Abstract:
Objective This study also aimed to evaluate the rationality of assigning pancreaticoduodenectomy cases to the HB and GB groups under the CHS-DRG 2.0 classification system based on differing principal diagnoses, to compare intergroup differences in hospitalization costs and length of stay (LOS), and to perform stratified analyses at the diagnosis-related group (DRG) subgroup level to inform DRG optimization. A secondary aim was to discuss the potential clinical implications of DRG grouping for surgical decision-making, doctor-patient communication, departmental performance evaluation, and equitable reimbursement for complex surgical procedures.
Methods A retrospective analysis was conducted on 505 patients who underwent pancreaticoduodenectomy at the Cancer Hospital of Chinese Academy of Medical Sciences from January 2021 to October 2025. The cases were divided into either an HB group (432 cases) or a GB group (73 cases) according to the DRG classification. The Mann-Whitney U test was used to compare hospitalization costs and LOS between the two groups, with effect sizes evaluated using Cliff's delta and rank-biserial correlation. Stratified subgroup comparisons were performed for HB11 vs GB11, HB13 vs GB29, and HB15 vs GB15. The Kruskal-Wallis test was used to analyze the differences in hospitalization costs across principal diagnoses within the HB group. Given the imbalanced sample size, particularly the small number of cases in .certain GB subgroups, subgroup findings are interpreted as exploratory.
Results The two groups were comparable in age and gender (P>0.05). Median hospitalization costs were 135,400 yuan (IQR: 115,000~162,000) in the HB group and 139,300 yuan (IQR: 127,000~166,500) in the GB group (U=14,174, P=0.167). Median LOS was 20 days in both groups (HB: IQR 14~29 days; GB: IQR 13~25 days; U=14,135, P=0.157). Overall differences were not statistically significant, with non-parametric effect sizes indicating small differences (|δ|<0.147, |r|<0.1). However, the distribution of complication/comorbidity levels differed significantly between groups (χ2=11.063, P=0.0040). Stratified subgroup analyses showed significant differences in hospitalization costs between HB11 and GB11 [136,600 yuan (IQR: 115,100~170,900) vs 167,700 yuan (IQR: 145,000~199,800), U=864, P=0.0198] and between HB15 and GB15 [127,700 yuan (IQR: 109,800-144,300) vs 142,000 yuan (IQR: 124,900-164,900), U=1134, P=0.0077], but no significant difference between HB13 and GB29 (P=0.0964). No statistically significant difference was found among the five principal diagnoses within the HB group (H=7.12, P=0.130).
Conclusion There is no significant difference in resource consumption between pancreaticoduodenectomy cases in the HB group and the GB group. The current principal diagnosis-based grouping approach fails to fully reflect the principle of "similar clinical processes and similar resource consumption". However, stratified subgroup comparisons revealed significant cost differences in certain subgroups, suggesting that overall comparisons may be confounded by differences in complication levels and case composition. From a clinical perspective, suboptimal grouping may lead to inconsistent payment standards across different principal diagnosis pathways, potentially affecting the fairness of surgical approach selection, complicating departmental performance management, and posing challenges to physician-patient communication regarding cost expectations. It is recommended that pancreaticoduodenectomy be considered for inclusion in an expanded MDCA pre-MDC subgroup, or alternatively, that a refined surgical subgroup specifically for pancreaticoduodenectomy be established within the HB group.
Key words:
CHS-DRG 2.0,
Pancreaticoduodenectomy,
Hospitalization cost,
Cost structure,
Clinical management
Yunhe Hu, Yuzhuo Zhou, Jianqiang Tang, Aidong Li. Optimization of diagnosis-related group classification for pancreaticoduodenectomy: an empirical analysis based on the HB and GB groups of CHS-DRG 2.0[J]. Chinese Journal of Clinicians(Electronic Edition), 2026, 20(06): 470-478.