Adverse events after antipsychotic discontinuation: an individual participant data meta-analysis

Summary

Background

Adverse events can occur after antipsychotic discontinuation but evidence from antipsychotic drug trials is scarce. We aimed to estimate the occurrence of adverse events after discontinuing antipsychotics.

Methods

For this two-stage individual participant data meta-analysis, we searched the Yale University Open Data Access Project's database for randomised controlled trials of antipsychotics from database inception until May 6, 2021. We included placebo-controlled antipsychotic randomised controlled trials with individual participant data of participants (aged ≥ 18 years, of any sex and ethnicity) with schizophrenia, schizoaffective disorder, or bipolar disorder. Studies were excluded if treatment with antidepressants, lithium, or antiepileptic drugs was initiated as additive therapy at the start of the placebo phase. Starting from the screening or washout phase, we divided participants who were randomised to placebo into two groups: the discontinuation group (participants who discontinued prestudy antipsychotics at the start of the screening or washout phase) and control group (participants who did not take prestudy antipsychotics for at least 4 weeks before the start of the screening or washout phase). Participants were excluded from the discontinuation and control groups if they discontinued prestudy treatment with antidepressants, lithium, or antiepileptic drugs up to 4 weeks before baseline, received an antipsychotic as a tolerability test, or received a long-acting injection of an antipsychotic within 12 weeks before baseline. In the discontinuation group, individuals were excluded if they discontinued prestudy antipsychotic treatment more than 3 days before, or any day after, the start of screening or washout phase. The prespecified primary outcome was occurrence of at least one new somatic adverse event with an onset within 4 weeks after the start of the screening or washout phase. We implemented a generalised linear model that accounted for potential confounders, to estimate the effect of antipsychotic discontinuation. This study is registered with PROSPERO (CRD42021224350).

Findings

We identified 409 records of which 18 were eligible and included in the analysis. From these 18 studies, 692 individuals (242 [35·0%] women and 450 [65·0%] men) were eligible for the discontinuation group and 935 individuals (339 [36·3%] women and 596 [63·7%] men) were eligible for the control group (median age in both groups: 39 years [IQR 30–47]). New somatic adverse events occurred in 295 (43%) individuals in the discontinuation group and 293 (31%) individuals in the control group (OR 1·74; 95% CI 1·27–2·39; τ²=0·15; moderate strength of evidence). New psychiatric adverse events were also more frequent in the discontinuation group than the control group (OR 2·01; 95% CI 1·38–2·94). Longer duration of treatment before discontinuation (OR for doubling the duration of treatment: 1·08; 95% CI 1·01–1·14) was associated with a higher probability of new somatic adverse events after antipsychotic discontinuation, and tapered discontinuation (compared with abrupt discontinuation: 0·54; 0·32–0·91) and no history of somatic illness (compared with history of somatic illness: 0·63; 0·43–0·91) were associated with lower probabilities of new somatic adverse events after antipsychotic discontinuation. The risk of bias was moderate in 13 (72·2%) studies and serious in five (27·8%) studies.

Interpretation

We detected moderate evidence of emerging somatic adverse events after discontinuation of first-generation and second-generation antipsychotics, particularly after discontinuation of longer durations of treatment. Tapered discontinuation can mitigate the risk of emerging somatic adverse events after antipsychotic discontinuation. These findings have implications for the safety of treatment discontinuation and could be used for tailored treatment planning.

Funding

German Research Foundation.