Deep divefrom Episode 13Sleep & recovery12 November 202618 min

Does Sleeping Next to Your Dog Change Your Heart?

A deep dive into one paper: what was known before, what the researchers did, what the numbers say and what they don't.

The paper: Who's Been Sleeping in My Bedroom? Co-sleeping with dogs, cats or children among WHOOP users (Nature and Science of Sleep, 2026)

MP3

Chapters

Transcript

The script the two AI voices read from. The times are worked out from the script, not heard, so a line can start a few seconds early or late.

Read the whole episode

Cold open

Host 1If I let my dog sleep in my bedroom, does my body actually sleep differently that night?

Host 2Measurably, yes, but only by a sliver: on nights with the dog in the room, people's resting heart rate was about one beat per minute lower, and they woke up a fraction more often. That comes from more than one and a half million nights of wearable data, and the authors themselves call every difference very small.

Host 1Welcome to the Hastrup Review of Science, where artificial intelligence reads real science, so you don't have to pretend you understood the paper.

Host 2Today it is one paper, from the journal Nature and Science of Sleep, with the title Who's Been Sleeping in My Bedroom. It compares nights with a dog, a cat or a child in the room to nights alone, in the same people.

Why the bedroom matters

Host 1Before we get to the dog, why would sleep scientists care who else is in the bedroom at all?

Host 2Because a lot of people don't sleep alone, and a lot of people sleep badly. The paper cites that roughly a third of adults in Western countries sleep under seven hours, and somewhere between thirty and fifty percent of adults share a bed with a pet, a child, or both.

Host 1So it's a very ordinary habit, and nobody has really checked what it does to the body?

Host 2Checked a little, but mostly by asking people how they slept. The wider picture is that sleep research grew up in laboratories, one person, wired up, alone in a quiet room, which is exactly the situation almost nobody sleeps in at home.

Host 1And the question of children in the parents' bed, hasn't that been argued about for a long time?

Host 2For decades, yes, though mostly about the baby's safety and development rather than the parent's heart rate. This paper sets that debate aside and only looks at the adult who wears the tracker.

What earlier studies found

Host 1What did the earlier studies, the ones before this paper, actually find about pets?

Host 2A mixed picture. One study with movement sensors found that when a dog moved in the night, the woman sharing the bed moved more too, and another found poorer sleep when the dog was on the bed than when it was just in the room.

Host 1And cats, are they any different from dogs as bedmates?

Host 2Biologically they keep different hours. The paper notes that dogs sleep in several bouts and wake a few times a night, while cats are most active around dusk and dawn, which is when people are still asleep, and the little evidence there is suggests owners of both report similar disruption.

Host 1Is there any reason to think a pet could make sleep better rather than worse?

Host 2There is a plausible one. A companion can make people feel safe and calm, and calm is linked to a more relaxed nervous system, so the authors say it could just as well lower heart rate as raise it.

Host 1And a child in the room, I'd guess that's mostly about being woken up?

Host 2Mostly, yes, with a twist. Being close makes night-time care easier, but it also means more wake-ups, and since mothers tend to do more of the night shift, the effect might differ between women and men.

The open question

Host 1So what was the gap this paper set out to fill?

Host 2Three weaknesses in the older work: small samples, sleep judged by the sleepers themselves, and comparisons between different people. Comparing co-sleepers with people who sleep alone is risky, because those two groups may differ in age, habits, jobs and much else.

Host 1Why does that comparison between different people cause so much trouble?

Host 2Because you can't tell whether the dog explains the difference or the kind of person who owns a dog does. If dog people happen to be more active, say, any gap in heart rate might just be their fitness talking.

Host 1So the cleaner question is the same person, on nights with the companion, compared with nights without?

Host 2Exactly, that's what they mean by a within-person comparison. Each person acts as their own reference, which strips out everything about them that stays the same from night to night.

How the study worked

Host 1Where did the data come from, and who were these people?

