to the Gulf War Theater during the time of the Khamisiyah demolition. Exposure status was determined by whether active-duty military were within the plume area defined by meteorologic-dispersion modeling, according to DOD's revised modeling in 2000 (Rostker 2000), and whether they were within a military unit determined, according geographic information systems data, to have been exposed during a 3-day period at concentrations set by CDC for the “general population limit”(GPL), below which no symptoms are expected. The GPL was adjusted because of the briefer duration of troops’ potential exposure. Troops were considered exposed if concentrations were above the adjusted GPL of 0.000003 mg/m3 for sarin and 0.00001 mg/m3 for cyclosarin.
Investigators studied hospitalization diagnoses from 15 ICD-10 categories, including “circulatory system diseases” and “endocrine, nutritional, and metabolic diseases”. On the basis of Cox's proportional hazard modeling, only one category of disease circulatory—system disease—was related to exposure (RR 1.10, 95% CI 1.05 -1.16), after adjustment for other variables in the model. The increased RR was due to one of 10 cardiac diagnoses: cardiac dysrhythmias (RR 1.23, 95% CI 1.04-1.44). While acknowledging that the finding could have been due to chance, the investigators concluded that the excess in dysrhythmia was "small in comparison with potential observational variability, but the findings are provocative and warrant further evaluation." The authors identified study limitations as use of hospitalizations because outpatient data were unavailable, restriction to DOD hospitals (because of the availability of computerized records), and restriction to hospitalizations of those who remained on active duty after the war. It also was not possible to adjust for confounding exposures. A further problem is the uncertainty in the exposure model (see Chapter 2). See Table 5.12 for a summary of the primary papers reviewed for cardiovascular disease and diabetes.
Several large-scale epidemiologic studies included self-reported cardiovascular outcomes or diabetes with new onset after the Gulf War. The committee regarded studies that relied on self-reports of cardiac outcomes as secondary studies. None of the studies, except the VA study, verified the diagnosis with physician records.
No postwar differences between deployed and nondeployed were found in two population-based cohort studies. Australian veterans deployed to the Gulf War were as likely as nondeployed veterans to report physician-diagnosed high blood pressure (OR 1.2, 95% CI 0.9-1.6) (Kelsall et al. 2004a). Similarly, Kansas veterans were as likely as their nondeployed counterparts to report physician-diagnosed high blood pressure (OR 1.24, 95% CI 0.82-1.89), heart disease (OR 1.56, CI 0.69-3.56), or diabetes (OR 1.22, 95% CI 0.45-3.30) (Steele 2000). Other studies had mixed findings. The large VA study, with its phase I questionnaire (Kang et al. 2000), found no statistical increase in physician-diagnosed stroke or diabetes in deployed veterans, but did find increases in physician-diagnosed hypertension; the difference in incidence between deployed and nondeployed was 3.84% (95% CI 3.75-3.93). That finding, however, occurred in the same cohort in which later medical evaluations found no statistical increase in hypertension (Eisen et al. 2005). Studying all Seabee commands, Gray and colleagues (2002) found no statistical increase in physician-diagnosed diabetes but did find increased hypertension (OR 1.63, 95% CI 1.36-1.95). McCauley and colleagues (2002) examined the effects of potential nerve-agent exposure on a cohort of veterans from Portland and Washington. The investigators found no statistical differences in self-reporting of physician-diagnosed diabetes, high blood pressure, or heart disease when they divided their cohort into potentially exposed and unexposed,