The SPRINTT trial, funded by the European Union’s Horizon 2020 program, was one of the largest randomized controlled trials ever conducted in community-dwelling frail older Europeans. It enrolled more than 1,500 men and women aged seventy and above across multiple European countries, all of whom met criteria for physical frailty and sarcopenia — a combination characterized by slowness, weakness, low physical activity and reduced muscle mass. Participants were randomized to either a moderate-intensity, multicomponent physical activity program, centered on walking and adapted to individual capacity, or to a control group receiving structured health education. The primary goal was to prevent major mobility disability, defined as the inability to walk 400 meters, a threshold with proven clinical relevance because it captures the capacity to function independently in everyday life. Earlier reports from the trial indicated that the intervention produced statistically significant but modest benefits, leaving open the crucial question of whether particular subgroups of patients benefited more than others.

The new study addresses that question through the lens of multimorbidity patterns rather than simple disease counts. Most previous analyses treated multimorbidity as a number — two diseases, three diseases, four or more — or as a crude index of cumulative burden. But clinicians have long observed that a patient with diabetes, peripheral vascular disease and osteoarthritis is not clinically equivalent to a patient with chronic obstructive pulmonary disease, heart failure and depression, even if both carry the same number of diagnoses. The researchers therefore applied statistical clustering techniques to the participants’ disease profiles, identifying groups of conditions that tended to occur together and grouping individuals according to the characteristic pattern of their chronic illnesses. This approach, sometimes described as person-centered rather than disease-centered, allowed the team to ask whether the protective effect of exercise against mobility loss was homogeneous across these clinically distinct constellations of disease.

The technical rationale for such heterogeneity is compelling. Different disease clusters disable the body through different physiological pathways. Cardiometabolic patterns, typically combining type 2 diabetes, hypertension and obesity, limit mobility largely through vascular damage, impaired muscle perfusion and peripheral neuropathy, all of which erode the capacity for sustained aerobic effort. Osteoarticular patterns, dominated by osteoarthritis and chronic pain, restrict movement through mechanical limitation and pain-avoidance behavior that accelerates deconditioning. Cardiorespiratory and neuropsychiatric patterns act through reduced aerobic reserve, breathlessness and fatigue, and through the motivational deficits associated with depression. A walking-based intervention that increases daily physical activity might, in principle, counteract deconditioning in all of these groups, but the magnitude of the achievable gain could differ substantially depending on which bottleneck — vascular, mechanical, respiratory or psychological — dominates the individual patient’s trajectory.

The analysis confirmed that suspicion in a clinically meaningful way. The benefits of the multicomponent physical activity program on the risk of developing major mobility disability were not uniform across multimorbidity patterns. For some clusters of conditions, the intervention produced clear and robust protection: frail older adults whose chronic disease burden followed certain patterns experienced a significantly lower hazard of losing the ability to walk 400 meters when they exercised regularly compared with their counterparts receiving health education alone. For other patterns, the observed benefit was attenuated and statistically uncertain, suggesting that standard exercise prescriptions may need to be adapted — or supplemented with other treatments — to reach patients whose disabling pathway is driven by disease mechanisms that physical activity alone cannot fully offset. The precise estimates, hazard ratios and interaction terms reported in the article quantify this differential effect, providing effect-size benchmarks that guideline developers and trial designers can build upon.

Methodologically, the study is a careful exercise in post-hoc subgroup science, a field notorious for false positives. The authors handled the inherent risks by defining multimorbidity patterns using prespecified statistical procedures, applying clustering algorithms to baseline disease data, testing interactions between intervention assignment and pattern membership, and adjusting for the covariates that typically confound mobility outcomes, including age, sex, baseline physical performance, body mass index and gait speed. Sensitivity analyses examined whether the results were robust to alternative cluster definitions and to variations in the handling of missing data. Although the analysis was not powered a priori for pattern-specific comparisons — a limitation the authors acknowledge — the coherence of the findings across analytic choices lends credibility to the central conclusion: multimorbidity is not a monolith, and its internal structure matters for prevention.

