Sauna and Mortality: A Beautiful Correlation With a Confounding Problem

The full story

A hazard ratio of 0.37 for sudden cardiac death is the kind of number that launches a product category. That is what a 2015 cohort analysis in JAMA Internal Medicine reported for Finnish men taking four to seven sauna sessions a week compared with one, across 2,315 middle-aged men followed a median 20.7 years, with the same inverse trend appearing for fatal coronary heart disease, fatal cardiovascular disease and death from any cause.

It is a lovely result. It is also observational, it is in men only, and the group that generated the headline number contained 201 of those 2,315 men, among whom 10 sudden cardiac deaths occurred. Everything difficult about sauna research lives in the gap between those two paragraphs.

What the Finnish cohort found

The 2015 paper reported both a frequency gradient and a duration gradient: sessions longer than 19 minutes carried a hazard ratio of 0.48 for sudden cardiac death compared with sessions under 11 minutes.

A 2017 analysis in Age and Ageing followed the same men over the same period, capturing 204 cases of dementia and 123 of Alzheimer's disease. Four to seven sessions a week were associated with hazard ratios of 0.34 for dementia and 0.35 for Alzheimer's, after adjustment for age, alcohol, body mass index, systolic blood pressure, smoking, diabetes, prior heart attack, resting heart rate and LDL cholesterol.

That is a long adjustment list, which is both reassuring and the tell. The upper bound of the Alzheimer's confidence interval reaches 0.90, close enough to 1 to be compatible with a fairly small effect. And adjusting for nine variables does not adjust for the tenth.

The confounding problem, quantified

The most useful paper in this literature is the one nobody selling a sauna quotes. A 2018 analysis in Annals of Medicine took 2,277 men from the same cohort, followed them a median 26.1 years through 520 cardiovascular and 1,124 total deaths, and put cardiorespiratory fitness and sauna frequency into the same model.

Fitness won, and not narrowly. High versus low fitness carried a cardiovascular mortality hazard ratio of 0.51. High versus low sauna frequency carried 0.74. The joint analysis is the part that should change how you read the headline: men with high fitness and low sauna use still had a hazard ratio of 0.50, while men with low fitness who used the sauna frequently managed only 0.72. In a population where sauna use is routine, the men going four to seven times a week are also, on average, the men who are well enough to.

A 2023 analysis in the Journal of Nutrition, Health and Aging makes the same point differently. Across 2,575 men with 744 cardiovascular deaths over a median 27.8 years, frequent sauna use did not abolish the excess risk carried by a systolic blood pressure of 140 mmHg or more: high pressure plus frequent sauna use still carried a hazard ratio of 1.52 against normal pressure with frequent use, versus 1.81 for high pressure with infrequent use. Heat is not an eraser.

What happens when you actually apply the heat

The mechanistic work is more convincing than the cohort work is decisive. A 2018 study in the Journal of Human Hypertension put 102 participants with a mean age of 51.9 and at least one cardiovascular risk factor through a single 30-minute session at 73 degrees C and 10 to 20 percent humidity. Mean carotid-femoral pulse wave velocity fell from 9.8 to 8.6 metres per second, systolic pressure from 137 to 130 mmHg and diastolic from 82 to 75 mmHg, and systolic pressure was still below its pre-sauna level after 30 minutes of recovery. Something real is happening to the vasculature.

The randomised evidence exists, and it is small. A 2022 multi-arm randomised trial in the American Journal of Physiology assigned 47 low-activity adults with a mean age of 49 to eight weeks of exercise plus 15 minutes of post-exercise sauna, exercise alone, or control. That is roughly 16 people per arm. Adding sauna to exercise lowered systolic pressure by a further 8.0 mmHg and improved cardiorespiratory fitness by 2.7 mL/kg/min and total cholesterol, relative to exercise alone. Both of those confidence intervals come close to zero at one end.

The strongest controlled heat-therapy dataset is a 2016 trial in the Journal of Physiology, and it has ten people per arm. Eight weeks of hot-water immersion, meaning a 40.5 degrees C bath with rectal temperature held at or above 38.5 degrees C for 60 minutes, four to five times a week, raised flow-mediated dilatation from 5.6 to 10.9 percent, reduced aortic pulse wave velocity from 7.1 to 6.1 metres per second, reduced carotid intima-media thickness from 0.43 to 0.37 mm and lowered mean arterial pressure from 83 to 78 mmHg. Nothing happened in the ten-person thermoneutral sham group. A 2018 review in Mayo Clinic Proceedings, by the cohort's own investigators, lists the postulated mechanisms, from endothelium-dependent dilatation to autonomic modulation and lower blood pressure, and explicitly flags what remains uncertain.

What the evidence does not show

There is no randomised trial with mortality as an endpoint, and there is not going to be one soon. Two decades of adherence to a bathing schedule is not a randomisable intervention, which means 0.37 will remain a correlation indefinitely.

It does not show the finding generalises beyond middle-aged Finnish men, because that is who was studied. It does not show that sauna substitutes for training; the fitness analysis argues the opposite. And it does not show that all heat is the same heat. A 2022 randomised crossover trial in Complementary Therapies in Medicine put 10 healthy women with a mean age of 36 through 45 minutes of rest, infrared sauna or indoor cycling. Infrared raised tympanic temperature more than exercise did, by 0.79 degrees C, and produced no significant difference in blood pressure, arterial stiffness or heart rate variability versus either comparison condition.

The authors' conclusion: infrared sauna effects are underpinned by thermoregulatory responses rather than exercise-mimetic cardiovascular activation.

In other words, you got hot. That is not the same as getting the cardiovascular stimulus.

The dose is temperature, humidity and time

The hardware difference is measurable, not philosophical. A 2015 study in SpringerPlus put 10 physically active men through 30 minutes of far-infrared sauna at 35 to 50 degrees C and 25 to 35 percent humidity and found no differences in any recovery variable after a hypertrophic strength session; after maximal endurance work, only countermovement jump height differed, 0.34 against 0.32 metres. The telling number is elsewhere in the same paper: post-sauna heart rate was 92 beats per minute after a traditional Finnish sauna and 71 after far-infrared. Same duration, very different physiological load.

If you are choosing hardware, the parameters worth caring about are the ones the trials actually reported:

• The acute vascular study ran 30 minutes at 73 degrees C with 10 to 20 percent humidity.
• The cohort signal sat at four to seven sessions a week, with sessions over 19 minutes outperforming sessions under 11.
• The randomised add-on was 15 minutes of sauna immediately after each exercise session, for eight weeks.
• Far-infrared cabins running at 35 to 50 degrees C impose a substantially lighter load than a traditional sauna does.

A blanket or a low-temperature infrared cabin is not a cheaper Finnish sauna. It is a different exposure, and the one trial that tested infrared against exercise found nothing cardiovascular.

None of this makes the sauna a bad idea. It is pleasant, the acute blood-pressure effects are reproducible, and the small randomised trial suggests it adds something on top of exercise rather than standing in for it. What it makes is a claim with an honest ceiling. The number that sold you on it came from men who were also fitter, and the fitness analysis in that same cohort suggests fitness is where most of the survival advantage was hiding. Buy the sauna because you will use it and it feels good, and keep doing the thing that actually carried the hazard ratio.