Utility of hospital data in determining community disease epidemiology and burden.

Background: Public health surveillance is critical for understanding disease epidemiology and burden. Low-and-middle income countries (LMICs) in sub-Saharan Africa (sSA), are heavily reliant on hospitals for both the provision of care and surveillance for monitoring disease burden. However, the representativeness of disease burdens at the community level remains a persistent constraint on the utility of hospital-based disease surveillance. Various efforts have been implemented towards improving the quality of hospital surveillance data by comprehensively detailing patients’ treatment paths and incorporating geocoordinates of both care-seekers and hospital centres. This doctoral project aims to leverage routine hospital data to develop methodologies aimed at overcoming the biases of using hospital surveillance data in defining disease epidemiology and burden within the community. Specifically, to improve our understanding of 1) hospital use/access and competition between hospitals of varying service delivery options and referral pathways; 2) optimal definitions of hospital service catchment areas to provide a reliable denominator to enable the use of hospital-based data to estimate community-level disease burdens; 3) the impact of travel time on hospital outcomes and; 4) understanding the utility of hospital data to infer population-level disease estimates with robust validation.

Methodology: The first step will be to conduct a systematic review according to the Preferred Reporting Items for Systematic Reviews (PRISMA) guidelines on existing methodologies in the construction of a composite measure of access to hospital care. Second, novel statistical methodologies under the multifactor analysis framework will be developed to estimate a composite measure of access that integrates the different facets of hospital care access (availability, affordability, acceptability, accommodation, and geographic accessibility). Third, utilising routine hospital surveillance data, a novel estimation technique will be developed under the model-based geostatistical analysis framework to delineate hospital catchments adjusting for access to hospital care. Fourth, a hierarchical Bayesian geostatistical model will be developed to assess the role of travel time and catchment on hospital outcomes. In addition, different optimal travel time thresholds will be defined for mortality outcomes for different diagnosis. Lastly, using malaria disease as an example, malaria disease burden will be estimated based on the delineated hospital catchments under a geostatistical framework across the study area and estimates compared to those obtained using the DHIS 2 data.

Expected results: The research is expected to provide insights on the statistical methodologies for utilising hospital surveillance to understand community disease epidemiology and burden. In addition, the project will comprehensively quantify access to hospital care and provide community-level representative estimates for disease burden to inform policy on the cost-efficient allocation of healthcare resources in the region.