2
Studies of the Health of Gulf War Veterans

Introduction

The second component of the charge to this committee is to identify issues to address in the development of study designs and methods used to answer identified questions regarding the health of Gulf War veterans. The committee began by reviewing and evaluating previous studies to determine what they can tell us about the health of these veterans. In this chapter we first review published studies that examine Gulf War veterans' health. This is followed by a discussion of the methodological strengths and limitations of the research to date. Finally, this chapter describes the evidence available to date on adverse health effects related to service in the Gulf War and the basis that current evidence provides for the design of the study described in Chapter 5.

A rich literature exists on adverse health effects of military conflicts, including effects attributed to exposure to mustard gas and Agent Orange. Additionally, a recent article by Hyams and colleagues (1996) summarized reports of poorly understood, multisymptom clusters recorded in conflicts dating back to at least the Civil War. They found reports of symptoms include fatigue, shortness of breath, headache, sleep disturbance, forgetfulness, and impaired concentration. No single etiologic entity has been discovered to account for these symptoms in conflicts prior to the Gulf War, and no generally accepted diagnostic label or set of clear clinical criteria has been developed out of earlier conflicts to use in the assessment of health problems among Gulf War veterans.

For earlier conflicts, various stress-related or psychogenic explanations were put forward to account for the cluster of symptoms observed among veterans. Shell shock, combat fatigue, irritable heart, or effort syndrome are all terms that have been used to both label and possibly explain the observed symptomatology. None of the labels was particularly useful in leading to clear diagnostic criteria, effective preventive interventions, or effective clinical treatments. There was a clear stigma of weakness associated with many of the labels and associ-



The National Academies | 500 Fifth St. N.W. | Washington, D.C. 20001
Copyright © National Academy of Sciences. All rights reserved.
Terms of Use and Privacy Statement



Below are the first 10 and last 10 pages of uncorrected machine-read text (when available) of this chapter, followed by the top 30 algorithmically extracted key phrases from the chapter as a whole.
Intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text on the opening pages of each chapter. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

Do not use for reproduction, copying, pasting, or reading; exclusively for search engines.

OCR for page 24
2 Studies of the Health of Gulf War Veterans Introduction The second component of the charge to this committee is to identify issues to address in the development of study designs and methods used to answer identified questions regarding the health of Gulf War veterans. The committee began by reviewing and evaluating previous studies to determine what they can tell us about the health of these veterans. In this chapter we first review published studies that examine Gulf War veterans' health. This is followed by a discussion of the methodological strengths and limitations of the research to date. Finally, this chapter describes the evidence available to date on adverse health effects related to service in the Gulf War and the basis that current evidence provides for the design of the study described in Chapter 5. A rich literature exists on adverse health effects of military conflicts, including effects attributed to exposure to mustard gas and Agent Orange. Additionally, a recent article by Hyams and colleagues (1996) summarized reports of poorly understood, multisymptom clusters recorded in conflicts dating back to at least the Civil War. They found reports of symptoms include fatigue, shortness of breath, headache, sleep disturbance, forgetfulness, and impaired concentration. No single etiologic entity has been discovered to account for these symptoms in conflicts prior to the Gulf War, and no generally accepted diagnostic label or set of clear clinical criteria has been developed out of earlier conflicts to use in the assessment of health problems among Gulf War veterans. For earlier conflicts, various stress-related or psychogenic explanations were put forward to account for the cluster of symptoms observed among veterans. Shell shock, combat fatigue, irritable heart, or effort syndrome are all terms that have been used to both label and possibly explain the observed symptomatology. None of the labels was particularly useful in leading to clear diagnostic criteria, effective preventive interventions, or effective clinical treatments. There was a clear stigma of weakness associated with many of the labels and associ-

