Background Cardiometabolic multimorbidity - the co-existence of heart disease, stroke, and type 2 diabetes - constitutes a growing public health burden due to its association with premature mortality and reduced quality of life. However, cardiometabolic multimorbidity has been limitedly studied as a combined condition, and risk factors are often examined in isolation, thereby overlooking their clustering and intersectional patterning in the population. This thesis aims to estimate socio-geographical disparities, identify clusters of cardiometabolic risk factors, and develop a prognostic model for predicting cardiometabolic multimorbidity among middle-aged adults in Sweden. Methods Cardiometabolic multimorbidity was assessed in a general population sample of 30,154 adults aged 50–64 years using self-reported information, registry data, and biomarkers available in the Swedish CArdioPulmonary BioImage Study (SCAPIS, 2013–2018). An intersectional multilevel analysis of individual heterogeneity was conducted to examine disparities in cardiometabolic multimorbidity across socio-geographical intersectional strata (Paper I). A three-step latent class analysis was used to identify the co-occurrence of cardiometabolic risk factors within individuals and to estimate class-specified predicted prevalence of subclinical atherosclerosis (Paper II). A regularised Bayesian logistic regression model was fitted to prospectively predict the risk of cardiometabolic disease and multimorbidity, incorporating social and cardiometabolic risk factors as predictors (Paper III). Results Males aged 60–64 years with low education, irrespective of country of birth, exhibited the highest prevalence of cardiometabolic multimorbidity, particularly in Göteborg and Malmö. The latent class characterised by both behavioural and metabolic risk factors had the highest predicted prevalence of subclinical atherosclerosis. However, a mild but clinically meaningful subclinical atherosclerotic burden was also observed in the class characterised by unhealthy diets, despite having normal lipid levels. The regularised Bayesian logistic regression model demonstrated good calibration and discrimination in predicting the risk of cardiometabolic multimorbidity, with the highest predicted risk concentrated among males aged 60–64 years with low HDL-C, high waist circumference, and current or former smoking. Conclusion Cardiometabolic risk is inadequately captured by average effects or single-factor analyses; rather, it is shaped by the intersection of multiple risk factors. These findings support preventive strategies grounded in proportionate universalism and precision public health. The identified co-occurrence pattern of risk factors underscores the need to move beyond single-factor interventions toward tailored prevention strategies that align intervention intensity with individuals’ sociodemographic and risk factor profiles to prevent atherosclerosis, cardiometabolic disease, and multimorbidity.