The government's research program includes epidemiologic studies and toxicologic evaluations of Gulf War risk factors. Epidemiology measures the occurrence of disease in human populations and the factors that influence their occurrence, severity, and outcome. Epidemiologic studies are particularly useful at the early stages of an investigation when the exact nature of a public health problem is unclear or poorly understood. They are also essential to determining whether an array of symptoms occurs more frequently in a particular population.
The Legionnaires' Disease outbreak in 1976 is a good example of epidemiology's role in investigating disease. The Centers for Disease Control and Prevention (CDC) investigated patterns of disease outbreak and transmission to determine that the cause was the previously unknown bacterium, Legionella pneumophila, which is found in a variety of water sources. This work laid the foundation for effective treatment of the disease using appropriate antibiotics. In this example, CDC was able to establish a clear link between a specific disease and a specific microorganism.
In comparison, epidemiologic investigations of chronic diseases that might be due to multiple factors often can be less clear cut. Nevertheless, epidemiologic studies will be crucial for helping us understand the impacts of service in the Gulf War on the health of veterans today.
Toxicologic studies of the risk factors encountered during service in the Gulf War are also essential elements of the government's research program. Suspected risk factors include:
Toxicologic studies use animal and other models to assess indirectly the possible health effects of Gulf War risk factors. Toxicologic studies might be the only means available to evaluate some hypotheses about the effects of certain risk factors on Gulf War participants.
This Committee is not the first external review body to assess federally funded research on Gulf War veterans' illnesses, nor is it the only current effort. The IOM's Medical Follow-Up Agency ( 6 ) has an ongoing evaluation of the research program, as does the VA Persian Gulf Expert Scientific Committee. We are making every effort to coordinate our activities with these other groups while maintaining the necessary independence.
Many other groups, including the DSB Task Force ( 3 ) and the NIH Workshop Panel ( 4 ), have made findings and recommendations concerning the importance of various risk factors (e.g., oil fire smoke, or infectious diseases) in the government's research scheme. We will take into consideration these findings and recommendations as we proceed with our own analysis over the next 10 months, but our Committee will independently assess health risk factors in the context of our review of the government's research efforts on Gulf War veterans' illnesses.
For this interim report, we have focused on two areas: the major epidemiological studies and the Persian Gulf Registry of Unit Locations.
The Committee initially has focused on epidemiologic studies being planned or carried out in fairly large populations of Gulf War veterans ( Table 1 ). Our analysis encompassed the following questions:
Study Design. Epidemiologic studies to examine the health status of Gulf War veterans face several different methodological challenges. Some challenges are common to any epidemiologic study, but others are specific to the circumstances of the Gulf War.
Lack of a case definition. Typically, epidemiologic studies measure the occurrence of a specific disease in populations. Researchers often have a good sense of how that disease manifests itself, how established epidemiologic methods can be applied to measure the frequency and/or severity of the disease in a group of people, and how to search for and interpret linkages with potential risk factors. With the Gulf War veteran population, however, this assessment is more difficult because no specific disease has been defined as the source of reported health problems. Commonly reported symptoms include fatigue, joint or muscle pain, headache, rashes, memory loss, abdominal pain, diarrhea, sleep disturbances, or difficulty in concentrating. The array of illnesses reported by Gulf War veterans has become popularly known as "Gulf War Syndrome."
External review groups that have examined the existence of a syndrome specifically related to the Gulf War experience conclude that a single, coherent syndrome cannot be defined, even though many illnesses reported by veterans might be attributable to Gulf War service. For example, the April 1994 NIH Workshop Panel found that no single disease or syndrome is apparent, but rather found evidence for multiple illnesses with overlapping symptoms and causes ( 4 ). The NIH panel concluded it was impossible to establish a single case definition at that time, and that instead, "an evolving case definition might be more appropriately used in developing a research strategy." At this stage, the Committee concurs: no case definition derived in a single population should be applied widely in all studies.
To better understand the illnesses described to date requires more information about the symptoms and their occurrence in the Gulf War veteran population. To this end, the NIH panel made the broad recommendation that DOD and VA establish a more accurate estimate of symptom prevalence. VA has responded to this recommendation in part by launching the "National Health Survey of Persian Gulf Veterans and Their Family Members." This survey is intended to help measure the occurrence of medical problems reported by veterans, and thereby assist both scientific and clinical efforts. Preliminary results of the study should be available in Fall 1996 and final data available in 1998.
