Cigarette smoking and completed suicide: Results from 3 prospective cohorts of American adults

Background Prior reports have indicated a potential dose-response relationship between smoking and suicide. However, this relationship is controversial. Methods This study evaluated the association between smoking and risk of death from suicide in three large-scale cohorts of U.S. men and women (n=253,033). Suicides were identified from death certificates among 43,816 men enrolled in the Health Professionals Follow-up Study (HPFS) between 1986 and 2008, 116,566 women in the Nurses' Health Study (NHS) between 1976 and 2008, and 92,651 women in the NHS II between 1989 and 2007. Information on smoking was obtained at baseline and updated every 2 years. Relative risks (RRs) of suicide were estimated using Cox proportional hazards regression models. Cohort specific RRs were pooled using random-effects models. Suicide deaths were determined by physician review of death certificates. Results A total of 457 deaths from suicide were documented. Compared to never smokers, the pooled multivariate RR (95% confidence interval [CI]) of suicide was 1.15 (0.91-1.45) for former smokers and 2.69 (2.11-3.42) for current smokers. A nonmonotonic dose-response relationship was noted between the number of cigarettes smoked per day (CPD) and suicide risk (P_trend<0.001). Compared to never smokers, the pooled multivariate RR (95% CI) was 2.59 (1.77-3.79) for those with 1-14 CPD, 2.03 (1.39-2.94) for those with 15-24 CPD, and 4.13 (2.96-5.78) for those with ≥25 CPD. Limitations Smoking was self-reported and had some degree of measurement error. Participants were not a representative sample of the U.S. population. Conclusions Results from three large cohorts suggest a nonmonotonic dose-response association between smoking and suicide risk.