Management of data quality--development of a computer-mediated guideline.

Appropriate data quality is a crucial issue in the use of electronically available health data. As source data verification (SDV) and feedback are two standard procedures for measuring and improving data quality it would be worthwhile to adapt these procedures to a current level of quality in order to reduce costs in data management. This project aims to develop a guideline for the management of data quality with special emphasis on this adaptation against the backdrop of research networks in Germany, which operate registers and conduct epidemiological studies. The first step in guideline development was a thorough literature review. The literature offers many measurements as candidates for quality indicators, however, systematic assessments and concepts of SDV and feedback are missing. We assigned possible quality indicators to the levels plausibility, organization, and trueness. Each indicator must be operationally defined to allow automatical calculation. The SDV sample size calculation leads to lower numbers for sites providing data of good quality and larger numbers for sites with poor data quality. The guideline's implementation in a software tool combines two cycles, one for the adaptation of recommendations to a given study/register, the other for the improvement of data quality in a PDCA-like approach. The recommendations will address needs common to medical documentation in daily health care, clinical, epidemiological, and observational studies as well as in surveillance data bases and registers. Further work will have to supplement other aspects of data management.