September 2, 2026 · Science · Men's health

What Was Published

Men at both ends of the testosterone range show higher rates of atrial fibrillation. That is the argument of a review in the Journal of the Endocrine Society from Anand Reddy Maligireddy at the University of Kansas Medical Center, with Rajat S. Barua and Shumedha T. Barua at the Kansas City Veterans Affairs Medical Center. The two ends reach that outcome by separate mechanisms, according to the review, which names a mid-normal target of roughly 350 to 550 ng/dL as the range carrying the lowest observed risk. As reported by MedicalXpress, the authors call the paper "a narrative summary of existing literature rather than a meta-analysis," and say on that basis its conclusions are interpretive.

A narrative review selects, weighs and interprets studies that already exist. It enrolls nobody, measures nothing, and produces no pooled estimate that another team can check. Every figure underneath this one was published somewhere else first, and the framing is the only new part.

The short version

The U-shape is a defensible reading of the observational data, and it is not a new finding. What makes this week worth a page is that a second narrative review, published in the World Journal of Men's Health in the same year, worked through much of the same literature and concluded that testosterone therapy at physiological levels either does nothing to atrial fibrillation risk or lowers it. The single randomized trial in the middle of the argument, TRAVERSE, found atrial fibrillation in 3.5% of men on testosterone against 2.4% on placebo over about 22 months. That gap is real and small in absolute terms, and two teams of reviewers have pulled it in opposite directions while reading the same shelf of studies. Anyone presenting the question as settled this week has picked one of the two reviews.

Where the U-Shape Comes From

The high-testosterone half of the curve rests substantially on a further analysis of the ASPREE study published in eClinicalMedicine in 2024. It followed 4,357 healthy older men, median age 73.7 years, and recorded 286 new cases of atrial fibrillation over a median of 4.4 years. Men in the fourth quintile of testosterone had a hazard ratio of 1.91 against the middle quintile, with a 95% confidence interval of 1.29 to 2.83. The fifth quintile came in at 1.98, interval 1.33 to 2.94. In absolute terms those two quintiles ran at 18.7 and 18.2 cases per 1,000 person-years.

Two features of that study decide how far its numbers travel. It measured endogenous testosterone, meaning the level each man's own body produced, and it was observational. A man whose natural testosterone sits in the top fifth of a healthy 73-year-old cohort differs from a man whose testosterone sits there because of a weekly injection, in ways no statistical adjustment fully captures. The hazard ratios describe the first group. They get quoted as though they describe the second.

The low-testosterone half draws on different cohorts, and its effect estimates swing sharply with age. The World Journal of Men's Health review cites work reporting a hazard ratio of 1.30 per one-standard-deviation decrease in testosterone among men aged 55 to 69, interval 1.07 to 1.59, rising to 3.53 in men aged 80 and over, interval 1.96 to 6.37. A relationship that moves that much across age bands is not a constant an individual reader can apply to themselves.

Two Reviews Read the Same Studies and Disagree

The two 2026 papers do not dispute the raw numbers. They disagree about which studies carry the weight.

The review in the World Journal of Men's Health, by Abdulmaged M. Traish, Karim Sultan Haider and Farid Saad, assembles the therapy-side literature and finds it pointing the other way. It cites a UK Biobank analysis of 179,988 men followed 11.8 years, in which 13,847 developed atrial fibrillation, reporting 15% lower risk among men with testosterone at or above 300 ng/dL. It cites an analysis of 133,584 men with hypogonadism where atrial fibrillation occurred in 3.6% of treated men against 4.0% of untreated, a relative risk of 0.900. It cites a study of 76,639 veterans in which normalizing testosterone was associated with fewer cases, and a meta-analysis finding testosterone therapy not significantly associated with atrial fibrillation in either direction.

That same review also sets out the studies cutting against its conclusion: a TriNetX analysis of 117,908 men with hypogonadism showing higher risk at five years, the ARIC cohort with 1,664 cases among 9,282 participants over 13.7 years, and a multi-ethnic cohort of 4,883 followed a median of 10.9 years. Its conclusion is that adequate therapy reaching physiological levels has no effect or a beneficial one, and that deficiency itself contributes to risk.

