalgia are sleep disturbance, morning stiffness, and cognitive impairment. The ACR has developed a definition of chronic widespread pain (CWP) as part of the diagnostic criteria for fibromyalgia: CWP is pain reported in the axial skeleton and two contralateral quadrants that persists for 3 months or longer, that is both right and left sides of the body and both above and below the waist. One of the predesignated pain sites is considered a true tender point only if the person feels pain on application of 4 kg of pressure to the site. There are no laboratory tests to diagnose fibromyalgia or CWP, and their etiology is unknown. It has been noted that fibromyalgia cannot be diagnosed without a physical examination (Buskila 2000) and it has been suggested that CWP and fibromyalgia are points on a continuum of chronic pain disorders (Kuzma and Black 2006). Chronic pain, as distinct from CWP, is discussed in the final section of this chapter, âSymptom Reporting.â Fibromyalgia has a prevalence in the general population of about 2.0%â3.4% in women and 0.5% in menâand its prevalence increases with age (Johnson 1989). As many as 4 million Americans, mostly women, may have fibromyalgia (American Pain Foundation 2007). CWP has been estimated to occur in 5% of patients seen in general internal-medicine practice and up to 20% of rheumatology-clinic patients (Wolfe 1989). In the United Kingdom, 10-11% of the general population reported having symptoms consistent with the ACR definition of CWP (Papageorgiou et al. 2002). Primary studies of fibromyalgia include a diagnosis based on symptom reporting and physical examination, preferably using ACR criteria. Primary studies of CWP also needed to include a diagnosis based on ACR criteria. Self-reports, even those using pain manikins (a drawing of a person on which areas of pain can be identified), were not considered to be sufficient for a primary study of either condition. The primary studies for fibromyalgia and CWP are summarized in Table 6-11. Primary Studies The committee identified three primary studies: two of fibromyalgia (Eisen et al. 2005; Smith et al. 2000) and one of CWP (Ang et al. 2006). In 1999-2001, Eisen et al. (2005) reported on the prevalence of 12 medical conditions in 1061 Gulf War-deployed veterans and 1128 nondeployed veterans who were randomly selected from a national cohort of 11,441 deployed and 9476 nondeployed veterans in all service branches. The veterans were part of the National Health Survey of Gulf War Era Veterans and Their Families, whose first phase was a mail and telephone interview conducted in 1995 (Kang et al. 2000b). Researchers were blinded to deployment status, and combat exposure was not assessed. On the basis of self-reports using the SF-36, the prevalence of fibromyalgia in deployed and nondeployed veterans was 0.6% and 0.8%, respectively, for a nonsignificant OR of 1.21 (95% CI 0.36-4.10). However, when fibromyalgia was diagnosed by physical examination following the ACR criteria (Wolfe et al. 1990), the prevalence was 2% in deployed veterans and 1.2% in nondeployed for a significant
HEALTH EFFECTS 223 OR of 2.32 (95% CI 1.02-5.27, p = 0.04) adjusted for age, sex, race, cigarette-smoking, duty type, service branch, and rank. Deployed veterans were slightly younger, less educated, less likely to be married, and of lower income, but the analysis adjusted for most of those factors. Strengths of the study include the population-based sampling strategy, blinding of evaluating physicians, and use of validated diagnostic criteria based on physical examination. Limitations include the potential for substantial selection bias due to modest participation ratesâ53% of deployed veterans and 39% of nondeployed veteransâand lack of information on potential exposures. Using a different approach, Smith et al. (2000) examined all hospital records in DoD medical facilities from 1991 to 1997 to assess whether 551,841 Gulf War-deployed and 1,478,704 nondeployed active-duty military personnel were at increased risk for hospitalization with a diagnosis of fibromyalgia. During the 6-year period, 239 Gulf War veterans and 621 nondeployed veterans were hospitalized for fibromyalgia. Cox proportional-hazards models showed a slightly higher hospitalization rate for fibromyalgia among the deployed than among the nondeployed veterans (RR 1.23, 95% CI 1.05-1.43). The risk of hospitalization was 3 times greater for female veterans than for male veterans, twice as likely if the person had been hospitalized for any cause before the war, and more than twice as likely for Army veterans than veterans in other service branches. The authors attributed the increase to the DoD Comprehensive Clinical Evaluation Program (CCEP)âwhich began in June 1994 and ended in mid-1995âduring which many veterans were admitted to the hospital only for purposes of evaluation. Before the inception of the CCEP, there was no difference in hospitalization for fibromyalgia between deployed and nondeployed veterans (RR 0.92, 95% CI 0.74-1.13); after CCEP in 1994, there was almost twice the rate of hospitalization for fibromyalgia (RR 1.76, 95% CI 1.39-2.22). CCEP participants were 26 times more likely to be hospitalized for fibromyalgia than were nonparticipants, and adding a CCEP covariate to the analysis resulted in a reduction in the risk for deployed veterans (RR 0.56, 95% CI 0.41-0.78). The Smith et al. study has the advantage of being a large population-based sample and having good statistical power for detection of an effect. Its major limitations are the inclusion of only active-duty personnel, changes in hospitalization rates for fibromyalgia in association with the practices of the CCEP, use of hospitalization data only from DoD medical facilities, lack of assessment of combat exposure or deployment stress, and the fact that fibromyalgia is rarely severe enough to warrant hospitalization. The committee identified only one primary paper that looked specifically at CWP in deployed and nondeployed Gulf War veterans. A random sample of a population-based cohort of regular military and National Guard and reserve veterans (Iowa Persian Gulf Study Group 1997), 1896 deployed and 1799 nondeployed, who listed Iowa as their home state at the time of enlistment were surveyed in 1995-1996. Veterans were identified through the DMDC. The study was conducted through structured telephone interviews to determine the prevalence of CWP on the basis of responses to the SF-36. Gulf War veterans reported significantly more bodily pain than did nondeployed veterans (p < 0.01). In a followup study of a subset of this cohort 5 years after the baseline survey, Ang et al. (2006) conducted in-person followup examinations of 370 Gulf War veterans who had not met the case definition of CWP at baseline. The goal of the followup study was to identify predictors of delayed-onset CWP. Of the 370 veterans, 69 (18.6%) had met the classification criteria for CWP at the followup evaluation: 51 in the deployed group and 18 in the nondeployed group. According to a logistic multiple-regression model, CWP was significantly associated with perceived life stress (based on responses to the
224 GULF WAR AND HEALTH Brief Life Stress Questionnaire) at the time of the Gulf War, whether military-related or not (OR 1.4, 95% CI 1.0-2.0), and with perceived life stress in the 6 months after returning home (OR 1.3, 95% CI 1.0-1.8). CWP also correlated with combat exposure during deployment (OR 1.5, 95% CI 1.1-2.0) although not specifically with deployment to the gulf itself (OR 1.1, 95% CI 0.6-2.0). Symptoms of alcohol use at the 5-year baseline survey were protective for CWP at 10 years (OR 0.2, 95% CI 0.1-0.6, p = 0.0039). The authors used the Expanded Combat Exposure Scale in the baseline survey and reported that for every 5-point increase in combat exposure score, there was a 50% increase in the likelihood that a veteran would develop CWP. Although the study had the advantage of using an in-person evaluation for the medical diagnosis of CWP and had a relatively large population of deployed and nondeployed veterans, there was a possibility of recall bias for life and deployment stressors reported 5 years after the conflict, and only veterans from Iowa were evaluated. Furthermore, only veterans who did not meet the CWP criteria at baseline were considered for the followup evaluation; veterans who may have developed CWP during the first 5 years after the conflict were not included in the followup examination. Secondary Studies Several studies have looked at self-reports of fibromyalgia, symptoms of fibromyalgia, or CWP in Gulf War veterans. The Iowa Persian Gulf Study Group (1997) surveyed 1896 Gulf War-deployed and 1799 nondeployed veterans who listed Iowa as their home state at the time of enlistment to determine whether there was a unique Gulf War illness. In telephone interviews, veterans were asked about symptoms of fibromyalgia on the basis of questions keyed to the symptom criteria of Wolfe et al. (1990), which included the presence of widespread pain for at least 3 months. Symptoms of fibromyalgia were present in 18.2% of 985 deployed regular military veterans and 23.8% of 911 deployed National Guard or reserve veterans compared with 9.2% of 968 nondeployed regular military veterans and 13.2% of 831 nondeployed National Guard or reserve veterans. The authors found a statistically significant Cochran-Mantel-Haenszel rate difference of 9.3 (95% CI 7.3-11.2) after adjustment for age, sex, race, branch of military, and rank. Although the study used a large population-based sample, the determination of fibromyalgia was based solely on self-reported symptoms, not a physical examination. Participants were asked about a number of military exposures during deployment, including chemical and psychologic stressors, although combat was not specifically mentioned. The study had a high response rate: interviews were completed with 78% of the eligible deployed veterans and 73% of the eligible nondeployed veterans. In a similar study conducted by Steele (2000), 1545 deployed and 435 nondeployed Gulf War veterans who were residents of Kansas in 1998 were asked about their health status in telephone interviews. Specifically, they were asked whether they had ever received a physicianâs diagnosis of or treatment for fibromyalgia and, if so, when it had developed. Of the deployed and nondeployed veterans, 2% (n = 24) and less than 0.5% (n = 2), respectively, reported having a diagnosis of fibromyalgia with new onset between 1990 and 1998 (OR 3.69, 95% CI 0.86-15.84 adjusted for sex, age, income, and education level). Although the study included a population- based sample, its goal was to ascertain the prevalence of symptoms that might be indicative of Gulf War illness and the circumstances that increased their prevalence. No physical examinations were conducted to confirm a diagnosis of fibromyalgia or to link it to specific exposures.
HEALTH EFFECTS 225 A 1997 mail survey of the entire cohort of Canadian Gulf War-deployed veterans and a comparison group of nondeployed veteransâgroup-matched to cases on sex, age, and regular or reserve statusâfound that the Gulf War veterans were more likely (16%) to report symptoms of fibromyalgia than nondeployed veterans (about 10%) (Goss Gilroy Inc. 1998). Veterans were also asked about exposure to psychologic stressors, physical trauma, ionizing radiation, and other CNS agents. After adjustment for income, service branch, rank, and age, an OR of 1.81 (95% CI 1.55-2.13) was derived from the prevalence of fibromyalgia in deployed and nondeployed veterans. However, among deployed veterans, exposure to psychologic stressors and physical trauma also increased the likelihood of symptoms of fibromyalgia (OR 1.86, 95% CI 1.47-2.37 and OR 1.64, 95% CI 1.28-2.11, respectively), as did reported exposure to CNS agents and radiation and having lower rank and income. In both deployed and nondeployed veterans, the prevalence of fibromyalgia symptoms did not vary according to whether the veteran had other theater experience in the preceding 12 years. The main study limitation was diagnosis of fibromyalgia based on symptoms or self-reported diagnosis without physical examination. Using data from the Iowa Persian Gulf Study Group (1997), Forman-Hoffman et al. (2007) analyzed information from the structured telephone interview conducted with 1896 deployed regular military, National Guard, and reserves and 1799 nondeployed veterans in 1995- 1996. They looked for self-reports of symptoms of CWP according to the following criteria: the veteran reported having fibromyalgia or fibrositis in the previous 12 months or reported overall body pain that occurred almost every day for at least 3 months during the previous 12 months and had body pain in the 24 hours before the interview. The deployed veterans reported significantly more symptoms of CWP than did nondeployed veterans (OR 2.03, 95% CI 1.60- 2.58), whether in the regular military or in the National Guard or reserves; the OR was adjusted for age, sex, race, rank, branch of service, military status, smoking, and current income. Veterans who met the researchersâ definition of CWP also had more impairment, such as bed days and VA disability and compensation, and they were more likely to be unemployed, to report their health status as fair or poor, and to have greater use of health care than those without CWP. Stimpson et al. (2006) surveyed UK veterans who had served only in the Gulf War (n = 2959), only in Bosnia (n = 2052), or both in the Gulf War and in Bosnia (n = 570) and a comparison era group of veterans who had not been deployed to either the Gulf War or Bosnia (n = 2614) for self-reports of CWP. A mailed questionnaire containing a pain manikin to ascertain the pattern and intensity of pain was sent to 12,592 male and female veterans in 1997; the response rate was 60-70%. Veterans were selected for each deployment group on the basis of stratified random sampling of all UK Gulf War veterans. Data from the shaded manikins were used to determine whether the pain pattern met the ACR definition of CWP. The prevalence of reporting of CWP in the Gulf War-deployed group (16.8%) and the Gulf War- and Bosnia- deployed group (15.8%), but not in the Bosnia only-deployed group (7.6%), was statistically significantly higher (p < 0.0001) than that in the era group (8.5%) even when adjusted for socioeconomic and demographic factors. The pattern of pain was similar in all the groups; the most common sites of pain were the back and knees. Veterans who reported pain in one limb were also 30 times more likely to report pain in the symmetrically opposite limb rather than a second limb on the same side of the body; the authors found this suggestive of âsystemic painâ rather than pain from an injury. Although the sample was large, the study is limited by a lack of physical examination and a lack of indication as to whether the veterans had sustained injuries during deployment or were using pain medication at the time of the survey.
226 GULF WAR AND HEALTH A similar study by Cherry et al. (2001a,b) 6-8 years after the Gulf War also used a pain manikin to identify whether and where veterans had experienced pain for at least 24 hours in the preceding month. Among the 9588 male and female UK Gulf War veterans in all service branches, 12.2% reported widespread pain on a manikin compared with 6.5% of 4790 nondeployed veterans; widespread pain was considered to be present if the manikin showed axial skeletal and contralateral body pain. CWP was not associated with exposure to combat although it was associated with other deployment exposures, specifically to insect repellent, medical attention, and side effects of nerve-agent prophylaxis. The study had response rates of 93% of the active-duty military and 80% of those who had left the military. Posttraumatic Stress Disorder, Fibromyalgia, and Chronic Widespread Pain As noted earlier, PTSD is highly comorbid with other health problems, particularly chronic pain and psychiatric disorders. Some studies have assessed whether PTSD is comorbid with fibromyalgia and CWP. Two studies that looked at the comorbidity of PTSD and fibromyalgia were identified. Amital et al. (2006) found that 49% of 55 male Israeli veterans suffering from combat- related current PTSD met the ACR criteria for fibromyalgia, whereas only 5% of 20 veterans with MDD had fibromyalgia; none of 49 healthy control veterans had fibromyalgia. The PTSD veterans also had more tenderness points (assessed by a rheumatologist) than the MDD veterans (mean, 8.85 vs 2.85). Patients who had both PTSD and fibromyalgia also had more severe PTSD symptoms, primarily re-experiencing symptoms, and scored significantly higher on the CAPS than did veterans without fibromyalgia (97.6 Â± 13.2 vs 88.2 Â± 14.0). The traumatic event that preceded PTSD had occurred during military service 22-36 years before the study. Fibromyalgia and PTSD are more prevalent in women than in men. Dobie et al. (2004) sent a mail survey to 1206 female Gulf War veterans who received care at VA medical centers in 1996-1998. The presence of PTSD was determined with the PTSD Checklist-Civilian Version, and fibromyalgia was self-reported. Of the 1206 women, 266 (22%) screened positive for current PTSD; these 266 were more likely to screen positive for fibromyalgia (19.2% of the 266) than were the 940 women without PTSD (8.0%) for an age-adjusted OR of 3.00 (95% CI 1.98-4.45). This cross-sectional survey has several limitations, including lack of identification of the trauma associated with the PTSD, use of a treatment-seeking population, lack of diagnostic examinations for either PTSD or fibromyalgia, and reporting bias. The committee identified only one study that looked at CWP in veterans with PTSD. Ang et al. (2006) found that CWP was associated with symptoms of PTSD in Gulf War veterans (OR 4.5, 95% CI 0.6-32.3, p = 0.1383) more than in nondeployed veterans. Because the veterans were screened for current PTSD with the PTSD Checklist-Military Version, this is a secondary study. Summary and Conclusions The diagnosis of fibromyalgia is based on meeting the ACR criteria in a physical examination for tender points and the presence of CWP. The committee identified only two primary studies that diagnosed fibromyalgia on the basis of the ACR criteria and included both Gulf War-deployed and nondeployed veterans: Eisen et al. (2005) and Smith et al. (2000). The committee was unable to locate any studies of fibromyalgia in Vietnam veterans or veterans of other U.S. conflicts. The Eisen et al. (2005) study found that Gulf War-deployed veteransâ risk of fibromyalgia was more than twice that of nondeployed veterans. Smith et al. (2000) found no
HEALTH EFFECTS 227 association between Gulf War deployment and hospitalization for fibromyalgia, but it should be noted that fibromyalgia is rarely severe enough to warrant hospitalization, so this finding can not be used to contradict that of Eisen et al. (2005). Steele (2000) also found fibromyalgia in about 2% in Gulf War-deployed forces but had no nondeployed comparison group. Both the Iowa study and the Canadian study found significantly increased fibromyalgia symptoms in deployed Gulf War veterans than in nondeployed veterans, but the findings from these studies are limited because of the self- reporting of the diagnosis of fibromyalgia. Two studies looked at the presence of fibromyalgia in veterans with PTSD. Amital et al. (2006) found that almost half the Israeli veterans with combat-related PTSD also had fibromyalgia, but the sample was small. Similar but less dramatic results were seen by Dobie et al. (2004), who found that female Gulf War veterans with PTSD were three times as likely to have fibromyalgia as those without PTSD. The committee reviewed one primary study and three secondary studies on deployment- related stress and CWP. Although each of the studies found a higher prevalence of CWP in deployed than nondeployed veterans, all had considerable limitations. In Ang et al. (2006), the prevalence of CWP was found to increase both with increased combat exposure and with increased perception of life stress at the time of deployment; the study is limited in that only veterans with no pain 5 years after the conflict were evaluated 10 years after the conflict. The Stimpson et al. study (2006) also found an increase in CWP associated with deployment to a war zone. The other two secondary studies also showed more CWP in deployed than in nondeployed veterans. The committee reviewed one study of PTSD and CWP, which found a strong association between the two disorders (Ang et al. 2006). Several studies have reviewed the presence of chronic pain in veterans but the definition of chronic pain varied with the study (Hyams et al. 1996; Kuzma and Black 2006; Thomas et al. 2006). Kuzma and Black (2006) noted that many studies of Gulf War veterans reported increased pain symptoms that could be clustered into CWP, but the terminology used in the studies was not consistent and included joint pain and general aches and pain; these pain clusters may or may not have met the ACR criteria for CWP. The committee concludes that there is limited but suggestive evidence of an association between deployment to a war zone and both fibromyalgia and chronic widespread pain.
