National Academies Press: OpenBook
« Previous: Environment and Exposures
Suggested Citation:"Health Outcomes." Institute of Medicine. 1996. Health Consequences of Service During the Persian Gulf War: Recommendations for Research and Information Systems. Washington, DC: The National Academies Press. doi: 10.17226/5272.
×
Page 67
Suggested Citation:"Health Outcomes." Institute of Medicine. 1996. Health Consequences of Service During the Persian Gulf War: Recommendations for Research and Information Systems. Washington, DC: The National Academies Press. doi: 10.17226/5272.
×
Page 68
Suggested Citation:"Health Outcomes." Institute of Medicine. 1996. Health Consequences of Service During the Persian Gulf War: Recommendations for Research and Information Systems. Washington, DC: The National Academies Press. doi: 10.17226/5272.
×
Page 69
Suggested Citation:"Health Outcomes." Institute of Medicine. 1996. Health Consequences of Service During the Persian Gulf War: Recommendations for Research and Information Systems. Washington, DC: The National Academies Press. doi: 10.17226/5272.
×
Page 70
Suggested Citation:"Health Outcomes." Institute of Medicine. 1996. Health Consequences of Service During the Persian Gulf War: Recommendations for Research and Information Systems. Washington, DC: The National Academies Press. doi: 10.17226/5272.
×
Page 71
Suggested Citation:"Health Outcomes." Institute of Medicine. 1996. Health Consequences of Service During the Persian Gulf War: Recommendations for Research and Information Systems. Washington, DC: The National Academies Press. doi: 10.17226/5272.
×
Page 72
Suggested Citation:"Health Outcomes." Institute of Medicine. 1996. Health Consequences of Service During the Persian Gulf War: Recommendations for Research and Information Systems. Washington, DC: The National Academies Press. doi: 10.17226/5272.
×
Page 73
Suggested Citation:"Health Outcomes." Institute of Medicine. 1996. Health Consequences of Service During the Persian Gulf War: Recommendations for Research and Information Systems. Washington, DC: The National Academies Press. doi: 10.17226/5272.
×
Page 74
Suggested Citation:"Health Outcomes." Institute of Medicine. 1996. Health Consequences of Service During the Persian Gulf War: Recommendations for Research and Information Systems. Washington, DC: The National Academies Press. doi: 10.17226/5272.
×
Page 75
Suggested Citation:"Health Outcomes." Institute of Medicine. 1996. Health Consequences of Service During the Persian Gulf War: Recommendations for Research and Information Systems. Washington, DC: The National Academies Press. doi: 10.17226/5272.
×
Page 76
Suggested Citation:"Health Outcomes." Institute of Medicine. 1996. Health Consequences of Service During the Persian Gulf War: Recommendations for Research and Information Systems. Washington, DC: The National Academies Press. doi: 10.17226/5272.
×
Page 77
Suggested Citation:"Health Outcomes." Institute of Medicine. 1996. Health Consequences of Service During the Persian Gulf War: Recommendations for Research and Information Systems. Washington, DC: The National Academies Press. doi: 10.17226/5272.
×
Page 78
Suggested Citation:"Health Outcomes." Institute of Medicine. 1996. Health Consequences of Service During the Persian Gulf War: Recommendations for Research and Information Systems. Washington, DC: The National Academies Press. doi: 10.17226/5272.
×
Page 79
Suggested Citation:"Health Outcomes." Institute of Medicine. 1996. Health Consequences of Service During the Persian Gulf War: Recommendations for Research and Information Systems. Washington, DC: The National Academies Press. doi: 10.17226/5272.
×
Page 80
Suggested Citation:"Health Outcomes." Institute of Medicine. 1996. Health Consequences of Service During the Persian Gulf War: Recommendations for Research and Information Systems. Washington, DC: The National Academies Press. doi: 10.17226/5272.
×
Page 81
Suggested Citation:"Health Outcomes." Institute of Medicine. 1996. Health Consequences of Service During the Persian Gulf War: Recommendations for Research and Information Systems. Washington, DC: The National Academies Press. doi: 10.17226/5272.
×
Page 82
Suggested Citation:"Health Outcomes." Institute of Medicine. 1996. Health Consequences of Service During the Persian Gulf War: Recommendations for Research and Information Systems. Washington, DC: The National Academies Press. doi: 10.17226/5272.
×
Page 83
Suggested Citation:"Health Outcomes." Institute of Medicine. 1996. Health Consequences of Service During the Persian Gulf War: Recommendations for Research and Information Systems. Washington, DC: The National Academies Press. doi: 10.17226/5272.
×
Page 84
Suggested Citation:"Health Outcomes." Institute of Medicine. 1996. Health Consequences of Service During the Persian Gulf War: Recommendations for Research and Information Systems. Washington, DC: The National Academies Press. doi: 10.17226/5272.
×
Page 85
Suggested Citation:"Health Outcomes." Institute of Medicine. 1996. Health Consequences of Service During the Persian Gulf War: Recommendations for Research and Information Systems. Washington, DC: The National Academies Press. doi: 10.17226/5272.
×
Page 86
Suggested Citation:"Health Outcomes." Institute of Medicine. 1996. Health Consequences of Service During the Persian Gulf War: Recommendations for Research and Information Systems. Washington, DC: The National Academies Press. doi: 10.17226/5272.
×
Page 87
Suggested Citation:"Health Outcomes." Institute of Medicine. 1996. Health Consequences of Service During the Persian Gulf War: Recommendations for Research and Information Systems. Washington, DC: The National Academies Press. doi: 10.17226/5272.
×
Page 88
Suggested Citation:"Health Outcomes." Institute of Medicine. 1996. Health Consequences of Service During the Persian Gulf War: Recommendations for Research and Information Systems. Washington, DC: The National Academies Press. doi: 10.17226/5272.
×
Page 89
Suggested Citation:"Health Outcomes." Institute of Medicine. 1996. Health Consequences of Service During the Persian Gulf War: Recommendations for Research and Information Systems. Washington, DC: The National Academies Press. doi: 10.17226/5272.
×
Page 90
Suggested Citation:"Health Outcomes." Institute of Medicine. 1996. Health Consequences of Service During the Persian Gulf War: Recommendations for Research and Information Systems. Washington, DC: The National Academies Press. doi: 10.17226/5272.
×
Page 91
Suggested Citation:"Health Outcomes." Institute of Medicine. 1996. Health Consequences of Service During the Persian Gulf War: Recommendations for Research and Information Systems. Washington, DC: The National Academies Press. doi: 10.17226/5272.
×
Page 92
Suggested Citation:"Health Outcomes." Institute of Medicine. 1996. Health Consequences of Service During the Persian Gulf War: Recommendations for Research and Information Systems. Washington, DC: The National Academies Press. doi: 10.17226/5272.
×
Page 93
Suggested Citation:"Health Outcomes." Institute of Medicine. 1996. Health Consequences of Service During the Persian Gulf War: Recommendations for Research and Information Systems. Washington, DC: The National Academies Press. doi: 10.17226/5272.
×
Page 94
Suggested Citation:"Health Outcomes." Institute of Medicine. 1996. Health Consequences of Service During the Persian Gulf War: Recommendations for Research and Information Systems. Washington, DC: The National Academies Press. doi: 10.17226/5272.
×
Page 95
Suggested Citation:"Health Outcomes." Institute of Medicine. 1996. Health Consequences of Service During the Persian Gulf War: Recommendations for Research and Information Systems. Washington, DC: The National Academies Press. doi: 10.17226/5272.
×
Page 96
Suggested Citation:"Health Outcomes." Institute of Medicine. 1996. Health Consequences of Service During the Persian Gulf War: Recommendations for Research and Information Systems. Washington, DC: The National Academies Press. doi: 10.17226/5272.
×
Page 97
Suggested Citation:"Health Outcomes." Institute of Medicine. 1996. Health Consequences of Service During the Persian Gulf War: Recommendations for Research and Information Systems. Washington, DC: The National Academies Press. doi: 10.17226/5272.
×
Page 98
Suggested Citation:"Health Outcomes." Institute of Medicine. 1996. Health Consequences of Service During the Persian Gulf War: Recommendations for Research and Information Systems. Washington, DC: The National Academies Press. doi: 10.17226/5272.
×
Page 99
Suggested Citation:"Health Outcomes." Institute of Medicine. 1996. Health Consequences of Service During the Persian Gulf War: Recommendations for Research and Information Systems. Washington, DC: The National Academies Press. doi: 10.17226/5272.
×
Page 100
Suggested Citation:"Health Outcomes." Institute of Medicine. 1996. Health Consequences of Service During the Persian Gulf War: Recommendations for Research and Information Systems. Washington, DC: The National Academies Press. doi: 10.17226/5272.
×
Page 101
Suggested Citation:"Health Outcomes." Institute of Medicine. 1996. Health Consequences of Service During the Persian Gulf War: Recommendations for Research and Information Systems. Washington, DC: The National Academies Press. doi: 10.17226/5272.
×
Page 102
Suggested Citation:"Health Outcomes." Institute of Medicine. 1996. Health Consequences of Service During the Persian Gulf War: Recommendations for Research and Information Systems. Washington, DC: The National Academies Press. doi: 10.17226/5272.
×
Page 103
Suggested Citation:"Health Outcomes." Institute of Medicine. 1996. Health Consequences of Service During the Persian Gulf War: Recommendations for Research and Information Systems. Washington, DC: The National Academies Press. doi: 10.17226/5272.
×
Page 104
Suggested Citation:"Health Outcomes." Institute of Medicine. 1996. Health Consequences of Service During the Persian Gulf War: Recommendations for Research and Information Systems. Washington, DC: The National Academies Press. doi: 10.17226/5272.
×
Page 105
Suggested Citation:"Health Outcomes." Institute of Medicine. 1996. Health Consequences of Service During the Persian Gulf War: Recommendations for Research and Information Systems. Washington, DC: The National Academies Press. doi: 10.17226/5272.
×
Page 106
Suggested Citation:"Health Outcomes." Institute of Medicine. 1996. Health Consequences of Service During the Persian Gulf War: Recommendations for Research and Information Systems. Washington, DC: The National Academies Press. doi: 10.17226/5272.
×
Page 107
Suggested Citation:"Health Outcomes." Institute of Medicine. 1996. Health Consequences of Service During the Persian Gulf War: Recommendations for Research and Information Systems. Washington, DC: The National Academies Press. doi: 10.17226/5272.
×
Page 108
Suggested Citation:"Health Outcomes." Institute of Medicine. 1996. Health Consequences of Service During the Persian Gulf War: Recommendations for Research and Information Systems. Washington, DC: The National Academies Press. doi: 10.17226/5272.
×
Page 109
Suggested Citation:"Health Outcomes." Institute of Medicine. 1996. Health Consequences of Service During the Persian Gulf War: Recommendations for Research and Information Systems. Washington, DC: The National Academies Press. doi: 10.17226/5272.
×
Page 110
Suggested Citation:"Health Outcomes." Institute of Medicine. 1996. Health Consequences of Service During the Persian Gulf War: Recommendations for Research and Information Systems. Washington, DC: The National Academies Press. doi: 10.17226/5272.
×
Page 111
Suggested Citation:"Health Outcomes." Institute of Medicine. 1996. Health Consequences of Service During the Persian Gulf War: Recommendations for Research and Information Systems. Washington, DC: The National Academies Press. doi: 10.17226/5272.
×
Page 112
Suggested Citation:"Health Outcomes." Institute of Medicine. 1996. Health Consequences of Service During the Persian Gulf War: Recommendations for Research and Information Systems. Washington, DC: The National Academies Press. doi: 10.17226/5272.
×
Page 113
Suggested Citation:"Health Outcomes." Institute of Medicine. 1996. Health Consequences of Service During the Persian Gulf War: Recommendations for Research and Information Systems. Washington, DC: The National Academies Press. doi: 10.17226/5272.
×
Page 114
Suggested Citation:"Health Outcomes." Institute of Medicine. 1996. Health Consequences of Service During the Persian Gulf War: Recommendations for Research and Information Systems. Washington, DC: The National Academies Press. doi: 10.17226/5272.
×
Page 115
Suggested Citation:"Health Outcomes." Institute of Medicine. 1996. Health Consequences of Service During the Persian Gulf War: Recommendations for Research and Information Systems. Washington, DC: The National Academies Press. doi: 10.17226/5272.
×
Page 116

Below is the uncorrected machine-read text of this chapter, intended to provide our own search engines and external engines with highly rich, chapter-representative searchable text of each book. Because it is UNCORRECTED material, please consider the following text as a useful but insufficient proxy for the authoritative book pages.

4 Health Outcomes OVERVIEW This chapter reviews the empirical evidence available to the committee through May 1996 regarding the health experience of Persian Gulf War (POW) veterans and potential risk factors for adverse health experiences. With the broad question of what adverse health consequences veterans have suffered as a result of their service, the range of relevant experiences is also very broad. The strength of evidence for or against increased risks of specific health outcomes among those who served in the Persian Gulf (PG) depends in part on what research studies have been conducted, and hence on numerous explicit and implicit decisions made by large numbers of research investigators and funding agencies? often acting individually with little perspective on overall needs and priorities. As a result, the research record is of uneven depth and quality. Our task is to summarize the data available to date that appear relevant to our charge of examining possible health consequences of POW service and to recommend the nature of future studies that would provide more and better-answers to this question. Although medical scientists often can use clinical data and individual reports of health experiences to identify areas of concern, such data and reports cannot in themselves provide proof of cause and effect about the health outcomes of POW service. No matter how well documented an illness may be, or how moving a personal story, unexplained illnesses also occur in the civilian population and in troops not deployed to the Gulf. A basic question regarding 67

68 HEALTH CONSEQUENCES OF THE PERSIAN GULF WAR the connection between illness in veterans and their service is not whether specific illnesses or adverse health experiences occurred, but whether the frequency or severity of such outcomes was increased over what occurs in otherwise similar populations that were not in the PG. Battle injuries are universally recognized as a hazard of war. Diseases and infections historically have produced casualties in past wars. Veterans from every war have suffered stress-related symptoms variously known as shell shock; battle fatigue; combat exhaustion; traumatic neurosis; and, since 1980, posttraumatic stress disorder (PTSD). In addition, service members on active duty are subject to the same health hazards as the civilian population, although not necessarily to the same degree: accidents, cancer, heart attacks, stroke, and the like. With the exception of imbedded depleted uranium (DU) fragments and possibly leishmaniasis, very few illnesses or injuries were clearly connected to EG service. However, this committee was convened, in part, in response to congressional concern that there might be something unique to the PG region or to the war fought there that resulted in specific illnesses, some even becoming manifest long after veterans had left that region. Over time, the military services, the Department of Veterans Affairs (DVA), and other federal agencies have expanded the scope and scientific rigor of investigations into the health of PGW veterans in an effort to systematically develop a body of knowledge about their health experiences and risks associated with exposures described in the previous chapter. The committee and other investigators are seeking to determine whether the risk of illnesses was increased among PGW veterans and what research should be carried out to make such determinations. In this overview, we make some general observations about the strengths and limitations of the evidence as a whole. The range of possible PGW-related health effects that can be studied at this time is intrinsically limited. Illnesses and symptoms that occurred during the deployment and were transient in nature were not studied or monitored systematically then and are very difficult or impossible to study retrospectively now. For example, possible temporary decrements in lung function associated with exposure to pollutants from the oil well fires were not evaluated at the critical time and are not very amenable to study now, although they may be important. Likewise, health effects that first come to attention years after the precipitating exposure cannot easily be studied. Many of the known causes of chronic diseases, such as cancer and coronary artery disease, operate over longer periods than have passed since the PGW and, therefore, cannot yet be evaluated in Gulf War veterans. For example, it is commonly believed that most cancers have a minimum 10-year latent period between exposure and detection of the first extra cases of disease. Thus, although no excess adverse cancer effects have yet been reported, delayed effects that have not yet come to light are still possible. What can be examined now are effects that appear early and are

