Expert knowledge sourcing for public health surveillance: National tsetse mapping in Uganda.

Abstract: In much of sub-Saharan Africa, availability of standardized and reliable public health data is poor or negligible. Despite continued calls for the prioritization of improved health datasets in poor regions, public health surveillance remains a significant global health challenge. Alternate approaches to surveillance and collection of public health data have thus garnered increasing interest, though there remains relatively limited research evaluating these approaches for public health. Herein, we present a case study applying and evaluating the use of expert knowledge sources for public health dataset development, using the case of vector distributions of Human African Trypanosomiasis (HAT) in Uganda. Specific objectives include: 1) Review the use of expert knowledge sourcing methods for public health surveillance, 2) Review current knowledge on tsetse vector distributions of public health importance in Uganda and the methods used for tsetse mapping in Africa; 3) Quantify confidence of the presence or absence of tsetse flies in Uganda based on expert informant reports, and 4) Assess the reliability and potential utility of expert knowledge sourcing as an alternative or complimentary method for public health surveillance in general and tsetse mapping in particular. Information on tsetse presence or absence, and associated confidence, was collected through interviews with District Entomologist and Veterinary Officers to develop a database of tsetse distributions for 952 sub-counties in Uganda. Results show high consistency with existing maps, indicating potential reliability of modeling approaches, though failing to provide evidence for successful tsetse control in past decades. Expert-sourcing methods provide a novel, low-cost and rapid complimentary approach for triangulating data from prediction modeling where field-based validation is not feasible. Data quality is dependent, however, on the level of expertise and documentation to support confidence levels for data reporting. Results highlight the need for increased evaluation of alternate approaches and methods to data collection. [Copyright &y& Elsevier]