A Weighty Problem: Identification, Characteristics and Risk Factors for Errors in EMR Data

EMR data are used for decision support, research and quality control; it is important to assure their accuracy. Researchers have reported poor accuracy of categorical EMR records but little information is available on the accuracy of quantitative EMR data.We designed an algorithm for identification of errors in EMR weight data. The algorithm achieved precision of 98.9% with upper boundary for sensitivity of 57.6%.We employed the algorithm to analyze 420,469 weight records of 25,000 patients. The algorithm identified errors in 0.58% of entries in up to 7% of all patients. Users who previously made an error were nearly twice as likely to make another (p < 0.001) and physicians were less likely to make an error than non-physicians (p = 0.034). User practice location had a significant effect on their error rate (p =0.015).Rapid and accurate error identification could be used to improve quality of EMR weight data.