1 April 2021 in Drug and Alcohol Dependence
3 March 2026 in Annals of Epidemiology
This study aimed to develop a cluster-based measure of multiple co-occurring social determinants of health by applying unsupervised machine learning to a population-based cohort, offering a data-driven approach to organize complex social exposures. Unsupervised clustering was applied to a population-based cohort of Ontario respondents to six-cycles of the Canadian Community Health Survey (2001-2012) linked to the Canadian census and vital statistics data. Clusters were evaluated using internal metrics, visualization techniques, descriptive analysis and theoretical considerations to determine the optimal number of clusters. Sensitivity analyses were integrated across the iterative clustering process. Premature mortality rates were generated assess validity. Optimal clustering solutions included 4-clusters and 6-clusters. Both cluster solutions revealed distinct social typologies. The 6-cluster solution offered greater granularity and theoretical interpretability. The 4-cluster solution showed greater heterogeneity within certain marginalized groups. Premature mortality rates differed meaningfully across clusters, supporting the clustering approach in capturing risk associated with social exposure. Unsupervised machine learning methods identified meaningful population subgroups reflecting complex patterns of social exposures. This approach offers a flexible, data-driven method for characterizing social exposures that can be considered alongside theoretical frameworks and used for equity monitoring, intervention planning and policy development.
28 December 2024 in SSM - Population Health
Multimorbidity, the co-occurrence of two or more chronic conditions, is associated with the social determinants of health. Using comprehensive linked population-representative data, we sought to understand the combined effect of multiple social determinants on multimorbidity incidence in Ontario, Canada. Ontario respondents aged 20–55 in 2001–2011 cycles of the Canadian Community Health Survey were linked to administrative health data ascertain multimorbidity status until 2022. Additive total social exposure (TSE) was generated by summing 12 measures of social disadvantage captured from the survey. Weighted-additive TSE included 15 measures of social disadvantage summed across 5 equally weighted domains. Hazard ratios for the association between each TSE measure and multimorbidity were estimated using competing risk Cox-proportional hazards models. All analyses were sex-stratified. Both additive and weighted-additive TSE were associated with an increased risk of multimorbidity among females and males. A social gradient was observed for multimorbidity risk in all models. While adjusted models were attenuated, an increased risk of multimorbidity was observed among those experiencing the most social disadvantage, compared to those with the least social disadvantage in additive (HR Females = 2.16; 95%CI = 1.63, 2.86; HR Males = 1.90; 95%CI = 1.52, 2.38) and weighted-additive (HR Females = 1.94; 95%CI = 1.49, 2.53; HR Males = 1.72; 95%CI = 1.41, 2.10) models. The observed social gradient was retained. These findings demonstrate the importance of considering the cumulative effects of multiple social determinants of health on multimorbidity.
21 April 2022 in SSM - Mental Health
18 July 2026 in American Journal of Epidemiology
Multimorbidity follows a strong social gradient, yet how the co-occurrence of social determinants shapes risk remains poorly understood. Most research examines individual determinants in isolation, potentially obscuring socially patterned subgroups relevant to multimorbidity risk. This study examines the association between multidimensional social exposure profiles (derived from individual- and area-level determinants) and multimorbidity risk. Profiles were derived through unsupervised clustering applied to population-based cohort of Ontario respondents to the Canadian Community Health Survey (2001-2011) and linked area-level Census measures. Chronic disease and subsequent multimorbidity status were identified through linkage to administrative health data until 2022. Sex-stratified Weibull models were used to estimate hazard ratios for the association between multidimensional social exposure and incident multimorbidity. Unsupervised clustering identified six distinct social exposure profiles, with multimorbidity onset differing across profiles and elevated risk observed across multiple patterns of social disadvantage, rather than following a simple graded gradient. In sex-stratified analyses, disadvantage social exposure profiles were associated with an earlier multimorbidity onset among females (adjusted HR: 1.70, 95%CI, 1.18-2.53) and males (adjusted HR: 1.59, 95%CI, 1.20-2.05). These multidimensional social exposure profiles capture population-level social patterning associated with differential multimorbidity risk. These findings have implications for upstream policy aimed at reducing inequities in multimorbidity.