HEALTH OUTCOMES 69 persistent or become manifest at some time up to several years after the relevant exposure. Concerns about unusual illnesses among PGW veterans arose initially through reports of individuals and then through "outbreak" studies, in which teams of epidemiologists studied groups of soldiers who reported a high prevalence of a cluster of symptoms later proposed to be characteristic of a "Gulf War Syndrome." To provide some support to those veterans concerned about their health, to enable them to receive a clinical work-up, and to gather information on a possible connection to service in the PG, the DVA and the Depa~l~ent of Defense (DoD) created registries and voluntary referral programs for troops, including DVA's National Referral Center and PG Health Registry (PGHR) and DoD's Comprehensive Clinical Evaluation Program (CCEP). Veterans who have voluntarily participated in these registries have not been found to have any unusual rates of diagnosable conditions (DoD, 1996; Kang, 1996; Kang et al., 1995) but do report a pattern of symptom complaints similar to that seen in the outbreak studies. The registries also share the scientific limitations of the outbreak studies, in that participants are self-selected, symptoms are self-reported, exposures are self-reported and could not be validated, and there is no suitable control group. Because of these limitations, the committee has concluded that the information on veterans' health that exists in the registries cannot serve alone as a basis for scientific study of the health effects of the PGW. The committee does consider these registries and their affiliated clinical referral programs as useful in assisting veterans who need clinical services and possibly useful as a source of hypotheses regarding the nature and extent of health problems experienced by PGW veterans. The DVA and various units of the DoD also have undertaken a variety of scientific studies of the health status of PGW veterans. The number and scope of these studies have increased rapidly over the past several years, but few studies had been completed as of May 1996. Findings of several of the completed studies are summarized in this overview. A more complete discussion of these studies and of studies planned or under way is provided in subsequent sections of this chapter, and a listing of research completed, under way, and planned is included in Appendix F. Most of these studies are limited by the absence of detailed exposure information related to individual troops or units. Consequently, studies have had to be designed to seek effects that are sufficiently widespread to be evident when comparing troops who served in the Gulf with those who did not (PG-era veterans). If an environmental exposure or experience was sufficiently widespread among PGW troops and if the health effects of that exposure were sufficiently severe, effects on mortality rates during and after deployment might be demonstrated. To investigate this question, DoD and DVA individually

70 HEALTH CONSEQUENCES OF THE PERSIAN GULF WAR undertook a study of mortality rates of deployed troops and appropriate comparison groups. These studies of mortality rates during deployment documented the rate of battle-related mortality, and both studies found increased rates of death due to accidents and unintentional injury. Despite isolated findings of slightly increased or decreased cause-specific mortality among the cohort of troops deployed to the PG, the overall finding from these studies was that no disease-specific mortality rate was increased among PGW troops. In summary, these mortality studies provide no evidence for an increase in total mortality in the cohort of 697,000 PG veterans that is consistent with a common health complaint or exposure, although mortality from accidents and unintentional injuries appears to be increased in the PG veteran group. The information provided is, however, of limited value in identifying the nature and extent of health effects of PGW service. These studies provide no information, of course, on health problems that did not result in increased mortality. Moreover, because the entire cohort of troops or veterans is studied without identification of exposure-based subgroups, these studies are intrinsically insensitive to increased mortality in a subset of troops subjected to a specific, so far unidentified, exposure. Finally, the period of follow-up has, thus far, been only 2 years. If they occulted, health effects of PGW service might not affect mortality rates for many years. To gather information about morbidity rates among PGW veterans, the Naval Health Research Center (NHRC) at San Diego has undertaken two studies of hospitalization rates, one in PGW veterans who remained on active duty and a second in residents of California who separated from military service after the PGW. Preliminary findings from the first study found no consistent evidence of increased risk of hospitalization for any cause among PGW veterans compared with nondeployed veterans, even after adjusting for a possible healthy worker effect, as suggested by a 10% lower hospitalization rate of PGW veterans prior to deployment. Strengths of this study included the large sample size, ability to adjust for differences in demographic characteristics of PGW and PGW-era veterans, and completeness of the hospitalization data. Yet this study is also of limited value in assessing the health consequences of the PGW. First, it was not feasible to include any but those who remain on active duty. Evaluation of this limitation may be possible with completion of the study of separated troops residing in California. Furthermore, this second study will permit examination of effects that were not evident until after the PGW. As with the mortality studies, an analysis of hospitalization rates in the entire cohort would be unlikely to detect health effects of exposures that affected a small geographic or occupational subset of the cohort. Nevertheless, the mortality and hospitalization studies, when taken together, provide no evidence of an association between any health experience and environmental exposure of sufficient severity to increase the risk

HEALTH OUTCOMES 71 of early death or hospitalization for the entire cohort of deployed troops in the 2- year period after deployment. Recognizing that mortality and hospitalization rates may not be affected by an increase in the symptoms described in the outbreak studies of military units, investigators have initiated several studies designed to assess the prevalence of self-reported symptoms among PGW veterans and appropriate comparison groups. At this time, only one population-based survey of veterans is nearing completion (Kaiser et al., 1995~: a survey of all regular Navy Seabees who were on active duty during the PGW deployment and were still on active duty between September 1994 and June 1995. This study found no differences in measures of physical health or symptom rates between troops deployed to the Gulf (N= 527) and those on duty elsewhere (N= 970~. However, Seabee veterans who had been deployed to the Gulf did have higher mean levels of psychological symptom scores. Again, a limitation of this study was the absence of any specific information about exposures experienced by individual troops. The NHRC is now conducting a broader study of active, reserve, and separated Navy Seabees who were on active duty during the era of the PGW. Psychological sequelae are important and somewhat predictable consequences of service in war. We recognize that they can disable otherwise healthy individuals, but psychiatric diagnoses should not be adopted in default of other medical diagnoses. Furthermore, we believe these are war-related illnesses that deserve attention by the military and DVA in terms of both prevention and treatment. Several studies of stress and responses to stress are described below. Many veterans have expressed concern about the possibility of adverse effects on their conception of children subsequent to the PGW. Four pilot studies of reproductive health experiences of PGW veterans have been completed, and three large population studies are now under way. None of the four pilot studies have detected increased rates of adverse reproductive outcomes in PGW veterans. Given the preceding overview, the committee has not identified scientific evidence to date demonstrating adverse health consequences specifically of PGW service other than the documented incidents of leishmaniasis, combat- related or injury-related mortality or morbidity, and increased risk of psychiatric sequelae of deployment. At the same time, the committee recognizes that studies provided to us thus far do not comprise a comprehensive scientific investigation of the health consequences of service in the PGW. In addition to the recommendations for research given in Chapter 2, the committee believes that value to the military services and the country will result from completion of the NHRC reproductive studies, well-designed scientific studies of determinants of vulnerability (and resilience) to stress arising in deployment and combat, investigations of unexplained illnesses and of relatively obscure syndromes (e.g., chronic fatigue syndrome), and research into

72 HEALTH CONSEQUENCES OF THE PERSIAN GULF WAR improved methods to diagnose leishmaniasis. If proposals for such studies are identified through established procedures for peer review, the committee recommends that they be favorably considered for funding. We have organized the accumulated evidence by study type and endpoint into the following sections: mortality studies, hospitalization studies, diagnosed diseases in PG veterans, studies of self-reported symptoms (outbreaks and surveys), adverse reproductive outcome studies, mental health studies, and women's health studies. The research activities known to the committee as of May 1996 are listed in Appendix F. We turn now to a review of the epidemiologic evidence. MORTALITY STUDIES Records of deaths during the period of deployment are available, and the mortality experience of the POW veteran cohort, both during and after deployment, is reviewed in this chapter. Death records are uniquely informative and available, in that without taking any special measures, the information is likely to be very nearly complete and accurate regarding the person and the time of occurrence. Even during the POW itself, both battle casualties (Helmkamp, 1994) and nonbattle casualties (Writer et al., 1996) were comprehensively ascertained. There is a restriction on the types of events that can be suitably monitored through mortality data. Only fatal injuries can be evaluated, and they may or may not follow the same pattern as nonfatal injuries. Diseases with a high case- fatality rate such as lung cancer can also be evaluated through mortality data. However, for some diseases, death is an unusual consequence; examples are arthritis, asthma, and nonmelanoma skin cancer. Relative to mental health, severe depression may be reflected to some extent in suicide mortality, but mortality data may tell us very little about severe nonfatal psychiatric disorders. Although multiple causes of death are often recorded and coded, the interpretation of nonunderlying causes of death is still difficult and uncertain. Not all potentially relevant ancillary causes will be noted, and the extent of completeness may well vary in relation to social and demographic factors and in relation to the source and quality of health care, including military versus nonmilitary providers. The adequacy, accuracy, and detail of mortality data are suitable for some assessments of health risk but not for others. In general, broad categories of fatal disease can be examined with some confidence (e.g., coronary heart disease or suicide), whereas death certificate classifications are less reliable for diseases that are more difficult to diagnose or require more specific classification (e.g., the chronic neurologic diseases or subtypes of leukemia). Two mortality studies of PG veterans have been conducted (Kang and Bullman, 1995; Writer et al., 1996~. Both studies encompass general PG

HEALTH OUTCOMES 73 veteran populations. Each study includes very large samples. One assessed mortality for veterans up to 2 years after active duty; the other assessed mortality during active duty. Both studies compared troops deployed to the PG with troops deployed elsewhere during the same period. Neither includes information about specific environmental exposures. No excess mortality was observed among PG personnel, with the exception of combat-related deaths and deaths due to accidents and unintentional injury. We do not consider battle deaths (evaluated by Helmkamp, 1994), although we do evaluate wartime nonbattle deaths (described by Writer et al., 1996~. Writer et al. (1996) studied troops with active duty status during or shortly after the POW. All men and women who were on active duty in the PG theater of operations between August 1, 1990, and July 31, 1991, were identified, with demographic information, dates of service in the PG and elsewhere, and date and cause of death. The control group was composed of service personnel stationed elsewhere at the same time. Among the 1,622 total nonbattle deaths during this year, 1,397 occurred in nondeployed service personnel (73 per 100,000 person-years) and 225 occurred in those deployed (85 per 100,000 person-years). More than half (N= 967) of all deaths were due to unintentional injury, with more than half of these (N = 501) being motor vehicle injuries. Other major contributors were deaths from illness (N= 294), suicide (N= 216), and homicide (N= 103~. By using the experience of nondeployed veterans to generate expected mortality rates for those deployed, relative risk (RR) estimates were generated, with adjustment for age. The RR= 1.12 (95% confidence interval [CI] = 0.97- 1.26) for all nonbattle deaths. This means that there were 12% more deaths in PG veterans than in veterans who had not been in the Gulf theater, but that random variation in the number of deaths is readily compatible with any figure from a 3% decrease to a 26% increase. Since this range includes "no difference," evidence for an effect of Gulf service on total mortality is weak. Similarly, RR= l.lS (95% CI = 1.01-1.34) for all injuries; RR= 1.54 (95% CI = 1.32-1.77) for unintentional injury; FOR= 0.34 (95% CI = 0.16-0.63) for suicide; and RR = 1.15 (95% CI = 0.73-1 .73) for a combination of cardiovascular disease and unexpected or undefined causes of death. Subject to the inherent limitations in mortality data noted above and the restricted study period, these data suggest that an increase in unintentional injury was associated with deployment, but most other comparisons were consistent with "no difference." A second major report concerns mortality in the 2 years after the end of hostilities. POW veterans were compared with veterans from the same era who did not serve in the Gulf (Kang and Bullman, 19951. This historical cohort mortality study was conducted among all 695,516 men and women who served in the PG between August 1990 and April 1991 and a sample of 746,291 veterans from the same period who served elsewhere, matched by branch of

74 HEALTH CONSEQUENCES OF THE PERSIAN GULF WAR service and unit type (active, reserve, and guard). The interval of follow-up was May 1, 1991, through September 30, 1993, or date of death, whichever occurred first. The investigators identified 1,765 deaths among PGW veterans and 1,729 deaths in the control group. Age was included in the multivariate analysis to control for possible differences in the distribution of this variable between the two groups. The risk of death was slightly higher among PG veterans who remained on active duty than among other active duty veterans from the same era, with a mortality rate ratio (relative risk) of 1 .15. Several causes of death were reduced, but not significantly, among PG veterans, including infectious disease, cancer, and respiratory disease, whereas a consistent pattern of significantly elevated risk was found for unintentional injuries, including motor vehicle injuries, for which the relative risk estimates were around 1.5. Suicide and homicide rates were similar in the two groups. The overall mortality (RR = 1.47) for female veterans of the PGW was slightly higher than the overall death rate for other women veterans. Investigators also found a significant twofold increase in unintentional injury and motor vehicle injuries, but no significant differences in mortality due to specific illnesses. The relative risks of homicide and suicide were elevated (RR = 2.0 and 1.5, respectively), but these were not statistically significant. The same overall pattern was present but less pronounced among reserves. Kang and Bullman (1995) compared mortality in both PG veterans and PG- era veterans with that in the total U.S. population of the same age and sex and found that for men, total mortality was markedly reduced among both groups of veterans compared with the U.S. population, as were deaths from unintentional injuries and motor vehicle injuries. Both groups of women veterans showed reduced mortality rates compared to those of the general population, with somewhat elevated rates of suicide and motor vehicle injuries for PGW veterans. Both of these studies (Kang and Bullman, 1995; Writer et al., 1996), particularly that of Kang and Bullman (1995), point to injury, both intentional and unintentional, during and after the war as the leading cause of excess mortality among PGW veterans. This is not to say that a potential effect on mortality due to illness has been disproved, only that such an effect could not be demonstrated with 2 years of follow-up. This is an important finding, and further follow-up will be appropriate for evaluating possible differences between groups for medical conditions with longer latency periods. HOSPITALIZATION STUDIES Computerized hospital discharge data provide a resource for assessing whether PGW veterans were at increased risk of hospitalization after the war relative to appropriate comparison groups and for investigating the distribution

