Thesis title: Exploring Disparities in Health and Mortality in Ageing Societies
Despite being a universal human right, health remains unequally – and unfairly – distributed across nearly all populations. Understanding the fundamental causes of disparities in health and mortality is crucial, especially in the backdrop of population ageing. As life expectancy continues to rise, a larger share of the population across the world is expected to experience health deterioration in later life, posing unprecedented challenges for healthcare systems’ ability to ensure effective and equitable healthcare. This thesis attempts a multidimensional exploration of disparities in health and mortality in ageing societies. Adopting a multidisciplinary approach, it examines how health in later life and mortality are shaped by multiple and intersecting social determinants, including education, gender, migration background, partnership status, and housing history. The thesis is a collection of three empirical studies, each shedding light on health disparities from a different angle. Study I adopts a life course approach to study how housing tenure trajectories over the life course (ages 16–65) predict disparities in later-life health among adults aged 65–75 in eleven European countries. Using SHARELIFE retrospective data (2017), sequence and cluster analysis are used for identifying the main patterns of housing tenure trajectories, and logistic regression is employed to investigate their relationship with self-rated health, chronic morbidity, and activity limitations, net of other relevant social determinants. The study explores potential heterogeneity in the association by gender and country group (Continental, Northern, and Southern Europe). Six patterns are identified. The standard – i.e., the most common – trajectory (‘early homeowners’) is characterized by early and sustained homeownership and comprises around 60% of the population. Overall, non-standard trajectories – ‘late-homeowners’, ‘never-leavers’, ‘private tenants’, ‘social tenants’, ‘rent-free and others’ – are associated with poorer health outcomes, compared to the standard pathway. Although evidence is mixed across health outcomes, some findings suggest that non-standard trajectories are more health-detrimental for women and for individuals living in Southern Europe. Study II investigates educational disparities in dementia incidence and subsequent healthcare utilization in Lazio Region (Italy) during the years 2012–2022. Drawing on census-linked data from the Lazio Region Longitudinal Study, the study pursues a two-fold aim: it first analyses educational disparities in dementia incidence among 1.9 million dementia-free adults aged 50–90, and secondly, it analyses disparities in subsequent healthcare utilization (all-cause and potentially preventable hospitalizations, and emergency visits) among around 70 thousand incident dementia cases. In all analyses, Cox proportional hazards models are applied. Findings suggest that low-educated individuals are at a significantly higher risk of developing dementia, especially early onset (ages 50–64). Following dementia onset, low-educated dementia cases also experience a higher risk of all-cause hospitalizations, potentially preventable hospitalizations, and emergency visits. These disparities are mostly explained by pre-existing health conditions, particularly for hospitalizations, though less so for emergency visits, for which other non-clinical factors may play a role. Study III investigates the extent to which the mortality advantage of being partnered (married and cohabitant) changes across migration generations (natives, first-generations, and second-generations). Partnered individuals generally experience lower mortality than their unpartnered peers, due to both selection and protection mechanisms. However, this advantage varies across population subgroups. The study uses Swedish register data (2012–2022) to examine how the mortality advantage of being partnered differs by migration generation. Using Gompertz proportional hazards regression models, it explores differences in all-cause mortality between ages 18-79 mortality in Sweden, across combinations of partnership status and migration generation. Results confirm a mortality advantage for all partnered individuals; overall, this advantage tends to be smaller among first-generation migrants than natives. Among second-generation migrants, instead, unpartnered men show a larger excess mortality, while women show similar patterns as native Swedes. Differential selection, protection, and compositional effects are likely responsible for heterogeneity in the mortality advantage.