IPM Take
Cardiometabolic medicine is learning to think in systems. Diabetes is no longer treated as a glucose problem alone, cardiovascular disease is increasingly linked with kidney health and obesity, and the newest cardiovascular-kidney-metabolic frameworks explicitly recognise that these conditions reinforce one another across the life course.
Sleep still sits awkwardly at the edge of that model.
A new study from Karolinska Institutet suggests that may be increasingly difficult to justify. Among more than 400,000 UK Biobank participants, people with type 2 diabetes, heart disease or stroke who also reported healthier sleep performed slightly better on cognitive testing and, in the imaging subgroup, had larger total, grey and white matter brain volumes than those with cardiometabolic disease and poor sleep.
That is not evidence that better sleep will prevent dementia.
It is evidence that sleep may identify another layer of vulnerability within a population already known to face elevated cognitive risk.
The policy implication is therefore less about prescribing an extra hour in bed and more about recognising that cardiometabolic care already has reasons to ask about sleep. Diabetes guidance recommends screening for sleep health, while current cardiovascular-kidney-metabolic guidance recommends targeted assessment for obstructive sleep apnoea in higher-risk patients.
The unanswered question is whether routine sleep assessment should become broader still.
If sleep turns out to modify cognitive vulnerability among people with cardiometabolic disease, then treating it as a lifestyle footnote may eventually look as outdated as treating kidney function as somebody else’s problem.
Executive Summary
A study published in the Journal of the American Heart Association analysed data from 408,977 adults in the UK Biobank to examine whether healthier sleep patterns were associated with better cognitive and brain outcomes among people with cardiometabolic disease.
At baseline, 54,017 participants, or 13.2%, had at least one cardiometabolic disease, defined in the study as type 2 diabetes, heart disease or stroke.
Sleep was assessed using a composite score incorporating five self-reported characteristics: sleep duration, chronotype, insomnia symptoms, snoring and daytime sleepiness. Participants were categorised as having healthy, intermediate or poor sleep patterns.
Among people with cardiometabolic disease, those with healthier sleep performed modestly better on measures of overall cognition, verbal and numerical reasoning, and processing speed than those reporting poor sleep.
Brain MRI data were available for 34,913 participants. Among participants with cardiometabolic disease, healthier sleep was associated with larger total brain volume as well as larger grey and white matter volumes.
The relationship between sleep and several structural brain measures appeared stronger among people with cardiometabolic disease than among participants without these conditions.
The study remains observational. Sleep was largely self-reported, healthier sleep may cluster with other favourable behaviours and health characteristics, and the analysis did not test a sleep intervention or measure whether improving sleep reduces dementia incidence.
The findings therefore support sleep as a potentially modifiable factor worthy of further study rather than establishing sleep improvement as a neuroprotective treatment.
Why it matters
- HTA bodies: Sleep interventions are increasingly relevant to chronic-disease pathways, but evidence needs to move beyond associations with cognition, imaging or intermediate biomarkers. The key assessment question is whether diagnosing and treating specific sleep problems in cardiometabolic populations improves clinically meaningful outcomes, including functional cognition, dementia incidence, cardiovascular events and quality of life.
- Payers: Screening for sleep disorders may already be justified in selected cardiometabolic populations, particularly where obstructive sleep apnoea is suspected. Broader routine sleep assessment could become attractive because it is relatively low cost, but reimbursement policy should distinguish simple sleep-history screening from formal sleep studies and treatments that require evidence of clinical benefit.
- Industry / innovation partners: Wearables and digital sleep tools can measure duration, timing and nocturnal physiology at unprecedented scale, but the commercial opportunity should not outrun the evidence. Tools that merely generate sleep scores will have limited clinical value unless they can identify actionable sleep problems and demonstrate that intervention improves outcomes in high-risk populations.
Cardiometabolic medicine has spent the past several years becoming less fragmented.
