Background: Smokers have excesses of injuries. Randomized, controlled trials (RCT) could assess whether cigarette smoking (smoking) causes (and smoking cessation (cessation) prevents) those injuries. We analyzed injury, accident, suicide, and homicide deaths as secondary endpoints in RCT that induced cessation.
Methods: RCT were sought via MEDLINE, bibliographies, the CDC Smoking and Health database, and the GLOBALink Tobacco listserver. RCT were included if (1) net (intervention minus control) cessation totaled over 10% of all intervention smokers and (2) injury data were available. Relative risks (RR) were summarized by fixed effects modeling.
Results: Three trials were located (the Lung Health Study, MRFIT, and Whitehall RCT). In-trial annual point prevalence cessation averaged 41, 46, and 59% in the intervention group smokers versus 17, 22, and 28% in the respective controls. Intervention (cessation) was associated with pooled injury RR of 0.65 within the trials (95% confidence interval (CI) 0.36-1.19) and 0.68 (95% CI 0.43-1.09) with additional follow-up in the Lung Health and MRFIT populations.
Conclusions: The smoking/injury death association is of borderline statistical significance in currently available RCT data. It modestly supports the significant forensic, cohort, case-control, cross-sectional, challenge-rechallenge, in vitro, and animal data suggesting that smoking may cause injury. Direct tests of the hypothesis could strengthen this inference.
Copyright 1999 American Health Foundation and Academic Press.