A method for measuring and reporting manual data extraction reliability.

Abstract

As health care costs have risen dramatically, the use of clinical data to analyze the quality of health care provided has increased. Central to this analysis is the means by which the clinical data itself is obtained. The reliability and validity of the data must be established in order to insure credible use of the data. The Wisconsin Ambulatory Review Project (WARP) is a three-year study of the nature of care in the ambulatory setting based on clinical data extracted retrospectively from written medical charts. This paper focuses on the method used to measure and report the reliability of the extraction process in the WARP study. The implementation of this methodology as a custom C program is described in detail.