Host 2From users of a commercial fitness wrist strap called WHOOP, aged eighteen to seventy-nine, using data from the year twenty twenty-four. Two of the three authors work for the company, which funded the study, and I'll come back to that.

Host 1How did the strap know whether a dog was in the room?

Host 2It didn't, the people told it. Each morning the companion app asked yes or no questions about the night before, such as had a dog in the room while sleeping, had a cat in the room, and slept with a child in your bedroom.

Host 1So if I just didn't answer one morning, what happened to that night?

Host 2It was left out of the main analysis, and that happened a lot. About thirty-one percent of nights were unanswered in the dog group, thirty-four in the cat group, and fifty-five percent in the child group.

Host 1Who got into the study, was it anyone who ever ticked yes?

Host 2No, they set rules. A person needed at least fourteen days of complete data, and had to report the companion present on at least three nights and absent on at least three, otherwise there's nothing to compare.

Host 1What about someone with a dog and a toddler both in the room?

Host 2They were removed. Anyone who reported more than one kind of companion was dropped, so each group shows one effect on its own.

Host 1How big was the pool before they started trimming it?

Host 2Sixty-one thousand three hundred and forty-seven people, almost forty-five thousand of them with dogs, about twelve thousand seven hundred with cats, and three thousand nine hundred and eleven with children. Together that was nearly ten million days of data.

Host 1Those groups are wildly uneven, so how did they make them comparable?

Host 2By matching. The child group was smallest, so for each parent they found the most similar dog owner and the most similar cat owner, on age, sex, body mass index, and each person's usual heart rate, heart rate variability, sleep and activity.

Host 1So it's a bit like a lookalike contest, where every parent gets a dog twin and a cat twin?

Host 2That's a fair picture, with a statistical yardstick deciding how alike counts as alike. It left three thousand nine hundred and eleven people in each group, eleven thousand seven hundred and thirty-three in all, and threw out over forty thousand unmatched dog owners.

Host 1Did the twins end up really alike?

Host 2Mostly. Ages were within two years, heart rates within a third of a beat, and about thirty-six percent of every group was female, but the parents were clearly less active than the pet owners, a gap matching couldn't close.

Host 1What did the strap actually measure overnight?

Host 2It uses a motion sensor and a light sensor that reads the pulse through the skin. From those it estimates resting heart rate, heart rate variability, breathing rate, how long you slept, how often you were disturbed, and which sleep stage each thirty-second slice of the night was in.

Host 1Heart rate variability comes up a lot, what is it in plain words?

Host 2It's how much the gap between heartbeats wobbles from one beat to the next, measured in milliseconds. Higher variability at rest is usually read as the calming branch of the nervous system being in charge, and poor sleep tends to go with lower variability and a higher resting heart rate.

Host 1And what do they mean by restorative sleep?

Host 2Deep sleep plus dream sleep, the stage called rapid eye movement sleep, added together and given as a share of the night. The paper also looked at next-day exercise, as a score that weights minutes by how hard the heart was working.

Host 1How trustworthy is a wrist strap compared with a proper sleep lab?

Host 2Good for heart rate, weaker for sleep stages. Earlier validation work cited here found ninety-nine percent agreement with a medical heart recording, and eighty-six to eighty-nine percent agreement with a sleep lab on the simple question of asleep or awake, and the authors say plainly that the finer staging is less accurate.

Host 1Once they had all these nights, how did they actually compare them?

Host 2With a statistical model that gives every person their own starting level, then asks how far that person's nights shift when the companion is there. It also accounts for age, sex, body mass index, season, and whether it was a weekday or a weekend.

Host 1Did they try other versions of the model, in case the first one was fooling them?

Host 2Several. One added bedtime and sleep timing, another added the previous day's exercise and the previous night's sleep, and they also allowed for the fact that one night tends to resemble the next.

Host 1With that many comparisons, wouldn't some come out significant just by luck?

