Dog Owners Showed a 24% Lower Risk of Death in a 3.8-Million-Person Analysis. The Dog May Not Deserve the Credit.

The Story Everyone Knows

Ask a dog owner why their dog is good for them and you will hear about love, not epidemiology: dogs are family, not medicine, and prescribing one sounds like a joke a cardiologist tells after bad news. The conventional wisdom runs the other way: pets cost money, wake you at dawn, and tie you to the house.

The American Heart Association said as much in 2013: pet ownership was probably associated with lower cardiovascular risk, but nobody had shown the pet caused it, and no guideline treats a dog as a statin.

Then a University of Toronto and Mount Sinai Hospital team did something new: they gathered every prospective study tracking dog owners and non-owners until death and pooled them into a single number, which made headlines worldwide while the journal spent the next year hosting arguments about it.

The Study

Caroline Kramer and colleagues searched Embase and PubMed for studies published before May 24, 2019, keeping only prospective studies that recorded dog ownership at baseline and followed participants until death; the analysis was pre-registered (PROSPERO CRD42018111048). Ten studies survived the screen, covering 3,837,005 participants and 530,515 deaths over a mean follow-up of 10.1 years.

Dog owners died at lower rates: compared with non-owners, a 24 percent lower risk of death from any cause (relative risk 0.76, 95% CI 0.67 to 0.86) and a 31 percent lower risk of cardiovascular death (relative risk 0.69, 95% CI 0.67 to 0.71), with the cardiovascular estimate barely varying across studies (I-squared 5.1 percent).

The most dramatic number came from the sickest people: among participants with a prior coronary event, living with a dog was associated with a 65 percent lower risk of death (RR 0.35, 95% CI 0.17 to 0.69). A companion study the same week followed 182,000 Swedish heart attack survivors and 155,000 stroke survivors: dog owners living alone had a 33 percent lower risk of death after a heart attack, versus 15 percent for those living with a partner or child.

The Swedish registry study behind those numbers followed 3,432,153 Swedes aged 40 to 80 for up to 12 years: single-person dog owners died at a rate 33 percent lower than single non-owners (HR 0.67), versus 11 percent lower for multi-person households (HR 0.89).

The Arithmetic Nobody Did

Here is the comparison the headlines skipped: the Swedish single-household benefit is three times the multi-person benefit, 33 percent versus 11 percent, because the dog matters most where no other human provides the walking, the schedule, and the company, which is also where loneliness does its worst damage: a meta-analysis of 308,849 people previously found social isolation deadlier than obesity. If dogs were simply magical, household composition would not change the effect.

Apply the same lens to the meta-analysis itself: the all-cause estimate of 0.76 averages disagreeing studies, while the cardiovascular estimate of 0.69 barely varied, so a reader who quotes only the 24 percent figure is quoting the shakier headline number.

And the 65 percent figure deserves its own warning label: its confidence interval runs from 0.17 to 0.69, so the data are compatible with anything from an 83 percent reduction to a 31 percent reduction.

The Strongest Counterargument

Now the case for throwing the whole thing out, stated at full strength. The meta-analysis pooled unadjusted estimates: the analyses did not account for confounders such as fitness or a healthier lifestyle, and dog owners differ from non-owners in almost every way that predicts survival, tending to be younger, wealthier, more active, and less disabled, so that in observational data a 24 percent gap can be manufactured entirely by who selects into dog ownership. No randomization means no causal claim.

The counterevidence is not theoretical: a Norwegian cohort of 25,031 dog owners followed for a median of 18.5 years found a hazard ratio of exactly 1.00 (95% CI 0.91 to 1.09), with dog owners dying at the same rate as everyone else and no more physically active. Three older American survey analyses likewise found no effect on all-cause mortality, with hazard ratios of 1.00, 1.17, and 0.82, each statistically indistinguishable from nothing.

The journal itself hosted the fight. In 2020, the journal published two letters challenging the methods, a response from Kramer and colleagues, and a Bauman perspective urging caution about meta-analyses of observational studies. A separate Swedish analysis found no large reduction in blood pressure or cholesterol among dog owners, removing the most obvious biological pathway, and inside the Swedish Twin Register, where twins share genes and early environment, the mortality benefit vanished.

Put bluntly: the largest studies say dog owners live longer, the best-controlled comparisons say maybe not, and the mechanism studies say the obvious pathways do not explain it.

What We Didn't Prove

This evidence does not prove that getting a dog extends your life, and it does not prove that walking is the mechanism: in the Norwegian cohort, dog owners walked no more than anyone else, and the Swedish risk-factor analysis found no large blood pressure or cholesterol advantage. It does not prove the effect holds outside the pooled populations, and since ownership was measured at one point in time, a dog acquired after a health scare or given up during decline could distort the picture either way.

It also does not weigh the costs: dog bites, falls over leashes, zoonotic infections, allergies, and veterinary bills fall hardest on the same older and sicker people the mortality numbers celebrate, and none of the pooled studies subtracted them.

The Bottom Line

So what survives contact with the critics? A large, consistent cardiovascular signal in the biggest pooled analysis ever run, concentrated in people who live alone and people recovering from heart attacks, alongside clean null results in Norway and America. The pattern fits a modest story: for some people, a dog supplies the daily walk, the fixed routine, and the reason to get out of the house that they would otherwise lack, which is not a longevity drug, and the Heart Association was right to warn against adopting a pet to reduce cardiovascular risk.

The honest headline is less shareable and more useful: dog owners tend to outlive non-owners, the gap is biggest where companionship is scarcest, and nobody has shown the dog is the reason.

What You Can Do

Walk a dog, even if it is not yours. Volunteer at a shelter, borrow a neighbor's dog, or join a dog-walking group, and you get the routine and the movement without the 15-year commitment.

If you already own a dog, make the walk the point. A fixed morning route, brisk enough to raise your breathing, beats an occasional weekend ramble.

Do not adopt a dog as a health intervention. A decade-plus commitment with real costs, the mortality evidence cannot promise you a single extra day, so decide on lifestyle fit and treat any health benefit as a bonus.

If you live alone, treat companionship as a health variable. The strongest thread here is not dogs but isolation, and regular social contact, a walking group, or a neighbor who expects you at a set time supply the same routine, without a leash.

If you have had a heart attack or stroke, talk to your clinician about activity first. The post-event numbers are the most dramatic and the least precise, and returning to regular walking needs a clinician's sign-off, not a puppy.