Abstract Background and aims Advanced trial design features are increasingly used in stroke. The MSU Telemedicine trial is a recent example, demonstrating that on a Mobile Stroke Unit (MSU), a telemedicine model of care improved resource efficiency with minimal delays in clinical decision-making and no difference in safety. We provide an in-depth discussion of how advanced trial design features can be combined to effectively address research questions in stroke, using MSU Telemedicine as a case study. Methods MSU Telemedicine was a covariate-adjusted, cluster-randomised clinical trial with the Win Odds as its primary outcome measure. It had three key features: use of the “Win Odds” to address a multifaceted research question; covariate-adaptive randomisation to prospectively balance prognostic covariates; and a cluster-randomised design, which facilitated randomisation at the daily level, as individual participant randomisation was impractical. Results MSU Telemedicine achieved this combination of features by applying a novel modification to Simon and Pocock’s Biased Coin Algorithm that balanced the number of participants seen by nurses with different levels of experience using the anticipated number of participants that would be seen each day. To account for this randomisation procedure, p-values for the Win Odds were estimated using re-randomisation tests in accordance with FDA guidelines on adaptive randomisation, while confidence intervals were estimated using clustered bootstrap that accounted for daily clustering. Conclusions The combined use of advanced trial design features equips stroke researchers to improve stroke care. This combined approach is achievable in practice, as evidenced by the MSU Telemedicine Trial. Conflict of interest HJ: Nothing to disclose. VY: Nothing to disclose. AB: Nothing to disclose. BC: Nothing to disclose. SD: Nothing to disclose. GD: Nothing to disclose. LC: Nothing to disclose.
Johns et al. (Fri,) studied this question.