Understanding how to innovate in clinical trial design
WASHINGTON, D.C.—Randomized clinical trials (RCTs) are the gold standard for testing medical interventions. However, researchers are using multiple new and creative ways to design these trials to account for real-world scenarios and produce information relevant to practicing clinicians. At ACR Convergence 2024, the session Innovative Clinical Trials: Precision, Design and Optimization provided tremendous insights into these important topics.
Heterogeneity of Treatment Effects
The first speaker, Kelli D. Allen, PhD, professor of medicine, research health scientist, Durham VA Medical Center, University of North Carolina at Chapel Hill, spoke about intervention optimization and heterogeneity of treatment effects (HTE). HTE refers to when a treatment has different effects on different individuals or subgroups within a study population. Quantitative subgroup interactions refers to a large improvement for some patients and little to no improvement for others. Qualitative subgroup interactions describes when one treatment may work better for one subgroup and the other treatment may be better for another subgroup.
Dr. Allen noted that such heterogeneity is common in clinical trials, and, knowing that often groups of patients do not respond to a specific intervention or treatment, researchers ought to plan for what they should do for these patients. For this, researchers can use adaptive interventions. An example of this approach is from a study conducted by Dr. Allen et al. on a stepped exercise program for patients with knee osteoarthritis (OA). The initial intervention was an internet-based exercise program, and participants were assessed for clinically relevant improvement in pain and function after three months. For those patients who did not benefit, telephone-based physical activity coaching was added to the exercise program. If still no benefit was seen with this additional intervention after three months, then physical therapy visits were implemented.1
Dr. Allen explained that HTE can affect analysis of results, and this issue can be addressed through different strategies meant to provide individual predictions of treatment effects that take into account multiple relevant characteristics simultaneously. Risk modeling involves developing a multivariable model that predicts risk for an outcome and applies that model to stratify patients within a trial to examine risk-based variation in treatment effects. Effect modeling uses a regression model developed from trial data to understand how much of the average effect is modified by baseline predictors.
Data-driven approaches are also used, and these employ technologies, such as machine learning, that take advantage of all the rich data collected in a trial. Dr. Allen referenced the qualitative interaction tree (QUINT) method that can identify subgroups of study patients for whom treatment effect differs significantly based on the characteristics of those patients. Using QUINTS may both help identify which patients are most likely to benefit from a specific intervention based on characteristics unique to them, such as body mass index in patients with knee OA, and may help with the design of future clinical trials, where patients may be randomized based on specific characteristics.