Effective factors on accuracy of principal diagnosis coding based on International Classification of Diseases, the 10th revision (ICD-10)

Quality of diagnostic data depends on accurate coding. The purpose of this study was to assess the accuracy of principal diagnosis coding and its effective factors. To achieve this aim, three hundred and seventy medical records were randomly selected and recoded blindly (as gold standard). The effects of possible factors on accuracy of coding which was gathered through observation method were analyzed by Chi-square (chi(2)), Fisher exact test, odds ratio (OR), and confidence interval 95% for OR. There were 84 (22.7%) errors in principal diagnosis codes, 28 errors (33.3%) of which were major ones. Less experienced coders showed fewer errors (p < 0.0001); however, these errors were mainly major (p < 0.0001). Diagnosis coding in the general hospital was significantly more accurate, but most errors in the general hospital were major (p < 0.0001). Lack of memory-based coding (p < 0.0001) and not using abbreviation (p = 0.001) reduced errors. Further, reviewing the record thoroughly increased coding accuracy and reduced major errors insignificantly. More thorough documentation about topography (p = 0.204), subtype (p = 0.708) and etiology (p < 0.0001) of diseases decreased the coding accuracy. Most errors in readable records were minor. In conclusion, not using abbreviation, records' readability, paying more attention to the available information and no memory-based coding can improve the quality of diagnosis classification. (C) 2009 Elsevier Ltd. All rights reserved.