How to Improve the Medical Quality of the Coding Reports Based on Who-Art and Costart Use

Abstract

At Rhône-Poulenc Rorer, the Clinical Coding and Dictionaries Unit is responsible for managing large databases to support adverse experiences reporting. An autoencoding system allows the unit to match investigator literal terms to an in-house dictionary of synonyms linked to the most appropriate preferred WHO-ART and COSTART terms. For well-defined reasons listed below, the efficiency and accuracy of the encoding process are far from being optimal: 1. Poor quality of investigator literal terms, 2. Difficulty in creating appropriate synonyms, 3. Weaknesses of the autoencoding algorithm, and 4. Weaknesses of the WHO-ART and COSTART dictionaries.

These problems are explained in detail and solutions are suggested (eg, investigators could be asked to make more diagnoses, could be allowed to use their mother tongue, reference dictionaries should contain more specific terms as well as vague preferred terms, a precise clinical definition should be assigned to each of them; not only adverse drug reactions but all intercurrent events should be taken into account, and so forth). In conclusion, the need for a new faultless and user friendly dictionary recognized worldwide will be emphasized.