UCSF

Malaria is one of the leading causes of morbidity and mortality in sub-Saharan Africa, causing over 200 million cases and 405,000 deaths in 2017 (1). The World Health Organization states that quality malaria surveillance is essential to target regions and populations at highest risk, accurately measure changes in disease burden, and evaluate the impact of interventions (2). Surveillance data can also be used to produce meaningful indicators of malaria burden in order to identify determinants of disease and produce high resolution maps of risk (3, 4). This is essential because malaria transmission is heterogenous across space and time (5). Despite the critical need for high quality data, malaria surveillance is particularly poor in high burden countries (1).

The ideal indicator for estimating malaria morbidity in high transmission areas is case incidence, which is defined as the number of cases of malaria per unit time divided by the population at risk (3, 6, 7). The approach currently employed by most countries in sub-Saharan Africa to estimate malaria incidence involves cartographic methods that link parasite prevalence to incidence rates (8, 9). This method has several weaknesses. It relies on cross-sectional surveys that are costly, time-consuming, and infrequently performed. In addition, its accuracy is limited because the relationship between prevalence and incidence is variable and non-linear, particularly in high transmission settings. An alternative method of estimating incidence that is more direct, inexpensive, and comprehensive is the use of data reported through routine health management information systems (HMIS). Unfortunately, this is challenging because HMIS data are typically not available at an individual patient level and lack information on where patients come from. Cases of malaria detected by the HMIS often lack laboratory confirmation. Furthermore, quantifying the denominator for incidence using these data is ambiguous because catchment areas around health facilities are not well defined.

The overall goals of this dissertation are to leverage surveillance data routinely collected at public health facilities in Uganda to improve measurement of malaria burden and to accurately capture changes in malaria burden across space and time. This dissertation is organized into three chapters. The first chapter used enhanced HMIS data to estimate care-seeking populations around health facilities and, with these estimates, to generate measures of malaria incidence over time. These estimates were then compared to “gold standard” measures of incidence measured concurrently in cohorts. Through developing this new method to generate catchment area populations, we found that our estimates of incidence accurately captured gold standard incidence and its dynamics over time. The second chapter aimed to map malaria incidence across Uganda using this new measure of incidence captured through enhanced HMIS data. Using spatio-temporal modeling, we generated predictions of malaria incidence in 2019-2020 at the parish-level in Uganda. The model performed well, particularly in areas where sentinel surveillance sites were more densely located across space. The findings from this analysis suggest that routinely collected health facility data could be used for risk mapping purposes. The third chapter used enhanced HMIS data to evaluate the impact of indoor residual spraying (IRS) in Uganda. We used data from 5 districts that received over 7 years of sustained IRS and compared monthly malaria burden to a baseline pre-IRS period. Our findings suggest that malaria burden declined 85% in the 4th and 5th year of sustained IRS, but increased in the 6th and 7th years. The timing of the increase in burden coincided with a switch in IRS active ingredient to a new chemical, clothianidin. These findings are an important call for more research aimed at studying the real world effectiveness of IRS with clothianidin.

Taken together, these chapters underscore the potential for routinely collected health facility surveillance data to answer important scientific questions and to inform policy.