Is malaria messaging working? African biostatisticians have the data
- Wits University
Edson Mwebesa, Fellow at the Wits-based Sub-Saharan Africa Advanced Consortium for Biostatistics (SSACAB), knows children who died from late-diagnosed malaria.
His research has also revealed that malaria is more prevalent in pregnant Ugandan women than in any other population. Malaria is also endemic across Africa, impacting all aspects of social and economic life.
Mwebesa, a biostatistician, wanted to dive deeper into what made people choose methods to help prevent malaria. He focused on social and behavioural change messaging, an essential part of encouraging people to use insecticide-treated nets (ITNs). While messaging campaigns have been implemented in the media, hospitals, and schools, the question of whether these messages actually change behaviour remains uncertain.
Robust biostatistics methods used by African experts are set to change this.
Mwebesa applied a quasi-experimental causal inference method, known as Propensity Score Matching, to investigate this phenomenon. This approach can measure, with precision, whether messaging actually changes behaviour.
Traditionally, measuring messaging impact relied on simple correlations and on counting how many people heard a message and whether they used a mosquito bed net. But correlation cannot reveal whether the messages caused the behaviour, which in this case is using the mosquito net. Those who heard or saw the messages might differ from those who did not, and thus, the use of mosquito nets might not be comparable between these groups. Those who hear malaria messages are often wealthier, more educated, or better connected, and these factors independently increase bed net use. This has made it difficult for policymakers to know whether expensive behaviour-change campaigns genuinely deliver impact.
Mwebesa measured the impact of messaging aimed at women in Uganda. In that country, about one in four children under five years old tested positive for malaria. In some districts, incidence rates over six-month periods have exceeded 500 cases per 1000 people, demonstrating how quickly the disease can spread in high-transmission settings. Malaria causes fever, chills, weakness, anaemia, and in some cases, complications involving the brain, lungs and other organs. Beyond the clinical burden, malaria disrupts schooling, reduces productivity and places immense strain on household finances, particularly among poorer families.
Understanding how to drive preventive behaviour in this context is both a scientific and operational priority. Historically, however, researchers lacked the tools and data required to measure the impact of health communication. In 2001, economists John Luke Gallup and Jeffrey Sachs, writing in The Intolerable Burden of Malaria, explained that reliable data on malaria incidence were lacking for many of the most severely affected countries.
They constructed an indirect malaria index using historical risk maps, estimates of the proportion of falciparum malaria and population distribution data. Their work was pioneering but constrained by limited information, underscoring how weak data systems restricted the ability to study malaria’s economic and social effects. They could not measure behaviour or the effectiveness of prevention campaigns. They could only infer risk.
“We’re in a different position now. We have conducted repeated Malaria Indicator Surveys, geocoded demographic data, collected extensive health record data, and improved surveillance systems. What’s most exciting, though, is how we are growing the capacity to analyse data using modern causal methods,” says Mwebesa.
Institutions such as SSACAB have trained a new generation of African biostatisticians who can use advanced techniques to answer complex policy questions that were previously out of reach.
Mwebesa used nationally representative data from the 2018–19 Malaria Indicator Survey and examined whether exposure to malaria messages increases the use of insecticide-treated nets among women of reproductive age and children under five in Uganda. The descriptive findings show that 37.6 per cent of women aged 15–49 and 37.9 per cent of caregivers of young children had been exposed to malaria prevention messages in the six months preceding the survey. Net use was higher but not universal: 69.3 per cent of women reported sleeping under an insecticide-treated net the night before the survey, as did 71.8 per cent of children under five.
Mwebesa’s use of propensity score matching paired each woman who was exposed to malaria messages with another woman who was not, but who shared similar characteristics, such as age, education level, wealth, household size, region, and urban or rural residence. By ensuring that the two groups are comparable, the method isolates the effect of the messaging itself.
“For years, people assumed that malaria messages influenced behaviour, but this analysis shows, with causal evidence, exactly how much they matter. Our findings demonstrate that communication increases net use. We can now quantify this,” he said.
After matching, women who were exposed to malaria messages were 5.1 percent more likely to sleep under an insecticide-treated net than similar women who were not exposed. Among children, caregivers’ exposure to messages increased ITN use by 4.3 percent.
These differences, when applied nationally, translate into tens of thousands of additional protected households and reduced malaria direct and indirect costs. The study also identified which communication channels were major sources of malaria messages. Radio emerged as the dominant messaging channel, reaching roughly two-thirds of women and caregivers. Community health workers and interpersonal communication were also influential. Digital platforms were used far less frequently, suggesting an untapped potential.
Professor Tobias Chirwa, Principal Investigator for SSACAB and Head of the Wits School of Public Health, explains why this type of work matters for Africa’s statistical future. “This study shows what becomes possible when we combine strong African data with strong African statistical capacity. We move from describing problems to measuring impact. That is the essence of statistical innovation. African biostatisticians are leading analyses that were impossible twenty years ago.”
In light of African Statistics Day, this work reflects an important shift in statistical justice and evidence sovereignty. Africa is no longer dependent on global assumptions or incomplete models. It can produce high-quality causal evidence that speaks to households' lived realities and informs local policy with accuracy and confidence. This, in turn, supports more efficient investments in communication campaigns, builds equity by identifying which groups are least reached and strengthens the effectiveness of malaria prevention efforts across the continent.
African Statistics Day 2025 highlights the theme of leveraging innovations in data and statistics to promote a just, peaceful, inclusive and prosperous society for Africans. In the field of malaria prevention, this theme is particularly resonant.