This chapter describes the major cohort studies and their derivative studies used by the Volume 10 committee to provide the evidence for the conclusions presented in Chapter 4. It provides an overview of the major cohort studies of Gulf War veterans, and it describes in detail the populations studied, the methods used to select those populations, and the approaches used to identify the health status of the veterans—including questionnaires, clinical examinations, and laboratory tests. Much of this information was previously described in Volumes 4 and 8 (IOM, 2006b, 2010); however, new studies on established cohorts are discussed where available, and the committee has included a description of three additional Gulf War cohorts that were not discussed in the previous volumes. The findings from the studies described in this chapter are evaluated in Chapter 4 where appropriate.
Cohort studies are important for understanding the health of Gulf War veterans. Some of these cohorts were brought together in the first few years after the Gulf War; others were assembled more recently. The largest studies of Gulf War veterans have been conducted in countries that were members of the Gulf War coalition, including Australia, Canada, Denmark, the United Kingdom (UK), and the United States. Most of the studies compare sizable groups of deployed veterans with groups of nondeployed veterans or with veterans who were deployed to locations other than the Persian Gulf region (for example, Bosnia or Germany). The first publication describing these cohorts is referred to as the reference study.
Some cohorts, once established, led to numerous studies, or multiple publications that examined more detailed questions about Gulf War veterans’ health; the committee refers to those studies as derivatives. A derivative study1 is included and summarized under the original cohort (reference study) from which the study population was drawn. This organization helped the committee identify populations that have been studied and understand which studies were independent of one another; establishing which studies rely on the same population sample is important because it helped the committee avoid double counting when weighing the evidence.
1 Derivative studies may simply be publications reporting additional results, subcohort analyses, or nested case-control studies based on the population described by the reference study.
The cohort studies of Gulf War veterans and their derivative studies have contributed greatly to the understanding of veterans’ health, but they have limitations that are encountered in many epidemiologic studies (see Chapter 2), including lack of representativeness, selection bias, lack of adjustment for potential confounding factors, self-reports of exposures, lack of a diagnosis by a health professional for some health effects, and outcome misclassification.
For each separately assembled cohort included in this chapter, the reference study is described first, followed by a summary of any derivative studies cited in Chapter 4. Cohorts of U.S. veterans are presented first, followed by population-based cohort studies of Australian, British, Canadian, and Danish veterans. The committee identified few new studies published since 2009 for most of the cohorts. The Volume 10 committee included studies on three additional Gulf War veteran cohorts not described in depth in Volumes 4 or 8. However, given that those cohorts have resulted in multiple publications (e.g., Iannacchione et al., 2011) and specifically address concerns identified in the statement of task (e.g., multiple sclerosis and amyotrophic lateral sclerosis), they receive more in-depth consideration in this volume.
Not all derivative studies for a particular cohort study are included in this chapter or in Chapter 4. In many cases, the derivative studies were highly specialized; for example, they reported on family issues, subclinical changes, or treatment outcomes. Table 3-1 at the end of the chapter lists the reference and derivative studies for each cohort cited in Volumes 4 and 8 and in this volume.
Most cohort studies are population-based and drawn from the entire population of veterans (Kang et al., 2000) while others are based on state of residence (Iowa Persian Gulf Study Group, 1997; McCauley et al., 1999a; Steele, 2000), and some are drawn based on military unit or base (Fukuda et al., 1998; Gray et al., 1999a; Haley et al., 1997a; Proctor et al., 1998; Stretch et al., 1995). Additional groups of U.S. veterans have been drawn based on disease (Horner et al., 2003; Wallin et al., 2012).
Department of Veterans Affairs National Health Survey
The Department of Veterans Affairs (VA) is conducting a longitudinal survey of 30,000 veterans known as the National Health Survey of Gulf War Veterans and Their Families (NHS). Thus far, three survey waves have been conducted: wave 1 in 1993–1995 (Kang et al., 2000) with physical examinations in 1999–2001 (Eisen et al., 2005), wave 2 in 2003–2005 (Kang et al., 2009), and wave 3 in 2012–2013 (Bossarte, 2014; Dursa et al., 2016). Volume 4 described seven derivative studies of the NHS, and Volume 8 identified six additional derivative studies of this cohort that examine specific health outcomes or conducted a follow-up survey and analysis. Since 2009, four new studies have been published based on the second wave survey. Preliminary results from wave 3 were presented to the committee by a VA representative (Bossarte, 2014), and some of the results have been recently published (Dursa et al., 2016).
A major population-based study of U.S. veterans was mandated by Public Law 103-446 in 1994 to estimate the prevalence of symptoms and other health outcomes (including reproductive outcomes in spouses and birth defects in children) in Gulf War deployed vs nondeployed veterans. This retrospective
study was designed to be representative of the nearly 700,000 U.S. veterans sent to the Persian Gulf and 800,680 veterans who were not deployed but who were in the military between September 1990 and May 1991.
For wave 1 of the NHS, VA mailed questionnaires to a stratified random sample of 15,000 deployed and 15,000 nondeployed Gulf War veterans identified by the Defense Manpower Data Center (DMDC) (Kang et al., 2000). Women and those serving in the National Guard and reserves were oversampled, resulting in a study population that was approximately 20% women, 25% National Guard, and 33% reservists. The controls were stratified by gender, unit component, and branch of service to mirror the population of deployed veterans. The self-administered structured health questionnaire contained a 48-symptom inventory (somatic and psychological symptoms) and questions about chronic medical conditions, functional limitations, use of medical services, and environmental exposures (e.g., immunizations, use of the prophylactic antinerve agent pyridostigmine bromide (PB), smoke from oil-well fires, and pesticides and insecticides).
Wave 1 also used telephone interview software in an attempt to capture those who did not respond to the mailed questionnaire. A total of 11,441 (75%) deployed and 9,476 (64%) nondeployed veterans participated in the study; 15,817 veterans responded to the questionnaire, and 5,100 responded to the telephone interview (Kang and Bullman, 2001; Kang et al., 2000). In addition, medical records were obtained for a random sample of 4,200 respondents to validate self-reports of clinic visits or hospitalizations within the last year. Of the 2,233 veterans with at least one clinic visit, 43.2% provided medical record release consent; of the 310 with at least one hospitalization, 45.2% provided medical record release consent. Medical record reviews verified more than 90% of self-reported reasons for clinic visits or hospitalizations (Kang et al., 2000).
Kang et al. (2000) did not assess exposure–symptom relationships but rather noted the percentage of veterans who reported each of 23 environmental exposures and nine vaccine or prophylactic exposures (such as to PB). The five most common environmental exposures reported by more than 60% of survey participants were diesel, kerosene, or other petrochemical fumes; local food other than that provided by the armed forces; chemical protective gear; smoke from oil-well fires; and burning trash or feces.
Davis et al. (2004) studied the presence of distal symmetric polyneuropathy determined by medical history, physical examination by a neurologist, blood tests, and standardized electrophysiologic assessment of motor and sensory nerves in a cohort subset of 1,061 deployed veterans and 1,128 nondeployed veterans from wave 1 of the NHS. Spouses of deployed (n = 484) and nondeployed (n = 533) veterans were studied to evaluate whether an infectious agent or environmental contaminant brought back from the gulf could be responsible for any adverse health outcomes. Evaluations of 244 Khamisiyah-exposed (data provided by the Department of Defense [DoD]) vs 817 nonexposed deployed veterans for the presence of distal symmetric polyneuropathy were conducted.
In 1999–2001, Eisen and colleagues (2005) performed a cross-sectional study on numerous health outcomes of Gulf War veterans who had participated in wave 1 of the NHS. The study population consisted of a stratified random sample of the 11,441 deployed and 9,476 nondeployed veterans who had responded to the mailed questionnaire or telephone interview described above. This study included a comprehensive medical examination and laboratory testing. Of the 1,996 eligible deployed veterans, 1,061 (53.1%) were examined; 680 (34.1%) declined; and 255 (12.8%) were not located. Of the 2,883
eligible nondeployed veterans, 1,128 (39.1%) were examined; 1,316 (45.7%) declined and 439 (15.2%) were not located. Despite extensive recruitment efforts, the participation rate for this study was low—60.9% of deployed veterans and 46.2% of the nondeployed.
Study participants were assigned a medical center closest to their residence where physicians and nurses performed medical, neurologic, psychiatric, and gynecologic histories and examinations; laboratory, nerve conduction, pulmonary function, and neuropsychological tests were also performed. Twelve primary health outcome measures and physical functioning were examined using the SF-36.2 Outcome measures were chosen by the authors to cover the most common symptoms reported by veterans, such as musculoskeletal pain, fatigue, rashes, and neuropathy (Kang et al., 2000).
Kang and colleagues (2001) assessed the association between self-reported adverse pregnancy outcomes and deployment to the gulf using data from the wave 1 questionnaire. Results are based on the 3,397 (2,761 males, 636 females) deployed and 2,645 (1,951 males, 695 females) nondeployed veterans who reported their or their partner’s first pregnancy ending after June 30, 1991.
A nested case-control analysis was performed on the 277 (2.4%) deployed veterans from wave 1 who met the case definition for a possible neurological cluster of symptoms including blurred vision, loss of balance or dizziness, tremors or shaking, speech difficulty, concentration or memory problems, and irregular heartbeat, to determine which of 23 self-reported exposures were more common among cases than among the controls (6,730 Gulf War veteran respondents who lacked symptoms) (Kang et al., 2002). Exposure to a variety of chemical agents was reported to be higher among cases than controls, specifically to chemical-agent-resistant compound paint, depleted uranium, nerve gas, food contaminated with oil or smoke, and bathing in or drinking water contaminated with oil or smoke.
Kang et al. (2003) used the wave 1 participants to assess the prevalence of posttraumatic stress disorder (PTSD) and chronic fatigue syndrome (CFS) in Gulf War veterans. The questionnaire administered to the veterans in 1993–1995 had included eight symptoms to be used to diagnose CFS, and the PTSD Checklist was used to identify symptoms of PTSD. Assessment of CFS was based on the Centers for Diseases Control and Prevention (CDC) case definition after exclusion of alternate medical causes of the symptoms.
Kang and colleagues (2005) conducted a nested case-control study evaluating the role of sexual assault on the risk of PTSD from the 11,441 Gulf War veteran respondents of the 1995 questionnaire described above. A score of 50 or higher on the PTSD Checklist was necessary to have met the criteria for PTSD; 1,381 (12.1%) Gulf War veterans (336 females and 1,045 males) screened positive for PTSD, while 10,060 (1,795 females and 8,265 males) screened negative and were used as a comparison group. Adjustments for age, race, branch, combat, rank, and unit component, and self-report of sexual harassment and assault were made.
