Robert W. Haley, M.D.
Epidemiology Division, Department of Internal Medicine,
University of Texas Southwestern Medical Center, Dallas, Texas
Three widely cited, Federally sponsored studies published
in The New England Journal of Medicine purported to demonstrate that the
approximately 695,000 U.S. military veterans deployed to the Persian Gulf
War suffered no higher rates of mortality, hospitalization and birth defects
than the 1.4 million era veterans who were not deployed. These three studies
as well as the Iowa survey of Gulf War veterans reached inferences by
comparing populations of Gulf War veterans (deployed veterans) and era
veterans not deployed to the war theater (non-deployed veterans). Two
types of selection bias, however, ensured that the postwar expectation
of illness would be higher in the non-deployed group in the absence of
illness caused by the war. First, the fact that populations of military
personnel deployed to a war zone are healthier than those not deployed
("healthy-warrior effect") ensured that expected postwar rates
of morbidity and mortality would be higher in the non-deployed than in
the deployed populations. Second, events occurring in non-military hospitals
in veterans who separated from active duty after the war were excluded
from analyses comparing the rates of postwar hospitalization and birth
defects in the deployed and non-deployed groups. Evidence in the papers
indicates that veterans with war-related illness were disproportionately
separated from duty soon after the war, thus reducing the observed rates
of events in the deployed group more than in the non-deployed group. Consequently,
relative risks of death, hospitalization and birth defects near unity
were mistakenly interpreted as the absence of excess risk from deployment.
It is likely that deployed veterans have suffered excess postwar deaths
from motor vehicle accidents, suicide and respiratory illnesses and excess
hospitalizations for ill-defined conditions and mental and musculoskeletal
complaints, and excess rates of birth defects have not been ruled out.
"Keywords": Meta-Analysis; Selection Bias;
Healthy Worker Effect
No support but findings relevant to Federally supported
Gulf War veterans illnesses research
|