Statistical Methods for Small-Area Estimation and Reporting of Perinatal Mortality

About the Project

Project overview

The UK’s national surveillance programme MBRRACE-UK provides robust estimates of perinatal mortality to maternity and neonatal care providers. However, geographically defined organisations, such as Integrated Care Boards (ICBs) and local authorities, also require reliable estimates for smaller areas to target interventions effectively. Direct estimates for these areas are often unstable due to small sample sizes and rare events.

Using national perinatal surveillance data from MBRRACE-UK, this PhD project will develop and evaluate advanced statistical methods for producing accurate and stable small-area estimates of perinatal mortality and related outcomes. The research will also focus on improving how these estimates are reported and communicated to local organisations for decision-making, ensuring outputs are both statistically sound and practically useful.

Key Research Questions

• Develop and compare Bayesian and frequentist approaches (e.g., hierarchical models, EBLUP, GLMMs) for small-area estimation.
• Investigate the potential appropriateness of covariates (e.g., deprivation, maternal age, ethnicity).
• Explore methods for communicating uncertainty and risk to local authorities and ICBs.
• Validate methods through simulation studies and historical data analysis.

Key Areas of Exploration

• Small-area estimation techniques: Develop and compare hierarchical models for rare events, including frequentist mixed models and Bayesian approaches.
• Spatial correlation: Incorporate geographic structure using spatial random effects to improve precision for neighbouring areas.
• Covariate integration: Assess the impact of adding maternal, clinical, and socioeconomic factors on model accuracy and interpretability.
• Simulation studies: Design scenarios to evaluate bias, variance, and interval coverage under different estimation strategies.
• Reporting and communication: Create practical methods for presenting uncertainty (e.g., funnel plots, confidence intervals) to local authorities and ICBs.
• Impact assessment: Quantify how different modelling choices influence national surveillance metrics and equity assessments.

Outputs

The student will produce a robust framework for small-area estimation of perinatal mortality. They will develop practical guidance for reporting results to local organisations and recommendations for national reporting standards.

This work will directly influence how perinatal mortality data are interpreted and acted upon at local levels, helping healthcare systems allocate resources more effectively and reduce inequalities in maternal and neonatal outcomes. This research has the potential to shape national policy and improve care for thousands of families.

Candidate Requirements

We are looking for a candidate with:

• Strong quantitative background (statistics, biostatistics, epidemiology, or related field).
• Experience with statistical software (R, Python) and interest in hierarchical modelling.
• Knowledge of healthcare data and perinatal outcomes is desirable but not essential.

Training and Environment

The successful candidate will join a vibrant research environment with access to training in statistical methods, health data science, and research communication. Opportunities for collaboration with clinicians, policy-makers, and statisticians will be available throughout the project.

Apply at:

https://le.ac.uk/study/research-degrees/research-subjects/school-of-healthcare

Supervisor contact details:

Prof Bradley Manktelow - brad.manktelow@leicester.ac.uk

Dr Sarah Seaton - sarah.seaton@leicester.ac.uk

Dr Ruth Matthews - rjm81@leicester.ac.uk