TABLE 6-11 Fibromyalgia and Chronic Widespread Pain 228 Study Study Design Population Outcomes Results Adjustments Comments Eisen et al. Population- 1061 GW-deployed Symptoms and OR 2.32, 95% CI 1.02-5.27 Age, sex, race, years of Uses gold standard for 2005 based, cross- veterans, 1128 physical examination education, cigarette- diagnosis of sectional, nondeployed for fibromyalgia with smoking, duty type, fibromyalgia; NHSGWEVTF prevalence, veterans ACR criteria service branch, rank participation rates 53% medical deployed, 39% (Derived from evaluation nondeployed Kang et al. 2000b) Smith et al. Postwar 551,841 GW- Hospitalization RR 1.23, 95% CI 1.05-1.43; Sex, age, branch of No increase after 2000 hospitalization deployed, records in 1991-1997 however, survival curves service accounting for CCEP 1,478,704 for fibromyalgia indicate excess due to effect; limited to active nondeployed hospitalization for purposes duty; most cases of of evaluation during CCEP; fibromyalgia not severe before CCEP RR 0.92, 95% enough to warrant CI 0.74-1.13 hospitalization Ang et al. 2006 Cohort of 370 veterans who Structured telephone Neither deployment to nor Controls matched for Potential for recall bias; veterans from were free of CWP at interview about 5 time in gulf significantly age, sex, branch of only veterans who were IPGWSG 5 years were years after GW; in- correlated with CWP: OR service free of CWP at 5 years examined 10 years person followup1.1, 95% CI 0.6-2.0 and OR were assessed 10 years after war: 267 GW- medical examination 1.0, 95% CI 0.7-13, after war deployed, 103 10 years after war of respectively; combat nondeployed 370 veterans who did exposure correlated: OR not report chronic 1.5, 95% CI 1.1-2.0; widespread pain 5 perception of stress due to years after warmilitary experience at time of GW correlated more significantly with CWP: OR 1.6, 95% CI 1.1-2.3, p = 0.0084 NOTE: ACR = American College of Rheumatology, CCEP = Comprehensive Clinical Evaluation Program, CI = confidence interval, CWP = chronic widespread pain, GW = Gulf War, IPGWSG = Iowa Persian Gulf War Study Group, NHSGWEVTF = National Health Survey of Gulf War Era Veterans and Their Families, OR = odds ratio, RR = relative risk.
HEALTH EFFECTS 229 REPRODUCTIVE EFFECTS Several studies of deployed veterans have indicated that veterans express concerns about the effects of deployment on child-bearingâboth their ability to conceive children and the health of those children. This section reviews studies on three subjects related to reproductive health: birth defects and miscarriage, fertility, and sexual function. To determine the association between deployment-related stress and adverse reproductive effects, the committee identified six primary studies and nine secondary studies. Primary studies obtained information from both deployed and nondeployed veterans and assessed relevant outcomes with methods beyond self-reporting, such as review of hospital or public-health records. Secondary studies obtained information relevant to the effects in question but had methodologic limitations, such as reliance on self-reports of symptoms or effects, or the examination of small or potentially biased samples of veterans. The primary studies of reproductive effects are summarized in Table 6-12. Birth Defects and Miscarriage Primary Studies The committee identified five citations of four studies that examined the prevalence of birth defects or miscarriage in veterans. One study was on veterans of the Vietnam War and four of the citations were to three studies of offspring of Gulf War veterans. The VES was a cross-sectional retrospective cohort study of all U.S. Army veterans conducted 17 years after the Vietnam War (CDC 1988b). It surveyed a random sample of the 5 million Army veterans who served during the Vietnam era. In the 1985-1986 phase 1 telephone survey, 7924 theater veterans and 7364 era veterans were asked about various reproductive outcomes in their offspring: live births, pregnancies that ended early (including miscarriages, induced abortions, and tubal pregnancies), stillbirths, major health problems in the first five years of life, low birth weight, leukemia or other cancer, birth defects, infant and child mortality, and cerebrospinal malformations (CDC 1988c). In a substudy, hospital birth records of the children of a random subsample of 1237 theater veterans and 1045 era veterans, selected without regard to phone interview results, were sought. Records were obtained for 1791 children of theater veterans and 1575 children of era veterans. During the interview phase, Vietnam-theater veterans reported more pregnancies that ended in miscarriage (OR 1.3, 95% CI 1.2-1.4) and more birth defects in their children (OR 1.3, 95% CI 1.2-1.4) than did era veterans, but there was no significant difference between the veteran groups in the other reproductive outcomes examinedâpregnancies ending in induced abortions, tubal pregnancies, stillbirths, and infant mortality (CDC 1988c). However, in the birth-defects substudy, the rates for all birth defects as documented by hospital birth records were similar for theater veterans and era veterans (72.6 and 71.1 per 1000 total births, respectively) for an OR of 1.0 (95% CI 0.8-1.4), adjusted for veteranâs age at the time of the childâs birth, race, year of entry into Army, enlistment status, general technical test score, primary military occupational specialty, years between entry and birth, maternal age, and gravidity. Cerebrospinal malformations, including anencephaly, spina bifida, and hydrocephalus, were observed in 26 children of theater veterans (expected number 18.3-32.4, based on rates from the nationwide Birth Defects Monitoring Program) and in 12 children of era veterans
230 GULF WAR AND HEALTH (expected number 17.0-30.3). Limitations of the study include the lack of data on mothersâ risk factors and possible exposures of the fathers in Vietnam, possible recall bias, and lack of verification of many of the reproductive effects reported by the veterans. Cowan et al. (1997) examined routinely collected data on all live births in 135 military hospitals in 1991-1993 to compare the frequency of birth defects in children of active-duty Gulf War veterans and nondeployed active-duty veterans. Information on 33,998 infants born to Gulf War-deployed veterans (30,151 men and 3847 women) and 41,463 born to nondeployed veterans (32,638 men and 8825 women) was reviewed. The risk of any birth defect was 7.45% for deployed veterans and 7.59% for nondeployed veterans (RR 0.98, 95% CI 0.93-1.03). There was no significant association between service in the Gulf War and the risk of any birth defect for male veterans (OR 0.97, 95% CI 0.91-1.03) or female veterans (OR 1.07, 95% CI 0.94-1.22) after adjustment for motherâs age at delivery, race or ethnicity, and marital status of parent at the time of the Gulf War. The unadjusted risk of having an infant with severe birth defects was 1.03 (95% CI 0.92-1.15) for male active-duty veterans, 0.92 (95% CI 0.71-1.20) for female active- duty veterans, and 1.00 (95% CI 0.90-1.10) for men and women combined; the authors note that when the adjusted ORs were calculated, no associations were seen but did not provide the data. A limitation of the study is that it examined data only from live births to active-duty personnel in military hospitals. After conducting a pilot study (Araneta et al. 2000) in Hawaii on birth-defects in the offspring of Gulf War veterans, Araneta et al. (2003) conducted an expanded study examining hospital records of births in 1989-1993 to military personnel in Arizona, Hawaii, Iowa, and selected counties of Arkansas, California, and Georgia. Data from the DMDC on people in military service in February 1991 were linked to data from state and county birth-defects surveillance programs to identify infants born to military personnel who served during the Gulf War. The researchers identified 11,961 infants born to Gulf War veterans (including 450 female veterans) and 33,052 infants born to Gulf War-era veterans (including 3966 female veterans). In infants conceived before and during the Gulf War, there was no significant difference in birth defects between those born to Gulf War veterans and those born to era veterans. In infants conceived after the war, the rate of hypospadias was significantly higher in male infants born to Gulf War mothers than to era mothers (RR 6.3, 95% CI 1.5-26.3, p = 0.015). The prevalence of congenital tricuspid valve insufficiency was higher in infants born to Gulf War male veterans than to era male veterans (RR 2.7, 95% CI 1.1-6.6, p = 0.039), as was the prevalence of aortic valve stenosis (RR 6.0, 95% CI 1.2-31.0, p = 0.026). A significant limitation on the interpretation of these findings is that comparisons were made for 26 birth-defect categories without correction for multiple comparisons; this raises the possibility that the statistical significance of the associations observed is due to chance (Ryan et al. 2004). In a case-control study performed by Werler et al. (2005), birth records of infants born with the malformation hemifacial microsomia were examined to determine whether there was an association between Gulf War service of the parents and the birth defect. Hemifacial microsomia was identified in 232 cases from craniofacial clinics in 26 cities and matched to 832 controls by pediatrician and childâs age. Mothers of case subjects and controls were interviewed by telephone in 1996-2002 to identify pregnancy exposures, including military service, particularly in the Gulf War, of the mother or father 5-11 years before the childâs birth. Of the cases, four mothers and 30 fathers had served in the military, as had 10 control mothers and 100 control fathers; of those, four case parents (all in the Army) and 23 control parents (including nine in the Army) had served in the Gulf War. The risk associated with Army service overall was significant
HEALTH EFFECTS 231 (OR 2.4, 95% CI 1.4-4.2), but the risk associated with having served in the Gulf War was not (OR 0.8, 95% CI 0.3-2.3). Secondary Studies Four secondary studies, one of Vietnam veterans and three of Gulf War veterans, were identified that assessed birth defects or miscarriages. Kang et al. (2000a) compared the self-reported pregnancy outcome of 3392 female Vietnam-theater veterans and 3038 female veterans not deployed to Vietnam. A structured health questionnaire was administered by telephone interview in 1992, and hospital records were obtained to verify reports of moderate to severe birth defects in 13 children. Although there were no significant differences between groups in rates of miscarriage or stillbirth, low birth weight, preterm delivery, or infant death, reports of âmoderate to severeâ birth defects were significantly higher in female Vietnam veterans (OR 1.46, 95% CI 1.06-2.02), after adjustment for demographic and military characteristics and a number of factors associated with pregnancy. Researchers obtained information on a cohort of 52,811 Gulf War veterans from the United Kingdom and a randomly selected comparison group of 52,924 nondeployed veterans regarding adverse reproductive outcomes (Doyle et al. 2004; Maconochie et al. 2003). Data were collected with a validated mailed questionnaire in 1998-2001, and 25,084 Gulf War veterans and 19,003 nondeployed veterans responded. There were 27,959 pregnancies reported by male veterans and 861 by female veterans. Miscarriages were reported more frequently by male Gulf War-deployed veterans than nondeployed veterans (OR 1.4, 95% CI 1.3-1.5). Malformations in infants were also more frequently reported among male Gulf War veterans than among nondeployed veterans (OR 1.5, 95% CI 1.3-1.7). No association between Gulf War service and miscarriage was found among women; when restricted to cases that had been clinically confirmed, the maternal results were too few to analyze for malformations and stillbirths. Kelsall et al. (2007) obtained information on reproductive function from 1424 Australian Gulf War-deployed veterans and 1548 nondeployed veterans with a mail questionnaire in 2000- 2002. There was no increased risk in veterans of miscarriage, stillbirth, or terminations. Children of male Gulf War veterans born after the Gulf War were not at greater risk of being born prematurely, having a low birth weight, or having a birth defect or chromosomal abnormality (OR 1.0, 95% CI 0.6-1.6). A mail survey with telephone followup of 10,000 married U.S. veterans deployed to the Gulf War and 10,000 married U.S. veterans not deployed to the Gulf War was conducted in 1996-1997 (Wells et al. 2006). The response rate in both groups was 51%. Among the subset of 2233 female deployed and nondeployed veterans and 2159 male deployed and nondeployed veterans who had reported one or more pregnancies, no significant differences were found between deployed men and women and nondeployed men and women in reported pregnancies, birth weight, ectopic pregnancies, stillbirths, or miscarriages. Fertility Difficulties Primary Studies The VES examined fertility (CDC 1988b) in theater and era veterans. During the 1985- 1986 physical-examination phase, fertility difficulties were reported in 21.0% and 14.5% of the 2490 Vietnam-theater and 1972 Vietnam-era veterans, respectively, for an OR of 1.5 (p < 0.05). Because of concerns about fertility, the 705 veterans without vasectomies who had physical
232 GULF WAR AND HEALTH examinations in the last 5 months of the study were asked to participate in a study of semen characteristics; 571 (81%) participated. The characteristics examined were sperm concentration, percentage of motile cells, and percentage of morphologically normal cells. A significant difference was seen between theater and era veterans only in sperm concentration (OR 2.3, 95% CI 1.2-4.3). Theater and era veterans were similar in terms of proportion who had not fathered any children (31% and 25%, respectively) as well as the average number of children fathered after primary tour of duty (1.4 and 1.5, respectively). The sperm of that VES subsample of 705 veterans was reanalyzed in another study that used a more sensitive measure of semen abnormalities (DeStefano et al. 1989). The results confirmed those of the original analysis, showing no difference in sperm characteristics except that more theater veterans than era veterans had low sperm concentration (â¤ 20 million cells/mL) for an OR of 2.7 (95% CI 1.3-5.7) adjusted for year of entry into the service, age at entry, enlistment status, enlistment general technical score, military occupational specialty, race, smoking status, illicit drug use, alcohol consumption, abstinence from sexual activity, time between sample collection and analysis, and videotaping magnification (used to measure sperm motility). Secondary Studies The study of Kelsall et al. (2007) described above also provided self-reported information on the fertility of male Australian veterans of the Gulf War. Compared with nondeployed veterans, the deployed veterans reported more fertility difficulties after the war (OR 1.4, 95% CI 1.0-1.8) although they were more successful at later fathering children (OR 1.8, 95% CI 1.3-2.6). The groups reported similar rates of pregnancies and live births. The UK Gulf War veteran study included an examination of infertility based on self- reported inability to achieve conception (type I infertility) and to achieve a live birth (type II infertility) (Maconochie et al. 2004). Reported infertility was significantly higher among male Gulf War veterans than among nondeployed veterans (type I infertility OR 1.41, 95% CI 1.05- 1.89; type II OR 1.50, 95% CI 1.18-1.89). In addition, among men not reporting infertility, time to conception was longer among Gulf War fathers than nondeployed fathers (OR 1.18, 95% CI 1.04-1.34). The studies had several limitations including low response rates that made response and recall biases possible. Sexual Dysfunction In a primary study, researchers conducted a clinical examination of 661 Danish peacekeepers who served in the gulf in 1990-1997 and 215 Danish military personnel who were not deployed to the gulf; all had previously completed a health interview that included questions about sexual problems (Ishoy et al. 2001). Veterans were asked whether they experienced any sexual problems (decreased libido or nonorganic erectile dysfunction) that they attributed to service in the gulf. Self-reported sexual problems were higher among Gulf War veterans (12%) than among controls (3.7%) for an age-adjusted OR of 2.9 (95% CI 1.4-6.0, p = 0.003). A clinical evaluation for serum concentrations of reproductive hormonesâincluding luteinizing hormone, follicle-stimulating hormone, serum hormone-binding globulin, and testosteroneâ found no significant differences between the deployed and nondeployed veterans. The deployed veterans were more likely to report sexual problems if they had seen killed or wounded people (p = 0.002), watched a friend or colleague being threatened or shot at (p = 0.02), or been threatened
HEALTH EFFECTS 233 with arms themselves (p = 0.04) than if they had not had these experiences. Deployed veterans with sexual dysfunction also reported more perceived psychologic stress during deployment than veterans without sexual dysfunction. Simmons et al. (2004) used a mail questionnaire to survey all UK Gulf War veterans and demographically similar veterans who had served at the same time but were not deployed to the gulf. Of the 42,818 male veterans who responded, 24,379 had been deployed and 18,439 had not. Sexual dysfunction or a lack of sexual drive was reported by 0.8% and 0.2% of the deployed and nondeployed veterans, respectively, for an OR of 4.6 (95% CI 3.2-6.6, p < 0.001) adjusted for age and service status at the time of the survey, service and rank at the time of the war, alcohol consumption, and smoking. Posttraumatic Stress Disorder and Reproductive Effects One study examined the association between PTSD and sexual dysfunction in combat veterans. Cosgrove et al. (2002) administered the International Index of Erectile Function to 44 combat veterans undergoing treatment for PTSD at a VA clinic and 46 age-comparable combat veterans without PTSD. Of the veterans with and without PTSD, 85% and 22% had erectile dysfunction, respectively. Severity of PTSD was associated with severity of erectile dysfunction; however, more than half the PTSD veterans were using psychotropic medications compared with only 17% of non-PTSD veterans. Summary and Conclusions There is some evidence that deployed veterans of the Vietnam War and the Gulf War report more difficulties with reproductive function, such as increased rates of miscarriage and birth defects, than do nondeployed veterans, based on four primary studies. Several studies have examined the frequency of birth defects, as ascertained from hospital records or registries, in the children of deployed and nondeployed veterans; in general, the studies have not detected a higher rate of birth defects in the children of deployed veterans. The one study in Vietnam veterans and the three studies in Gulf War veterans found no increases in birth defects in children born to deployed fathers (or mothers for the Gulf War) compared to nondeployed fathers (or mothers), although the Vietnam veterans reported more miscarriages and birth defects. One study of Gulf War veterans found an increase in the risk for male children of deployed female veterans having hypospadia and infants of male Gulf War veterans were at increased risk of congenital tricuspid valve insufficiency and aortic valve stenosis. The studies did not examine the relationship between exposure to deployment-related stressors and such self-reported problems. Furthermore, virtually none of the studies has examined the potential influence of deployment-related stress. One primary study in male Vietnam veterans did find theater veterans had lower sperm counts than did era veterans; there were no primary studies of fertility difficulties in Gulf War veterans. The two secondary studies found that Australian and UK Gulf War veterans reported more fertility difficulties than their nondeployed counterparts. One secondary study examined sexual dysfunction in deployed Danish peacekeepers and found it to be linked it to deployment- related stressors (Ishoy et al. 2001) and a second study found sexual dysfunction was also reported by UK Gulf War veterans. The committee concludes that there is inadequate/insufficient evidence to determine whether an association exists between deployment to a war zone and reproductive effects.
