The design of an adaptive clinical trial to evaluate the efficacy of platelets stored at low temperature in surgical patients.

Abstract

Storage of platelets at 4°C compared with 22°C may increase both hemostatic activity and storage duration; however, the maximum duration of cold storage is unknown. We report the design of an innovative, prospective, randomized, Bayesian adaptive, "duration finding" clinical trial to evaluate the efficacy and maximum duration of storage of platelets at 4°C. Patients undergoing cardiac surgery and requiring platelet transfusions will be enrolled. Patients will be randomized to receive platelets stored at 22°C up to 5 days or platelets stored at 4°C up to 5 days, 10 days, or 15 days. Longer durations of cold storage will only be used if shorter durations at 4°C appear noninferior to standard storage, based on a four-level clinical hemostatic efficacy score with a NIM of a half level. A Bayesian linear model is used to estimate the hemostatic efficacy of platelet transfusions based on the actual duration of storage at 4°C. The type I error rate, if platelets stored at 4°C are inferior, is 0.0247 with an 82% probability of early stopping for futility. With a maximum sample size of 1,500, the adaptive trial design has a power of over 90% to detect noninferiority and a high probability of correctly identifying the maximum duration of storage at 4°C that is noninferior to 22°C. An adaptive, duration-finding trial design will generate Level I evidence and allow the determination of the maximum duration platelet storage at 4°C that is noninferior to standard storage at 22°C, with respect to hemostatic efficacy. The adaptive trial design helps to ensure that longer cold storage durations are only explored once substantial supportive data are available for the shorter duration(s) and that the trial stops early if continuation is likely to be futile.