When Relapse Comes Fast: Withdrawal, Illness—or Something More Complicated?
Rapid relapse after an antipsychotic is stopped is sometimes interpreted as evidence of pharmacological withdrawal. A new individual-participant-data meta-analysis in The Lancet Psychiatry tested whether that explanation fits the observed clinical pattern.
The answer is nuanced. Rapid and delayed relapse trajectories were identified, but rapid relapse was not significantly concentrated among people randomized to discontinue paliperidone. This weighs against a common, distinct withdrawal mechanism—but it neither rules withdrawal out nor reduces the established relapse risk associated with stopping antipsychotic treatment.
What the researchers studied
Louie and colleagues used individual-level data from the Yale Open Data Access database. They identified five randomized, double-blind, placebo-controlled relapse-prevention trials involving people with schizophrenia or schizoaffective disorder.
Participants had received antipsychotic treatment for more than three months, had become clinically stable, and were then randomized either to continued paliperidone or to placebo. Both oral and long-acting injectable formulations were represented.
The analysis focused specifically on 417 participants who experienced relapse:
271 from long-acting injectable trials
146 from oral trials
The investigators used longitudinal PANSS scores and latent-class mixed modeling to identify patterns of symptom change preceding relapse. PubMed abstract
Two relapse trajectories emerged
The analysis identified rapid-onset and delayed-onset relapse.
The investigators then examined whether rapid relapse occurred disproportionately among people whose antipsychotic had been discontinued.
Formulation | Discontinuation | Continued treatment | Comparison |
Long-acting injectable | 39 of 197 (20%) | 8 of 74 (11%) | p=0.12 |
Oral | 29 of 108 (27%) | 10 of 38 (26%) | p=0.95 |
No statistically significant difference was detected in either formulation group. Symptom profiles at relapse also did not differ by whether treatment had been discontinued or continued.
Across both formulations, participants in the rapid-relapse class had significantly higher baseline PANSS scores than those experiencing delayed relapse. Rapid relapse was also associated with greater symptom severity at the point of relapse.
Evidence versus interpretation
The evidence: Among participants who relapsed, rapid relapse was not statistically overrepresented after paliperidone discontinuation, and relapse symptom profiles did not differ by treatment status.
The authors’ interpretation: This is not the pattern expected if rapid relapse after discontinuation were commonly caused by a distinct pharmacological withdrawal effect. Higher baseline symptom severity may reflect greater underlying vulnerability.
The authors also acknowledge a possible trial artifact: symptom ratings near eligibility thresholds could make subsequent deterioration appear unusually abrupt.
What the study cannot establish: It cannot prove that withdrawal-related relapse never occurs, determine the mechanism of an individual patient’s relapse, or demonstrate that discontinuation is low-risk.
Why this matters clinically
The study challenges a simple inference:
Relapse happened quickly after stopping medication; therefore, withdrawal caused it.
Temporal proximity raises an important clinical question, but timing alone does not establish mechanism.
The more useful signal may concern baseline status. People in the rapid-relapse class already had greater symptom burden when entering the trial. Before considering dose reduction, clinicians should therefore examine residual positive, negative, and cognitive-disorganization symptoms rather than relying only on a broad description of “stable.”
A dose-reduction or discontinuation plan should include:
Individual assessment of residual symptoms and previous relapses
Shared decision-making about benefits, adverse effects, and uncertainty
A cautious, supervised strategy when reduction is pursued
Frequent monitoring during vulnerable periods
A clear contingency plan for emerging symptoms
Rapid access back to treatment
These are clinical implications drawn from the findings; they were not tested as interventions in this meta-analysis.
What the study does not change
Antipsychotics reduce relapse risk, and discontinuation increases it. This analysis examined the trajectory among people who relapsed; it was not designed to test whether maintenance treatment prevents relapse.
It should not be used to reassure patients that stopping medication is safe or to support routine discontinuation. Its narrower contribution is to question whether the rapidity of relapse, by itself, identifies a withdrawal mechanism.
Important limitations
All five trials involved paliperidone.
Only participants who relapsed were included.
The search was limited to trials available through the Yale database.
The meta-analysis was not preregistered.
Gradual or individualized tapering strategies were not compared.
The nonsignificant LAI comparison—20% versus 11%—does not prove equivalence.
Latent classes are statistical groupings, not established biological mechanisms.
Conclusion
Relapse after antipsychotic discontinuation is clinically heterogeneous. In this analysis, rapid relapse was associated more clearly with baseline symptom burden than with discontinuation status among those who relapsed.
The appropriate takeaway is not that withdrawal is impossible or that discontinuation is benign. It is that the timing of relapse does not, by itself, reveal its cause.
For clinicians, the enduring priorities remain individualized risk assessment, careful monitoring, shared decision-making, and rapid intervention when symptoms begin to return.
Reference
Louie K, Jauhar S, Rubio J, et al. Relapse trajectories and antipsychotic discontinuation in schizophrenia from individual participant data of five trials from the Yale Open Data Access database: a meta-analysis. The Lancet Psychiatry.
Published online August 24, 2026;13(10):829–840.
Comments