234 TABLE 6-12 Reproductive Effects Reference Study Design Population Outcomes Results Adjustments Comments CDC 1988c Retrospective 3366 birth records of Reproductive outcomes Fathering pregnancy that Age of veteran Lack of data on mothers of cohort, prevalence, veteransâ children: and child health ended in miscarriage OR at birth of child children studied concerning VES population-based, 1791 of Vietnam measured through 1.3, 95% CI 1.2-1.4); (or adverse potentially important telephone veterans, 1575 of era telephone interview and other reproductive pregnancy exposures: tobacco, alcohol, interview with veterans hospital birth records: outcomes no difference; outcome), race, drug use, exposure of father medical pregnancies that ended birth defects OR 1.0, 95% year of entry examination at early, including CI 0.8-1.4; based on into Army, Recall bias of veteran followup miscarriages, induced hospital birth records enlistment regarding adverse abortions, tubal status, general reproductive outcomes, pregnancies, DOB, sex technical such as miscarriages of child, status (live aptitude test birth or stillbirth), score, primary major health problems, military low birth weight, occupational leukemia or other specialty, time cancer, birth defects between and CSMs, infant and enlistment and child mortality birth of child, maternal age, gravidity Cowan et al. Cohort, routinely 33,998 infants born Occurrence of birth Any birth defect: men OR Motherâs age, Obtained information only 1997 collected data on to GW-deployed defects recorded in 0.97, 95% CI 0.91-1.03; ethnicity, marital on live births in military all live births in veterans, 41,463 to medical file women OR 1.07, 95% CI status hospitals and only on military hospitals nondeployed 0.94-1.22 children of active-duty 1991-1993 veterans personnel; no data from Severe birth defects (men later monitoring; and women) OR 1.00, information was only on 95% CI 0.90-1.10 live births, not, for example, miscarriages
TABLE 6-12 Reproductive Effects Reference Study Design Population Outcomes Results Adjustments Comments Araneta et al. Cohort, prevalence 684,645 GW- Military records of Male veterans: congenital Significant limitation is that 2003 deployed veterans, veterans linked to birth tricuspid valve comparisons were made for 1,587,102 certificates in Arizona, insufficiency RR 2.7, 26 birth-defect categories nondeployed Hawaii, Iowa, and 95% CI 1.1-6.6, p = without correction for veterans; 11961 selected counties in 0.039; aortic valve multiple comparisons, so infants from Arkansas, California, stenosis RR 6.0, 95% CI observed associations may deployed parents, and Georgia that had 1.2-31.0, p = 0.026 be due to chance; no 33,052 from active case information on possible nondeployed parents ascertainment of birth Female veterans: deployment exposures defects hypospadias RR 6.3, 95% CI 1.5-26.3, p = 0.015 CDC 1988b Retrospective 2490 Vietnam- Telephone interview Fertility difficulties OR Age at Low participation rate in cohort, prevalence, theater veterans, followed by medical 1.5, p < 0.05; sperm enlistment, race, control group, CI not given, VES population-based, 1972 Vietnam-era examination, including concentration OR 2.3, year of conducted 17 years after telephone veterans randomly semen analysis 95% CI 1.2-4.3; no other enlistment, war, participation bias in interview with selected from 7924 characteristics enlistment status semen substudy by men screening medical theater veterans and significantly different (volunteer vs who perceived negative examination at 7364 era veterans draftee), score reproductive outcomes followup who entered Army in on general 1965-1971; of 705 technical test, veterans without primary military vasectomy who were occupational eligible for sperm specialty substudy, 571 (81%) participated Werler et al. Case-control 232 cases of Mothers interviewed in Risk of hemifacial Family income, Small sample; possible 2005 hemifacial 1996-2002 about microsomia based on race, BMI in confounding by lifestyle microsomia pregnancy events and parental GW service OR pregnancy factors identified in exposures, including 0.8, 95% CI 0.3-2.3 craniofacial clinics in GW deployment of 26 cities, 832 either parent 5-11 years controls matched by before child was born age and pediatrician 235
TABLE 6-12 Reproductive Effects 236 Reference Study Design Population Outcomes Results Adjustments Comments DeStefano et Cross-sectional, 2490 Vietnam- Computer-assisted Sperm concentration (â¤20 Six covariates as Semen analysis performed al. 1989 prevalence, theater veterans, semen-analysis system million cells/mL) OR 2.7, above plus race, only on subsample of men population based, 1972 Vietnam-era for more sensitive 95% CI 1.3-5.7 age at entry into (Derived from telephone veterans randomly analysis than CDC Army, year of VES) interview followed selected from 7924 1988b entry, type of by screening theater veterans and enlistment, medical 7364 era veterans general technical examination who entered Army in test score, 1965-1971; of 705 military veterans without occupational vasectomy who were specialty eligible for semen- analysis substudy, 571 (81%) participated Ishoy et al. Cross-sectional, 661 Danish Clinical examination, Sexual problems OR 2.9, Age Deployed veterans reported 2001 prevalence peacekeepers including hormone 95% CI 1.4-6.0, p = more sexual problems if deployed to Gulf in measurements; 0.003; no difference in they had seen killed or 1990-1997, 215 interview concentrations of wounded people (p = Danish nondeployed reproductive hormones 0.002), watched friend or military controls colleague be threatened or shot at (p = 0.02), or been threatened with arms themselves (p = 0.04), or perceived psychologic stress during deployment NOTE: BMI = body-mass index, CDC = Centers for Disease Control and Prevention, CI = confidence interval, CSM = cerebrospinal malformation, DOB = date of birth, GW = Gulf War, OR = odds ratio, RR = relative risk, VES = Vietnam Experience Study.
HEALTH EFFECTS 237 SUICIDE AND ACCIDENTAL DEATH Suicide occurs when a person takes his or her own life. By definition, suicide is fatal and it affects not only those who die from or attempt suicide, but families and communities as well. In the United States, it is the 11th leading cause of death in the general population. Over 32,000 people die from suicide in the United States each year (CDC 2007b). Men are 4 times more likely than women to die from suicide, although women are more likely to attempt suicide. Accidental death is considered to be death resulting solely through violent, external, and accidental means. Primary studies provided the basis of the committeeâs findings on the relationship between deployment-related stress and suicide. Primary studies had to meet the criteria established in Chapter 2. Primary studies were considered to be studies in which exposure was determined by deployment status or studies that used a diagnosis of PTSD as a marker of war- related trauma. Primary studies also included those in which the cause of death was confirmed through at least one supplemental source, for example, through linkage of death-certificate data to the National Death Index or by a qualified nosologist or medical-review team that was blinded to subjectsâ deployment status. Secondary studies typically had methodologic limitations, such as assessment of suicidal behaviors solely on the basis of self-report data. Most of the secondary studies did not confirm causes of death through a supplemental source. Understanding of the relationship between deployment-related stress and suicide or accidental death is hampered by the use of different case definitions for suicide or accidental death among cohorts of veterans. Suicide is an outcome that is rare even in the highly vulnerable population being considered here, so studies with low statistical power would have limited ability to establish an association between it and deployment-related stress. Primary studies are summarized in Table 6-13. Primary Studies The committee identified seven primary articles (Boehmer et al. 2004; Bullman and Kang 1996; CDC 1987; Kang and Bullman 1996, 2001; Kaplan et al. 2007; Thomas et al. 1991) and one government report (MHAT 2006a) relevant to determining whether there is a relationship between deployment-related stress and suicide or accidental death. The VES is a major comprehensive study of the health of Vietnam veterans, an initial historical cohort of 9324 Vietnam-theater veterans who were compared with 8989 Vietnam-era veterans who served in the military during the Vietnam War but were deployed in Korea, Germany, or the United States. To be included in the study, veterans had to be male U.S. Army veterans who initially entered military service from 1965 to 1971, served a single term of enlistment, and were discharged alive in enlisted pay grades E-1 through E-5; participants were randomly selected from military personnel files of veterans discharged after 1971. The first component of the VES to be completed was an assessment of mortality (CDC 1987). Nearly complete ascertainment of vital status of both veteran cohorts was obtained with multiple methods. In addition to determination of cause of death through inspection of all death certificates by persons blinded to the veteransâ Vietnam status, a medical review board, also blinded to Vietnam status, independently coded the underlying causes of all 426 deaths by using data from additional sources, including physicians, hospital records, autopsy records, and coroner and law-enforcement files. In the first followup period, which began at discharge from
238 GULF WAR AND HEALTH active duty (1965) and ended at death or on December 31, 1983, whichever came first, the total death rate of theater veterans was 17% higher than that of era veterans, and theater veterans had a 25% higher mortality due to external causes. The excess mortality during the early followup period (considered to be the first 5 years) was due largely to external causes (rate ratio 1.25, 95% CI 1.00-1.55), including motor-vehicle accident (rate ratio 1.93, 95% CI 1.16-3.22), suicide (rate ratio 1.72, 95% CI 0.76-3.88), homicide (rate ratio 1.52, 95% CI 0.59-3.91), and accidental poisoning (rate ratio 1.69, 95% CI 0.49-5.77); the death rate ratio for all causes of death was 1.45 (95% CI 1.08-1.96) for the first 5 years after discharge and 1.01 (95% CI 0.79-1.28) after 6 or more years. Death from motor-vehicle crashes was almost 2 times more likely in theater veterans than in era veterans during the early followup period. Alcohol use did not appear to account for the excess risk of a fatal motor-vehicle crash. A similar increase in risk was found for suicide during the early followup period; adjustment for covariates increased the risk ratio for suicide in the early followup period from 1.7 (95% CI 0.76-3.88) to 2.59 (95% CI 1.09-6.17) based on death certificates analyses and to 2.56 (95% CI 1.11-5.87) based on medical review panel assessment. Thereafter, the ratio for suicide declined until it was no more than 1.0. The rate ratios for homicide and other accidents also were below 1.0 in the later followup period and the rate ratio for motor vehicle accidents was 1.16 (95% CI 0.72-1.87). Boehmer et al. (2004) studied postservice mortality in a 30-year followup of Vietnam veterans by obtaining data on vital status and underlying causes of death on 18,313 male U.S. Army veterans in the VES cohort, thus extending followup an additional 17 years. Vital status was retrospectively ascertained from the end of the original study (1984) through 2000 by using three national databases: the VA BIRLS, the Social Security Administrationâs Death Master File, and the National Death Index Plus. Potential matches from each data source were reviewed manually, and underlying cause-of-death codes were obtained from the National Death Index Plus. If data on cause of death were not available from the National Death Index Plus, official copies of death certificates were obtained, and an experienced nosologist at National Center for Health Statistics (NCHS) coded the cause of death. NCHS converted ICD-9 and ICD-10 causes of death to the NCHS list of 113 selected causes of death, which were then categorized for external causes to allow comparisons with the original study. Crude death rates were calculated separately for Vietnam-theater and Vietnam-era veterans by using person-years at risk, beginning with the day of discharge from service and ending with the date of death or December 31, 2000, whichever came first. Cox proportional- hazards models were used to approximate the adjusted relative risks for all-causes mortality; for each model, potential confounding and effect modification were assessed. Standardized mortality ratios were used to compare veteran death rates with those in the U.S. male population; they were adjusted for age, race, and calendar year but were limited to deaths through 1998 because U.S. rates had not yet incorporated ICD-10 codes when the analysis was conducted. There were 1138 new deaths during the additional followup period, five of which occurred during the previous study but had not been identified then. At 5 years since discharge, the only significant external cause of death was accidental poisoning (rate ratio 2.58, 95% CI 1.09-6.14). Mortality due to motor-vehicle accidents or suicide was not significantly higher in theater veterans than in era veterans more than 5 years after discharge from the service (rate ratio 1.02, 95% CI 0.73-1.43 and rate ratio 0.93, 95% CI 0.64-1.34, respectively). However, there continued to be an excess of drug-related deaths among Vietnam veterans during the entire followup period, and there was some overlap between the definitions of unintentional poisoning and drug-related death. The authors noted that even though theater veterans were more likely than era veterans to die from
HEALTH EFFECTS 239 drug-related causes, the contribution to the overall number of deaths was small. The strengths of the study lie in the relatively long followup time after the Vietnam War and, as in the other primary studies reviewed above, the use of multiple databases. In a retrospective cohort study, Thomas et al. (1991) examined excess mortality in 4582 female veterans who served in Vietnam during the period July 1965-March 1973, and 5324 female veterans who never served in Vietnam. The investigators reported a nonsignificant excess mortality due to external causes, including suicide, in female theater veterans compared with era veterans (RR 1.33, 95% CI 0.80-2.23). There was a significant excess of motor-vehicle accidents (RR 3.19, 95% CI 1.03-9.86). There was no increase in risk of suicide in this cohort of female theater veterans (RR 0.96, 95% CI 0.39-2.39). Some caution is advisable in using these findings to support a relationship between deployment-related stress and accidental death in that many women deployed to Vietnam were nurses and generally did not witness combat although they were exposed to dead bodies and other traumatic events. However, the investigators estimated cause-specific mortality in the female theater veterans relative to the female era veterans by using a proportional-hazards multivariate model that adjusted for rank (officer or enlisted), military occupation (nursing or other), duration of military service, age at entry into followup, and race. With regard to the nonsignificant finding when suicide alone was used as the outcome, the small number of female veterans who served in Vietnam may not provide sufficient statistical power to detect an association between serving in Vietnam and suicide. Although the study cohort of female Vietnam veterans had sufficient power (95%) to detect an increase in relative risk of death from all causes of 1.3 or greater, the investigators noted that the study had insufficient statistical power to detect moderate increases in deaths from rare causes. Taking the power calculations into consideration may explain why when suicide, accidental death, and nonintentional poisoning are included together as an external cause of death there is sufficient power to detect an increased risk of death in female theater veterans from external causes, especially accidental death, but not suicide. Bullman and Kang (1996) sought to determine whether there was an association between Vietnam veteransâ exposure to trauma, as indicated by being wounded in combat, and risk of suicide. Subjects were identified from a computerized database, the Casualty Information System, which covered casualties sustained by U.S. Army military and civilian personnel and their dependents worldwide from 1961 to 1981. Of about 70,000 veterans who received nonlethal wounds in 1969-1973, 34,534 were selected randomly for inclusion in the study. Cause-specific mortality in all wounded veterans was compared with that in the U.S. population after adjustment for age, race, and calendar year. In study subjects, there was a slight decrease in overall mortality (standardized mortality ratio 0.97, 95% CI 0.93-1.02). However, study subjects had a statistically significant increase in risk of death from all motor-vehicle accidents (standardized mortality ratio 1.23, 95% CI 1.10-1.37), and all accidents (standardized mortality ratio 1.18, 95% CI 1.09-1.29). The rate of suicide was slightly, but not significantly, higher in wounded veterans than in U.S. men (standardized mortality ratio 1.12, 95% CI 0.96-1.30). The relative risk of suicide for veterans wounded two or more times compared to those wounded once was 1.50 (95% CI 1.01- 2.24). There was a statistically significant increase in risk of suicide in veterans who were wounded more than once and hospitalized for a wound (rate ratio 1.82, 95% CI 1.12-2.96) compared with those wounded once and not hospitalized. In addition, there was a significant trend of increasing risk of suicide with increasing occurrences of combat trauma (Ï2 = 5.3, p <
