Right sizing for vaccine effectiveness studies: how many is enough for reliable estimation?

Area of Science:

• Epidemiology
• Public Health
• Immunology

Background:

• Vaccine effectiveness (VE) estimates rely on sample size and sampling methods.
• Sentinel surveillance in Victoria involves general practitioners (GPs) collecting respiratory samples from influenza-like illness (ILI) patients.
• Resource limitations necessitate determining the minimum sample size for reliable VE estimates.

Purpose of the Study:

• To assess the impact of reduced swab rates on VE estimates.
• To determine the minimum sample size required for precise VE estimates.
• To evaluate the representativeness of GP sampling across different age groups.

Main Methods:

• Utilized a test-negative design with ILI patients recruited by GPs and tested for influenza.
• Calculated VE using logistic regression, adjusted for presentation week and age.
• Assessed the effect of swab rates by analyzing random 20% and 50% samples without replacement.

Main Results:

• GPs swabbed a lower proportion of patients aged ≥65 years (45.9%) compared to other age groups.
• Decreasing swab rates did not significantly alter VE point estimates.
• Reduced swab rates diminished the precision of VE estimates and, in some cases, led to insufficient sample sizes.

Conclusions:

• Arbitrary swabbing rates (e.g., 20%, 50%) yield less robust VE estimates.
• The required number of swabs should align with seasonal severity, not fixed rates.
• Systematic patient swabbing by age group is recommended for sufficient data to investigate VE across age demographics.