Enumerating and estimating maternal and neonatal deaths in the Western Cape Province, South Africa

Doctoral Thesis

2022

Permanent link to this Item
Authors
Journal Title
Link to Journal
Journal ISSN
Volume Title
Publisher
Publisher
License
Series
Abstract
Measuring and monitoring progress towards global development goals requires valid and reliable estimates of maternal and child mortality. This thesis has aimed at enumerating and estimating maternal and neonatal deaths from 2010 to 2013 in the Western Cape Province; and determining factors associated with these outcomes during the same period. This thesis comprises nine chapters, of which six present the research findings. The first results chapter has presented the findings from a systematic review, determining trends of maternal and neonatal mortality from 1990 to 2015 in South Africa. The review found that estimates of maternal and neonatal mortality are widely divergent across data sources and estimation methods, with conflicting trends over the analysis period. The second results chapter compared the performance of the existing decision-rule based linkage approach (provincial linkage) which uses fuzzy linkage to an independent fully probabilistic record linkage (PRL) implementation for identifying mortality records across the Western Cape Provincial Health Data. The PRL was shown to be a feasible method for future implementation, while the existing linkage performed similarly to the independent linkage exercise, providing reassurance on the adequacy of the linked datasets on which the subsequent chapters were based. The third and fourth results chapters involved the applications of three-source capture-recapture methods, to estimate maternal and neonatal mortality under-reporting in the Western Cape province. Based on these models, maternal and neonatal mortality under-reporting were estimated at 45.6% and 17.7% over the full 4-year period respectively. The last two results chapters focused on determining factors associated with maternal and neonatal mortality in this setting and exploring whether the estimates of association were altered through using an expanded number of outcome events based on database linkage across multiple data sources. Most findings were consistent with known associations, as well as estimates from single-source analyses in the same setting. The thesis concludes that estimates of maternal and neonatal mortality are widely divergent in South Africa, and single-source reporting likely under-estimates the event rates. The application of capture recapture methods is a viable approach in South Africa to resolve the problems of under ascertainment in estimation of these outcomes.
Description

Reference:

Collections