Host 2They corrected for that with a standard method that controls the share of false alarms across all the outcomes. And because so many results were tested, they also reported effect sizes, which say how big a difference is, not just whether it's there.

A word from our sponsor

Host 1This episode is supported by the Yesno pencil, the pencil that can only write yes or no. The following terms and conditions apply.

Host 2By picking up the pencil you agree that it was in the room while you slept. You may skip a question. The pencil will not.

Host 1Questions left blank will be imputed five times and pooled across five imaginary versions of you.

Host 2I have used mine for eleven months. It now answers questions nobody asked, mostly about the dog.

Host 1The pencil must be present on at least three nights and absent on at least three. On absent nights it continues writing from inside the drawer, and those answers are final.

Host 2From the fourth week the pencil also records whether you were in the room. So far it says no.

The results

Host 1So on a night with the dog in the room, what happens to my heart?

Host 2It runs slightly slower, about one point one beats per minute lower for dog nights, zero point nine for cat nights and zero point eight for child nights. Heart rate variability went up by about one point four to two milliseconds in all three groups.

Host 1Is one beat a minute a lot, when my resting rate is around sixty?

Host 2It's small. These people averaged about sixty-one beats, so it's under a two percent shift, and the authors note there's no agreed threshold for when a night-to-night change like this matters clinically.

Host 1And breathing, did that change as well?

Host 2Slightly, by a few hundredths of a breath per minute slower. That's tiny, it only shows up because there are so many nights.

Host 1What about how long people slept, did the dog cost them sleep?

Host 2The opposite, on the face of it. Dog nights had about nine and a half extra minutes of sleep, cat nights about seven, child nights about five and a half, against a typical night of roughly seven hours.

Host 1Then why did you say at the start that people woke more?

Host 2Because both things happened at once. Disturbances rose by about zero point four per night in every group, a bit under half an extra disturbance, and the share of restorative sleep dropped by a third of a percentage point with a dog, a quarter with a cat, and over one full point with a child.

Host 1So if I have this right, people slept a bit longer and their hearts were calmer, but the sleep was slightly more broken up?

Host 2That's the pattern, yes, with the important footnote that it's an association, not proof the companion caused any of it. The authors offer no firm explanation for longer but choppier sleep.

Host 1Could something else about those nights explain it, like simply being at home?

Host 2That's the kind of thing the design can't rule out, and the authors list unmeasured factors that change from night to night as a limitation. Think of a night in a hotel without the dog, or a night the child stayed with grandparents: those no nights may differ in many ways besides the companion.

Host 1And the next day, did people exercise differently?

Host 2After a pet night they did a bit more, about seven and a half points on that weighted exercise score for both dogs and cats. After a child night it went the other way, about two and a half points lower.

Dogs, cats and children

Host 1Does that mean the dog gets people out of the door in the morning?

Host 2Perhaps for dogs, and other research does link dog ownership to more activity, but cats rose by almost exactly the same amount, which is harder to put down to walks. For children, the authors guess at tiredness, or simply no free time for an early workout, and they label that as speculation.

Host 1Of the three, which companion disrupted sleep the most?

Host 2The child, fairly clearly. Child nights had the smallest gain in sleep, the largest rise in disturbances per hour, and a drop in restorative sleep roughly three to five times bigger than with pets.

Host 1And did the extra sleep with a child survive those stricter models you mentioned?

Host 2No, that one fell apart. Once bedtime, the previous day's exercise and the previous night's sleep were accounted for, the gain shrank to under half a minute and was no longer distinguishable from zero, while everything else held its direction.

Men and women

Host 1Did men and women respond differently, the way the night-shift idea would predict?

Host 2In several places, yes. With a child in the room, in the strictest model, women slept about three minutes less and men about two minutes more, so it pointed in opposite directions.

Host 1So that fits the idea that mothers do more of the getting up?

