The IPD dose—response estimation Study Group was formed in February 2026. Research groups with relevant data sets were contacted. Data gathering and curation have commenced. 

Plasmodium vivax is the most geographically widespread cause of human malaria and can cause relapsing acute infections through activation of its dormant stage in the liver (i.e., hypnozoite), resulting cumulative morbidity and treatment challenges.

At present, primaquine or tafenoquine, both belonging to the 8-aminoquinoline drug class, are the most effective options for preventing Plasmodium vivax relapses. In most endemic countries, primaquine is the only drug that is routinely available; as tafenoquine has only recently been approved and requires a quantitative G6PD test prior to use. The optimal dose of primaquine must trade-off the absolute number of relapsing malaria episodes averted against the length of treatment duration, its tolerability, and safety. Accurate and precise estimation of drug efficacy is therefore essential for clinical and policy decision making.

The availability of individual patient data contributed through and pooled by IDDO has been a major advance in addressing this objective. These data include both randomised and non-randomised studies of various primaquine regimens conducted worldwide. For estimating drug efficacy, randomised data are likely to be more accurate but may offer less precision, for instance, when assessing treatment-effect heterogeneity; and they may not include the direct dose comparisons of primary interest. Pooling individual data with non-randomised studies could enhance precision and allow more flexible dosing contrasts; however, in our context, the impact of such pooling on accuracy remains uncertain as non-randomised data are prone to confounding. The current approach is simply to treat each patient, regardless of randomisation, as equally informative for quantifying comparative efficacy.

We aim to answer whether (1) dose–response curves can be robustly estimated using pooled individual patient data and (2) to what extent non-randomised variation in dose provides additional information.

Essential inclusion criteria of eligible studies

• Prospective clinical efficacy studies of uncomplicated vivax malaria; including randomised and non-randomised therapeutic trials, and prospective cohort studies with active follow-up
• A minimum follow-up of 42 days
• Treatment with a daily primaquine regimen given within the first three days of a schizontocidal treatment, i.e., chloroquine or one of six common artemisinin-based combination therapies (artemether-lumefantrine, artesunate-mefloquine, artesunate-amodiaquine, dihydroartemisinin-piperaquine, artesunate-sulfadoxine-pyrimethamine, artesunate-pyronaridine) 

Essential data for inclusion in the analysis

• Follow-up time and vivax malaria outcome
• Information on schizontocidal treatment (e.g., drug, dose)
• Information on primaquine (e.g., use, timing, dose, assignment)
• Baseline data on age, sex, body weight, study site
• Asexual parasite density at baseline
• Reporting of parasite presence or absence during follow up
• Status of pregnancy and lactation
• Presence of fever/temperature

• Individual tablet or mg dosing
• History of fever within the last 24 hours at baseline and during follow up

Shared individual patient data were uploaded to the Worldwide Antimalarial Resistance Network repository and curated in accordance with the IDDO SDTM Implementation Guide.

We aim to answer whether (1) dose–response curves can be robustly estimated using pooled individual patient data and (2) to what extent non-randomised variation in dose provides additional information.

To allow comparisons across alternative approaches, this study is structured using the ICH E9(R1) estimands framework, complemented by the target trial emulation approach to target the same precise estimand. Attributes to standardise include the population (patient eligibility criteria), treatment strategies, assignment procedure (or its emulation), time zero and follow-up, outcome, causal contrast, handling of intercurrent events, summary measure, and analysis plan. Our main target population will be individuals with incident, uncomplicated P. vivax malaria, confirmed at enrolment (symptomatic or asymptomatic), and eligible for multi-day primaquine radical cure. Exclusion criteria are contraindications to this primaquine regimen, including known G6PD deficiency, pregnancy, or lactation at enrolment. Patients who received adjunctive antimalarial treatments within 28 days of the initial schizontocidal treatment will also be excluded.

We will describe the characteristics of the pooled data—particularly the exposure of interest (continuous body-weight-adjusted primaquine total dose); compare different models based on different assumptions (randomisation-based approaches such as network meta-analysis and methods based on unconfoundedness such as direct adjustment using outcome regression and instrumental variable analysis); compare point and interval estimates; illustrate alternative presentations of effect measures (e.g. hazard ratios, survival curves, differences in restricted mean survival time) and non-linearity; and conduct sensitivity analyses (quantitative bias analysis and negative controls) to explore the impact of residual confounding. We will also assess the extent to which non-randomised data derived from single-arm studies contribute to the overall pooled dataset.

Ultimately, we will suggest a framework for how analysts can improve their analysis workflow using IPD and provide suggestions to IDDO and study contributors on how best to improve the IDDO platform.

The IPD dose-response estimation study group comprises participating investigators who contribute relevant datasets to the pooled analysis will retain ownership of their datasets. These will not be shared without their consent. Ihsan Fadilah, James Watson, Robert Commons, Kevin Baird, Ric Price, and Qinyu Li will oversee the analysis. Participating investigators will be recognised in publications as contributors.  A writing committee will be assigned to coordinate activities including planning (starting 1 Feb 2026), data analysis, and drafting of publications and reports for complete group review. The writing committee will comprise a group of investigators undertaking the data analysis and preparation of the manuscript. Authors will be recognised according the ICMJE guidelines and the IDDO publication policy. Results will be published in an open-access, peer-reviewed journal. 

For further information, please contact Ihsan Fadilah on email: ifadilah@oucru.org