Chronic Multisystem Illness Among Gulf War Veterans—Reply

In Reply: Dr Haley repeatedly contends that our study confirms some findings of his study¹ but also suggests that our study was flawed where our findings or interpretations do not agree with his. We would like to clarify several points.

First, our study was never intended to replicate or confirm the findings of Haley et al.¹ We are skeptical of his study findings and conclusions because of substantial study design flaws already described by others.^2- 3 Second, we used a very different approach than Haley et al for choosing symptoms to study by factor analysis. To prove a hypothesis that neurotoxic exposures were causing illness among selected Gulf War veterans, Haley et al used unstated criteria to choose symptoms from reports from various Gulf War registries. Because we had no preconceived ideas about a Gulf War syndrome, we used open-ended interviews and examinations of 59 Gulf War veterans and a survey of more than 3700 military personnel from 4 military bases to identify relevant symptoms for study. This lengthy process provided the most epidemiologically and clinically sound basis (ie, least open to bias) for identifying important symptoms among Gulf War veterans. Third, our objective as stated was to study illness in Gulf War veterans who remained in active service. Fourth, stress is a difficult factor to measure, because its manifestations and relations to other disorders are not clear. Although we found only 1 person who met formal criteria for posttraumatic stress disorder, such criteria are just 1 approach for measuring stress. A more fundamental observation is that we found multisymptom illness cases among both Gulf War veterans and, at lower levels, among their nondeployed peers. It is inescapably logical that important etiologic factors probably were common to both groups and were not unique to the Gulf War experience; such factors could include stress. Fifth, we flatly disagree with Haley that our symptom-based case definition is "incongruous" with our factor analysis results and is "invalid." The results were highly concordant, and this finding helped convince us that the factor analysis results were believable. The role of factor analysis in any study should not be overstated. The meaningfulness of symptom groupings (ie, factors) identified by factor analysis fundamentally depends on the symptoms that are selected for study. Claims that factors constitute disease syndromes should be received with skepticism unless strong supporting biological or clinical evidence is presented. Sixth, we agree with Haley that our findings do not necessarily implicate a psychological basis for symptoms reported by Gulf War veterans. However, unlike Haley, we do not have a particular etiology to champion. Given the nature of war, it remains probable that psychological factors have an important contributing role in the development of unexplained symptoms in some personnel after all wars.⁴