CHS-DRG 2.0 and insurance-specific patient cost sharing in gestational diabetes admissions: a regression discontinuity in time study

Vollständiger Abstract

Worum geht es in dieser Arbeit?

Objectives Diagnosis-related group reform aims to restrain inpatient spending, but obstetric evaluations commonly emphasize average hospital costs rather than insurance-specific patient cost sharing. We evaluated early discontinuities associated with CHS-DRG 2.0 in hospitalization costs, out-of-pocket (OOP) payment, self-pay ratio, and selected perinatal outcomes among gestational diabetes mellitus (GDM) admissions. Methods We analyzed 8,690 singleton GDM delivery admissions at a tertiary maternal and child health hospital in Fujian, China. The main cohort excluded eight stays exceeding 60 days, consistent with the policy DRG-management scope. We classified policy period from the recorded admission date under the provincial rule: admissions before 1 January 2025, including cross-cutoff stays, remained under the prior payment arrangement; admissions on or after that date entered the CHS-DRG 2.0 policy period. Actual DRG assignment and claim settlement were unavailable. Local linear regression discontinuity in time models used triangular kernels, data-driven bandwidths, robust bias-corrected confidence intervals, and discharge-day-clustered standard errors. Results The admission-date cutoff was not associated with a precise discontinuity in total hospitalization cost (estimate, −42.10 CNY; 95% CI, −1,158.46 to 1,074.26; p = 0.941), OOP payment (−102.15 CNY; 95% CI, −1,289.55 to 1,085.25; p = 0.866), or overall self-pay ratio (−2.75 percentage points; 95% CI, −8.06 to 2.56; p = 0.310). The self-pay ratio declined among resident-insured admissions (−12.41 percentage points; 95% CI, −20.05 to −4.77; p = 0.001), but not employee-insured admissions (−0.33 percentage points; 95% CI, −5.79 to 5.13; p = 0.905). The interaction estimate was −11.99 percentage points (95% CI, −19.85 to −4.13; p = 0.003). No exploratory perinatal outcome remained significant after false-discovery-rate correction. Conclusion The admission-date policy boundary was associated with a larger reduction in relative patient cost sharing among resident-insured GDM admissions. Discontinuous admission density at the cutoff and possible concurrent calendar-time changes limit causal attribution and preclude inference about a specific payment mechanism. Payment-reform evaluations should jointly assess insurance-specific patient cost sharing, provider expenditure, and clinical outcomes.