Using aggregated site features computed with the MPsQ to allow a better monitoring of sites in clinical trials
The MPsQ is an assessment characterizing the profile of the subjects participating to a Randomized Controlled Trial (RCT) to predict their behaviors in the study (placebo response, adherence, …). This analysis aimed to determine whether individual responses collected through the MPsQ could serve for characterizing clinical sites in RCTs. The objective is to enhance study data comprehension by profiling clinical site.
Historical data from five diverse RCTs across varying medical indications (Chronic Pain, Degenerative Disease, and Ophthalmology) were employed. Analysis was restricted to clinical sites with a minimum of 10 subjects, resulting in a dataset comprising 1164 subjects distributed across 56 clinical sites. For this first analysis, two site scores were aggregated using the subjects answers to MPsQ—namely, subjects’ expectations of improvement and interaction with clinical sites.
The analysis showed significant correlations between subject scores and the scores of their associated sites: 36% (95% CI: [0.31, 0.40]) for Expectations and 20% (95% CI: [0.13, 0.27]) for Interactions. Leveraging the 2 aggregated site scores allowed also to characterize other site characteristics with, e.g., the explanation of approximately 31% of the variance in drop-out rates across individual sites.
This analysis underscores the potential of MPsQ features to effectively profile clinical sites in RCTs. These profiles, reflective of intrinsic site attributes, demonstrably impact subject behaviors. Moreover, these profiles can explain disparities in subject behaviors across different sites. Characterizing the sites could improve the post-hoc data understanding and the conduct of studies through relevant feedback.