OCR for page 24
ated explanatory theories, so they were understandably not widely embraced by veterans or their families. The Vietnam experience contributed case criteria and a label for posttraumatic stress disorder (PTSD). The disorder itself was not new, but the recognition of it and the eventual incorporation of PTSD into the standard medical and psychiatric diagnostic coding systems (ICD-9 and DSM-III [Diagnostic and Statistical Manual], respectively) date from the Vietnam era. Additionally, concerns regarding health effects related to Gulf War exposures are similar to the concerns that emerged regarding Agent Orange exposure for Vietnam veterans. Health problems in Gulf War veterans, then, are studied with the recognition of their potential similarity to problems in other conflicts, but not necessarily with the acceptance of prior causal explanations. Most studies take Gulf War deployment (yes/no) as the measure of exposure, because the set of potential causal factors for postwar health problems was not well understood at the time of the conflict and is not much more clearly understood today. A few studies, however, focus on a specific chemical or biological exposure (again, usually yes/no rather than a graded scale). The studies discussed below typically are designed to detect an unusually high frequency or severity of health problems in a group of veterans who served in the Gulf, compared to either veterans of the same period who were not deployed or a more general population who did not serve. Specific Studies of Gulf War Veterans Mortality Two studies focused on mortality among Gulf War-deployed veterans compared to similar veterans not deployed to the Gulf. The study by Kang and Bullman (1996) used the Beneficiary Identification and Records Locator System (BIRLS) of the Department of Veterans Affairs (VA) to track deaths in nearly the entire population of deployed Gulf War veterans. Death rates in the deployed group were compared with the rate in a similar-sized control group of active duty, National Guard, and reserve personnel who served during the Gulf War period but were not deployed. Deaths for any cause were tracked through September 1993. Writer et al. (1996) compared deaths occurring during the conflict among the population of deployed veterans and among a large population of nondeployed veterans in the 1990–1991 period. Data on deaths were obtained from Report of Casualty forms (DD Form 1300). The Kang and Bullman study is perhaps more directly relevant to the committee's charge, because it was designed to detect excess mortality among deployed veterans in at least the immediate post war period. No such excess was found, except for deaths attributed to accidents. A similar finding was also noted among Vietnam veterans and may reflect a set of risk-taking phenomena among veterans returning from conflict rather than a chemical, biological, or psychological exposure leading to physiological change. During the 1998 meeting of

OCR for page 24
Federal Investigators on Gulf War Illness, Kang presented similar results based on an updated analysis of mortality data through 1996. The Writer study also found no excess mortality among deployed veterans once deaths directly related to combat were excluded. Data from these studies indicate no measurable increase in mortality, other than accidents, either during the Gulf War itself or in the period through 1996. Disease Incidence and Prevalence Some studies have specifically focused on incidence of diagnosed disease (using ICD-9 or ICD-10 labels and codes) in cohorts of Gulf War veterans. Most notably, diagnostic data are available from the VA's Persian Gulf Registry (DVA, 1999) and the DoD's Comprehensive Clinical Evaluation Program (IOM, 1998). Tables 2-1 and 2-2 present the distribution of diagnoses by major disease categories. It is important to note that data collection began in 1992 but in 1996 a revised code sheet was implemented. This revision makes it impossible to aggregate complaint, symptom, and diagnostic data for the total population of veterans whose data are recorded in the VA registry. The revision does, however, allow for a greatly expanded self-report exposure history questionnaire. In addition, although the original recording form allowed up to 3 symptoms and up to 3 diagnoses to be listed, the revised form allows expanded recording of up to 10 symptoms and up to 10 diagnoses. Although absolute prevalence rates are different in the two series, the relative distribution across major categories is very similar, with diseases of the musculoskeletal system and mental disorders being the two leading categories in both sets of patients. Because the data come from self-selected case series (i.e., veterans who presented for care in special Gulf War health assessment and treatment programs), the relative prevalence of various diagnoses is informative, but no conclusions can be drawn about risk for various diseases among Gulf War veterans versus nondeployed veterans or a civilian population with similar demographic characteristics. Data obtained in a similar manner were reported in a study of British Gulf War veterans by Coker and colleagues (1999). They found that 59 percent of veterans who presented for care at a Ministry of Defense Gulf War Assessment Programme had more than one diagnosis. The five most common categories of diagnosis (in order of prevalence) were diseases of the musculoskeletal system, mental or behavioral disorders, chronic fatigue syndrome, diseases of the respiratory system (mainly asthma), and diseases of the digestive system. Twelve percent of the participants had a PTSD diagnosis. Because the study did not seek to compare the symptom experience of deployed and nondeployed veterans, the prevalence of diagnoses and the relative distribution of diagnoses across organ systems are useful as descriptions of the problems found in a selected case series of program participants, but these data are not generalizable to all British Gulf War veterans or Gulf War veterans of other nationalities.

