A limitation of this study is that although information about some risk factors for suicide were avalaible, information about other risk/preventive factors were lacking. Indeed, this study has a lack information on mental illness (e.g., specific diagnoses, severity), previous suicide attempts, substance abuse/dependence, alcohol abuse/dependence (a distinct variable from alcohol intake itself), firearm availability, sexual orientation, impulsivity-aggressiveness, emotional state (such as hopelessness or anhedonia), social and family support, and spirituality or beliefs (incuding moral objections to suicide) (Nock et al. 2008; Hawton and van Heeringen 2009). Nevertheless, even if complete and detailed information on all the relevant factors were available, the potential confounding is very laborious, because of the complex and reciprocal relation between smoking behavior, psychosocial stress, and mental health. Thus, neither this nor previous investigations can determine causality. Some insight on causality could perhaps be derived by investigating whether trends in smoking behavior at the population level are correlated with opposite trends in suicide rates. Finally, because the participants were predominantly non-Hispanic white health professionals, the generalizability of the observed associations may be limited to similar populations.
