Cost Homogenization and System-Level Drivers in Plateau Laparoscopic Cholecystectomy: Failure and Reconstruction of Traditional Cost-Control Models in the DRG Era
Discuss this preprint
Start a discussion What are Sciety discussions?Listed in
This article is not in any list yet, why not save it to one of your lists.Abstract
Background
Under the DRG/DIP payment reform, the cost structure and its driving factors for laparoscopic cholecystectomy (LC) in resource-limited plateau regions remain unclear. Traditional cost-control models focus on clinical process factors such as length of stay and operative time, but their applicability in the DRG era requires validation.
Methods
Based on a single-center cohort of 605 plateau LC patients from May 2020 to October 2025, natural log transformation was applied to total hospitalization costs. Pearson/Spearman correlation analysis, multivariate linear regression (two nested strategies: traditional clinical model and system-driven model incorporating year dummies), and quantile regression were used to identify cost drivers. VIF testing was performed for multicollinearity, and scenario simulation was conducted based on the regression model (operative time reduction to 40 min set according to best historical surgeon efficiency at our center).
Results
The mean hospitalization cost was ¥8097.49±936.85 with a coefficient of variation of only 11.6% and a Gini coefficient of 0.062, demonstrating high homogenization. The traditional six-variable clinical model (length of stay, operative time, age, BMI, systolic and diastolic blood pressure) yielded R²=0.008 (F=0.78, P=0.587), with no significant predictors, indicating that traditional clinical factors had largely lost explanatory power for cost variation under DRG control. The system-driven model (incorporating year dummies, season, and clinical variables) achieved R²=0.143 (F=12.41, P<0.001), with year effects (2024: +7.3%, P<0.001; 2025: −5.4%, P=0.001, comprising diagnosis composition changes and DRG policy transition) and seasonal effects (spring: +5.9%, P<0.001) as significant drivers, operative time marginally significant (P=0.039), and length of stay non-significant (P=0.638). Quantile regression revealed that age had a significant positive effect on high-quantile costs (Q90: β=0.0021, P=0.003) but not on median costs. Scenario simulation showed that clinical process optimization (reducing operative time to 40 min) could only reduce costs by approximately 0.6%, far below the annual policy fluctuation (12.1% decline from 2024 to 2025).
Conclusion
Plateau LC costs demonstrate high homogenization under DRG control, and traditional clinical process-based cost-control models have minimal explanatory power. System-level factors (annual effects comprising diagnosis composition and policy transition, seasonal resource fluctuations) explain significantly higher cost variation than clinical process factors. We recommend extending the classic Donabedian Structure-Process-Outcome (SPO) framework to a System-Allocation-Outcome (SAO) paradigm, shifting the focus of cost control from clinical process optimization to policy window management and system resource allocation.
