Predictive Modeling of Heterogeneous Treatment Effects in RCTs: A Scoping Review.

Area of Science:

• Clinical Epidemiology
• Biostatistics
• Health Services Research

Background:

• The 2020 Predictive Approaches to Treatment Effect Heterogeneity (PATH) Statement introduced predictive modeling for identifying heterogeneity in treatment effects (HTE) in randomized clinical trials (RCTs).
• Two primary approaches were proposed: risk modeling and effect modeling.
• Risk modeling predicts individual baseline risk, examining treatment effects across risk strata, while effect modeling directly predicts individual treatment effects.

Purpose of the Study:

• To identify, describe, and evaluate findings from studies that cited the PATH Statement.
• To assess the application and credibility of predictive modeling for HTE in RCTs.
• To determine the clinical importance of identified HTE.

Main Methods:

• A comprehensive literature search was conducted across PubMed, Google Scholar, Web of Science, and SCOPUS.
• Reports were systematically reviewed using a double-review process with adjudication.
• Credibility and clinical importance of HTE findings were assessed using established criteria.

Main Results:

• 65 reports analyzing 162 RCTs were identified, with 31 using risk models and 41 using effect models.
• Credible, clinically important HTE was found in 37% of the reviewed reports.
• Risk modeling demonstrated higher credibility (87%) compared to effect modeling (32%), with external validation enhancing the credibility of effect models.

Conclusions:

• Multivariable predictive modeling identified credible and clinically important HTE in 37% of reports on RCTs.
• Risk modeling was more consistently credible for identifying HTE than effect modeling.
• External validation is critical for establishing the credibility of findings from exploratory effect models.