Testosterone and Blood Clots: The Case-Crossover Study That Found a Signal

Walker and colleagues studied 39,622 men who had a venous thromboembolism and compared each man's testosterone exposure in the six months before the clot against his own exposure in the six months before that. Testosterone use was associated with roughly double the short-term odds of VTE — odds ratio 2.32 (95% CI, 1.97–2.74) in men with evidence of hypogonadism and 2.02 (95% CI, 1.47–2.77) in men without.[1]

Most of the testosterone safety literature is about the heart. This one is about the veins, and it is the reason clot risk sits on the label.

The design controls for the man himself

A case-crossover study uses each participant as his own control. The comparison is not between treated and untreated men — it is between the same man during a period when he was exposed and a period when he was not.

That structure removes, by construction, every stable characteristic of the individual: genetics, baseline clotting tendency, body weight, smoking history, chronic conditions. Confounders that plague cohort studies of testosterone simply cannot operate here, because they do not change between a man's case period and his own control period.

What it cannot remove is anything that changed over those months for reasons connected to both the prescription and the clot.

What it did

• Data: IBM MarketScan Commercial Claims and Encounters and the Medicare Supplemental Database, 1 January 2011 to 31 December 2017.
• Population: 39,622 men (mean age 57.4) with a first VTE event, cancer-free at baseline, with 12 months of continuous enrolment before the event. 3,110 (7.8%) had evidence of hypogonadism.
• Case periods: 6 months, 3 months and 1 month before the VTE, each matched to an equivalent control period in the 6 months preceding the case period.
• Exposure: billed testosterone prescriptions identified by national drug codes.
• Outcome: first VTE event, stratified by presence or absence of hypogonadism.

What it found

In age-adjusted models, testosterone use across all case periods was associated with higher VTE risk in men with hypogonadism (OR 2.32; 95% CI, 1.97–2.74) and without hypogonadism (OR 2.02; 95% CI, 1.47–2.77).[1]

The finding that the association held in men without a hypogonadism diagnosis is the one with the most practical weight. Testosterone prescribing has expanded well beyond classical hypogonadism, and this suggests the clot signal travels with the drug rather than with the diagnosis.

Among men without hypogonadism, the point estimate in the 3-month window was higher for men under 65 (OR 2.99; 95% CI, 1.91–4.68) than for older men (OR 1.68; 95% CI, 0.90–3.14), though the interaction was not statistically significant (P = .14). The authors describe this as some evidence the association is more pronounced in younger men — appropriately hedged, and it should stay hedged.

Relative versus absolute

Doubling sounds alarming, and relative risk without a baseline is how alarm gets manufactured. VTE is uncommon in middle-aged men; doubling an uncommon event still leaves an uncommon event. This study, by design, cannot supply the absolute rate — it enrolled only men who already had a clot.

So the correct reading is directional: exposure is associated with a meaningfully elevated short-term risk, in a population where the underlying risk is low to begin with. That is enough to matter for a man with additional clot risk factors and not enough to contraindicate therapy generally.

The plausible mechanism

Testosterone raises haematocrit, and higher haematocrit raises blood viscosity. The authors name this among the possible mechanisms. It is the same physiology that makes polycythemia the most common side effect of therapy and that T4DM measured directly, with 22% of treated men crossing a 54% haematocrit trigger.

A mechanism that is coherent across three independent lines of evidence is harder to dismiss than an isolated statistical association.

What TRAVERSE adds

The randomized evidence points the same direction at a smaller magnitude. TRAVERSE reported pulmonary embolism numerically higher in the testosterone group (0.9% versus 0.5%) — not a primary endpoint, not powered for it, but consistent.

Two very different designs, one claims-based and one randomized, both finding more clotting on treatment, is the shape of a real effect rather than an artefact of either method.

What follows clinically

• A personal or family history of VTE is a genuine consideration, not a formality, when weighing therapy.
• Haematocrit monitoring is the lever available. It is measurable, it is on every standard panel, and it responds to dose reduction or dose interruption.
• The first months may carry the most risk. The design detected the association in windows as short as one month.
• "Not hypogonadal" is not protective. The association was present in men without the diagnosis.

Bottom line

The strongest available evidence on testosterone and venous clots is a case-crossover study of 39,622 men that found roughly doubled short-term odds of VTE on treatment, in men with and without hypogonadism. The design eliminates fixed personal confounders, the mechanism is plausible and corroborated, and the randomized data lean the same way. Against a low absolute baseline risk, this is a reason for real monitoring and honest counselling — not a reason to avoid therapy in men who need it.

References

1 source

1. Walker RF, Zakai NA, MacLehose RF, et al. Association of Testosterone Therapy With Risk of Venous Thromboembolism Among Men With and Without Hypogonadism. JAMA Intern Med. 2020;180(2):190–197. doi:10.1001/jamainternmed.2019.5135 · PMID 31710339