Comorbidity Patterns and Temporal Pathways in Metabolic Multimorbidity: A Five-Year Retrospective Cohort Study of Dyslipidemia, Hyperuricemia, Obesity, and Diabetes

Abstract Background: Dyslipidemia, hyperuricemia, obesity, and type 2 diabetes frequently co-occur, yet the structural organization, temporal ordering, and directional predictive relationships among these conditions remain poorly characterized. Methods: This retrospective cohort study included 8,383 adults undergoing annual health examinations at Beijing Chao-Yang Hospital (2020–2024), contributing to five observations each. Four metabolic diseases were defined using established diagnostic criteria. Several complementary methods were applied: comorbidity network analysis, generalized estimating equations (GEE), sequence analysis, group-based trajectory modeling, multi-state Markov modeling, time-varying covariate Cox regression, and latent class analysis. Multiple-comparison corrections were applied throughout. Results: Among 8,383 participants (62.27% female; mean age 45.82 ± 13.50 years) with 41,915 person-observations over five years, dyslipidemia was the most prevalent condition (32.53%), followed by hyperuricemia (15.75%), obesity (13.59%), and diabetes (7.23%). Dyslipidemia occupied the central position in the comorbidity network (node strength = 0.59) and served as the most common metabolic gateway (one-year onset probability 12.28%). The dyslipidemia–hyperuricemia pair showed the strongest co-occurrence (odds ratios [OR] = 3.74, 95% confidence intervals [CI] 3.46–4.06), while the diabetes–hyperuricemia association was fully attenuated after adjustment (OR = 1.02; p -FDR = 0.782). Four distinct longitudinal trajectories were identified: persistently healthy (54.38%), stable single-disease (29.01%), stable dual-disease (11.15%), and persistently high-burden (5.46%). All four trajectories remained flat over five years, indicating that metabolic comorbidity patterns persist once established. Directional analyses revealed that obesity preceded diabetes in 83.87% of sequential onsets, with the obesity → diabetes pathway representing the strongest unidirectional effect (hazard ratios [HR] = 2.52, 95% CI 2.06–3.09). Diabetes-involving multimorbidity states showed high retention (probability 0.73) and obesity had the lowest remission probability (11.32%). Four comorbidity phenotypes were identified, including a dyslipidemia–obesity–diabetes cluster with low hyperuricemia probability, consistent with sex-stratified findings showing the diabetes–hyperuricemia association reversed direction between males and females. Conclusions : Metabolic multimorbidity is a structurally ordered process in which dyslipidemia acts as the central hub and the most common gateway condition, obesity serves as the dominant unidirectional driver of diabetes, and established multi-disease states, particularly those involving diabetes, are largely irreversible. Early lipid screening and weight management may offer the most effective intervention opportunities before comorbidity patterns become self-sustaining.