Big data: Using databases and registries

Abstract

The field of vascular surgery is in constant evolution. Administrative data and registries can provide important contemporary evidence to inform clinical decision making and delivery of health services. This review outlines some important considerations for retrospective studies using administrative health databases and registries. First, these data sources have advantages (e.g., real-world applicability, timely data access, and relatively lower research cost) and disadvantages (e.g., potential missing data, selection bias, and confounding bias) that may be more or less relevant to different administrative databases or registries. Second, a framework to guide data source selection and provide a summary of frequently used data sources in vascular surgery research is discussed. Third, a retrospective study design warrants planned exposure, outcome, and covariate definitions and, when studying an exposure-outcome association, careful consideration of confounders through direct acyclic graphs. Finally, investigators must plan the most appropriate analytic approach, and we distinguish descriptive, explanatory, and predictive analyses.