Assessment of vaccine herd protection in a cluster-randomised trial of Vi conjugate vaccine against typhoid fever: results of further analysis
Khanam F., Kim DR., Liu X., Voysey M., Pitzer VE., Zaman K., Pollard AJ., Qadri F., Clemens JD.
Background: A cluster-randomised trial of Vi-tetanus toxoid (Vi-TT) conjugate vaccine conducted in urban Bangladeshi children found a high level of direct protection by Vi-TT but no significant vaccine herd protection. We reassessed the trial using a “fried egg” analysis to evaluate whether herd protection might have been obscured by transmission of typhoid into the clusters from the outside. Methods: A participant- and observer-blind, cluster-randomised trial was conducted between February 14, 2018 and August 12, 2019 in three wards of Mirpur, a densely populated urban area of Dhaka, Bangladesh. Children 9 months to under 16 years of age in 150 geographic clusters, which had a total of 311,289 persons present at baseline or entering during follow-up, were randomised by cluster to a single-dose of Vi-TT or Japanese encephalitis (JE) vaccine. Vi-TT protection against typhoid fever, detected at 8 treatment centres serving the study population, was compared in the original clusters for the trial, and for progressively more central subclusters (“yolks” of the “fried egg”) of the cluster residents. If transmission of typhoid into the clusters had diluted observed vaccine herd protection, we hypothesised that analysis of the innermost “yolks” would reveal vaccine herd protection that was not evident in analysis of the entire clusters. The trial is registered at www.isrctn.com as ISRCTN11643110. Findings: At ≤18 months of follow-up, total vaccine effectiveness (protection of Vi-TT recipients relative to JE vaccine recipients) was 85% (95% CI: 76%, 90%); indirect effectiveness (protection of non-Vi-TT recipients in Vi-TT clusters relative to non-JE vaccine recipients in JE vaccine clusters) was 17% (95% CI: −13%, 40%); and overall effectiveness (protection of all residents in the Vi-TT clusters relative to all residents of the JE vaccine clusters) was 57% (95% CI: 44%, 66%). Analyses of subpopulations in inner 75%, 50% and 25% “yolks” of the clusters failed to reveal significant changes in any of these estimates. Interpretation: Our analysis did not reveal Vi-TT herd protection in the trial. Consideration should be given to exploring whether targeting adults as well as children with Vi-TT yields appreciable levels of vaccine herd protection. Funding: Bill & Melinda Gates Foundation (OPP1151153, INV-025388).