The association between vaccine coverage and herd protection: exploratory analyses of a cluster-randomised trial of Vi conjugate vaccine.

Background: Indirect protection from the typhoid conjugate vaccine (TCV) has been negligible, possibly due to vaccination being restricted to children, which may have limited the overall vaccine coverage needed to sufficiently interrupt transmission dynamics across endemic populations. In this study herd protection was evaluated in households and schools with different TCV coverages by conducting secondary analyses of a double-blind cluster-randomised trial on TCV in Bangladesh. Methods: TyVAC Bangladesh was a cluster-randomised trial conducted in Dhaka, Bangladesh between April 2018 and March 2020 (ISRCTN11643110 at www.isrctn.com). For this study, 150 geographical clusters were randomly assigned to receive TCV or Japanese encephalitis (JE) vaccine. Children aged 9 months to <16 years were vaccinated. Household-level herd protection was estimated by indirect vaccine effectiveness (iVE), comparing typhoid incidence among residents of all ages who did not receive the assigned study vaccine in TCV clusters with those in JE clusters using mixed-effects Poisson regression. The iVE was calculated as (1 - adjusted incidence rate ratio [aIRR]) × 100%. Household-level coverage was stratified by the number of vaccinated children (0, 1, ≥2) and the presence of eligible children. School-level herd protection was evaluated among students attending 35 primary and secondary schools who did not receive TCV by comparing typhoid incidence in high-coverage (>median) versus low-coverage schools (≤median), where school-level coverage was measured by the proportion of TCV-vaccinated students. The aIRR was estimated by mixed-effects Poisson regression. Findings: Household-level analysis included 129,601 unvaccinated residents from 40,435 households in the TCV arm and 129,973 unvaccinated residents from 40,577 households in the JE arm (median household size: 4.0 [IQR: 3.0, 5.0]). During 24-month follow-up between April 2018 and March 2020, the overall aIRR comparing all unvaccinated residents in TCV versus JE clusters was 0.91 [95% CI: 0.43, 1.91]. Stratified by household vaccine coverage, aIRR was 0.81 [95% CI: 0.15, 4.52] in households without eligible children and 0.97 [0.31, 3.04] in households with ≥ 1 eligible children but none vaccinated, and iVE was 25% [95% CI: -234, 83] and -26% [-766, 82] in households with ≥ 1 eligible children and 1, or ≥ 2 vaccinated children, respectively, with no significant interaction between number of vaccinees and arm (p = 0.85). School-level analysis (9209 children, 8%-55% TCV coverage) showed typhoid incidence of 349 [95% CI: 219, 528] and 329 [180, 552] per 100,000 person-years among non-TCV vaccinated students in low- and high-coverage schools, respectively (aIRR: 0.96 [95% CI: 0.28, 3.37]). Interpretation: Our study did not find significant herd protection across vaccine coverage levels in households and schools, although estimates were imprecise with wide confidence intervals. Funding: Gates Foundation.