Type 2 diabetes is increasingly managed with cardiovascular and kidney outcomes in mind. Heart failure pathways now account for obesity, kidney dysfunction and metabolic disease. The concept of cardiovascular-kidney-metabolic syndrome has formalised what clinicians have observed for years: disease does not remain inside the specialty box in which it was first diagnosed.
The brain may be the next organ forcing that conversation to widen.
People with type 2 diabetes, heart disease and stroke are already known to face greater risk of cognitive decline and dementia. Some of that vulnerability is likely to emerge through vascular injury, metabolic dysfunction, inflammation and other pathways shared across cardiovascular and neurological disease.
A new analysis from Karolinska Institutet asks whether sleep may influence that relationship.
Researchers studied 408,977 middle-aged and older adults enrolled in the UK Biobank. Just over 54,000 had at least one of three cardiometabolic conditions: type 2 diabetes, heart disease or stroke.
Rather than looking only at sleep duration, the study treated sleep as a multidimensional behaviour.
Participants were scored across five characteristics including how long they slept, whether they tended toward an earlier or later circadian preference, insomnia symptoms, snoring and daytime sleepiness. Those characteristics were then combined into healthy, intermediate and poor sleep patterns.
That approach matters because asking whether somebody sleeps seven hours is not the same as asking whether they sleep well.
A person can spend an apparently adequate amount of time in bed and still experience fragmented sleep, insomnia, sleep-disordered breathing or significant daytime sleepiness. Conversely, a single sleep metric may miss problems occurring across other dimensions.
When researchers compared sleep patterns with cognitive performance, people with cardiometabolic disease who fell into the healthier sleep category performed slightly better than those with the same diseases who had poor sleep.
The differences appeared in overall cognitive performance as well as verbal and numerical reasoning and processing speed.
The word slightly matters.
This was not a dramatic separation between cognitively healthy and impaired groups, and the findings should not be translated into a promise that better sleep can reverse the cognitive consequences associated with diabetes or cardiovascular disease.
The imaging results added a second signal.
Nearly 35,000 participants had undergone brain MRI. Among people with cardiometabolic disease, healthier sleep was associated with larger overall brain volume as well as greater grey and white matter volumes.
Brain volume naturally changes with age, and smaller volumes in particular regions or tissue compartments can accompany neurodegenerative and vascular processes. But an imaging association is still not a clinical outcome.
A larger brain volume in one observational comparison does not prove that sleep preserved brain tissue, just as a smaller volume does not prove that poor sleep caused brain atrophy.
The study design cannot establish the direction of the relationship.
People in better overall health may sleep better and have healthier brains. People with early neurological changes may sleep worse. Depression, medication use, socioeconomic circumstances, physical activity, obesity and many other factors can affect both sleep and cognition.
Statistical adjustment can reduce some of those differences.
It cannot make an observational study behave like a randomised trial.
That limitation is especially important because sleep has become an increasingly popular target for health optimisation, and the commercial ecosystem around trackers, sleep scores, supplements and digital tools can make associations sound more actionable than they really are.
The current evidence supports paying attention to sleep.
It does not support promising that optimising a wearable sleep score will prevent dementia.
Where the study becomes more interesting is in the context of cardiometabolic care that is already beginning to take sleep more seriously.
The American Heart Association includes sleep among its Life’s Essential 8 components of cardiovascular health, alongside diet, physical activity, tobacco exposure, body weight, cholesterol, blood glucose and blood pressure.
That inclusion reflects a wider understanding that sleep is not simply recovery time after the medically important part of the day has ended.
Poor and insufficient sleep are associated with hypertension, obesity, diabetes and cardiovascular disease, while disorders such as obstructive sleep apnoea are particularly common in people with obesity and cardiometabolic disease.
The 2026 cardiovascular-kidney-metabolic guideline goes further in selected populations by recommending regular assessment for symptoms of obstructive sleep apnoea among people with CKM syndrome and obesity.
