Document Type : Original Article
Center for Data Science, New York University, New York City, NY, USA
Roche Hong Kong and Macau, Hong Kong SAR, China
University Institute of Lisbon (ISCTE-IUL), Lisbon, Portugal
Faculty of Sciences, University of Lisbon, Lisbon, Portugal
Many countries with universal healthcare have a parallel private healthcare sector due to the waiting time in the public sector. People purchase individual health insurance to pay for private services. Past studies on the relationship between the public sector’s waiting time and the demand for health insurance have two limitations: not considering the capacity of the private sector, and subsequently, the omission of a feedback loop. These limitations are also present in the health insurance policy discussion in Hong Kong, where the public sector is overstretched. A lack of understanding of market dynamics might lead to unrealistic expectations of public policy. This study highlights these limitations, and tries to answer the research question: whether the historical dynamics between the intersectoral imbalance of burden and the demand for health insurance in Hong Kong could be quantitatively explained.
A system dynamics model was created based on a negative feedback loop. The model’s initial input was the percentage of population with health insurance in 2009, and to simulate the percentage continuously until 2019. Results from 2015 to 2019 were compared with actual figures to examine the model’s explanatory power. Multivariable sensitivity analysis was performed.
With initial fluctuation, the simulated result stabilized and was within the acceptable error range from 2015 to 2019. The mean absolute percentage error (MAPE) was 0.94%. At the end of 2019, the simulated percentage of population with health insurance is 36.6% versus the “real value” of 36.7%. Simulated patient admissions and occupancy rates also approximate the reality. Sensitivity analysis demonstrates the robustness of the model.
We can quantitatively explain the feedback loop between health system burden and demand for health insurance. With local parameterization, this model should be transferable to other universal health systems for a better understanding of the system dynamics and more informed policy-making.