The implications reach well beyond the walls of geriatric medicine. Health systems across Europe, North America and Asia are confronting a demographic transition in which the majority of people over sixty-five live with at least two chronic conditions. Current guidelines for these patients are typically assembled disease by disease — a cardiology recommendation, a diabetes recommendation, an orthopedic recommendation — with little attention to how the combination of conditions changes what prevention can achieve. The SPRINTT analysis provides an empirical basis for a different model, one in which the pattern of multimorbidity becomes a stratification variable in clinical decision-making. A physical activity prescription for a frail older patient with a cardiometabolic cluster may be among the most effective interventions available; for a patient whose mobility is limited by a different cluster, the same prescription may need reinforcement with pain management, nutritional support, depression treatment or assistive technology to translate into preserved walking ability.

The findings also intersect with a growing body of research on physical resilience — the capacity of an organism to resist and recover from health stressors. Frailty, understood as a state of diminished physiological reserve, has often been treated as a single construct measured with composite scores such as the frailty phenotype or the frailty index. The SPRINTT results suggest that the trajectory of frail older adults is better predicted by a vector than by a scalar: the direction of their disease burden, not merely its magnitude, determines how plastic their mobility remains. This resonates with mechanistic work on the biology of aging, in which distinct molecular hallmarks — chronic inflammation, mitochondrial dysfunction, cellular senescence, neuromuscular junction degeneration — are differentially engaged by different chronic diseases and may respond differently to exercise as a systemic intervention. Physical activity is one of the few therapies known to act simultaneously on most of these pathways, which helps explain why it remains effective, if unevenly so, across clinically diverse populations.

Questions inevitably remain. The trial population consisted of community-dwelling Europeans selected for physical frailty and sarcopenia, so the generalizability of the pattern-specific results to frailer institutionalized populations, to non-European cohorts or to younger adults with early multimorbidity has not been demonstrated. The clustering solution chosen by the investigators is one of several statistically defensible partitions of the disease space, and alternative algorithms might yield patterns with different boundaries. And although major mobility disability is a validated and consequential endpoint, future work should examine whether multimorbidity patterns also moderate the effects of exercise on other outcomes, including falls, hospitalization, cognitive decline and mortality. Longer follow-up, larger samples and replication in independent cohorts will be needed before pattern-stratified exercise prescriptions become standard clinical practice.

Even with those caveats, the study marks a turning point in how the prevention of disability in old age can be conceptualized. It moves the field away from the blunt arithmetic of disease counting and toward a nosology of combinations, in which the specific constellation of conditions a person carries is treated as clinically actionable information. It also delivers a pragmatic message of hope with nuance: exercise remains one of the most powerful tools available for keeping frail older adults on their feet, but its power is conditional, and understanding those conditions is the key to unlocking it for everyone. As populations age and multimorbidity becomes the norm rather than the exception, the lesson from SPRINTT is that precision geriatrics — matching preventive interventions to the pattern of disease, not just to its burden — is no longer an aspiration but an evidence-backed necessity.

The study, led by Joel Cohen of Rockefeller University’s Laboratory of Populations and Jonathan Tobin, director of Community-Engaged Research at Rockefeller’s Center for Clinical and Translational Science, identifies a mathematical pattern in the way multimorbidity varies with age. The researchers found that as the average burden of chronic disease rises, the differences between individuals also expand. In practical terms, two people of the same age are likely to have increasingly dissimilar combinations of illnesses as they grow older. The finding challenges the widespread use of chronological age as a simple guide for medical screening and treatment decisions and points toward a more individualized approach based on each patient’s actual health profile.

The analysis emerged from the Tipping Points clinical trial, a large effort focused on patients who are frequently missing from conventional medical research. These patients often receive care through Federally Qualified Health Centers, which provide primary and preventive services to millions of people in low-income communities across the United States. Because people with multiple serious conditions can introduce numerous confounding variables into a clinical trial, they are often excluded from studies that eventually shape medical guidelines. Tobin and his colleagues instead placed multimorbid patients at the center of their research, seeking to understand how their health changes and whether targeted support can prevent hospitalizations.