OCR for page 24
Two studies focused specifically on diagnosis of birth defects among infants born to U.S. Gulf War veterans. Cowan and colleagues (1997) examined the incidence of diagnosed birth defects in children born in the 1991–1993 time period to Gulf War veterans remaining on active duty. Records for all live births at 135 U.S. military hospitals were examined. Birth defects were identified from medical records that included any ICD-9-CM codes in the range 740–759, plus codes for neoplasms or hereditary diseases. The study included 33,998 infants born to veterans deployed to the Gulf and 41,463 infants born to nondeployed veterans. The risk of birth defects was the same for deployed as for nondeployed veterans (odds ratio .97 for male veterans, 1.07 for female veterans), and rates for both groups were very similar to rates in the civilian population. Length of service in the Gulf was not associated with risk of birth defects, and a separate analysis of severe birth defects also showed no association with service in the Gulf. Araneta and colleagues (1997) found among newborns of Gulf War veterans a relative risk of 3.0 for Goldenhar Syndrome (a rare abnormality of facial structure), but the increased risk was not statistically significant. A small number of cases produced very broad confidence intervals (0.63–20.6) around this relative risk estimate, indicating that it could have been produced by chance alone. TABLE 2-1 Distribution of Diagnoses for Participants in the VA Persian Gulf Registry (PGR)a   Original Code Sheetb (n = 53,935) Revised Code Sheetc (n = 19,721) Diagnosis No. % No. % Musculoskeletal and connective tissue 13,299 25.5 7,286 36.9 Mental disorders 7,995 15.1 6,887 34.9 Respiratory system 7,540 14.3 3,626 18.4 Skin and subcutaneous tissue 7,144 13.5 3,813 19.3 Digestive system 6,028 11.4 3,451 17.5 Nervous system 4,398 8.3 3,441 17.4 Circulatory system 3,747 7.1 2,083 10.6 Infectious diseases 3,715 7.0 1,785 9.1 Injury and poisoning 2,485 4.7 2,020 10.2 Genitourinary system 1,774 3.4 1,126 5.7 Neoplasm 232 0.4 149 0.8 a As of February 1999; data were prepared by the DVA Environmental Epidemiology Service. b Data were collected using the original code sheet implemented in 1992 and used until 1996. c Data were collected using the revised code sheet implemented in 1996.

OCR for page 24
TABLE 2-2 Distribution of Diagnoses for Participants in the DoD CCEP   Original Code Sheeta (n = 18,495) Revised Code Sheetb (n = 10,242) Diagnosis No. % No. % Musculoskeletal 3,419 18.8 2,140 22.1 Psychoses and mental disorders 3,385 18.6 1,645 17.0 Signs, symptoms, and ill-defined conditions 3,236 17.8 1,721 17.7 Healthy 1,712 9.4 583 6.0 Respiratory system 1,226 6.7 597 6.2 Digestive system 1,141 6.3 626 6.5 Skin and subcutaneous tissue 1,152 6.3 526 5.8 Nervous system and sensory organs 1,047 5.8 512 5.3 Infectious diseases 457 2.5 268 2.8 Circulatory system 383 2.1 302 3.1 Endocrine/nutritional metabolic diseases 366 2.0 248 2.6 Genitourinary system 226 1.2 140 1.4 Injury and poisoning 140 0.8 117 1.2 Neoplasms 153 0.8 102 1.1 Blood and blood-forming organs 105 0.6 102 0.7 a Data were collected from inception until April 1996 using the original code sheet. b Data were collected from April 1996 through March 1998 using the revised code sheet. Signs, Symptoms, and Specific Impairments Preliminary reports on unusual clusters of symptoms among Gulf War veterans began to appear in the media in late 1992, and publications in peer-reviewed journals began to appear in 1993. Perconte (1993) and Southwick (1993) reported on what they described as psychological and war stress symptoms and trauma-related symptoms among relatively small numbers of Army, Marine, and National Guard reservists who had been deployed to the Gulf. The analysis focused on mental health issues in general, and symptoms of PTSD in particular. Both studies noted relatively high prevalence of PTSD symptoms among deployed veterans, and the Perconte study also noted higher levels of depression and general psychiatric symptomatology in deployed veterans than in a comparison group of nondeployed veterans. Sostek and colleagues (1996) found a higher prevalence of gastrointestinal (GI) complaints (e.g., abdominal pain, gas, loose stools, and nausea and vomiting) among members of a National Guard unit who had been deployed to the Gulf than among members of the same unit who had not been deployed. The

