A Fuzzy Patient Matching Algorithm Based on HIV/AIDS Patient Identifiers in Kampala Uganda

Accurate patient information including transfer-ins and outs at every hospital visit is crucial for successful HIV care and treatment. In Uganda, one of the challenges to achieving this is lack of universally unique identifiers, making matching across different health facilities challenging. In this paper, we propose a set of identifiers for patient matching, data cleaning methods and a Fuzzy Matching approach to determine relations in two or more datasets. We used an experimental approach to develop and test a set of identifiers that when used in combination provide a match result. We implemented it in Open Medical Records System. Results show that patient identifiers often collected at health facilities can be used to fuzzily match a patient record. In the national roll out of patient matching using Finger print and National ID, Ministry of Health could consider Fuzzy Patient Matching as a possible solution to patient matching in the absence of these unique identifiers.