Validating a Population Measure of Early Childhood Development in Low- and Middle-Income Countries: The Early Human Capability Index

Abstract

Increased focus on population monitoring of early childhood development has been spurred by the Sustainable Development Agenda, together with burgeoning evidence for the need to support children to reach their developmental potential. Tracking children’s development in low- and middle-income countries (LMICs) has been challenged by a lack of appropriate measurement tools and resources to implement measurement. The early Human Capability Index (eHCI) was designed to measure holistic development among children aged 3-5 years, be feasible for large-scale use in low resource settings, and capture locally relevant information. The aim of this thesis was to investigate the reliability, validity, and sensitivity of the eHCI across diverse settings to advance understanding of how the tool can be used to facilitate population measurement of early childhood development. Four research studies utilised pre-existing data collected using the eHCI from 2013-2020 among children aged 2-6 years in seven LMICs, including Brazil, China, Kiribati, Lao PDR, Samoa, Tonga, and Tuvalu. The first study investigated whether data fit the theoretical structure of the eHCI (nine developmental domains) across seven countries, given the necessary adaptation of the instrument in each country. Confirmatory factor analyses indicated the eHCI maintained the same factor structure across countries, providing evidence for the tool’s construct validity. The second study explored the convergent, divergent, discriminant, and concurrent validity of the eHCI, and whether results varied across seven countries. Results provided evidence that the tool captured aspects of early childhood development it was designed to measure. Although the eHCI was intended to measure development among children aged 3-5, results from this study indicated it may be validly applied to children aged 2-6 years. The first two studies established the eHCI was psychometrically robust using cross-sectional data. The third study used longitudinal data to explore the ability of the eHCI to predict children’s later abilities in Lao PDR, establishing predictive ability of the tool. Specifically, whether scores on the eHCI at 2-5 years predicted cognitive development (literacy, numeracy, executive function) at 6-9 years, four years later. Receiver Operator Characteristic curve analyses demonstrated the summary indicator, eHCI overall development, signalled risk for poor future cognitive development with similar ability to measures of socioeconomic position. The eHCI was designed to have adequate sensitivity to detect variation in children’s development to facilitate program evaluation, which is a limitation of many existing population measures. The final study tested the sensitivity of eHCI scores to inputs promoting children’s development, namely quality of early childhood education. Using cross-sectional data in Lao PDR, adjusted linear regressions demonstrated small, positive associations between quality and children’s development measured via the eHCI, as was hypothesised. Together, studies demonstrated that the eHCI, a pragmatic, freely available and locally adapted tool, can be validly applied to children aged 2-6 years across diverse LMICs, for the purposes of locally relevant population monitoring of early childhood development, as well as program evaluation. Ultimately, information collected using the eHCI may be used to inform policy and practice in terms of resourcing and supports to promote children’s development.