Karlinsky and colleagues (2004) examined pulmonary function and self-reported respiratory symptoms in 1,036 deployed and 1,103 nondeployed veterans who completed the clinical examination component of the NHS. Results of pulmonary function tests were classified into five categories: normal pulmonary function, nonreversible airway obstruction, reversible airway obstruction, restrictive lung physiology, and small-airway obstruction. The authors also reported on the pattern of pulmonary function test results in those exposed (n = 159) and those not exposed (n = 877) (according to DoD exposure estimates developed in 2002) to nerve agents from destruction of munitions at the storage site at Khamisiyah in 1991.
2 The Medical Outcome Study’s 36-item questionnaire, known as the Short Form-36 or SF-36, is a standardized instrument to measure physical and mental health, physical and social functioning, and general well-being.
The Volume 8 committee identified six studies of the NHS cohort (Blanchard et al., 2006; Kang et al., 2009; Page et al., 2005a,b; Toomey et al., 2007, 2009) published after Volume 4. The committee notes that VA uses the term chronic multisymptom illness rather than Gulf War illness and thus the discussions of the NHS cohort below use the former term.
Page and colleagues (2005a) assessed the possible health effects of Khamisiyah exposure (determined from models developed by DoD and Central Intelligence Agency) in 5,555 Army veterans drawn from the cohort of 11,441 deployed veterans who responded to the wave 1 questionnaire. When the survey was completed in 1995, veterans were not yet notified of possible chemical agent exposure in Khamisiyah. Comparisons were made between the 1,898 exposed and 3,336 unexposed veterans.
Page and colleagues (2005b) also examined the association between notification of possible exposure at Khamisiyah and self-reported morbidity. In 2000, a subsample of 1,056 deployed Army veterans was surveyed; of the 600 notified subjects, 438 (73%) responded, and of the 456 nonnotified subjects, 318 (70%) responded.
Blanchard and colleagues (2006) assessed the prevalence and severity of chronic multisymptom illness (CMI) in the same cohort of 1,061 deployed and 1,128 nondeployed veterans as described by Eisen and colleagues (2005). Combat exposure was significantly associated with CMI. The prevalence of CMI in the nondeployed population remained relatively constant at 4, 7, and 10 years postwar. Among the deployed veterans, CMI prevalence decreased from 44.7% at 4 years to 28.9% after 10 years (Fukuda et al., 1998; Steele, 2000). Blanchard et al. (2006) also assessed for the presence of CMI based on the possible exposure of deployed veterans to nerve agents as a result of the Khamisiyah demolition. Based on DoD modeling, 236 (22.2%) of the deployed veterans were exposed; 92 (39.0%) had CMI and 144 (61.0%) did not.
Toomey and colleagues (2007) examined the prevalence of mental health disorders, self-report of symptoms, and quality of life 10 years postconflict in the same cohort of 1,061 Gulf War deployed and 1,128 nondeployed veterans as that of Eisen et al. (2005). Neuropsychological functioning was also evaluated in the same population (Toomey et al., 2009). The measures (e.g., general intelligence, attention or executive functioning, motor ability, visuospatial processing, and verbal and visual memory) were based on those previously found to be different between the same deployed and nondeployed groups as in earlier studies.
Kang and colleagues (2009) conducted a 10-year follow-up general health assessment of Gulf War veterans in wave 2 of the NHS. For this wave, VA and Social Security records through December 2002 were used to identify and mail health questionnaires to the 29,607 living veterans who had been sampled in wave 1 (15,000 Gulf War deployed and 15,000 nondeployed). Telephone interviews were conducted with 2,000 nonresponsive participants and a sample of 1,000 participants who had indicated a clinic visit or hospitalization within the previous 12 months in order to obtain permission for medical record retrieval. As a result of those recruitment efforts, 6,111 (40%) deployed and 3,859 (27%) nondeployed participants responded to the survey; the overall response rate was low, only 34%. The administered questionnaire was a modified version of that used in the 1995 survey and included the PTSD Checklist, the Patient Health Questionnaire, and the SF-123 in addition to other items used to assess general health status.
The Volume 10 committee identified five new published studies that met its inclusion criteria, which were all based on data from the second wave of the NHS (Coughlin et al., 2011a,b; Li et al., 2011a,
3 “The 12-Item Short Form Health Survey (SF-12) was developed for the Medical Outcomes Study, a multiyear study of patients with chronic conditions. The instrument was designed to reduce respondent burden while achieving minimum standards of precision for purposes of group comparisons involving multiple health dimensions” (RAND Corporation, 2010).
In 2003–2004, Wallin et al. (2009) assessed neuropsychologic performance in Gulf War deployed veterans with Gulf War illness (n = 25) and without (n = 16) who were recruited from the first wave of the NHS and who lived in the mid-Atlantic region (248 veterans were contacted). Cases were defined based on the CDC criteria for Gulf War illness. Participants completed an interview and full day of cognitive testing. The neuropsychological testing battery included assessment of verbal abilities, attention, memory and learning, problem solving, and motor skills. Other assessments included the Personality Assessment Inventory and SF-36. Additional data was extracted from responses to the first wave of the NHS. Regression analyses were adjusted for demographic and military variables.
In an assessment of 8,822 veterans who completed both the 1995 and 2005 surveys, Li et al. (2011a) reported changes in health and chronic diseases. Baseline data from the 1995 survey and follow-up data from the 2005 survey were used to assess longitudinal change for each individual. Weights for gender, service branch, and unit component were applied to estimate population prevalence rates.
Using data from the second wave of the NHS in 2003–2005, Coughlin et al. (2011b) assessed the association between alcohol consumption in 9,970 respondents and unexplained multisymptom illness (defined as having unexplained physical symptoms and illnesses that persisted for 6 months or longer and were not explained by other diagnoses) while adjusting for age, sex, race/ethnicity, branch of service, rank, and deployment status. The authors used the same survey results to assess the association between body weight and various health conditions (Coughlin et al., 2011a).
In a small case-control study conducted by Li et al. (2014), among the respondents of the 2005 survey, Li et al. sampled 200 Gulf War veterans with “chronic fatigue-like multi symptoms” and 398 controls veterans. From those contacted, 16 Gulf War veterans and 12 control veterans completed a telephone interview, met eligibility criteria, and completed a clinical exam with autonomic testing (conventional large fiber nerve conduction, quantitative sensory testing, baseline autonomic testing, quantitative sudomotor axon reflex test, and diagnostic tilt-table test).
During an open session, VA made a presentation to the Volume 10 committee on the most recent results of the third survey wave of the NHS (Bossarte, 2014). Results from this survey were not published until early 2016 (Dursa et al., 2016). This survey consisted of the same 30,000 Gulf War deployed and era veterans and the same administration methods as the two prior surveys, that is via mail, website, or a computer-assisted telephone interview. A total of 14,252 veterans responded (8,104 deployed and 6,148 era veterans), for a response rate of 50% (57% deployed, 43% nondeployed). Respondents were more likely to have served in the Army compared with other branches, been deployed, been an officer, be older, and to identify as white than were nonrespondents (Dursa et al., 2016). In a follow-up email to the committee, VA stated that 7,399 veterans participated in both the 2005 and 2013 surveys (Personal communication, Robert Bossarte, VA, January 26, 2015).
The wave 3 questionnaire was modified from the earlier versions, but it again collected information about the presence of various symptoms such as fatigue, muscle or joint pain, headaches, memory problems, and respiratory problems; functional status; activity limitations; health perceptions; chronic medical conditions (self-report of provider diagnoses); mental health disorders; health care utilization; medications; women’s health; and potential confounders such as the use of alcohol and cigarettes. No questions on environmental exposures were included. The questionnaire contained modules of validated self-report assessments, including the PTSD Checklist-Civilian version. Groups of symptoms were used to identify the presence of CMI (Bossarte, 2014).
A sample of 2,500 veterans who had completed the questionnaire and who had provided a valid response to either “reason for clinic or doctor visit in the past 12 months” or “reason for hospitaliza-
tion within past 12 months” were selected for medical record validation. In the medical record review subset, 86% of self-report responses for reasons for clinic visit or hospitalization were verified by medical records (Bossarte, 2014). Weighted prevalence estimates and adjusted odds ratios (ORs) (models were adjusted for age, sex, race, body mass index, smoking status, service branch, and unit component) were presented for several self-reported medical conditions among deployed and nondeployed veterans (Dursa et al., 2016).
Although the data in the presentation by Bossarte (2014) were unpublished and not peer reviewed during the course of the committee’s deliberations, they are included in Chapter 4 when data were not included in the Dursa et al. (2016) publication because these surveys are among the few sources of data on many health conditions in a large group of U.S. Gulf War veterans. The Dursa et al. (2016) publication includes additional information about the survey and the statistical methods and weights used in the analysis. It is cited in the discussion of health conditions in Chapter 4 where appropriate.
The Iowa Study
In Volume 4, the Iowa Persian Gulf Study was presented as a reference study with eight derivative studies, and the Volume 8 committee identified three studies derived from the original Iowa cohort. No new studies using this cohort were identified by the Volume 10 committee.
The Iowa study was a cross-sectional survey of a representative sample of 4,886 military personnel who listed Iowa as their home of record at the time of enlistment and served between August 2, 1990, and July 31, 1991 (Iowa Persian Gulf Study Group, 1997). The DMDC identified a representative sample of 28,968 potentially eligible military personnel.
Study subjects were divided into four groups: Gulf War deployed active duty, Gulf War deployed National Guard or reserve, Gulf War nondeployed active duty, and Gulf War nondeployed National Guard or reserve. Random samples were evenly selected from each of the four groups for a total of 4,886 study subjects. Of the study subjects who were contacted, 3,695 (76%) completed a telephone interview. Trained examiners used standardized questionnaires, instruments, and scales from the following to collect information: National Health Interview Survey; Behavioral Risk Factor Surveillance Survey; National Medical Expenditures Survey; Primary Care Evaluation of Mental Disorders; Brief Symptom Inventory; CAGE questionnaire (for alcoholism)4; PTSD Checklist–Military; CDC Chronic Fatigue Syndrome Questionnaire; Chalder Fatigue Scale; American Thoracic Society questionnaire; and the Sickness Impact Profile.
No clinical examinations were performed; rather, before conducting their telephone survey, researchers grouped sets of symptoms from their symptom checklists into a priori categories of diseases or disorders. After a veteran identified himself or herself as having the requisite set of symptoms, researchers analyzing the responses considered the veteran as having symptoms “suggestive” of or consistent with a particular disorder but not as having a formal diagnosis of the disorder.
Gulf War veterans scored significantly lower on all eight subscales for physical and mental health on the SF-36. The subscales for bodily pain, general health, and vitality showed the greatest absolute differences between deployed and nondeployed veterans. The Iowa study assessed exposure–symptom
4 The CAGE is a 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).
relationships by asking veterans to report on their deployment exposures including to solvents or petrochemicals, smoke or combustion products, lead from fuels, pesticides, ionizing or nonionizing radiation, chemical warfare agents, PB use, infectious agents, and physical trauma (Iowa Persian Gulf Study Group, 1997).
