PREDICTING MALARIA SURGES IN SOUTH SUDAN

During the rainy season from April to November, hundreds of thousands of people in South Sudan are at risk of contracting malaria. Due to missing data and unreliable reporting, predicting peaks in the number of malaria cases is challenging. Operational research suggests linear regression as a novel model to estimate surges of malaria patients, allowing for better planning of preventive measures and treatment capacities.

Caused by parasites transmitted by female Anopheles mosquitoes, malaria remains a dangerous and often deadly disease. 228 million people were estimated to be infected globally in 2018 alone, and over 400,000 died. Infants and young children are especially at risk, making for over two thirds of malaria deaths worldwide.

In South Sudan, malaria accounts for 68 percent of all illnesses and 72 percent of deaths reported in health facilities. In some months during the rainy season, patient numbers can double at hospitals and paediatric wards.

While a preventable disease, protective measures such as bed nets and environmental spraying, as well as sufficient treatment facilities need to be made available when patients need them the most. To predict sudden increases in malaria cases, Médecins Sans Frontières (MSF) and other humanitarian actors use mathematical models estimating future cases based on historical data.

Yet in South Sudan, fragmented, missing, or noncomparable data impede the validity of predictive standards like weekly averages with standard deviations.

A LuxOR-supported study therefore tested a moving linear regression model as an alternative, requiring only eight weeks of previous data to predict disease surges.

To benchmark the new model with older standards, the study additionally used a complete influenza data set from Norway as a testing ground. Here, the linear regression performed comparably well to calculating the weekly averages and adding standard deviations – the current standard method used by MSF. For malaria data from five MSF projects in South Sudan, linear regression correctly identified several exceptional malaria occurrences including one direct operational response by MSF.

The study results suggest the linear regression model is a valid early warning indicator to predict surges in malaria cases where only limited disease surveillance data is available. It has the potential to improve emergency preparedness and reduce outbreak response times, saving lives in the fight against malaria.

In the wake of the global Coronavirus disease pandemic, LuxOR supports MSF epidemiologists and medical coordinators in the field with the design and implementation of disease surveillance and monitoring tools. Research advisors are part of MSF’s COVID-task force in Brussels, advising on evidence gaps, testing, surveillance strategies and operational research requests. The team also helps assembling and reviewing all medical guidance materials developed in-house, making it publicly available for partners and medical actors.