Short Sleep and Too Little Exercise May Reshape How Chronic Diseases Cluster Together

Two of the most familiar pieces of health advice — get enough sleep and move your body regularly — have just acquired a new layer of scientific significance. A large cross-sectional study drawing on data from more than 18,000 middle-aged and older Chinese adults suggests that the combination of insufficient physical activity and short sleep is not merely linked to a higher chance of living with multiple chronic conditions. It also appears to change the very architecture of disease clustering, altering which illnesses tend to travel together in the same person and how tightly those illnesses are woven into a network of comorbidity. The findings, published in BMC Public Health by a team at Hainan Medical University, offer one of the most detailed behavioral maps of multimorbidity to date, and they arrive at a moment when aging societies worldwide are grappling with the mounting burden of patients who carry two, three, or more chronic diseases at once.

Multimorbidity — conventionally defined as the coexistence of two or more chronic conditions in one individual — has become one of the defining public health challenges of the twenty-first century. Health systems were largely designed around single-disease specialties, yet the typical older patient rarely presents with just one problem. Hypertension arrives with diabetes, dyslipidemia follows, arthritis limits mobility, digestive complaints accumulate, and the clinical picture becomes a tangle rather than a single thread. The research team, led by Lianhua Liu and Weixia Li with corresponding authors Li Cao and Bo Bi, set out to ask whether two modifiable behaviors — how much people sleep and how much they move — are associated not only with the likelihood of multimorbidity but with the specific patterns in which diseases cluster. That second question is far less commonly asked, and answering it required a method borrowed from a very different field: data mining.

The evidence base came from the 2020 wave of the China Health and Retirement Longitudinal Study, known as CHARLS, a nationally representative survey of adults aged 45 and older. The analysis included 18,664 participants, a sample large enough to support both conventional statistical modeling and more exploratory pattern discovery. Multimorbidity was defined as the presence of two or more chronic diseases among the fifteen assessed in the survey, a list spanning cardiometabolic conditions such as hypertension, diabetes, and dyslipidemia; respiratory illnesses including asthma and chronic lung diseases; osteoarticular problems; and digestive disorders. Participants were classified by physical activity level and by sleep duration, and the researchers then examined how multimorbidity prevalence and disease patterns varied across the resulting behavioral subgroups.

The headline numbers from the conventional analysis are striking on their own. In fully adjusted multivariable logistic regression models — meaning the researchers accounted for sociodemographic and behavioral covariates that could otherwise confound the picture — short sleep was associated with roughly 64 percent higher odds of multimorbidity, with an odds ratio of 1.640 and a 95 percent confidence interval of 1.505 to 1.788. Insufficient physical activity carried a more modest but still meaningful association, with an odds ratio of 1.187 (95 percent CI: 1.060 to 1.329). But the most eye-catching result emerged when the two behaviors were considered together: participants who both slept too little and moved too little faced the highest odds of all, with an odds ratio of 2.148 (95 percent CI: 1.818 to 2.539). In plain terms, the combination more than doubled the odds of multimorbidity compared with the reference group, a magnitude that exceeds what either behavior alone would predict and hints at a compounding relationship worth probing in future longitudinal work.

To go beyond prevalence, the team turned to association rule mining, or ARM, a technique developed in the world of market-basket analysis, where retailers use it to discover which products tend to be purchased together. Applied here, each chronic disease becomes an item in a basket, and the algorithm searches for rules of the form: if a person has disease A, how likely are they also to have disease B? The researchers also constructed comorbidity networks, in which diseases are nodes and statistically supported co-occurrences are edges, allowing network density — how interconnected the disease web is — to be compared across behavioral subgroups. It is important to note, as the authors do, that these ARM analyses were exploratory and unadjusted, so they generate hypotheses rather than confirm causal pathways. Even so, the contrast between subgroups is arresting.

Across every subgroup, the highest-support multimorbidity patterns clustered around two broad territories: cardiometabolic disease and the combined osteoarticular-digestive cluster. At the center of the web sat hypertension, which functioned as the central hub — the disease most frequently connected to others and the most common anchor of strong association rules. This hub status is biologically plausible. Hypertension shares risk factors, including vascular dysfunction, metabolic stress, and chronic low-grade inflammation, with diabetes, dyslipidemia, stroke, and heart disease, and it is among the most prevalent conditions in middle-aged and older populations, giving it ample opportunity to co-occur with nearly everything else. What changed between subgroups was not the identity of the hub but the density of the network built around it.