Likely low response rates. In an ideal epidemiologic study of Gulf War veterans' illnesses, each participating veteran or active duty service member would have a thorough physical examination, face-to-face interview, and medical record review to determine health status and exposure history. Carrying out such complete examinations in a large population, however, is costly and time consuming. Thus, many planned or ongoing studies use postal or telephone surveys to gather data.
Postal surveys, while significantly less expensive than face-to-face interviews, frequently have low response rates. Standard epidemiologic principles caution that if completed surveys are received from less than about 70 percent of the surveyed population, it is difficult to draw conclusions from the study because the responses of a substantial proportion of the sample remain unknown. Nonrespondents can differ from respondents in many important ways, and absence of information from 30 percent or more of a population can introduce biases in the results. For example, veterans with health problems might respond at a higher rate than veterans without problems. To address this issue, studies include efforts to measure the characteristics of a sample of those people who did not respond to determine how they differ from survey respondents. These efforts are necessary, but might not be sufficient to determine the extent and effect of bias from low response rates. Since several of the large epidemiologic studies-including VA's National Health Survey-rely on extensive postal surveys, low response rates could jeopardize some of the major epidemiologic research underway on Gulf War veterans' illnesses.
Selection of appropriate study and comparison populations. Early studies to explore illnesses in Gulf War veterans focused specifically on volunteer respondents from specific units or groups in which Gulf War-attributed symptoms first were reported. Though these studies were useful in characterizing the illnesses in these groups and ruling out some hypotheses, their results cannot be generalized to the Gulf War veteran population as a whole. To obtain data about an entire group, data must be collected from everyone in that group or from a randomly selected sample of the whole population. Collecting comprehensive health data from the approximately 697,000 troops who participated in Operations Desert Shield/Desert Storm is impossible. Hence, to obtain an accurate assessment of the health status of Gulf War veterans, researchers must collect information from a representative sample of that entire group.
Current epidemiologic studies are designed to gather information from more broadly representative samples of the entire Gulf War veterans population. For example, VA's National Health Survey targets 15,000 military personnel who participated in Operations Desert Shield/Desert Storm. The sampling is being carried out so that the military branch and unit status of those surveyed represents the Gulf War veteran population, with extra sampling of women and reserve/guard veterans so that the sample of these relatively smaller groups will be large enough to draw useful conclusions. A second VA study of Gulf War veterans intends to capture information on mortality in virtually all Gulf War veterans and a comparison veteran population to provide as complete information as possible on comparative rates and causes of deaths.
The ability to draw conclusions about whether Gulf War veterans are experiencing more or different health problems than expected clearly depends on the ability to identify a suitable comparison population. Worker populations are characteristically healthier than the general population, because people with serious health problems are less likely to be in the work force; workers thus would typically be compared to another similar worker population. Moreover, active duty military personnel are likely to be healthier than typical U.S. workers because of the physical demands of the military. Hence, it would be inappropriate to compare the health of Gulf War veterans to the general civilian or specific civilian working populations. The ideal comparison population would be identical to the Gulf War veteran population in every way except for deployment to the Gulf region.
To address this important issue, the Defense Manpower Data Center (DMDC) constructed a roster of all people assigned to military units that served in the Gulf area and also compiled a list of about 1.5 million troops who served in the military at the same period as the Gulf War but who did not serve in the Gulf War theater. This group of 'era veterans' still might not be the ideal control group if it differs from troops deployed to Southwest Asia in ways for which researchers cannot adjust-e.g., service members deployed to the Gulf War could have been the most physically fit. Nonetheless, era veterans are considered the best group for comparison to Gulf War veterans, and this roster will be used as the comparison population in VA's National Health Survey and other studies.
Self-reporting health outcomes and exposure. As noted by DOD's Armed Forces Epidemiology Board, self-reporting of health outcomes and exposures is a major and significant limitation of all current epidemiologic studies of Gulf War veterans. In most ongoing studies researchers ask veterans to recollect or self-report any health problems-depending on the case, this recollection can amount to a self-diagnosis of one's own conditions. To some extent, self-reported health problems can be validated by cross-checking with other sources of information, such as hospital records, state disease and birth defects registries, or clinical examinations of veterans. The government's current epidemiologic studies stand to gain tremendously by their planned physical examinations and medical record searches to validate self reports on health problems for a subset of study participants.
