Subsystem and deaths reported to the Social Security Administration.17 It compared deployed veterans with a similarly sized cohort of veterans who did not serve in the Gulf War. A further analysis extended the mortality data through 1997 with no change in the results (Kang and Bullman, 1999).

A second mortality study of active duty military personnel focused exclusively on the Gulf War period. This study compared noncombat mortality rates among troops stationed in the Gulf War versus troops on active duty elsewhere. There was no excess noncombat mortality in deployed veterans, except for unintentional injury (due to vehicle accidents and other causes; Writer et al., 1996).

The principal limitation of published mortality studies is the short duration of follow-up observation. More time must elapse before excess mortality would be expected from illnesses with long latency (e.g., cancer) or a gradually deteriorating course (e.g., multiple sclerosis).18 An ongoing, long-term study of all U.K. veterans of the Gulf War in relation to contemporaneous controls is assessing the incidence of cancer and all-cause mortality (Cherry and Macfarlane, 1999).

Hospitalization Studies

The risk of hospitalization was the subject of two large studies of active duty personnel discharged from DoD hospitals before and after the Gulf War. The first study, compared almost 550,000 Gulf War veterans with almost 620,000 nondeployed veterans and found no significant and consistent differences in hospitalizations after the war (Gray et al., 1996). Before the Gulf War, from 1988 to 1990, those subsequently deployed to the Gulf were at lower risk of hospitalization than their nondeployed counterparts, probably due to the healthy-warrior effect. In order to permit valid “before-versus-after” comparisons, the investigators used statistical methods to remove bias introduced by the healthy-warrior effect (also “healthy-worker effect”; see Chapter 3).

A second hospitalization study reexamined the same data set of active duty personnel discharged from DoD hospitals to search for excess hospital admissions because of unexplained illnesses. The authors reasoned that the first study


The degree of completeness using these record systems was assessed by a validation study using state vital statistics data. Ascertainment was estimated at 89 percent of all deaths in the Gulf War cohort and comparison group.


Critics assert that the mortality study by Kang and Bullman (1996) made errors in calculating confidence intervals around mortality rates and did not adequately account for the “healthy-warrior effect” (i.e., the possibility that troops mobilized to the Gulf War were healthier than nondeployed troops, thereby biasing the study toward not finding a mortality difference) (Haley, 1998). The study authors disagreed with this assertion and demonstrated that other statistical techniques, recommended by Haley, had negligible impact on their confidence intervals (Kang and Bullman, 1998). To counter the charge of selection bias by Haley (1998), the study authors point out that effects of any potential selection bias are minimal because they found no differences in mortality risk between troops mobilized to sites other than the Persian Gulf and troops not mobilized at all (Kang and Bullman, 1998).

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