Spatial heterogeneity in the coverage of full immunization among children in India: Exploring the contribution of immunization card

•This paper analyzes the spatial heterogeneity and contribution of immunization card in the full immunization coverage among children in India.•The slow progress and a wider socio-economic and spatial heterogeneity in full immunization coverage.•Ensuring immunization card to each eligible child for age appropriate vaccination has contributed significantly. This paper analyzes the mesoscale correlates and spatial heterogeneity in coverage of full immunization among children age 12–23 months in India using data from three rounds of National Family Health Survey conducted in 1998–99, 2005–06 and 2015–16. The analytical methods used in this paper are predicted probabilities to explain changes in likelihood of full immunization among children over time. Propensity score matching to estimate contribution of vaccination card using counterfactual approach. Bivariate LISA maps and spatial error model have been used to understand spatial heterogeneity in full immunization coverage among children. Overall, 62 percent of children aged 12–23 months were fully immunized in 2015–16, that has increased from 41 percent in 1998–99. Availability of vaccination card at the time of survey and coverage of full immunization have strong positive association, which further increases with increasing wealth status in each of the three rounds of NFHS. Changes in predicted probability of full immunization among children indicate that the immunization program in India has made concerted efforts to reach out to all the socially deprived and economically marginalized children and ensured availability of immunization card and increased probability of full immunization. The counterfactual approach portrays that the estimated contribution of immunization card in the full immunization coverage were significantly higher in almost half of Indian states. Bivariate LISA Cluster map of full immunization have identified around 109 districts in the country as hotspots. When spatial weights were taken into consideration, the auto regression model noticeably became stronger in predicting the prevalence of full immunization. From the Spatial Error Model the estimated coefficients were −0.74 (p-value