was significantly associated with the multisymptom syndrome among all Gulf War veterans (odds ratio [OR] = 1.2, 95% confidence interval [CI] 1.1–1.4); however, when the analysis was restricted just to veterans who had formal records of their vaccinations, the association was not significant (OR = 1.0, 95% CI 0.7–1.3). There was also a significant association between reporting biological warfare vaccination and the multisymptom outcome in the Gulf War cohort (OR = 1.5, 95% CI 1.3–1.7). Data were available in the Bosnia cohort only for anthrax vaccination, where no significant association was seen (OR = 1.5, 95% CI 0.7–2.9). An association was seen for Gulf War veterans receiving the anthrax vaccine (OR = 1.5, 95% CI 1.3–1.7). Servicemen who recalled experiencing adverse effects immediately after vaccination were more likely to have current symptoms (Gulf War cohort: OR = 2.8, 95% CI 2.4–3.3; Bosnia cohort: OR = 2.2, 95% CI 1.6–3.1). Controlling for these perceived adverse effects immediately after vaccination weakened the association in the entire Gulf War cohort between vaccination and the long-term multisymptom outcome, except for tetanus vaccination (OR = 1.2, 95% CI 1.0–1.4). Controlling for immediate short-term adverse effects after vaccination in the statistical analysis allows an examination of a possible direct association between vaccination and long-term effects. However, this statistical procedure would tend to underestimate the association between vaccination and long-term effects if short-term adverse effects were really correlated with long-term adverse effects; such a correlation is biologically plausible.

The total number of vaccinations received bore a weak but significant relationship to the occurrence of the multisymptom outcome among all Gulf War veterans, even when the cohort was subdivided into groups according to whether the subject possessed his vaccination records or not. The association still existed after controlling for the receipt of biological warfare vaccines and for experiencing side effects after vaccination (although the addition of this independent variable weakened the association). Among Bosnia veterans, no association between the number of vaccinations and the occurrence of the multisymptom outcome was observed. Although recall bias may be the reason for the significant results for individual vaccines, this is unlikely to be the case when the overall number of vaccinations is considered. Thus, the U.K. Gulf War study provides some limited evidence of an association between multiple vaccinations and long-term multisymptom outcomes. The respondents were more likely to be older, to still be in service, and to have attended the Ministry of Defence’s assessment program for Gulf War veterans with symptoms. This study was conducted through questionnaire and relied primarily on self-reports.

A recently released study (Hotopf et al., 2000) reported on a further analysis of the United Kingdom data. This study focused on the subcohort of U.K. Gulf War veterans who reported that they had copies of their vaccine records (n = 923; 28 percent of responders in the original survey). The analysis examined the vaccines received, the timing of vaccinations, and six health outcome measures (the CDC-defined multisymptom outcome, psychological distress, posttraumatic stress reaction, fatigue, health perception, and physical functioning). The ques-



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