Leveraging datasets and insisting on quality to address unmet research needs—the epidemiologists’ dream

By aggregating data from large numbers of heterogeneous patients managed in everyday routine care, electronic health records can be used to explore clinical realities and related outcomes to wide effect. They can be used to evaluate current care practices, examine differential safety and effectiveness treatments (and across key patient subgroups of interest) and, at an international level, to compare and contrast system-based outcomes, demonstrate differences in disease prevalence (between regions and countries) and possible differences in etiology. As recognition of the potential value of database studies grows, there is a parallel need to raise awareness of what constitutes quality in database research. The robustness of database research is not only impingent upon high quality database research methodologies and selection of the appropriate data source for the question at hand, but also on the very quality of the data contained available (e.g., variables, the completeness and quality of data). This presentation will review some of the strengths and limitations of databases for medical research and offer recommendations for how to optimize quality at the development/design stage and opportunities to support high quality data entry.