Causal Vaccine Effects on Binary Postinfection Outcomes.

Area of Science:

• Epidemiology
• Biostatistics
• Vaccinology

Background:

• Prophylactic vaccination's impact on postinfection outcomes (disease, death, transmission) is crucial for public health.
• Evaluating vaccine effects conditional on infection can introduce selection bias, complicating causal interpretation.
• Existing methods may not accurately capture the true causal effect of vaccines on postinfection health.

Purpose of the Study:

• To define and estimate a causal vaccine efficacy estimand for binary postinfection outcomes within the principal stratification framework.
• To address the identifiability challenges of causal vaccine effects conditional on post-treatment infection.
• To propose and derive statistical models and estimators for causal vaccine effects on postinfection outcomes.

Main Methods:

• Utilized the principal stratification framework to define a causal estimand for vaccine effects on postinfection outcomes.
• Developed selection models to address identifiability issues under standard assumptions.
• Derived closed-form maximum likelihood estimators (MLEs), including scenarios with maximum selection bias.
• Compared the proposed MLE with commonly used estimators like the intent-to-treat (ITT) estimator.

Main Results:

• The causal postinfection vaccine efficacy estimand is not identifiable under standard assumptions, necessitating selection models.
• Maximum likelihood estimators were derived, providing a method to estimate the causal effect.
• The ITT estimator serves as an upper bound for the causal postinfection effect under certain conditions.
• Analysis of rotavirus and pertussis vaccine trials demonstrated the application of the methods.

Conclusions:

• Pertussis vaccination was shown to have a significant causal effect in reducing disease severity.
• The proposed methods offer a robust approach to estimating causal vaccine effects on postinfection outcomes, accounting for selection bias.
• Accurate causal inference is essential for understanding the full impact of vaccination programs.