Equity in Maternal, Newborn and Child Health in Low and Middle-Income Countries.

Abstract

Reproductive, Maternal, newborn and child health (RMNCH) coverage in Low and Middle- Income Countries (LMIC) have improved over the past few decades. However, coverage distribution remains inequitable. Despite relative improvement in coverage of RMNCH services, there is still a disproportionately high level of preventable maternal and child mortality in LMIC. Many of these preventable child deaths are also due to poor quality of care and interventions. Although a significant amount of work has been done in measuring RMNCH intervention coverage, very little work has been done in assessing the quality of these interventions. Furthermore, it has been shown that achieving equity in coverage is underpinned by evaluating contextual differences at community and household levels as against exploring national trends. However, some previous measures of progress in RMNCH coverage have focused on exploring trends at national level. The aim of this project is to explore trends and patterns of RMNCH inequity at community and household levels and to extend the literature on the assessment of quality of RMNCH services in LMICs, while also investigating the application of new approaches such as Artificial Intelligence (AI) for service quality improvement in LMIC. Through a series of publications, this project investigated: (1) the effects of inequality on selected indicators of reproductive, maternal, newborn and child health using a meta-analysis; (2) the efficacy of a machine learning approach (deep learning) for the prediction of Under-five Mortality (U5M); (3) the contextual variations and association between birth weight and under-five mortality in LMIC; (4) geospatial approach for assessing quality trends in basic and emergency obstetric care services in a low-income country; (5) a review of the adoption of AI for maternal and child health service quality improvement in Low and Middle-Income Countries. Data from the Demographic and Health Survey (DHS) were utilised in investigating the efficacy of machine learning for the prediction of under-five mortality. The DHS data was also used to explore contextual variations and association of birth weight with U5M, while data from the Service Provision Assessment (SPA) survey was used to map the quality of obstetric care services in a low-income country. The results showed that: (1) maternal and child health coverage remains highly inequitable and access to maternal and child health services are governed by factors such as income, level of education, and place of residence; (2) machine learning has significant potentials and are more sensitive and specific for the prediction of U5M; (3) several contextual factors nested at community and household levels, affects U5M, birth weight and duration of breastfeeding are the most important determinants of U5M; (4) the quality of obstetric care services varies significantly at sub-national levels, and primary care facilities are making the most progress in obstetric care quality improvement; (5) Artificial Intelligence (AI) is under-utilised for maternal and child health services in LMIC and the data currently used to explore AI-applications in these settings may result in significant research-practice gaps. Overall, this project shows that considerable inequity persists in RMNCH intervention coverage and quality. It also shows that predictors of RMNCH indicators such as U5M, are nested at community and household levels; hence underscoring the importance of investigating quality and coverage at household levels as against measuring national trends. Finally, the potential of geospatial monitoring and machine learning as a more sensitive and specific approach for investigating sub-national service quality trends and for quality improvement is underscored.