Prevalence, Determinants, and Socioeconomic Inequality of Anaemia Among Women of Reproductive Age in Odisha and Jharkhand, India: A Survey-Weighted Analysis with Machine Learning Insights Using NFHS-5 Data

Abstract

Background: Anaemia among females of reproductive age (15-49 years) is a major community health crisis in India. Odisha and Jharkhand, two states with large tribal and economically disadvantaged populations, report anaemia burdens substantially above the national average. Objective: This study sought to define the dominance of anaemia in Odisha and Jharkhand, investigate its socioeconomic and nutritional causes using survey-weighted multivariable logistic regression, measure wealth-related inequality via the Concentration Index, and pinpoint key predictors using Random Forest classification. Methods: The analysis performed on pooled data from the National Family Health Survey-5 (NFHS-5, 201921), comprising 27,971 women from Odisha and 26,495 women from Jharkhand, for a entire sample of 54,466 women age bearing 15 to 49 years. Survey-weighted logistic regression models (svyglm with quasibinomial family), Random Forest classification ( ntree = 500), Variance Inflation Factor ( VIF) analysis, Concentration Index ( CI), and Erreygers’ Normalized Concentration Index ( ECI) were used. Jharkhand, the middle wealth category, and higher education were used as reference groups. Results: The survey-weighted prevalence of anaemia was 67.1% in Jharkhand and 65.5% in Odisha. There was no statistically significant difference between states (AOR=1.00; 95% CI: 0.951.07; p=0.879). Key factors included: higher BMI (Protective effect, AOR=1.00 per unit; p < 0.001), no education (AOR=1.00 1.20; p < 0.001), primary education (AOR=1.00; p=0.003), secondary education (AOR=1.00; p < 0.001), poorest wealth status (AOR=1.00; p < 0.001), and poorer wealth status (AOR=1.00; p=0.010). The VIF analysis indicated the absence of multicollinearity. According to the Mean Decrease Gini metric, Random Forest identified BMI and Age as the most significant predictors. The Concentration Index (CI = -0.065; Erreygers' ECI = -0.177) indicated that anaemia was significantly more prevalent among poorer populations. Conclusion: Anaemia in Odisha and Jharkhand is driven primarily by low BMI, lower education, and poverty rather than state of residence. Targeted nutrition, education, and poverty-alleviation interventions are urgently needed in both states.