Vaccine efficacy estimands and power considerations.

Area of Science:

• Biostatistics
• Clinical Trials
• Epidemiology

Background:

• The ICH E9(R1) addendum emphasizes pre-specifying clinically interpretable estimands and handling intercurrent events.
• Vaccine efficacy trials require robust treatment effect measures, especially when proportional hazard assumptions are violated.

Purpose of the Study:

• To evaluate and compare the statistical power of different vaccine efficacy estimands.
• To assess estimands interpretable beyond the proportional hazard assumption in prophylactic vaccine trials.

Main Methods:

• Comparison of estimands including average hazard ratio, cumulative incidence ratio, and restricted mean time lost ratio.
• Power analysis applied to various estimands in the context of vaccine efficacy.
• Illustration using data from a human papillomavirus phase III vaccine trial.

Main Results:

• Average hazard ratio and cumulative incidence ratio demonstrate comparable power in classical prophylactic vaccine efficacy settings.
• Restricted mean time lost ratio is found to be less powerful compared to the other estimands.
• All considered estimands provide interpretable results even when the proportional hazard assumption is not met.

Conclusions:

• The choice of estimand in vaccine efficacy trials significantly impacts statistical power.
• Average hazard ratio and cumulative incidence ratio are often preferred for their power and interpretability.
• Restricted mean time lost ratio may be less suitable for prophylactic vaccine efficacy trials due to lower power.