Systematic literature review and meta-analysis of the efficacy of artemisinin-based and quinine-based treatments for uncomplicated falciparum malaria in pregnancy: methodological challenges

Makoto Saito, Mary Ellen Gilder, François Nosten, Rose McGready, Philippe J. Guérin

There is no agreed standard method to assess the efficacy of anti-malarials for uncomplicated falciparum in pregnancy despite an increased risk of adverse outcomes for the mother and the fetus. The aim of this review is to present the currently available evidence from both observational and interventional cohort studies on anti-malarial efficacy in pregnancy and summarize the variability of assessment and reporting found in the review process. Efficacy methodology and assessment of artemisinin-based treatments (ABT) and quinine-based treatments (QBT) were reviewed systematically using seven databases and two clinical trial registries (protocol registration-PROSPERO: CRD42017054808). Pregnant women in all trimesters with parasitologically confirmed uncomplicated falciparum malaria were included irrespective of symptoms. This review attempted to re-calculate proportions of treatment success applying the same definition as the standard WHO methodology for non-pregnant populations. Aggregated data meta-analyses using data from randomized control trials (RCTs) comparing different treatments were performed by random effects model. A total of 48 eligible efficacy studies were identified including 7279 treated Plasmodium falciparum episodes. While polymerase chain reaction (PCR) was used in 24 studies for differentiating recurrence, the assessment and reporting of treatment efficacy was heterogeneous. When the same definition could be applied, PCR-corrected treatment failure of ≥ 10% at any time points was observed in 3/30 ABT and 3/7 QBT arms. Ten RCTs compared different combinations of ABT but there was a maximum of two published RCTs with PCR-corrected outcomes for each comparison. Five RCTs compared ABT and QBT. Overall, the risk of treatment failure was significantly lower in ABT than in QBT (risk ratio 0.22, 95% confidence interval 0.07-0.63), although the actual drug combinations and outcome endpoints were different. First trimester women were included in 12 studies none of which were RCTs of ABT. Efficacy studies in pregnancy are not only limited in number but use varied methodological assessments. In five RCTs with comparable methodology, ABT resulted in higher efficacy than QBT in the second and third trimester of pregnancy. Individual patient data meta-analysis can include data from observational cohort studies and could overcome some of the limitations of the current assessment given the paucity of data in this vulnerable group.