Background: Acutely occurring, life-threatening side-effects of antipsychotic drugs might contribute to the reduced life expectancy observed in patients with severe mental disorders. We aimed to assess this hypothesis by doing a systematic review and meta-analysis of deaths occurring in placebo-controlled trials of antipsychotic drugs.
Methods: For this systematic review and meta-analysis, we included randomised controlled trials comparing second-generation antipsychotics with placebo across several diagnostic categories. We searched MEDLINE, EMBASE, Cochrane CENTRAL, BIOSIS, PsycINFO, PubMed, ClinicalTrials.gov, and WHO ICTRP from inception (the last search was done on Jan 21, 2017), and contacted pharmaceutical companies and regulatory authorities for further eligible trials. We examined mortality from any cause (the primary outcome) and mortality from natural causes, suicide, and other non-natural causes. We synthesised the results with odds ratios (ORs) in a common-effects meta-analysis. We investigated the effects of age, diagnostic category, sex, study duration, antipsychotic drug used, drug dose, and polypharmacy in subgroup and meta-regression analyses. This study is registered with PROSPERO, number CRD42016033930.
Findings: We identified 596 randomised controlled trials published between 1978 and 2017, comprising 108 747 participants. 352 studies (comprising 84 988 participants) with mortality data available constituted the main dataset for our meta-analysis. 207 (0·4%) deaths were reported in 53 804 patients on an antipsychotic drug and 99 (0·3%) deaths in 31 184 patients on placebo. 300 (85%) of 352 trials were 13 weeks (3 months) or shorter in duration (median 6 weeks; IQR 4-10). We found no evidence of a difference between antipsychotic drugs and placebo in mortality by any cause (OR 1·19; 95% CI 0·93-1·53), from natural causes (1·29; 0·85-1·94), from suicide (1·15; 0·47-2·81), and from other non-natural causes (1·55; 0·66-3·63). Most subgroup and meta-regression analyses did not indicate any important effect moderators. The exceptions were increased mortality in patients with dementia (OR 1·56; 95% CI 1·10-2·21), in elderly patients (1·38; 1·01-1·89), in aripiprazole-treated patients (2·20; 1·00-4·86), and in studies with a higher proportion of women (regression coefficient 0·025; 95% credible interval 0·010-0·040). However, the effects in elderly patients, aripiprazole-treated patients, and women were mainly based on the included dementia trials. For patients with schizophrenia there was no evidence of an increased mortality risk (OR 0·69; 95% CI 0·35-1·35).
Interpretation: Overall, and for the main indication of schizophrenia, there is no evidence from randomised trials that antipsychotic drugs increase mortality. However, vulnerable populations (particularly patients with dementia) might be at increased risk. This meta-analysis could only address acute treatment effects leading to death in the short-term, and not long-term effects of antipsychotic drugs on mortality.
Funding: German Ministry of Education and Research.
Copyright © 2018 Elsevier Ltd. All rights reserved.