Evaluation of breast cancer coding quality and its influence on diagnosis-related groupings: a cross-sectional study

BACKGROUND: This study aimed to analyze the diagnosis and procedure coding of breast cancer in our hospital, summarize the types of error codes, and analyze the impact on disease diagnosis-related grouping (DRG). METHODS: The data source was the cases of breast cancer discharged from the oncology department according to DRG settlement from 1 June 2024, to 31 May 2025, and the diagnostic codes and surgical codes on the first page of the medical history were classified and statistically analyzed. Factors Influencing the Occurrence of Encoding Defects Using Logistic Regression Analysis. RESULTS: Coding errors were identified in 93 out of 752 cases. These included 28 main diagnosis code errors (with 15 cases failing DRG enrollment), 49 surgical procedure code errors (with 33 cases failing DRG enrollment), and 16 cases involving both diagnosis and surgical procedure code errors (with 9 cases failing DRG enrollment). Diagnostic information deficiencies included 55 major diagnostic errors and 29 coding errors. Surgical information comprised 23 major procedural errors and 30 coding errors. The main causes of coders' coding defects are over-reliance on coding databases, insufficient mastery of coding rules, and failure to carefully review medical records. The causes of coding defects caused by physicians are non-standardized documentation by clinicians, diagnostic errors by clinicians. The comparison results of the common defect medical records before and after modification show that the surgical coding has a certain impact on the DRG allocation of cases involving breast cancer surgery treatment. The 752 cases were divided into the defect-free group (n = 659) and the defective group (n = 93) based on the presence or absence of defects. Logistic regression analysis showed that number of diagnoses, resuscitation performed, years of experience as a coder, and years of experience as a senior physician are independent predictors of coding defects were independent predictors of coding errors. CONCLUSION: This study reveals that coding errors in breast cancer cases at our hospital primarily stem from coders' insufficient understanding of relevant rules, overreliance on coding libraries, and inadequate medical record review. Additionally, non-standard documentation practices and diagnostic errors by clinicians are significant contributing factors. Independent predictors of coding errors include the number of diagnoses, length of hospital stay, resuscitation status, and the years of experience of both coders and senior physicians.