Abstract

OBJECTIVES:

To determine the completeness and accuracy of ICD-9-CM codes allocated by primary health care physicians in their computerized medical records and evaluate the effects of improvement procedures.

METHODS:

The codes of 87,806 patients assigned to 56 primary care physicians in the Basque National Health Service in Spain, were evaluated 3 times over a 1-year period according to the following criteria: correspondence to a valid ICD-9-CM code, agreement between diagnosis and code, and the percentage of visits with an unspecified reason for consultation. Finally, the mean number of unique diagnoses and rates of diagnostic groups in the 84,136 patients that remained with the same physician for a minimum of 6 months were contrasted with another previously registered morbidity database. Two interventions were performed to improve coding: detected errors were corrected centrally and physicians were assessed and given information on their individual results.

RESULTS:

Diagnoses lacking an ICD-9-DIC code decreased from 59% in the first assessment to 2% at the end of the study period. The percentage of coding mistakes (discrepancies in episode diagnosis and ICD-9-CM code) decreased from 17% to 3%. The mean annual number of diagnoses per patient was slightly lower than that in the reference database (2.26 versus 2.43). The same result was observed in the rates of some diagnostic groups.

CONCLUSIONS:

Primary care doctors can achieve a high degree of quality in ICD-9-CM diagnosis coding. Implementing procedures for evaluating coding, rectifying mistakes, and providing information to physicians markedly improved the initial results.