The Use of Administrative Data to Generate Information for Decision-Making in Healthcare: A Narrative Review

Information technology has recently led to vast data generation, including in healthcare services. This data can be used to generate evidence for decision-making and has been widely used in high-income countries (HICs) to inform their decision-making process. However, the use of data in low- and middle-income countries (LMICs) is lacking. The burden of diseases between HICs and LMICs differs, with LMICs still struggling to battle infectious diseases. Thus, we aim to review the available literature on the status of using administrative data in generating evidence and how to use this data for decision-making on infectious diseases. Most studies were conducted in HICs, with the majority in Taiwan, the United States, South Korea, France, and Japan, compared to LMICs. Research in LMICs is scarce, highlighting inadequate information on prevalent diseases in LMICs, but not in HICs, such as neglected tropical diseases. HICs’ research utilized data from insurance claims, electronic medical records, and linkage from various databases. Administrative data are used for five main purposes, as follow 1.) to describe the epidemiology of diseases, 2.) to identify risk factors or associations among diseases, 3.) to evaluate healthcare services or policies, 4.) to identify any temporal changes in diseases, and 5.) to improve the validity of studies. The methodological approaches are diverse, ranging from regression analysis to advanced analysis such as machine learning. Yet, machine learning's potential in this domain remains underutilized since the applied analyses are mostly limited to data visualization, and conventional machine learning such as XGBoost, and decision tree analysis. An important highlight is that there are still gaps and opportunities for LMICs to further explore insights for infectious disease management, due to the current lack of data science research from LMICs in this area.