Fig. 5From: Cohort Multiple Randomised Controlled Trials (cmRCT) design: efficient but biased? A simulation study to evaluate the feasibility of the Cluster cmRCT designPercentage bias and power of the four analysis methods for varying levels of patient refusal and correlation between individual patient refusal probabilities and risk. Clinician refusal = 0, correlation is negative, recruitment method 2 is used and ICC =0.025. The black line in the power graph represents the expected power in the trial. A robust marginal frailty model is fitted to the dataBack to article page