240 GULF WAR AND HEALTH 0.05). When study subjects were stratified by hospitalization status and number of times wounded and compared with U.S. men, those who received at least one wound that required hospitalization had a higher risk of suicide than those not hospitalized (standardized mortality ratio 1.22, 95% CI 1.00-1.46), and those wounded more than once had a higher risk of suicide than those wounded only once (standardized mortality ratio 1.58, 95% CI 1.06-2.26). Vietnam veterans who were wounded only once and never required hospitalization had the lowest risk of suicide (standardized mortality ratio 0.96, 95% CI 0.72-1.24), and those who were wounded more than once and required hospitalization for at least one wound had the highest risk of suicide (standardized mortality ratio 1.73, 95% CI 1.10-2.60). Strengths of the study included a 95% statistical power to detect a 1.2-fold increase in relative risk of motor-vehicle accidents and a 1.28-fold increase in relative risk of suicide in comparison with the U.S. general population using an alpha level of 0.05 and a two-sided test. Furthermore, although most studies of exposure to combat trauma rely on interview data to ascertain exposure, this study was able to minimize that source of potential bias by using military records (that is, military reports of casualties) to determine exposure. However, this study, like many, was unable to include data on potentially predisposing psychologic and behavioral characteristics that might be important risk factors for suicide after military service. Kang and Bullman (1996) obtained death certificates of Gulf War veterans, of a control group that was mobilized and deployed elsewhere, and of a control group of veterans who were not deployed at all. Death certificates were provided through VA regional offices, as identified from combined data sources, which included the BIRLS database and the files from Social Security Administration. A qualified nosologist used the ICD-9 to code all causes of death and was blinded to subjectsâ deployment status. Of 695,516 Gulf War veterans, 1765 had died, and death certificates were located for 1654 (93.7%). Of 746,291 controls deployed elsewhere or not deployed, 1729 had died, and death certificates were located for 1615 (93.4%). Gulf War veterans had a significantly higher mortality than veterans deployed elsewhere or never deployed during the period (adjusted mortality rate ratio 1.09, 95% CI 1.01-1.16). Accidental death accounted for the largest part of the difference (rate ratio 1.25, 95% CI 1.13-1.39), but the suicide rate in Gulf War veterans was not higher (rate ratio 0.94, 95% CI 0.79-1.12); rate ratios were adjusted for age, race, sex, branch of service, and type of unit. In general, the contrast between mortality for all causes in female veterans who served in the Gulf War and the general U.S. population was greater (standardized mortality ratio 0.56, 95% CI 0.44-0.71); however, the excess risk of suicide in female veterans approached significance in comparison with women in the general U.S. population (standardized mortality ratio 1.81, 95% CI 0.90-3.24). Kang and Bullman (2001) extended their 1996 study for an additional 4 years and evaluated changing patterns of mortality in U.S. veterans of the Gulf War for almost 7 years of followup. They compared cause-specific mortality in 621,902 Gulf War veterans who arrived in the Persian Gulf before March 1, 1991, with nondeployed veterans. That excluded people who arrived in the Persian Gulf after hostilities had ended and therefore were not likely to have received the exposures of interest, which included the psychologic stresses of combat. A stratified random sample of all military personnel who served during the war but were not deployed to the Persian Gulf made up the control group of 746,248 veterans. Vital status was determined by using multiple databases, including those of the VA BIRLS and Social Security Administration. Death certificates were requested from VA regional offices. For death certificates that could not be located in that manner, the National Death Index was used to capture causes of death. As in their previous study of Vietnam veterans, the investigators used a
HEALTH EFFECTS 241 qualified nosologist who coded causes of death according to the ICD-9 without knowledge of the subjectsâ deployment status. Kang and Bullman (2001) reported four analyses. The first yielded unadjusted rate ratios calculated from crude death rates. The second used Cox proportional-hazards models to account for possible confounding and effect modification by selected covariates related to veteransâ risk of dying (or hazard) from specific causes from the time they entered the cohort. Adjusted rate ratios derived from the models were used to approximate relative risk. The third analysis compared cause-specific mortality in Gulf War veterans and nondeployed veterans with expected numbers of deaths in the overall U.S. population. Finally, changes in relative mortality in Gulf War veterans and nondeployed veterans during four 20-month followup periods were examined with a chi-squared test for trend. For each of followup periods beginning in January 1993 and extending to December 1997, an adjusted rate ratio was derived from the Cox proportional- hazards model after adjustment for age, race, marital status, branch of service, and type of unit. In the analyses, there was no excess all-cause-specific mortality and no excess mortality from all external causes combined, but there was an excess risk of death from motor-vehicle accidents in Gulf War veterans compared with nondeployed veterans that decreased steadily from the first followup (rate ratio 1.32, 95% CI 1.13-1.53) to the last followup (rate ratio 1.00, 95% CI 0.82- 1.22). The chi-squared value (Ï2 = 7.53) indicated that there was a significant decreasing trend (p = 0.0061) in the risk of motor-vehicle accidental death with time since the Gulf War in contrast with a steadily increasing trend in disease-related causes of death in Gulf War veterans compared with nondeployed veterans; deaths from disease was about the same for deployed and nondeployed veterans by the end of the last followup period. The authors concluded that 7 years after the Gulf War, mortality in both Gulf War and nondeployed veterans was half that in their civilian counterparts. A report from the Mental Health Advisory Team-III (MHAT-III) (MHAT 2006a) issued by the Office of the Surgeon General, U.S. Army Medical Command, is included as a primary source because it used a direct measure of deployment exposure in Iraq. The MHAT-III found that in 2005 the confirmed OIF suicide rate for soliders deployed in Iraq was 19.9 per 100,000 troops. That was compared with a suicide rate in the entire Army of 12.3 per 100,000 in the same year and an average suicide rate of 11.6 per 100,000 in 1995-2005 in the Army. Secondary Studies Eight publications were identified as secondary studies (Adams et al. 1998; Boscarino 2005; Kaplan et al. 2007; Kramer et al. 1994; Macfarlane et al. 2000, 2005; Price et al. 2004; Writer et al. 1996). Two publications by Macfarlane et al. (2000, 2005) identified veterans from the UK who served in the Gulf War and veterans who were not deployed to the region. Data on the deployed and era cohorts were obtained from the National Health Service but not confirmed with a supplemental source. The initial study (2000) by the investigators found that the deployed cohort had higher mortality from external causes than the era cohort. However, the difference was nonsignificant and was due primarily to accidents. The second study (2005) by the investigators found no differences in external causes of death between Gulf War and era veterans at followup in June 2004. The study was considered secondary because it did not confirm cause of death. Writer et al. (1996) used the same retrospective cohort data as Kang and Bullman (1996). Cause-specific mortality from August 1990 though July 1991 was compared for Gulf War- deployed veterans and active-duty troops deployed elsewhere. All active-duty military deaths
242 GULF WAR AND HEALTH were reported on a Report of Casualty Form collected routinely and unrelated to the study. Cause of death was reviewed on every casualty report, and the investigators changed the casualty code on the form if it was found to be clearly incorrect. When possible, the cause of death of those who served in the Persian Gulf was verified through a registry of autopsies maintained by the Office of the Armed Forces Medical Examiner; however, no central repository of autopsy data was available to determine the accuracy of deaths of service members deployed elsewhere. The investigators reported that although death rates from unintentional injury were significantly higher in those deployed to the gulf region than in those deployed elsewhere (69.1 vs 41.2 per 100,000 person-years), suicide rates were significantly lower in those deployed to the gulf (3.78 vs 10.82 per 100,000 person-years). The standardized mortality ratios were 0.34 (95% CI 0.16- 0.63) for suicides and 1.54 (95% CI 1.32-1.77) for unintentional deaths. This study was considered secondary because no supplemental source was used to determine causes of death in the control group. Adams et al. (1998) used the Southeast Asia Combat Area Casualties Database to assess the risk of suicide in U.S. ground troops during their tours of duty in Vietnam between 1957 and 1973. They found a higher risk of suicide in men serving in the Army than in other branches of the military (OR 7.86, 95% CI 1.64-5.31). The study was included as a secondary study because no confirmatory sources were used to check the accuracy of the cause of death and only deaths during deployment in country were considered. Kaplan et al. (2007) examined suicide rates in a nationally representative sample of male community residents in the United States who had completed National Health Interview Surveys in 1986-1994. Vital status and causes of death were determined for survey respondents from 1986 through 1997. Data were collected with face-to-face household interviews; response rates were 94-98%. The outcome variable in the study was death by suicide. The relative risk of suicide was estimated with the Cox proportional-hazards model and adjusted for demographics and socioeconomic, health, and functional status. Time to death was measured from the month of the interview to the month of suicide. Analyses that adjusted for self-reports of demographic characteristics, socioeconomic factors, and health status found that respondents who reported having ever served on active duty in the armed forces had twice the risk of suicide of those who reported being nonveterans (hazard ratio 2.04, 95% CI 1.10-3.80). The committee emphasizes that a major limitation of this study that the investigators did not include a measure of deployment, which is the exposure of interest for this report. Several studies of suicide in veterans that confined their samples to VA populations (Fontana and Rosenheck 1995b; Lambert and Fowler 1997; Thompson et al. 2002) could not avoid the inherent bias of a self-selected population that elects to receive health care from VA, and so were not considered in this review. PTSD and Suicide or Accidental Death Boscarino (2005) and Kramer et al. (1994) found that the risk of suicide or suicidal behavior was higher in Vietnam veterans with PTSD than in those without it. Boscarino (2005) used data from the VES to examine causes of death in 15,288 Vietnam Army veterans 30 years after the war and 15 years after the phase I telephone surveys and personal interviews. When the DIS was used, 10.6% of the theater veterans and 2.9% of the era veterans were found to have PTSD. Vital status and cause-specific mortality were determined with the VA BIRLS, the Social Security Administration Death Master File, and the National Death Index Plus. Veterans with PTSD were more likely than those without PTSD to have died since the 1985-1986 survey
HEALTH EFFECTS 243 (11.8% vs 4.9%, p < 0.001). The hazard ratio for death from external causes in theater veterans with PTSD compared with those without PTSD was 2.3 (95% CI 1.4-3.9). External causes of death were motor-vehicle collision, accidental poisoning, suicide, homicide, and injury of undetermined intent. The hazard ratio was adjusted for race, Army volunteer status, Army entry age, Army discharge status, Army illicit drug use, age at interview, and intelligence. Price et al. (2004) used data from phase III of the VES survey, conducted in 1996-1997, to assess the effect of PTSD on the risk of nonfatal suicidality (suicidal thinking or behavior) from 1972 to 1996 in 637 male Vietnam-theater veterans. Comparing veterans with PTSD to those without PTSD, the hazard ratio was 1.51 (risk limits 0.65-3.48); during that time, nine of 943 veterans in the cohort database died from suicide, and 15.7% of the 641 veterans interviewed in 1996-1997 reported suicidality. MDD was also significantly associated with nonfatal sucidality over the 25 years (hazard ratio 3.21, risk limits 1.93-5.34), as was drug dependence (hazard ratio 2.06, risk limit 1.31-3.24) but not alcohol dependence (hazard ratio 1.18, risk limit 0.84-1.68). Both PTSD and drug dependence prolonged the duration of suicidality; although the influence of PTSD remaining high during the 25 years, the influence of drug dependence on suicidality decreased with time. Kramer et al. (1994) assessed the impact of PTSD on suicidality in 131 Vietnam veterans who lived in the community, 64 veterans in a clinical outreach program, and 37 veterans who were psychotherapy patients. PTSD and suicidality were assessed with the Schedule for Affective Disorders and Schizophrenia-Lifetime Version. Suicidal thoughts were more highly correlated with lifetime and current PTSD (p < 0.001) than were suicidal behaviors. Summary and Conclusions The results of the studies based on cohorts from the Vietnam War and the 1991 Gulf War demonstrate consensus and divergence with regard to the strength of an association between deployment and suicide or accidental death. The conclusions on suicide and accidental death are presented separately. One study from the Vietnam era provides evidence of a dose-response relationship between the degree of traumatic injury suffered during deployment and suicide. Considering that study with two other primary studies, one in the early followup period after the Vietnam War and the other a study of female Vietnam veterans, the committee concluded that there is sufficient evidence that, at least for some period after deployment, Vietnam veterans were at increased risk of dying from suicide compared with nondeployed veterans. That conclusion is supported in part by the recent report of a higher suicide rate in deployed OIF troops in 2005 than in the Army in that year. The risk may also be increased for some veteran groups, namely, those suffering from PTSD, depression, or substance abuse, and those with specific war-related traumas. The association was not strongly supported by studies conducted in those deployed to the Gulf War, but the brevity of that war and the limited involvement of troops in the region may have kept the statistical power of those studies too low to detect an association between a rare effectâ suicideâand deployment. The committee concluded that there is also sufficient evidence for an association between deployment and accidental death, again, especially potent during the early years postexposure. These conclusions are based on evidence from both the Vietnam War and the Gulf War and from primary and secondary studies. The studies considered in this section have limitations. In the United States, the coding of mortality data changed substantially in 1999 from ICD-9 to ICD-10, and the number of deaths
244 GULF WAR AND HEALTH and death rates due to suicide and accidental death before 1999 may not be readily comparable with data from 1999 on (Anderson et al. 2001; Hoyert et al. 2001). In 2003, the National Center for Health Statistics (NCHS) revised the ICD-10 Injury Mortality Matrix to finalize groupings of external cause-of-injury classifications; this affected groupings related to both suicide and motor- vehicle accidental deaths. The best way to address discrepancies in how clinicians and researchers define suicidal behaviors is still being debated. Thus, underlying all the studies considered in this section is an inevitable limitation related to a potential for bias due to the inconsistency of case definition of suicide and accidental death among studies of veterans that were conducted over decades. Further limitations may be related to the presence of unknown bias due to misclassification of accidental deaths, some of which may be suicides, and to potential underreporting. However, those potential sources of bias are likely to occur equally among veterans and nonveterans, so it was critical to assess the inclusion of an appropriate control or reference group in the studies evaluated here. The committee concludes that there is sufficient evidence of an association between deployment to a war zone and suicide in the early years after deployment. The committee also concludes that there is sufficient evidence of an association between deployment to a war zone and accidental death in the early years after deployment.
