Commentary on Roberts et al. (2016): Bupropion and varenicline are efficacious and well-tolerated cessation medications for smokers with serious mental illnessThis network meta-analysis concluded that bupropion and varenicline are efficacious and well-tolerated cessation medications for smokers with serious mental illness who are stably treated and motivated to quit. Poorly graded data quality point to the need for direct comparisons of cessation medications, including nicotine replacement, in larger psychiatric samples.Varenicline, bupropion and nicotine replacement therapy (NRT) improve success with quitting smoking [1]. A network meta-analysis conducted with studies of non-psychiatric samples of smokers indicated equal efficacy for bupropion and single forms of NRT and superior efficacy for varenicline and combination NRT [2]. Unknown is the relative efficacy and tolerability of cessation medications among people with serious mental illness (SMI), a group with elevated smoking prevalence and increased risk for tobacco-related harms [3].Roberts et al.'s network meta-analysis sought to address this gap [4]. The authors defined SMI as any non-organic disorder with psychotic features that results in substantial disability. Of the 14 randomized controlled trials (RCTs) included, 11 were conducted with smokers with schizophrenia or schizoaffective disorder and three with smokers with bipolar disorder, two of which enrolled only five participants, more akin to case reports than clinical trials. Excluded were samples of smokers with unipolar depression, post-traumatic stress disorder (PTSD), other anxiety disorders and non-nicotine substance use disorders. Participants were recruited from the community, screened as stably treated, medication adherent and motivated to quit.Nine of the 14 trials, with 356 total participants, reported on abstinence outcomes, and 10 trials with 423 participants reported on tolerability (defined as trial discontinuation due to any adverse event). This averaged to about 40 participants per trial or about 20 per treatment group; retention rates were unreported. Although most of the individual trials were underpowered, the synthesized findings indicated statistically significant treatment effects with wide credibility intervals (CrI), for both bupropion [odds ratio (OR) = 4.51, 95% CrI = 1.45, 14.04] and varenicline (OR = 5.17, 95% CrI = 1.78, 15.06). Treatment efficacy for the two medications was comparable (OR = 1.15, 95% CrI = 0.24, 5.45), with no significant difference in tolerability for the two active drugs relative to placebo.Roberts et al. rated the evidence as very low quality [4], and with few studies there were no closed loops within