OCR for page 24
deployed veterans had GI symptoms while serving in the Gulf that continued after return to the United States. These data were obtained in a self-report survey focusing on GI symptoms. No clinical examinations were performed, and no attempt was made to provide clinical diagnoses for the symptoms reported. In a larger study of deployed veterans, Stretch and colleagues (1995) surveyed all active duty and reserve personnel in Army, Navy, Marine, and Air Force units in Hawaii and Pennsylvania. Questionnaires were distributed to approximately 16,000 individuals through their units; responses were received from 4,334, of whom 1,739 had been deployed to the Gulf. Deployed veterans were two to three times as likely as nondeployed veterans to report a variety of symptoms. The investigators concluded that the symptoms were related to deployment and that the physical symptoms were more strongly associated with deployment than psychological health measures (Stretch et al., 1996a). An algorithm based on a cluster of symptoms (e.g., fatigue, malaise, rash, headache, and respiratory symptoms) that might characterize a possible "Gulf War syndrome" identified 178 deployed veterans, as well as 55 nondeployed veterans. The investigators noted a four-fold difference in PTSD prevalence rates between deployed veterans (9.2%) and nondeployed veterans (2.1%) (Stretch et al., 1996b). A study by Proctor and colleagues (1998) compared the symptom experience of Gulf War veterans from New England and New Orleans to that of members of a single National Guard unit who had been sent to Germany during the same time period. Symptoms from the 52-item Expanded Health Symptom Checklist were grouped into nine body-system clusters; scores for the body system groupings were the sum of 0–4 frequency ratings for either the three symptoms in the group or, if more than three symptoms were included in a body-system group, the three most representative symptoms (determined by expert judgment). Veterans also reported on exposure to various toxic substances, biological agents, and combat stressors, and were assessed for PTSD using two different instruments. Gulf War veterans reported a higher level of symptom experience than those deployed to Germany on all but one of the symptom groups. The largest differences between the Gulf-deployed and Germany-deployed cohorts (in terms of odds ratios) were for dermatological, neuropsychological, and gastrointestinal symptoms. Rates of PTSD were also higher in Gulf-deployed veterans, and their self-reported general health status scores were lower. The meaning of these differences is uncertain, though, as individuals from the New England and New Orleans cohorts were selected for study based on symptom reports in 1992–1993 while the Germany-deployed cohort was not. Even though associations were found between specific wartime exposures and presence of symptoms, there is uncertainty about the significance of this association because the data on exposures were from self-reports several years after the fact. Another approach to defining Gulf War illness was reported by Haley, Kurt, and Horn (1997b), based on a study of 249 U.S. naval reservists from a single battalion that had been called to active service in the Gulf. Members were