TABLE 6-13 Suicide and Accidental Death Reference Study Design Population Outcomes Results Adjustments Comments CDC 1987 Retrospective 9324 Vietnam- Follow-up began at date Mortality 17% higher in theater than Unadjusted cohort, theater veterans, of discharge, ended at era; all deaths â¤ 5 years rate ratio 1.45, VES mortality, 8989 Vietnam-era death or December 31, 95% CI 1.08-1.96, > 5 years rate ratio prevalence veterans who 1983, whichever came 1.01, 95% CI 0.79-1.28; all external entered Army in first; mortality data from causes rate ratio 1.25, 95% CI 1.00- 1965-1971 death certificates coded 1.55; death from motor-vehicle by ICD-9 and blind to accidents â¤ 5 years rate ratio 1.93, 95% Vietnam status; review CI 1.16-3.22), > 5 years rate ratio 1.16, panel of two physicians 0.72-1.87; suicide â¤ 5 years rate ratio reviewed cause of death, 1.72, 95% CI 0.76-3.88 (adjusted for using medical and legal covariates based on death certificates records for each of 426 risk ratio 2.59, 95% CI 1.09-6.17 or deaths; all causes of medical review panel risk ratio 2.56, death were coded by 95% CI 1.11-5.87), > 5 years rate ratio ICD-9 and blind to 0.64, 95% CI 0.32-1.30; homicide â¤ 5 Vietnam status years rate ratio 1.52, 95% CI 0.59-3.91, > 5 years rate ratio 0.78, 95% CI 0.39- 1.55; accidental poisoning (mostly drugs) â¤ 5 years rate ratio 1.69, 95% CI 0.49-5.77 Boehmer et Cohort, 9324 Vietnam- Vital-status and cause- All external causes â¤ 5 years rate ratio Crude death rate Used multiple al. 2004 mortality 30 theater veterans, of-death followup using 1.62, 95% CI 1.16-2.26, > 5 years rate per 100,000 sources for vital- years after war 8989 Vietnam-era VA BIRLS, SSA, and ratio 1.08, 95% CI 0.89-1.29; motor- person-years status ascertainment; (Derived veterans NDI Plus databases; vehicle accidents â¤ 5 years rate ratio death certificates may from VES) mortality data from 1.93, 95% CI 1.16-3.22, > 5 years rate not be accurate as to death certificates; causes ratio 1.02, 95% CI 0.73-1.43; suicide cause of death, of death coded by ICD-9 â¤ 5 years rate ratio 1.72, 95% CI 0.76- particularly suicide; and ICD-10 and blind to 3.88, > 5 years rate ratio 0.93, 95% CI long followup after deployment status 0.64-1.34; accidental poisoning (mostly Vietnam war; large drugs) â¤ 5 years rate ratio 1.69, 95% CI number of deaths 0.50-5.77, > 5 years rate ratio 2.58, 95% CI 1.09-6.14; homicide â¤ 5 years rate ratio 1.52, 95% CI 0.59-3.91, > 5 years rate ratio 0.80, 95% CI 0.50-1.26; undetermined intent > 5 years rate ratio 3.19, 95% CI 0.88-11.58 245
TABLE 6-13 Suicide and Accidental Death 246 Reference Study Design Population Outcomes Results Adjustments Comments Thomas et Retrospective 4582 female Mortality data from Cause-specific mortality rates; all Rank, military Small cohort of al. 1991 cohort, Vietnam-theater death certificates coded causes of death combined rate ratio occupation, women, short mortality veterans, 5324 by ICD-9 and blind to 0.93, 95% CI 0.74-1.16; external duration of followup period, female Vietnam-era Vietnam status causes rate ratio 1.33, 95% CI 0.80- military service, small number of veterans; theater 2.23; motor-vehicle accidents rate ratio age at entry into deaths veterans served in 3.19, 95% CI 1.03-9.86; suicide rate followup, race Vietnam in 1965- ratio 0.96, 95% CI 0.39-2.39 1973 Bullman Retrospective 34,534 wounded Followup from date of Suicide among veterans who were 1990 data showing and Kang cohort, Vietnam-theater wounding until death or wounded once and not hospitalized rate military population in 1996 average length veterans selected December 31, 1991, ratio 1.00; wounded once and general had lower of followup 21 randomly from whichever was earlier; hospitalized rate ratio 1.19, 95% CI suicide risk than U.S. years sample of 70,000 cause of death was 0.85-1.66; wounded two or more times population wounded veterans based on data from each and not hospitalized rate ratio 1.24, (standardized included in Casualty veteranâs death 95% CI 0.53-2.88; wounded two or mortality ratio = 0.69, Information System certificate coded by more times and hospitalized for wound 95% CI 0.56-0.82); database in 1969- ICD-9; Casualty rate ratio 1.82, 95% CI 1.12-2.96; risk is almost twice 1973 Information System data compared with U.S. population, that for peace time, used to measure date of wounded veterans had slightly higher supporting wounding, severity of risk of death from suicide standardized association between wound, number of times mortality ratio 1.12, 95% CI 0.96-1.30 exposure to combat wounded trauma and increased risk of suicide Kang and Retrospective 695,516 GW- Vital-status followup Mortality in men: all causes rate ratio Crude death rate Interview data, small Bullman cohort, deployed veterans, using VA and SSA 1.09, 95% CI 1.01-1.18; per 10,000 sampleâalcohol- 1996 postwar 746,291 databases; mortality data external causes rate ratio 1.17, 95% CI person-years; related accident mortality, 2.4- nondeployed from death certificates; 1.07-1.29; all accidents rate ratio 1.26, Cox group may have year followup veterans, including causes of death coded by 95% CI 1.11-1.42; motor-vehicle proportional- differed from suicides 49,919 female GW ICD-9 and blind to accidents rate ratio 1.27, 95% CI 1.09- hazards model in undetected ways veterans, 84,517 deployment status 1.48; suicide rate ratio 0.88, 95% CI adjusted for age, (unrecognized female active-duty 0.72-1.08 race, sex, service confounders) controls branch, type of Mortality in women: all causes rate unit ratio 1.32, 95% CI 0.95-1.83; external causes rate ratio 1.78, 95% CI
TABLE 6-13 Suicide and Accidental Death Reference Study Design Population Outcomes Results Adjustments Comments 1.16-2.73; all accidents rate ratio 1.83, 95% CI 1.02-3.28; motor-vehicle accidents rate ratio 1.81, 95% CI 0.96- 3.41; suicide rate ratio 1.47, 95% CI 0.63-3.43 Kang and Retrospective 621,902 GW- Vital status followup Mortality in men: all causes rate ratio Crude death rate Interview data, small Bullman cohort; initial deployed veterans, using VA and SSA 0.95, 95% CI 0.92-0.99; external per 10,000 sampleâalcohol- 2001 study was 2.4 746,248 databases; mortality data causes rate ratio 1.04, 95% CI 0.99- person-years; related accident years; this nondeployed from death certificates; 1.10; all accidents rate ratio 1.15, 95% Cox group may have (Followup study was veterans (stratified causes of death coded by CI 1.07-1.23; motor-vehicle accidents proportional- differed from suicides to Kang and nearly 7 years random sample of ICD-9 and blind to rate ratio 1.19, 95% CI 1.09-1.30; hazards model in undetected ways Bullman all military deployment status suicide rate ratio 0.92, 95% CI 0.83- adjusted for age, (unrecognized 1996) personnel, including 1.02 race, sex, service confounders) active-duty, branch, type of reserves, National Mortality in women: all causes unit Guard units who rate ratio 1.16, 95% CI 0.97-1.38; served during external causes rate ratio 1.39, 95% CI conflict but not 1.08-1.80; all accidents rate ratio 1.36, deployed in Persian 95% CI 0.95-1.96; motor-vehicle Gulf) accidents rate ratio 1.63, 95% CI 1.09- 2.45; suicide rate ratio 1.29, 95% CI 0.78-2.31 MHAT Survey of 1320 U.S. Army MHAT Soldier and Suicide rate in Iraq (per 100,000) 18.8 Report states Report is prepared 2006a OIF troops soldiers (79% Marine Well-Being (2003), 9.6 (2004), 19.9 (2005), 16.1 suicide rates periodically as during active-duty, 8% Survey of anonymous (2006), p < 0.01; U.S. Army 10-year were adjusted, requested by Army; it deployment in reserve, 13% troops; observations by average suicide rate 11.6 per 100,000 but no details is not a formal, Iraq National Guard), authors; focus-group year given methodologically 447 Marines interviews; behavioral- rigorous survey, but health surveys, primary is included because it care, unit ministry-care contains recent data surveys on troops surveyed in Iraq NOTE: BIRLS = Beneficiary Identification Record Locator Subsystem Death File, BMI = body-mass index, CI = confidence interval, GW = Gulf War, ICD-9 = International Statistical Classification of Diseases, 9th edition, MHAT = Mental Health Advisory Team, NDI = National Death Index, OIF = Operation Iraqi Freedom, OR = odds ratio, SSA = Social Security Administration, VA = Department of Veterans Affairs, VES = Vietnam Experience Study. 247
248 GULF WAR AND HEALTH SYMPTOM REPORTING Numerous studies, many of them discussed earlier in this chapter, indicate that war-zone- deployed veterans, both men and women and regardless of the conflict in which they served, consistently report more symptoms, more adverse health effects, and poorer health status than do veterans who served in the military at the same time but were not deployed or were deployed but not to a war zone (CDC 1988b; Cherry et al. 2001a; Eisen et al. 2005; Goss Gilroy Inc. 1998; Gray et al. 1999, 2002; Hotopf et al. 2003b; Iowa Persian Gulf Study Group 1997; Ishoy et al. 1999; Kang et al. 2000b; Kelsall et al. 2004a; Kulka et al. 1990; OâToole et al. 1996b; Ozakinci et al. 2006; Pierce 1997; Proctor et al. 1998; Simmons et al. 2004; Steele 2000; Unwin et al. 1999; Wolfe et al. 1999). In particular, Gulf War veterans reported markedly more symptoms compared to their nondeployed counterparts, whether the veterans were from the United States, the UK, Canada, Australia, or Denmark. Increased symptoms have also been reported in Vietnam veterans from the United States and Australia (CDC 1988b; OâToole et al. 1996b). This section considers three aspects of this excess symptom reporting by deployed veterans from the Vietnam War, the Gulf War, and OEF and OIF that do not readily fit with the health outcomes already discussed: general symptoms that do not appear to be indicative of a specific illness or disorder, symptoms that appear to cluster into the ICD category of unexplained illness, and chronic pain of unknown origin. The committee has included the primary studies for general symptoms and unexplained illness in Table 6-14. General Symptoms and Signs Gulf War-deployed veterans reported many symptoms at rates 2-3 times higher than those seen in nondeployed veterans in several large studies from five countries: the United States (Kang et al. 2000b; Gray et al. 2002), the United Kingdom (Cherry et al. 2001a; Simmons et al. 2004; Unwin et al. 1999), Denmark (Ishoy et al. 1999), Canada (Goss Gilroy Inc. 1998), and Australia (Kelsall et al. 2004a). Similar results have been seen in veterans of the Vietnam War (CDC 1988b; Kulka et al. 1990) and World War II (Villa et al. 2002). Furthermore, the symptoms and reports of poor health persist, often for many years after the war (Ozakinci et al. 2006). Among the symptoms most commonly reported by Gulf War and Vietnam veterans are fatigue, headaches, irritability or feeling anxious, poor memory, joint stiffness or pain, sleep difficulties (including problems in falling asleep or staying asleep and unrefreshing sleep, such as waking up feeling tired), and poor concentration. Several of the symptomsâsuch as sleep problems, abdominal pain, and chest painâand neurologic problems have been discussed by the committee in previous sections on sleep disturbances, neurocognitive effects, cardiovascular diseases, and digestive system diseases. CDC undertook the VES to assess the health status of Vietnam-theater and Vietnam-era veterans who served in the U.S. Army during 1965-1971; the study was completed in 1988, about 15-20 years after the war (CDC 1988b). A nationally representative random sample of 7924 theater veterans and 7364 era veterans completed a phase 1 telephone interview; in phase 2, a subsample of 2490 of the theater veterans and 1972 of the era veterans also completed physical- and psychologic-health screening examinations in 1985-1986. In phase 1, 19.6% of the 7924 theater veterans reported their health as fair or poor compared with 11.1% of 7364 era
HEALTH EFFECTS 249 veterans (OR 1.8, 95% CI excludes 1.0, p < 0.05); similar results were found on examination (OR 1.9, 95% CI excludes 1.0). On examination, 10.2% of theater veterans and 6.2% of era veterans had somatic symptoms, which included nervousness, fatigue, gastrointestinal tract ailments, dizziness, and headaches for a significant OR of 1.7 (95% CI excludes 1.0); symptoms of peripheral neuropathy (numbness, tingling, burning sensation, or weakness of arms or legs) were found in 3.5% of the theater veterans and 1.9% of the era veterans (OR 1.5, 95% CI 1.0- 2.2). The OR was adjusted for age at enlistment, race, year of enlistment, enlistment status, score on a general technical test, and primary military occupational specialty. The study had the advantage of including a physical examination and a large study population; the response rate was 75% for theater veterans and 63% for era veterans. Ishoy et al. (1999) assessed the health status of 686 Danish peacekeepers deployed to the Persian Gulf during 1991-1996 and compared them with 231 nondeployed military personnel. All study participants underwent a health examination in 1997-1998. The deployed veterans had significantly (p < 0.001) more repeated headaches (19.2% vs 6.5%), balance disturbances or dizziness (13.65 vs 3.9%), concentration or memory difficulties (31.2% vs 8.2%), abnormal fatigue (26.4% vs 10.8%), sleep problems (19.8% vs 6.9%), and feeling nervous, irritable, or agitated (21.0% vs 9.1%) with onset during or after deployment than the nondeployed controls. Some surveys have asked veterans about more than 95 symptoms and found that for all symptomsâranging from the less severe, such as loss of appetite, to the more severe, such as chest infections and abscessesâdeployed veterans report having more symptoms and being more troubled by the symptoms than nondeployed veterans (Cherry et al. 2001a). In a telephone survey conducted in 2000-2002, 674 World War II veterans, 983 Korean War veterans, 1420 Vietnam War veterans, and 137 Gulf War veterans living in southern California and Nevada were asked about their health and activities of daily living (Villa et al. 2002). World War II and Korean veterans reported the best mental health, but World War II veterans were also more likely to report their health status as poor and to have more impairment in activities of daily living than the other veteran groups even when socioeconomic status, disease prevalence, and mental-health status were held constant. Vietnam veterans, however, were more likely than Korean War or Gulf War veterans to report difficulty in performing activities of daily living. In a comprehensive review, Barrett et al. (2002b) examined symptom prevalence in various Gulf War veteran populations. They concluded that reports of symptoms were higher in Gulf War veterans than in controls but the pattern of symptoms being reported was neither abnormal for, nor peculiar to, a veteran population. The primary precipitating event for increased reporting of adverse health effects after deployment is combat experience. Combat stressors were discussed in Chapter 3. McFarlane (1997) reported on World War II veterans who were followed until the age of 65 years for PTSD. They found that men who had experienced heavy combat had died earlier, and this finding was independent of PTSD. Of the men who experienced heavy combat, 56% were dead or chronically ill by the age of 65 years compared with only 39% of the men who had not experienced heavy combat (Lee et al. 1995). Increasing combat exposure results in an increase in the reporting of stress-related symptoms (Sutker et al. 1993b; Wolfe et al. 1998). Kulka et al. (1990) found that male Vietnam veterans who had experienced high war- zone stress were twice as likely to report their physical health as poor as were veterans who experienced low war-zone stress (25% vs 13%). In a study of Australian Vietnam veterans; however, OâToole et al. (1996b) did not find a significant increase in symptoms, signs, and ill-
250 GULF WAR AND HEALTH defined conditions with increasing combat exposure (p = 0.029). The impact of combat on PTSD and other psychiatric disorders is discussed in Chapter 5. PTSD and General Symptom Reporting PTSD is associated with increased reports of poor physical health in veterans and civilian populations, regardless of how physical health is determined, that is, through physical examination or self-reports (Baker et al. 1997; Barrett et al. 2002a; Beckham et al. 1998; Schnurr and Jankowski 1999; Sloan et al. 2005). That has been seen in studies of male and female Vietnam veterans (Beckham et al. 1998; Boscarino 1997; Kulka et al. 1990; Taft et al. 1999; Zatzick et al. 1997a,b), Gulf War veterans (Baker et al. 1997), and World War II and Korean War veterans in the VA Normative Aging Study (Schnurr and Spiro 1999). Several researchers have attempted to identify links between war-zone exposures and physical health of veterans (Friedman et al. 1995; Taft et al. 1999; Wolfe et al. 1994). In each case, PTSD was the major mediator between war-zone exposure and poor physical health; the presence of PTSD was a better predictor of poor health than was being in a war zone. People with PTSD also tend to engage in poor behavioral practices, such as increased alcohol consumption and smoking, which in turn put them at risk for other health problems (Friedman et al. 1995). Asmundson et al. (2002) studied the effects of PTSD on health in 1187 Canadian men deployed to war zones for UN peacekeeping missions and compared them with the health of 669 Canadians who had served in the military but had never been deployed. According to the PTSD Checklist-Military Version, 11% of the deployed and 3% of the nondeployed troops met the screening criteria for current PTSD. Those with PTSD had more self-reported poor health than those without PTSD regardless of deployment status. PTSD symptoms also contributed to depression, which in turn, resulted in even more poor health. PTSD was also predictive of alcohol use, however, the latter, unlike depression, was not associated with poorer health beyond that associated with symptoms of PTSD alone. In a study of 107 Harvard graduates who fought in World War II discussed above (Lee et al. 1995), symptoms of PTSD in both 1946 and 1988 were correlated significantly (p < 0.001) with combat exposure and with the number of physical symptoms experienced by the veterans during their combat exposure; however, the PTSD symptoms were not associated with any premorbid vulnerabilities, such as low socioeconomic status or childhood emotional problems. Combat exposure was also a predictor of later poor health: 59% of those with both heavy combat and PTSD were chronically ill or dead by the age of 65 years, compared with only 39% of those without heavy combat experience. World War II veterans (70-74 years old) who had participated in secret military tests of mustard gas during the war were assessed for current PTSD in 1996 using the PTSD Checklist (Schnurr et al. 2000). Veterans with PTSD had significantly higher rates of the following self- reported illnesses than veterans without PTSD: coronary heart disease, pulmonary disease, dermatologic conditions, ophthalmologic diseases, GI disorders, sexual dysfunction, and urologic disorders. Veterans with PTSD also had greater pain and fatigue, greater impairment in physical and psychosocial functioning, and were more likely to have lifetime disability, including lifetime VA psychiatric disability. Health care use was also significantly higher for veterans with PTSD. The VA Normative Aging Study, begun in 1963, consists of 2280 men, 95% of whom are World War II or Korean War veterans. At study entry, 84% of the veterans reported combat exposure; in 1990, using the Combat Exposure Scale, 79% reported combat exposure. In 1990,