Diabetes guidance is broader still.
The 2026 American Diabetes Association Standards of Care recommend screening people with prediabetes or diabetes for sleep health, including sleep disorders and diabetes-related sleep disruption, and recommend counselling patients on sleep-promoting routines.
In other words, the health system already has reasons to ask cardiometabolic patients about sleep.
What remains uncertain is whether brain health provides another reason to make that assessment more systematic.
The new study suggests that sleep may matter particularly in people whose cardiometabolic disease has already placed them at elevated risk of cognitive decline.
That is an important distinction.
A prevention intervention does not necessarily need to produce the same benefit in everyone to be worthwhile. It may have greater value in populations with less physiological reserve or multiple overlapping risks.
But demonstrating that requires intervention studies.
Researchers now need to move from asking whether healthier sleepers have healthier brains to asking whether improving sleep changes the trajectory.
That means identifying the problem first.
“Poor sleep” is not one disease.
For one patient, the relevant condition may be obstructive sleep apnoea. Another may have chronic insomnia. Another may work rotating shifts. Someone else may spend too little time sleeping because of caring responsibilities, work schedules or socioeconomic stress.
Those mechanisms are unlikely to respond to the same intervention.
This is why generic advice to “sleep better” has limited clinical value.
A patient with untreated sleep apnoea needs diagnostic assessment and appropriate therapy. Someone with chronic insomnia may benefit from structured behavioural treatment. A shift worker may face an occupational problem that individual sleep hygiene cannot fully solve.
That difference also matters for policy because it determines who is responsible.
Some sleep problems belong primarily to clinical medicine.
Others are shaped by employment conditions, housing, mental health, caregiving and social circumstances that are much harder to change inside a consultation.
If sleep is eventually shown to modify cognitive vulnerability in cardiometabolic populations, care pathways will therefore need more than a checklist asking patients how many hours they sleep.
They will need to identify which sleep problems are actionable and provide somewhere for patients to go next.
The current study cannot show that such a pathway prevents dementia.
It did not randomise participants to a sleep intervention, and dementia incidence was not the main outcome being tested. Its strongest evidence lies in the consistency of the association across several cognitive and structural brain measures within an exceptionally large population.
The UK Biobank itself also introduces limitations.
Participants tend to be healthier and less socioeconomically deprived than the general UK population, which means associations observed in the cohort may not translate perfectly to populations with a higher burden of disease or poorer access to healthcare.
Most sleep measures were self-reported rather than objectively recorded, introducing the possibility of reporting error.
Yet the scale of the dataset allows researchers to examine an increasingly important clinical question with considerably greater precision than most sleep studies can achieve.
The result is a signal worth following rather than a treatment recommendation ready to implement.
That distinction should shape the policy response.
Health systems do not need to wait for a dementia-prevention trial before asking about sleep in patients with diabetes or cardiovascular disease, because current guidelines already provide reasons to identify sleep disorders and poor sleep health.
But they should resist claiming that sleep improvement has been proven to preserve brain volume or prevent cognitive decline.
The more immediate opportunity is integration.
A patient with type 2 diabetes should not have glucose treated while chronic insomnia goes unmentioned. A person with obesity and cardiovascular disease should not repeatedly present with daytime sleepiness and loud snoring without somebody considering sleep apnoea. A stroke survivor’s sleep problems should not be treated as an unrelated quality-of-life complaint if they may interact with neurological recovery and broader cardiometabolic health.
Those are less dramatic changes than a new drug.
They may also be easier to implement.
Cardiometabolic care has already accepted that the heart, kidneys and metabolic system cannot be managed independently.
The new evidence suggests that the brain, and the hours spent sleeping, may belong in the same conversation.
The study does not show that sleep can erase the neurological consequences of cardiometabolic disease. It does suggest that a health system serious about integrated cardiometabolic prevention may eventually have to stop treating sleep as what happens between appointments.

