An adaptive two-arm clinical trial using early endpoints to inform decision making: Design for a study of sub-acromial spacers for repair of rotator cuff tendon tears

Abstract

Background: There is widespread concern across the clinical and research communities that clinical trials, powered for patient-reported outcomes, testing new surgical procedures are often expensive and time-consuming, particularly when the new intervention is shown to be no better than the standard. Conventional (non-adaptive) randomised controlled trials (RCTs) are perceived as being particularly inefficient in this setting. Therefore, we have developed an adaptive group sequential design that allows early endpoints to inform decision making and show, through simulations and a worked example, that these designs are feasible and often preferable to conventional non-adaptive designs. The methodology is motivated by an ongoing clinical trial investigating a saline-filled balloon, inserted above the main joint of the shoulder at the end of arthroscopic debridement, for treatment of tears of rotor cuff tendons. This research question and setting is typical of many studies undertaken to assess new surgical procedures. Methods: Test statistics are presented based on the setting of two early outcomes, and methods for estimation of sequential stopping boundaries are described. A framework for the implementation of simulations to evaluate design characteristics is also described. Results: Simulations show that designs with one, two and three early looks are feasible and, with appropriately chosen futility stopping boundaries, have appealing design characteristics. A number of possible design options are described that have good power and a high probability of stopping for futility if there is no evidence of a treatment effect at early looks. A worked example, with code in R, provides a practical demonstration of how the design might work in a real study. Conclusions: In summary, we show that adaptive designs are feasible and could work in practice. We describe the operating characteristics of the designs and provide guidelines for appropriate values for the stopping boundaries for the START:REACTS (Sub-acromial spacer for Tears Affecting Rotator cuff Tendons: a Randomised, Efficient, Adaptive Clinical Trial in Surgery) study.