Coding Errors: Classification,Detection,and Prevention

Abstract

Coding is necessary for summarizing adverse events in clinical trial reports. Although caricatural errors are easy to detect, more subtle errors are common. The importance of detecting subtle errors is discussed with particular reference to the assessment of safety during drug development. A classification of coding errors is presented.

The seriousness of coding errors is highly variable. The most serious errors are those which are difficult to detect and one should consider in particular those resulting indirectly in attenuation of the severity of adverse events, or in a false sense of security. Detection of coding errors is achieved by the “decode” procedure, preferably computer-assisted, and if possible by a different person. Prevention of coding errors can be achieved by: qualified coders with a good knowledge of medical terminology and of the coding dictionaries, training sessions during which difficult terms are coded by consensus, and group discussions of coding errors detected.