Physician-clinical coder collaboration effectively improves coding accuracy: A single-centre prospective study in China.

Background: Clinical coding is important for reimbursement, resource planning, administration and medical research. Objective: This study aimed to evaluate clinical coding accuracy and its influencing factors, especially the benefits of physician-clinical coder collaboration. Method: Twenty-four physicians and one senior clinical coder participated in the quality audit. The audit checklist, assessment criteria, training program and physician-clinical coder collaboration mechanism were clearly defined. The homepage filling standards, homepage filling guidelines and the guidelines of the International Classification of Diseases were used as the assessment criteria for evaluating accuracy. Results: A total of 323,320 medical records were reviewed. The average accuracy of homepage completion was 60.4% and poor-quality homepages accounted for 89.9% of coding errors. The average coding accuracy and correction percentage were 83.4% and 62.3%, respectively. After physician-clinical coder collaboration, the coding accuracy increased from 78.9% to 87.1% (χ² = 799.904, p < 0.001) and correction percentage increased from 52.0% to 73.0% (χ² = 1628.015, p < 0.001). Multivariate logistic regression revealed that complexity of medical records (odds ratio (OR) = 0.625), quality of homepages (OR = 20.445), month of physician-clinical coder collaboration (OR = 1.133-2.297), coder's major (OR = 1.616), coding experience (OR = 1.953), work engagement (OR = 1.290) and day of the week (OR = 1.054) were factors influencing coding accuracy. Conclusion: Physician-clinical coder collaboration effectively improved clinical coding accuracy and clinical coders benefited greatly. However, homepage quality was not improved. Furthermore, homepage quality and psychological factors influenced coding accuracy. Implications: Managers should implement regular standardised training for homepage completion, alongside ongoing improvements in coding practices and training.