You are viewing the site in preview mode

Skip to main content
Figure 1 | Emerging Themes in Epidemiology

Figure 1

From: Hard, harder, hardest: principal stratification, statistical identifiability, and the inherent difficulty of finding surrogate endpoints

Figure 1

Schematic of a vaccine trial design incorporating closeout placebo vaccination. The four horizontal lines represent four subjects (two each assigned to placebo and active vaccine at time T=0). Subjects have biomarker S obs measured at time τ, identifying S0 and S1 for placebo and vaccine recipients, respectively; the counterfactual biomarker values (S1 and S0 respectively) remain unidentified, as indicated by ‘?’. Subjects with the solid diamond represent those infected during the trial (yielding Y0=1 and Y1=1). Subjects assigned to placebo and uninfected during the trial (i.e., those with Y0=0) are eligible for closeout vaccination at the end of the study at time T=t. Post-closeout vaccination biomarker measurements S closeout are obtained on these subjects at time T=t+τ. The dashed curved arrow represents the “time constancy” assumption that allows S closeout to be used to identify S1, by definition the counterfactual measurement that would have been obtained at T=τ under assignment to the active vaccine. For ease of readability, this figure does not represent subjects who were infected prior to τ (i.e., those with Aτ=0) and who would have Y=1 and S undefined.

Back to article page