- Design
- Dose-response meta-analysis of cohort studies and trials
- Population
- 134,111 people with schizophrenia spectrum disorders on clozapine, 34 studies
- Primary outcome
- Efficacy and safety outcomes by plasma clozapine level
- Effect
- Turning points: BPRS 535, seizures/EEG 452, blood dyscrasia 357, tachycardia 503 ng/mL
This dose-response meta-analysis, published in July 2026, combined 34 cohort studies and trials with 134,111 participants exposed to clozapine. It modelled how plasma clozapine and norclozapine levels related to symptom response and adverse effects.
Exploratory turning points for symptom improvement were about 535 ng/mL on the BPRS and 898 ng/mL on the PANSS. Turning points for harms were lower: salivary hypersecretion about 225, blood dyscrasia about 357, EEG abnormality or seizures about 452 and tachycardia about 503 ng/mL; overall tolerability about 707. Certainty was moderate for 11 of 21 outcomes and low for 10.
The authors are explicit that these are not universal thresholds or dose targets. The practical message is narrower: pushing levels well above the usual range in pursuit of response carries rising risk without clear added benefit. That supports individual level-guided dosing rather than escalation by default.
- Use plasma levels to guide clozapine dosing rather than escalating on clinical impression alone.
- Be cautious about pushing levels high when response is incomplete; added benefit was uncertain.
- Watch for seizures and EEG change as levels rise; consider prophylaxis per local practice.
- Treat these turning points as population estimates, not individual targets.
Why it matters
Clozapine is often pushed higher when response is partial, and this suggests caution in doing so.
Don't overread it
Exploratory model turning points; the authors warn against using them as targets.
The statistics, in plain English
A turning point from a quadratic model is where a curve changes direction on average across studies. It is not a threshold for any one patient, and mixing many different studies makes the exact numbers uncertain. Low certainty for half the outcomes means these numbers could shift with better data.
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 clinical pharmacology, written by the desk, every claim tied to its paper. Six minutes.
Get the app — free