Mathematical modelling for vaccine efficacy trials during the future epidemics of emerging respiratory infections

Assessing vaccine efficacy (VE) during emerging epidemics is challenging due to unpredictable disease transmission dynamics. We aimed to investigate the impact of vaccine randomized controlled trials (RCTs) timing on estimates of VE and sample sizes during future epidemics of emerging respiratory diseases. We developed an age-structured susceptible-exposed-infected-asymptomatically infected-removed (SEIAR) compartment models using 2022 Korean population, and COVID-19 and 2009 A/H1N1 pandemic influenza parameters. Various RCT scenarios were tested to calculate VE estimates, sample size and power by varying RCT timings (using the epidemic peak as the base, $\pm$10%, $\pm$20%, $\pm$30% relative to the time of peak) with follow-up durations (4 weeks as the base, and 8 and 12 weeks), recruitment durations (4 weeks as the base, and 2, 8, and 12 weeks), and non-pharmaceutical intervention (NPI) levels in reducing R₀ by 10% and 20%. Additionally, assumptions regarding baseline cumulative incidences were evaluated for sample size calculations. The results showed that VE remained relatively stable across trial timings; however, required sample sizes varied significantly with timing. Sample size requirements initially decreased after a peak and then increased steeply as the epidemic progressed. Initiating RCTs 30% earlier than the peak, along with extended recruitment duration, could reduce sample sizes without compromising VE. NPIs effectively extended the feasible timeframe for RCTs. Sample size estimates based on simulated case numbers in the placebo group were slightly underestimated, with power consistently above 85%. In contrast, calculations using cumulative incidence over the 4 weeks pretrial or the entire study duration could lead to overpowered or underpowered studies.