In contrast, self-reported exposure estimates are much more difficult to validate by other external sources. Epidemiologists recognize that relying on self reporting to determine exposure information carries problems of possible biases in recollection.
Expectations of current studies. Current major epidemiologic studies will compare health problems experienced by large subsets of Gulf War veterans. These studies are not likely to detect a small veteran subpopulation that has health problems because of unique exposure situations. Such questions cannot be effectively addressed until testable hypotheses are developed about the specific smaller groups that might be expected to have those greater risks. The Persian Gulf Registry of Unit Locations database could help identify potentially higher-exposure subgroups. Other ongoing studies might clarify the absence or presence of susceptibility factors relevant to particular exposures.
External Review. The methodological issues just mentioned are only a few of the many important limitations and challenges faced in the design and execution of epidemiologic studies of Gulf War veterans. Other critical issues include the adequacy of population sample size, effectiveness of survey/questionnaire design, and limitations in the use of medical records or registries for gathering or validating health information. Because of resource constraints, difficult choices frequently must be made in designing such studies. For example, more questions on a survey questionnaire allow for more specific information to be gathered, but the increased length often results in fewer people completing the questionnaire.
External scientific review is invaluable as these issues are faced and tradeoff decisions are made. No matter how large and qualified an individual study team, any study benefits from external and independent perspective and input throughout research design, data collection, and analysis. External scientific review of the current major epidemiologic studies has ranged from nonexistent, to one-time review of protocols, to standing scientific advisory panels with an ongoing role in the design and execution of the studies. The responsiveness of principal investigators to external review varied as well.
A public advisory committee also has proved useful to at least one study now underway. Iowa researchers funded by CDC have a science advisory committee and a 20-member public advisory committee composed of Gulf War veterans, spouses of veterans, and representatives of veterans service organizations. The Iowa researchers report that the public advisory committee has played an important role in simplifying the wording of the study questionnaire, in stressing the need to safeguard the confidentiality of questionnaire responses, and in disseminating information about the study to veterans groups.
Study Coordination and Oversight. External scientific review of individual studies can greatly improve quality, but it cannot ensure that the overall research agenda encompassing all of the individual studies is adequately served. In fact some studies deemed not useful and potentially misleading by external reviewers have been continued. Coordinating the government's epidemiologic studies and other studies being funded within the larger context of research into Gulf War veterans' illnesses can ensure priorities are set, appropriate research questions are addressed, and unnecessary duplication is avoided. Since resources are limited, judgments must be made about which research endeavors can provide the most useful information.
The Persian Gulf Veterans Coordinating Board, comprising the Secretaries of Defense, Health and Human Services, and Veterans Affairs, was established in January 1994. It is charged with providing direction and coordination on health issues related to the Gulf War. The Coordinating Board's Research Working Group, which has primary responsibility for Gulf War-related research, describes its responsibilities as:
The major contribution of the Research Working Group has been the publication of A Working Plan for Research on Persian Gulf Veterans' Illnesses in August 1995 and a series of meetings at which principal investigators have discussed details of their research. In testimony before the Committee, there has been no clear indication that this group has had a significant impact on the existing major epidemiologic studies, even in cases where external reviews have suggested that certain studies would not make a positive contribution to the overall research effort.
Several previous external reports that reviewed various portions of federally funded research on Gulf War veterans' illnesses recommended overall, centralized coordination. In 1995, the IOM Committee on Health Consequences of Service During the Persian Gulf War ( 6 ) wrote that studies completed before 1994 on Gulf War veterans' health issues had been piecemeal, and recommended that VA and DOD determine the specific research questions that need to be answered and design epidemiologic studies accordingly. IOM specifically recommended that VA and DOD "collaborate to obtain populationbased and controlled data on symptom prevalence, health status, and diagnosed disease." Our own findings and recommendations concerning the major epidemiologic studies reinforce this view.
