Estimating Plasmodium falciparum Parasite Rate using Test Positivity Rate from 2016-2024: Health Management Information Systems in Uganda

BackgroundMalaria transmission in Uganda is heterogenous, so the national malaria program needs information about the distribution of malaria to develop appropriate policies. While population-based community surveys estimate Plasmodium falciparum parasite rate (PfPR), they are too infrequent and sparse for routine malaria management. Health facility data is routinely collected and covers a large geographic scope, but the data is collected passively, variable in quality, and potentially highly biased. We aimed to triangulate test positivity rate (TPR) from health facility data to survey estimated PfPR data in Uganda to create monthly, high-resolution PfPR estimates. MethodsUsing matched health facility and survey data, we fit a multi-level logistic regression model that accounted for clustering at the district and region level, to predict PfPR from TPR. Additional covariates were explored to select a final model that reduced bias while prioritizing its utility for programmatic tasks. Model predictions were validated against observed PfPR and used to generate monthly district-level prevalence estimates from 2016 to 2024. Regional and national level estimates were made by weighting district level estimates by population. ResultsThe final model included a smoothed TPR term and proportion of severe malaria cases at a district-month level. Predicted PfPR was strongly positively correlated with the observed survey PfPR (Pearsons rank correlation rho =0.79, p<0.001). National estimates derived from predicted PfPR aligned well with survey estimates from the same time and area. ConclusionHealth Management Information System (HMIS) data, when paired with research data, can be used to estimate malaria prevalence with high spatial and temporal resolution. Estimates can be tested and models can be updated to help malaria programs best leverage facility data. In the context of declining survey frequency, HMIS-based modeling offers a resilient and cost-effective alternative for malaria surveillance and programmatic decision-making in Uganda and similar high-burden settings.