Dynamics of Chronic Disease Burden in Older Adults: State-Level and Disease-Level Markov Transition Models Using the English Longitudinal Study of Ageing

The coexistence of two or more chronic conditions, known as multimorbidity, represents one of the key health challenges associated with an ageing population. This study merges two complementary longitudinal analyses of chronic disease accumulation in older adults in England, utilizing data from Waves 6–11 (2012–13 to 2023–24) of the English Longitudinal Study of Ageing (ELSA). The first analysis employs a first-order discrete-time Markov framework, estimated via multinomial logistic regression, to model health transitions at the aggregate state level (healthy, single morbidity, multimorbidity). The second analysis expands this framework to the disease level, estimating a comprehensive 7×7 matrix of disease-to-disease transition odds ratios among seven common chronic conditions to identify “gateway” diseases that promote the accumulation of multimorbidity .The two analyses together demonstrate that the deterioration of health in older adults is highly path-dependent: those with multimorbidity have an 81.3% chance of staying multimorbid and only a 1.5% chance of returning to a healthy state. They are expected to remain in this condition for 5.35 waves, in contrast to 3.62 waves for individuals in a healthy state. Hypertension and diabetes create a robust bidirectional reinforcing pathway at the disease level, with odds ratios of 2.89 and 4.25, respectively. Heart disease serves as the most significant gateway condition (out-degree = 3), while arthritis is revealed to be an overlooked gateway connecting musculoskeletal and cardiometabolic disease clusters. Sojourn times stratified by disease show that heart disease and lung disease hasten the progression to multimorbidity the most, while diabetes, despite its high network centrality, correlates with the slowest temporal progression. These findings reconceptualize multimorbidity as a dynamic, path-dependent, and disease-specific process rather than a static endpoint, directly impacting the timing and targeting of preventive interventions.