Under-five mortality in high focus states in India: a district level geospatial analysis

Abstract

Background: This paper examines if, when controlling for biophysical and geographical variables (including rainfall, productivity of agricultural lands, topography/temperature, and market access through road networks), socioeconomic and health care indicators help to explain variations in the under-five mortality rate across districts from nine high focus states in India. The literature on this subject is inconclusive because the survey data, upon which most studies of child mortality rely, rarely include variables that measure these factors. This paper introduces these variables into an analysis of 284 districts from nine high focus states in India.

Methodology/principal findings: Information on the mortality indicator was accessed from the recently conducted Annual Health Survey of 2011 and other socioeconomic and geographic variables from Census 2011, District Level Household and Facility Survey (2007-08), Department of Economics and Statistics Divisions of the concerned states. Displaying high spatial dependence (spatial autocorrelation) in the mortality indicator (outcome variable) and its possible predictors used in the analysis, the paper uses the Spatial-Error Model in an effort to negate or reduce the spatial dependence in model parameters. The results evince that the coverage gap index (a mixed indicator of district wise coverage of reproductive and child health services), female literacy, urbanization, economic status, the number of newborn care provided in Primary Health Centers in the district transpired as significant correlates of under-five mortality in the nine high focus states in India. The study identifies three clusters with high under-five mortality rate including 30 districts, and advocates urgent attention.

Conclusion: Even after controlling the possible biophysical and geographical variables, the study reveals that the health program initiatives have a major role to play in reducing under-five mortality rate in the high focus states in India.

Figure 1. Study area and under-five Mortality.

A. Location of study area in India B. Under-5 Mortality Rate (per 1000 live births) across 284 districts in high focus states of India, 2010–11.

Figure 2. Maps depicting spatial clusters and outliers in under-five mortality rate across 284 districts in high focus states of India, 2010–11.

A. Cluster and Outlier analysis map (Anseline Local Moran's I = 0.45) using polygon contiguity (first order) weight in ArcGIS. B. Hot Spot analysis map (Getis–Ord Gi*) using polygon contiguity (first order) weight in ArcGIS. C. Univariate LISA Cluster map (Moran's I = 0.439) using Rook's weight in GeoDa. D. Univariate LISA Significance map of Figure C.

Figure 3. Residual maps of OLS and Spatial Error Model for under-five mortality across 284 districts in high focus states of India, 2010–11.

A. Univariate LISA Cluster map (Moran's I = 0.416) plotting residuals of OLS regression model. B. Univariate LISA Cluster map (Moran's I = 0.200) plotting residuals of Spatial Error regression model.

Figure 4. Bivariate LISA (Cluster and Significance) maps depicting spatial clustering and spatial outliers of under-five mortality by coverage gap index across 284 districts in high focus states of India, 2010–11.

A. Bivariate LISA Cluster map of Under-5 Mortality Rate and Coverage Gap Index. B. Bivariate LISA Significance map of Under-5 Mortality Rate and Coverage Gap Index.

Figure 5. Clusters of districts with high under-five mortality rate.