OCR for page 24
from five southeastern states. Study participants completed surveys in special face-to-face sessions conducted in two cities in December 1994 and January 1995. The investigators used factor analysis to identify underlying relationships among the individual symptoms and identified six possible clusters. The investigators did not find an association between psychological factors and having one or more of the six clusters; they concluded that the six symptom clusters could represent six syndromes or six variants on a single syndrome. It is important to note that the participants in this study were members of a single unit of Gulf War veterans that had already been extensively studied. Additionally, the response rate was low and there was no control group. A second report from this group (Haley and Kurt, 1997) on the same 249 veterans noted associations between specific types of chemical exposures and specific symptom clusters. However, the results are suggestive rather than conclusive given the limitations which include relatively small sample size, conceptual overlap among some of the lists of symptoms in the six symptom clusters, and reliance on retrospective self-report data on possible chemical exposures. A third study by Haley and colleagues (1997a) reported on a series of neurological tests in a small group (23) of veterans who met their case definition for a Gulf War-related symptom cluster compared to 10 controls who were deployed to the Gulf but did not meet case criteria and 10 veteran controls who were not deployed. A series of laboratory investigations on neurologic function was performed, and several specific dysfunctions were noted with more frequency in the case group than in either of the control groups. The authors concluded that the three factors identified in their prior studies represented three variants of a more general neurologic injury related to service in the Gulf (Haley et al., 1997a). No specific diagnoses could be made, and no specific wartime exposures were identified as causal factors for the neurologic dysfunctions. Some demographic factors (e.g., age) were associated with the symptom cluster experience. Unfortunately, there is no clear definition for a "case," rather the authors studied a group of subjects scoring high on these symptom factors for a broad range of neurologic tests, compared to groups of asymptomatic controls. Other limitations to this study include concern about the use of pooled scores for comparisons and multiple comparisons. Fukuda et al. (1998) conducted a factor analysis study of symptoms on a much larger sample of Air Force and Air National Guard veterans (3,723) from four units in Pennsylvania. A detailed set of survey, clinical evaluation, and laboratory test data was assembled. Approximately one-third of the sample had been deployed to the Gulf, so the investigators had the opportunity to compare symptom experience, and clinical, and laboratory findings for the deployed and nondeployed groups. The factor analysis originally identified 10 possible factors, but confirmatory factor analysis produced two factors accounting for most of the variance in symptom experience: mood-cognition-fatigue and musculoskeletal. A total of 10 symptoms were included in the two factors. A third factor was identified based on clinical experience with chronic fatigue. An individual was labeled a case if

OCR for page 24
he or she had one chronic symptom from two of these three categories. Although deployed veterans were significantly more likely than nondeployed veterans to meet case criteria, perhaps the most notable finding from this study was that approximately 15% of the nondeployed also met the case definition. Clinical and laboratory findings were generally not different between the deployed and nondeployed groups, and no specific characteristics of service in the Gulf were predictive of the multisymptom illness. Perhaps the strongest study on Gulf War veterans' experience of symptoms related to deployment in the Gulf is the "Iowa Study," a population-based study of all military personnel who listed Iowa as the home of record at enlistment (Iowa Persian Gulf Study Group, 1997). A total of 29,000 individuals were potential study participants; a stratified random sample was created to study approximately 750 individuals in each of four groups: Gulf War regular military; Gulf War National Guard or reserve; non-Gulf War regular military; and non-Gulf War National Guard or reserve. Gulf War status was based on deployment to the Gulf War theater. Other stratification variables (e.g., age, sex) were built into the sampling design; 4,886 individuals were selected for possible inclusion in the study and 3,695 were actually contacted by telephone and interviewed. Topics in the interview included symptom experience, wartime exposures, occurrence of injuries or specific illnesses (e.g., cancer), and PTSD. The investigators developed a priori definitions for a series of health outcomes of interest, such as cognitive dysfunction, fibromyalgia, depression, bronchitis, asthma, and anxiety. They used structured instruments such as the PRIME-MD and the SF-36 to arrive at these outcomes. Not only did those deployed to the Gulf report symptoms significantly more often, but most of the medical and psychological a priori outcomes were also elevated in the deployed group. The greatest differences between deployed and nondeployed respondents were seen for symptoms of cognitive dysfunction, fibromyalgia, and depression. The associations between exposures and the outcomes of symptoms of fibromyalgia, depression, and cognitive dysfunction were with "categories" of exposure thus, for example, associations were seen between one or more petrochemical exposures, and each of these three major outcomes. Similar findings about higher prevalence of clusters of cognitive and other symptoms for deployed veterans were reported by Wolfe et al. (1998) in a study of approximately 3,000 veterans from New England, by Pierce (1997) in a study of 525 women veterans, and by Unwin and colleagues (1999) in a study of 8,000 veterans from the United Kingdom. The Unwin study, a mail-out survey assessing symptoms, is noteworthy for its ability to compare veterans who served in the Gulf War to veterans from the same era who served in Bosnia and those who served in the Gulf War era but were not deployed to either conflict. Service in the Gulf War was associated with significantly higher rates of symptom experience than service in Bosnia or nondeployment. The authors did report an association between vaccinations and the multisymptom cluster identified by Fukuda et al. (1998).

