IDENTIFYING PLAUSIBLE RANGES FOR DIFFERENTIAL VACCINE EFFICACY ACROSS HIGH- AND LOW-INCOME SETTINGS: A SYSTEMATIC REVIEW, DESCRIPTIVE META-ANALYSIS, AND ILLUSTRATIVE EVIDENCE ANALYSIS

Abstract

ABSTRACTBackgroundRandomized clinical trials provide the highest standard of evidence about vaccine efficacy. Modelling exercises such as in evidence synthesis and health economic models where efficacy estimates are combined with other data to obtain effectiveness and cost-effectiveness estimates help inform policy decisions. The main challenge with such sensitivity analyses is in deciding on which assumptions to model.PurposeTo identify plausible ranges for differential vaccine efficacy across high- and low-income settings.Data Sources and Study SelectionMEDLINE, EMBASE, clinicaltrials.gov, and the World Health Organization International Clinical Trials Registry Platform (WHO-ICTRP) were searched for multi-site randomized clinical trials of bacterial and viral vaccines for the period of 01/01/1990 to 31/12/2020. Articles were restricted to those where at least one trial had included a low- or lower-middle-income setting, published in English, and conducted in humans.MethodsA Bayesian random-effects meta-analysis was used to estimate the difference in vaccine efficacy in high-(high or upper middle) and low-(low or lower middle) income settings. A single hierarchical model that included all trials was used so that the degree to which estimates of vaccine efficacy against different diseases influenced one another was estimated from the observed data.ResultsAcross 65 eligible trials (37 high-income, 21 low-income, and 7 both) covering 7 pathogens, only one trial reported efficacy estimates stratified by setting. Trials were similar in terms of design across settings. There was evidence of heterogeneity by vaccine target, typhoid vaccine demonstrated higher vaccine efficacy in low-income settings than in high-income settings but for all other vaccines, the point estimates indicated efficacy was lower in low-income settings; however, all credible intervals crossed the null.ConclusionsThe percentage of trials in low-income settings poorly reflects the burden of disease experienced in low-income settings. While there is evidence of lower vaccine efficacy in low-income settings relative to high-income settings, the credible intervals were very wide. Vaccine efficacy trials should report treatment effects stratified by settings.