Accuracy of clinical coding from 1210 appendicectomies in a British district general hospital.

AIMS: The primary aim of this study was to assess the accuracy of clinical coding in identifying negative appendicectomies. The secondary aim was to analyse trends over time in rates of simple, complex (gangrenous or perforated) and negative appendicectomies. METHODS: Retrospective review of 1210 patients undergoing emergency appendicectomy during a five year period (2006-2010). Histopathology reports were taken as gold standard for diagnosis and compared to clinical coding lists. Clinical coding is the process by which non-medical administrators apply standardised diagnostic codes to patients, based upon clinical notes at discharge. These codes then contribute to national databases. Statistical analysis included correlation studies and regression analyses. RESULTS: Clinical coding had only moderate correlation with histopathology, with an overall kappa of 0.421. Annual kappa values varied between 0.378 and 0.500. Overall 14% of patients were incorrectly coded as having had appendicitis when in fact they had a histopathologically normal appendix (153/1107), whereas 4% were falsely coded as having received a negative appendicectomy when they had appendicitis (48/1107). There was an overall significant fall and then rise in the rate of simple appendicitis (B coefficient -0.239 (95% confidence interval -0.426, -0.051), p = 0.014) but no change in the rate of complex appendicitis (B coefficient 0.008 (-0.015, 0.031), p = 0.476). CONCLUSIONS: Clinical coding for negative appendicectomy was unreliable. Negative rates may be higher than suspected. This has implications for the validity of national database analyses. Using this form of data as a quality indictor for appendicitis should be reconsidered until its quality is improved.