Eight derivative studies of the Iowa cohort were described in Volume 4 (Barrett et al., 2002; Black et al., 1999, 2000, 2004a,b; Doebbeling et al., 2000; Lange et al., 2002; Zwerling et al., 2000). The Volume 8 committee identified three new studies of this cohort.
Subsequent cross-sectional studies of the Iowa cohort examined the presence of possible multiple chemical sensitivity (MCS) in these veterans, including the impact of MCS on quality of life and utilization of health services (Black et al., 1999) and the prevalence of and risk factors for the development of MCS (Black et al., 2000). Self-reports of postwar injuries were also assessed (Zwerling et al., 2000).
Doebbeling et al. (2000) used factor analysis to attempt to determine if the symptoms reported by Gulf War veterans after their deployment were different than those reported by nondeployed veterans and if the symptoms seen in the deployed veterans could possibly constitute a unique Gulf War syndrome.
Three studies assessed the prevalence of psychiatric disorders in Gulf War deployed and nondeployed veterans. Barrett et al. (2002) examined the association between PTSD and self-reported physical health status. The prevalence of and risk factors for current anxiety disorder was studied by Black et al. (2004b), who used the PRIME-MD in a structured telephone interview to identify symptoms of anxiety. In a case-control study, however, Black et al. (2004a) used the Structured Clinical Interview for Diagnostic and Statistical Manual of Mental Disorders, fourth edition (SCID-IV) to diagnose current or lifetime depression in 608 of the Iowa veterans, of whom 192 met the case definition for lifetime depressive disorder (132 deployed and 60 nondeployed). The prevalence of comorbid psychiatric diagnoses was determined.
Lange et al. (2002) examined the impact of exposure of Gulf War veterans to Kuwaiti oil-well fires and the prevalence of asthma and bronchitis 5 years after the war. Modeled exposures were developed using a geographic information system to integrate spatial and temporal records of smoke concentrations with troop movements ascertained from global positioning system records. Results for modeled exposures were compared with self-reported exposures.
The Volume 8 committee identified three new studies based on the Iowa cohort: Ang et al. (2006) and Forman-Hoffman et al. (2007) both looked at the presence of chronic widespread pain (CWP), and Black et al. (2006) assessed the prevalence of borderline personality disorder.
Approximately 5 years after the initial survey, Ang and colleagues (2006) designed a follow-up evaluation to determine predictive factors for development of CWP in Gulf War veterans. Of the 3,695 veterans in the baseline study, a sample of 602 veterans who previously met the criteria for cognitive dysfunction, depression, or CWP, and a sample of veterans without these problems were evaluated at follow-up.
Forman-Hoffman et al. (2007) also conducted a cross-sectional survey 5 years postconflict to determine the effect of deployment on the development of CWP in the Iowa cohort. The authors used the same 3,695 participants as sampled in Black et al. (2006) below.
Black and colleagues (2006) assessed the prevalence of borderline personality disorder traits in a sample of the Iowa Persian Gulf War veterans using the Schedule for Adaptive and Nonadaptive Personality. The population was drawn from the initial 3,695 surveyed individuals; a second assessment (in-person interviews and medical examinations) was administered to 602 veterans who previously met
the criteria for one or more of the following: depression, CWP, and cognitive dysfunction. The overall response rate was 95.7% (n = 576).
Oregon and Washington Veteran Studies
One reference study discussed in Volume 4 examined Gulf War veterans who listed Oregon or Washington as their residence at the time of their deployment (McCauley et al., 1999a); two derivative studies were also described in that volume (Bourdette et al., 2001; Spencer et al., 2001). Neither the Volume 8 nor the Volume 10 committees identified any new studies that used this data set.
Investigators from the Portland Environmental Hazards Research Center examined numerous health outcomes in Gulf War veterans who were deployed between August 1, 1990, and July 31, 1991, and who listed Oregon or Washington as their home state of record at the time of deployment; data was obtained from the DMDC (McCauley et al., 1999a). Beginning in November 1995 and ending in June 1998, a mailed questionnaire was distributed to a representative and random sample (n = 2,343) of the eligible 8,603 Gulf War veterans to assess general health through symptom self-reports; the response rate was 48.4%. The study did not include a nondeployed comparison group. The next phase consisted of a clinical examination of the first 225 participants who showed differences between the symptoms they reported on questionnaires and the symptoms they reported at time of clinical examination.
In Volume 4, Bourdette et al. (2001) was considered a reference study although it used the same data set as McCauley et al. (1999a). Bourdette and colleagues (2001) compared 244 potential cases of unexplained illness based on clinical examination and the author’s definition of unexplained illness with 113 potential controls that did not meet the case definition from 799 of those eligible for the clinical study, and located participants who had completed the questionnaire described above.
A nested case-control analysis of the cohort examined 142 items related to Gulf War self-reported exposure that might account for cases of unexplained illness (Spencer et al., 2001). The sample consisted of 241 veterans with unexplained illness and 113 healthy controls (drawn from those who completed the 1995–1998 questionnaire above). Multivariate analyses were used to assess associations between Gulf War exposures and unexplained illness. One strength of this study was its elimination of numerous self-reported exposures (such as anthrax and botulinum toxoid vaccines) with questionable validity as determined by lack of test-retest reliability or time-dependent information (for example, chemical weapon exposure reported by precombat veterans or postcombat veterans who could not have been so exposed) (McCauley et al., 1999b).
Kansas Veteran Study
In Volume 4, the study of Kansas Gulf War veterans (Steele, 2000) was considered to be a secondary study for most health outcomes. No derivative studies using the Kansas cohort were identified in that volume, or by the Volume 8 committee. However, since 2009, two nested case-control studies using the Kansas veterans cohort have been published (Steele et al., 2012, 2015).
Using lists of eligible veterans from the DMDC, Steele (2000) conducted a population-based survey to determine health problems of veterans who listed Kansas as their home state of record. A stratified random sample of 3,138 veterans was selected; 2,396 (76%) were located with in-state contact information, of whom 2,211 met further eligibility requirements. Of those, 2,030 agreed to be interviewed; the response rate was 92%. The sample included 1,548 deployed veterans and 482 nondeployed veterans.
The survey, mailed out in 1998, inquired about 16 specific physician-diagnosed or physician-treated medical or psychiatric conditions; 37 symptoms; service branch and locations during the Gulf War (including whether the veterans were notified about the Khamisiyah demolitions); and vaccinations. Using her own definition of Gulf War illness (see the Chapter 4 section on Gulf War illness for the definition), which was similar to that used by CDC (Fukuda et al., 1998), Steele assessed the prevalence of Gulf War illness and associations with Gulf War deployment.
Steele et al. (2012) conducted a case-control study in 2000 that compared Gulf War veteran’s exposures in a population-based sample of 304 deployed veterans residing in the greater Kansas City metropolitan area, 144 with Gulf War illness and 160 healthy controls out of 906 households contacted (86% response rate, 68% participation rate). Gulf War illness case status was determined based on screening using the CDC definition and subsequent inclusion in the study based on the Kansas definition for Gulf War illness. Potential participants were excluded if they reported being diagnosed with a chronic health condition that could explain their symptoms (e.g., diabetes or lupus), had persistent problems due to chronic infection or injury, reported diagnoses of schizophrenia or bipolar disease, or were hospitalized for alcohol or drug dependence, depression, or PTSD. At the study location, participants completed self-administered questionnaires about their deployment locations and duration and whether they had any of 19 specific exposures or experiences during deployment, and provided blood samples for another study on genetic factors and enzyme activity.
The blood samples were used to assess the relationship between butyrylcholinesterase genotype and Gulf War illness (Steele et al., 2015). All 304 participants were included in this second study. Genotype and butyrylcholinesterase enzyme activity for Gulf War illness cases were compared with healthy controls. Post hoc analyses examined effects of confounding factors and interaction between multiple factors, including reported exposures.
Fort Devens and New Orleans Cohort Studies
Gulf War veterans from military units stationed in Fort Devens, Massachusetts, and in New Orleans are among the most studied groups of U.S. Gulf War veterans. The reference study (Proctor et al., 1998) and three derivative studies (Proctor et al., 2001a,b; White et al., 2001) from the cohort were described in Volume 4. The Volume 8 committee identified one additional study (Proctor et al., 2006), but the Volume 10 committee did not identify any new studies of these veterans.
The symptom experience of two deployed cohorts of Gulf War veterans was studied by Boston-based researchers. The first, an Army cohort based in Fort Devens, Massachusetts, was surveyed longitudi-
nally. From 1994 to 1996, Proctor et al. (1998) surveyed the deployed cohorts (Fort Devens and New Orleans) with a medical and occupational history questionnaire, an environmental interview, a battery of neuropsychological tests, and scales assessing psychological symptoms. Psychological symptoms were assessed with the Brief Symptom Interview and the Mississippi Scale for PTSD; psychiatric disorders were diagnosed using the Clinician Administered PTSD Scale and the Structured Clinical Interview for DSM-IIIR Axis I Disorders, and structured interviews assessing psychological outcomes. A second deployed cohort from New Orleans and a unit deployed to Germany during the Gulf War were studied only at the second time period. The unit deployed to Germany, serving as controls (n = 47), was an air ambulance company of National Guard from Maine that handled wounded personnel evacuated from the gulf. The study’s nearly 300 subjects were a stratified random sample of 2,949 troops (n = 220) from Fort Devens and 928 (n = 73) from New Orleans; both groups consisted of active duty, reserve, and National Guard troops.
Psychiatric interviews and other clinical evaluations were administered between 1993 and 1994 to all three cohorts. White et al. (2001) assessed neuropsychological functioning in the cohorts and examined whether any links could be made between performance on a battery of tests and self-reported exposure to a variety of toxicants experienced during deployment. The neuropsychological test battery was designed to assess abilities across the following functional domains: general intelligence, attention/executive function, motor ability, visuospatial processing, verbal and visual memory, mood, and motivation. In addition, eight exposures were surveyed, and respondents were asked to rate each on a scale of 0–2 (0 = no exposure; 1 = exposed; 2 = exposed and felt sick at the time). The authors used standardized regression to identify associations between musculoskeletal, neurologic, neuropsychologic, and psychologic symptoms and several exposures-debris from Scuds and chemical and biologic warfare agents.
The health-related quality of life among the Fort Devens cohort (n = 141) and the Germany deployed cohort (n = 46) was evaluated by Proctor et al. (2001a). The SF-36 was administered to a stratified, random sample of the original cohort approximately 4 years after the war. Proctor et al. (2001b) assessed 180 deployed veterans from the Fort Devens cohort and 46 Germany deployed veterans for symptoms of chronic fatigue and chemical sensitivity to assess the prevalence of the symptoms and whether there was an overlap between these symptoms and the CDC case definition of CMI.
The Volume 8 committee identified one new derivative study. Proctor et al. (2006) assessed neurobehavioral functioning in relationship to potential exposure to sarin and cyclosarin as a result of the demolition of the Khamisiyah munitions dump. Data had been collected in 1994 to 1996, before veterans had been notified about their potential exposure to the nerve agents. The neurobehavioral tests included those for attention, executive function, psychomotor function, visuospatial abilities, and short-term memory.
