The Epistemology of Extrapolation: When Can Pragmatic Trial Evidence Travel?

Abstract: Pragmatic Clinical Trials (PCTs) are often seen as a solution to the external validity limitations of traditional Randomized Controlled Trials (RCTs). RCTs prioritize internal validity but often lack applicability to real-world clinical settings. PCTs, in contrast, aim to assess real-world effectiveness and are assumed to be more generalizable. However, this assumption lacks clear epistemological justification. Philosophers like Cartwright (2009, 2011) and Reiss (2019) argue that effectiveness is highly context-dependent, and recent critiques (e.g., Tresker 2022) highlight the limitations of the explanatory/pragmatic trial distinction. Despite these critiques, no structured framework exists for assessing when PCT evidence can be extrapolated. This paper addresses that gap by developing an epistemology of extrapolation, arguing that generalizability is best understood as a spectrum rather than a binary property. I propose a three-dimensional framework for evaluating PCT extrapolation, based on (1) causal similarity (whether the intervention interacts with patients similarly across contexts), (2) contextual similarity (whether healthcare infrastructure, clinician behavior, and patient adherence align), and (3) implementation feasibility (whether the necessary structural conditions exist to reproduce the intervention’s effects). I apply this framework to the PREMIER and ADAPTABLE trials, demonstrating the risks of overgeneralization when pragmatic evidence fails to travel despite its real-world design. By offering a new philosophical framework for generalizability, this paper provides practical guidance for policymakers, trial designers, and clinicians translating pragmatic evidence into practice.

Keywords: Extrapolation of Clinical Trial Evidence, Generalizability in Pragmatic Trials, Philosophy of Evidence in Medicine