Skip to main content

Table 1 Prentice framework simulation results

From: Assessing correlates of protection in vaccine trials: statistical solutions in the context of high vaccine efficacy

logit model 4 \(\hat {VE}\) p(S)<α p(Z)<α p(S)<α & p(Z)≥α
Linear 0.41 1.00 0.05 0.95
Non-linear 0.41 1.00 0.05 0.95
Scaled logit 0.41 1.00 0.03 0.96
Linear 0.75 1.00 0.10 0.90
Non-linear 0.75 1.00 0.04 0.96
Scaled logit 0.75 1.00 0.04 0.96
Linear 0.86 1.00 0.22 0.78
Non-linear 0.86 1.00 0.05 0.95
Scaled logit 0.86 1.00 0.04 0.96
Linear 0.96 1.00 0.34 0.66
Non-linear 0.96 1.00 0.04 0.96
Scaled logit 0.96 0.99 0.03 0.96
  1. Power (α=0.05) to assess Prentice criterion 4 using classical (linear) and flexible (non-linear) model 4 in case of full mediation (data generated using scaled logit model 3). \(\hat {VE}\): estimated VE; p(S)<α: power to detect the Surrogate effect; p(Z)<α: type-I error of the treatment effect; p(S)<α & p(Z)≥α: power to meet Prentice criterion 4