Social Connections and Public Healthcare Utilization

Abstract

How can public healthcare administrators predict tertiary healthcare utilization when hospitals face highly variable patient flows but large and infungible costs of specialty treatment? Using administrative claims data from a public health insurance program in India shows that utilization by social connections is associated with 26% increase in first-time utilization. The social connections model decreases root mean squared error by 11.25% compared to a baseline model. Social connections predict treatment in private rather than public facilities, but not in specialty use. The enhanced model predicts higher levels of optimum healthcare provision when the costs of undertreatment are severe.