Design: Individual participant data pairwise, dose-response and network meta-analyses
Population: 7,062 adults with acute schizophrenia in 18 fixed-dose trials (2,168 women)
Primary outcome: Change in PANSS total score at about 6 weeks, by sex
Effect: No sex difference (DoSMD 0.04, −0.09 to 0.18); prolactin rise greater in women (DoSMD 0.83)

An international team pooled individual participant data from 18 fixed-dose trials (7,062 adults with acute schizophrenia; 2,168 women) of haloperidol, olanzapine, paliperidone, risperidone or placebo, using the Vivli and YODA data-sharing platforms.

There was no sex difference in effect on overall symptoms at about six weeks (difference in SMD 0.04, 95% CI −0.09 to 0.18). Dose-response curves were similar, levelling off at 3–5 mg risperidone equivalents in both sexes. Antipsychotics reduced all-cause dropout more in women (ratio of odds ratios 0.74, 0.56 to 0.98; moderate credibility) and raised prolactin considerably more (difference in SMD 0.83, 0.67 to 0.99; high credibility).

Clinicians sometimes start women on lower doses on the assumption they respond differently. For the drugs studied, these data do not support that on efficacy grounds. The difference that is credible lies in prolactin.

The practical change is to dose by response and tolerability, not by sex, and to monitor prolactin-related effects more closely in women, particularly on risperidone, paliperidone and haloperidol.

Dose antipsychotics by response and tolerability, not by sex.
Expect little extra symptom benefit above about 3–5 mg risperidone equivalents in either sex.
Monitor prolactin and related symptoms more closely in women.
These findings cover five drugs in acute schizophrenia, not every antipsychotic or phase of illness.

Why it matters

It removes a common assumption about sex-based dosing and puts the attention where the difference actually lies.

The statistics, in plain English

A difference in standardised mean difference of 0.04 with an interval from −0.09 to 0.18 means any sex difference in efficacy is small and could go either way. The prolactin difference of 0.83 is large, and its interval stays well away from zero. Individual participant data let the authors test the sex interaction directly rather than compare trial averages.