Integrating a pilot study and large trial into one adaptive design

Adaptive clinical trial designs are widely adopted in industry-sponsored trials but remain limited in NIH-funded public health research due to regulatory and funding challenges. This manuscript presents the Nutricity study as a proposed framework for implementing a seamless phase II/III trial design by integrating an NIH-funded pilot trial with a potential confirmatory trial into a single adaptive design to evaluate a pediatric nutrition intervention.

The Nutricity study proposes a seamless design combining a phase II pilot (NIH COBRE program funded) with a phase III trial (pending R01 grant application). This randomized trial focuses on changes in child diet quality measured by HEI scores. Simulations compared operation characteristics including sample size, study duration, and statistical power, between the seamless design and traditional two-stage approaches across various effect scenarios.

Simulations demonstrated that the proposed seamless design achieved a 37 % sample size reduction and a 34 % reduction in duration, while maintaining high probability of success (99.4 %) when the effect size was as expected. Under the null scenario, the seamless design maintained an empirically estimated Type I error rate of 5.047 %. Across other scenarios with similar resource utilization, it enhanced efficiency through early stopping for futility or improved power compared to traditional approaches.

The Nutricity study illustrates the potential of seamless phase II/III adaptive designs to improve efficiency in public health research, by integrating pilot and confirmatory phases while preserving scientific integrity. This proposed approach offers a valuable model for future NIH-funded studies adopting innovative trial methodologies.