Supporting Information.

<p>Systematic search terms for the databases. Signalling questions for risk of bias assessment using the QUIPS tool adapted to the current analysis. Table A. PRISMA-IPD checklist. Table B. Risk of bias assessment. Table C. Studies included in analysis. Table D. Studies that were eligible for analysis but not included in the pooled data. Table E. Comparison of characteristics of patients (as originally reported) who received primaquine between included and eligible but not included studies. Table F. Regression table output for the main Cox proportional hazards model. Fig A. Study sites that contributed to the pooled data in this individual patient data meta-analysis. Fig B. Distribution of weight-adjusted primaquine daily dose by primaquine regimen. Fig C. Dynamics of primaquine-induced increases in blood methaemoglobin over time. Fig D. Day 7 methaemoglobin concentrations by primaquine regimen and dose group. Fig E. Inverse J-shaped association between patient age and day 7 methaemoglobin by recurrence status, after controlling for daily mg/kg primaquine dose. Fig F. Distribution of day 7 methaemoglobin among the patients on (A) the original scale and (B) the logarithmic scale. Fig G. Mixed-effects estimates for the intercept and slope (mean difference). Fig H. Mixed-effects estimates for the slope (hazard ratio). Fig I. Example of sample size calculations for future studies. Fig J. Comparisons of estimates from the current analysis with those previous studies.</p> <p>(DOCX)</p>