Editor—Professors Graham Dunn and Brian Everitt have pointed out to us in a personal communication that a different statistical method would have been more appropriate in our paper on the effects of multiple vaccines on the health of Gulf war veterans.1 In that paper we assessed the effect of multiple vaccines received either before or during deployment to the Gulf as two separate exposures. We found that vaccines received before deployment were not associated with most of the outcome measures we looked at, whereas vaccines received during deployment were. We performed the analysis in this way as we were testing a specific a priori hypothesis based on a theory put forward by Rook and Zumla.2 However, we did not directly compare these two exposures.
We have performed a series of further analyses in which we entered the two separate exposures (vaccines received before deployment and vaccines received during deployment) in the same model and calculated the differences in regression coefficient for each of the outcome measures, with 95% confidence intervals (table). None of these differences was significant. The 95% confidence intervals of four outcome measures—“multisymptom illness” as classified by the Centers for Disease Control and Prevention, fatigue, post-traumatic stress reaction, and “health perception” (a subscale of the SF-36 questionnaire)—are asymmetrical around zero, which is consistent with a difference in effect between multiple vaccines received during deployment and vaccines received before deployment.
Our paper showed large differences in the effects of multiple vaccines on health according to whether they were received before or during deployment. The results of this alternative method indicate that there may be no difference. However, these findings do not necessarily contradict those reported in our paper: an effect might be detected in the revised analysis only with a larger sample. It is reasonable to conclude that multiple vaccines received during deployment have a greater effect than those received before deployment, but we have not shown that multiple vaccines received before deployment are without risk. We say again that these are preliminary findings that require replication and also that they have no bearing on public vaccination programmes. Our results still indicate, as did our earlier paper,3 that there is a link between multiple vaccines and illness in veterans of the Gulf war.
Table.
Differences in regression coefficients between vaccines received before and vaccines received during deployment to the Gulf. Positive values indicate greater effect of vaccines received during deployment
| Outcome measure | Difference in regression coefficient (95% CI) |
|---|---|
| CDC syndrome* | 0.18 (−0.02 to 0.39) |
| Fatigue | 0.17 (−0.03 to 0.37) |
| General health questionnaire | 0.02 (−0.18 to 0.22) |
| Post-traumatic stress reaction | 0.29 (−0.01 to 0.59) |
| SF-36† physical function | 0.65 (−0.58 to 1.88) |
| SF-36† health perception | 1.56 (−0.70 to 3.82) |
Multisymptom illness as classified by the Centers for Disease Control and Prevention.
Short form 36, a health status questionnaire.
References
- 1.Hotopf M, David A, Hull L, Ismail K, Unwin C, Wessely S. Role of vaccinations as risk factors for ill health in veterans of the Gulf war: cross sectional study. BMJ. 2000;320:1363–1367. doi: 10.1136/bmj.320.7246.1363. . (20 May.) [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Rook GAW, Zumla A. Gulf war syndrome: is it due to a systemic shift in cytokine balance towards a Th2 profile? Lancet. 1997;349:1831–1833. doi: 10.1016/S0140-6736(97)01164-1. [DOI] [PubMed] [Google Scholar]
- 3.Unwin C, Blatchley N, Coker W, Ferry S, Hotopf M, Hull L, et al. Health of UK servicemen who served in the Persian Gulf war. Lancet. 1999;353:169–178. doi: 10.1016/S0140-6736(98)11338-7. [DOI] [PubMed] [Google Scholar]