Rising DIP-weighted case mix, restructuring of cost composition, and expanding settlement deficits under diagnosis-intervention packet payment: evidence from case-level data in a tertiary grade a traditional Chinese medicine hospital in Anhui Province, China, 2022-2024.

BACKGROUND: Diagnosis-Intervention Packet (DIP) payment operates under regional global budgets in China. Multi-year evidence from tertiary traditional Chinese medicine (TCM) hospitals on settlement performance, cost composition, and departmental heterogeneity remains limited. We described temporal changes in DIP-weighted case mix, resource use, cost structure, and settlement performance and examined factors associated with case-level settlement deficit and any-deviation status. METHODS: We retrospectively analysed inpatient DIP records from a tertiary Grade A TCM hospital in Anhui Province during 2022-2024. Outcomes were adjusted DIP settlement deficit (Bi < 0) and any-deviation status (high- or low-multiple versus normal). Logistic models used two-way cluster-robust standard errors for repeated admissions within patients and clustering within departments. Restricted cubic splines, year-specific upper-tail exclusions, and observed 30-day same-hospital readmission provided sensitivity and limited quality assessments. RESULTS: Among 176,253 admissions from 118,520 patients, 27,912 patients (23.6%) had multiple admissions. From 2022 to 2024, median disease point value rose from 76.54 to 84.14 and case-mix index from 1.0777 to 1.2300, whereas mean length of stay fell from 9.47 to 8.71 days and mean cost per admission from 7,377 to 6,823 CNY. The aggregate examination and laboratory cost share rose from 19.30 to 22.65%; within-department changes accounted for +3.43 percentage points and departmental expenditure-mix changes for -0.08 percentage points. The unadjusted aggregate settlement shortfall increased from 16.40 million to 42.10 million CNY, and deficit cases increased from 56.0 to 63.2%. High-multiple classification declined from 10.4 to 2.6%, whereas low-multiple classification increased from 8.9 to 12.3%. Clustered models showed higher deficit odds in 2024 versus 2022 (OR 1.424, 95% CI 1.206-1.682) and with longer stay (linear-slope OR 1.123/day, 1.080-1.168), but lower odds with higher disease point value (OR 0.946/10 points, 0.930-0.962). Same-hospital 30-day readmission was 11.18, 10.89, and 11.54% across 2022-2024. CONCLUSION: At this single tertiary TCM hospital, higher DIP-weighted case mix and lower mean resource use coincided with less favourable administrative settlement performance. The observational design precludes causal attribution to DIP, and settlement deficit is not an accounting net loss. Findings support auditable department- and disease-group review and prospective evaluation incorporating TCM-service and clinical-quality indicators.