OCR for page 24
A separate study by Ismail and colleagues (1999) used factor analysis of symptom data from United Kingdom servicemen deployed to the Gulf. This report identified 10 possible factors, which were then reduced to three by confirmatory factor analysis. The three factors were labeled mood-cognition, respiratory system, and peripheral nervous system. The authors noted some similarities between their factor structure and that reported by Haley (1997a); however, they also noted significant differences, attributable perhaps to differences in sample definitions and wartime experiences, or to somewhat different original lists of symptoms being used for the analyses. The absence of a musculoskeletal factor in the Unwin group's analyses is perhaps the most noteworthy difference. The survey of Canadian Gulf War veterans (Goss Gilroy, 1998) showed significantly higher rates of self-reported chronic conditions and symptoms of a variety of conditions (fibromyalgia, cognitive dysfunction, PTSD, depression, bronchitis, asthma, multiple chemical sensitivity, and anxiety) among deployed veterans compared to controls. Odds ratios for various conditions ranged from 1.35 (respiratory disease) to 5.27 (chronic fatigue). Rates of self-reported chronic conditions were also generally higher in both the deployed and control veterans than in an age-matched general population sample of the Ontario Health Survey. The Canadian Study and the Iowa Study produced very similar results. Some of the differences in prevalence between the two studies may be attributable to the use of a priori outcomes in the Iowa Study and differences in deployment of the Canadian veterans who were more often in supporting, nonfrontline areas. Most of the results discussed in this section come from self-report data obtained through surveys or telephone interviews. Some questions can be raised about the accuracy or validity of reports, particularly reports about specific exposures during the Gulf War that occurred several years before the survey and were not well documented or perhaps even noted at the time. Nevertheless, the consistency of findings of higher prevalence of cognitive, musculoskeletal, and energy/fatigue symptoms among veterans deployed to the Gulf is striking. Given the subjective nature of most of the symptoms some psychogenic origin for the complaints may exist. However, the consistency of reports across services, parts of the country, and across countries suggests that these are very real experiences, even if not fully understood. Gulf War veterans who participated in the CCEP program completed detailed symptom surveys. A report by Kroenke (1998) described the experience of more than 18,000 veterans who participated in the CCEP program through April 1996. The same general set of symptoms identified in the Iowa studies and most other studies appears here: fatigue, cognitive problems, muscle and joint pains, rash, and so on. Because the study reports on self-selected program participants, drawing any inferences about relative frequency of symptom experience as a result of deployment to the Gulf is not possible. The similarity of distribution of symptoms in this large group to that reported in more population-based studies is striking. No specific exposures during the war were associated

OCR for page 24
with individual symptoms, but individuals who reported exposures to more things also reported more symptoms. Three other studies related some aspect of symptom experience to either specific aspects of Gulf War experience or demographic factors. Morgan et al. (1998) found that PTSD symptoms among deployed veterans were more severe during months associated with the anniversary of particularly stressful events such as seeing a fellow soldier killed or being in a missile attack. Sutker (1994) noted a particularly high prevalence of PTSD and other psychological symptoms among veterans who had been assigned to graves registration duties during the Gulf War. Sutker (1995) also reported higher levels of psychological distress (assessed through a combination of psychological tests) among women and minority veterans who had served in the Gulf. On the basis of all these studies it appears that veterans who served in the Gulf are more likely than their nondeployed comrades or civilians to experience a set of symptoms that include cognitive, musculoskeletal, and energy/fatigue elements. However, no study has yet included a representative sample of the entire population of Gulf War veterans with appropriate comparison groups. In some cases, the symptoms are severe enough to be totally debilitating. Not all veterans experience the same cluster of symptoms; therefore, assuming a single underlying pathology or single cause for the complaints would not be appropriate. Despite intensive study in a number of large cohorts, it is impossible to say which exposure(s) in the Gulf War are associated with the symptoms being reported and what the underlying mechanism(s) may be. Functional Status and Well-Being The study of Canadian Gulf War veterans (Goss Gilroy, 1998) indicated a slightly higher proportion of bed days due to health in the 2 weeks prior to survey administration in deployed veterans compared to controls. Deployed veterans also reported more days in the past 2 weeks with "activity cut down" due to health (27% versus 15% in controls). Deployed veterans were more likely to respond in the "less favorable" half of a scale of general health than were controls. The Iowa Study (1997) included the RAND Short Form-36 (SF-36) questionnaire in its data collection battery. Deployed veterans reported statistically significant lower scores than did the nondeployed on all eight subscales of the SF-36. Absolute differences on a 0–100 scale were smaller for physical functioning (2 points) and social functioning (3 points), but were larger for general health (7 points) and vitality (8 points). Because differences of even 1 point or a fraction of a point have been considered meaningful in other large population studies (Allen, 1997), a 7- or 8-point difference is substantively, as well as statistically, significant. The Fukuda study of Air Force and Air National Guard veterans used SF-36 questionnaires for the subset of participants who volunteered for a clinical