HEALTH OUTCOMES 75 of primary diagnoses as defined at hospital discharge. Because hospital discharge data are collected and reported by individual hospitals, which serve poorly defined populations, studies of hospital discharge rates often focus on geographically defined populations, such as persons who report that their permanent residence is in a specified state. Virtually all hospitalizations of active duty personnel take place in military hospitals, and all military hospitals participate in a hospital discharge record system maintained by DoD. However, hospitalizations that occur after discharge can take place in private; public; VA (Veterans Administration); and for those who are retired, military hospitals, making total case ascertainment extremely difficult. Since the population who served in the Gulf is fairly well defined and a very high percentage of hospitalizations during military service can be recovered among those who remain on active duty, studies of hospital discharge data can provide sound comparisons of hospitalization experience based on large numbers of observations. This minimizes the effect of sampling variability on the findings. Hospitalization records typically provide detailed demographic data that can be used to control for important potential sources of confounding. The data reflect the diagnosis assigned at hospital discharge and are only as accurate as the diagnostic categories used in hospitalization records and the diagnostic evaluations conducted at the hospitals under study. Comparisons of large samples will not provide information about the risk of hospitalization among subsets of the POW veteran population subject to specific exposures or hazards unless exposed populations can be defined. Additionally, military treatment centers may not provide the majority of care for certain health conditions of interest (e.g., obstetric care is often provided in the civilian sector for some of these military groups). Finally, hospitalization studies of those who remain on active duty will be subject to biased ascertainment if health-status- specific discharge rates differ for PG and PG-era veterans. To date, one study of the hospitalization experience of POW veterans has reported preliminary findings (Coate et al., 1995~. These investigators, at the NHRC in San Diego, used a retrospective cohort design to compare prewar and postwar hospitalization rates of 547,076 regular Army, Navy, Marine, and Air Force active duty troops deployed to the PG between August 8, 1990, and July 31, 1991, with those of 618,335 nondeployed veterans. The study sample consisted of all regular service personnel on active duty at the beginning of the study period who were deployed during that period and for whom records were available, and a random sample of nondeployed personnel matched for service. Four periods were selected for study: October 1, 1988, to July 31, 1990 (prewar); August 1, 1991, to December 31, 1991; all of 1992; and January 1, 1993, to September 30, 1993. The period August 1, 1990, to July 31, 1991 (roughly the period of the POW), was not included because of known differences in care between the two groups during that time. In each period, the study was restricted to people who were still on active duty on the first day of

76 HEAL TH CONSEQ UENCES OF THE PERSIAN G ULF WAR the study period. Hospitalization data were drawn from the DoD hospital discharge records provided by all U.S. military hospitals and were matched to records of service personnel by social security number. Hospitalizations for any cause and with a diagnosis in each of 14 broad ICD-9 (International Classification of Diseases, version 9) diagnostic categories were compared for deployed and nondeployed personnel, both before and after adjustment for demographic characteristics. This study found no increased risk of hospitalization for any cause among PGW veterans compared with nondeployed veterans, even after adjusting for a possible "healthy worker effect." (This is a phenomenon commonly observed in studies of workers, whereby their death and disease rates are lower than those of the general population. Reasons may include favorable health status associated with the ability to find or keep a job, the comprehensive health care offered by some employers, errors in reporting work status, and other factors. In the present context, perhaps it could be called the "healthy veteran effect.") Examination of 14 broad diagnostic categories in each of the three postwar periods found only four instances of possible increased risk of hospitalization among PGW veterans: neoplasms (largely benign), 1991; diseases of the genitourinary system, 1991; diseases of blood and blood-forming organs (mostly anemias), 1992; and psychiatric disorders, 1992. No diagnostic category of hospitalizations was elevated in all three postwar periods. PGW veterans did experience higher rates of hospitalizations for psychiatric disorders. Detailed analysis of these hospitalizations showed an increased risk of hospitalization related to the use of alcohol and drugs. This study (Coate et al., 1995) had several important strengths. The large sample size and availability of demographic data reduced the effects of sampling variation and confounding on the results. Virtually all hospitalizations of active duty personnel (with the possible exception of childbirth) take place in military hospitals and are captured in the hospital discharge data set, so that bias in ascertainment of hospitalizations is not likely to explain the observed results. Limitations of the study include its restriction to regular active duty personnel who remained on active duty throughout the study period. Of the initial cohort of 1,279,931 individuals, 91% remained on active duty on August 1, 1991; 84% remained on January 1, 1992; and 66% remained on January 1, 1993. Increased hospitalization rates among those who left active duty because of their illnesses would not have been detected unless these illnesses had led to hospitalization prior to discharge. Diseases with latency periods longer than 3 years would not be detected by this study, nor would illnesses that did not lead to hospitalization. Increased hospitalizations due to public awareness of the putative Gulf War Syndrome would have been limited to PGW veterans. Since this was not observed, response bias appears not to have appreciably affected the data.

HEALTH OUTCOMES 77 The team at NHRC also is conducting a study of rates of hospitalization of veterans from all service branches who have separated from active service. This study will compare the hospitalization rates of PGW veterans and a random sample of PGW-era veterans in nonmilitary hospitals in California between 1989 and 1994. The PGW cohort will be defined as military personnel who deployed to participate in the PGW at any time between August 2, 1990, and July 31, 1991. The study population will be further limited to PG veterans who, when deployed, had resided in California for at least 1 year. Prewar hospitalization rates will be collected to compare the predeployment health status of PGW and PGW-era veterans. Hospitalization rates also will be compared to those of active duty personnel. This study has the potential to complement the previously described hospitalization study of the same group of PGW veterans who remained on active duty. It will address the possibility that PGW veterans experiencing adverse health effects of service were differentially separated from service, thereby masking a health impact on the cohort as a whole. The study will provide information on hospitalization rates after separation for about 12% of the PGW cohort, the percentage of troops who satisfy the California residency requirements. Limitations of the study include its inability to identify hospitalizations outside California for the same individual and the possibility that admission criteria differ for nonmilitary hospitals and VA Medical Centers (VAMCs). Despite these limitations, this study will provide information about hospitalization rates after discharge that complement the previously described study of PGW veterans who remained on active duty after the war. It may also help to characterize the potential for differential follow-up in the study of military hospitalizations resulting from earlier separation of veterans experiencing health problems after the war. This is a good plan that may provide a model for future deployments. It will be important for researchers to acknowledge study design limitations as they interpret the findings. A limitation of both mortality analyses and hospital discharge analyses as commonly performed is that they are usually restricted to a single cause of death on the certificate (usually the "underlying cause of death") or a "principal reason for hospitalization." A second limitation of both study designs is that even an examination of the entire PGW veteran cohort would be unlikely to detect the health effects of exposures that affected only a small geographical or occupational subset of the cohort or were rare events. When taken together, however, completed studies of mortality and hospitalization rates of PGW veterans provide no evidence of an association for any widespread or common environmental exposure or other threat to health of sufficient severity to increase the risk of early death or hospitalization among a large segment of the cohort of deployed troops in the 2 to 3 years after deployment.

78 HEALTH CONSEQUENCES OF THE PERSIAN GULF WAR DIAGNOSED DISEASES IN PG VETERANS One disease that has been diagnosed in a small number of veterans is leishmaniasis. As discussed in Chapter 2, the diagnosis of viscerotropic leishmaniasis is difficult to make, and only 12 cases of the viscerotropic and 19 cases of the cutaneous form have been diagnosed among PG veterans (Hyams et al., 1995~. Leishmaniasis has been associated with service in the PG. DoD and DVA have established registries to which active duty service personnel (DoD) and veterans of the PG war (DVA) can report any illnesses or symptoms that they believe are related to POW service and receive a health review and examination according to a standard protocol. Associated with each of the registries are follow-up centers that provide more extensive examinations as needed. Since 1992, DVA has had three national referral centers in Washington, D.C.; Houston, Texas; and West Los Angeles, California. In 1995, a fourth referral center was added at the VAMC in Birmingham, Alabama. The DoD Specialized Care Centers (SCCs) offer, in addition, medical treatment and rehabilitation services. Both registries were established to provide veterans with health concerns access to comprehensive physical examinations and baseline laboratory and other appropriate diagnostic tests. These registries could be used to identify individuals with illnesses and symptoms that may have resulted from service in the PG. However, because both registries were designed to provide a clinical series of self-referred persons for whom common health-related information is collected systematically, they are not suitable for use in formal studies of the frequency of PGW-related adverse health effects. Both registries and their related referral centers continue to operate and to register additional self-referred individuals. Besides providing service to individuals who attend the health review and follow-up exams, these registries provide minimal ongoing morbidity surveillance. Public Law (PL) 102-585 included instructions to the secretary of Veterans Affairs to establish and maintain a PGHR, whose purposes and nature were not specified in the law. As indicated in the committee's first report (IOM, 1995a), the PGHR may be a valuable clinical tool, but it was not designed to provide information on etiology or disease frequency and cannot be used alone for research on these matters. As also indicated in the committee's first report, registry data are computerized, but considerable lag times remain from data collection to data entry to availability of data for review and use. The data collection instrument was modified and fully fielded in 1996 to allow reporting of up to ten complaints and diagnoses, where three had previously been the maximum. If there is an error or omission in completion of the form, it is rejected at the point of data entry and returned to the field for correction. Although this quality control check is important, the rejection rate is high and contributes to the slow turnaround time. As a consequence, recently collected data are not fully compatible with data collected earlier, particularly when

HEALTH OUTCOMES 79 reporting "any illness." Data reviewed and reported to the committee through January 1996 still contain no more than three symptoms or diagnoses. We review these data below, but caution that base rates and comparisons are subject to considerable potential bias. DVA Persian Gulf Health Registry Data from the DVA PGHR have been analyzed separately for women (N= 5,429) and men (N = 46,7841. Nine of the ten most frequently reported complaints were the same for men and women (Table 4-1~. Of those with symptoms, a diagnosis was made for 77% of women and 77°/O of men. Men and women without a diagnosis after examination reported fatigue, headache, skin rash, muscle and joint pain, memory loss, shortness of breath, sleep disturbances, chest pain, and diarrhea (Table 4-2~. Women reported abdominal pain among the 10 most Sequent symptoms, whereas men reported cough. As seen in Table 4-3, men with a diagnosis more frequently had infectious diseases arid circulatory system diagnoses, whereas women more frequently had genitourinary system diagnoses. Whether these genitour~nary diagnoses among women are related to the relative unavailability of gynecological care in the Gulf theater cannot be determined, but may warrant further consideration. Regardless of symptom or diagnosis status, 69% of women reported their health as all right, good, or very good compared with 73% of men (Kang et al., 1995~. Table 4-1. Ten Most Frequent Complaints of Female (N= 4,919) and Male (N= 42,705) Veterans in the Persia Gulf Health Registry with Complaint Data Available Complaintsts) . Fatigue Headache Skin rash Muscle, joint pain Loss of memory and other general symptoms Shortness of breath Sleep disturbances Abdominal pain Other symptoms involving skin and integument Diarrhea and other gastrointestinal symptoms 3.6 . . . . SOURCE: Data were received May 1996 from the DVA Environmental Epidemiology Service for registrants through January 1996 (Kang, 1996). Women (%) 23.3 23.0 18.2 14.5 13.9 7.6 5.3 3.9 3.8 Men (DO) 20.7 17.7 18.5 16.5 14.2 8.0 5.9 2.5 3.2 4.5

80 HEALTH CONSEQUENCES OF THE PERSIAN GULF WAR Table 4-2. Ten Most Frequent Complaints of Female (N = 1,009) and Male N= 8,705) Veterans in the Persian Gulf Health Registry with Symptoms but No Diagnosis Complaints Women (%) Men (%) Fatigue 31.5 29.5 Headache 30.3 21.4 Skin rash 20.8 19.8 Muscle, joint pain 15.4 15.6 Memory loss 14.5 15.5 Shortness of breath 10.1 9.6 Sleep disturbances 5.8 6.6 Chest pain 3.9 4.9 Diarrhea and other gastrointestinal symptoms 3.4 3.8 Abdominal pain 3.3 3.0 Cough 3.2 4.3 SOURCE: Data were received May 1996 from the DVA Environmental Epidemiology Service for registrants through January 1996 (Kang, 1996~. Table 4-3. Distribution of Diagnoses for Female (N = 4,919) and Male (N = 42,725) Veterans in the Persian Gulf Health Registry Women Men Diagnosis No. % No. % Musculoskeletal/ connective tissue1,128 22.910,763 25.2 Mental disorders839 17.16,198 14.5 Respiratory system825 16.85,974 14.0 Skin/subcutaneous tissue619 12.65,719 13.4 Nervous system474 9.63,438 8.0 Digestive system464 9.44,938 11.6 Genitourinary system440 8.91,134 2.7 Infectious diseases224 4.63,144 7.4 Injury and poisoning212 4.32,016 4.7 Circulatory system203 4.13,109 7.3 Neoplasm18 0.4179 0.4 . SOURCE: Data were received May 1996 from the DVA Environmental Epidemiology Service for registrants through January 1996 (Kang, 1996~.

HEALTH OUTCOMES 81 DoD Comprehensive Clinical Evaluation Program In June 1994, DoD developed a clinical protocol, based on the one being used by the DVA, to provide a comprehensive health exam. Active duty service members who were POW veterans were encouraged to refer themselves. In 1995, DoD established a Specialized Care Center at Walter Reed Army Medical Center to provide intensive treatment to symptomatic veterans. Referrals for this intensive 4-week treatment are accepted from clinicians who identify individuals that are unable to perform their duty or meet fitness and retention standards. In August 1995, DoD published the results of the first 10,020 participants in the CCEP (DoD, 1995~. That report and the June 1995 draft were analyzed extensively by the Institute of Medicine (IOM) Committee on the DoD Persian Gulf Syndrome Comprehensive Clinical Evaluation Program. This committee, separate from ours, was established in July 1994, at the request of DoD. Its evaluation noted that the principal shortcoming of CCEP exams was that they were "not, however, designed to answer epidemiological questions. Instead, it was designed as a medical evaluation and treatment program" (IOM, 1996, p. 2~; however, the committee recognized the "compassionate and comprehensive effort to address the clinical needs of thousands of active duty personnel" (p. 1~. Finally, the committee made several recommendations designed to improve the quality of the clinical information obtained. The most recent evaluation of CCEP data summarizes the completed diagnostic results of 18,598 participants (DoD, 1996), addressing concerns indicated in the IOM (1996) report. CCEP participants include a greater percentage of women, blacks, and older persons than their actual representation in the total deployed POW force. The data from this report are subject to limitations, as DoD clearly has recognized and reported. The CCEP was designed as a clinical rather than a research program; the findings are subject to selection bias since persons had to agree to be evaluated, symptoms are self-reported, reported exposures cannot be validated, and there is no comparable control group. The findings are summarized in Tables 4-4 and 4-5. One significant advance in military medicine was generated by this program, according to the CCEP report (DoD, 1996~. DoD has implemented a comprehensive medical surveillance program for U.S. forces in Bosnia, incorporating lessons learned from the CCEP. The information gathered prior to deployment will allow for better assessment of any subsequent health problems. Stress management programs will be made available to service members and their families. These efforts may help to prevent or reduce the development of illness, psychosocial problems, and other adverse conditions that could result from deployment of U.S. service members in future combat and military operations.

82 HEALTH CONSEQUENCES OF THE PERSIAN GULF WAR Table 4-4. Twelve Most Cited Complaints and Their Frequency as Chief Complaint Among 18,075 CCEP Participants Reported as "Any Reported as "Chief Complaint Complaint' (%) Complaint" (%) Joint pain 49 11 Fatigue 47 10 Headache 39 7 Loss of memory 34 4 Sleep disturbance 32 2 Rash/dermatitis 31 7 Difficulty concentrating 27 <1 Depression 23 1 Muscle pain 21 1 Diarrhea 18 2 Shortness of breath 18 3 Abdominal pain or gastrointestinal symptoms 17 3 SOURCE: DoD, 1996. Table 4-5. Frequency of Primary Diagnosis Among 2, 1 31 Female and 1 5,944 Male CCEP Participants and Frequency of Any Diagnosis for All 18,075 Participants . _ . Primary Diagnosis All Participants, Women (DO) Men (%) Any Diagnosis (%) Psychological conditions 19.1 18.3 36.0 Symptoms, signs, and ill-defined conditions 16.5 18.1 43.1 Musculoskeletal system diseases 15.9 18.6 47.2 Nervous system diseases 8.8 5.3 17.8 Healthy 8.6 9.9 10.2 Respiratory system diseases 6.1 6.9 17.5 Skin and subcutaneous tissue diseases 6.0 6.3 19.9 Digestive system diseases 4.9 6.5 20.4 Genitourinary system diseases 3.6 1.0 5.4 Endocrine diseases 2.7 1.9 7.9 Infectious diseases 2.5 2.6 9.0 Circulatory system diseases 1.6 2.3 8.0 SOURCE: DoD, 1996.

