Processing medical records: Data related problems

Abstract

Computer processing of medical data can sometimes give a poor result. This can be due to several reasons. First, medical data may be heterogeneous, sometimes inacurate and often subjective. Secondly, the complex process of codification, data capture and validation is rarely carried out as carefully as possible. Finally, the processing itself can be the cause of some inaccuracy when it does not correspond to the defined objectives for which the data have been collected.

MIRACLE: A Time-oriented Data Validation and Analysis System for Clinical Research

“…A major problem with all collections of clinical information is how to ensure the quality of the data [13]. Manual checking before entering the data into the computer is mandatory [21]; data acquisition programs routinely perform syntactical checking while entering new data [3,5,8,11]. Separate data validation systems also exist [22,23].…”

“…As a rule, one error per 100 values will be found by the program even after the data had been very thorougly checked manually, after check-punching, structural and syntactic validation. Manual checking of the coding forms before they are entered is nevertheless indispensable because even extensive plausibility validation can never detect all possible errors [21], especially if the data are formally correct.…”