Numerous studies have been conducted on the Seabees, members of the U.S. Naval Construction Force battalions. Two reference studies were included in Volume 4. Haley et al. (1997b) began a study of one cohort of Seabees, and Gray et al. (1999a) surveyed a subset of the Haley cohort that excluded Gulf War veterans who were no longer in the service at the time of their study. Three derivative studies for Haley et al. (1997b) were presented in Volume 4 along with two derivative studies for Gray et al. (1999a). The Volume 8 committee identified five additional derivative studies of the Haley et al. (1997b)
cohort, and the Volume 10 committee identified eight additional derivative studies. However, neither the Volume 8 nor the Volume 10 committee identified any derivative studies of the Gray et al. (1999a) cohort.
Haley et al. Studies
In 1994, Haley et al. (1997b) recruited members of the Twenty-Fourth Naval Reserve Construction Force Battalion into an epidemiologic study in an effort to identify syndromes that might help define the variety of symptoms experienced by Gulf War veterans. Study participants were selected from the 606 reserve Seabees living in five southern states who had been called to active duty for the Gulf War. The researchers were able to directly contact 350 of the 429 Seabees for whom the investigators had presumptive addresses from the official in-theater battalion roster; efforts to reach the other 250 Seabees were made by battalion commanders, media coverage, and networking by veterans. Of the 606 Seabees, 41.1% (n = 249) agreed to participate in the study (42% were still on active duty and the rest of the participants had retired); there was no comparison group of nondeployed veterans. A majority of participants (70%) reported having had a serious health problem since returning from the Gulf War. A telephone survey of a random sample of nonparticipants found that, while they were demographically similar to participants, only 43% reported having had serious health problems since the war. Eleven percent of participants and 3% of nonparticipants were unemployed. Of those who participated, more than half (58%) had left the military by the time of the study.
Study participants completed a standardized survey measuring the anatomical distributions or characteristics of each symptom and wartime exposures, and a standard psychological personality assessment inventory. Two-stage factor analysis was used to derive syndromes from the survey responses. Six syndrome factors were identified based on results from 63 of the 249 Seabee veterans.
- Syndrome 1—impaired cognition—was characterized by symptoms indicative of problems with attention, memory, reasoning, insomnia, depression, daytime sleepiness, and headaches.
- Syndrome 2—confusion-ataxia—was characterized by problems with thinking, disorientation, balance disturbances, vertigo, and impotence.
- Syndrome 3—arthromyoneuropathy—was characterized by joint and muscle pains, muscle fatigue, difficulty lifting, and extremity paresthesias.
The three other syndromes—syndrome 4 phobia-apraxia; syndrome 5 fever-adenopathy; and syndrome 6 weakness-incontinence—demonstrated weaker clustering and overlapped to some extent with syndromes 1 through 3 (Haley et al., 1997b). In most of the derivative studies discussed below, only veterans with syndromes 1 through 3 were assessed.
The three syndromes identified by Haley and colleagues (1997b) were the focus of a case-control study that examined the relationship of the syndromes to self-reported exposures to neurotoxicants. The study tested the hypothesis that exposure to organophosphates and related chemicals that inhibit cholinesterase are responsible for the three nervous system–based syndromes (Haley and Kurt, 1997).
Another study by Haley and collaborators (1999) examined whether genetic susceptibility could play a role in placing some veterans at risk for neurologic damage by organophosphate chemicals. The
investigators studied 45 veterans: 25 with chronic neurologic symptoms identified through their earlier factor analysis study and 20 healthy controls from the same battalion. Investigators measured blood butyrylcholinesterase and two types, or allozymes, of the enzyme paraoxonase/arylesterase-1. The genotypes encoding the allozymes were also studied.
The Volume 8 committee identified one study by Haley et al. (2009), conducted in 1997–1998, that looked at abnormal brain response to cholinergic challenge in a small group of Gulf War veterans. Twenty-one Gulf War veterans with symptom complexes used by Haley to define three Gulf War illness syndromes and 17 age-, sex- and education-matched controls, underwent a 99mTc-HMPAO-SPECT brain scan with and without an infusion of PB. The Volume 8 committee also summarized other studies by Haley and colleagues on neurologic outcomes as follows:
Haley and colleagues performed detailed neurologic assessments in several case-control studies of the original cohort of Seabee reservists. The cases were veterans who had met criteria for factor-derived syndromes. Under the hypothesis that those veterans were ill from neurotoxic exposures, especially to organophosphates, the assessments covered broad neurologic function (Haley and Kurt, 1997), autonomic function (Haley et al., 2004), vestibular function (Roland et al., 2000), basal ganglia injury (Haley et al., 2000a,b), normalized regional cerebral blood flow (Haley et al., 2009); and paraoxonase genotype and serum concentrations (Haley et al., 1999).
The Volume 10 committee identified many derivative studies of the Seabee cohort initially described by Haley et al. (1997b). In general, these studies were further assessments of specific aspects of Haley’s factor-analysis defined syndromes of Gulf War illness, with an emphasis on brain structure and function and neurologic endpoints, particularly the neural and cognitive effects of exposure to cholinesterase inhibitors. Many of the derivative studies drawn from the original 249 Seabees fail to provide adequate description of the methods used such as how their subjects were selected or when they were assessed. Because of the small number of veterans that were assessed in the majority of the studies by Haley and colleagues, and because the focus of Volume 8 was not on linking Gulf War exposures to health outcomes, these studies were not discussed in detail in the previous volumes. The Volume 10 committee summarizes the goals of the studies here briefly and discusses them in Chapter 4 in the section on Gulf War illness.
Between June 2008 and July 2009, 48 members of the Seabee cohort attended a week-long clinical, neuroimaging, and neuropsychologic and psychiatric sessions at the University of Texas Southwestern Medical Center in Dallas. Tests included face-name associative memory and face recognition tests, the SCID-IV, and the Clinician-Administered PTSD Scale. Participants also underwent a blinded hippocampus perfusion study in two sessions separated by 2 days, one session with a saline infusion and the second with physostigmine infusion (Li et al., 2011b). The testing goals and protocols are summarized briefly below and results, where relevant, are presented in Chapter 4.
Two studies used magnetic resonance imaging (MRI) to determine blood flow in the brain of veterans with and without Gulf War illness. Li et al. (2011b) used arterial spin labeling perfusion MRI with a physostigmine challenge to compare hippocampal blood flow in a group of veterans with one of three Gulf War illness syndromes (11 veterans with syndrome 1, 13 veterans with syndrome 2, and 11 veterans with syndrome 3) and a control group of 13 well veterans, 11 years after initial testing and 20 years after the war. For this study, 34 Seabees from the 1998 cohort and 23 new Gulf War veterans from the original epidemiologic study (assumed by the Volume 10 committee to be from the 1994 Haley cohort although this was not explicitly stated) were evaluated. Initial testing was conducted with computed tomography. Liu et al. (2011) reported a similar study with a small number of veterans who met Haley’s definition for Gulf War illness syndromes 1, 2, and 3 (n = 11, 12, and 10, respectively) and 14 veterans without Gulf War illness. The authors found that using arterial spin labeling with a physostigmine challenge was
a cost-effective method for characterizing Gulf War illness syndromes. It was unclear to the committee whether the Li et al. (2011b) and Liu et al. (2011) studies were conducted on the same group of veterans.
In the same population of veterans with and without Gulf War illness as reported by Li et al. (2011b), Odegard et al. (2013) used functional magnetic resonance imaging (fMRI) to measure activation of the hippocampus while participants memorized information. The veterans then underwent a face-name associative recall test developed by the authors to assess recall of faces’ names as well as intermediate memory states. In yet another study in the same group of Seabees, Moffett et al. (2015) compared measures of complex language (verbal fluency, word generation) and fMRI data between 50 Seabees with Haley’s syndromes and 14 age-, sex-, education-matched controls.
In a series of studies conducted in 2008–2009, Tillman and colleagues attempted to differentiate responses to stimuli in veterans who met the criteria for Haley’s three Gulf War illness syndromes. In the first study, Tillman et al. (2010) compared 25 Gulf War veterans from Haley’s original cohort of Seabees who reported major cognitive complaints (i.e., the veterans met the criteria for Gulf War illness syndromes 1 or 2) with 23 age- and education-matched Gulf War veterans who were not ill or who reported predominantly peripheral pain symptoms. Event-related potentials as measured by electroencephalographic (EEG) activity were collected while the veterans performed a go/no go task that required semantic decisions and inhibitory processing (responding to drawing stimuli).
In a further study of event-related potentials for hyperarousal, six veterans who met Haley’s criteria for Gulf War illness syndrome 1, eight veterans who had syndrome 2, six veterans who had syndrome 3, and eight control veterans who did not have Gulf War illness performed an auditory three-condition oddball task with gunshot and lion roar sounds as the distractor stimuli (Tillman et al., 2012). Hyper-arousal was evaluated with a subset of questions from the Mississippi Scale for Combat-Related PTSD and was measured by means of EEG activity; only 4 of the 20 veterans with Gulf War illness and none of the controls had been diagnosed previously with PTSD. Differences in hyperarousal responses were associated with the Gulf War illness syndromes. In the third study, Tillman and colleagues (2013) had 22 veterans with Gulf War illness (seven with syndrome 1, nine with syndrome 2, and six with syndrome 3) and eight controls perform a visual three-condition oddball task where images authenticated to be associated with the Gulf War were the distractor stimuli. Again, hyperarousal was determined by event-related potentials based on EEG activity. The Volume 10 committee notes that all three studies may have been performed on the same individuals—although there is no explanation as to why the number of veterans in each syndrome category varied—but it is difficult to determine this from the methods described in each paper.
Further testing of this group of veterans with Gulf War illness (11 with syndrome 1, 16 with syndrome 2, and 12 with syndrome 3) and 14 controls used a quantitative sensory testing fMRI protocol to evaluate brain activation in response to innocuous and noxious heat stimuli (Gopinath et al., 2012). Subjects were tested both in the scanner and outside the scanner. Differences in response were seen among the four groups.
U.S. Military Health Survey
A new sample of veterans was selected to further investigate the case definition of the Gulf War illness syndromes identified by Haley and colleagues on the basis of the earlier Seabee studies. While this is a different cohort of veterans, it is discussed here because it includes Haley’s original group of Seabees and has been specifically evaluated for the validation of Haley’s case definition of Gulf War illness and any associated factors. Because results of this survey were not published until 2011, Volumes 4 and 8 did not report on this cohort of veterans or any derivative studies of them.
