unexposed. The study reported an increased risk of brain cancer deaths in the exposed population (relative risk [RR] 1.94, 95% confidence interval [CI] 1.12-3.34; 25 exposed cases vs 27 unexposed cases) and there was a suggestion of a dose-response relationship with increased risk among those who were considered exposed for 2 days (6 cases) relative to 1 day (19 cases) (RR 3.26, 95% CI 1.33-7.96 and RR 1.72, 95% CI 0.95-3.10, respectively). The authors also discussed modeling exposure to smoke from oil-well fires as a confounder, and the effect estimates for exposure to Khamisiyah nerve agents remained elevated. There was no significant elevation in risk associated with exposure to oil-well fires as a main effect. Because brain cancer likely has a latent period of 10-20 years and Bullman et al. (2005) had fewer than 9 years of follow-up, the Volume 4 committee concluded that additional follow-up is needed to draw any definitive conclusions concerning the association between deployment to the Gulf War and the development of brain cancer.
The Volume 4 committee identified two studies that specifically examined testicular cancer among US servicemen during the Gulf War: Knoke et al. (1998) and Levine et al. (2005), and one study of military hospitalizations where a positive association for testicular cancer was observed (Gray et al., 1996). Gray and colleagues (1996) examined all-cause hospitalizations in DoD hospitals from August 1, 1991, through September 30, 1993, for 517,223 deployed and 1,291,323 nondeployed Gulf War servicemen. They observed an increased risk of hospitalization for testicular cancer among the deployed (standardized rate ratio 2.12, 95% CI 1.11-4.02) in the last 5 months of 1991, the period immediately after the end of deployment. However, the increased risk did not carry over into 1992 or into the first 9 months of 1993. Knoke et al. (1998) focused on the cases of first diagnosis of testicular cancer in this cohort, continuing follow-up until March 31, 1996. They observed no association with deployment status (standardized rate ratio 1.05, 95% CI 0.86-1.29). This pattern of increased incidence immediately after the war with a tapering off with time likely demonstrates a healthy warrior effect. In other words, the peak probably represented a regression to the mean after healthier people were selected for deployment and there was deferment of care during deployment. The limitations of these studies are that they were restricted to active-duty military personnel and did not include veterans who may have left the service because of poor health or those who sought treatment elsewhere.
Levine et al. (2005) conducted a pilot study matching data from the District of Columbia and New Jersey cancer-registry cases with the records of 621,902 deployed Gulf War veterans and 746,248 veterans serving at the same time as the Gulf War but not deployed. Testicular cancer cases yielded a crude proportional incidence rate (PIR) of 3.05 (95% CI 1.47-6.35) that was attenuated after adjustment for state of residence, deployment status, race, and age (PIR 2.33; 95% CI 0.95-5.70). No definitive conclusions could be made until additional registries are added.
The Volume 4 committee included results from two primary mortality studies. Kang and Bullman (2001) compared cause-specific mortality rates in the same database of Gulf War deployed veterans and nondeployed veterans used for the Levine et al. (2005) study described above. Vital status was determined using databases of the VA and the Social Security Administration (SSA). Over the follow-up period of 1991 to 1997, there were no significant excesses of overall cancer deaths or deaths from cancer at any specific site among deployed