Estimating Levels and Trends of Early and Late Neonatal Mortality in Countries Worldwide

Marija Pejchinovska , University of Toronto
Monica Alexander, University of Toronto

In recent decades, increased attention has been given to monitoring neonatal mortality rates (NMR) — defined as the number of deaths in the first 28 days of life per 1000 live births — largely as a result of international efforts like the Millenium Development Goals (MDG) and Sustainable Development Goals (SDG). Less attention has been devoted to understanding the timing of neonatal mortality, that is, whether the deaths occur early, within 0-7 days from birth, or late, within 8-28 days. Early neonatal deaths are believed to make up a vast proportion of neonatal deaths, and consequently, as under-5 child mortality rates (U5MR) have declined, early neonatal deaths have become an increasing portion of under-5 mortality. Having accurate and up-to-date estimates is, therefore, essential for policy makers and healthcare providers. However, estimating the timing of neonatal mortality is challenging as complete, good quality data are often lacking, especially in places where the mortality burden is the highest. In this work we present a bivariate Bayesian hierarchical penalized splines regression framework to jointly estimate early and late neonatal mortality for all countries in the world for the period 1990-2022, with projections through to 2030. Our model exploits known relationship between NMR and U5MR, allows for data-driven time trends, pools information across geographic regions, and accounts for varying data characteristics. We present our preliminary results in four distinct data and mortality contexts.

See extended abstract

 Presented in Session 68. Dealing with Incomplete or Deficient Data from Surveys and Censuses