Abstract
BACKGROUND: This study aims to establish a multi-agent system model to provide accurate suggestions for the policy proposal of controlling the unreasonable growth of medical expenses charged by public hospitals in China. METHODS: A multi-agent system model was employed in this study. Agents of this model were divided into patients, doctors, medical institutions, the government, and medical insurance agencies. The model was composed of two subsystems: the disease and medical-seeking subsystem, and the medical expenses subsystem. Policy intervention experiments were conducted on patients' medical-seeking preferences, doctors' public welfare behaviors, and the government's financial investment. RESULTS: At present, medical expenses in China are unreasonable and keep increasing, and the proportion of medicine and physical examination expenses to total medical expenses for public hospitals is unreasonable. Intervention experiments suggested that expanding the promotion and application of the community first-visit system could rationalize patients' medical-seeking preferences, increasing doctors' incomes and reducing workload could significantly restrict doctors' over-prescription behaviors. Also, improving the government's financial investment could guide public hospitals to strengthen their commitment to public welfare responsibilities. These interventions could decrease the unreasonable growth of medical expenses of public hospitals. The combined intervention effects on suppliers, demanders, and the government were better than the effect of these agents independently. CONCLUSIONS: The main reasons for the unreasonable increase in patient medical expenses at public hospitals could be attributed to patients' unreasonable medical-seeking preferences, doctors' weak public welfare incentives, and the government's inadequate financial investment. Policy-makers should consider proposals to restrict and guide the behaviors of suppliers, demanders, and the government, simultaneously. The government should consider the feasibility, response speed, and implementation cost of policies as well.