Comparative Study of Inpatient Utilization between Automobile Insurance and National Health Insurance in Korea

Abstract

Background: South Korea operates the National Health Insurance (NHI) system, which provides universal coverage to all citizens, along with an automobile insurance (AI) system that includes private liability insurance offering full coverage without patient cost-sharing. These structural and institutional differences offer divergent incentives for both patients and providers, potentially shaping distinct patterns of healthcare utilization. Particularly, the combination of a fee-for-service payment model and full coverage under AI increases the risk of moral hazard, provider-induced demand, and unnecessary care. Despite growing concerns—especially regarding rising expenditures in traditional Korean medicine under AI—empirical comparisons of inpatient utilization between AI and NHI remain limited, and evaluations of recent policy efforts to reduce excessive hospitalization are scarce. This study aims to address these gaps by comparing inpatient healthcare utilization across insurance types and empirically evaluating the impact of the December 2022 policy that strengthened claims review for mild-condition hospitalizations. Methods: This retrospective cohort study used national claims data from 2018 to 2024, focusing on the five most frequently claimed diagnostic groups under AI, which account for approximately 72% of total AI expenditures. First, inpatient episodes for these conditions under both NHI and AI were extracted and matched 1:1 based on sex, age, medical department, primary diagnosis, surgery status, ICU stay, and admission timing, to ensure comparability between groups. Generalized estimating equations (GEE) were employed to account for repeated admissions. Second, a Difference-in-Differences (DID) analysis assessed the effect of the strengthened claims review policy implemented in December 2022. A Difference-in-Difference-in-Differences (DDD) analysis was also conducted to evaluate heterogeneous effects by targeted diagnosis groups. Primary outcomes included length of stay, total expenditure per episode, and per diem expenditure. Results: Compared to NHI patients, AI patients had 2% longer hospital stays (exp(β) = 1.02, 95% CI: 1.01–1.03), 9% higher total inpatient expenditures (exp(β) = 1.09, 95% CI: 1.08–1.10), and 10% higher per diem expenditures (exp(β) = 1.10, 95% CI: 1.09–1.10). Among NHI patients, those receiving traditional Korean medicine care had 24% longer stays than those receiving Conventional medicine (exp(β) = 1.24, 95% CI: 1.21–1.27), while under AI, traditional Korean medicine care was associated with 40% higher medical costs (exp(β) = 1.40, 95% CI: 1.37–1.42). Following the policy implementation in December 2022, the average length of stay for AI patients decreased by approximately 6% (exp(β) = 0.94, 95% CI: 0.92–0.96), and total expenditure per episode declined by around 9% (exp(β) = 0.91, 95% CI: 0.89–0.93). However, the DDD analysis showed limited effects among the specifically targeted diagnostic groups. Conclusions: This study empirically analyzed the characteristics of inpatient healthcare utilization and the effects of policy interventions under the automobile insurance (AI) system, which has been relatively understudied in Korea. Even after rigorous matching to adjust for differences in patient characteristics between the two insurance types, utilization under AI was significantly higher than under National Health Insurance (NHI). This suggests that in a zero cost-sharing system like AI, both provider- and patient-side moral hazard may arise. The analysis of policy effects further indicates that, even in a system without patient cost-sharing, policy interventions can partially mitigate excessive hospitalizations. The limited effects observed in the triple-difference analysis may be attributable to the non-disclosure of the targeted diagnosis groups. To enhance the sustainability of the insurance system and promote the efficient allocation of healthcare resources, it is essential to establish clear policy objectives to address moral hazard and to develop a systematic framework for continuous monitoring and evaluation.