Optimal allocation strategies in platform trials

TL;DR

This paper derives optimal patient allocation strategies for platform trials with changing treatment arms, considering shared controls and different analysis methods, to minimize estimator variance and improve trial efficiency.

Contribution

It introduces a novel method for determining optimal allocations in dynamic platform trials, accounting for entry times and analysis strategies, which differs from traditional fixed-ratio approaches.

Findings

Optimal allocations depend on arm entry times.

The optimal solution often deviates from the classical square root rule.

Case study shows improved efficiency and error control.

Abstract

Platform trials are randomized clinical trials that allow simultaneous comparison of multiple interventions, usually against a common control. Arms to test experimental interventions may enter and leave the platform over time. This implies that the number of experimental intervention arms in the trial may change over time. Determining optimal allocation rates to allocate patients to the treatment and control arms in platform trials is challenging because the change in treatment arms implies that also the optimal allocation rates will change when treatments enter or leave the platform. In addition, the optimal allocation depends on the analysis strategy used. In this paper, we derive optimal treatment allocation rates for platform trials with shared controls, assuming that a stratified estimation and testing procedure based on a regression model, is used to adjust for time trends. We consider both, analysis using concurrent controls only as well as analysis methods based on also non-concurrent controls and assume that the total sample size is fixed. The objective function to be minimized is the maximum of the variances of the effect estimators. We show that the optimal solution depends on the entry time of the arms in the trial and, in general, does not correspond to the square root of $k$ allocation rule used in the classical multi-arm trials. We illustrate the optimal allocation and evaluate the power and type 1 error rate compared to trials using one-to-one and square root of $k$ allocations by means of a case study.