# An age-structured spatially varying coefficient model for high-resolution mapping of vaccination coverage

## TL;DR

This study introduces a new method to map vaccination coverage by age and location, revealing where and when children in Côte d’Ivoire are missing measles vaccines.

## Contribution

A novel Bayesian age-structured spatially varying coefficient model for high-resolution vaccination coverage mapping.

## Key findings

- Significant delays in measles vaccination during the first year of life were identified.
- Substantial spatial differences in vaccination coverage by age were revealed in Côte d’Ivoire.
- The method enables targeted interventions to address immunity gaps in specific age groups and regions.

## Abstract

High-resolution maps of vaccination coverage are valuable for uncovering heterogeneities in coverage to inform vaccine delivery strategies. Coverage maps stratified by age can reveal additional heterogeneities in the timeliness of vaccination and critical immunity gaps among birth cohorts. Here, we propose a spatially varying coefficient model relying on a Bayesian approach for age-structured mapping of vaccination coverage using geolocated individual level household survey and geospatial covariate data. Our flexible modelling framework includes parameterizations capturing spatial (non-)stationarity in differences in coverage between age groups, as well as a modification to allow coverage mapping for single age points through the inclusion of a smoother over age. The proposed models are fitted using the INLA-SPDE approach implemented in the inlabru package in R. We choose between competing model parameterizations by examining their out-of-sample predictive performance via cross-validation and using Bayesian model choice criteria. The methodology is applied to age-structured mapping of measles vaccination coverage in Cote d’Ivoire using the 2021 Demographic and Health Survey. Our results reveal a significant delay in measles vaccination in the first year of life and substantial spatial differences in coverage by age, highlighting the need for targeted interventions to achieve equity and attain vaccine-derived immunity goals.

This study presents a new method to create spatially detailed, age-specific maps of vaccination coverage, with a focus on measles vaccination in Côte d’Ivoire. Unlike traditional national or provincial estimates, these maps show where children are being vaccinated on a fine geographical scale, including differences between age groups. This matters because children who miss their vaccines early in life can leave pockets of vulnerability that fuel disease outbreaks. We used data from a 2021 national health survey implemented in the country, combined with maps of local conditions such as how far families travel to access health care, living standards, and the environment. We then applied a modern statistical technique – an age-structured spatially varying coefficient model – that flexibly adjusts how age-related vaccine coverage varies across space, and which can estimate coverage for age groups and single age points. This enabled us to create maps that uncovered not just where overall vaccination is low, but where in specific age groups (like children under 15 months) it’s falling behind. The results revealed important delays in measles vaccination during early childhood and highlighted geographic areas with particularly low coverage. By showing gaps in both age and location, this approach enables health workers to target immunization efforts more precisely, helping ensure that more children have timely protection against measles and potentially other diseases.