Response adaptive intervention allocation in stepped-wedge cluster randomized trials.

Area of Science:

• Clinical Trials Methodology
• Biostatistics
• Public Health Research

Background:

• Stepped-wedge cluster randomized trials (SW-CRT) are often implemented with a pre-determined plan to offer interventions to all clusters, raising ethical questions about equipoise.
• Response adaptive (RA) designs offer a parallel solution in pharmaceutical development, enabling treatment adjustments based on observed patient responses.

Purpose of the Study:

• To introduce a novel framework for incorporating response adaptive (RA) designs within stepped-wedge cluster randomized trials (SW-CRT).
• To enable dynamic modification of intervention allocation during SW-CRTs, balancing statistical power with patient benefit.

Main Methods:

• A simulation study was conducted to evaluate the proposed methodology for integrating RA designs into SW-CRTs.
• The framework allows for adaptive adjustments to intervention allocation based on emerging data.

Main Results:

• In simulations, one RA design saw the proportion of cluster-periods in the intervention condition increase from 32.2% to 67.9% with increasing intervention effect.
• This adaptability came at the cost of a 6.2% power reduction compared to a fixed-allocation design (45.0% intervention time).

Conclusions:

• RA designs are particularly suitable for SW-CRTs where interventions offer significant individual or societal benefits, or have safety concerns.
• The proposed methodology provides adaptability in intervention roll-out speed with a potentially small impact on study power.