Identifying subgroups with decreased performance on malaria rapid diagnostic tests in Malawi

We sought to identify groups in which performance of point-of-care malaria rapid diagnostic tests (MRDTs) was decreased to improve the accuracy of MRDT-based diagnosis. We evaluated the performance of Plasmodium falciparum MRDTs compared to quantitative polymerase chain reaction (PCR) in population subgroups based on 1108 patient visits to rural health centers in Malawi from 2019 to 2021. Approximately 48% of malaria PCRs and 37% of MRDTs were positive. Overall, the fraction of positive MRDTs among participants with positive PCR (sensitivity) was 67% and the fraction of negative MRDTs among participants with negative PCR (specificity) was 91%. Sensitivity varied with age, lowest among subjects aged 25-40 years (38%) and < 1 year (33%) and highest among those 5-8 years (86%). Nearly all cases of MRDT-negative/PCR-positive discordance (false negative MRDTs) had < 200 parasites/µl of blood. In regression models, male sex, younger age, and presence of fever, headache, chills, and vomiting reduced the likelihood of a false negative result. Presence of fever, headache, chills, and vomiting were more likely to result in a false positive result. Using classification tree analysis, combinations of clinical and demographic characteristics predicted true positive and true negative MRDT results with 79% and 84% overall accuracy, respectively. Age and parasite density were negatively correlated; age was included in regression models and classification trees rather than parasite density because it would be information readily available to clinicians. Though there are several possible explanations for discordance between PCR and MRDT (including incidental malaria infections), MRDTs performed less well in certain subgroups, which may inform deployment and development of these important tools in clinical settings and for malaria surveillance.