Mental health effects of varenicline

BMJ 2015; 350 doi: http://dx.doi.org/10.1136/bmj.h1168 (Published 17 March 2015) Cite this as: BMJ 2015;350:h1168
  1. Mira Harrison-Woolrych, honorary research associate professor
  1. 1Dean’s Department, Dunedin School of Medicine, University of Otago, New Zealand
  1. miraharrison-woolrych{at}outlook.com

Results from a new meta-analysis seem at odds with patients’ real life experiences

The safety of drugs for smoking cessation, in particular varenicline (Chantix, Champix; Pfizer) and its effects on mental health, has been debated by regulatory authorities, researchers, prescribers, and patients since varenicline was first marketed nine years ago. The debate is likely to be sharpened by the linked paper by Thomas and colleagues (doi:10.1136/bmj.h1109), a systematic review and meta-analysis of randomised placebo controlled trials that found “no evidence of an increased risk of suicide or attempted suicide, suicidal ideation, depression, or death with varenicline.”1 This finding seems to be at odds with many patients’ experience of psychiatric adverse effects associated with varenicline, recently summarised in a citizen petition to the US Food and Drug Administration (FDA).2 It is timely to assess where the balance of evidence lies so that patients struggling to stop smoking can be advised about the risks of varenicline and consensual decisions about treatment can be made.

Mental health effects of varenicline—including depression, suicidal behaviour, and psychotic reactions—have been reported worldwide3 4 5 6 and are documented on the product information, including a “black box” warning on the Chantix label.7 Since the addition of these warnings in 2009, two British cohort studies8 9 and a meta-analysis of 17 industry sponsored trials10 found no increased risk of psychiatric adverse effects with varenicline and discussions regarding a causal association have continued (www.bmj.com/content/347/bmj.f5704/rapid-responses).

This new meta-analysis includes 44 randomised controlled trials in which 5817 patients received varenicline and 4944 patients received placebo.1 Primary outcomes of the meta-analysis were treatment emergent effects of suicide and attempted suicide (analysed together), suicidal ideation, and depression. No significant differences were shown between varenicline and placebo for these endpoints, although there were more completed suicides in patients taking varenicline (two) than in those taking placebo (none). For some secondary outcomes, including sleep disturbance and abnormal dreams, the meta-analysis showed an increased risk with varenicline compared with placebo.1 The authors concluded that their results “provide some reassurance for users and prescribers regarding the neuropsychiatric safety of varenicline.”

Patients who have experienced mental health effects while taking varenicline might be surprised by these conclusions. In October 2014, five non-profit organisations submitted a citizen petition to the FDA requesting strengthening of the warnings regarding neuropsychiatric effects of varenicline.2 The petition summarised evidence relating to four psychiatric event categories: suicidal ideation and behaviour, aggression and violence, psychosis, and depression. In addition to review of the published literature and two unpublished meta-analyses, several striking case reports were submitted. The features of these cases are consistent with other published reports of psychiatric adverse events4 6 and with findings during intensive post-marketing monitoring of varenicline in New Zealand.11 In a nationwide prospective cohort study of 3415 patients prescribed varenicline, about 3% reported symptoms of depression and more than half (65%) of these patients reported improvement after stopping varenicline. This “real life” evidence from pharmacovigilance causality assessments of hundreds of case reports suggests that varenicline is genuinely associated with mental health side effects.

To understand apparent differences between the conclusions of the new meta-analysis and patients’ real life experience of varenicline, we should compare these two different settings from which safety data can be obtained. Randomised clinical trials are usually designed to test drug efficacy against a comparator or placebo, with safety endpoints as secondary outcomes. These trials have strict inclusion and exclusion criteria, which narrow the study population or might mean it differs in clinically important ways from a general population taking varenicline. In the new meta-analysis, at least 18 of 44 (41%) studies seem to have excluded patients with psychiatric illness.1 People in these randomised controlled trials might have a lower risk of psychiatric effects with varenicline than patients trying to stop smoking in real life, in whom past or current mental health issues are common.

Differences in the sex distributions of study populations might also be relevant. In a post-marketing study in New Zealand, varenicline was prescribed to roughly equal numbers of men (48% of cohort) and women (52%).12 In the new meta-analysis, about a third of the studies included at least 75% men, with two studies having an almost all male population.1 Post-marketing research has also shown that women are over-represented in the group who experience psychiatric effects with varenicline.11 Thus the low proportion of women in some varenicline trials could reduce the incidence of psychiatric events reported.

There are other important differences between clinical trials and real life. More than half the studies (27/44) in the new meta-analysis investigated 12 weeks of treatment, as recommended on the product information,7 but only a small minority of patients receive exactly 12 weeks’ treatment in real life.12 The number of patients (and outcomes) in randomised controlled trials is usually much smaller than in post-marketing use studies and if an adverse event is rare (such as suicide) even meta-analyses might not have the statistical power to detect differences in rates of these events. Closer monitoring in clinical trials could reduce the risk of certain adverse events—for example, patients feeling suicidal might be less likely to commit suicide if this is detected early and action taken. Reporting and analysis of results could also affect safety conclusions: in the linked paper only 46% of the included trials reported data on suicidal ideation1 and combination of the endpoints of completed and attempted suicide for analysis for statistical reasons might not have been clinically appropriate.

What does all this mean for patients considering varenicline for smoking cessation? Continuing reports from real life experience of varenicline suggests that the concerns of patients and doctors are not unfounded, despite the linked paper’s conclusions, and that patients should still be warned of possible psychiatric effects. Patients who notice changes to their mental health while taking varenicline should discuss these with their doctor and consider stopping varenicline; this could resolve adverse psychiatric effects and serious consequences such as suicide could be prevented.


Cite this as: BMJ 2015;350:h1168


  • Research, doi:10.1136/bmj.h1109
  • Competing interests: I have read and understood the BMJ policy on declaration of interests and declare the following interests: none.

  • Provenance and peer review: Commissioned; not externally peer reviewed.


View Abstract