Host 2It's consistent with it, and the authors cite research that mothers are the main night-time caregivers, but the journal doesn't record who got up. And they stress that the sex differences, like everything else, are very small.

Host 1What about with pets, any difference between men and women there?

Host 2Some. With dogs, men tended to show bigger drops in heart rate and smaller increases in disruption than women, and with cats the differences were mostly in disturbances and breathing.

How big is very small

Host 1You keep saying very small, but how do they put a number on small?

Host 2With a standardised effect size, which converts every result onto one common scale. By the usual convention, zero point one is small, zero point three medium and zero point five large.

Host 1And where did these results land on that scale?

Host 2Every single one fell below small, between zero point zero zero eight and zero point zero four five. The largest were dogs and resting heart rate, dogs and sleep length, and children and restorative sleep.

Host 1So how can something be highly significant and still barely matter?

Host 2Because with one and a half million nights, the statistics can detect differences far smaller than anyone could feel. Significant here means probably not zero, it doesn't mean big.

What it means for you

Host 1If I'm a dog owner lying awake wondering, what should I take from this?

Host 2That the dog is very unlikely to be the deciding factor in how well you sleep or how your heart behaves. The authors conclude that choices about co-sleeping can rest on comfort, caregiving and family routine rather than fear of a large effect.

Host 1And for a parent with a small child in the room, is the message the same?

Host 2Broadly, though the child signal is a little more disruptive, especially for women. It's still small, and this study doesn't speak to the child's safety or sleep at all, only the adult's wrist.

Host 1What's the general lesson here, beyond pets?

Host 2That huge datasets make small effects look important, so the size matters more than the stars next to the number. Here the authors did the honest thing and led with how small their effects were.

The caveats

Host 1What are the biggest reasons to be careful with these findings?

Host 2The first is that the question only asked whether the companion was in the bedroom, not in the bed. A dog on the floor and a dog on your pillow count the same, and the authors say this likely understates what actually sharing a bed does.

Host 1Does the app know how many dogs, or how old the child is?

Host 2No. Number of companions, size or breed of pet, age of the child, size of the bed, none of it was recorded, and neither was a human partner in the bed.

Host 1What about all those mornings nobody answered, could that tilt the results?

Host 2It could, and the gaps weren't random: younger people, short sleepers and those with higher heart rates skipped more often. The authors filled the gaps two different ways and the results held, but they admit they can't completely rule out a bias from it.

Host 1And the people themselves, are they a fair sample of everybody?

Host 2Not really. Users of this strap tend to be younger and more active than average, only about a third were women, and the study kept only people who sometimes sleep with the companion and sometimes don't, which leaves out the every-night co-sleepers.

Host 1And what about the company paying for the study, should that worry me?

Host 2It's worth knowing. The company funded it, two authors are employees and one holds stock options, though they state the business side had no role, and frankly the result is not a flattering sales pitch for anything.

Host 1Is there anything the study simply couldn't see?

Host 2Plenty: income, illness, medication, alcohol, caffeine, sleep disorders. The within-person design deals with things that stay the same about you, but not with things that change from night to night, and the data and code aren't public, only available on request.

What to watch next

Host 1What would a better follow-up study need to do?

Host 2Above all, separate bed from room. A journal question that asks where the companion slept, how many there were, and the child's age would let the next study test whether actually sharing the bed matters more.

Host 1Could anyone measure the animal side of it too?

Host 2That's where the field seems to be heading, more broadly: putting a movement sensor on the dog as well as the owner, so you can see whether the animal's restlessness lines up with the person's wake-ups. A study that did that at scale would get much closer to cause and effect.

Sign-off

Host 1So, after all of this, is the dog allowed to stay?

Host 2On this evidence, the dog's effect on your heart and your sleep is roughly the size of a rounding error, so that's between you and the dog. That was the Hastrup Review of Science.

Host 1And every weekend this same feed carries the roundup, the week's new papers in one short episode, which no dog has ever interrupted.