Verifying keyed medical research data.

Although diminished use of double keying for the verification of keyed medical research data has occurred, there are no published data that demonstrate the existence of equivalent or better replacement. The simple replacement of double keying with visual comparison is not recommended because visual comparison is too susceptible to the transient energy levels of the individual who makes the comparison. Some system design innovations applicable in specific situations allow elimination of verification by providing nearly equivalent or possibly even superior data quality assurance opportunities. These alternatives include designs where one replaces the transcription to data forms by direct keying from original source documents coupled with extensive quality reviews, and designs with direct data base input of data subjected to intense automated consistency checking and immediate analytic use. In both cases, the absence of keyed data verification places a greater burden for data quality assurance on other subsystems, such as quality review, and ultimately the analytic process. Data entry systems based on new technologies, such as tablet computers, electronic image processing, and voice input, will require similar considerations for assuring data quality.