Rapid relapse after stopping an antipsychotic may not be a stopping effect

Design

Individual-participant-data meta-analysis of five double-blind, placebo-controlled discontinuation trials

Population

271 long-acting injectable and 146 oral paliperidone trial participants who relapsed

Primary outcome

Trajectory of symptom change preceding relapse, by discontinuation status

Effect

Rapid relapse not over-represented on discontinuation (20% vs 11%, p=0.12; 27% vs 26%, p=0.95); linked to higher baseline PANSS (p<0.001)

Pooled individual-participant data from five double-blind paliperidone discontinuation trials asked whether rapid relapse after stopping an antipsychotic is a pharmacological rebound. Two relapse patterns emerged — rapid and delayed — but rapid relapse was no more common among those who discontinued than among those who continued treatment, in both oral and long-acting injectable trials.

Instead, rapid relapse tracked with higher baseline symptom severity. If stopping the drug drove rapid relapse pharmacologically, you would expect it to cluster in the discontinuation arm; it did not.

For de-prescribing, this reframes the task. The patients at risk of a fast, severe relapse are identifiable by baseline severity rather than by the act of stopping, so dose reduction should be stratified by that risk and paired with closer monitoring in more severely ill patients — not abandoned, but individualised.

• Rapid relapse was not over-represented in those who stopped treatment (long-acting injectable 20% vs 11%, p=0.12; oral 27% vs 26%, p=0.95).
• Rapid relapse was associated with higher baseline PANSS scores (p<0.001), pointing to pre-existing severity.
• Stratify antipsychotic dose reduction by baseline severity, with closer monitoring for more severely ill patients.
• The data cover paliperidone discontinuation trials, so extrapolate to other antipsychotics with caution.

Why it matters

It shifts de-prescribing risk from the act of stopping to the patient's baseline severity, changing who you watch most closely.

Don't overread it

These were paliperidone trials in people who had already relapsed; it does not say discontinuation is safe, only that rapid relapse is not primarily a pharmacological rebound.

The statistics, in plain English

The non-significant p values (0.12 and 0.95) mean the proportion of rapid relapses was statistically indistinguishable between stopping and continuing — the opposite of what a pharmacological rebound effect would produce. The strong association with baseline PANSS (p<0.001) points to patient severity, not the act of stopping.

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.

Get Daily Doctor — free

Scan to keep reading on your phone. No account needed to start.