HEALTH OUTCOMES 83 The DoD is currently in the process of making the CCEP database available for use by outside investigators, with all identifying information on individuals in the database removed. The details of accessing these data were not available at the time this report was being finalized, but may be available from DoD by the time this report is released. Predictors of Enrollment in the Persian Gulf Health Registry The availability of databases containing information on Army soldiers, including Army reserve and National Guard, who were deployed to the POW theater and the information on those who registered in the DVA PGHR have led to an investigation of possible determinants of registration. A population of 50,000~0,000 Army soldiers (including reserve and National Guard) who deployed to the PG in Operation Desert Shield/Storm will be identified, including registrants in the PGHR and three controls for each registrant who deployed to the Gulf but who, at the time of analysis, had not registered. All personal identifiers are to be removed, and results presented in aggregate. This study uses information available from DVA and DoD to describe, very roughly, the kinds of individuals who are entered on the PGHR. Variables to be examined among these groups for a possible association with registry enrollment include demographic, medical, and health risk behaviors. The investigators indicate two main objectives to this exercise: (1) to evaluate the techniques and statistical methods necessary for combining and analyzing medical and demographic records before, during, and after deployment; and (2) to determine differences between soldiers enrolling and not enrolling in the registry. Hypotheses for future studies will be generated and information can be used to identify soldiers at risk for reporting problems. This could allow for development of possible interventions before deployment, thus reducing risk. It will be important for the analyses to group factors into exogenous (age, gender, race, and distance Tom nearest VAMC) and endogenous (disease) categories and to study relative influences. If data are available on whether a registrant or control has health coverage, and its type, information may be generated on the importance of this variable for registry participation. We encourage a substantial increase in statistician involvement in this study. STUDIES OF SELF-REPORTED SYMPTOMS Outbreak Studies Studies in which investigators focused on reported clusters of symptoms or illnesses among POW veterans are similar in many respects to the frequent "cluster studies" of illness in the U.S. with a possible environmental cause

84 HEALTH CONSEQUENCES OF THE PERSIAN GULF WAR (Caldwell, 1990~. This analogy is instructive because many of the investigators who have participated in such cluster studies have become skeptical about their scientific value (Rothman, 1990) (also called outbreak studies or "hot pursuit" studies, characterized as fast, preliminary investigations). The typical cluster study is characterized by small sample size, selection problems (i.e., many other groups could have been studied but were not, because the participants did not report a cluster), poorly identified exposures, and a significant potential for information bias resulting from respondent awareness of the underlying concern. In the environmental domain, these studies rarely have been fruitful (Bender et al., 1990; Cutler et al., 1986~. A few well-done cluster studies can be useful at the early stages of an investigation to help define the problem and to rule out some statistical flukes that have been misinterpreted and some possible etiologies. That phase of research on the reported Gulf War Syndrome has been completed. Exploratory studies failed to generate useful leads about either the condition or the exposures that might cause such a condition. This report reviews three such studies: the 123rd Army Reserve Command (DeFraites et al., 1992), the Reserve Naval Mobile Construction Battalion 24 (Berg, 1994), and a unit of the Pennsylvania National Guard (MMW1l, 1995; Reeves, 19951. In each case, the unit came to medical attention because of a report of what appeared to be an unusually high rate of unexplained illness. The troops were then carefully studied by a medical and epidemiologic team. 123rd Army Reserve Command (ARCOM) The first "hot-pursuit" study was completed by investigators at Walter Reed Army Institute of Research (WRAIR) (DeFraites et al., 1992), who investigated reports of symptomatic complaints among reservists belonging to the 123rd ARCOM, Lafayette, Indiana. Staff of the 123rd ARCOM Surgeon's Office became aware of these complaints early in 1992. Similar complaints were reported subsequently by members of the 417th Quartermaster Company, in Scottsburg9 Indiana. In response to growing concern about these reports, the team from WRAIR visited Fort Benjamin Harrison and neighboring facilities in April 1992. During this visit, 79 reservists who were concerned about their health were evaluated. All 79 participants completed a medical questionnaire, and 78 completed a Brief Symptom Inventory (BSI) and were available for a detailed interview. Each reservist interviewed completed a brief psychiatric intake-type interview and had vital signs measured. All but one of the reservists received a dental examination. All 78 who participated in the interviews also had blood drawn for complete blood count, white blood cell differential, platelet count, erythrocyte sedimentation rate, and liver function studies. All sera were tested for antibodies to Leishmania tropica. The performance characteristics of this

HEALTH OUTCOMES 85 test for use in screening have not been fully defined (DeFraites et al., 1992~. S era from selected individuals were tested for antibodies for brucellosis. Limited comparative data were available from other groups of veterans. Fatigue was the most common symptom (70%~. Other systemic symptoms, including fever, abdominal pain, and diarrhea, were less common. The onset of fatigue and associated symptoms tended to occur after redeployment from the PG; the onset of diarrhea was more frequent during deployment. No cases of leishmaniasis, brucellosis, or Lyme disease were detected. The findings did not suggest a common pattern of illness among study group members. Review of potential exposures during Operation Desert Shield/Storm (ODS/S) provided no evidence that the respondents had exposures to microwaves, chemicals, radiation? or other suspected environmental hazards. These reservists did report high levels of stress. Investigators noted that the rapid deployment and subsequent redeployment were stressful for many reservists and their families. They believed, however, that PTSD was present in few, if any, of these reservists. The investigators concluded that this study provided no objective evidence of an outbreak of any diagnosable disease in this group. They concluded that the documentable medical problems and illnesses found were typical of a general population with similar demographic characteristics. As investigators noted, this study provided no basis for defining a "case" of disease. No evidence of a common exposure was found. DeFraites et al. (1992) reported that symptoms and objective findings seemed to appear in two peaks, one coincident with return from the Gulf and another some 6-8 months later. They could find no calendar month associations, clustering by activities, or evidence of any "dose-response" relationships with increasing length of time in the Gulf theater to suggest that at least some of the reported symptoms could be related to the reentry of these reservists. The largest proportion of illnesses that resulted in time lost from work was attributed to injuries and, thus, was recognized and explained. The morbidity assessments reported in this investigation do not include data collected by civilian providers, so only a portion of the health records are available for a limited subset of participants. This study is useful in documenting the absence of a common underlying malady or environmental exposure among a group of reservists who initially appeared to have similar medical problems. It also shows that the complaints of this group did not represent a common pattern of illness or environmental exposure. Somewhat reassuring, however, was the fact that one of the first studies of aspects of a PGW-related illness did not show alarming patterns of disease or exposure, other than the possible impact of stress on the troops.

86 HEALTH CONSEQUENCES OF THE PERSIAN GULF WAR Reserve Naval Mobile Construction Battalion 24 (SeabeesJ The second investigation reported by DoD involved the Reserve Naval Mobile Construction Battalion 24 (Seabees) (Berg, 19941. Many members of detachments of this battalion complained of symptoms that they believed were related to their service in the Gulf theater. The Asheville, North Carolina, detachment first reporting symptoms was contacted in November 1992, and additional detachments of this Seabees battalion were followed a year later. A team of physician epidemiologists and a preventive medicine technician visited four detachments of the unit between November 1993 and February 1994 to evaluate Seabees who believed they were affected by PGW-induced health problems. A total of 154 of 232 ODS/S veterans in four detachments were surveyed. After investigators obtained proper consent from the reservists, diagnoses were verified by medical record review and by interviews of medical care providers. The findings centered on the evaluations of these 154 ODS/S reservists. These veterans reported from 1 to 20 symptoms, including fatigue, joint and muscle pains, and confusion and irritability, but the symptoms did not point to a common underlying medical diagnosis. This study did not identify patterns of disease or exposure that seemed to be useful to define further research. Pennsylvania Air National Guard In December 1994, the Centers for Disease Control and Prevention (CDC) began an investigation of a report of unexplained illnesses among veterans of the POW who belonged to a unit of the Pennsylvania Air National Guard. The initial cluster investigation expanded into a study being conducted in three stages. The first stage focused on the identification and characterization of the signs and symptoms of disease in these troops. The second stage was designed to compare the frequency of symptoms in POW veterans and guardsmen who had not been deployed to the Gulf. The third stage was designed to characterize the illness and identify risk factors. A further discussion of each stage follows. Stage 1. A team of CDC medical epidemiologists visited the VAMC and surveyed primary care physicians and regional hospitals in south-central Pennsylvania, identifying 59 veterans reported to be symptomatic. These veterans were evaluated by standardized interviews and physical examinations. They were found to have a high prevalence of symptoms considered to be moderate or severe, including fatigue (61%), joint pain (51%), nasal congestion (51%), diarrhea (44%), joint stiffness (44%), unrefreshing sleep (42%), and a variety of others suggestive of mood disturbances and musculoskeletal disorders. No consistent abnormalities were identified on physical examination (MMWR, 1995).

HEALTH OUTCOMES 87 Stage 2. From January to March 1995, 3,927 members of the index unit and three comparison units were surveyed. This survey established that in all units, symptom rates were higher in veterans who had been deployed to the Gulf than in those not deployed. Symptom prevalence was greater in the index unit than in the comparison units; however, the index unit had nearly twice the deployment rate of the comparison units. Controlling for deployment accounted for the apparently higher rates in the index unit (MMWR, 1995; Reeves, 1995~. The investigation then focused on identifying case-defining symptoms. Two approaches were used: clinical epidemiologic and statistical. In the epidemiologic approach, a symptom was considered case defining if it had been present for 5 or more months, was reported by at least 25% of PGW veterans, and was reported at least 2.5 times more often by PGW veterans than by those not deployed. This approach yielded three clusters of symptoms: (1) fatigue; (2) mood and cognition disorders (impaired memory or concentration, moodiness, or difficulty sleeping); and (3) musculoskeletal complaints (joint stiffness or pain). The statistical approach employed factor analysis to identify two symptom factors. The first was characterized by fatigue and impairment of mood or cognition; the second was characterized by joint or muscle pain and joint stiffness. The epidemiologic and statistical approaches produced similar findings, and investigators focused on the epidemiologic approach. Based on these findings, they proposed a working case definition of illness as one or more chronic symptoms from two or more of three symptom categories: (1) fatigue, (2) mood and cognition, and (3) musculoskeletal. Preliminary findings by CDC investigators found that 42% of PGW veterans and 13% of nondeployed troops in the study met the working case definition. Investigators concluded that this symptom complex was very common among PGW veterans, whereas the prevalence rates were similar for nondeployed veterans and a civilian population involved in a study of chronic fatigue syndrome (Reeves, 1995~. Stage 3. This stage will compare cases, using the working case definition above, and controls from the Air National Guard unit to identify possible risk factors. The proposed study design includes provisions for structured questionnaire administration, structured physical examination and specimen collection, and proposed medical record review. Stage 3 was not complete at the writing of this report. Investigators concluded from the observations to date that the statistical and epidemiologic approaches yield highly concordant working case definitions for a definable syndrome, which can be used to define severity of illness based on the severity of underlying symptoms. They also concluded that the syndrome defined in this way is common in PGW veterans but is also found in nondeployed members of the Air National Guard.

88 Summary HEAL TH CONSEQ UENCES OF THE PERSIAN G ULF WAR These studies came to the similar conclusion that troops reported high rates of a variety of nonspecific symptoms, including fatigue, joint pain and stiffness, disturbed or unrefreshing sleep, some gastrointestinal complaints, and a variety of complaints suggestive of mood and musculoskeletal disorders. Although these studies demonstrated considerable similarity in the nature of the reported health problems across units, they had limited ability to determine the existence or nature of a unique Gulf War Syndrome. First, because these studies were initiated in response to reports of unusual medical problems in the three units, they do not provide information about the prevalence of such problems in the larger cohort of troops deployed to the Gulf. Because symptoms are self-reported, it is difficult to determine whether the higher rates represent true differences or whether they result from recall bias among troops aware of the general public debate about a Gulf War Syndrome as well as the medical concerns of others in their unit. There is also the possibility of an increased rate of reporting among troops sensitized to notice and bring to medical attention minor and transient symptoms, in part because of concern about the availability of health coverage and eligibility for DVA medical care on a "service-connected" basis should those symptoms become more severe. A second concern is that the outbreak studies have not consistently linked these symptoms to either a medical diagnosis or a specific exposure. Investigators were severely limited in their ability to associate symptoms with exposures in the Gulf since military records were not designed to maintain this type of information and no common exposure was evident among those interviewed. A third concern is that the symptoms reported in these units are also common in the general population (Kroenke and Price, 1993~. When a study population is selected because of a reported cluster of affected troops and no scientifically valid comparison group is available, the investigation cannot provide a scientific basis for determining whether rates of self-reported symptoms differ from those expected in the general population of military veterans, or in the population at large, by more than random variation. Certain factors are also known to influence symptom perception fairly dramatically, including cognition (beliefs), context (setting in which the person has a symptom), attention (degree paid to it), and mood (Barsky, 1995~. Thus, although these outbreak studies were successful in demonstrating a common pattern of perceived health problems across a range of military units deployed to the Gulf, they were not successful in demonstrating that these symptoms occurred at a higher rate among POW veterans than among PGW-era veterans, or that these symptoms could be linked to specific medical diagnoses or exposures. Proposals for fixture studies of this type should be scrutinized

HEALTH OUTCOMES 89 carefully; they are unlikely to be useful, and they may divert attention and resources away from potentially more useful studies. Surveys This section reviews the results of several surveys of PGW veterans and comparison groups. In some studies, the survey was supplemented by examination of all or a subset of study participants. These studies provide information on the frequency of self-reported symptoms and illnesses of soldiers deployed to the Gulf theater and may be of value in identifying diagnoses or illnesses of special concern. The major limitation of such studies is that they reflect the respondents' reports of perceptions of their health status, and such reports commonly reflect more than health per se. To the extent that responses are influenced by reporting and public discussion of the health effects of the PGW, individual reports may not correspond to the findings that would be obtained by medical examination. Even if the possibility of reporting bias is taken into account, however, well-designed surveys can provide information about the number and characteristics of veterans who, by their own assessment, do have specific illnesses or symptoms. There are good reasons to achieve consistency in format and style, but those reasons do not necessarily define the most scientific approach to studies that depend on symptom lists and questionnaires or that use subjective psychological interviews. The most important reasons for consistency are to better ensure the utility of results and the robustness of findings for cross-collaboration with other studies. However, judgments about exactly how to achieve such consistency must remain with the scientists designing the studies. From time to time, under the mandate of the Reduction in Paperwork Act and its general charge to protect both the public and the public purse, the Office of Management and Budget (OMB) has required consistency in the questions included in a wide variety of survey instruments. However, changes in the thrust of the research projects have resulted sometimes. Substantial delay in initiation of some potentially important research has occurred. OMB should give high priority to rigorous and timely PG research by ensuring that DVA, DoD, and other agencies make adequate and appropriate use of scientific review by expert peers and encourage the development of individual study designs. Naval Health Research Center (NHRC) Studies At this time, initial findings from only one population-based survey of veterans of the PGW have been reported (Kaiser et al., 19959. This was a survey of all regular Navy Seabees who served in the Gulf or elsewhere at the same time and were still on active duty between September 1994 and June 1995. A total of 1,497 Seabees responded to the survey, 527 of whom had been

