Pragmatic Randomized Trials Using Claims or Electronic Health Record Data

Abstract

Randomized clinical trials have been the accepted standard for addressing key questions in medicine for well over 60 years. The structure and process, while well documented and characterized, have been historically described in the context of “efficacy” in a targeted population as opposed to “effectiveness” objectives in the greater population. Efficacy can be defined as the performance of an intervention under ideal and controlled circumstances, whereas effectiveness refers to its performance under “real-world” conditions. Another way to think about an effectiveness trial is that it often tests a treatment strategy or “policy” of applying an intervention as opposed to the actual action of the intervention itself. Trials looking at real-world outcomes may have some aspects of greater variability in the data and therefore the precision of the question and rigor of the hypothesis are as essential as they are in any trial. Therefore, a key component of implementing a pragmatic clinical trial (PCT) is using data that one can more easily obtain (i.e., pragmatically). In recent years, one solution to this has been to use Electronic Health Record (EHR) and/or administrative billing (claims) data. These data fall under the larger and growing rubric of “real-world data,” roughly defined as data that are not primarily collected for research purposes. Instead, EHR and claims data are generated as a product of a patient’s healthcare encounters. As such, from a researcher’s perspective, there appears to be minimal cost with using these data for research purposes. However, the trade-off is that there is often minimal control of which data are collected and how.

In this chapter, we discuss the characteristics of these data sources, their usage within pragmatic trials, and considerations that can be challenging to practitioners’research.