Space-time smoothing models for sub-national measles routine immunization coverage estimation with complex survey data

TL;DR

This paper introduces a space-time smoothing model to accurately estimate sub-national measles vaccination coverage using complex survey data, accounting for supplementary immunization activities and survey design variations.

Contribution

The paper develops a novel space-time model that integrates multiple survey data sources and SIA information for precise sub-national immunization coverage estimation.

Findings

Model effectively incorporates SIA impact on coverage estimates.

Method handles multiple survey schemes with uncertainty quantification.

Implementation via INLA enables efficient analysis.

Abstract

Despite substantial advances in global measles vaccination, measles disease burden remains high in many low- and middle-income countries. A key public health strategy for controling measles in such high-burden settings is to conduct supplementary immunization activities (SIAs) in the form of mass vaccination campaigns, in addition to delivering scheduled vaccination through routine immunization (RI) programs. To achieve balanced implementations of RI and SIAs, robust measurement of sub-national RI-specific coverage is crucial. In this paper, we develop a space-time smoothing model for estimating RI-specific coverage of the first dose of measles-containing-vaccines (MCV1) at sub-national level using complex survey data. The application that motivated this work is estimation of the RI-specific MCV1 coverage in Nigeria's 36 states and the Federal Capital Territory. Data come from four Demographic and Health Surveys, three Multiple Indicator Cluster Surveys, and two National Nutrition and Health Surveys conducted in Nigeria between 2003 and 2018. Our method incorporates information from the SIA calendar published by the World Health Organization and accounts for the impact of SIAs on the overall MCV1 coverage, as measured by cross-sectional surveys. The model can be used to analyze data from multiple surveys with different data collection schemes and construct coverage estimates with uncertainty that reflects the various sampling designs. Implementation of our method can be done efficiently using integrated nested Laplace approximation (INLA).