A good example of the need for coordination and oversight is the debate about standardizing study questionnaires. Most experts agree that it is not desirable to make all questionnaires from all studies identical. Nevertheless, when specific questions from different studies are aimed at obtaining the same information, then consistency offers the advantage of allowing future inter-study comparisons. The Office of Management and Budget (OMB) initiated this type of coordination for epidemiologic studies of Gulf War veterans. Under the authority of the Paperwork Reduction Act, OMB has required that the principal investigators of several studies revise their individual questionnaires to reflect a common, core set of questions on symptoms and conditions.
OMB's role in enforcing some questionnaire standardization is controversial. Some researchers believe it is too early in the understanding of the Gulf War veterans' illnesses to stifle independent experimentation and innovation in the research process by dictating some portion of the questionnaires. The Research Working Group has not played an active role in the debate.
The Gulf War theater posed significant barriers for collecting exposure data on the various risk factors that might effect the health of service members months or years later. Plainly put, the military's goal is to successfully carry out the mission of the war, not to collect research data. Nevertheless, the lack of good exposure data for Gulf War veterans has certain consequences today for evaluating the long-term health impacts of service in Southwest Asia.
For example, DOD has precise information about the pesticides shipped to the Gulf, but information about who used them and when is fragmented at best. Such information might have allowed an estimation of individual exposures. The only available information to estimate individual exposures for most Gulf War risk factors is recollections by veterans, and these will be difficult or impossible to validate.
There is no way to compensate fully for the lack of good exposure data related to Gulf War service. As a consequence, it will be difficult to link health problems discovered in epidemiologic studies with specific exposures. In this, the country has not avoided repeating the mistakes of the past. For example, the 1994 IOM report Veterans and Agent Orange found the numerous health studies on Vietnam veterans were hampered by the lack of good individual data about exposure to dioxin or herbicides, though even without such data epidemiologic studies were useful for evaluating veterans' health status. Nevertheless, to evaluate the potential impact of a specific Gulf War related risk factor, quantitative information about the size and timing of the exposure is crucial, and few data are generally available for many risk factors. New draft guidance under development by DOD indicates the issue of better exposure records and medical surveillance of troops has become a higher priority for future conflicts. The Committee will evaluate these efforts in the coming months.
Some investigators anticipate DOD's Persian Gulf Registry of Unit Locations database will provide exposure information that will be unavailable from any other source. In December 1991, Congress required that DOD produce a database on those who served in the Gulf War (Public Law 102-190). Although Congress mandated the database in response to concerns about smoke from Kuwait oil fires, the database is likely to be relevant to other possible exposure assessments.
The Environmental Support Group (ESG) of DOD intends for the database to provide personnel and unit data for research purposes; the database also establishes the location of units, which may be useful in evaluating future health claims. ESG hopes to produce a database with at least one location coordinate for each unit (company level) from January 15, 1991, to the time that unit left the Gulf region. Both the IOM Committee ( 6) and this Committee have commented that the Persian Gulf Registry of Unit Locations database needs to be completed as quickly and accurately as possible. However, DOD projects the database will be available to researchers by April 1996 at the earliest-over a year after IOM's initial call and the original goal set by DOD.
Though location data will be at the resolution of an individual unit and will not be specific to an individual, it could prove an important resource for exposure information on Gulf War troops. For example, a time series of geographic location of troop units might be useful to distinguish between units that were in the vicinity of Kuwait oil fires from troops that were not in the vicinity. Such data would be useful in epidemiologic studies evaluating the possible health effects of exposure to Kuwait oil fire smoke.
Some investigators hope to use the Registry of Unit Locations database to assess other exposures. For example, it might be possible to determine which troop units were in the vicinity of depleted uranium (DU) weapons (e.g., tank battles). Such a population might be expected to have a greater potential exposure to DU than other troop units. Investigating possible disease clusters is another potential for this database. Some investigators are intrigued with the possibility of looking for clusters of certain diseases or symptoms among specific units-e.g., clusters in units closest to damaged Iraqi chemical weapon depots.
The Registry of Unit Locations is unlikely to be informative for most missing exposure data. It will reveal little about exposure to pesticides, pyridostigmine bromide, vaccines, or other health risk factors because little information exists about how, when, or by whom such agents were used during the Gulf War. The limited resolution at the unit level means exposure information for a specific service person might be prone to error; individuals did not always physically remain with their units.