That density gradient tracked the behavioral profile with remarkable consistency. Participants with both insufficient physical activity and short sleep exhibited the highest network density and generated the most association rules — 54 in total — suggesting that in this group, diseases were more tightly and more variously interconnected, with more statistically supported co-occurrence pathways linking one condition to the next. At the opposite extreme, participants with both sufficient physical activity and optimal sleep duration showed the lowest network density and produced only 8 association rules, the fewest of any subgroup. The implication, though it cannot be read causally from a cross-sectional design, is that favorable behavioral profiles are associated with a sparser disease web, in which chronic conditions are less likely to pile up in tightly bound clusters. If future longitudinal studies confirm this, behavior-stratified prevention strategies could target not just individual diseases but the clustering process itself.

Perhaps the most intriguing findings are the subgroup-specific association rules — disease pairs that appeared in some behavioral strata and were entirely absent in others. The rule linking asthma to chronic lung diseases surfaced exclusively in subgroups characterized by short sleep or insufficient physical activity, hinting that respiratory vulnerability may be a signature of these behavioral profiles. The rule connecting hypertension and diabetes to dyslipidemia was absent in the long-sleep subgroup, and rules with stomach or other digestive diseases as the consequent — meaning digestive disease appearing as the outcome in the rule — were identified only in the short-sleep subgroup. These are exploratory observations, but they are the kind of pattern that can seed targeted research questions: does sleep restriction alter gut physiology, immune regulation, or stress-hormone pathways in ways that make digestive disease a more common companion of other chronic conditions? The authors point toward plausible mechanisms, including dysregulation of the hypothalamic-pituitary-adrenal axis, the body’s central stress-response system, which is known to be perturbed by both sleep deprivation and sedentary lifestyles and to influence metabolic, immune, and inflammatory processes.

The study’s strengths lie in its scale, its nationally representative sampling frame, and its methodological creativity in pairing traditional regression with network-based pattern discovery. Its limitations are equally clear and worth stating plainly: the cross-sectional design captures a single moment in time, so it cannot establish whether short sleep and inactivity cause multimorbidity, whether illness disrupts sleep and activity, or whether both are driven by shared underlying factors. The ARM analyses were unadjusted, and self-reported sleep and activity data carry well-known measurement limitations. The authors are careful to frame the subgroup differences as hypothesis-generating — directions for research on behavior-stratified prevention rather than prescriptions. Still, the practical resonance is hard to ignore. If the constellation of chronic diseases a person develops is associated with how they sleep and move, then interventions promoting adequate sleep duration and sufficient physical activity may do more than lower the risk of any single illness; they may help keep the entire web of comorbidity looser, simpler, and more manageable.

For a world in which populations are aging rapidly and multimorbidity threatens to overwhelm health systems built for single diseases, that reframing matters. The message emerging from this study is not simply that sleep and exercise are good for you — a conclusion few readers will find surprising. It is that these everyday behaviors may shape the topology of chronic disease itself, influencing which conditions cluster together, how densely they interconnect, and how predictable one diagnosis becomes from another. As the researchers suggest, the next step is to test these hypotheses prospectively, in designs that can untangle timing and causation. Until then, the study stands as a compelling piece of evidence that the fight against multimorbidity may begin not in the clinic, but in the bedroom and on the walking path.

Subject of Research: Associations of physical activity and sleep duration with the prevalence and patterns of multimorbidity in middle-aged and older adults

Article Title: Physical activity, sleep duration, and multimorbidity: a cross-sectional study of prevalence associations and association rule mining of multimorbidity patterns

Article References: Liu, L., Li, W., Gui, M., Ju, F., Wang, Q., Cao, L., & Bi, B. (2026). Physical activity, sleep duration, and multimorbidity: a cross-sectional study of prevalence associations and association rule mining of multimorbidity patterns. BMC Public Health. https://doi.org/10.1186/s12889-026-29846-0

Image Credits: AI Generated

DOI: 10.1186/s12889-026-29846-0

Keywords: multimorbidity, physical activity, sleep duration, association rule mining, comorbidity network, chronic disease, hypertension, CHARLS, public health, aging, cardiometabolic disease, cross-sectional study