Choppy rest-activity rhythms linked to more deaths among 3,061 adults with heart, kidney or metabolic risk

Most fitness watches now hand down a verdict on the night: how long their wearer slept, how restless they were, sometimes how regular their bedtimes have been. Researchers have read the same kind of signal for years from research-grade wrist sensors, and a new US analysis suggests that for people with heart, kidney or metabolic disease, the shape of the whole 24-hour day tracks with how long they live.

Miao Huang, Renjie Huang and Dongyan Wang looked at wrist recordings from 3,061 American adults in a national health survey, all of them somewhere on the scale doctors now use for overlapping heart, kidney and metabolic disease. People whose daily activity was more broken up, or whose days and nights looked more alike, were more likely to die over the following years. So were people who slept less, slept less efficiently or fell asleep at shifting times. The study appeared on October 1 in the journal Sleep Health, and its authors are careful about what it shows: “These findings do not establish causality or clinical thresholds.”

What a rest-activity rhythm is, and how a wrist sensor reads it

The body keeps time on a clock of roughly 24 hours, and one of the easiest places to see that clock is in movement. A wrist accelerometer, a motion sensor worn like a watch, logs how much a person moves minute by minute, day and night, over several days. Researchers describe the resulting pattern as one of the most prominent outputs of the circadian system, the body’s internal clock, and they boil those days down to a few numbers.

Relative amplitude compares the most active continuous 10-hour period of an average day with the quietest five hours. Busy days and still nights give a high score, while a flattened rhythm, with less going on by day or more stirring at night, gives a low one. One research team computes it as the gap between those two activity levels divided by their sum, so it runs from 0 to 1, and describes it as a measure that reflects the robustness of the 24-hour rhythm. Intradaily variability tracks how broken up the pattern is. It sits near zero for a perfect sine wave, one smooth rise and fall a day, and climbs toward 2 for pure random noise. In practice a high score belongs to someone whose activity arrives in fits and starts, with naps and still spells scattered through the day. A third score, interdaily stability, asks how closely each day repeats the one before.

The same recordings yield sleep estimates. In earlier work on these survey recordings, sleep efficiency was calculated as sleep duration divided by the length of the night’s sleep window, so a score of 0.9 means asleep for 90% of it. The new study also tracked how far the midpoint of each person’s sleep moved from one night to the next, which captures how regular their sleep timing was.

Who was studied: adults with heart, kidney or metabolic disease

Cardiovascular-kidney-metabolic syndrome, shortened to CKM, is a framework the American Heart Association set out in 2023 for what it calls the interplay among metabolic risk factors, kidney disease and the heart. It comes in stages. As a 2026 review sums up the scheme, people in Stage 0 have no risk factors for CKM, and stage 1 begins with excess or poorly working body fat. Stage 2 shows risk factors of metabolic syndrome such as type 2 diabetes or raised blood fats, often with kidney damage. Stage 3 adds early heart disease that causes no symptoms yet, and stage 4 is heart disease that has already shown itself. By that review’s reading of national survey data, approximately 89.5% of US adults fall within the CKM stages from 1 to 4. The label therefore takes in most of the adult population, from someone carrying extra weight around the middle to someone who has survived a heart attack.

The recordings came from the National Health and Nutrition Examination Survey, or NHANES, a federal survey designed to yield a weighted, representative sample of the US population. In its 2011 to 2014 rounds, participants were asked to wear a wrist sensor around the clock. The Sleep Health team took the adults from those rounds whose archived survey data placed them in stages 1 to 4 and linked them to national death records, using NHANES 2011-2014 data linked to mortality through 2019. By the end of that window the records showed 274 all-cause and 79 heart-disease deaths among the 3,061, so roughly one participant in eleven had died.

Which daily rhythms and sleep habits tracked with deaths

The clearest results concerned the daily rhythm. A more broken-up day went with more deaths: higher intradaily variability was associated with higher death rates from any cause, by about 15% for each 0.1 step on that 0-to-2 scale. A stronger rhythm went the other way, and the paper reports that higher relative amplitude was associated with lower mortality, about a quarter lower for each 0.1 rise on its 0-to-1 scale.

The sleep measures lined up behind the rhythm. Longer mean sleep duration and higher sleep efficiency both went with fewer deaths, about 18% fewer for each extra hour of average sleep and about 30% fewer for each 10-point rise in efficiency, from 80% of the night asleep to 90%, say. Sleep timing pointed the same way: greater sleep-midpoint variability was associated with higher death rates, about a quarter higher for each extra hour that the middle of a person’s sleep wandered from one night to the next.

Two details temper those figures. The links for relative amplitude and sleep length were nonlinear, meaning the death rate did not move by the same amount at every point on the scale, so a figure “per hour” or “per 0.1” is an average across a curve. The results for sleep efficiency and timing also came with wide margins of error, the range for sleep timing running from barely raised risk to about half again higher, as might be expected when 274 deaths are spread across many measures. For the 79 heart-disease deaths taken alone, the researchers report that the results pointed the same way, though so few deaths leave more room for chance.