HEALTH EFFECTS 251 Schnurr et al. (2000) screened 605 of the veterans (98% white and 85% World War II veterans) for PTSD symptoms with the Mississippi Scale for Combat-Related PTSD. They found that 1% of the study participants had a score on the scale that indicated symptoms of PTSD. More combat exposure was correlated with more PTSD symptoms, and both combat and PTSD symptoms were correlated with increased self-reports of poor physical and mental health (Schnurr and Spiro 1999). When PTSD symptoms were associated with various physician- diagnosed health outcomes, a 10-point increase in PTSD symptom scores was found to increase the risk of having an arterial disorder by 27%, a lower GI disorder by 23%, a dermatologic disorder by 18%, and a musculoskeletal disorder by 9%. Increases in arterial, pulmonary, and upper GI disorders, but decreases in onset of other heart disorders, were associated with increased combat exposure. No association was found between PTSD and cancer, genitourinary disorders, or endocrine disorders (Boscarino 1997; Schnurr et al. 2000). The ambulatory care veterans from the VA Veterans Health Study who screened positive for PTSD were found to have significantly more chronic lower back pain than veterans without PTSD (OR 2.85, 95% CI 2.25-3.63, p â¤ 0.05) adjusted for age and for depression (Spiro et al. 2005). Women with PTSD have a greater risk of poor health status than do women without PTSD. Using NVVRS data, Zatzick et al. (1997b) found that female veterans with PTSD reported poorer health status and well-being, had more days in bed, were more likely not to be working currently, and had more limitations in physical functioning than female veterans without PTSD. In a mailed survey of female veterans who attended a VA medical facility, those who screened positive for PTSD (n = 266) using the PTSD Checklist-Civilian Version were significantly more likely than those without PTSD (n = 940) to also screen positively for a drinking problem (OR 1.68, 95% CI 1.22-2.30), a drug problem (OR 3.56, 95% CI 2.36-5.37), being a victim of domestic violence (OR 2.58, 95% CI 1.92-3.46), and various psychiatric disorders, including panic disorder and major depression. Female veterans who screened positive for PTSD were also more likely to have several self-reported medical problems, including fibromyalgia, stroke, irritable bowel syndrome, chronic pelvic pain, premenstrual syndrome, and polycystic ovary disease (all ORs > 2.0, 95% CIs > 1.0) (Dobie et al. 2004). Unexplained Illness Unexplained illnesses have been described primarily in the Gulf War veteran literature. As discussed above, veterans who were deployed to the Persian Gulf region report more symptoms than their nondeployed counterparts. The numerous symptoms or clusters of symptoms have been referred to by a variety of terms, such as Gulf War syndrome, chronic multisymptom illness, and âunexplainedâ illness. They are âunexplainedâ not in the sense that they are of unknown etiology (which is true of many medical conditions) but rather in the sense that they do not fit into established medical diagnostic categories (IOM 2006). The ICD includes a category âunknown and unspecified causes of morbidity,â R69, that might be appropriate for this health effect. Several studies for unexplained illness that met the committeeâs criteria for primary studies because they did not rely on self-reports are included in Table 6-14. The committee identified eight studies of unexplained illness in Gulf War veterans beginning with a study by Fukuda et al. (1998) that established the CDC case definition for chronic multisymptom illness. In response to a request from DoD, VA, and the Commonwealth of Pennsylvania, Fukuda et al. (1998) assessed the health status of Air Force veterans who had been deployed to the Gulf War. Their focus was to assess the prevalence and causes of an unexplained illness in members of one Air National Guard unit. They administered a 35-item
252 GULF WAR AND HEALTH symptom inventory that included symptom severity (mild, moderate, or severe) and duration (less than 6 months or 6 months or longer) and randomly divided the 3255 participants who had answered all symptom questions into subsamples of 1631 and 1624. They used factor analysis to organize symptoms into a case definition for the CDC. The case definition consisted of having one or more chronic symptoms (present for 6 months or longer) in each of at least two of three categories: fatigue, mood-cognition (symptoms of feeling depressed, difficulty in remembering or concentrating, feeling moody, feeling anxious, trouble in finding words, and difficulty in sleeping), and musculoskeletal (symptoms of joint pain, joint stiffness, and muscle pain). A case was classified as severe if each reported symptom that was used to meet the case definition was rated as severe. Of the survey participants, those deployed to the Gulf War experienced a higher prevalence of chronic symptoms than nondeployed veterans (33 of 35 symptoms with more than 6-month duration were reported to be more prevalent). According to the case definition, 39% of Gulf War-deployed veterans and 14% of nondeployed veterans had mild to moderate cases, and 6% and 0.7%, respectively, had severe cases. On the basis of a total of 158 clinical examinations performed in one Air National Guard unit, there were no abnormal physical or laboratory findings that differentiated those who met the case definition from those who did not meet the case definition. Case subjects, however, reported significantly lower functioning and well-being. Because such a large fraction (14%) of nondeployed veterans met the definition of mild to moderate cases, the investigators concluded that the case definition could not specifically characterize Gulf War veterans who had unexplained illnesses (Fukuda et al. 1998). The study has several limitations, including the fact that its coverage of only current Air Force personnel several years after the Gulf War makes it difficult to generalize its results to other branches of service and to those who might have left the service because of illness. The use of self-reported symptoms introduced the possibility of reporting bias, and the low participation rates in two of the four units (62% and 35%) introduced the possibility of selection bias. Nonetheless, symptom reporting and prevalence were similar among the four units. A particular strength of the study was its use of a symptom inventory rather than asking veterans about specific diagnoses, such as CFS, multiple chemical sensitivity, depression, and various neurologic abnormalities. Its use of a more intensive examination of Gulf War veterans from one unitâincluding an additional clinical questionnaire, a variety of laboratory tests, and interviewer-administered modules on major depression, somatization disorder, and panic disorderâprovided important additional data even though participation rates were low (62%). A nested case-control secondary study of the Fukuda et al. (1998) cohort (n = 1002) sought to identify self-reported exposures associated with cases of chronic mutisymptom illness (Nisenbaum et al. 2000). Having an injury requiring medical attention was associated with having a severe case of chronic mutisymptom illness. Symptom clustering in the Fort Devens cohort was studied in 1997 with CDC's case definition of chronic mutisymptom illness (Wolfe et al. 2002). The case definition was applied to findings from use of the 52-item health checklist. About 60% of respondents met the CDC case definition. That group was divided between âmild to moderateâ and âsevereâ cases. Both Nisenbaum et al. (2000) and Wolfe et al. (2002) found that many Gulf War exposuresâincluding exposure to pyridostigmine bromide, anthrax vaccination, tent-heater exhaust, oil-fire smoke, and chemical odorsâand psychologic distress such as fear of a chemical attack, were associated with meeting the case definition of chronic mutisymptom illness.
HEALTH EFFECTS 253 In a primary study, Blanchard et al. (2006) sought to determine the prevalence of unexplained illnesses in Gulf War veterans 10 years after the Gulf War. The study applied CDCâs definition of chronic multisymptom illness (Fukuda et al. 1998). Data were collected from 1035 deployed veterans and 1116 nondeployed veterans. Participants were asked about symptoms in face-to-face interviews, and those who reported at least one symptom in each of three clustersâfatigability, mood and cognition, and musculoskeletalâwere considered to meet the case definition. Cases were classified as severe if at least one symptom in each cluster was rated as severe. The investigators found that overall 29% of deployed participants and 16% of nondeployed participants met the criteria for chronic mutisymptom illness (OR 2.16, 95% CI 1.61-2.90); deployed veterans were more likely than nondeployed veterans to have severe chronic mutisymptom illness (OR 4.65, 95% CI 2.27-9.52), and among deployed veterans it was associated with a higher score on the Combat Exposure Scale (p < 0.001). Both deployed and nondeployed veterans who met the case definition had lower mean scores on the SF-36 for physical and mental health, more nonroutine clinic visits, more prescriptions, and were more likely to be using psychotropic medications, than deployed or nondeployed veterans without chronic mutisymptom illness. Veterans with chronic mutisymptom illness were also more likely than veterans without it to also have fibromyalgia, CFS, symptomatic arthralgia, dyspepsia, metabolic syndrome, PTSD, anxiety disorders, major depression, nicotine dependence, alcohol dependence (deployed veterans only), and more than one psychiatric diagnosis during the year preceding the examination. The study was limited by low participation rates and the self- reporting of symptoms, but it provided evidence that the cluster of symptoms used to define chronic mutisymptom illness persisted in Gulf War veterans 10 years after the war. Ozakinci et al. (2006), in a secondary study, also investigated widespread symptomatic illness in Gulf War veterans. Participants were identified from the VA Gulf War Health Registry, contacted twice (in 1995 and 2000), and asked to respond to symptom survey questionnaires. Statistical analyses were conducted to assess changes in symptoms over time. No significant changes were found in the cohort in symptom number or severity. Thus, Gulf War Health Registry veterans 10 years after deployment continued to experience significant symptoms. Limitations of the study include the problem of generalizability to all Gulf War veteransâthe study included only veterans in a registry, and there was no nondeployed comparison groupâ and the possibility of reporting bias because of the self-reporting nature of the questionnaire. Self-reports of health status in Australian Vietnam veterans 20-25 years after the war, and the impact of combat exposure, were investigated. OâToole et al. (1996b) used the Australian Bureau of Statistics Health Interview Survey questionnaire to interview in-person a random sample of Army veterans posted to Vietnam during 1964-1972 and compared the veterans with the general Australian population. A 21-item combat exposure index was also used to measure the relationship between combat and physical health. Combat exposure was related to recent but not chronic symptoms, signs, and ill-defined conditions. Comparison of Australian Vietnam Veterans with the Australian population for prevalence of self-reported symptoms, signs, and ill- defined conditions, adjusted for response bias, yielded an RR of 2.77 (99% CI 1.88-3.66). In the first of three primary hospitalization studies, Gray et al. (1996) used a retrospective cohort and data from DoD hospitals for the period 1991-1993. The study examined hospitalizations of 547,076 Army, Navy, Marine Corps, and Air Force Gulf War veterans and 618,335 nondeployed era veterans. Multivariate logistic-regression analyses yielded ORs of less than 1 for ill-defined conditions for all 3 years. The authors noted that in an attempt to reduce the
254 GULF WAR AND HEALTH potential bias due to attrition, only members of the study population who remained on active duty for at least half the study period were included in the multivariate models. A second hospitalization study expanding on Gray et al. (1996) study compared the postwar records of Gulf War-deployed veterans (n = 552,111) and nondeployed veterans (n = 1,479,751) from DoDâs hospital-discharge data from August 1991 to April 1996 to search for admissions for unexplained illnesses in military hospitals (Knoke and Gray 1998). The study defined unexplained illnesses as diagnoses in 77 ICD-9 diagnostic categories that comprised ill- defined conditions. The study examined only first hospitalizations to avoid overcounting medical conditions that required repeated hospitalizations of the subset of patients who had at least one unexplained illness coded on a discharge summary. Up to eight discharge diagnoses were examined per hospitalization. The authors found that deployed active-duty military members were less likely to have been hospitalized for unexplained illnesses than nondeployed (RR 0.93, 95% CI 0.91-0.96) (Knoke and Gray 1998). That finding included adjustment for a variety of covariates and removed the effect of participation in the CCEP after June 1994. Participants in the CCEP were more likely to have been hospitalized only for evaluation. This study has the advantage of a large sample that allowed detection of even minimal effects. Its major limitations are its inclusion of only active-duty personnel and its inability to detect illnesses that did not warrant hospitalization. Although the previous studies demonstrated no increase in unexplained illness among active-duty Gulf War veterans, Gray et al. (2000) sought to expand their investigation of Gulf War veterans to include reserve and separated military personnel who may not have been treated in a DoD facility. They investigated hospitalization data from DoD, VA, and nonfederal hospitals in California for 1991-1994. Hospitalization rates could not be directly compared among the three sources, because of the unreliability of state-of-residence data in DoD and VA datasets. Therefore, PMRs of hospital-discharge diagnoses (14 diagnostic categories from ICD-9) were compared for deployed vs nondeployed veterans. For VA hospitals, but not for DoD or California hospitals, the PMR for the ICD code of symptoms, signs, and ill-defined diseases was increased for deployed vs nondeployed veterans (PMR 1.24, 95% CI 1.16-1.33). A previous IOM report (IOM 2006) has carefully described and evaluated studies of unexplained illness and increased symptom self-reporting. Cluster or factor analysis has been used by several researchers (for example, Fukuda et al. 1998) to determine whether the many symptoms reported constituted a new syndrome or a variant of an existing syndrome. However, the 2006 IOM report concluded that outcomes based primarily on symptoms or self-reports constituted âno unique syndrome, unique illness, or unique symptom complex in deployed Gulf War veterans. Veterans of the Gulf War report higher rates of nearly all symptoms or sets of symptoms than their nondeployed counterparts; 29% of veterans meet a case definition of âmultisymptom illness,â as compared with 16% of nondeployed veteransâ (IOM 2006). Posttraumatic Stress Disorder and Unexplained Illness Two studies of PTSD and unexplained illness were identified. In a nested case-control study drawn from a large, population-based study of Iowa veterans of the Gulf War, Barrett et al. (2002a) investigated the relationship between PTSD and perceived physical health. Of the 53 veterans who screened positive for PTSD (37 deployed and 16 nondeployed), over 50% had symptoms that corresponded with ill-defined conditions according to the ICD-9 compared with less than 10% of the 3629 veterans without PTSD. The study was conducted by telephone interview 5 years after the Gulf War. The prevalence of PTSD among the Gulf War-deployed
HEALTH EFFECTS 255 veterans was 3.4% and 1.4% for those who had participated in combat and for those who did not, respectively. The study is limited by the inclusion of both deployed and nondeployed veterans in the PTSD-positive group and the lack of specific data on the prevalence of various health outcomes. A study of posttraumatic stress symptomatology (PTSS) and unexplained illness was conducted by Ford et al. (2001). They sought to determine whether there was an association between war-zone trauma or PTSS and illnesses reported by Gulf War veterans. Participants were randomly selected from a DoD database of 8603 eligible Gulf War veterans from Oregon or Washington who were deployed from August 1, 1990, through July 31, 1991. Of those deemed eligible and who completed questionnaires, 237 cases and 113 controls were identified by medical examination. A 4-hour test battery of 19 tests was administered to assess psychologic status and neurobehavioral function. Findings indicate that 13 of the 14 psychologic variables were significantly associated with case (vs control) status in unadjusted univariate logistic regression analyses. Case subjects reported significantly higher levels of somatic distress; health problems, fatigue, pain, and deterioration in physical health; global and specific psychiatric distress; negative effects of recent life events; and war-zone trauma exposure than controls. The study is limited by its retrospective analysis of war trauma and its lack of representativeness of the entire Gulf War veteran population. Chronic Pain Chronic pain, defined as âan unpleasant sensory and emotional experience, associated with actual or potential tissue damage or described in terms of such damageâ that persists for 6 months or longer (Otis et al. 2003), is one of the most frequently reported symptoms in veteran populations. Such unexplained pain does not help the body to prevent injury. It can persist for weeks to years as pain signals continue to stimulate the nervous system. Common chronic pain complaints include headache, low-back pain, joint pain, neurogenic pain (pain resulting from damage to the peripheral nerves or to the central nervous system), and psychogenic pain (pain not due to disease, injury, or any visible sign of damage inside or outside the nervous system). Chronic pain is different from CWP (discussed earlier in the chapter with fibromyalgia) because it does not meet the ACR criteria necessary for a diagnosis of CWP. It has been estimated that 50 million Americans have serious chronic pain annually (American Pain Foundation 2007). Studies of various types of unspecified and unexplained pain in deployed and nondeployed veteran groups invariably find that deployed veterans report significantly more pain symptomsâincluding joint pain, backache, chronic back pain, muscular pain, neck ache, neuralgia, and headacheâthan nondeployed troops (Gray et al. 1999; Kang et al. 2000b; Kelsall et al. 2006; Kuzma and Black 2006; Proctor et al. 1998, 2001; Simmons et al. 2004; Unwin et al. 1999). In a study of 970 OEF and OIF veterans seeking treatment at a VA medical center, 38% reported some level of pain; of those, 59% had pain that was clinically significant and likely to interfere with functional activities (Gironda et al. 2006). In over half the patients with chronic pain, the pain could not be attributed to any type of injury. Thomas et al. (2006) conducted a meta-analysis of 20 studies that compared self-reports of pain in Gulf War-deployed veterans with era veterans. They found that deployment was most strongly associated with abdominal pain (six studies, OR 3.23, 95% CI 2.31-4.51), but deployment was also associated with reports of other pain, including muscle pain (eight studies, OR 3.06, 95% CI 2.18-4.30), joint pain (12 studies, OR 2.81, 95% CI 2.31-3.42), chest pain (seven studies, OR 2.52, 95% CI 2.23-2.85), and back pain (six studies, OR 1.58, 95% CI 1.23-
256 GULF WAR AND HEALTH 2.04). The authors commented on the potential effects of the survey procedures, response bias, symptom measurement, and confounding in each study. They noted that although the methodologic quality of the 20 studies varied considerably, all but one of the studies found more self-reports of pain in Gulf War-deployed than in nondeployed veterans. The DSM-IV also has a category of somatoform disorders that includes pain disorder that has been studied in some veterans. Toomey et al. (2007) found that Gulf War veterans had substantially more pain disorders, as diagnosed with the CIDI using the DSM-IV criteria, than did nondeployed veterans (OR 91.66, 95% CI 10.52-798.21). Ikin et al. (2004) interviewed a random sample of Australian Gulf War veterans after the war to determine the prevalence of various psychiatric disorders using the CIDI. They found that deployed veterans (n = 1381) were no more likely than nondeployed veterans (n = 1377) to have any somatic disorder (OR 2.6, 95% CI 1.0-6.3) but not pain disorder (OR 1.4, 95% CI 0.2-16.4). ORs were adjusted for service type, rank, and age; for any somatic disorder, the OR was also adjusted for education and marital status. Combat exposure has been linked to somatoform pain disorder (OâToole et al. 1996a). The ORs for a current (1-month) diagnosis of somatoform pain disorder in Australian Vietnam veterans, based on the DIS, compared with each quartile increase in combat exposure (based on a 21-item combat index) were 1.00, 1.76, 3.07, and 5.08 (p < 0.0005); ORs for lifetime somatoform pain disorder and increasing combat exposure were 1.00, 1.05, 1.88, and 2.47 (p < 0.001). Posttraumatic Stress Disorder and Chronic Pain Although chronic pain is common in deployed veterans, veterans with PTSD are at particular risk. Chronic pain is one of the most commonly reported physical complaints of people (veterans and nonveterans) who have PTSD (McFarlane et al. 1994). And PTSD is common in people who have chronic pain as the result of an accident or trauma (Otis et al. 2003). Several studies that have examined the relationship between PTSD and chronic pain in veterans are discussed below. Asmundson et al. (2004) reported on a sample of 221 female veterans who used a VA Health Center clinic in 1998-1999 for general health purposes. The women were identified as having PTSD on the basis of responses to a mailed questionnaire that included the PTSD Checklist-Civilian Version, the SF-36, and two additional questions about pain in the preceding 6 months. Female veterans with PTSD reported significantly greater painâincluding bodily pain, pain interference, severe headache or migraine, and back painâthan did those without PTSD. Beckham et al. (1997) investigated patterns of chronic pain in Vietnam veterans with PTSD. Of 129 combat veterans with PTSD, 80% reported chronic pain as determined by the Pain Disability Index, the McGill Pain Questionnaire, the Visual Analog Scale, and a pain drawing; PTSD was diagnosed with the CAPS. Combat veterans with PTSD and chronic pain reported significantly higher somatization, as measured by the hypochondriasis and hysteria subscales of the Minnesota Multiphasic Personality Inventory-2, than did combat veterans with PTSD but without chronic pain. In the sample of 103 veterans with PTSD and chronic pain, Minnesota Multiphasic Personality Inventory-2 hypochondriasis scores and PTSD symptoms from the re- experiencing symptom cluster were significantly related to pain disability, overall pain index, and current pain level.
