Linkage of Administrative Datasets: Enhancing Longitudinal Epidemiological Studies in the Era of “Big Data”

Abstract“Modern epidemiology” has consolidated the direct collection of individual data as the most valued approach for conducting epidemiological research. An essential feature of powerful epidemiological studies (in whatever design, observational, quasi-experimental or experimental) is a longitudinal structure, so that in the course of the study, data are collected over time and measurements can be repeated for each participant. Notably, the amount and variety of individual health data routinely collected from different sources and available in digital media have increased exponentially. This growing amount of data has caused scientific disciplines to confront essential challenges in operational (data management, infrastructure, training), methodological (new approaches to analyze and to derive inferences from “big data”), and epistemological (several argue that the hypothesis-driven science is outdated, and we live now in a data-driven era) realms. There is no doubt that the use of large administrative databases in particular when enriched through linkage with other sources of data, while in its infancy, is a powerful tool with the potential to bolster medical and epidemiological longitudinal research. Being relatively fast and low cost, it can enable the study of essential research questions previously unfeasible for among others, budgetary, or ethical reasons.