Cost deviation as an operational signal of financial performance in a tertiary public hospital in China: a department-level panel data analysis

Department expenditure is difficult to compare across Chinese public hospitals because workload and case complexity vary over time and between clinical units. A variance from expected cost may provide a more informative monthly signal than expenditure alone. To examine the concurrent association between model-based cost deviation and department financial performance in one tertiary public hospital in China. The analysis used a balanced panel of 30 clinical departments observed monthly from January 2021 through December 2024 (1,440 department-months). Expected cost was estimated from case-mix index (CMI), workload, length of stay, bed occupancy, and department and month fixed effects; predictions were retransformed with Duan’s smearing estimator. We related cost deviation, (actual cost − expected cost)/expected cost, to surplus rate in two-way fixed-effects models with standard errors clustered by department. Sensitivity and additional analyses assessed model diagnostics and temporal prediction, lagged and nonlinear associations, surgical-department heterogeneity, cost components, and four quality outcomes with Bonferroni correction. The primary expected-cost model had an adjusted log-scale R² of 0.979, an in-sample mean absolute percentage error of 5.93%, and variance inflation factors below 4.2. Cost deviation was associated with surplus rate in the same month (β=−0.309; clustered standard error, 0.026; 95% confidence interval, − 0.362 to − 0.256; p < 0.001). A 10-percentage-point higher deviation corresponded to a 3.09-percentage-point lower surplus rate. Associations at 1, 3, and 6 months were not significant. The additional slope for surgical departments was − 0.174 (standard error, 0.038; p < 0.001). No quality association met the Bonferroni threshold, and the quadratic test did not support a common threshold. In this hospital, cost deviation identified departments under current financial pressure but showed no association with subsequent surplus at the tested 1-, 3-, or 6-month lags. Because actual cost contributes to both measures, the contemporaneous coefficient has an accounting component and should be interpreted as a monitoring association. Validation in other hospitals should use outcomes independent of this shared component.