Dementia, a leading cause of death and disability globally, affects over 55 million people and is expected to rise sharply with ageing populations. While biomedical risk factors for dementia have been extensively studied, growing evidence highlights the potential role of social determinants (which link to social risk factors such as education, income, housing, and social isolation) in potentially influencing dementia onset. However, research in this area is often limited by methodological challenges, including confounding, reverse causality, and selection bias. This PhD project aims to address these gaps using advanced causal inference methods within a novel linked dataset that has linked detailed clinical records from South London and Maudsley NHS Foundation Trust with individual-level social and economic data from the 2011 UK Census and matched population controls.
The study will apply methods such as propensity score matching and inverse probability weighting to assess social risk factors with dementia incidence in older adults. It will also explore heterogeneity across key demographic groups (e.g. by ethnicity, gender, and socioeconomic position), and, where appropriate, the student will be supported to apply intersectionality-informed approaches such as Multilevel Analysis of Individual Heterogeneity and Discriminatory Accuracy (MAIHDA), or Latent Class Analysis (LCA) approaches to understand how multiple social risk factors may overlap in individuals. The project offers a rare opportunity to analyse one of England’s large-scale linkage study of individual-level social and clinical data, with potential to inform upstream public health prevention strategies. Training will be provided in advanced statistical methods, social epidemiology, and data access/ ethical access. The findings are expected to contribute substantively to the understanding of social inequalities in dementia and to support a policy-relevant public health approach to dementia prevention.