HEALTH EFFECTS 257 Benedikt and Kolb (1986) examined case histories of 225 veterans who were referred to a VA pain clinic for the treatment of chronic pain between 1978 and 1984. Of the 225 patients, 22 later received a diagnosis of PTSD (two were World War II veterans, and 20 were Vietnam veterans); this suggests a high prevalence of PTSD (10%) in patients who have chronic pain. Hoge et al. (2007) found that soldiers who screened positive for PTSD (n = 468) a year after their return from combat duty in Iraq reported more pain symptoms than those negative for PTSD (n = 2347). Half the soldiers with PTSD indicated that they were âbothered a lotâ by pain in their arms, legs, or joints (OR 2.89, 95% CI 2.35-3.57); 40% were bothered by back pain (OR 3.36, 95% CI 2.72-4.16); and almost 32% had headaches (OR 4.25, 95% CI 3.32-5.42), compared with 26%, 22%, and 10%, respectively, of soldiers without PTSD. Several other studies report that Gulf War veterans with PTSD had more pain symptoms than did veterans without PTSD (Engel et al. 2000; Spiro et al. 2006). It has been reported that one-fifth of U.S. Army soldiers returning from OIF have a diagnosis of migraines and that this group has nearly twice as high a risk of depression, PTSD, and other psychiatric disorders as returning soldiers who do not have migraines. When assessed within 90 days of their return from a 1-year tour of duty, 19% of the 2200 veterans had migraines, 32% had depression, 22% had PTSD, and 13% had anxiety. Of those with migraines, 50% were also depressed, 39% also had PTSD, and 22% also had anxiety disorders compared with 27%, 18%, and 10% of those who did not have migraines (Erickson and Diamond 2007). Those studies suggest that PTSD and chronic pain are frequently comorbid and that each disorder has the potential to exacerbate the symptoms of the other (Otis et al. 2003). Summary and Conclusions Male and female veterans who have been deployed to a war zone, regardless of the war in which they served, report more symptoms and poorer health than do their nondeployed counterparts. Symptoms range from severe, such as chest pain and numbing in the extremities, to minor, such as loss of appetite. Combat exposure was associated with increased number and severity of symptoms. The committee identified eight studies that assessed the prevalence of unexplained illness in Gulf War veterans compared with nondeployed veterans and found mixed results. Some researchers have attempted to cluster the symptoms into new diseases but in general the symptoms are too broad and nonspecific to suggest the presence of a new illness specific to the Gulf War (see IOM 2006). Fukuda et al. (1998) developed the case definition for chronic multisymptom illness based on an Air Force unit deployed to the Gulf War. A later study by Blanchard et al. (2006) found that deployed veterans reported higher rates of nearly all symptoms or sets of symptoms than their nondeployed counterparts 10 years after the war. Four other studies also found higher rates of unexplained illness in deployed Gulf War veterans than in nondeployed Gulf War veterans, but the use of self-reported symptoms introduced the possibility of reporting bias, and the low participation rates in some of the studies introduced the possibility of selection bias. The three hospitalization studies did not find a consistent association between deployment and a diagnosis of unexplained illness. Although the Knoke and Gray (1998) and Gray et al. (1996) hospitalization studies did not demonstrate an increase in unexplained illness in active-duty Gulf War veterans, the Gray et al. (2000) study reviewed the hospital discharge diagnoses of Gulf War veterans and found an increase in hospitalization for unexplained illness
258 GULF WAR AND HEALTH in only one of the three hospital systems examined. The committee notes that symptoms of unexplained illness may not be severe enough to require hospitalization in many cases. Chronic pain of unexplained origin is one symptom reported more frequently by deployed than nondeployed Gulf War veterans. A meta-analysis of 20 studies of Gulf War veterans found that deployment was most commonly associated with abdominal pain, but back, joint, and muscle pains were also frequently reported. One study of pain in Australian Vietnam veterans found that somatoform pain disorder, a psychiatric diagnosis, was associated with combat exposure. PTSD is also associated with increased reporting of symptoms, medical conditions, and poor health in veterans, both male and female. Self-reports of more health problems in veterans with PTSD than in those without PTSD have also been confirmed by physical examination. Although combat exposure is associated with increased symptom reporting, PTSD appears to be an even stronger predictor of reports of poor health. PTSD is widely associated with self-reports of pain in both civilian and veteran populations. Veterans of World War II, the Vietnam War, the Gulf War, and OIF and OEF who have PTSD all report more chronic pain than veterans without PTSD. The chronic pain is not specific; back pain, headaches (including migraines), and joint pain are all reported frequently by both male and female veterans. The studies cited above have important limitations. Researchers used different terminology for symptoms, and definitions or descriptions of what was meant by the symptom, such as backache or cough, are lacking. Some asked veterans to indicate how much a symptom bothered them over a specific period; others simply asked for an indication of whether the veterans had ever had the symptom after the war. The onset and duration of the symptom were not always assessed or reported. For the studies of PTSD, the veterans were often not diagnosed but only screened for the disorder. And none of the studies indicated whether the veterans had had any of the symptoms before the war in which they were deployed. All those factors make it difficult to compare symptoms among studies. The committee concludes that there is limited but suggestive evidence of an association between deployment to a war zone and increased symptom reporting, unexplained illness, and chronic pain.
TABLE 6-14 Symptom Reporting Reference Study Design Population Outcomes Results Adjustments Comments General symptoms and signs CDC 1988b Retrospective 2490 Vietnam- Perceived health Health status poor or fair Age at enlistment, Low participation rate cohort, theater veterans, status; somatic OR 1.9, 95% CI race, year of in control group, CI VES prevalence, 1972 Vietnam-era symptom; neurologic excludes 1.0, p < 0.05; enlistment, enlistment not given population-based, veterans randomly examination somatic symptoms OR status (volunteer vs telephone selected from 7924 1.7, 95% CI excludes draftee), score on interview theater veterans and 1.0, p < 0.05; general technical test, followed by 7364 era veterans peripheral neuropathy primary military screening medical who had entered symptoms OR 1.5, 95% occupational specialty examination Army in 1965-1971 CI 1.0-2.2 Ishoy et al. 1999 Cross-sectional 686 Danish Health examination by Prevalence: Participation rate peacekeepers physician, including repeated headaches 83.6% deployed, deployed to gulf in lung function and self- 19.2% vs 6.5%; 57.8% nondeployed 1990-1997 vs 231 report questionnaire balance disturbances or age- and sex- dizziness 13.65% vs matched armed 3.9%; concentration or forces nondeployed memory difficulties controls 31.2% vs 8.2%; abnormal fatigue 26.4% vs 10.8%; sleep problems 19.8% vs 6.9%; feeling nervous, irritable, or agitated 21.0% vs 9.1% Unexplained illness Blanchard et al. Cross-sectional 1035 GW-deployed CMI determined by CMI (all cases) OR 2.16, Age, sex, race, 2006 veterans, 1116 medical examination 95% CI 1.61-2.90; mild education, duty type, nondeployed in 1999-2001 to moderate cases OR service branch, rank, veterans 1.92, 95% CI 1.41-2.63; income, combat severe cases OR 4.65, exposure score, 95% CI 2.27-9.52 Khamisiyah exposure, psychiatric conditions and other conditions diagnosed before GW 259
TABLE 6-14 Symptom Reporting 260 Reference Study Design Population Outcomes Results Adjustments Comments Gray et al. 1996 Retrospective 547,076 active-duty Hospital-discharge All results for each year Prewar ORs statistically cohort, GW veterans, diagnoses for ill- NS hospitalization, sex, significantly below 1, hospitalization 618,335 non-GW defined conditions for age, race, service but no values given; veterans 1991-1993 branch, marital status, no separation of rank, length of specific illnesses service, salary, occupation Knoke and Gray Cross-sectional, 552,111 GW- 10 most common Hospitalizations for all Race, rank, salary, Did not separate 1998 military deployed active-duty unexplained illnesses unexplained illnesses RR branch of service, analyses for specific hospitalization veterans, 1,479,751 1.06, 95% CI 1.03-1.09; health-care worker illnesses; adjusted for nondeployed active- when adjusted for CCEP status, prewar participation in CCEP duty veterans participation RR 0.93, hospitalization, sex after June 1994 followed from 1991 95% CI 0.91-0.96 to March 1996 Gray et al. 2000 Retrospective 652,979 GW Hospital-discharge DoD PMR 0.97, 95% CI Age, sex, race Able to assess only cohort, veterans, 652,922 diagnoses for 0.94-1.00 illnesses that resulted hospitalization randomly selected endocrine, nutritional, in hospitalization; non-GW-deployed metabolic disease in VA PMR 1.24, 95% CI possible undetected veterans three hospital systems: 1.16-1.33 confounders DoD, VA, COSHPD CA PMR 1.04, 95% CI 0.87-1.20 NOTE: CCEP = Comprehensive Clinical Evaluation Program, CI = confidence interval, CMI = chronic multisymptom illness, COSHPD = California Office of Statewide Health Planning and Development, DoD = Department of Defense, GW = Gulf War, NS = not significant, OR = odds ratio, PMR = proportional morbidity ratio, RR = relative risk, VA = Department of Veterans Affairs, VES = Vietnam Experience Study.
HEALTH EFFECTS 261 REFERENCES Adams DP, Barton C, Mitchell GL, Moore AL, Einagel V. 1998. Hearts and minds: Suicide among United States combat troops in Vietnam, 1957-1973. Social Science and Medicine 47(11):1687-1694. American Lung Association. 2007. American Lung Association Site. [Online]. Available: http://www.lungusa.org [accessed July 29, 2007]. American Pain Foundation. 2007. American Pain Foundation: A United Voice of Hope and Power Over Pain. [Online]. Available: http://www.painfoundation.org/ [accessed August 1, 2007]. Amital D, Fostick L, Polliack ML, Segev S, Zohar J, Rubinow A, Amital H. 2006. Posttraumatic stress disorder, tenderness, and fibromyalgia syndrome: Are they different entities? Journal of Psychosomatic Research 61(5):663-669. Anderson R, Minino A, Hoyert D, Rosenberg H. 2001. Comparability of cause of death between ICD-9 and ICD-10: Preliminary estimates. National Vital Statistics Reports 49(2):1-32. Ang DC, Peloso PM, Woolson RF, Kroenke K, Doebbeling BN. 2006. Predictors of incident chronic widespread pain among veterans following the first Gulf War. The Clinical Journal of Pain 22(6):554-563. Araneta MR, Destiche DA, Schlangen KM, Merz RD, Forrester MB, Gray GC. 2000. Birth defects prevalence among infants of Persian Gulf War veterans born in Hawaii, 1989-1993. Teratology 62(4):195-204. Araneta MR, Schlangen KM, Edmonds LD, Destiche DA, Merz RD, Hobbs CA, Flood TJ, Harris JA, Krishnamurti D, Gray GC. 2003. Prevalence of birth defects among infants of Gulf War veterans in Arkansas, Arizona, California, Georgia, Hawaii, and Iowa, 1989-1993. Birth Defects Research 67(4):246-260. Archibald HC, Tuddenham RD. 1965. Persistent stress reaction after combat: A 20-year follow- up. Archives of General Psychiatry 12:475-481. Asmundson GJ, Stein MB, McCreary DR. 2002. Posttraumatic stress disorder symptoms influence health status of deployed peacekeepers and nondeployed military personnel. Journal of Nervous and Mental Disease 190(12):807-815. Asmundson GJ, Wright KD, Stein MB. 2004. Pain and PTSD symptoms in female veterans. European Journal of Pain 8(4):345-350. Axelrod BN, Milner IB. 1997. Neuropsychological findings in a sample of Operation Desert Storm veterans. Journal of Neuropsychiatry and Clinical Neurosciences 9(1):23-28. Baker DG, Mendenhall CL, Simbartl LA, Magan LK, Steinberg JL. 1997. Relationship between posttraumatic stress disorder and self-reported physical symptoms in Persian Gulf War veterans. Archives of Internal Medicine 157(18):2076-2078. Barrett DH, Green ML, Morris R, Giles WH, Croft JB. 1996. Cognitive functioning and posttraumatic stress disorder. American Journal of Psychiatry 153(11):1492-1494.
262 GULF WAR AND HEALTH Barrett DH, Doebbeling CC, Schwartz DA, Voelker MD, Falter KH, Woolson RF, Doebbeling BN. 2002a. Posttraumatic stress disorder and self-reported physical health status among U.S. military personnel serving during the Gulf War period: A population-based study. Psychosomatics 43(3):195-205. Barrett DH, Gray GC, Doebbeling BN, Clauw DJ, Reeves WC. 2002b. Prevalence of symptoms and symptom-based conditions among Gulf War veterans: Current status of research findings. Epidemiologic Reviews 24(2):218-227. Beckham JC, Crawford AL, Feldman ME, Kirby AC, Hertzberg MA, Davidson JR, Moore SD. 1997. Chronic posttraumatic stress disorder and chronic pain in Vietnam combat veterans. Journal of Psychosomatic Research 43(4):379-389. Beckham JC, Moore SD, Feldman ME, Hertzberg MA, Kirby AC, Fairbank JA. 1998. Health status, somatization, and severity of posttraumatic stress disorder in Vietnam combat veterans with posttraumatic stress disorder. American Journal of Psychiatry 155(11):1565- 1569. Beckham JC, Feldman ME, Barefoot JC, Fairbank JA, Helms MJ, Haney TL, Hertzberg MA, Moore SD, Davidson JR. 2000. Ambulatory cardiovascular activity in Vietnam combat veterans with and without posttraumatic stress disorder. Journal of Consulting and Clinical Psychology 68(2):269-276. Beckham JC, Vrana SR, Barefoot JC, Feldman ME, Fairbank J, Moore SD. 2002. Magnitude and duration of cardiovascular responses to anger in Vietnam veterans with and without posttraumatic stress disorder. Journal of Consulting and Clinical Psychology 70(1):228-234. Beckham JC, Taft CT, Vrana SR, Feldman ME, Barefoot JC, Moore SD, Mozley SL, Butterfield MI, Calhoun PS. 2003. Ambulatory monitoring and physical health report in Vietnam veterans with and without chronic posttraumatic stress disorder. Journal of Traumatic Stress 16(4):329-335. Beckham JC, Gehrman PR, McClernon FJ, Collie CF, Feldman ME. 2004. Cigarette smoking, ambulatory cardiovascular monitoring, and mood in Vietnam veterans with and without chronic posttraumatic stress disorder. Addictive Behaviors 29(8):1579-1593. Beebe GW. 1975. Follow-up studies of World War II and Korean War prisoners. II. Morbidity, disability, and maladjustments. American Journal of Epidemiology 101(5):400-422. Benedikt RA, Kolb LC. 1986. Preliminary findings on chronic pain and posttraumatic stress disorder. American Journal of Psychiatry 143(7):908-910. Black DW, Carney CP, Forman-Hoffman VL, Letuchy E, Peloso P, Woolson RF, Doebbeling BN. 2004a. Depression in veterans of the first Gulf War and comparable military controls. Annals of Clinical Psychiatry 16(2):53-61. Black DW, Carney CP, Peloso PM, Woolson RF, Schwartz DA, Voelker MD, Barrett DH, Doebbeling BN. 2004b. Gulf War veterans with anxiety: Prevalence, comorbidity, and risk factors. Epidemiology 15(2):135-142. Blanchard EB, Kolb LC, Prins A, Gates S, McCoy GC. 1991. Changes in plasma norepinephrine to combat-related stimuli among Vietnam veterans with posttraumatic stress disorder. The Journal of Nervous and Mental Disease 179(6):371-373. Blanchard MS, Eisen SA, Alpern R, Karlinsky J, Toomey R, Reda DJ, Murphy FM, Jackson LW, Kang HK. 2006. Chronic multisymptom illness complex in Gulf War I veterans 10 years later. American Journal of Epidemiology 163(1):66-75.
