Projecting vaccine efficacy: accounting for geographic strain variations.

Researchers must often make assumptions about the efficacy of an intervention in a target population without the benefit of trial data specific to that population. Such assumptions may be particularly tenuous with models of vaccination strategies, since the distribution of pathogen strains in target populations may differ substantially from the strain distributions in trial sites. We describe a technique for projecting expected vaccine efficacy in settings where applying unadjusted trial-based efficacy data may overestimate the benefits of immunization. This simple method uses data describing setting-specific strain distributions of pathogens and strain-specific vaccine efficacies to generate a weighted overall efficacy. An example of estimating the expected efficacy of a new rotavirus vaccine in India is used to illustrate the technique. The method is shown to perform very well in a validation population for whom actual efficacy had been observed and can therefore aid those in the international health community in determining the optimal uses of scarce resources.