Hierarchical Clustering of Indian States Based on Child Mortality Indicators: Evidence from NFHS-5
Sanjay Karande *
Department of Statistics, Sathaye College, Mumbai, Maharashtra, India.
Ramkrishna Lahu Shinde
Department of Statistics, School of Mathematical Sciences, Kavayitri Bahinabai Chaudhari North Maharashtra University, Jalgaon, India.
*Author to whom correspondence should be addressed.
Abstract
Background: Child mortality remains a critical public health concern in India, with substantial disparities in infant, child, and under-five mortality across states.
Aims: The study aims to classify Indian states according to their child mortality profiles using hierarchical cluster analysis based on infant mortality rate (IMR), child mortality rate (CMR), and under-five mortality rate (U5MR), and to identify regional patterns that may support targeted public health interventions.
Study Design: Cross-sectional analytical study using secondary data.
Place and Duration of Study: The study was conducted using state-level data from the National Family Health Survey-5 (NFHS-5), India (2019–2021). The analysis was carried out during 2026.
Methodology: State-level estimates of IMR, CMR, and U5MR for 30 Indian states were analysed. Descriptive statistics and Pearson's correlation analysis were performed to examine the distribution and relationships among the mortality indicators. The variables were standardised using a Z-score transformation, and Mahalanobis distance was used to assess multivariate outliers. Hierarchical agglomerative cluster analysis using Ward's minimum variance method and Euclidean distance was performed. The Elbow method was used to determine the optimal number of clusters. The cluster solution was validated using one-way analysis of variance (ANOVA) and Tukey's Honest Significant Difference (HSD) test. Spatial distribution maps and boxplots were used to visualise regional mortality patterns.
Results: The Elbow method identified a three-cluster solution. Cluster 1 included 19 states with intermediate mortality levels, Cluster 2 comprised 4 states with the lowest mortality, and Cluster 3 contained 7 states with the highest mortality. Significant differences were observed among the clusters for IMR (F = 38.21, P < .001), CMR (F = 16.77, P < .001), and U5MR (F = 44.81, P < .001). The spatial distribution of the clusters revealed marked regional disparities, with high-mortality states predominantly concentrated in northern and central India.
Conclusion: Hierarchical cluster analysis successfully classified Indian states into statistically distinct child mortality groups. The identified clusters provide a practical framework for region-specific public health planning, efficient resource allocation, and targeted interventions aimed at reducing child mortality and regional inequalities across India.
Keywords: Child mortality, infant mortality rate, child mortality rate, hierarchical cluster analysis, ward’s method, mahalanobis distance, NFHS-5