Small Area Estimation of Fertility in Low- and Middle-Income Countries

2025-07-04

AI Summary

In low- and middle-income countries (LMICs), national fertility indicators—such as age-specific fertility rates (ASFR) and total fertility rates (TFR)—lack fine spatial resolution, hindering localized public health interventions. To address data sparsity in small areas, narrow age groups, and short time periods, we propose a unified framework integrating direct estimation with Bayesian hierarchical modeling. The method leverages Demographic and Health Surveys (DHS) fertility history data and incorporates spatiotemporal smoothing alongside key covariates—including educational attainment, contraceptive use, and urbanization. It jointly models areal and unit-level structures and is validated via cross-validation. Applied to the 2021 Madagascar DHS, our approach produces the first subnational (district-level) ASFR and TFR maps, markedly improving estimation stability and accuracy. This scalable methodology advances small-area fertility monitoring and evidence-informed policy design in LMICs.

Technology Category

• Machine Learning: Bayesian Learning
• Data Mining & Knowledge Management: Mining of Spatial, Temporal or Spatio-Temporal Data
• Reasoning under Uncertainty: Graphical Models

Application Category

• Graph Algorithms and Modeling for the Web: Representation, reconstruction, and subgraph or motif discovery in Web-related graphs
• User Modeling, Personalization and Recommendation: Federated recommendation systems and personalization
• Search and Retrieval-Augmented AI: Web evaluation methodologies and metrics

Abstract

Accurate fertility estimates at fine spatial resolution are essential for localized public health planning, particularly in low- and middle-income countries (LMICs). While national-level indicators such as age-specific fertility rates (ASFR) and total fertility rate (TFR) are often reported through official statistics, they lack the spatial granularity needed to guide targeted interventions. To address this, we develop a framework for subnational fertility estimation using small-area estimation (SAE) techniques applied to birth history data from household surveys, in particular Demographic and Health Surveys (DHS). Disaggregation by geographic area, time period, and maternal age group leads to significant data sparsity, limiting the reliability of direct estimates at fine scales. To overcome this, we propose a suite of methods, including direct estimators, area-level and unit-level Bayesian hierarchical models, to produce accurate estimates across varying spatial resolutions. The model-based approaches incorporate spatiotemporal smoothing and integrate covariates such as maternal education, contraceptive use and urbanicity. Using data from the 2021 Madagascar DHS, we generate district-level ASFR and TFR estimates and evaluate model performance through cross-validation.

Problem

• Estimating subnational fertility rates in LMICs
• Addressing data sparsity in small-area fertility estimation
• Developing Bayesian models for reliable spatial fertility data

Innovation

• Uses small-area estimation (SAE) techniques
• Applies Bayesian hierarchical models
• Incorporates spatiotemporal smoothing and covariates