Depot antipsychotics halve manic admissions in bipolar disorder

Design

Self-controlled case series using territory-wide electronic health records, conditional Poisson regression

Population

2086 people with bipolar disorder prescribed both long-acting injectable and oral antipsychotics, Hong Kong, 2004 to 2023

Primary outcome

Health care utilisation, relapse and safety during depot vs oral periods

Effect

Psychiatric hospitalisation aIRR 0.67 (95% CI 0.61 to 0.74); manic 0.51 (0.44 to 0.59); mixed 0.31 (0.14 to 0.66); depressive 1.34 (0.94 to 1.93)

Among 17,841 people with bipolar disorder in Hong Kong's records, 2086 had periods on both a long-acting injectable and an oral antipsychotic, allowing each to act as their own control. Compared with their oral periods, depot periods carried lower rates of emergency attendance, all-cause admission and psychiatric admission. Admissions for mania roughly halved and mixed-episode admissions fell further still.

The pattern is specific rather than global. Depressive admissions showed no benefit — the point estimate went the other way and the interval crossed 1.0 — and non-psychiatric and cardiovascular admissions were unchanged. So the gain is in the pole where adherence failure and relapse are most tightly linked, and the feared metabolic and cardiac cost did not appear over these periods.

One safety signal matters for how you start. Extrapyramidal side effects were nearly three times as frequent during the first 90 days, after which the difference disappeared. That is an argument for a slower, watched initiation with a clear review point, not against the drug class.

• Consider a depot where the relapse pattern is manic or mixed and adherence is the problem; the evidence is weakest for depressive relapse.
• Review for extrapyramidal symptoms within the first 90 days, and use a rating scale rather than an open question.
• Do not offer a depot as a way of improving depressive symptoms in bipolar disorder.
• Discuss the choice of molecule on its own merits; this study pooled long-acting injectables as a class.
• In Indian practice, check supply continuity for the specific depot before starting it — an interrupted depot is worse than a continued oral.

Why it matters

It answers whether the depot argument from schizophrenia transfers to bipolar disorder, and shows it transfers to one pole only.

Don't overread it

This is observational within-person data, not a randomised comparison of depot against oral treatment.

The statistics, in plain English

The self-controlled design compares each person with themselves, which removes confounding by anything stable about that patient — illness severity, comorbidity, socioeconomic position — but not by anything that changes with time. Depots are often started after a bad period, so regression to the mean pushes in the direction of benefit. The mixed-episode estimate (aIRR 0.31, 95% CI 0.14 to 0.66) rests on few events and is correspondingly imprecise; the psychiatric admission estimate (0.67, 0.61 to 0.74) is the sturdiest number here. The depressive-admission result (1.34, 0.94 to 1.93) crosses 1.0, so no difference was shown either way.

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.