9o HEALTH CONSEQUENCES OF THE PERSIAN GULF WAR deployed to the Gulf. The Seabees were selected for study because Mobile Construction Battalion 24 had previously reported a high prevalence of symptoms, and the more current work built on these initial investigations (Berg, 1994~. The demographic characteristics of deployed and nondeployed respondents were similar. Veterans who had been deployed to the Gulf had higher mean scores on all five dimension scales of the Hopkins psychological symptoms profile (somatization, obsessive-compulsive disorder, interpersonal sensitivity, anxiety, or depression); higher scores indicate greater severity or extent of symptoms. Their mean scores were low relative to the mean scores of patients with clinically diagnosed depression or anxiety disorders. Active duty Seabees are concentrated in California and Mississippi, and investigators traveled to those sites during the study period. The reliability of the study instrument was evaluated through retesting 260 study participants on a subsequent visit. For a random sample of 150 of the 260 participants in the reliability study, medical and service records were obtained and compared to survey responses. Physical measurements of grip strength and spirometry were obtained for study participants. Mean values of grip strength and FEV1/FVC ratio (ratio of forced expiratory volume in one second to forced vital capacity), a measure of airway flow, did not differ between Seabees who went to the Gulf and those who did not. An outside panel reviewing the studies at NHRC was skeptical that a study of active duty Seabees was adequate to address PG health issues, due to the number of Seabees that did not remain on active duty because they were in the reserves or separated from service. Therefore, a more comprehensive study has been designed to investigate self-reported post-PGW symptoms of physical and functional illness among all Navy Seabees (active, reserve, or separated from the military) who were deployed to the PG between August 1, 1990, and June 3, 1991. The symptom rates will be compared to those of PG-era veterans. The population identified for this study includes all men and women Seabees (approximately 17,500) who served at least 1 month during the interval specified. For this cohort, baseline information and a determination of physiological and psychological events and syndromes will be assessed by a telephone interview and follow-up database searches of hospitalizations and mortality. Initial analyses will examine participation rates, questionnaire data (descriptive), and questionnaire data in conjunction with the geographical information system (GIS). Investigators plan to track study participants using the Defense Manpower Data Center (DMDC) database and will obtain death certificates from the DVA or National Death Index (NDI). If an acceptable response rate is obtained from the initial survey, the study will proceed to Phase II surveys, with periodic reanalysis of follow-up data in the years 2001, 2006, and 201 1. Because the population is relatively young and

HEALTH OUTCOMES 91 healthy, investigators do not anticipate sufficient numbers for hypothesis testing until the later study years. This is a reasonable study, although it is disappointing that the study was not initiated by DoD several years ago. The main drawbacks are problems with the telephone survey, overstudying of the Seabee population, incomplete follow-up, and data capture in general. Refusal bias is another major threat. The investigators have indicated that they will document correlates of refusal; in this way, they will know whether the study is valid, but there is no reliable way to protect against these sorts of biases. Although the study is restricted to Seabees, they constitute a well-def~ned target population; it has the potential to be informative, but exposures will be self-reported. Findings and associations detected in specific study populations may serve to define needs for studies of other PG veteran populations. D VA National Health Survey The National Health Survey of PG veterans and their family members, funded by DVA, is a mailed survey that will compare self-reported prevalence of health outcomes between 15,000 PG and 15,000 PG-era veterans randomly selected from the DMDC database, oversampling for women and reserve troops. Exposure data will be self-reported and, eventually, linked with DoD unit location files. At the time of this writing, surveys have been mailed and a repeat mailing has been sent. This study may have adequate power for certain relative risks to eliminate associations if participation rates are high and if recall bias can be ruled out. The investigators, in their proposal, comment on many of the issues and threats to validity. This a well-designed and well-intended study, but it has started at least several years too late. Recall problems and the inability to obtain accurate information on those who died before the study started are major threats to its validity. If the response rate is low, if there is a strong differential in the propensity to participate between PG veterans and PG-era veterans, or if there is differential recall bias, the study will be limited. There appeared to be little statistical input in the analysis plan reviewed, and these data will require sophisticated statistical adjustment. The survey instrument is well formatted to increase completeness and accuracy. However, symptom checklists can be of questionable value. Definitions are fuzzy, and higher prevalences can arise when checklists are used than when respondents are invited to report their symptoms without a checklist. Additional aspects of the DVA study include a validation study, special efforts to reach a sample of nonrespondents, and physical examination of both PG and PG-era veterans. The review of military records is a good idea, but if the record and the questionnaire disagree, which is the gold standard? Inviting these individuals for physical examination is a major undertaking and may or

92 HEAl TH CONSEQUENCES OF THE PERSIAN GULF WAR may not provide useful information on health outcomes. Investigators should address what they will do if in fact the validation study shows that study results may not be valid. Survey ofIowa Veterans A smaller survey of PG veterans (approximately 3,000) is being conducted in Iowa, funded by CDC at the request of Congress. Study objectives are to determine differences in the prevalence of health outcomes between (1) PG veterans and PG-era service personnel; (2) PG active duty military and non-PG active duty military; (3) PG National Guard and reserve troops and non-PG National Guard and reserve troops; and (4) PG active duty and PG National Guard and reserve troops. The protocol is a comprehensive document. The level of detail indicates what is intended and how tasks will be accomplished. The project will have three advisory committees: internal, external scientific, and external community. The statistical power analysis indicates that the study can detect a 50% rate increase in the "exposed" group; exposure is not specified but is assumed to be "service in the PG." Available information on health consequences indicates that a 50% increase may be unrealistically high. The actual power of the Iowa study to detect a smaller increase will be substantially lower than the 80% indicated in the analysis. The survey methods proposed include a 5°/0 reinterview rate that will allow estimation of interview reliability. Despite this study's strengths, it has two major limitations. First, the telephone interview is critical to the entire approach. There are problems inherent in trying to collect the necessary information by telephone self-report. In addition, the interview will be long; there will be many who refuse to participate, and others will terminate the interview part way through. Because of these factors, the sample will be highly selected. Information collected from surrogates is likely to be very different from that obtained directly from study participants. A second and greater concern is that this study has been limited to Iowa. Much work has already gone into a study that will be underpowered and too narrow. Sample size calculations were based on examining "risk" among those exposed versus those not exposed (presumably exposed or not exposed to the POW). Because it is likely that subpopulations or less than the whole study group will be of interest for analyzing outcomes, the study is unlikely to have adequate size to detect significant differences. The population of Iowa veterans is also likely to be too narrow a focus because veterans with an Iowa home of record may differ in some explained or unexplained way from the total population of PG veterans, which will limit the interpretation of results. The "Iowa" constraint is very limiting.

HEALTH OUTCOMES ADVERSE REPRODUCTIVE OUTCOME STUDIES Pathways for Environmental Influences on Reproduction 93 The following section on reproductive outcomes and methods for studying them is purposely more detailed than other sections in this chapter, to serve as an example of the methodologic difficulties involved in a specific area of research, which can be broadly applied to other study designs. Its length is not meant to reflect or imply a greater degree of importance of this potential health outcome. To study reproductive health consequences of PG service and determine the potential for further productive research, one must consider how environmental exposure might exert such effects. These pathways or mechanisms help to define how reproductive health might be influenced by some aspect of having served in the POW. One possible mechanism for adverse effects on reproduction is through direct harm to the nonpregnant woman, such as genetic damage to the ovum, which can lead to chromosomal anomaly in the fetus (Kline et al., 1989), or scarring of the fallopian tubes, which leads to an inability to conceive (Healy et al., 1994~. These changes, although permanent, may not be detected for many years after they have occurred, until the woman attempts conception, pregnancy, or delivery. Other types of direct harm may be temporary or reversible. Males also may have adverse reproductive effects from exposure to environmental agents. Exposures that damage sperm production can produce temporary or permanent infertility (Schrader and Kesner, 1993; Skakkebaek et al., 1994), and some types of genetic damage to the sperm could result in fetuses or children who do not develop normally (Olshan and Faustman, 1993~. Since the production of sperm cells is continuous throughout adult life, most genetic damage that affects spermatozoa directly will be manifest only in pregnancies conceived relatively soon after the exposure (within about 74 days) (Scialli, 1992~. Thus, male exposures related to the POW that were soon followed by conception would be most plausibly affected in this manner. It is possible for stem cells to be permanently harmed, although they appear to be less susceptible than later stages of spermatogenesis (Scialli, 1992~. If they do undergo genetic changes, however, these would result in continuous production of genetically defective sperm. Exposures that produce this effect would lead to a permanent decrease in fertility or an increase in risk of abnormalities in the offspring. Most research on the reproductive consequences of environmental exposures in general has focused on pregnant women and, hence, on exposure of the fetus. There is abundant evidence that the fetus is highly vulnerable to insults that come through the pregnant woman (Shepard et al., 1993), with potential increases in a range of different adverse outcomes. If severe harm early in pregnancy causes miscarriages prior to the recognition of pregnancy, the woman is likely to be misdiagnosed as infertile. Fetal loss at a later stage

94 HEAL TH CONSEQ UENCES OF THE PERSIAN G ULF WAR results in miscarriage or stillbirth. These are generally recognized and reported, although definitions of miscarriage versus stillbirth versus postpartum death are difficult to apply consistently. Thus, differences in reported data may simply reflect differences in reporting standards. Structural damage to the developing fetus can result in congenital malformations, whereas reduced growth and low birth weight may reflect a range of fetal insults. Preterm delivery can be a consequence of either maternal or fetal problems. A lengthy interval between maternal exposure and fetal effects can occur only if the maternal exposure itself is persistent, which may occur with exposure to heavy metals or chlorinated hydrocarbons that persist in the bone or fat of the mother and cause a sustained elevation in blood levels. The relatively small number of pregnancies among women while they were serving in the Gulf limits opportunities to study immediate fetal effects, but the potential for continuing effects of a persistent maternal body burden of Gulf-related toxic agents needs study. Definition of Outcomes Many facets of reproductive function have been associated with exposure to drugs, chemicals, and other environmental agents (Mattison, 1994), ranging from loss of libido to failure of spermatogenesis or lactation to major birth defects. Many of these reproductive events are difficult to define and diagnose accurately. Infertility is, in part, a function of the opportunity to conceive and, thus, is likely to remain undetected for some time, perhaps long after biological damage has occurred. Miscarriages are notoriously difficult to identify with certainty, readily missed when they occur early in gestation, and potentially overdiagnosed when menstrual periods are unusually heavy or delayed (Kline et al., 1989~. Many congenital anomalies cannot be detected at birth or in very young infants, and accurate and complete diagnosis requires careful examination at the proper time in development. Congenital heart defects, for example, are rarely identifiable in live newborns and require monitoring the child for up to 5 years to be comprehensively ascertained. Because there are often no standardized definitions and the number of identified reproductive events depends on the level of scrutiny, valid research designs must identify the frequency of adverse outcomes in truly comparable groups with and without exposures. For example, given the lack of standardization in the definition of miscarriage, there is no benchmark or "normal" rate of miscarriage to which an exposed population may be compared. Instead, an exposed population must be compared with an unexposed population, with the same miscarriage definitions and the same opportunities for observation of miscarriage. Conditions subject to underascertainment require comparable levels of scrutiny in the populations to be compared: infertility rates will appear higher in a population that has a greater desire to have children than in a population that has

HEALTH OUTCOMES 95 less desire simply because failures to conceive will come to attention more often in the former. As another example, about 5% of all newborns can be found at the time of birth to have abnormalities, increasing to about 10% after the first year or two as problems become manifest, but many of these may be diagnosed only by a specialist experienced in this field. Findings in an exposed population subject to special searching are not comparable to findings in an unexposed, unscrutinized population. Even the term "congenital defect" encompasses a wide range of processes, including DNA damage of parental germ cells prior to pregnancy, adverse intrauterine environment, and birth injury. Much of the scientific literature on environmental exposures and reproductive outcomes has been based on studies of special, often high-risk, populations. Thus, direct comparisons of rates of adverse outcomes in those special studies with the rates among POW veterans may be misleading. Some male veterans and their partners have reported that contact with their semen causes a burning sensation in the partner. This issue was discussed at some length by an expert panel convened by the Department of Veterans Affairs (DVA, 1995a), with efforts to identify how many such cases had been reported, whether any known condition could account for such a phenomenon, and what might be done to further evaluate the reports. Based on a few case reports of allergic responses to seminal constituents (Bernstein et al., 1993), the most plausible candidate explanation for this symptom is some type of local hypersensitivity reaction. This hypothesis is based on the absence of male discomfort associated with seminal fluid and the temporal pattern of symptoms after exposure to ejaculate. Dr. David Bernstein, an immunologist at the University of Cincinnati Medical Sciences Center, reported to the panel that detailed evaluations of approximately 40 couples as described in the scientific literature are consistent with this phenomenon. The women complain of allergic symptoms after exposure to semen that are not accounted for by infection, which is a much more common explanation for such symptoms. Symptoms are prevented by use of condoms, and in some cases the hypersensitivity has been treated successfully. In theory, such a phenomenon could be induced by male exposure to agents that contribute to the formation of new allergens in the seminal fluid. However, as noted by the panel, there is no clear documentation at this time of how frequently the phenomenon is reported by either veterans or He general population. The recommendation of the DVA's expert panel for a clear case definition followed by accurate information on the magnitude of the problem among PG veterans is a reasonable approach.

