Infectious Disease
Target Trial Emulation in Vaccine Evaluation: Challenges and Opportunities Kayoko Shioda* Kanako Fuyama Fuyama Fuyama Fuyama Boston University School of Public Health
Target trial emulation (TTE) is a causal inference framework that emulates a randomized trial using observational data. Although TTE has been increasingly applied in vaccine evaluations, infectious disease outcomes often violate a key causal inference assumption: no interference. The purpose of this simulation study is to assess the performance of conventional TTE using the clone-censor-weight (CCW) approach in the presence of interference. We compared three dosing schedules: a single-dose protocol; a two-dose protocol administered at the recommended interval (second dose given 28 days after the first); and a delayed two-dose protocol (41 days), assuming that the delayed protocol produced the highest protection. Using an agent-based transmission model, we simulated transmission in a population of 1 million individuals with mixed vaccination protocols and estimated cumulative infection risks using CCW, while true counterfactual risks were derived from simulations in which all vaccinated individuals followed the same vaccination schedule. We considered scenarios with varying levels of vaccine coverage, adherence to dosing schedules, and timing of the first dose. Although discrepancies were observed in cumulative risks and risk ratios between the simulated counterfactuals and CCW estimates, both indicated that the delayed protocol yielded the lowest cumulative risk of infection. However, in scenarios where adherence to the dosing protocols varied over time, CCW using Cox proportional hazards regression for inverse probability of censoring weighting did not capture this trend, whereas the Kaplan-Meier method and stratified Cox regression did. This discrepancy may be explained by violations of the proportional hazards assumption induced by time-varying adherence. Overall, our analysis suggests that CCW may be able to identify the superior dosing schedule; however, caution is warranted when interpreting the estimated cumulative infection risks under each dosing schedule.
