participants were older, had more illnesses, and were more likely to be unemployed than nonparticipants. The authors believe that such biases were avoided because participants were demographically representative of the entire battalion and because retired veterans were included in the study. Nevertheless, nonparticipants were less likely to report having had a serious illness since the war and were more likely to be employed. Moreover, the average age of this group of Seabees was greater than that of most active duty units, suggesting that study subjects might not accurately reflect the nature of illnesses in other military units.
At least 25 percent of ill veterans in the battalion studied had symptoms that the authors believe suggested generalized neurological injuries, mainly combinations of damage to the brain or brain stem (e.g., cognitive and vestibular dysfunction), the spinal cord and peripheral nervous system (e.g., paresthesias of the extremities, muscle pain and weakness, joint pain, urinary incontinence), and the autonomic nervous system (e.g., chronic diarrhea).
There is concern that the survey sample used by Haley and colleagues was small, increasing the potential to generate spurious results (Gray et al., 1998). Another potential source of bias is the numerous medical examinations and media contacts of study subjects before the survey was conducted and the reliance on self-reports of symptoms and adverse responses to PB that occurred many years earlier (Gray et al., 1998). The study population was a reserve naval command, whose members were often employed full-time in nonmilitary careers, with occupational exposures and subsequent confounding health risks, and thus may not be representative of the general population of Gulf War veterans.
The most important of the Haley reports with regard to an association with PB exposure is described as “a cross-sectional epidemiologic study” (Haley and Kurt, 1997). This study of the association between self-reported wartime exposures and self-reported symptoms in a small proportion (41 percent) of the 606 members from the RNMCB-24 relies heavily on the syndromes developed by factor analysis of symptoms reported by these same veterans (Haley et al., 1997b). The survey instrument used by Haley and colleagues to elicit self-reported exposures and symptoms in participating members of the battalion was developed by Haley and colleagues and pretested on five Gulf War veterans. After revision, the survey instrument (exposure and symptom booklet) was again pretested on five additional Gulf War veterans. It is important to note that associations reported by Haley and Kurt (1997) are based on comparisons of responses by ill and non-ill Gulf War veterans and do not include comparisons of responses from nondeployed veterans.
The authors report that the prevalence of syndrome 1 (impaired cognition) was greater among veterans who reported wearing flea collars during the war (5 of 25, 20 percent) than in those who never wore them (7 of 229, 3 percent; RR [relative risk] 9.7 [3.0–24.7], p < .001). Syndrome 1 was not associated with subjects having taken PB or reporting adverse effects from PB.
The prevalence of syndrome 2 (confusion–ataxia) was eight times greater among veterans who reported having experienced a likely CW attack. The prevalence of syndrome 2 was not higher in people who reported having taken a