96 HEALTH CONSEQUENCES OF THE PERSIAN GULF WAR Frequency of Events in the General Population Many of the broad categories of reproductive problems are quite common, with typical ranges for infertility of 10% of all couples, and 10~15% of recognized pregnancies resulting in fetal loss (perhaps 30% or more including those not recognized), and with 5-10% of those resulting in preterm delivery or low birth weight. Even congenital malformations, which are individually rare (less than 1 per 1,000), are not as rare in the aggregate, being recognized in 3-5% of infants at the time of birth. Congenital malformations range from cosmetic (birthmarks or webbed toes) to significant impairment (Down's syndrome or congenital dislocation of the hip) to severely life threatening or lethal (aneuploidy, anencephaly, or transposition of the great vessels). Thus, anecdotal case reports of a given exposure (whether associated with the POW or with a drug or environmental chemical) must be interpreted very cautiously because of the high baseline rate of adverse reproductive outcomes. For many reproductive outcomes, not all individuals in a given population will contribute information. In a population of exposed and unexposed individuals, only a subset will attempt to conceive (needed to address infertility), a subset of those will have conceptions (needed to address miscarriage rates), and a subset of those will result in births (needed to address birth weight or congenital malformations). Reports of clusters of adverse reproductive outcomes (particularly miscarriages or birth defects) have often come to attention as possible clues to a hazardous environmental exposure. None of these episodes of reported clusters in the general population have yielded new insights into the causes of reproductive health problems, in spite of many investigations by health depa~l~ents, CDC, and others. Often, despite what appears to be an unusual number of events, the small population of interest with the resulting limited opportunity to examine dose-response gradients and confounding leaves these episodes unresolved. In addition, once potentially affected individuals become aware of and concerned with that possibility, they are often more attentive to potential hazardous exposures and adverse outcomes and may be inclined to report more exposure and more disease (perhaps more accurately) than comparison populations, so that further reports of an excess must be questioned. Therefore, especially in light of heightened awareness, nothing short of a rigorous evaluation of exposures, reproductive outcomes, and potential confounders is likely to yield informative results. Confounding Much like the healthy worker effect, in which mortality rates are reduced in employed populations due to screening and selection of physically and mentally fit individuals, there may well be a healthy soldier effect that applies to

HEALTH OUTCOMES 97 reproductive outcomes. Screening to participate in the armed services and further selection to serve in the PGW may have produced a population with above-average reproductive health as well as good general physical health and, hence, with a lower-than-normal expected frequency of reproductive health problems. For example, persons with some kinds of hereditary defects may be excluded from military service. Studies intended to address potential adverse effects of exposures in the Gulf theater must take such selection into account, perhaps by identifying equally fit military populations that did not serve in the PG. Another phenomenon that has been observed in groups of employed women is the selection of populations that have not had live births (i.e., continued employment of women who have been infertile or had pregnancy losses) since those who had live births tended to leave employment (Lemasters and Pinney, 1989; Savitz et al., 19901. Thus, those who remained in the work force more often had an unfavorable reproductive history as a result of factors unrelated to their work. An analogous phenomenon might occur in studying women in the military: those who fail to conceive or conceive but fail to deliver a live birth may be most likely to remain on active duty, even if the reproductive problem had origins before or unrelated to military service. A number of"life-style" exposures are known to influence reproductive health, including the use of tobacco, alcohol, and illicit drugs; diet; and sexual behavior involving the risk of sexually transmitted infections. Studies must take such risk factors into account in trying to isolate any effect of the PG experience per se. Service in the Gulf may have affected these behaviors during and after the war, so that isolating the effects of environmental agents encountered during the war on reproductive health requires measurement and adjustment for behavioral risk factors. Given such considerations, national surveys such as the National Health and Nutrition Examination Survey or the National Maternal and Infant Health Survey may not provide suitable standards to evaluate rates of outcomes for comparison to PGW veterans. Reproductive Outcome Studies in PG Populations Pilot Studies Although many studies of PGW veterans are in progress, four pilot projects have evaluated the possibility of adverse reproductive health experiences, and three larger population-based studies are under way. The pilot projects include studies of: (1) birth defects among children born to veterans from four Mississippi National Guard Units (Penman et al., 1996~; (2) pregnancy outcomes among veterans at Robins Air Force Base (Eggert, 1994~; (3) miscarriage rates in the William Beaumont Army Medical Center population (Rosa, 19939; and

98 HEALTH CONSEQUENCES OF THE PERSIAN GULF WAR (4) analysis of miscarriage rates at six Army hospitals (Broadnax, 1992), comparing the proportions of pregnancies ending in miscarriage prior to and after the war. Each of these studies found no pattern of adverse reproductive effects from which hypotheses could be generated. In sum, inferences about the effect of some aspect of service on reproductive health are not warranted at this time. Perhaps the strongest conclusion that can be drawn at present is that no ubiquitous exposure associated with PGW service has been found to produce an adverse impact on common reproductive outcomes. NHRC Studies A study of the reproductive health of veterans is currently in progress at the NHRC in San Diego. A multifaceted study of the health of PGW veterans ("Epidemiologic Studies of Morbidity Among Gulf War Veterans: A Search for Etiologic Agents and Risk Factors") includes three projects focused on reproductive outcomes: (1) "A Comparative Study of Pregnancy Outcomes Among Gulf War Veterans (Male and Female) and Other Active Duty Personnel"; (2) "Reproductive Outcomes"; and (3) "Prevalence of Congenital Anomalies Among Children Born to Gulf War Veterans." The first project, comparing birth defects among offspring born in military hospitals to deployed versus nondeployed veterans, has been presented in preliminary form (Cowan et al., 1995~. Active duty personnel who were deployed to the PG (approximately 580,000) and a sample of troops not so deployed (approximately 700,000) were identified and followed to identify live births in military hospitals. The study included all births conceived after return from the Gulf by deployed persons and after January 1, 1991, by nondeployed persons. Date of birth was on or before September 30, 1992. Outcomes examined were live birth rates, birth defects, and infant sex ratio. Nearly 16,000 live births occurred among spouses of deployed men and 18,000 among spouses of nondeployed men; approximately 2,000 live births occurred to deployed women and 5,000 to nondeployed women. Birth rates were similar for men who were and were not deployed (29.1 versus 29.6 per 1,000), as well as for deployed and nondeployed women (56.4 versus 59.5 per 1,0003. Sex ratios (male to female) were in the range of 0.99-1.06 for all groups, well within the expected range for samples of these sizes. Discharge diagnosis codes for offspring of active duty veterans who had been deployed in the Gulf showed birth defects in the range of 7% for offspring of men and 9°/O for offspring of women. Relative risks comparing deployed to nondeployed men and women were 1.02 (95% CI = 0.9~1.11) and 1.13 (95% CI = 0.97- 1.32), respectively, prior to adjustment for known differences in confounders. These relative risks did not depend in any systematic way on duration of service in the PGW. Adjustment for race, age of mother, branch of service, and rank

HEALTH OUTCOMES 99 reduced relative risk estimates for both men and women (RR = 0.98 and 1.04? respectively). Preliminary results of this study provide no evidence for any major association between simple deployment in the Gulf and birth defects in the aggregate. They do not address specific exposures encountered in the war or specific birth defect outcomes, nor do they address the experience of veterans who are no longer on active duty or who had a child born outside of a military hospital. The second NHRC study focuses on self-reported infertility and miscarriage based on a mailed survey to be sent to married couples. The sample includes strata in which women served in the Gulf, men served in the Gulf, women served in the military but not in the Gulf, and men served in the military but not in the Gulf, with 4,000 to be chosen from each of the four groups. All couples will have been married at least 1 year, and the sample will include active duty, reserve, and nonactive duty personnel. The survey includes demographic factors, life-style, medical history, contraception, reproductive history, and other environmental exposures. Study plans include repeated mailings of the questionnaire, with telephone follow-up to improve response rates. Based on a review of the questionnaire and study plans, the methods seem to be carefully developed and suitable for addressing the endpoints of infertility and miscarriage, despite the considerable challenges in eliciting accurate data on miscarriage and other outcomes. The instrument is brief and designed to encourage complete and accurate reporting. Plans for follow-up of nonrespondents are likely to be successful. The overall design is well suited to screen for potential adverse effects on fertility and pregnancy outcome. The third study of reproductive health focuses on congenital abnormalities in the offspring of veterans. Although this outcome was considered in the hospitalization study, the third study will include more complete information from birth defects registries for evaluation of any association with PGW experience. Records of veterans who served in the PGW and those who did not will be linked to population-based registries of birth defects in Arizona, Arkansas, Hawaii, Iowa, Oklahoma, Georgia (Atlanta), and California (selected counties), each of which has an active monitoring program in place. Births to veterans who lived in these states will be identified from birth certificates, which will allow linkage to the birth defects registry. In addition, there are plans to examine fetal deaths, birth weight, and gestational age using vital records. The proportion of births with specific adverse outcomes, such as malformations or low birth weight, will be compared among PGW and PGW-era veterans, as well as with the nonveteran populations in those states. Approximately 9,500 live births are expected to have occurred to PGW veterans in the selected states. This is an ambitious study for this research team, proposing to conduct record linkage on a scale not previously attempted in the United States. Investigators are conducting a large-scale feasibility study in Hawaii to evaluate

100 HEALTH CONSEQUENCES OF THE PERSIAN GULF WAR study methods. If feasible, the strategy is a good one for studying the overall birth prevalence of malformations. Even though there may be incomplete identification of all births to PG veterans ("all births" serving as the denominator for the calculation of adverse outcomes), the actual proportion of infants with adverse outcomes may not be affected. Within the groups of births identified as having occurred among PGW and PGW-era veterans, the birth defects registries should be quite effective in comprehensively ascertaining and diagnosing early malformations. Despite the study's limitations, the proposed study, if feasible, will be a reasonable approach to a broad survey of congenital defects in the offspring of PGW veterans. Based on a review of the protocols, critique by a panel of independent experts, and response from the team of investigators, some general characterization of the set of projects can be made. Subject to practical constraints (budgets and feasibility), quality of data sources (medical records and self-report), and selection of subsets of all possible reproductive outcomes for detailed study, the projects represent a reasonable set of choices. Through this series of studies, investigators will be able to consider infertility, miscarriage, low birth weight, preterm delivery, and congenital malformations. It is critical to note, however, that unless there are major modifications, investigators will address a potentially broad or universal effect of PGW service rather than specific exposures. Although the global hypothesis of some PGW effect is less precise than one or more specific hypotheses about exposure to individual agents, no highly specific hypotheses are now considered to have much scientific support, and a global view is a reasonable starting point. If results are positive, they could be used to identify more refined issues for further study. Overall, the ongoing efforts to address reproductive health appear to be responsive to the broad and somewhat diffuse array of present concerns. Without some particular, focused hypothesis for a specific environmental agent or outcome, there seems to be little need for expansion beyond the present effort. If the results of those studies provide specific leads to an increased risk of particular outcomes, such leads should be pursued with more detailed research and larger samples to evaluate exposures encountered in the war. It is also possible that some specific exposures producing known reproductive health consequences will be identified as having occurred, which would warrant more focused study of reproductive outcomes. However, unless a broad effect or some specific exposures associated with adverse reproductive effects are identified, there is little reason to continue research on the reproductive health consequences of service in the PGW beyond what is now planned.

HEALTH OUTCOMES 101 MENTAL HEALTH STUDIES The mental state of troops is an important dimension of military effectiveness because it both reflects and affects unit cohesion, morale, and leadership. DoD has invested in programmatic research, focusing on the stressors associated with deployment, as well as the mental and physical well- being of PGW-era personnel. The DVA also has conducted several studies of the health of veterans beginning with their return home (NEPEC, 1992) and continues such research through activities based at specific VAMCs. Issues in Studies of Mental Health Investigators have begun to. study some of the questions concerning the mental health of POW veterans and possible associations with service in the PG. To do so, they have had to address several issues challenging all researchers studying mental health. Measurements of mental health phenomena rely strongly on subjective perceptions of events, emotional responses to them (how one feels about a life experience), and cognitions (how one thinks or gives meaning to an experience). For research purposes, the subjective dimensions of an individual's experiences must be assumed to be valid for that individual. Objective indicators of mental health that can be observed by another person are included in mental health assessments: for example, behavior, such as the ability to function in one's work and social interactions; demeanor; test performance; and the like. To promote valid and reliable mental health diagnoses, the American Psychiatric Association (APA) has defined research and diagnostic criteria for psychiatric diagnoses that represent consensus among many experts in the field and has published them in its Diagnostic and Statistical Manual of Mental Disorders (DSM) (APA, 1987, 1994~. As the field of mental health has developed, diagnostic criteria and the nosology of diagnoses have been continually refined. One such example is PTSD, a phenomenon that has been described in conjunction with prior wars but was not named PTSD in the DSM until 1980. Most studies of POW soldiers produced psychiatric findings that were based on criteria for diagnoses such as PTSD. It has been noted that one of the difficulties in analyzing mental health scores for tests and scales used in the research described is that the "norms" to which comparisons are made have not been developed for a population of men or women in the military. The nature of many of the scales may have the tendency to show high scores (more illness) for military populations compared with the general population because of the way some questions are phrased (Marlowe, 1996~. Diagnosis of psychiatric problems upon return from the Gulf may capture many of the mental health problems that originated during ODS/S or were exacerbated by service. It could be assumed that troops deployed in ODS/S

102 HEALTH CONSEQUENCES OF THE PERSIAN GULF WAR were mentally fit for duty at the time. However, unless baseline predeployment data are available, it is difficult to discern which problems were present before deployment and did not change significantly, which originated with or were exacerbated by service, and which were worse after return. To complicate diagnosis, some problems are acute and remit quickly; some are chronic; still others have a delayed presentation, that is, they may not be apparent until months or years after return from the theater. Mental Health Problems and Military Experience PTSD has become a mental health outcome of primary interest since the Vietnam War era. Many veterans who served in Vietnam were exposed to shocking or horrifying events associated with the war and, in its aftermath, returned to a hostile environment in the United States. Cumulative experiences with "combat stress" after service during earlier wars and growing recognition of the multiple presentations of illness (acute, chronic, or delayed) and sequelae led to recognition of a syndrome (often called "traumatic neurosis") associated with severe trauma. In 1980, the American Psychiatric Association first formulated PTSD among its diagnoses of anxiety disorders in DSM-III. A revision, published in 1987, known as DSM-III-R, was the version used for the studies reported herein (APA, 1987~. The version in current use is known as DSM-IV and was published in 1994 (APA, 1994~. Four criteria are listed for PTSD diagnosis in DSM III-R: (1) exposure to an event that would be stressful to almost anyone; (2) one or more specified symptoms of reliving the traumatic experiences (nightmares, intrusive thoughts, or flashbacks); (3) three or more specified symptoms of withdrawal, avoidance, or numbing of emotions; and (4) two or more symptoms of hyperarousal, such as exaggerated startle or difficulty concentrating. A diagnosis of PTSD requires that criteria 2 through 4 have been met during the previous month. In addition to PTSD, military personnel are subject to other psychiatric disorders including depression, anxiety, substance abuse, and (rarely) a major mental illness such as schizophrenia. These have been documented in some assessments completed by DVA-sponsored projects. In comparison with the atte action given PTSD, little detailed information is available on other mental health problems resulting from service in the POW. Mental Health: Comparison of Deployed and Nondeployed Troops A diagnosis of PTSD is usually made by means of a clinical interview, but algorithms have been developed to identify persons likely to have PTSD by using data obtained by questionnaire, such as the Impact of Event Scale (IES) or the Brief Symptom Inventory (BSI). The latter approach allows an

HEALTH OUTCOMES 103 approximation of risk for PTSD in large sample studies where clinical interviews are not feasible. The IES is a 15-item scale used to assess reactions to trauma and related characteristics associated with stress disorders. The IES scale emphasizes two dimensions of PTSD, intrusion and avoidance (Horowitz, 1986~. High scores indicate more severe reactions to stressors. In a comparative survey, conducted in 1993, of PG active duty personnel and reserve troops from Hawaii and Pennsylvania, reserve troops scored higher than active duty troops on the avoidance and intrusion scales of the IES, but their scores were remarkably lower than those of active duty personnel in the XVIIIth Airborne Corps and the VIIth Corps, who were studied in 1991-1992 and who had engaged in main-force actions during the ground war. These lower scores in the Pennsylvania and Hawaii samples may reflect attenuation due to passage of another year since ODS, a difference in combat exposure, or other unidentified factors (USAMRMC, 19941. The BSI is a 53-item scale of symptoms derived from the 90-item Symptom Check List-90-Revised (SCL-90-R) (Derogatis and Melisaratos, 1983~. It includes nine dimensions or subscales: somatization, obsessive-compulsive, interpersonal sensitivity, depression, anxiety, hostility, phobic anxiety, paranoid ideation, and psychoticism, as well as three global indices of pathology, including a General Severity Index (GSI). Higher scores indicate more severe symptoms. Of concern in interpreting the results is that military samples differ somewhat from civilian samples. For example, a "normal" military sample might not appear the same as a "normal" civilian sample of nonpatients for reasons specific to military selection (Stretch et al., 1996~. A comparison of BSI scores for active duty and reserve personnel who were deployed to the Gulf versus those who were not deployed indicated that the former were at greater risk of PTSD as assessed using the GSI of the BSI. In addition, the nondeployed active duty personnel of the XVIII Airborne had relatively high scores (USAMRMC, 1994), perhaps reflecting concerns over the unit disestablishments in Europe and turbulence and downsizing in the continental United States. A PTSD risk algorithm was developed from the IES and BSI and used to identify those at moderate and high risk for a diagnosis of PTSD. Criteria included exposure to a traumatic event outside the range of normal human experience, intrusive thoughts and memories associated with the trauma, symptoms of persistent avoidance of stimuli associated with the trauma, and persistent symptoms of increased arousal associated with the trauma. Although the ground war was short, scores indicating moderate or high risk of PTSD were more common among those who were deployed than among the nondeployed. Scores of those deployed in the VIIth and XVIIIth Corps indicated moderate to high risk for PTSD in 13-15% of these troops, respectively (studied 6 months to 1 year after ODS), compared with 8% for

