Aims/Background: Accurate and standardized diagnostic coding is essential for hospital performance, reimbursement, and the secondary use of health data for management and research. Despite its significance, how diagnostic coding structure and complexity shift over time, and how these trends differ between institutions, remain poorly defined at the health system level. This study aimed to assess long-term trends and inter-hospital variability in diagnostic coding practice in Chinese public hospitals, focusing on the secondary diagnosis number (SDN), International Statistical Classification of Diseases and Related Health Problems, 10th Revision (ICD-10) chapter composition, and diagnostic diversity over a 12-year period. Methods: The retrospective study analyzed inpatient discharge data from two public tertiary hospitals (Hospital A and Hospital B) at three time points: 2012, 2019, and 2024. Key indicators included the mean SDN per admission, ICD-10 chapter-level diagnosis composition, the shannon diversity index, and heatmap-based profiling of coding structure. Non-parametric tests were applied to assess temporal trends and differences between hospitals. Results: Mean SDN increased significantly in both hospitals between 2012 and 2024 (Hospital A: 1.45 vs. 1.62; Hospital B: 1.37 vs. 1.61; p < 0.01), suggesting progressively deeper documentation of comorbidities and complications. ICD-10 chapter distributions also changed over time, with a gradual increase in chronic disease-related diagnoses and a substantial reduction in non-specific categories, such as symptoms and abnormal findings (p < 0.001). The shannon diversity index showed a modest downward trend, reflecting increasing concentration and standardization in coding patterns. Heatmap analysis further revealed a decrease in inter-hospital variation and a clear convergence toward standardized diagnostic structures over time. Conclusion: Over the past decade, diagnostic coding practices in Chinese public hospitals have become more comprehensive, more structured, and increasingly standardized. These trends are consistent with the combined effects of policy reforms, institutional learning, and the digitalization of health information systems. Furthermore, we observed that improvement in coding depth and structure is likely to contribute to greater accuracy and reliability of Diagnosis-Related Group (DRG)-based payment systems.
Wu et al. (Fri,) studied this question.