- Design
- Individual participant data meta-analysis of five double-blind discontinuation trials
- Population
- 417 stabilised adults with schizophrenia or schizoaffective disorder (271 injectable, 146 oral)
- Primary outcome
- Relapse trajectory (rapid or delayed) by discontinuation or continuation
- Effect
- Rapid relapse 20% vs 11% (injectable, p=0.12) and 27% vs 26% (oral, p=0.95), discontinuation vs continuation
Researchers pooled individual participant data from five randomised, double-blind, placebo-controlled paliperidone discontinuation trials, 271 from long-acting injectable trials and 146 from oral trials, all clinically stable before randomisation.
Among people who relapsed, two trajectories emerged: rapid and delayed. Rapid relapse was not over-represented after discontinuation. In the injectable trials 20% (39 of 197) of discontinuation relapses and 11% (8 of 74) of continuation relapses were rapid (p=0.12); in the oral trials the figures were 27% (29 of 108) and 26% (10 of 38; p=0.95). People who relapsed rapidly had higher baseline PANSS scores.
This argues against treating quick relapse as mainly a pharmacological withdrawal effect, though the injectable comparison was underpowered. Higher baseline severity could reflect susceptibility or a trial artefact. The practical message is risk stratification and individual monitoring when deprescribing, not a new threshold for stopping.
- Rapid relapse was linked to higher baseline symptom scores, not to stopping itself.
- Stratify risk before deprescribing, using baseline severity and past relapse history.
- Plan closer early monitoring for patients with higher baseline severity.
- The data cover paliperidone only; do not assume the same pattern for other antipsychotics.
Why it matters
It challenges the assumption that fast relapse after stopping is a discontinuation phenomenon, and points to severity instead.
Don't overread it
The analysis covers paliperidone trials only, was not preregistered, and the injectable comparison was underpowered.
The statistics, in plain English
A p-value of 0.12 for the injectable trials means the data neither show nor rule out a difference; with only 74 continuation relapses the comparison has little power. The finding is reassuring about one explanation, not proof against it.
Read the rest in the app
You have read your two free briefings this month. The app carries all 27 specialties, every morning, free — and this finding is waiting in it.
Scan to keep reading on your phone. No account needed to start.
Tomorrow morning, before your first patient
One edition a day for psychiatry, written by the desk, every claim tied to its paper. Six minutes.
Get the app — free