Psychosis relapse during treatment with long-acting injectable antipsychotics in individuals with schizophrenia-spectrum disorders: an individual participant data meta-analysis

Lancet Psychiatry. 2020 Sep;7(9):749-761. doi: 10.1016/S2215-0366(20)30264-9.


Background: Most individuals with schizophrenia-spectrum disorders have relapses, which increase the risk of morbidity and mortality. Because non-adherence to antipsychotic maintenance treatment could affect more than half of individuals with schizophrenia-spectrum disorders, psychosis relapse can often be confounded by unnoticed treatment interruption. Research of relapse during confirmed antipsychotic exposure has basic clinical and neurobiological implications, but data are scarce. We aimed to generate reliable estimates of incidence and predictors of relapse during assured antipsychotic treatment.

Methods: We did a systematic review and individual participant data (IPD) meta-analysis of clinical trials of long-acting injectable antipsychotics (LAIs) for psychosis relapse-prevention, following IPD-PRISMA guidelines. Datasets were identified by searching relevant repositories from inception to Aug 1, 2019. Each LAI group was reanalysed as a separate cohort, further identifying subcohorts of individuals with and without prospectively determined symptom remission (PSR). Summary incidence rate of relapse, incidence rate ratios (IRRs) of relapse between individuals with and without PSR, hazard ratios (HRs) of covariates on risk of relapse, and standardised mean difference (SMDs) in changes in overall functioning associated with relapse were generated by pooling results from the harmonised reanalysis of each study. This study is registered with PROSPERO, number CRD42019137439.

Findings: 19 treatment cohorts consisting of 5130 individuals (2938 with PSR, 2192 without PSR), with 3959·53 observed participant-years, were meta-analysed. Pooled incidence of relapse was 22·97 per 100 participant-years (14·76 per 100 participant-years for the PSR subcohort, 31·51 per 100 participant-years for the non-PSR subcohort), with an IRR of 0·19 (95% CI 0·07 to 0·54). Relapse was associated with functional decline (overall SMD -0·76, 95% CI -1·14 to -0·37; PSR SMD -0·52, 95% CI -0·80 to -0·21; non-PSR SMD -0·72, 95% CI -1·18 to -0·26). The strongest predictor of relapse was tardive dyskinesia at treatment onset (HR 2·39, 95% CI 1·05 to 5·42).

Interpretation: Despite the established efficacy of antipsychotics in preventing relapse, these data indicate that these drugs might not prevent subsequent exacerbations for a proportion of individuals whose illness is stabilised on continuous antipsychotic treatment. Tardive dyskinesia in particular might have pathophysiological implications for relapse.

Funding: Northwell Health.

Publication types

  • Meta-Analysis
  • Systematic Review

MeSH terms

  • Antipsychotic Agents / administration & dosage*
  • Clinical Trials as Topic
  • Delayed-Action Preparations
  • Humans
  • Injections
  • Psychotic Disorders / prevention & control*
  • Schizophrenia / drug therapy*
  • Schizophrenia / prevention & control
  • Secondary Prevention


  • Antipsychotic Agents
  • Delayed-Action Preparations