Research methodology: coding perceived morbidity in general practice—an evaluation of the read classification and the international classification of primary care (ICPC)

Abstract

Objectives—To evaluate the Read Classification and the International Classification of Primary Care (ICPC). Methods—The Read Classification was used to code the diagnoses for 3474 patient encounters, in a pilot sample of three volunteer practices (11 general practitioners), and the ICPC was used to code 21,416 patient encounters in a stratified quota sample of 22 practices (59 general practitioners), in a survey aiming to relate prescribing to perceived diagnosis. Results/Experience—The Read Classification was found to be a detailed and exhaustive classification of medical diagnoses, but it was more time consuming to use than the ICPC, due to the complexity of the classification, the over-use of alpha characters compared to the ICPC, and the mixing of alpha characters with numeric digits within the codes. Encoding, decoding and statistical analysis were found to be more straightforward using the ICPC compared with the Read Classification. The ICPC was found to be deficient in 40 important diagnoses, and these are listed. Conclusion—The Read Classification was of limited value in this drug utilization survey, in that the design of the code reduced its utility in statistical analyses. The ICPC was an efficient code, which met the criteria of exclusiveness, usefulness and hierarchy. The classification is not exhaustive enough to prevent loss of information as a result of coding, but the authors' amendments virtually eliminated this problem. © 1997 John Wiley & Sons, Ltd.