To validate the Gulf War illness syndromes posed by Haley et al. (1997b, 2001), Iannacchione et al. (2011) conducted the U.S. Military Health Survey, a study of a population-based sample of more than 8,000 Gulf War deployed and nondeployed (but fit for deployment) service members. The random sample of 14,817 veterans was selected on the basis of 229 sampling strata of demographic and military characteristics. All members of the original Seabee battalion and parents of children with Goldenhar syndrome (a craniofacial abnormality) were included in the sample. The survey, conducted in 2007–2009 by mailed questionnaire and telephone interview, contained questions about symptoms used to develop the syndromes derived by Haley et al. (1997b) and symptoms used for other case definitions—such as that of the CDC—and similar conditions, for example, CFS and fibromyalgia. The telephone interview was completed by 8,020 veterans, a 60.1% response rate.
The authors did not replicate the exploratory factor analysis conducted by Haley and colleagues to develop their case definitions of Gulf War illness. Rather, the factor weights from the original Haley et al., study were used to create factor scales and to determine which syndrome fit each person in the study, as was done in both earlier studies of the factor case definition (Haley et al., 1997b, 2001). Results showed similar goodness-of-fit statistics for all three studies. Some 14% of the deployed and 4% of the nondeployed fit any of the six syndromes that make up the factor case definition.
Haley and Tuite (2013) explored the association between Gulf War illness (as defined by Haley and also by CDC) in association with indicators of exposure to chemical warfare agents. Indicators of exposure included self-reports of hearing chemical warfare detection alarms while deployed and modeled exposure to fallout from Khamisiyah based on unit location. This analysis included all 8,020 survey respondents. Results were weighted and reported as a reflection of the entire Gulf War veteran population.
Haley et al. (2013) performed a nested case-control study of 66 veterans who had Gulf War illness syndromes 1, 2, or 3 and 31 control veterans randomly selected from the Military Health Study. Controls were deployed to the Kuwaiti theater of operations but did not meet criteria for Gulf War illness. Participants were admitted to the University of Texas Southwestern Medical Center’s Clinical and Translational Research Center where they completed questionnaires and underwent psychologic evaluation, self-reported autonomic symptoms, and completed the following autonomic tests: pupilometry, lacrimation, quantitative sudomotor axon reflexes, heart rate response to deep breathing and the Valsalva maneuver, sensory testing to heat and pain, and blood pressure and heart rate response to head-up tilt. The 24-hour Holter electrocardiography was completed at home.
Hubbard et al. (2014) sought to investigate whether the cholinergic system is impaired in veterans with Gulf War illness by assessing working memory. The investigators combined event-related fMRI results from participants of two studies: 61 veterans with Gulf War illness and 28 matched controls from the sample described by Iannacchione et al. (2011) and 35 veterans with Gulf War illness and 16 matched controls from the sample described by Haley et al. (1997b). Gulf War illness was defined by Haley’s syndromes 1, 2, and 3. Controls were matched on age, sex, education, handedness, and rank. Functional MRI was used to examine task-related blood oxygen-level-dependent activity in the dorsolateral and ventrolateral prefrontal cortexes during delayed-response task performance.
Gray et al. Studies
The first in a series of studies by Gray and colleagues (1999a) surveyed Seabees who remained on active duty for at least 3 years after the Gulf War. The Seabees were from 14 commands at two locations (Port Hueneme, California, and Gulfport, Mississippi). Those who were deployed to the Gulf War were in mobile construction battalions performing the same tasks and at the same sites as the reserve Seabee battalion studied by Haley et al. (1997b); however, Gray et al. excluded Gulf War veterans who were no longer on active duty at the time of study.
In 1994, 1,497 study subjects were enrolled: 527 Gulf War veterans and 970 nondeployed veterans. The participation rate of eligible Seabees was 53%. The following were administered to the study participants: an eight-page questionnaire regarding medical history, Gulf War exposures, postwar symptoms, hospitalization, and pregnancy outcomes; questions regarding the presence of CFS and PTSD; laboratory tests on sera, blood, urine; and pulmonary function and handgrip strength tests.
Beginning in May 1997, Gray et al. (2002) distributed a mailed questionnaire to all regular and reserve Navy personnel (n = 18,945) who served on active-duty Seabee command during the Gulf War period. The questionnaire collected information regarding medical history, current health status, symptoms, and environmental exposures such as PB. Of the 17,559 participants who were located, 11,868 completed and returned the survey: 3,831 Gulf War deployed, 4,933 deployed elsewhere, and 4,933 nondeployed.
Knoke et al. (2000) used the same Seabee cohort to conduct a factor analysis of the symptoms reported by the veterans on the Hopkins Symptom Checklist to determine whether there was a unique Gulf War syndrome.
Pennsylvania Air National Guard Study
The Volume 4 committee described an investigation of symptom reporting among members of the Air National Guard in Pennsylvania and two derivative studies. However, no additional derivative studies were identified by the Volume 8 or Volume 10 committees.
In response to requests from DoD, VA, and the Commonwealth of Pennsylvania, in 1995, Fukuda and colleagues (1998) conducted a factor analysis study to assess health status and prevalence and causes of an unexplained illness in Gulf War deployed and nondeployed members of a currently active Pennsylvania Air National Guard unit (n = 667). Three demographically similar Air Force units were used as comparison groups (n = 538, 838, and 1,680). Questionnaires regarding military characteristics, demographics, health status, and 35 specific symptoms previously identified to be of concern were distributed and completed by 3,723 participants (1,163 Gulf War deployed, 2,560 nondeployed). Participation rates were as follows: 62% index unit; 35% unit A; 73% unit B; and 70% unit C. To assess symptom prevalence, investigators combined the four units and compared questionnaire responses of deployed and nondeployed. The authors further studied health outcomes in a subset of participants from
the index unit. Of the 490 (45%) deployed members of this unit, 173 (35%) volunteered to participate in the clinical evaluation and completed a mailed clinical questionnaire and the SF-36.
A nested case-control study of the same cohort (n = 1,002) sought to identify self-reported exposures associated with cases of CMI (Nisenbaum et al., 2000). Results indicate that meeting the case definition of severe and mild to moderate illness was associated with use of PB, use of insect repellent, and belief in a threat from biologic or chemical weapons. Having an injury requiring medical attention was also associated with having a severe case of CMI. Nisenbaum et al. (2004) conducted a factor analysis of the symptoms reported by the UK Gulf War veterans combined with those reported by the Pennsylvania veterans.
Hawaii and Pennsylvania Active Duty and Reserve Study
The Volume 4 committee described an early study of active duty and reserve personnel stationed in Hawaii and Pennsylvania who had served in the Gulf War, and two derivative studies. However, no additional derivative studies were identified by the Volume 8 or Volume 10 committees.
One of the first epidemiologic studies of U.S. Gulf War veterans was a congressionally mandated study evaluating the psychologic and physical health of active-duty and reserve Army, Navy, Air Force, and Marine Corps personnel from bases in Hawaii and Pennsylvania (Stretch et al., 1995). Questionnaires were mailed to 16,167 potential study participants and inquired about the following: demographics; physical, psychological, and psychosocial symptoms; deployment type; and perceived sources of stress prior to, during, and after combat or deployment. A total of 4,334 veterans returned the questionnaires for a response rate of 31%. Of those, 715 active duty and 766 reserves were deployed to the Gulf War; 1,576 active duty and 948 reserves were not deployed.
Two derivative studies of Stretch et al. (1995) were identified. In response to the questionnaire, deployed veterans commonly reported significant levels of stress during deployment, including operating in desert climates, long duty days, extended periods in chemical-protective clothing, lack of sleep, crowding, lack of private time, physical workload, and boredom. Significant levels of stress continued postdeployment (Stretch et al., 1996a). Another publication examined PTSD in this cohort (Stretch et al., 1996b). The prevalence of PTSD symptoms was measured by the Impact of Event Scale and the Brief Symptom Inventory.
New Orleans Reservist Studies
The Volume 4 committee described an investigation of symptom reporting among reservists stationed in New Orleans, and two derivative studies. However, no additional derivative studies were identified by the Volume 8 or Volume 10 committees.
A study by Sutker et al. (1995) and colleagues analyzed psychologic outcomes in a cohort of New Orleans reservists (n = 1,520). The cohort consisted of Louisiana National Guard and reservists from the Army, Air Force, and Navy who had been deployed to combat. Of the 1,272 who responded (overall response rate of 83.7%), 876 had been deployed and 396 had not been deployed. A discriminant function model was used to assess the relationship between personal and environmental resources and psychological outcomes.
One derivative study of Sutker et al. (1995) was identified. Assessed by survey at an average of 9 months after the war, veterans completed the Beck Depression Inventory, the Brief Symptom Inventory for Anxiety and Depression, the PTSD checklist, and the Mississippi Scale for PTSD (Brailey et al., 1998).
Multiple Sclerosis Cohort
The Volume 10 committee identified a new cohort of veterans diagnosed with multiple sclerosis (MS), and one derivative study, both published since 2009.
Wallin et al. (2012) identified veterans with MS through VA and DoD disability claims databases. Veterans must have served on active duty between 1990 and 2007. Demographic and medical data were reviewed, and diagnoses were confirmed by study neurologists using standardized criteria. To calculate incidence, the investigators collected annual data on the entire active-duty population between 1990 and 2010 from the Armed Forces Health Surveillance Center. Estimates were stratified by sex, race, and branch of service. The resulting cohort included 2,691 Gulf War era veterans with MS or other clinically isolated syndromes. This cohort is not limited to veterans from the 1990–1991 Gulf War era, and results specific to each era of conflict are not presented. The main limitation of this study may be potential under-ascertainment of cases because any veterans who did not apply for service-connected disability would be missed. Furthermore, because service-connected disability is limited to cases presenting within 7 years of military service, veterans who have MS diagnosed more than 7 years after service may not have applied for service connection. Lastly, this investigation did not consider deployment data, and therefore did not compare MS in deployed vs nondeployed groups.
A second publication from the same authors (Wallin et al., 2014) looked at the incidence of MS and other demyelinating diseases in Gulf War veterans only and investigated disease risk based on self-reported exposures. This study examined 1,841 MS cases (387 deployed; 1,454 nondeployed) who served in 1990–1991. Deployment characteristics were collected from the DMDC, and exposure to sarin and cyclosarin at Khamisiyah were provided by DoD plume modeling. As with the reference study, these results may be affected by incomplete case ascertainment.
Amyotrophic Lateral Sclerosis Cohort
The Volume 10 committee identified a new cohort of veterans diagnosed with amyotrophic lateral sclerosis (ALS), and one derivative study, both published since 2009.
Horner and colleagues (2003) conducted a nationwide, epidemiologic case-ascertainment study in 1990–1999 to determine if Gulf War veterans have elevated rates of ALS. They used active and passive methods of case ascertainment: active methods included screening of inpatient, outpatient, and pharmacy medical databases from VA and DoD; passive methods included establishment of a toll-free telephone number, solicitations through relevant Internet sites, and mass mailings of study brochures to practicing neurologists in VA and to members of the American Academy of Neurology. ALS diagnosis was verified by a review of medical records.