Both papers are narrative reviews. Neither pooled its inputs, so neither can be checked on the thing that separates a systematic review from an argument: whether the choice of studies was specified before the studies were read. Two teams reaching opposite conclusions from an overlapping literature, with no pooled estimate between them, tells a reader about the state of the field rather than about testosterone.

What This Changes for Someone Buying Testosterone Therapy Online

Several platforms benchmarked here prescribe testosterone inside a subscription built around a blood panel and a dashboard. Suppose the practical claim in the Endocrine Society review holds in outline, and the level reached matters more than the binary decision to treat. The questions a buyer should ask then shift toward things a panel does not display.

  • What trough level does the protocol target, and is it written down before you subscribe? A program titrating toward symptom relief and one titrating toward a stated mid-normal target will land patients in different places, and only the second can be audited against a number.
  • How often is a level rechecked once you are stable, and is that draw included or billed separately? The answer changes the annual cost as well as the monitoring.
  • Who reads the result? A value flagged red on a dashboard is not the same as a clinician deciding whether the dose should change. Our guide to doctor-led options sets out how the clinical models behind these subscriptions differ on exactly that point.

There is also a gap worth naming plainly. Atrial fibrillation is diagnosed on an electrocardiogram, and no testosterone panel contains one. A platform can report that your testosterone is 620 ng/dL. It cannot report whether your heart is in rhythm. Readers weighing that gap can compare devices in our wearables coverage, which sets out which consumer hardware records an FDA-cleared single-lead ECG and which only infers rhythm irregularity from a pulse signal.

Our guide to TRT side effects and risks covers the broader monitoring picture, including the haematocrit and PSA checks that belong in any protocol, and free versus total testosterone explains which number the ranges quoted above refer to. The men's health hub collects the rest.

What Would Change This Read

A randomized trial with a prespecified rhythm endpoint and two testosterone targets would change it most. The whole argument turns on whether the level reached matters, and no trial has randomized men to different target levels and then counted arrhythmias. Until one does, the U-shape stays an inference drawn from cohorts of men who were never assigned anything.

A properly systematic review with a registered protocol would also move it. The current disagreement exists partly because both 2026 papers were free to choose their inputs after seeing them. A pooled estimate with inclusion criteria set in advance would produce a single number that can be argued with on its merits, which neither review currently offers.

Related Coverage

A separate briefing covers the STAREE trial in adults over 70, another large study whose headline result and primary endpoint point in different directions. The men's health hub and the biomarker guides cover what the panels behind these subscriptions measure.

Sources

  • MedicalXpress, "Men with low or high testosterone may have a higher risk of irregular heartbeat, review indicates," September 1, 2026, medicalxpress.com (accessed September 2, 2026). Source of the Journal of the Endocrine Society review's existence and 2026 publication, its authors and their institutions, the quoted description of it as a narrative summary rather than a meta-analysis, the authors' note that its conclusions are interpretive, the mid-normal target of approximately 350 to 550 ng/dL, and the claim that low and high testosterone act through two different mechanisms.
  • "Testosterone and the risk of incident atrial fibrillation in older men: further analysis of the ASPREE study," eClinicalMedicine, 2024, PMC11067494 (accessed September 2, 2026). Source of the 4,357 men, the median age of 73.7 years, the 286 incident cases over a median 4.4 years, the quintile 4 hazard ratio of 1.91 (95% CI 1.29 to 2.83) and quintile 5 hazard ratio of 1.98 (95% CI 1.33 to 2.94) against quintile 3, the incidence rates of 18.7 and 18.2 per 1,000 person-years, and the TRAVERSE figures of 3.5% versus 2.4% with p = 0.02 in over 5,000 men with baseline testosterone below 10.4 nmol/L treated for about 22 months.
  • Traish AM, Haider KS, Saad F, "Does Testosterone Therapy Increase Risk of Atrial Fibrillation? Appraisal of the Contemporary Scientific and Clinical Evidence," World Journal of Men's Health, 2026, wjmh.org (accessed September 2, 2026). Source of this review's narrative (non-systematic) design and its conclusion, the TRAVERSE case counts of 91 versus 63, the UK Biobank figures (179,988 men, 11.8-year follow-up, 13,847 cases, 15% lower risk at testosterone at or above 300 ng/dL), the 133,584-man analysis reporting 3.6% versus 4.0% and a relative risk of 0.900, the 76,639-veteran normalization finding, the meta-analysis reporting no significant association, the TriNetX cohort of 117,908, the ARIC figures of 1,664 cases among 9,282 over 13.7 years, the multi-ethnic cohort of 4,883 over a median of 10.9 years, and the age-stratified hazard ratios of 1.30 (95% CI 1.07 to 1.59) at ages 55 to 69 and 3.53 (95% CI 1.96 to 6.37) at ages 80 and over.
  • The Journal of the Endocrine Society paper itself did not return a matching record on PubMed or a locatable publisher page at the time of writing. Every statement about that specific paper on this page is attributed to the MedicalXpress report above, and no figure is taken from the paper itself. The conversion of 10.4 nmol/L to approximately 300 ng/dL is a unit conversion of a sourced value, not a figure any source here reports.

