OBJECTIVES: Pragmatic clinical trials (PCTs) are interventional studies that incorporate design elements which reflect how the investigational therapy would be applied under real-life circumstances. For example, they recruit patients based on relatively few selection criteria, employ flexibility in the delivery of and adherence to the allocated treatment, and avoid subjecting participants to any follow-up procedures that go beyond standard practice. As such, PCTs are considered to be sources of real-world evidence. However, since the PCT label is mostly self-assigned, it may be inappropriately used by researchers. The objective of this study was to evaluate whether oncology trials that call themselves pragmatic are actually pragmatic.

METHODS: We conducted a systematic review of cancer clinical studies that labelled themselves as pragmatic and for which we could find publications in PubMed and Embase, with trials focusing on systems interventions or complementary therapies being excluded from further analysis. Subsequently, we used the validated PRECIS-2 tool to assess the degree of pragmatism exhibited by the included studies, based on scores from one to five given across nine different domains by a team of reviewers blinded to each other’s work.

RESULTS: The initial search returned 803 unique articles, of which 38 were retained upon conclusion of the screening process. The trials in question encompassed 12 different cancer types and were mostly randomized (71.0%), open-label (84.2%) studies. The majority (65.8%) were initiated rather recently, having had their results published in the last five years. The average overall PRECIS-2 score of the examined trials was 3.2 (standard deviation: 0.6) and the median score was 3.3 (interquartile range: 2.9-3.6). For nearly all (97.4%) studies, one or more PRECIS-2 domains could not be appraised due to a lack of information provided in the corresponding publications.

CONCLUSIONS: Cancer trials that describe themselves as pragmatic are generally not very pragmatic in practice.