Accuracy of cause-of-death coding in Taiwan: Types of miscoding and effects on mortality statistics

Background. The objectives of this study were to assess the accuracy of cause-of-death coding, determine the extent to which coders follow the selection rules of coding set out in the International Classification of Diseases, 9th Revision (ICD-9), and the effects of miscoding on mortality statistics in Taiwan. Method. A systematic sample of 5621 death certificates was reviewed. The underlying cause of death (UCD) selected by the reviewer for each death certificate was compared with that selected by the original coder. The UCD was selected according to ACME (Automated Classification of Medical Entities) Decision Tables. Results. The overall agreement rates between the reviewer and coders according to the three-digit and two-digit categories of ICD-9 were 80.9% and 83.9%, respectively. Good agreement was found for malignant neoplasms (kappa = 0.94) and injuries and poisoning (kappa = 0.97), but there was poor agreement for nephrotic diseases (kappa = 0.74), hypertension-related diseases (kappa = 0.74), and cerebral infarction (kappa = 0.77). Reasons for disagreements included disagreement in nomenclature (42.8%), inappropriate judgement of causal relationships (41.5%), and incorrect interpretation of Selection Rule 3 and Modification Rules (15.7%). Conclusion. This study showed various levels of agreement for different diseases between the reviewer and the original coders in selection of the UCD. Owing to the 'compensatory effect of errors', the national mortality statistics were not affected significantly. The national administration should undertake routine internal studies to control the quality of UCD coding practices.