Frequently Asked Questions

What did the new review claim?

A review published in the Journal of the Endocrine Society, from authors at the University of Kansas Medical Center and the Kansas City Veterans Affairs Medical Center, argues that both unusually low and unusually high testosterone are associated with a higher rate of atrial fibrillation, through separate biological mechanisms. It names a mid-normal target of roughly 350 to 550 ng/dL as the range with the lowest observed risk. The authors describe it as a narrative summary of the existing literature rather than a meta-analysis, so it pools no data and produces no new estimate.

Is this new evidence?

No. A narrative review selects and interprets studies that were already published. Nobody was measured, randomized or followed for this paper. The observations underneath it, including the ASPREE cohort analysis and the TRAVERSE trial, were reported in 2023 and 2024. The framing is new. The data are not.

Does a high testosterone level on a lab report mean a higher risk of atrial fibrillation?

That is not something the cohort data can establish. The clearest figures come from a further analysis of the ASPREE study in eClinicalMedicine, covering 4,357 healthy men with a median age of 73.7 years. Men in the top two quintiles of testosterone had roughly twice the rate of new atrial fibrillation as men in the middle quintile. That is an observational association in older men, measured on testosterone their own bodies produced. It does not show that testosterone causes the arrhythmia, and it is not a measurement of what happens when a prescribed dose puts someone at that level.

What did the one randomized trial find?

TRAVERSE randomized more than 5,000 middle-aged and older men with hypogonadism and cardiovascular risk to testosterone or placebo for about 22 months. Atrial fibrillation was reported in 91 men on testosterone against 63 on placebo, which is 3.5% versus 2.4%, with a reported p value of 0.02. The absolute difference is about 1.1 percentage points, in a population selected for existing cardiovascular risk. It remains the strongest single piece of evidence on the question, and atrial fibrillation was not its primary endpoint.

Do the two 2026 reviews agree?

No. A separate 2026 narrative review in the World Journal of Men's Health surveyed much of the same literature and concluded that testosterone therapy achieving physiological levels either has no effect on atrial fibrillation or reduces it. It cites a UK Biobank analysis of 179,988 men reporting 15% lower risk at normal testosterone, and a meta-analysis finding no statistically significant association in either direction. Two narrative reviews published in the same year reached different conclusions from an overlapping evidence base, which is a description of an unresolved question rather than an answer to it.

What can a longevity platform's blood panel tell you about this?

Your testosterone level, and nothing about your heart rhythm. Atrial fibrillation is an electrical diagnosis made on an electrocardiogram, which no testosterone panel includes. A subscription that reports a hormone number and a trend line is not monitoring for the outcome discussed here. If rhythm is a concern, that is a conversation with a clinician and a recording device rather than a marker on a panel.

What should someone currently on testosterone therapy do with this?

Nothing on this page is a basis for starting, stopping or changing a dose. What the material does support is asking the prescribing clinician two process questions: what trough level the protocol targets, and how often it is rechecked. New palpitations, an irregular pulse, unusual breathlessness or light-headedness warrant prompt clinical contact whatever any review concludes, because those symptoms get evaluated on their own merits.