OCR for page 24
evaluation (Fukuda et al., 1998). The clinical evaluation was offered only to members of the index unit (the National Guard unit that was deployed to the Gulf), so one cannot draw any conclusion about any effect of deployment on SF-36 scores. It was noted, though, that SF-36 scores were significantly lower among those veterans who met criteria as a severe case of multisymptom illness (according to symptom criteria) than among those labeled as mild-to-moderate cases, who in turn had lower scores than noncases. Health Care Utilization The largest and most significant study of health care utilization, specifically hospitalization, among Gulf War veterans was conducted by Gray and colleagues (1996) at the Naval Health Research Center in San Diego. They studied the predeployment and postdeployment hospitalization experience of essentially all deployed military personnel who remained on active duty through 1993 to that of nondeployed military personnel for the same time period. Data for the study came from the Defense Manpower Data Center (used to identify eligible individuals) and discharge summary data for all DoD hospitals. The study sample was very large; approximately 1 million individuals were included in follow-up analyses for 1991, 1992, and 1993. The study tested the hypothesis that deployed veterans would be admitted to DoD hospitals more frequently than their nondeployed counterparts, either for any cause or for specific conditions suspected of being caused by exposures during the Gulf War. Examples of the specific conditions included infectious and parasitic diseases, cancers, nervous system diseases, and musculoskeletal diseases. The basic finding of the study was that risk of hospitalization for deployed veterans in the immediate postwar period was no different from that of nondeployed veterans, either for any cause or for virtually all the specific disease conditions. The one or two significant findings were for 1 year only and did not appear in previous or subsequent years (e.g., more admissions for disease of the blood in 1992 but not in 1991 or 1993); moreover, these findings were not always in the predicted direction (deployed veterans were significantly less likely to be admitted for ill-defined conditions in 1991 and 1992). A healthy soldier effect was noted: admission rates for deployed veterans were significantly lower than those for nondeployed veterans in the 2 years prior to the Gulf War, and the largest difference in the period occurred immediately before the war. The effect was not found before 1990, however, suggesting that those deployed to the Gulf were perhaps in better health at that time than their nondeployed counterparts, but were not necessarily healthier due to stable causes that would predict hospitalization rates for years into the future. The study was limited because of its focus on active duty personnel; conceivably, those suffering from Gulf War-related symptoms might leave active duty voluntarily or take a medical discharge. Hospitalizations for that group would appear in VA or private sector databases but not in the DoD database.

OCR for page 24
Despite this limitation, the study would have been able to detect any marked increase in hospitalizations among those who did stay on active duty (and perhaps would have been admitted at least once before going into any sort of medical discharge status); no such effect was seen. A study from the same data set on hospitalizations that focused on unexplained symptoms or conditions showed a slight excess of such admissions for deployed veterans compared to nondeployed veterans (Knoke and Grey, 1998). The exclusion of veterans who had been admitted through the CCEP for evaluation not only eliminated the difference, this analysis suggested a lower rate of hospitalization for unexplained symptoms in the deployed group. Summary of Findings from Studies to Date The committee found that the large and growing literature on the health of Gulf War veterans supports both the conclusions that follow and the list of questions to be addressed by the prospective cohort study described in Chapter 5. Military personnel who served in the Gulf War have had a significantly higher risk (at least through 1996) of suffering one or more of a set of symptoms that include fatigue, memory loss, difficulty concentrating, pains in muscles and joints, and rashes. Other symptoms are noted with reduced frequency, but still may be experienced more often by deployed than nondeployed veterans. The symptoms range in severity from barely detectable to completely debilitating. No single accepted diagnosis or group of diagnoses has been identified that describes and explains this cluster of symptoms. There is no single exposure, or set of exposures, that has been shown conclusively to cause individual symptoms or clusters of symptoms. Although some statistical associations have been seen in some studies, they have not been confirmed in other studies or confirmed through laboratory tests that would establish a cause—effect connection in individual patients. No diseases included in the ICD-9-CM or ICD-10 classification systems have been shown to be more frequent in deployed or in nondeployed veterans with the exception of PTSD symptoms. Mortality among deployed veterans is not higher in general than mortality among nondeployed veterans, at least through 1993. Deaths due to accidents are higher among deployed veterans. Health-related quality of life, as measured through instruments such as the SF-36, is lower on average among deployed veterans than among nondeployed veterans. The natural course of symptom experience over time is not known, as no longitudinal studies of symptom experience have been conducted and reported in the literature.

