Reducing variation in the risk of stillbirth and preterm birth in the United Kingdom

Abstract

This thesis examines adverse pregnancy outcomes (APOs), specifically stillbirth and preterm birth, using routinely collected clinical data to understand variations in risk and perform risk prediction. Around one-third of pregnancies experience complications leading to APOs, impacting both maternal and neonatal health. The research emphasises the need for individualised and population-level interventions to prevent APOs and supports these efforts using quantitative modelling to identify at-risk individuals and populations.

The first part of the thesis involves developing and validating risk prediction models based on early pregnancy maternal and fetal characteristics, including socio-demographic factors, obstetric history, and current pregnancy features like ultrasound and biochemical markers. Though model predictive performance was modest, key predictors for stillbirth and preterm birth were confirmed across various modelling methods.

The second part investigates patterns and clusters of preterm birth and stillbirth risk using multilevel modelling to assess variations across geographic areas and healthcare providers in England, as well as the effects of environmental factors like air pollution. Significant clustering was identified, revealing that some areas and providers experience disproportionately high levels of risk, suggesting health inequalities.

This research contributes to the development of predictive tools for pregnant women, with the goal of improving early risk assessments and guiding clinical decision-making. By identifying vulnerable populations and highlighting disparities in APO risks, this work aims to inform targeted interventions. The findings are expected to support more personalised and equitable maternity care, contributing to the prevention of preterm birth and stillbirth.

Future research should focus on improving predictive models by including additional early pregnancy predictors and developing dynamic models for continuous risk assessment as pregnancies progress. Modelling should also address heterogeneous and overlapping causes of APOs and predict their timing. Further investigation is needed to understand factors driving spatial clustering and organisational disparities, building on insights from this thesis.

Date of Award	30 Sept 2025
Original language	English
Awarding Institution	University of Bristol
Sponsors	Tommy's, Nicholas House, London, UK.
Supervisor	Andy Judge (Supervisor) & Erik Lenguerrand (Supervisor)