Associations between substance use disorders and suicide mortality risk should be adjusted for tobacco use disorder

Abstract

The journal publishes both invited and unsolicited letters. ASSOCIATIONS BETWEEN SUBSTANCE USE DISORDERS AND SUICIDE MORTALITY RISK SHOULD BE ADJUSTED FOR TOBACCO USE DISORDERIn January 2014, Bohnert et al. [1] reported in Addiction an excess risk of death by suicide in individuals diagnosed with tobacco use disorder. Their findings derived from a cohort of the US Veterans Health Administration (VHA) records and included all individuals who received VHA services in fiscal year (FY) 2005 and who were alive at the beginning of FY 2006 (n = 4 863 086). In an unadjusted, bivariate model, tobacco use disorder was associated significantly with an increased risk of suicide [hazard ratio (HR) = 1.88; 95% confidence interval (CI) = 1.76-2.02). After adjustment for a number of confounders-including substance use disorders-the association, although attenuated, remained significant (HR = 1.36; 95% CI = 1.27-1.46). This paper added strong evidence to the already well-demonstrated increased independent risk of suicide death among smokers [2][3][4][5].The reading of the recent paper of Bohnert et al.[6] in the same journal triggered some level of surprise. The authors, reportedly using the same database of individuals who received VHA services in FY 2005 and who were alive at the beginning of FY 2006 (n = 4 863 086), aimed to estimate associations between substance use disorders and suicide. The authors specified HRs between each alcohol, cocaine, cannabis, opioid, amphetamine, psychostimulant and sedatives use disorders, but not tobacco use disorder and death by suicide, in unadjusted and adjusted models. In the adjusted models, adjustment was made for many possible confounders (age, Charlson comorbidity index, co-occurring psychiatric conditions such as depression, schizophrenia, bipolar disorder, post-traumatic stress disorder and other anxiety disorders) but, astonishingly, not for tobacco use disorder or smoking.Thus, the authors demonstrated in 2014 [1] that tobacco use disorder was associated significantly with suicide mortality after adjusting for substance use disorders. The 2017 paper by the same authors [2] using the same database seems to ignore their own previous finding. Declarations of interestNone. RESPONSE TO AUBIN ET AL. (2017)We appreciate the comments on our paper by Aubin and colleagues [1], and agree that tobacco use disorder is a potentially important risk factor for suicide. Nonetheless, given preliminary evidence suggesting potential male-female differences with respect to other substance use disorders (SUDs) and the risk of suicide, the primary objectives of our most recent paper [2] were to estimate and compare other SUD-suicide associations between men and women in the US Veterans Health Administration (VHA). Within the paper, we acknowledged that incomplete adjustment for confounding factors, measured or unmeasured, was a potential study limitation. We elaborated further that,

Nicotine Inhalation and Suicide: Clinical Correlates and Behavioral Mechanisms

“…The complexity of temporal relationships among smoking, other risk factors, and suicide suggest that common mechanisms interact across risk factors for suicide 45 . Smoking can both predispose to and result from behaviors like impulsivity combined with negative affect (negative urgency, see below), which are involved in suicide 46‐48 .…”