Epidemiologic Risk Factors for Suicide and Attempted Suicide in the U.S. Air Force: Using Administrative Data Systems and Multiple Cause of Death Information to Improve Prevention Policy

摘要

In a series of retrospective studies, the investigator used data from personnel, hospitalizations, and mortality administrative information systems to elicit epidemiologic risk factors for suicide and suicide attempts occurring during 1990-1998. The reliability of the official source of military mortality information was compared against the newly-developed USAF Mortality Registry (AFMR) which uses multiple cause of death (certificate) information. The AFMR was fotmd to be the most valid and reliable source of mortality information, largely due to the '120-day retiree' sub-cohort missed by the official data. The AFMR was subsequently used to select suicide completers into a nested case- control study in which risk factors for both suicide and attempted suicide were investigated. Exposure information was drawn from existing administrative data. Completers and attempters had several independent risk factors in common, but the socio-demographic risk factors were altogether dffferent. Mental health- related hospitalizations for prevalent disorders had the strongest association with suicidal behaviors. Injury-related hospitalizations for 'accidents' were positively correlated with suicidal behaviors, suggesting intent misclassification, particularly in poisonings. Deployment and overseas assignment screening appeared ineffective in identifying personnel at risk of suicidal behavior. Suicide rates generally increased with increased levels of in-patient utilization. An all-military mortality registry, using the AFMR as a model, is needed. The opportunity to screen for suicidal risk factors should be taken during medical encounters. Pre-deployment and overseas assignment screening should incorporate more sensitive screening tools. Intent misclassification of injury hospitalizations is a threat to both public health policymaking and clinical interventions. (209 pages excluding CV).