104 HEALTH CONSEQUENCES OF THE PERSIAN GULF WAR active duty troops and 10% for reservists from Pennsylvania and Hawai (studied 2 years after ODS) (IJSAMRMC, 1994). Reservists who served in ODS were compared with those deployed in the continental United States and in Europe. Data collected in 1993 revealed that mean scores on a trauma scale constructed from the BSI and the GSI/BSI were essentially the same for personnel deployed in those two areas but significantly lower than scores for the group deployed to the Gulf theater. Thus, activation by itself did not seem to account for psychological symptom levels, which were connected specifically to deployment to the Gulf (Marlowe, 1995~. Inpatient psychiatric services at a U.S. Army combat support post reported no increase in hospitalizations during 1990-1991 by active duty soldiers, families, and retirees before, during, and after ODS/S, coincident with return from ODS. However, the psychiatric proportion of all hospitalizations increased among soldiers not deployed who had served less than 1 year (Koshes and Rothberg, 19949. In addition to DoD studies, DVA initiated several studies of veterans shortly after demobilization. A summary of the DVA assessments of 9,090 veterans seen through its readjustment counseling service after ODS/S (from May to September 1991) indicated a prevalence of PTSD of 2.2%, subdiagnostic PTSD of 6.4%,~ and other psychological problems of 3.1%. Prevalence of alcohol or drug abuse was 1.7%, and marriage and family problems were 3.9% (Blank and Gelsomino, 1992~. Site-specific studies of mental health problems among returning veterans revealed a range of prevalence of PTSD across sites. At the New Orleans VAMC, PG troops assessed within the first year after return from war zone duty had prevalence estimates as high as 16-19% (Sutker et al., 1992~. At the Boston VAMC, the prevalence estimate was 4% for men and 9% for women according to the Mississippi Scale (Wolfe et al., 1992c). One would expect differences across geographic areas due to differences in exposure to combat and differences in military occupational specialty, as well as differences in the demographic makeup of the populations studied and their desire for health services or counseling relative to that expressed by troops evaluated at other VAMC sites. Results of first-year findings from various VAMC studies have been compiled (NEPEC, 19929. We emphasize that many of these studies were small, several did not have control groups, and others were of self-selected study samples. 'Subdiagnostic PTSD includes veterans who meet some but not all of the DSM-III-R diagnostic criteria for PTSD.

HEALTH OUTCOMES Factors Associated with Mental Health Problems: Combat and Other Stressors 105 A variety of factors have been associated with mental health problems experienced by troops deployed to ODS/S. Although combat exposure is among these stressors, a variety of other stressors were greater after deployment to the Gulf theater than to other sites. For example, troops experienced prolonged disruption of normal living patterns and relationships, reunion and readaptation, and culture shock upon returning from the Gulf. These factors have been studied among active duty and reserve personnel and are reviewed in Chapter 3 of this report. Factors Increasing Vulnerability to PTSD and Other Psychiatric Disorders Although combat exposure has a strong influence on development of PTSD, the majority of troops exposed to combat or other major stressors did not develop PTSD. Thus, assessment of vulnerability may aid understanding of how to prevent or attenuate the likelihood of developing PTSD. Vulnerability to mental health problems has been related to person-situation interactions in the psychology and psychiatry literature. A range of factors has been associated with individual differences in response to some stressors; among these factors are ethnicity, gender, educational level, intellectual sophistication as it affects coping, and military rank. Both Sutker and Wolfe studied the relationship between vulnerability factors and mental health outcomes. Sutker et al. (1993) found in one study (N = 215) that female gender and nonwhite race were associated with modestly greater risk of development of PTSD symptoms. In a larger series At= 912; 653 deployed to the PG and 259 with stateside duty), Sutker et al. (199Sa) found that race was not significantly associated with depression. Likewise, Wolfe et al. (1993) found that race was not associated with a higher prevalence of mental health problems. Women had a higher prevalence of PTSD symptoms than men in both the Boston and the New Orleans studies (Sutker et al., 1993; Wolfe et al., 1993~. Higher educational level and officer versus enlisted status were associated with a lower prevalence of PTSD (Wolfe et al., 1993~. A variety of explanations have been offered for these associations. Education and intellectual sophistication may enhance the soldier's ability to understand events, to garner necessary support through social interaction, and to come to terms with the events of the war. In addition, officer status is associated with greater access to personal and social resources, including better training with which to deal with traumatic experience. Officers are generally older and often more experienced in war or in training for war than enlisted personnel.

106 HEALTH CONSEQUENCES OF THE PERSIAN GULF WAR Explanations for gender differences in mental health have included the likelihood that women are socialized to express symptoms whereas men are socialized to suppress expression. Another explanation is that women's experiences of trauma in war were different from those of men. This explanation may be warranted by consideration of data about abuse. As discussed in Chapter 3, an experience unique to women or at least much more prevalent among women who served in the Gulf was exposure to violence and sexual harassment (Wolfe et al. 1992b). Those who reported assault had higher Mississippi Scale scores for PTSD symptomatology, and those reporting assault or physical harassment had higher GSI/BSI scores. There is also evidence that past exposure to violence and abuse, such as childhood sexual or physical abuse, is associated with PTSD. Marlowe (1995) concluded that one could not determine from data collected by DoD whether there were factors that predisposed one to a highly stressed response to combat. He did note that those who were highly stressed before combat were also highly stressed after combat and that those with high psychological symptom scores at the time of the first postcombat study tended to have high scores a year later. These same individuals had high reactions to current life events as well as to ODS events. Factors Enhancing Resilience or Buffering Effects of Stress on Mental Health Sutker et al. (199Sb) compared 97 POW troops with PTSD diagnosis to 484 without psychological distress to determine differences in their personal resources (gender, education, intellectual sophistication, and personality dispositions such as hardiness and coping styles) and environmental resources (perceived social support and family relationship support). She found that less commitment, more avoidance of coping, less family cohesion, and lower social support satisfaction were factors associated with those who developed PTSD diagnoses. DoD studies indicated that the service person's military unit served as a social support mechanism to buffer the effects of exposure to acute and chronic stress on health. Postcombat sharing of experiences within studied units provided a source of aid for coping with the sequelae of war. Shared coping with traumatic events in the field seemed to be more beneficial to soldiers than dealing with such events after returning to the United States (demobilization) (WRAIR, 1992~. As identified by active duty troops deployed to the Gulf from Pennsylvania and Hawaii, sources of support that were somewhat helpful included family, friends, chaplains, and unit members. For reserve troops, similar sources of support were indicated. That report (USAMRMC, 1994) stated that more than 90°/O of active duty and reserve troops studied were coping moderately to extremely well 2 years after ODS/S, but 7.8% of the active duty

HEALTH OUTCOMES 107 deployed felt that they were coping poorly versus 5.5% of active duty nondeployed. Long-Term Mental Health Outcomes Most studies to date have been of short-term acute outcomes. DVA- sponsored studies discussed above, conducted by Wolfe at the Boston VAMC (Wolfe et al., 1993) and Sutker at the New Orleans VAMC (Sutker et al., 1993), indicated persistence of psychological distress among troops for at least 1 year after return from ODS/S. Marlowe (1995) of DoD has followed selected troops for as long as 2 years after return from ODS/S and found that those who were highly distressed before combat were also highly distressed after combat. Thus, these studies have made significant contributions, but follow-up so far is too short to resolve questions about factors associated with the persistence of PTSD symptoms and other types of illness. Symptoms of PTSD in some veterans have persisted for at least two decades after the Vietnam war. Physical Symptoms and Exposure to Stressors Physical symptoms in conjunction with stressors were also studied by DoD. Although physical symptoms do not in themselves signify PTSD or other mental health problems, there is a large literature linking acute and chronic life stress and mental health to future physical morbidities, including gastrointestinal and hyperimmune diseases (Vogt et al, 1994~. A comparison of physical symptoms experienced by deployed and nondeployed active duty and reserve troops indicated that those deployed were more than twice as likely to report head colds, sinus trouble, sore throat, difficulty swallowing, headaches, back problems, stomach upset, muscle aches, aching joints, cough, chills or fever, and other problems. This was the case when demographic factors such as age, rank, education, marital status, and branch of services were controlled (Stretch et al., 1 9951. An algorithm of physical and psychological symptoms was calculated as a proxy indicator for Gulf War Syndrome (GWS) symptoms (Stretch et al., 19953. These included headache, stomach or intestinal upset, muscle aches or cramps, aching joints and bones, weight loss or gain, cough, chills or fever, general level of spirits, level of energy, trouble with memory, pains in the chest or heart, low energy, difficulty getting breath, difficulty concentrating, and feeling weak in parts of the body. Those reporting five or more of the health symptoms and at least one of the psychological symptoms were identified as follows: 178 of troops deployed (12%) and 55 (2.2%) of those not deployed scored high on this index. Of those with high scores on the GWS index, 42% attributed their health problems to ODS/S service, 45% had concern over exposure to oil fires, 36% claimed to have been exposed to oil fires, and 30% met the criteria for possible

108 HEALTH CONSEQUENCES OF THE PERSIAN GULF WAR PTSD diagnosis (Stretch et al., 1995). The authors indicate that the most powerful factor discriminating these multiple-risk groups from those not at high risk was whether or not individuals had been deployed to the Gulf. Investigations by Wolfe and Sutker also addressed the prevalence of somatic symptoms or symptoms with a bodily referent (i.e., pertaining to the body rather than to emotion) among veterans studied immediately after return to the United States and 1 year later. The associations among physical symptoms, stress, and mental health were examined in ongoing studies of Fort Devens reservists. Wolfe et al. (1994) found that women veterans with PTSD At= 40) reported significantly more health symptoms than their counterparts without PTSD (N= 153~. ~ ^'' the to~iow~ng symptoms were reported by 85-100% Of women with PTSD, in decreasing order of prevalence: trouble concentrating, depression, nervousness lack of energy, aches or pains, insomnia, headaches, loss of interests and crying easily. Those with PTSD also reported more _ . · ~. ~. 1 1 ~1 combat exposure, less current social support, and more postclep~oyment bite stressors. When all health symptoms were included in a regression model, anxiety and PTSD were significant predictors of health symptoms, but combat exposure was not. Of interest is that PTSD symptoms of emotional numbing and physiologic hyperarousal were significantly related to the number of health symptoms reported. Understanding the connection between psychological and physical symptoms and their relation to exposure to traumatic events requires consideration of the temporal ordering of events and plausible mechanisms. Given the lack of predeployment data for most troops, it is difficult to ascertain the temporal order of events. Some may have experienced traumatic events prior to deployment (e.g., rape or sexual or physical abuse), and some may have had prior illnesses with mental and physical symptoms. The DoD and DVA studies document the cooccurrence of stressors (e.g., combat exposure), PTSD- like symptoms, and physical symptoms, many of which are nonspecific. (Chapter 5 of this report addresses consideration of somatization, chronic fatigue, and multiple chemical sensitivity [MCSl). Those with PTSD after ODS/s have had more severe health symptoms (Stretch et al., 1995; Wolfe et al., 1994). One of the mechanisms proposed to link traumatic events (stressors) with symptoms is the physiologic changes induced as part of the stress response. Studies dating from Selye's classic experiments (Selye, 1956) link stress to changes in the hypothalamic-pituitary-adrenal (HPA) axis hormones, resulting in an increase of glucocorticoid production from the adrenals. Cortisol and other glucocorticoids initiate suppression of immune, metabolic, and neural defensive reactions in response to stress. Over the past two decades, physiologic alterations in response to extreme stress have been studied in humans with PTSD related to a variety of traumatic events (combat exposure, holocaust survivors, rape victims, and survivors of natural disasters). A series

HEALTH OUTCOMES 109 of studies of hormonal alterations in response to stress has demonstrated persistent biological changes after extreme stress. Findings of lower basal cortisol levels in people with PTSD are opposite to those expected as part of the normative stress response and suggest possible exhaustion of the response. In contrast, exposures of PTSD patients to a novel environmental stressor and exacerbation of comorbid conditions are associated with elevated cortisol levels. Similar responses occur to acute treatment interventions that involve "reliving" exposure to the stressor (Yehuda et al., 1993~. Other studies of PTSD related to abuse show elevated levels of norepinephrine, epinephrine, dopamine, and cortisol (Lemieux and Coe, 1995~. Studies of women who have been sexually assaulted show an increased risk of both medically explained and unexplained symptoms (Golding, 19941. Studies of civilian populations in the Gulf also demonstrate somatic responses to stressors in the war. A telephone survey of the Israeli civilian population found that 38% had somatic reactions to the orders to wear gas masks and move into sealed rooms, but only 20% had somatic responses 12 days later, suggesting adaptation to stress through time. A complex pattern of relationships between expectations about chemical warfare and somatic interactions was evident. Those with specific expectations about future chemical attacks were more likely to have a somatic reaction. Also, women were more likely to react than men. Habituation occurred quickly (Carmeli et al., 1991~. In another telephone survey conducted during the third week of the war, 28% of community residents complained about sleep: 10% had awakenings, 4.5% had difficulty falling asleep, and 13.5% experienced bosh. Women and persons with less education reported more sleep problems (Lavie et al., 19919. Although PTSD may have psychophysiological components, to date the committee has found no published studies linking PTSD with physiologic aspects of stress response or symptoms with physical referents among ODS/S veterans. Future studies elucidating the psychophysiological consequences of extreme stress and their relationship to both mental health and symptoms with bodily referents are needed. Discussion of Mental Health Issues In addition to the literature reviewed above, the committee discussed several psychosocial variables that could influence the incidence of unexplained illnesses among POW veterans. These included the role of suggestion in shaping symptoms and the role of concerns regarding health care. Also discussed were various psychiatric syndromes that include physical symptoms, PTSD, and other psychiatric outcomes, with recognition of He facts that psychiatric diagnoses are not to be made by exclusion and that they are often comorbid. Each is discussed briefly here.