Of the 2.5 million eligible military personnel, nearly 700,000 had been deployed to the Gulf War. Most of the cases were found with active ascertainment methods. The major study limitation was potential under-ascertainment of cases, particularly among nondeployed veterans because nondeployed veterans had less incentive to participate. Because of the rarity of ALS, under-ascertainment of a few cases, particularly if the under-ascertainment is greater among the nondeployed, can substantially exaggerate results.
Volume 4 noted one study published using the same groups of veterans. The same authors undertook a secondary analysis to address concerns about differential case ascertainment among deployed vs nondeployed veterans. In this secondary analysis, Coffman et al. (2005) assessed case ascertainment bias by estimating the occurrence of ALS employing three capture–recapture analysis methods: log-linear models, sample coverage, and ecologic models. The investigators concluded that there might have been some modest under-ascertainment of cases in nondeployed military personnel but little under-ascertainment in the deployed. Results were corrected for case under-ascertainment.
Volume 8 included a study that extended follow-up for 1 year to December 2001 and investigated temporal patterns of ALS occurrence (Horner et al., 2008). A total of 124 ALS cases were confirmed: 48 among deployed and 76 among nondeployed. The authors emphasized that the follow-up period was still too short to draw any conclusions about ALS in nondeployed military personnel when compared to the general population.
The Volume 10 committee identified one derivative study of these ALS cases. Kasarkis et al. (2009) examined the medical records of 135 cases of ALS diagnosed in deployed and era Gulf War veterans between 1990 and 2002, an additional year of follow-up (28 more cases) beyond that studied by Horner et al. (2008). Details pertaining to diagnoses, age of onset, family history, atypical clinical features, and date of initial ventilation were evaluated by neurologists for 109 veterans for whom medical records were available.
After the United States, the United Kingdom provided the most troops for the coalition forces during the Gulf War. The UK veterans have been followed by several research teams. Three reference studies
of UK Gulf War veterans were identified in Volume 4. Two teams of researchers in the UK studied separate, nonoverlapping, stratified random samples of the more than 53,000 military personnel sent to the Gulf War. The first team was from the University of London (Guy’s, King’s, and St. Thomas Medical Schools) (Unwin et al., 1999); the second team was from the University of Manchester (Cherry et al., 2001a,b). In addition, a third team of researchers from the London School of Hygiene and Tropical Medicine surveyed the entire cohort of 53,000 veterans examining birth defects and other reproductive outcomes (Maconochie et al., 2003). The Volume 8 committee identified only one new derivative study that was based on the University of London cohort study. No new studies pertaining to the UK veterans were identified by the Volume 10 committee.
Although not traditional cohort studies per se, UK Defence Analytical Services Agency has periodically assessed the rates and causes of mortality for the entire population of UK Gulf War veterans. The Statistical Notices provide summary statistics on the causes of deaths that have occurred since April 1, 1991. The mortality rates of 53,409 UK Gulf veterans are compared with an era cohort of 53,143 UK Armed Forces personnel of similar age, gender, service, rank, and regular/reservist status who were in service on January 1, 1991, but who did not deploy to the gulf. The statistics include those who died while in service and those who died after they had left the services and are based on deaths reported to the Ministry of Defence. These notices are cited in Chapter 4 in the section on mortality (DASA, 2005, 2009, 2015).
University of London Veteran Studies
At the University of London, Unwin and colleagues (1999) studied the health effects of deployment by randomly sampling the entire UK contingent deployed to the Gulf War (n = 53,462)5; the control groups consisted of those deployed to the conflict in Bosnia (n = 39,217) and service members who were deployed in the same period to noncombat locations outside the United Kingdom (n = 250,000). The era control group was recruited from a subset of nondeployed service members who were fit for combat duty, thus avoiding selection bias related to the healthy warrior effect. Investigators distributed a mailed questionnaire that asked about symptoms (50 items), medical disorders (39 items), exposure history (29 items), functional capacity, and other topics. Potential confounding factors (including sociodemographic and lifestyle factors) were controlled for in multiple logistic regression analysis. Response rates were as follows: 70.4% Gulf War deployed, 61.9% Bosnia cohort, and 62.9% era cohort. The Gulf War deployed veterans reported a higher prevalence of symptoms and diminished functioning than did either comparison group.
The UK Gulf War cohorts completed a second questionnaire with details of the dates they were deployed to each location and the exposures they had experienced. The questionnaire listed 14 exposures, such as combat exposure, number of inoculations, number of days handling pesticides, days exposed to smoke from oil-well fires, and duration of stay in the gulf region. The main analysis involved a multiple regression of each of the seven factors identified through factor analysis on all exposures and other potential confounders. Many of the reported exposures correlated with one another.
Nine derivative studies of this cohort were described in Volume 4 (Hotopf et al., 2003a,b; Macfarlane et al., 2000, 2003, 2005; Nisenbaum et al., 2004; Reid et al., 2001; Rose et al., 2004; Sharief et al., 2002). The Volume 8 committee identified two new studies of this cohort (Ismail et al., 2008; Stimpson et al., 2006).
A follow-up study using a mailed survey was sent 11 years after the war to a stratified random sample of 3,305 participants (1,472 Gulf War deployed, 909 Bosnian deployed, 924 era veterans) from the total who completed the first study. The response rates were as follows: 74.0% Gulf War deployed, 70.2% Bosnia deployed, and 69.7% era veterans. To assess physical symptoms, respondents completed the SF-36. Receiving multiple vaccinations during deployment was weakly associated with five of the six health outcomes, including CMI (as defined by the CDC) (Hotopf et al., 2003a). Hotopf et al. (2003b) studied a subset of these veterans to assess paraoxonase-1 activity and genotype for paraoxonase-1–55 and paraoxonase-1–192. Four groups were selected: deployed veterans who reported physical symptoms after the war (n = 115); healthy deployed veterans (n = 95); symptomatic Bosnia peacekeeping veterans (n = 52); and symptomatic nondeployed military controls (n = 85).
Macfarlane and colleagues (2000) assessed mortality of the entire UK cohort of deployed veterans (n = 53,462) compared with frequency matched controls; the follow-up period was from the end of the Gulf War (April 1, 1991) to March 31, 1999. Results from a further 13-year follow-up (ending June 30, 2004) of the same cohort remained consistent (Macfarlane et al., 2005). The later study examined self-reported exposures (taken from the earlier morbidity studies in 1997–2001) and mortality in the population of UK Gulf War deployed veterans.
Incidence of cancer, as identified by the NHS register, was determined based on follow-up from the end of the Gulf War until July 31, 2002, for all deployed UK service members compared with nondeployed matched controls (era veterans). Of the 51,721 Gulf War veterans included in this analysis, 270 were diagnosed with cancer; of the 50,755 era veterans included in this analysis, 269 were diagnosed (Macfarlane et al., 2003).
Another separate analysis of a subgroup of veterans meeting case criteria for MCS symptoms assed whether they were more likely to report several types of pesticide exposures (Reid et al., 2001).
Rose et al. (2004) and Sharief et al. (2002) conducted a case-control study examining neuromuscular symptoms in 49 Gulf War veterans with more than four neuromuscular symptoms and lower functioning according to the SF-36 compared with 26 healthy Gulf War deployed veterans, 13 symptomatic Bosnia deployed veterans, and 22 symptomatic nondeployed controls. They tested peripheral nerves, skeletal muscles, or neuromuscular junctions.
Nisenbaum et al. (2004) used factor analysis on symptom data from both the UK reference study and the Pennsylvania Air Force unit study (Fukuda et al., 1998; discussed later in this chapter) to look at the interrelationships between symptoms. Each sample was split in half to provide an exploratory and a confirmatory sample. Four correlated factors were identified in each of the samples: respiratory, mood-cognition, gastrointestinal/urogenital, and peripheral nervous.
The Volume 8 committee identified two new studies based on the University of London study. Stimpson and colleagues (2006) used the same population and methods described in Unwin et al. (1999) to specifically examine the prevalence of reported pain and its association with deployment status. Ismail and colleagues (2008) assessed the prevalence of CFS by implementing a two-phase study approach that was a continuum of the Unwin et al. (1999) (phase I) cohort above. Phase II consisted of a random sampling of the 244 veterans (Gulf War and Bosnia deployed, and era veterans) who screened positive
for a physical disability (score less than or equal to 72.2 on the SF-36) in the Unwin study; 111 (45.5%) were Gulf War deployed and 133 (54.5%) were nondeployed veterans (Bosnia: n = 54, era: n = 79).
University of Manchester Veteran Study
Seven years after the Gulf War, the University of Manchester study surveyed a random sample of all UK veterans, distinct from that of Unwin et al. (1999), who deployed between September 1990 and June 1991, as identified by the Ministry of Defense (Cherry et al., 2001a,b). Eligible deployed veterans (n = 9,505) were divided into two groups—main cohort (n = 4,755) and validation cohort (n = 4,750) to permit replication of analysis and to assess consistency. The control population (n = 4,749) was nondeployed veterans in good general health. Veterans were sent a questionnaire about the extent to which they were burdened, within the last month, by any of 95 symptoms. By asking them to mark their answers on a visual analogue scale, investigators sought to determine the degree of symptom severity. Investigators also sought to determine areas of peripheral neuropathy by asking veterans to shade body areas on two mannequins in which they were experiencing pain or numbness and tingling
No derivative studies were identified for this UK cohort.
London School of Hygiene and Tropical Medicine Veteran Study
The third UK study was a very large mail survey that began in August 1998 (with reminders until 2001) (Maconochie et al., 2003). The study was designed to assess reproductive outcomes among Gulf War veterans and also contained open-ended questions regarding general health. The exposed cohort consisted of all UK Gulf War veterans, and the unexposed cohort consisted of a random sample of nondeployed UK military personnel from the same period. Although the participation rates were low (47.3% and 37.5% of male and female Gulf War veterans, respectively, and 57.3% and 45.6% of male and female nondeployed veterans), 25,084 Gulf War veterans and 19,003 nondeployed veterans returned survey responses. The survey included items on reproductive and child health, exposure history, current health, and health of sexual partners; it was supplemented by examination of medical records for pregnancies, live births, and outcomes. Male participants (n = 42,818) reported 27,929 pregnancies in their partners, and female participants (n = 1,269) reported 861 pregnancies. Reports of miscarriages and congenital malformation were clinically validated.
Three derivative studies were described in Volume 4; no new derivative studies were identified by the Volume 8 committee.
Based on results of Maconochie et al. (2003), Doyle et al. (2004) examined the associations between risk of miscarriage, stillbirth, or congenital malformations for deployed and nondeployed women. Maconochie et al. (2004) assessed the risk of infertility in male Gulf War veterans (females were not included in study); self-reports were validated with clinical diagnosis.
No additional derivative studies were identified by the Volume 10 committee.
