Haley (2003) found an excess incidence of ALS among deployed veterans in comparison with the expected incidence based on US vital statistics. Similar to the first Horner et al. study (2003), this analysis spanned a short war or postwar period from 1991 to 1998 only. In the first half of that period an increased incidence was not apparent; from 1995 to 1998, the incidence more than doubled (standardized mortality ratio [SMR] 2.27, 95% CI 1.27-3.88). Although the study used passive and active means of case ascertainment similar to those of Horner et al., it differed in several key aspects: it restricted cases to those below the age of 45 years (instead of all ages); it used 8 years of follow-up (instead of 10 years), and it used for comparison the age-adjusted rates from US mortality statistics (instead of the age-adjusted rates in nondeployed veterans). Use of mortality statistics for the general population to estimate an “expected” incidence is a major limitation (Armon et al., 2004), since this assumes that the case ascertainment methods for the comparison population are similar to those for the deployed military population when deriving SMRs. The case ascertainment methods used by Haley (2003) are not comparable to those in the general population and may have overascertained cases among veterans.
Several US and UK mortality studies have not found an excess risk of ALS, but they did not have sufficiently long follow-up or sufficiently detailed methods (DASA, 2005; Kang and Bullman, 1996; Macfarlane et al., 2000). Recently, the original Kang and Bullman (1996) mortality study has been updated (Barth et al., 2009) and included data through December 2004. This study did not find any increase in ALS mortality in Gulf War veterans compared to nondeployed veterans (adjusted RR 0.96; 95% 0.56-1.62), but these results were based on less than half the number of ALS cases identified in the primary study by Horner et al. (2003) (23 cases in deployed and 38 in nondeployed veterans). A hospitalization study (Smith et al., 2000) also found no difference in relative risk of ALS (RR 1.66, 95% CI 0.62-4.44), but the authors acknowledge that they had too few cases to make valid comparisons between deployed and nondeployed veterans. The study was also limited by inclusion of only active-duty military personnel and only 6 years of follow-up. Nicolson and colleagues, studying eight Gulf War veterans with ALS and two other comparison populations, found that ill Gulf War veterans had the highest frequency of systemic mycoplasm infections (Nicolson et al., 2002). Although the authors suggest that mycoplasma might be involved in the pathogenesis or progression of ALS, insufficient information was given regarding the selection of cases and controls to evaluate bias in ascertainment.
Horner et al. (2008) extended their follow-up for 1 year to December 2001 and investigated temporal patterns of ALS occurrence. Among all 2.5 million military personnel on active duty during the 1991 Gulf War a total of 124 ALS cases were confirmed; 48 of these cases occurred among those deployed, while 76 cases were found among the nondeployed; the percentage of young onset cases (< 45 years of age) was similar in both groups (69% vs 64%). The main increase in ALS cases in this study occurred within the decade following the war, that is, prior to 1999. The authors proposed that Gulf War-specific neurotoxins may have caused the short-period increase in ALS rates after the war among deployed military personnel. The results for nondeployed troops—found as in the original study to have experienced lower rates of ALS than the western Washington State population—contrast with another high-quality study that reported increased rates among men who had served in the military or had been deployed to a