110 HEALTH CONSEQUENCES OF THE PERSIAN GULF WAR Many pervasive and powerful suggestive influences have affected PGW veterans, including attention by the news media, Congress, and peers. Reports in the news media, meetings of veterans groups, and public and congressional hearings have served to draw attention to certain symptoms that may be present in PGW veterans or certain possible exposures. These are strong influences affecting patients' recall and reporting. How such forces affect the symptoms reported is not known, but this could be an important topic to evaluate. Veterans of the PGW returned to a nation in turmoil over health care and its financing. A key question for many veterans was whether DVA would provide the necessary treatment on a service-connected basis; this fear may have been put aside to some extent when legislation determined that unexplained illnesses were service connected, whereas many diagnosed diseases were not. However, this also meant that having a diagnosis did not guarantee treatment if the illness was not considered service connected, especially for reserve and National Guard troops who were left to prove the connection. Some persons suggested to the committee that certain known syndromes and physical illnesses could explain some veterans' undiagnosed illnesses. How these syndromes and illnesses may or may not be connected to purely physical versus psychiatric diagnoses is not well determined in either the scientific literature or the medical community. Some of these hypotheses are discussed later in this report. PTSD may be the most studied and expected of the war-related psychiatric illnesses, but there are others that could be manifest in less well defined ways. Our society has a greater acceptance now than in years past of stress as a cause of impairment, as well as a greater tendency to provide compensation for such impairment (e.g., disability claims, legal settlements, and medical retirements). After the Vietnam War, veterans were more forthcoming than previously in seeking treatment or claiming disability for mental, emotional, or psychophysiological distress. This trend should not be discouraged among PG veterans, whose stressors may include experiences from predeployment through combat to reentry to daily living. Finally, the committee has been concerned that a diagnosis of somatization disorder, PTSD, or other psychiatric illness not be made by default in those cases where a physical diagnosis has not yet been established. Psychiatric diagnoses require specific findings that the disorder is present. Physical illnesses are often accompanied by emotional disturbances or even gross psychopathology, and DSM-IV (APA, 1994) lists some physical illnesses that may present with psychiatric symptoms. A patient whose complaints are not accepted as legitimate over a period of time may also develop symptoms of an adjustment disorder superimposed on the underlying illness, thereby making a basic diagnosis even more difficult.

HEALTH OUTCOMES 111 WOMEN'S HEALTH STUDIES Approximately 50,000 of the troops deployed to the PG were women. Although women have been present in combat situations during earlier wars (Congress established the Nurse Corps of the Army in 1901 and of the Navy in 1908, and an estimated 5,000-10,000 women served in Vietnam), the number of women in forward combat support positions was uniquely high for the U.S. armed forces during ODS. With the exception of DoD studies conducted in- theater prior to the air war, most studies of the health of PG soldiers include women and men. Findings of these studies are summarized below, and issues related to women's experiences are enumerated along with questions that merit further consideration. The large number of women living in field conditions, working side by side with men, and doing the same work as men under "field expedient conditions" was also unique to this war. These conditions often required work (and exposures) for 16-hour days, 7 days per week, as discussed in Chapter 3. The health of women veterans is a very important issue in light of projections of a 17% increase in women veterans between 1990 and 2010, when women will represent 6.4% of the total U.S. veteran population (Sorensen and Feild, 1994~. In addition, in civilian populations, women's utilization of health services on average exceeds that of men (Nathanson, 1975), sO that the types and number of visits and the use of services could be significantly increased. In addition, women's health has been influenced by a social context in which men played the dominant roles. Issues of leadership or management and training raised by having women serve with men in combat roles also may have significant health effects. Health consequences of combat service for women, gender differences in health, and health issues related to men and women serving in combat situations are all considerations in our recommendations for further action. Health Effects of Combat Service for Women Given the novel experiences noted above and the large number of women deployed in the POW, it is important to determine whether there were health problems unique to women. Because limited data are available to answer this question, the topic requires further study. Long-term follow-up of women veterans who served in ODS is needed. Available data suggest a need for a data collection system that will capture important health services (utilization, needs, and issues) in a way that directs immediate action as well as support for future research. It has been well established in civilian populations that women are more willing to report symptoms and use health services more readily than men for symptoms and conditions that affect both sexes (Nathanson, 1975, 1980~.

112 HEALTH CONSEQUENCES OF THE PERSIAN GULF WAR Thus interpretation of health reports by veterans will require adjustment for gender differences in reporting. Reports from several studies, including the General Accounting Office (GAO) study of women, confirm that women deployed to the PG could and did accommodate to adverse conditions and, in the short term, did not present with more health and hygiene problems than men (GAO, 1993b). Data presented earlier in this chapter suggest some gender differences in symptom reporting in the DVA registry. Pregnancy and certain reproductive system problems (including malignancies such as breast, cervical, and ovarian cancer) are problems unique to women. Pregnancy presents a challenge to women's health under any circumstances and may produce unique health problems in combat situations. In the past, the military has forced women to separate because of pregnancy. Indeed, one of six women in the 1985 Veterans Administration survey of female veterans who entered the military in the post-Vietnam era reported being forced to separate from military service for reasons of pregnancy or having children (VA, 1985~. To date, there are no published reports or studies concerning women who became pregnant while deployed to the PG or the military management of these situations. Pierce (1996) identified a stratified sample of 638 women to study 2 years after the PGW and again 2 years later. The sample included active duty, guard, and reserve Air Force components; those deployed to the Gulf versus elsewhere; and those who were parents, as well as nonparents. General physical health, gender-specific health problems, emotional responses to war, and symptoms common in the unexplained illnesses were surveyed with 97% participation of those located (N= 5254. Women deployed to the PGW reported significantly more general and gender-specific health problems than those deployed elsewhere. Common health problems included skin rash, cough, depression, unintentional weight loss, insomnia, and memory problems. Active duty women reported more general health problems than those in the reserve or guard, despite their younger age. Among those deployed, women who were parents had significantly more health problems. Two years after the PGW, skin rash, depression, unintentional weight loss, and insomnia were significantly more prevalent among those deployed to the Gulf than those not deployed. Four years after deployment, skin rash remained significantly more prevalent among those deployed and cough and memory problems were more prevalent among the deployed. There were no differences in gender-specific health problems 2 years after PGW, but 4 years later, those deployed had a greater prevalence of lumps or cysts in the breasts, abnormal Pap smear results, and headaches. There was a trend for increased prevalence of herpes among those who served in the Gulf for more than 120 days compared with those who served for less than 120 days. Of those deployed to the Gulf, 24% met criteria for PTSD versus 15% of those deployed elsewhere (Pierce, 1996~.

HEALTH OUTCOMES 113 A final variety of health problems affecting women includes those producing acute episodes of symptoms (e.g., vaginitis) that indicate special personal support needs. A "field doctrine" for medically appropriate and field expedient approaches to contraception is also needed. This includes ensuring continuity while activated of medication (e.g., oral contraceptives) used prior to deployment. Gender Differences in Health Studies of the prevalence of symptoms during deployment in the POW revealed few differences between women and men of the First Cavalry Division. Men reporting to sick call were,more likely to be diagnosed with orthopedic and dermatologic disorders, and women were more likely to be diagnosed with psychiatric and optometric disorders (Hines, 1993~. In a cluster study after the occurrence of unexplained illnesses among POW veterans in Indiana, DeFraites et al. (1992) found that the prevalence of symptoms among men and women reserve troops was similar. Wolfe has reported preliminary results suggesting thnt Acme PTSD symptomatology is significantly associated with a range of self- reported symptoms in women (Wolfe et al., 1994~. Studies of gender differences in PTSD and other symptoms indicate that prior abuse, coupled with exposure to combat stress, may affect women and men differently. The need for prospective longitudinal studies of both women and men was reinforced by results of a study of PTSD symptoms and precombat sexual and physical abuse among Desert Storm veterans who attended a mental health clinic (297 veterans, including 28 women). Women veterans who reported previous abuse had higher Mississippi Scale scores, higher combat- related PTSD and nonspecific PTSD scores, and higher depression scores than those who had not been abused. Precombat-abused male veterans did not differ from the nonabused in PTSD symptoms (Engel et al., 19939. Still another study of reserve troops found no overall gender differences in PTSD indicators (Mississippi PTSD Scale, Beck Depression Inventory, and SCL-90-R) for noncombat reservists, but gender differences did appear for two specific combat units. The first unit had experienced fatalities and injuries Tom a SCUD missile attack and the second had morale problems. Thus, data to support gender differences were found only in units in which there had been combat-related trauma or impaired morale (Perconte et al., 1993~. Some investigators have found evidence of anticipatory anxiety about mobilization among Vietnam veteran women who subsequently were diagnosed with PTSD. A small sample of Army reserve nurses who anticipated mobilization during ODS experienced higher levels of anxiety than similar civilian registered nurses. Separation from loved ones and financial concerns were the greatest contributors to anxiety. Detailed and consistent information from army commands reduced anxiety (Wynd and Dziedzicki, 19929. In

114 HEALTH CONSEQUENCES OF THE PERSIAN GULF WAR another study, women who had had PTSD linked to Vietnam service experienced intensification of their distress during ODS regardless of whether they were deployed (Wolfe et al., 1992a). Aside from these reports, few have examined gender differences in health related to the war. The few reports to date have focused on the development of symptoms and PTSD. In a study of reservists (653 POW veterans and 259 stateside duty troops), Sutker et al. (1995a) examined the hypothesis that ethnic minority status and female gender were associated with greater psychological distress after war zone duty. All troops were studied within 1 year of return from the Gulf. Women reported more physical or somatic symptoms, including more headaches, lack of energy, cold or flu symptoms, and upset stomach, than men regardless of their deployment status. There were no statistically significant gender differences in depression, anxiety, or anger. Minority men reported more PTSD symptoms than white men. The small numbers of minority women in this study may have limited the ability to detect interactive effects of gender and ethnicity. Studies examining correlates of symptoms have identified combat stress, sexual harassment and assault, leadership and unit morale, and family care worries as important. One study of women and men immediately after their return from the Gulf demonstrated that both women and men experienced war zone stress related to combat exposure. No significant gender differences in the Lauder combat scores (describing the kinds of combat stressors to which people were exposed) were found, although men's scores were somewhat higher. Results of the more comprehensive ODS expanded checklist found that the three most commonly reported war zone stressors for both genders were formal alert for chemical or biological warfare, receiving incoming fire from large arms, and witnessing death and disfigurement of enemy troops. PTSD scores were higher for women (9% of women and 4% of men scored above 89 on the Mississippi Scale). Women reported more PTSD symptoms and had higher BSI scores than men. These findings suggest that women were more symptomatic during the initial postdeployment phase. Wolfe et al. (1993) suggest that this difference may be attributable to women's socialization; women's reporting symptoms more freely than men; family-related reentry stressors; prior stress, such as sexual or criminal victimization, leaving residual symptoms and greater sensitivity to new stress; and sexual harassment or assault experienced during deployment. Women with PTSD also reported more physical symptoms than men (Wolfe et al., 1995~. This finding suggests that some but not all of the physical symptomatology reported by POW women vets may be related to PTSD. Wolfe et al. (1995) found that both men and women reservists reported negative change in physical and psychological health since they returned from the PG and that more women than men reported a change for the worse in both physical and psychological health. More women had evidence of presumptive

HEALTH OUTCOMES 115 PTSD and reported health symptoms in conjunction with PTSD. Headaches, lack of energy, aches or pains, nervousness, tension, insomnia, and depression were endorsed more frequently by women than by men. Gender differences in needs for support on return home have received little attention, yet it is reasonable to believe that stresses related to separation from home and family affected women and men in different ways. According to preliminary findings by Pierce (1995) among 525 Air Force women who were activated, being a parent was associated with more health problems and depressive symptoms, and this effect was pronounced for women who were deployed in the theater. Nelson et al. (1996) conducted ethnographic research regarding family support needs associated with mobilization of the National Guard and reserve forces by interviewing 59 military members, family members (including children), commanders, and family support personnel from National Guard and reserve forces; by using participant observation in selected units; and by examining cultural artifacts such as documents, journals, and newspapers. This research involved interviewing in great detail this nonrandomly selected small group of individuals in a very open-ended fashion. Nelson found that staying connected in an uncertain environment was a universal concern that was exacerbated when units were fragmented. Living with the war was unusual for families who "watched" the war on television as if it were a staged theatrical production. Indeed, military members recounted watching themselves being attacked on television as the attack was occurring and they were putting on protective gear. Fluctuating emotions were common throughout families. Refocusing lives became the major task for the families of those who returned from the Gulf. Family support systems were regarded as helpful but not easily accessed. Moreover, the study revealed a need to train families better to cope with deployment, to improve information transmission, and to support reassimilation after the war. In particular, children needed preparation as did their caretakers. Financial issues were among some of the most difficult for families to resolve (because of the delay in receiving military pay, some families could not buy food). Nelson has indicated that although returning from the Gulf represented a vulnerable period, it also was an opportune time for helping families. Deployed women whose family issues were well managed may have had reduced anxiety or stress during deployment and reduced familial stress on return from deployment. Further emphasis by commanders of such leadership or management issues for both women and men in combat situations may alleviate some of the problems experienced in-theater and after deployment.

116 HEALTH CONSEQUENCES OF THE PERSIAN GULF WAR Health Issues Related to Men and Women Serving Together in Combat Situations In addition to assessing gender differences in health outcomes and exposures, researchers have begun to address the consequences of women and men serving together in combat. Most American military women and men carried themselves through the conflict without major problems, but it remains important to improve the training of both men and women on active duty and in the reserves or National Guard in handling these sensitive issues. As discussed earlier in this chapter, Wolfe and colleagues have begun to describe women's exposures to stressors such as physical or sexual harassment and assault during deployment. Also, Engel et al. (1993) found that precombat abuse led to higher rates of PTSD symptoms in female POW veterans and that deployed women with a prior history of abuse had greater PTSD symptomatology than those without a history of abuse. The long-term consequences of abuse prior to deployment, coupled with harassment, discrimination, and abuse during deployment and demobilization, merit further study. Sexual activity between women and men may eventuate in pregnancy and sexually transmitted disease, both preventable with access to appropriate preventive measures. Another committee of the IOM (IOM, l995b) recently reported on research needs related to the health effects of military service on women.

Next: Some Hypotheses Regarding Illnesses in Persian Gulf War Veterans »
Health Consequences of Service During the Persian Gulf War: Recommendations for Research and Information Systems Get This Book
×
Buy Paperback | $45.00 Buy Ebook | $36.99
MyNAP members save 10% online.
Login or Register to save!
Download Free PDF

In January 1995 the Institute of Medicine released a preliminary report containing initial findings and recommendations on the federal government's response to reports by some veterans and their families that they were suffering from illnesses related to military service in the Persian Gulf War.

The committee was asked to review the government's means of collecting and maintaining information for assessing the health consequences of military service and to recommend improvements and epidemiological studies if warranted. This new volume reflects an additional year of study by the committee and the full results of its three-year effort.

  1. ×

    Welcome to OpenBook!

    You're looking at OpenBook, NAP.edu's online reading room since 1999. Based on feedback from you, our users, we've made some improvements that make it easier than ever to read thousands of publications on our website.

    Do you want to take a quick tour of the OpenBook's features?

    No Thanks Take a Tour »
  2. ×

    Show this book's table of contents, where you can jump to any chapter by name.

    « Back Next »
  3. ×

    ...or use these buttons to go back to the previous chapter or skip to the next one.

    « Back Next »
  4. ×

    Jump up to the previous page or down to the next one. Also, you can type in a page number and press Enter to go directly to that page in the book.

    « Back Next »
  5. ×

    To search the entire text of this book, type in your search term here and press Enter.

    « Back Next »
  6. ×

    Share a link to this book page on your preferred social network or via email.

    « Back Next »
  7. ×

    View our suggested citation for this chapter.

    « Back Next »
  8. ×

    Ready to take your reading offline? Click here to buy this book in print or download it as a free PDF, if available.

    « Back Next »
Stay Connected!