A national study of all Australian Gulf War veterans conducted in 2000–2002 (Sim et al., 2003) has been extensively assessed and the cohort followed since the initial study. Seven derivative studies were described in Volume 4 (Forbes et al., 2004; Ikin et al., 2004, 2005; Kelsall et al., 2004a,b, 2005; McKenzie et al., 2004), and the Volume 8 committee identified two new derivative studies (Kelsall et al., 2006, 2007). The Volume 10 committee identified two derivative studies based on the original investigation (Kelsall et al., 2009, 2014), a report detailing the results of a recent follow-up assessment conducted in 2011–2012 (Sim et al., 2015), and two publications analyzing data from that follow-up assessment (Gwini et al., 2015; Ikin et al., 2015).
Investigators from Monash University in Australia conducted the Australian Gulf War Veterans’ Health Study by examining all 1,871 Australian veterans deployed to the Gulf War region from August 2, 1990, to September 4, 1991; naval personnel made up 86.5% of this cohort (Sim et al., 2003). The control group consisted of 2,924 nondeployed Australian Defence Force personnel matched by service type, sex, age, and military status. Participation rates were 81% (n = 1,456) for the deployed and 57% (n = 1,588) for the control group.
A mailed questionnaire was distributed in 2000–2002, which included the SF-12, General Health Questionnaire (GHQ)-12, and questions regarding physical and psychological health, military service history, and exposures during deployment. In addition, participants were asked to attend one of 10 Health Services Australia medical clinics to undergo a comprehensive health assessment, a full physical examination, blood work, and fitness tests. Interview-administered questionnaires such as the Composite International Diagnostic Interview (CIDI) were given to all participants to assess mental health (Sim et al., 2003). Gulf War veterans reported more general health symptoms and more severe symptoms than the nondeployed controls.
Kelsall et al. (2004a) reported the prevalence and severity of symptoms experienced by veterans assessed by Sim et al. (2003). Ikin et al. (2004) used responses to the CIDI and a service experience questionnaire to assess the relationship between the presence of a psychiatric disorder and the veterans’ perceptions of stressors during deployment. Very few personnel experienced direct combat; however, despite their lack of combat exposure, deployment was a stressful event with veterans experiencing higher rates of fear and threat of entrapment, attack (including nerve agent warfare), and death or injury. In a follow-up study, Ikin et al. (2005) reported associations between stressful experiences and demographic and deployment characteristics.
McKenzie et al. (2004) reported on the psychological health and stressful experiences of Australian Gulf War veterans using three standard instruments to measure functioning and psychological health. Symptom severity associated with Gulf War exposures (vaccinations, PB, pesticides and insect repellants, chemical weapons, and stressful situations) was assessed by Kelsall and colleagues (2004a). Kelsall et al. (2004b) also assessed the respiratory health status of a random sample of the veterans using a questionnaire, spirometric testing, and a physical examination. Health status was assessed in relation to reported exposure to smoke from the Kuwaiti oil-well fires and dust storms.
In 2005, Kelsall et al. studied the neurological status of the Australian cohort. Of the 1,424 deployed veterans who completed the mailed questionnaire, 1,382 undertook the neurological examination; 1,376 of the 1,548 controls completed both the questionnaire and neurological examination (described in Sim et al., 2003, above).
Forbes et al. (2004) used factor analysis to attempt to group symptom complexes for this cohort. This study confirms the greater extent and severity of symptoms in Gulf War veterans, even in a predominantly naval population with few direct military attacks, no deaths, and few casualties.
The Volume 8 committee identified two additional derivative studies of the Australian Gulf War veterans. Kelsall et al. (2006) conducted a study on the prevalence of CFS from August 2000 to April 2002. The participation rates were as follows: 80.5% (1,456) of the 1,808 eligible veterans and 56.8% (1,588) of 2,796 controls. Of those, 1,384 deployed veterans and 1,379 controls completed both the mailed questionnaire and medical assessment. In addition to questions on general health, fitness tests, laboratory work, and pulmonary function, clinical examiners specifically inquired about any tiredness or fatigue following normal activities and its duration within the last 12 months.
Male reproductive health was assessed for the 1,424 deployed and 1,548 nondeployed veterans who completed the mailed questionnaire; response rates were as noted in Kelsall et al. (2006). Questions of interest included those related to pregnancy outcomes (live birth, miscarriage, stillbirth), and for live births, participants were asked about date, weight, gestation, and birth defects (Kelsall et al., 2007).
Five new studies based on the Australian Gulf War Veterans’ Health Study cohort were identified by the Volume 10 committee.
Kelsall et al. (2009) assessed comorbidities associated with Gulf War illness among 1,381 eligible Australian male veterans who had deployed to the Gulf War, a comparison group of 1,085 veterans who had been on active service during the war but had not deployed, and 292 veterans who had deployed elsewhere during the war. All the veterans completed a 63-item symptom questionnaire, as well as the GHQ-12, the SF-12, and the Alcohol Use Disorders Identification Test. Veterans also received an in-person health assessment that included a full physical examination, with lung function and fitness tests, and a mental health assessment using the psychologist-administered, computer-assisted CIDI. Multisymptom illness was defined using a modification of the CDC definition.
In another study, Kelsall et al. (2014) compared the 1,381 Australian veterans with 1,377 veterans who were serving in the military at the time or had previously deployed. The assessment queried veterans about doctor-diagnosed arthritis or rheumatism, back or neck problems, joint problems, and soft tissue disorders. Medical practitioners then rated the self-reported diagnoses as nonmedical, unlikely, possible, or probable; only probable diagnoses were analyzed. This approach added a level of medical judgment to the self-reports but did not verify the self-reported diagnoses with a clinical evaluation.
Sim et al. (2015) reported the results of the Australian Gulf War Veterans’ Follow Up Health Study. This assessment was conducted in 2011–2013, and all participants of the original 2003 study were eligible to participate; 715 Gulf War veterans and 675 comparison group veterans participated in the follow-up study (50% participation rate). The follow-up study collected much of the same information as the original study, but it also inquired about additional outcomes including pain, sleep disturbance, injury, musculoskeletal disorders, demoralization, and measures of well-being (quality of life, life satisfaction, life events, financial distress, suicidal ideation, and community participation). The follow-up also extended exposure assessment efforts from those used in the baseline study. Data were collected by mailed questionnaire, telephone interview, collection of VA’s health data, and claims history from Medicare, Pharmaceutical Benefits Scheme, and Repatriation Pharmaceutical Benefits Scheme. Other databases included national mortality and cancer registries and information from the baseline study.
Ikin et al. (2015) reported additional results of the Australian follow up study compared to the results of the previous survey. Depression and depression symptoms were measured by the CIDI and other heath
questions included in the survey. The authors compared depression measured in the baseline survey (Ikin et al., 2004) to responses collected in the latest survey to report patterns of persistence, remittance, and new onset depression. The analyses included 715 Gulf War veterans and 675 nondeployed veterans (about a 50% participation rate).
To assess the persistence of symptoms in Gulf War veterans, Gwini et al. (2015) compared responses to a 63-item symptom checklist in the first survey (Sim et al., 2003) and the second survey (Sim et al., 2015). Fifty-four percent of veterans and 47% of comparison veterans surveyed responded to the second survey. The analysis adjusted for age, rank, and branch of service. The authors made three main comparisons: between groups to show longitudinal change; symptoms reported in the second survey among respondents who reported symptoms in the first survey to show persistence; and new symptoms reported by respondents who did not report symptoms in the first survey to show incidence.
Danish service members were sent to the Persian Gulf at the end of the Gulf War as peacekeepers (Ishoy et al., 1999b). Volume 4 cited four derivative studies based on the initial cohort. Neither the Volume 8 nor the Volume 10 committees identified any additional derivative studies on Danish Gulf War veterans.
Military personnel from Denmark were involved in peacekeeping or humanitarian missions occurring predominantly after the Gulf War ceasefire, but were located in the same areas as other coalition forces who served in Gulf War combat (Ishoy et al., 1999b). A total of 821 Danes, deployed between August 1990 and December 1997, were eligible for inclusion in this population-based cohort; 686 (83.6%) agreed to participate in the study. The deployed veterans were matched by age, sex, and profession to 400 members of the Danish armed forces who were not deployed to the Gulf War; the participation rate was 57.8% (n = 231). Participants completed a detailed questionnaire that included 22 neuropsychologic symptoms, and then participants received detailed clinical health and laboratory examinations (e.g., height, weight, blood pressure, battery of urinary and blood work, battery of neuropsychologic tests) and physician interviews about their medical history and symptoms. Gulf War participants were also asked about their exposures while in the gulf. The examinations were conducted between 1997 and 1998.
Proctor et al. (2003) assessed neuropsychologic symptoms in the Danish Gulf War veterans. Gastrointestinal symptoms and diseases and symptoms related to the skin or allergies were evaluated by Ishoy et al. (1999a) in relation to specific Gulf War exposures. The investigators also examined male participants for sexual dysfunction and reproductive health. Self-reports of sexual problems were validated with medical examinations and laboratory testing, such as reproductive hormone parameters (Ishoy et al., 2001a,b).
One analysis investigated whether 22 neuropsychologic symptoms were associated with 18 self-reported environmental exposures (physical, chemical, biologic exposures and psychological stressors) (Suadicani et al., 1999).6
6 Exposures did not include pyridostigmine bromide or vaccinations against chemical or biologic warfare agents, because Danish veterans had a peacekeeping role and thus were not at risk for chemical or biologic warfare.
In Volume 4, the Canadian Gulf War Veterans Study was not included in the presentation of major cohort studies. The Canadian study (Goss Gilroy Inc., 1998) was considered to be a secondary study in Volume 4, and the Volume 8 committee agreed with that classification because the study relied on self-reports gathered through a mailed questionnaire. The Volume 8 committee identified one new study based on the Canadian Gulf War veterans (Statistics Canada, 2005), although it is unclear if the cohort in the new study was the same as that studied by Goss Gilroy. The Volume 10 committee did not find any additional studies of this cohort.
The Canadian Department of National Defence tasked Goss Gilroy (1998) to compare the overall health status and prevalence of symptoms in all Canadian Forces personnel who deployed to the Gulf War conflict with a representative sample of Canadian personnel who served in the military at the same time but who had not been deployed to the Persian Gulf. Military personnel were also compared with the general population as represented by the 1990 Ontario Health Survey. A questionnaire was mailed out to 9,947 participants in 1997 and included information on sociodemographic factors, medical history and current status, and Gulf War deployment and exposure. The response rates were 73% (n = 3,113) for the Gulf War deployed and 60.3% (n = 3,439) for the deployed elsewhere controls. The report described the effects of confounders such as income and rank and presented associations between Gulf War exposures and self-reported adverse health outcomes (Goss Gilroy Inc., 1998).
A 9-year follow-up study was conducted on mortality and cancer incidence in the cohort of deployed Canadian personnel (n = 5,117) compared to Canadian Forces who were eligible but not deployed to the gulf (n = 6,093). The national mortality database and national cancer registry provided the information necessary for record linkage. There were 42 deaths in the deployed cohort and 54 deaths in the nondeployed (Statistics Canada, 2005).