The research team examined de-identified electronic health records from 238,156 people receiving care through 16 Federally Qualified Health Centers in New York City and Chicago. The data were assembled with the help of clinical research networks and health information exchanges, including INSIGHT, CAPriCORN, Healthix, BronxRHIO, and AllianceChicago. The records included each individual’s age, location, and score on the enhanced Charlson Comorbidity Index, or eCCI. Nearly 2,000 patients were subsequently enrolled in the Tipping Points trial, in which health coaches helped participants manage their conditions and recognize problems before they escalated into emergency visits or hospital admissions.

The eCCI provided the mathematical foundation for the new analysis. The index assigns weights to chronic conditions according to their association with hospitalization risk and healthcare costs. Less severe conditions, such as myocardial infarction, congestive heart failure, and peripheral vascular disease, receive lower scores, while conditions including metastatic solid tumors, AIDS, and organ transplants receive higher weights. Most of the 39 categories included in the index fall between these extremes. By combining the scores, researchers can estimate the overall burden of disease carried by an individual rather than simply counting the number of diagnoses.

Cohen examined how the average eCCI score changed across age groups and, crucially, how widely individual scores were scattered around each age-specific average. That second measurement—statistical variance—proved to be the key result. The average burden of multimorbidity increased with age, as expected, but the variance increased as well. Older age groups therefore contained a wider range of health profiles. While some older adults had relatively limited chronic disease, others had extensive and severe multimorbidity, producing a much broader spread than was seen among younger adults.

The pattern resembles Taylor’s law, a mathematical relationship Cohen has identified in diverse biological and social systems, from infectious diseases and wildlife populations to human censuses and weather. Taylor’s law generally describes a power-law relationship between the mean of a population and its variance: as the average level of a phenomenon changes, the amount of variation around that average changes in a predictable way. In this study, the researchers found that the variability of eCCI scores rose in a mathematically consistent relationship with the mean. The result indicates that aging is not simply associated with a steadily increasing number of diseases; it is associated with a widening divergence in the kinds and severity of diseases experienced by different people.

The contrast can be illustrated by comparing patients in their forties with patients in their seventies. Two 40-year-olds may differ in their health, but their overall chronic disease profiles tend to be more alike than those of two 70-year-olds. By the time people reach older age, their medical histories have been shaped by different genetics, environmental exposures, behaviors, social conditions, access to care, treatments, and chance events. One person may have accumulated cardiovascular disease and diabetes, while another may have cancer and chronic lung disease, and a third may have relatively few serious diagnoses. Chronological age alone cannot capture those differences.

That finding has direct implications for clinical guidelines, many of which use age thresholds to determine when screening or preventive care should begin or end. The U.S. Preventive Services Task Force, for example, recommends colorectal cancer screening for adults beginning at age 45 and continuing through age 75, while biennial mammography is recommended for many women between ages 40 and 74. Such recommendations are essential population-level tools, but the new analysis suggests that their application may need to account more explicitly for the medical complexity of individual patients. A treatment or screening strategy that is appropriate for one 70-year-old may be ineffective, burdensome, or even inappropriate for another with a very different combination of conditions.

The researchers say the mathematical relationship could also help identify patients approaching a “tipping point,” when the accumulation or interaction of chronic conditions sharply increases the likelihood of hospitalization or disability. If future studies confirm that certain multimorbidity patterns predict rapid deterioration, clinicians might be able to intervene earlier with medication adjustments, health coaching, social support, or closer monitoring. The study does not establish that the mathematical pattern itself can predict an individual hospitalization, and the eCCI is an aggregate measure rather than a detailed map of disease interactions. Even so, the results provide a framework for moving beyond age-based assumptions. As Cohen puts it, the central lesson is that medical care cannot be one-size-fits-all: the older people become, the more important it is to understand the particular constellation of conditions carried by each person.

Subject of Research: Multimorbidity, aging, chronic disease variation, population health, and personalized medicine.

Web References: Rockefeller University; Journal of Population Ageing article DOI: https://doi.org/10.1007/s12062-026-09571-7

References: Journal of Population Ageing, DOI 10.1007/s12062-026-09571-7.

Keywords: Multimorbidity, chronic diseases, aging, life expectancy, healthcare, hospitalization, Charlson Comorbidity Index, eCCI, Taylor’s law, mathematical modeling, personalized medicine, public health, clinical guidelines, health disparities.