Confidence interval estimation for vaccine efficacy against COVID-19

Area of Science:

• Biostatistics
• Epidemiology
• Infectious Diseases

Background:

• Accurate estimation of vaccine efficacy is crucial for public health.
• Confidence intervals provide a range of plausible values for vaccine efficacy.
• Fixed number of events designs are commonly used in clinical trials.

Purpose of the Study:

• To construct and compare confidence intervals for vaccine efficacy against COVID-19.
• To evaluate the performance of five different statistical methods.
• To assess the impact of diagnostic test under-sensitivity on efficacy estimates.

Main Methods:

• Comparison of five confidence interval construction methods.
• Evaluation based on coverage probability, non-coverage probability, and interval width.
• Analysis of diagnostic test sensitivity's effect on vaccine efficacy.

Main Results:

• Most methods, except the exact conditional, showed coverage probabilities exceeding the 5% significance level.
• Bayesian, approximate Poisson, and mid-P methods offered narrower intervals but with coverage instability.
• Under-sensitivity in diagnostic tests led to overestimated vaccine efficacy.

Conclusions:

• The exact conditional method is preferred for COVID-19 vaccine efficacy trials with >= 60 events.
• The mid-P method is a viable alternative for narrower intervals when case counts are lower.
• Accurate diagnostic test sensitivity is vital to avoid biased vaccine efficacy estimates.