Unlocking Private Healthcare in China: Large Language Model–Based Policy Analysis and Empirical Evidence on Patient Choice

Promoting private healthcare is widely recognized as a means to alleviate pressures on public systems and improve healthcare access. Yet in China, despite the rapid growth of private providers, patient uptake remains limited. This study investigates how different types of private sector promotion policies—market entry, government procurement, and insurance designation coverage—affect patients’ choices of private healthcare providers in China. We constructed a large-scale health policy database comprising over 500,000 official documents issued by Chinese central and local governments. Using a BERT-based large-language-model approach, we identified more than 15,000 policy statements promoting private providers. These were matched with four waves of nationally representative survey data from the China Health and Retirement Longitudinal Study, spanning 2011 to 2018. High-dimensional fixed-effect models were used to estimate the effects of policy exposure at the prefecture level on patients’ use of private primary care facilities (PCFs) and hospitals for outpatient and inpatient services. Findings reveal that local governments disproportionately emphasized market entry over post-entry policies such as government procurement and insurance designation. However, only government procurement and insurance designation policies significantly increased patients’ use of private PCFs for outpatient care. Market entry promotion policies had no significant impact on patient choices. Mediation analysis suggests that the presence of private providers alone does not fully explain patient choice, implying that service quality, affordability, and patient trust are likely critical. Our findings suggest that promoting private healthcare requires more than facilitating provider entry. Targeted post-entry policies that improve financial access and quality may be more effective in improving patient uptake—particularly for primary care. This study also demonstrates the potential of large language models to support policy analysis at scale, offering new tools for health system research globally.