Statistical Modelling and Mapping of Health Outcomes in Developing Countries

Khaki, Jessie and Giorgi, Emanuele (2025) Statistical Modelling and Mapping of Health Outcomes in Developing Countries. PhD thesis, Lancaster University.

Text (2025KhakiPhD)
2025KhakiPhD.pdf - Published Version
Download (6MB)

Abstract

The 2030 Sustainable Development Goals (SDGs) aim at improving the lives of people. To monitor the progress towards achieving the SDGs and effectively improve people’s lives, there is a need to efficiently use publicly available data to inform decisions. However, developing countries struggle to track the SDGs due to limited financial resources and technical skills. This thesis explores how health SDG outcomes can be tracked and modelled using publicly available datasets in low- and middle-income countries (LMICs). In Chapter 3, this thesis investigates how passive surveillance data arising from a typhoid point pattern process in Blantyre, Malawi, can be analysed using environmental and individual-level covariates such as age and gender. Chapter 4 applies multilevel and mixed effects models to publicly available geostatistical demographic and health survey data from Malawi to model and map the double and triple malnutrition burden among mother-child pairs without spatial correlation. Chapter 5 extends the work carried out in Chapter 4 by applying model-based geostatistics to publicly available geostatistical soil-transmitted helminth survey data from 35 African countries. Chapter 5 also discusses some challenges encountered when using sparse data from LMICs and provides recommendations on ideal data for geospatial predictions. Lastly, Chapter 6 characterises the dengue outbreak in 77 Nepalese districts between 2006 and 2022. Using district-level areal data and a modified Negative Binomial model, the thesis estimates the timing and duration of 3 outbreak intensity functions within each district. This thesis demonstrates the use of statistical modelling in tracking health outcomes in developing countries. The thesis additionally discusses the challenges associated with publicly available data in LMICs, such as sparse data, and proposes solutions to these challenges. Finally, the thesis suggests ways in which each aspect of the research can be extended in future studies