A7:    The nature of data in the electronic patient record: the promise and deceit of observational databases

Johan van der Lei MD PhD. Associate Professor, Dept. of Medical Informatics. Erasmus University Rotterdam (The Netherlands); Miriam Sturkenboom PhD. Assistant Professor of Pharmacoepidemiology, Erasmus University, Rotterdam (The Netherlands)

Target level: INTERMEDIATES

Researchers have argued that once physicians would use electronic patient records, a rich source of observational data would become available to study both the delivery of care and the outcome of that care. In a number of countries (e.g. UK or The Netherlands), population based general practitioners have started using electronic patient records and investigators are constructing observational databases for research purposes. One common use of these observational databases is to study the use of drugs (post-marketing surveillance or pharmacovigilance).

But are the data obtained from the routine delivery of care suitable to answer scientific issues related to the quality of care? This tutorial will discuss the possibilities and limitations of observational databases. First, we will discuss the design of these systems (issues such as privacy and the collection of additional data). Secondly, we will focus on the question whether the data in these observation databases is sufficiently reliable for research. Thirdly, we will discuss both possibilities and limitations of these databases by analyzing a number of studies that were performed using observational databases. The examples we will discuss will focus in particular on the area of post-marketing surveillance.