Levels Trends In Child Mortality Report 2022

Author : Roy Burstein
Publisher :
ISBN 13 :
Total Pages : 156 pages
Book Rating : 4.:/5 (19 download)

DOWNLOAD NOW!

Download or read book Mapping Child Health written by Roy Burstein and published by . This book was released on 2018 with total page 156 pages. Available in PDF, EPUB and Kindle. Book excerpt: This dissertation is comprised of three distinct chapters which each aim to improve understanding of child health and survival in developing countries through novel applications of statistical modelling. In each case, modelling was used to make estimates at more refined levels than previously existed, either geographically or in age categories, with the goal of improving the evidence base upon which trends and inequalities in child health could be monitored and assessed. In the first chapter, Mapping under-5 and neonatal mortality in Africa, a baseline analysis for the Sustainable Development Goals, a geostatistical model was developed to estimate the child mortality rate for each 5 x 5 kilometer piece of land in 46 countries in Africa, from 2000 to 2015. The study utilized geographically referenced data from 235 household surveys and censuses and a suite of geospatially disaggregated covariates to make high resolution estimates. Resultant estimates on a gridded surface were also aggregated to the district and province level in each country, to provide a full range of estimates at useful spatial resolutions. Despite large declines in mortality rates overall during the study period, there was substantial heterogeneity in both the absolute levels and rates of change in child mortality, both across and within countries. While declines in many areas met or surpassed the 2015 UN Millennium Development Goal for a 4.4% annual reduction, the hard threshold of 25 deaths per 1000 livebirths set for the 2030 Sustainable Development goal would require much of Africa to reduce mortality at unprecedented rate during the coming years. The second chapter, Development and validation of a new method for indirect estimation of neonatal, infant, and under-5 mortality using summary birth histories, dealt with the issue arising in child mortality estimation wherein the dominant source of data, summary birth histories (SBHs), do not on their own provide enough information to estimate age-specific trends in child mortality. To address this, a discrete hazards model was trained on complete birth history data from 243 surveys in 76 countries, representing over 8 million births. A novel approach was developed for prediction at the level of the hypothetical child, each weighted by their probability of birth. Three validation and verification approaches were developed: survey-wise cross validation, comparison against existing indirect methods, and external validation employed through the application of the new method to an addition 243 SBH-only data sources. The new method was found to produce results which are comparable to current best methods for under-5 mortality estimation while additionally producing valid age-specific estimates. Use of this method could allow researchers to utilize a massive amount of SBH data for estimation of trends in neonatal and infant mortality. The third chapter, Geographic accessibility and utilization of facility-based care in Zambia: a geostatistical analysis, explored the predictive ability of geographic factors, including travel time to nearest health facility, to explain healthcare utilization in Zambia, and developed on geostatistical model in order to predict a gridded surface representing treatment seeking rates for diarrhea and febrile illness in children across the country. This analysis overcame a number of important data and methodological limitations present in previous research, including using exact household locations, using flexible splines to represent the decay function over travel time, and using a full probabilistic model and providing uncertainty around all estimates. Results indicate that while at least three quarters of Zambians live within an hour of a health facility, small differences in travel time to healthcare are independently associated with large declines in utilization rates within the first hour. The decision to seek care at a health facility is a complex process that is not easily reduced to geography. As such, a univariate model based solely on geographic accessibility is not sufficient for accurate prediction of utilization across a gridded surface. Improved prediction using a probabilistic model is possible but uncertainty remains high.