These measures overlap. In the earlier analysis of the whole NHANES wrist sample, researchers noted a relatively strong correlation between relative amplitude and sleep efficiency, which makes sense: someone who sleeps soundly through the night also has a deeper overnight trough in activity. The new results read best as several views of one underlying pattern, and adding them up as five separate risks would overstate them.

How the rhythm result lines up with earlier wrist studies

A wrist sensor has been linked to survival before, and the new study adds to a fairly consistent record. In the Rotterdam Study in the Netherlands, 1,734 people aged 45 to 98 wore one, and their mean follow-up time was 7.3 years, during which 154 died. A more fragmented rhythm was associated with a higher mortality risk in that group, and a more stable one with a lower risk. Sleep told a different story there: once the researchers took account of people’s health, the sleep measures were not related to mortality. On that last point the CKM study disagrees, since sleep length, efficiency and timing all tracked with deaths in its participants.

The closest precedent uses the very same recordings. Yanyan Xu and colleagues at Augusta University analyzed 7,252 adults from NHANES 2011 to 2014, of whom 674 died during a median of 81 months. People with a stronger rhythm had lower death rates, and in Xu’s words relative amplitude “outperformed all the tested traditional predictors of all-cause mortality with the exception of age.” Because the new study draws on the same survey rounds, many of its participants were probably in that analysis too. It narrows the lens to people with CKM and sets the sleep measures beside the rhythm, which makes it a closer look at old data more than a fresh test.

A much larger study had already looked at CKM specifically. Bingtao Weng and colleagues followed 74,777 UK Biobank participants with the syndrome and checked the results in 6,046 Americans from NHANES. Those in the top third for relative amplitude had about a 40% lower death rate from any cause than those in the bottom third, and higher intradaily variability was associated with increased all-cause mortality. The pattern held when the team set aside people who had heart events or died in the first two years, a check against illness already flattening their days. Markers of inflammation in the blood explained only a small slice of it: inflammatory biomarkers modestly mediated these associations, by 1% to 5%.

The broadest view comes from a meta-analysis published in September in the Journal of the American Heart Association. Fengshuo Chang and colleagues pooled seven studies with 109,365 participants. Lower relative amplitude went with a death rate about 46% higher, and higher intradaily variability was associated with a smaller rise of about 5%. The pooled studies measured these scores in different ways, so those figures cannot be set directly against the new study’s per-step numbers. Day-to-day stability, the third score, showed no link with any outcome that stood clear of chance.

Sleep timing has its own line of evidence. Daniel Windred and colleagues worked through 10 million hours of accelerometer data from 60,977 UK Biobank participants to score how regular each person’s sleep was. Compared with the least regular fifth of sleepers, each of the four more regular fifths had a lower risk of all-cause mortality, by between 20% and 48% depending on the group, and regularity was a stronger predictor of death than the number of hours slept. The wandering sleep midpoint in the new study is a different measure in a different population, and it points the same way.

What a single recording of rhythm data leaves open

The Sleep Health study is observational, and its authors claim no more than an association. People with flattened, broken-up days may differ from others in ways that also shorten life, such as how ill they already were, how easily they could move, or which medicines they took, and statistical adjustment cannot rule out every such difference.

Reverse causation is the obvious worry. Heart failure, kidney disease and the drugs used for them can sap energy, break up the day with rests and disturb sleep, so part of the link may run from failing health to the rhythm and not the other way round. Dropping the earliest deaths, as the Weng team did, is one guard against that, though it cannot catch illness that takes years to unfold.

The rhythm and sleep were measured once, over a few days, at the start. Habits change, and the Augusta group points out that ignoring such changes over time might have underestimated the magnitude of the associations; in other words, a single snapshot may make the links look weaker than they are. The death count was modest, especially for heart disease, and the paper sets no cut-off for a healthy or unhealthy score. The data describe the United States in 2011 to 2014, and the per-unit figures are averages along a curve, not a dose one person could dial up or down.

What rest-activity rhythms could mean for risk checks

For now the scores are, at most, a warning light. A wrist sensor costs little, involves no needles and records days of ordinary life, and the Chang team concluded that the rhythm scores could serve as early, noninvasive risk indicators. Weng’s group found that adding the rhythm scores to basic models improved prediction of deaths, though only by a small margin.

Nobody has yet shown that changing a rhythm changes the risk. The meta-analysis authors call for research into whether appropriately timed physical activity and sleep, along with other approaches, can lower cardiovascular risk, and Windred’s group writes that sleep regularity “may be a simple, effective target for improving general health”, a hypothesis that still needs a trial. Whether any one person’s sleep or daily activity pattern needs attention is assessed by their doctor.

The new study adds US adults with heart, kidney or metabolic disease to the list of groups in whom a flatter, choppier day on a wrist sensor went with shorter survival, and it leaves the question of cause where the earlier studies left it.