The data giveth, but what do we take away?

Abstract

The data giveth, but what do we take away?News media, popular periodicals, and scientific literature describe the growth in health-related data as explosive. 1 As the availability of data grows, and methods to protect the privacy and improve sharing continue to develop, clinical research and epidemiologic studies related to pregnancy and perinatal outcomes become increasingly easier to conduct. Pair these growing resources with developments in data linkage methods, as well as the simultaneously explosive uptake in data science, and the possibilities are profound. And while these developments may lead to more straightforward analyses and insights into other fields, reproductive and perinatal epidemiology lends itself to nuances that may require special considerations.Given the reality of more data, tools, and computational power, we assembled this special issue of Paediatric and Perinatal Epidemiology touching on three major areas in research, and we highlight approaches, considerations, and some limitations to this work.The first section includes data source profiles from two countries (Mexico and Sweden). The second section considers issues around data fidelity. In other words, are we studying what we intended to study (or to what extent do we have accuracy and representativeness in our data of what we expect or anticipate it to represent)? And finally, the third section highlights data analysis, where authors discuss approaches such as Mendelian randomisation and instrumental variables, machine learning, recalibration of prognostic models, and mediational g-formula methods. We discuss these issues in more detail.