HEALTH EFFECTS 263 Boehmer TK, Flanders WD, McGeehin MA, Boyle C, Barrett DH. 2004. Postservice mortality in Vietnam veterans: 30-year follow-up. Archives of Internal Medicine 164(17):1908-1916. Boscarino JA. 1995. Post-traumatic stress and associated disorders among Vietnam veterans: The significance of combat exposure and social support. Journal of Traumatic Stress 8(2):317-336. Boscarino JA. 1997. Diseases among men 20 years after exposure to severe stress: Implications for clinical research and medical care. Psychosomatic Medicine 59(6):605-614. Boscarino JA. 2004. Posttraumatic stress disorder and physical illness: Results from clinical and epidemiologic studies. Annals of the New York Academy of Sciences 1032:141-153. Boscarino JA. 2005. Posttraumatic stress disorder and mortality among U.S. Army veterans 30 years after military service. Annals of Epidemiology 16(4):248-256. Boscarino JA, Chang J. 1999. Electrocardiogram abnormalities among men with stress-related psychiatric disorders: Implications for coronary heart disease and clinical research. Annals of Behavioral Medicine 21(3):227-234. Buckley TC, Kaloupek DG. 2001. A meta-analytic examination of basal cardiovascular activity in posttraumatic stress disorder. Psychosomatic Medicine 63(4):585-594. Bullman TA, Kang HK. 1996. The risk of suicide among wounded Vietnam veterans. American Journal of Public Health 86(5):662-667. Buske-Kirschbaum A, Geiben A, Hollig H, Morschhauser E, Hellhammer D. 2002. Altered responsiveness of the hypothalamus-pituitary-adrenal axis and the sympathetic adrenomedullary system to stress in patients with atopic dermatitis. Journal of Clinical Endocrinology and Metabolism 87(9):4245-4251. Buske-Kirschbaum A, Kern S, Ebrecht M, Hellhammer DH. 2007. Altered distribution of leukocyte subsets and cytokine production in response to acute psychosocial stress in patients with psoriasis vulgaris. Brain, Behavior, and Immunity 21(1):92-99. Buskila D. 2000. Fibromyalgia, chronic fatigue syndrome, and myofascial pain syndrome. Current Opinion in Rheumatology 12(2):113-123. Caldwell BA, Redeker N. 2005. Sleep and trauma: An overview. Issues in Mental Health Nursing 26(7):721-738. CDC (Centers for Disease Control and Prevention). 1987. Postservice mortality among Vietnam veterans. The Centers for Disease Control Vietnam Experience Study. Journal of the American Medical Association 257(6):790-795. CDC. 1988a. Health status of Vietnam veterans. I. Psychosocial characteristics. The Centers for Disease Control Vietnam Experience Study. Journal of the American Medical Association 259(18):2701-2707. CDC. 1988b. Health status of Vietnam veterans. II. Physical health. The Centers for Disease Control Vietnam Experience Study. Journal of the American Medical Association 259(18):2708-2714. CDC. 1988c. Health status of Vietnam veterans. III. Reproductive outcomes and child health. The Centers for Disease Control Vietnam Experience Study. Journal of the American Medical Association 259(18):2715-2719.
264 GULF WAR AND HEALTH CDC. 2007a. Chronic Fatigue Syndrome: The Revised Case Definition (abridged version). [Online]. Available at: http://www.cdc.gov/cfs/cfsdefinitionHCP.htm. [accessed August 31, 2007]. CDC. 2007b. Suicide Facts at a Glance. [Online]. Summer 2007. Available at: www.cdc.gov/injury [accessed August 31, 2007]. Chang L. 2006. Neuroendocrine and neuroimmune markers in IBS: Pathophysiological role or epiphenomenon? Gastroenterology 130(2):596-600. Cherry N, Creed F, Silman A, Dunn G, Baxter D, Smedley J, Taylor S, Macfarlane GJ. 2001a. Health and exposures of United Kingdom Gulf war veterans. Part I: The pattern and extent of ill health. Occupational and Environmental Medicine 58(5):291-298. Cherry N, Creed F, Silman A, Dunn G, Baxter D, Smedley J, Taylor S, Macfarlane GJ. 2001b. Health and exposures of United Kingdom Gulf War veterans. Part II: The relation of health to exposure. Occupational and Environmental Medicine 58(5):299-306. Cosgrove DJ, Gordon Z, Bernie JE, Hami S, Montoya D, Stein MB, Monga M. 2002. Sexual dysfunction in combat veterans with post-traumatic stress disorder. Urology 60(5):881-884. Cowan DN, DeFraites RF, Gray GC, Goldenbaum MB, Wishik SM. 1997. The risk of birth defects among children of Persian Gulf War veterans. New England Journal of Medicine 336(23):1650-1656. Creed F, Levy R, Bradley L, Fransisconi C, Drossman D, Naliboff B. 2006. Psychosocial aspects of functional gastrointestinal disorders. In: Drossman D, Corazziari E, Delvaux M, Spiller R, Talley N, Thompson W, editors. Rome III: The Functional Gastrointestinal Disorders. 3rd Ed. McLean, VA: Degnon Associates, Inc. Crowell TA, Kieffer KM, Siders CA, Vanderploeg RD. 2002. Neuropsychological findings in combat-related posttraumatic stress disorder. The Clinical Neuropsychologist 16(3):310-321. Dagan Y, Lavie P, Bleich A. 1991. Elevated awakening thresholds in sleep stage 3-4 in war- related post-traumatic stress disorder. Biological Psychiatry 30(6):618-622. Dalager NA, Kang HK, Thomas TL. 1995. Cancer mortality patterns among women who served in the military: The Vietnam experience. Journal of Occupational and Environmental Medicine 37(3):298-305. David AS, Farrin L, Hull L, Unwin C, Wessely S, Wykes T. 2002. Cognitive functioning and disturbances of mood in UK veterans of the Persian Gulf War: A comparative study. Psychological Medicine 32(8):1357-1370. David D, Woodward C, Esquenazi J, Mellman TA. 2004. Comparison of comorbid physical illnesses among veterans with PTSD and veterans with alcohol dependence. Psychiatric Services 55(1):82-85. DeStefano F, Annest JL, Kresnow MJ, Schrader SM, Katz DF. 1989. Semen characteristics of Vietnam veterans. Reproductive Toxicology 3(3):165-173. Dlugosz LJ, Hocter WJ, Kaiser KS, Knoke JD, Heller JM, Hamid NA, Reed RJ, Kendler KS, Gray GC. 1999. Risk factors for mental disorder hospitalization after the Persian Gulf War: U.S. Armed Forces, June 1, 1991-September 30, 1993. Journal of Clinical Epidemiology 52(12):1267-1278.
HEALTH EFFECTS 265 Dobie DJ, Kivlahan DR, Maynard C, Bush KR, Davis TM, Bradley KA. 2004. Posttraumatic stress disorder in female veterans: Association with self-reported health problems and functional impairment. Archives of Internal Medicine 164(4):394-400. Dohrenwend B, Turner J, Turse N, Adams B, Koenen K, Marshal R. 2006. The pychological risks of Vietnam for U.S. veterans: A revisit with new data and methods. Science 313(5789):979-982. Dow BM, Kelsoe JR Jr, Gillin JC. 1996. Sleep and dreams in Vietnam PTSD and depression. Biological Psychiatry 39(1):42-50. Doyle P, Maconochie N, Davies G, Maconochie I, Pelerin M, Prior S, Lewis S. 2004. Miscarriage, stillbirth and congenital malformation in the offspring of UK veterans of the first Gulf war. International Journal of Epidemiology 33(1):74-86. Drossman D. 2002. A biopsychosocial understanding of gastrointestinal illness and disease. In: Feldman M, Friedman LS, Sleisenger MH, editors. Sleisenger and Fordtransâs Gastrointestinal and Liver Diseases. 7th Ed. Philadelphia, PA: W.B. Saunders. Pp. 2372- 2385. Drossman D. 2006a. The functional gastrointestinal disorders and the Rome III process. In: Drossman D, Corazziari E, Delvaux M, Spiller R, Talley N, Thompson W, editors. Rome III: The Functional Gastrointestinal Disorders. 3rd Ed. McLean, VA: Degnon Associates, Inc. Drossman D, Chang L. 2003. Psychosocial factors in the care of patients with GI disorders. In: Yamada T, editor. Textbook of Gastroenterology. Philadelphia, PA: Lippincott-Raven. Pp. 636-654. Drossman D, Ringel Y. 2004. Psychosocial factors in ulcerative colitis and Crohnâs disease. In: Sartor B, Sandborn W, editors. Kirsner's Inflammatory Bowel Disease. 6th ed. London, UK: W.B. Saunders. Pp. 340-356. Drossman DA. 1998. Presidential address: Gastrointestinal illness and the biopsychosocial model. Psychosomatic Medicine 60(3):258-267. Drossman DA. 1999. Mind over matter in the postinfective irritable bowel. Gut 44(3):306-307. Drossman DA. 2006b. The functional gastrointestinal disorders and the Rome III process. Gastroenterology 130(5):1377-1390. Drossman DA, Talley NJ, Leserman J, Olden KW, Barreiro MA. 1995. Sexual and physical abuse and gastrointestinal illness. Review and recommendations. Annals of Internal Medicine 123(10):782-794. Drossman DA, Li Z, Leserman J, Tommey TC, Hu YJB. 1996. Health status by gastrointestinal diagnosis and abuse history. Gastroenterology 110(4):999-1007. Drossman DA, Camilleri M, Mayer EA, Whitehead WE. 2002. AGA technical review on irritable bowel syndrome. Gastroenterology 123(6):2108-2131. Dumitrascu DL, Baban A. 1991. Irritable bowel syndrome complaints following the uprising of December 1989 in Romania. Medicine and War 7(2):100-104. Dunlop SP, Jenkins D, Neal KR, Spiller RC. 2003. Relative importance of enterochromaffin cell hyperplasia, anxiety, and depression in postinfectious IBS. Gastroenterology 125(6):1651- 1659.
266 GULF WAR AND HEALTH Dunphy RC, Bridgewater L, Price DD, Robinson ME, Zeilman CJ, 3rd, Verne GN. 2003. Visceral and cutaneous hypersensitivity in Persian Gulf war veterans with chronic gastrointestinal symptoms. Pain 102(1-2):79-85. Eaton WW, Armenian H, Gallo J, Pratt L, Ford DE. 1996. Depression and risk for onset of type II diabetes. A prospective population-based study. Diabetes Care 19(10):1097-1102. Eisen SA, Goldberg J, True WR, Henderson WG. 1991. A co-twin control study of the effects of the Vietnam War on the self-reported physical health of veterans. American Journal of Epidemiology 134(1):49-58. Eisen SA, Griffith KH, Xian H, Scherrer JF, Fischer ID, Chantarujikapong S, Hunter J, True WR, Lyons MJ, Tsuang MT. 2004. Lifetime and 12-month prevalence of psychiatric disorders in 8,169 male Vietnam War era veterans. Military Medicine 169(11):896-902. Eisen SA, Kang HK, Murphy FM, Blanchard MS, Reda DJ, Henderson WG, Toomey R, Jackson LW, Alpern R, Parks BJ, Klimas N, Hall C, Pak HS, Hunter J, Karlinsky J, Battistone MJ, Lyons MJ. 2005. Gulf War veteransâ health: Medical evaluation of a U.S. cohort. Annals of Internal Medicine 142(11):881-890. Engdahl BE, Eberly RE, Hurwitz TD, Mahowald MW, Blake J. 2000. Sleep in a community sample of elderly war veterans with and without posttraumatic stress disorder. Biological Psychiatry 47(6):520-525. Engel CC Jr, Liu X, McCarthy BD, Miller RF, Ursano R. 2000. Relationship of physical symptoms to posttraumatic stress disorder among veterans seeking care for gulf war-related health concerns. Psychosomatic Medicine 62(6):739-745. Erickson DJ, Wolfe J, King DW, King LA, Sharkansky EJ. 2001. Posttraumatic stress disorder and depression symptomatology in a sample of Gulf War veterans: A prospective analysis. Journal of Consulting and Clinical Psychology 69(1):41-49. Erickson JC, Diamond S. 2007. Presentation, Annual Meeting, American Academy of Neurology. Boston, MA. Escobar JI, Canino G, Rubio-Stipec M, Bravo M. 1992. Somatic symptoms after a natural disaster: A prospective study. American Journal of Psychiatry 149(7):965-967. Falger PR, Op den Velde W, Hovens JE, Schouten EG, De Groen JH, Van Duijn H. 1992. Current posttraumatic stress disorder and cardiovascular disease risk factors in Dutch Resistance veterans from World War II. Psychotherapy and Psychosomatics 57(4):164-171. Fiedler N, Ozakinci G, Hallman W, Wartenberg D, Brewer NT, Barrett DH, Kipen HM. 2006. Military deployment to the Gulf War as a risk factor for psychiatric illness among U.S. troops. British Journal of Psychiatry 188:453-459. Fontana A, Rosenheck R. 1995a. Attempted suicide among Vietnam veterans: A model of etiology in a community sample. American Journal of Psychiatry 152(1):102-109. Fontana A, Rosenheck R. 1995b. An etiological model of attempted suicide among Vietnam theater veterans: Prospective generalization to a treatment-seeking sample. Journal of Nervous and Mental Disease 183(6):377-383. Ford JD, Campbell KA, Storzbach D, Binder LM, Anger WK, Rohlman DS. 2001. Posttraumatic stress symptomatology is associated with unexplained illness attributed to Persian Gulf War military service. Psychosomatic Medicine 63(5):842-849.
HEALTH EFFECTS 267 Forman-Hoffman VL, Carney CP, Sampson TR, Peloso PM, Woolson RF, Black DW, Doebbeling BN. 2005. Mental health comorbidity patterns and impact on quality of life among veterans serving during the first Gulf War. Quality of Life Research: An International Journal of Quality of Life Aspects of Treatment, Care and Rehabilitation 14(10):2303-2314. Forman-Hoffman VL, Peloso PM, Black, DW, Woolson RF, Letuchy EM, Doebbeling BN. 2007. Chronic widespread pain in veterans of the first Gulf War: Impact of deployment status and associated health effects. [Online]. Journal of Pain. Available at: http://www.ampainsoc.org/pub/journal. Freeman T, Roca V, Guggenhehn F, Kimbrell T, Griffin WST. 2005. Neuropsychiatric associations of apolipoprotein E alleles in subjects with combat-related posttraumatic stress disorder. Journal of Neuropsychiatry and Clinical Neurosciences 17(4):541-543. Friedman M, Charney D, Deutch A. 1995. Neurobiological and Clinical Consequences of Stress: from Normal Adaptation to Post-Traumatic Stress Disorder. Philadelphia, PA: Lippincott- Raven Publishers. Fu Q, Heath AC, Bucholz KK, Nelson EC, Glowinski AL, Goldberg J, Lyons MJ, Tsuang MT, Jacob T, True MR, Eisen SA. 2002. A twin study of genetic and environmental influences on suicidality in men. Psychological Medicine 32(1):11-24. Fukuda K, Straus SE, Hickie I, Sharpe MC, Dobbins JG, Komaroff A. 1994. The chronic fatigue syndrome: A comprehensive approach to its definition and study. International Chronic Fatigue Syndrome Study Group. Annals of Internal Medicine 121(12):953-959. Fukuda K, Nisenbaum R, Stewart G, Thompson WW, Robin L, Washko RM, Noah DL, Barrett DH , Randall B, Herwaldt BL, Mawle AC, Reeves WC. 1998. Chronic multisymptom illness affecting Air Force veterans of the Gulf War. Jo