OCR for page 24
Limitations of Previous Studies Although the studies to date have provided valuable information regarding the health problems experienced by some Gulf War veterans, these studies have important limitations in terms of assessing the health status of those veterans. One concern is that available data and studies are not representative of the entire population of U.S. Gulf War veterans. For example, the Gulf War veteran registries developed by DoD and VA include only those veterans who have chosen to participate. The health status of participating veterans may differ from that of other Gulf War veterans. The studies of hospitalizations (Gray et al., 1996; Knoke and Gray, 1998) and adverse birth outcomes (Araneta et al., 1997; Cowan et al., 1997) have been limited to personnel remaining on active duty and to events occurring in military hospitals. Health status or characteristics of active duty personnel could differ from those of personnel who have left active duty or who have been treated in nonmilitary hospitals. Moreover, economic and other non-health-related factors are likely to have a greater effect on the use of nonmilitary hospitals and health care services. As noted in the review above, other studies have focused on individual service units or other subpopulations of Gulf War veterans (e.g., Fukuda et al., 1998; Haley et al., 1997 a,b,c; Pierce, 1997; Stretch et al., 1995). The deployment experience of specific units (e.g., location of deployment, tasks during deployment) or other features of a unit (e.g., service branch, reserve or National Guard versus active duty) may be distinctive and therefore not generalizable to a larger Gulf War veteran population. Even the Iowa Study (Iowa Persian Gulf Study Group, 1997), which drew a representative sample from the population of all veterans with Iowa as home of record at time of enlistment without regard to service unit, may not be generalizable to all Gulf War veterans because of factors specific to Iowa such as a small minority population. Another factor that may affect the representativeness of the studies conducted so far is bias introduced by low participation rates or by participation rates that differ among subgroups (e.g., Fukuda et al., 1998; Haley et al., 1997a,b). Several factors may influence participation rates. For example, veterans still on active duty or attached to reserve or National Guard units may be easier to locate than those who have been discharged. Gulf War veterans on active duty may be reluctant to identify conditions that could lead to a change in their duty status, whereas those veterans who perceive that they have service-related health problems may have a greater incentive to participate in studies than those who do not have health problems. The studies to date are also limited in their assessments of the health status of Gulf War veterans. Some studies have used very specific indicators such as mortality (Kang and Bullman, 1996) or hospitalizations (Gray et al., 1996; Knoke and Gray, 1998). Many conditions, however, may impair health, functional status, and other aspects of well-being without causing death or requiring hospitalization. Furthermore, because the published mortality analysis covers only about 30 months following the return of many deployed troops, those data

OCR for page 24
reflect only the most severe health effects that Gulf War veterans might experience. A longer period of observation will be needed to detect changes in mortality related to health problems with a long latency (e.g., cancer) or with a more chronic course (e.g., multiple sclerosis). Other studies have focused on indicators such as the prevalence of reported symptoms. Although symptoms can signal health problems, determining their presence does not by itself provide enough information to assess whether those symptoms affect function or other aspects of health. Remainder of this Report The work of this committee is directed toward designing a research portfolio that will address the knowledge gaps and methodological problems identified thus far. Our overall aims are to develop a conceptual and operational framework that will produce a population-based assessment of the nature and extent of health problems among Gulf War veterans, assess the impact of these health problems on veterans' health status, and monitor changes in the health status of these veterans over time. Such a framework will generate information necessary for policymaking, clinical decision making, and shared decision making between health care providers and patients. The remainder of the report addresses the committee's proposed study framework for making such assessments.