Multimorbidity and Mortality: A Review of U.S. Evidence and Modelling Approaches

Multimorbidity, or the presence of two or more chronic conditions, represents a significant and expanding public health concern in the U.S., particularly among older adults and underserved subpopulations. This article combines results of various US longitudinal and cross-sectional studies on the association of multimorbidity and mortality with a focus on modeling strategies as well as public health implications. It was demonstrated that the presence of multi-morbidity is associated with an increased mortality risk in a dose-response-like fashion, as shown by steadily increasing hazard ratios. Combinations like cardiometabolic and mental health disorders have disproportionately high mortality burdens. Race, poverty, and access to healthcare are the key effect modifiers with significant race and rural disparities in results. Typical modeling methods include Cox Proportional Hazards models, competing risks models, latent class analysis, the more sophisticated survival tree methods, a wrapper of deep learning, and recurrent neural networks. However, interpretation and prediction strengths differ between these models, and unification of comorbidity definitions, lack of representation of minority groups, and longitudinal cohort-based causal modeling are some persistent challenges. Real-world data integration and generalizability gaps also hinder policy translation. We emphasize advanced modeling methods, diverse and representative cohorts, and integrating electronic health records with administrative databases to overcome the limitations inherent to forecasting. Dealing with multimorbidity is an action at the national level; in addition to being implemented, correct modeling has a great potential for defining health care policy for Medicare, preventing diseases, and even distributing resources more equitably.