Conditional probability and ratio-based approaches for mapping the coverage of multi-dose vaccines

Area of Science:

• Epidemiology
• Biostatistics
• Geographic Information Systems (GIS)

Background:

• Multi-dose vaccines require careful coverage mapping to ensure effectiveness.
• Ensuring monotonic decrease in coverage with successive doses is a key challenge.
• Accurate coverage maps guide investments in vaccination services and health systems.

Purpose of the Study:

• To explore conditional probability (CP) and ratio-based (RB) approaches for mapping multi-dose vaccine coverage.
• To address the challenge of ensuring monotonic coverage decrease in spatial mapping.
• To apply and evaluate these methods using real-world vaccination data.

Main Methods:

• Binomial geostatistical modeling framework.
• Implementation of a fully Bayesian model using Integrated Nested Laplace Approximation (INLA) and Stochastic Partial Differential Equations (SPDE).
• Comparison of CP and RB approaches via simulation studies and application to diphtheria-tetanus-pertussis (DTP) vaccine coverage data.

Main Results:

• Both CP and RB approaches performed comparably in out-of-sample estimation.
• Coverage maps for DTP vaccine doses in Nigeria were generated using both methods.
• The CP approach provided more precise estimates on average in the Nigerian application.

Conclusions:

• The developed geostatistical methodology effectively maps multi-dose vaccine coverage while ensuring monotonic decrease.
• The CP and RB approaches are robust and comparable, with CP offering potential for higher precision.
• The methodology is adaptable for other vaccines and geographical contexts, providing estimates for zero-dose children and dropout rates.