As the committee organized the broad body of evidence described above, they made several overarching observations. First, there are numerous studies on Gulf War veterans, from several countries, and by various research groups, but these studies are all limited in a few common ways, such as lack of exposure data or reliance on self-reported outcomes, that can never be overcome. As a result there will continue to be gaps in the epidemiologic literature on these cohorts that may never be resolved. Second, sex-specific and race/ethnicity-specific data are rarely reported. Even though women are oversampled in some studies such as those conducted by VA (Bossarte, 2014; Kang et al., 2000, 2009) they are excluded from analyses in others (for example, studies of the Australian cohort, Sim et al., 2003, 2015). VA also oversamples other groups such as National Guard and reservists, and results for these groups are occasionally reported separately. Race and ethnicity are usually adjusted for in analyses but are otherwise virtually ignored by the studies in this chapter. Lastly, many derivative studies have moved from characterizing the health of veterans to looking for etiology and biomarkers of certain health outcomes (such as use of fMRI in Gulf War illness). Those studies are discussed briefly in Chapter 4.
TABLE 3-1 Reference and Derivative Studies for the Major Gulf War Cohorts
|Cohort/Reference Study||Derivative Studies||Purpose/Outcome|
|VA National Health Survey of Gulf War Veterans and Their Families, Kang et al., 2000|
|Volume 4||Kang et al., 2001||Self-reported birth defects|
|Kang et al., 2002||Association of symptom clusters with self-reported exposures|
|Kang et al., 2003||Prevalence of PTSD and chronic fatigue syndrome|
|Davis et al., 2004||Presence of distal symmetric polyneuropathy|
|Karlinsky et al., 2004||Pulmonary function and self-reported respiratory symptoms|
|Eisen et al., 2005||Numerous health outcomes and general health assessment|
|Kang et al., 2005||Role of sexual assault and harassment on the risk of PTSD|
|Volume 8||Blanchard et al., 2005||Prevalence of chronic multisymptom illness|
|Page et al., 2005a,b||Possible exposure at Khamisiyah and self-reported morbidity|
|Toomey et al., 2007||Prevalence of psychiatric disorders, symptom self-report, and quality-of-life status|
|Kang et al., 2009||Self-reported general health status|
|Toomey et al., 2009||Neuropsychological functioning|
|New||Wallin et al., 2009||Neuropsychological performance and quality of life|
|Coughlin et al., 2011a||Relationship between obesity, PTSD, CFS-like illness, and multisymptom illness|
|Coughlin et al., 2011b||Relationship between alcohol use and PTSD, depression, multisymptom illness, and CFS-like illness|
|Li et al., 2011a||Prevalence, incidence, and persistence of self-reported health outcomes and health status|
|Bossarte, 2014; Dursa et al., 2016||Self-reported health conditions and symptoms (unpublished results of wave 3 survey)|
|Li et al., 2014||Autonomic function in veterans with Gulf War illness|
|Iowa Veterans/Iowa Persian Gulf Study Group, 1997|
|Volume 4||Black et al., 1999||Impact of multiple chemical sensitivity on quality of life and utilization of health services|
|Black et al., 2000||Risk factors and prevalence of multiple chemical sensitivity|
|Doebbeling et al., 2000||Factor analysis of self-reported symptoms (definition of Persian Gulf War syndrome)|
|Zwerling et al., 2000||Prevalence of self-reported postwar injuries|
|Barrett et al., 2002||Association between PTSD and self-reported physical health issues|
|Lange et al., 2002||Exposure to Kuwait oil fires and risk of asthma and bronchitis|
|Black et al., 2004a||Prevalence of psychiatric disorders|
|Black et al., 2004b||Prevalence and risk factors for anxiety disorders|
|Volume 8||Ang et al., 2006||Identification of predictors of chronic widespread pain|
|Black et al., 2006||Prevalence of borderline personality disorder|
|Forman-Hoffman et al., 2007||Prevalence of self-reports of symptoms of chronic widespread pain|
|Oregon and Washington Veterans, McCauley et al., 1999a|
|Volume 4||Bourdette et al., 2001||Prevalence of unexplained illness|
|Spencer et al., 2001||Self-reported exposure and unexplained illness|
|Cohort/Reference Study||Derivative Studies||Purpose/Outcome|
|Kansas Veterans Study, Steele et al., 2000|
|New||Steele et al., 2012||Wartime exposures related to Gulf War illness|
|Steele et al., 2015||Butyrylcholinesterase genotype and gene-exposure interaction in Gulf War illness|
|Ft. Devens and New Orleans Cohorts, Proctor et al., 1998|
|Volume 4||Proctor et al., 2001a||Assessment of health-related quality of life|
|Proctor et al., 2001b||Overlap between symptoms of chronic fatigue and chemical sensitivity, and the case definition for chronic multisymptom illness|
|White et al., 2001||Self-reported exposure and neuropsychological functioning|
|Volume 8||Proctor et al., 2006||Possible exposure at Khamisiyah and neuropsychological functioning|
|Seabees, Haley et al., 1997b|
|Volume 4||Haley and Kurt, 1997; Haley et al., 1997a||Self-reported exposure to neurotoxicants and nervous system–based syndromes|
|Haley et al., 1999||Genetic susceptibility and risk of neurologic damage|
|Volume 8||Haley et al., 2000a,b||Basal ganglia injury|
|Roland et al., 2000||Vestibular function|
|Haley et al., 2004||Autonomic function|
|Haley et al., 2009||Regional cerebral blood flow|
|New||Tillman et al., 2010, 2012, 2013||Event-related potentials in response to stimuli|
|Li et al., 2011b||Hippocampal blood flow assessed by MRI|
|Liu et al., 2011||Cerebral blood flow assessed by MRI|
|Gopinath et al., 2012||Sensory and pain processing assessed by fMRI|
|Odegard et al., 2013||Memory impairment assessed by fMRI|
|Moffett et al., 2015||Impairments in complex language assessed by fMRI|
|U.S. Military Health Survey, Iannacchione et al., 2011|
|New||Haley and Tuite, 2013||Indicators of exposure to chemical warfare and Gulf War illness|
|Haley et al., 2013||Measures of cholinergic autonomic dysfunction in Gulf War illness|
|Hubbard et al., 2014||Executive functioning assessed by fMRI|
|Seabees, Gray et al., 1999a|
|Volume 4||Knoke et al., 2000||Self-report of symptoms|
|Gray et al., 2002||Self-report of symptoms and general health status|
|Cohort/Reference Study||Derivative Studies||Purpose/Outcome|
|Pennsylvania Air National Guard Veterans, Fukuda et al., 1998|
|Volume 4||Nisenbaum et al., 2000||Self-reported exposures and chronic multisymptom illness|
|Nisenbaum et al., 2004||Factor analysis of self-reported symptoms|
|Hawaii and Pennsylvania Active-Duty and Reserve, Stretch et al., 1995|
|Volume 4||Stretch et al., 1996a||Prevalence of psychiatric disorders|
|Stretch et al., 1996b||Prevalence of PTSD|
|New Orleans Reservists, Sutker et al., 1995|
|Volume 4||Sutker et al., 1995||Prevalence of psychologic disorders|
|Brailey et al., 1998||Prevalence of psychiatric disorders|
|Multiple Sclerosis Cohort, Wallin et al., 2012|
|New||Wallin et al., 2014||Risk of MS|
|Amyotrophic Lateral Sclerosis Cohort, Horner et al., 2003|
|Volume 4||Coffman et al., 2005||Risk of ALS|
|Volume 8||Horner et al., 2008||Risk of ALS|
|New||Kasarkis et al., 2009||Clinical features of ALS|
|UK Veterans: University of London, Unwin et al., 1999|
|Volume 4||Macfarlane et al., 2000||Self-reported exposure and mortality|
|Reid et al., 2001||Self-reported exposure and multiple chemical sensitivity and chronic fatigue syndrome|
|Sharif et al., 2002||Neuromuscular symptoms evaluated through objective tests|
|Hotopf et al., 2003a,b||Neurologic assessments|
|Macfarlane et al., 2003||Incidence of cancer|
|Nisenbaum et al., 2004||Factor analysis of self-reported symptoms|
|Rose et al., 2004||Neuromuscular symptoms evaluated through objective tests|
|Macfarlane et al., 2005||Self-reported exposure and mortality|
|Volume 8||Stimpson et al., 2006||Self-report of chronic widespread pain|
|Ismail et al., 2008||Prevalence of chronic fatigue and related disorders through assessment|
|Cohort/Reference Study||Derivative Studies||Purpose/Outcome|
|UK Veterans: University of Manchester, Cherry et al., 2001a,b|
|UK Veterans: London School of Hygiene and Tropical Medicine, Maconochie et al., 2003|
|Volume 4||Doyle et al., 2004||Prevalence of miscarriage, stillbirth, and congenital malformations|
|Maconochie et al., 2004||Self-report of fertility problems|
|Simmons et al., 2004||Self-report of medical symptoms or disease|
|Australian Veterans, Sim et al., 2003|
|Volume 4||Forbes et al., 2004||Factor analysis of self-reported symptoms|
|Ikin et al., 2004||Prevalence of psychiatric disorders|
|Kelsall et al., 2004a||Association between self-reported exposures with numerous symptoms and medical conditions|
|Kelsall et al., 2004b||Self-reported exposure and respiratory health status|
|Mckenzie et al., 2004||Psychological health and functioning|
|Kelsall et al., 2005||Self-report of exposures and neurological symptoms|
|Ikin et al., 2005||Self-report of exposure to stressful events|
|Volume 8||Kelsall et al., 2006||Self-reported exposure and prevalence of chronic fatigue syndrome|
|Kelsall et al., 2007||Self-reported birth defects and other pregnancy outcomes|
|New||Kelsall et al., 2009||Comorbidities of multisymptom illness|
|Kelsall et al., 2014||Psychological comorbidities of musculoskeletal disorders|
|Gwini et al., 2015||Prevalence, persistence, and incidence of symptoms|
|Sim et al., 2015||Australian Gulf War Veterans Follow Up Health Study—all health outcomes|
|Danish Peacekeepers, Ishoy et al., 1999b|
|Volume 4||Ishoy et al., 1999a||Prevalence of gastrointestinal symptoms and diseases, skin disease, and respiratory symptoms and function|
|Suadicani et al., 1999||Associations between exposures and self-reported neuropsychological symptoms|
|Ishoy et al., 2001a,b||Self-report of sexual dysfunction and birth defects|
|Proctor et al., 2003||Prevalence of neuropsychiatric symptoms and neurobehavioral performance|
|Canadian Veterans, Goss Gilroy Inc., 1998|
|Volume 8||Statistics Canada, 2005||Mortality rate and cancer incidence|
NOTE: ALS = amyotrophic lateral sclerosis; CFS = chronic fatigue syndrome; fMRI = functional magnetic resonance imaging; GWI = Gulf War illness; MRI = magnetic resonance imaging; MS = multiple sclerosis; PTSD = posttraumatic stress disorder.