Stillbirths are under-researched in comparison to other child health outcomes. They rarely appear in national health targets or commitments, despite being an immense global burden with 2.6 million cases annually. Using routine national datasets, this thesis compares the social disparities in stillbirth rates within and between the UK and Brazil as a high-income and middle-income country, respectively.
The role of socio-economic position (SEP) in the likelihood of stillbirth, and how it is mediated through behavioural, health and pregnancy-related factors, is clarified comprehensively reviewed and a new conceptual model constructed. A comprehensive review of perinatal health policies in the UK and Brazil showcases Brazil’s evolution as a major emerging economy from a public health perspective, in relation to the UK.
Analysis of the UK required two separate datasets; aggregated birth counts from ONS for England and Wales, and Scotland’s Maternity and Neonatal Linked Database (MNLD). In England and Wales, higher stillbirth rates were associated with areas containing higher proportions of ethnic minority babies, after accounting for maternal age and other confounders. Whilst the data for England and Wales was limited to an aggregate analysis, the comprehensive Scottish data allowed a mediation analysis which was applied to the MNLD to investigate more fully the causal pathways between SEP and stillbirth. While socio-demographic factors such as maternal age, ethnicity, and marital status were found to have no direct effect on stillbirth risk, their mediating influence on gestational age and birth weight played a vital role in the cause of stillbirth.
Brazil’s National Household Sample Survey (PNAD) was analysed to determine the primary socioeconomic risk factors associated with stillbirth. The main challenge presented by the data was that the birth outcomes were presented as counts - the numbers of live births and stillbirths had by each woman. By converting the observed live birth and stillbirth counts for each woman into lists of possible birth sequences, and deriving the appropriate likelihood function, this thesis was able to perform regression on a new binary outcome for live birth verses stillbirth. Women with lower levels of education and inadequate household goods and services had a higher risk of stillbirth. Analysis of stillbirth rates based on women’s lifetime birth histories revealed some counter-intuitive relationships between geographic and social factors and stillbirth, which were distorted by high parity among disadvantaged women.
Data and methodologies were too different to compare the magnitudes of variable effects between countries. Whilst our new methodological approaches have allowed more sophisticated analyses than have been previously possible, ultimately, this thesis has highlighted the strengths and limitations that different data sources allow in describing stillbirths globally. Without more comprehensive monitoring and better defined outcomes, countries will struggle to see the full picture. As we have shown aggregate level data are inadequate for public health monitoring in England and Wales, and restrictions on data availability prohibit informative individual-level analyses. Looking to the future the Scottish data has shown that the gold standard of data linkage is both viable and recommended for 21st century public health management.