TRAVERSE and Bone: Testosterone Did Not Reduce Fractures

Testosterone treatment reliably increases bone mineral density, so the reasonable expectation was that it would reduce fractures. The TRAVERSE bone sub-study tested that directly in 5,204 men and found the opposite direction: a clinical fracture occurred in 91 men (3.50%) on testosterone versus 64 (2.46%) on placebo — hazard ratio 1.43, 95% CI 1.04 to 1.97.[1] The authors' conclusion is unambiguous: testosterone treatment did not result in a lower incidence of clinical fracture than placebo.

This is the single most misreported result in the TRAVERSE programme, and it is worth stating plainly before anything else.

Why anyone expected the opposite

Bone mineral density is a well-established response to testosterone. Trials going back decades show that treating hypogonadal men raises BMD at the spine and hip, and BMD is the standard surrogate marker for fracture risk across osteoporosis medicine. The inference — better density, fewer breaks — is the one nearly everyone made.

The problem is that a surrogate marker is not an outcome. Density is a measurement; a fracture is an event. The sub-study exists precisely because the field needed a trial large enough and long enough to measure events rather than infer them from the surrogate.

What the sub-study did

The fracture analysis was nested inside the main TRAVERSE cardiovascular trial, using the same randomized population.

• Population: 5,204 men in the full-analysis set — 2,601 assigned testosterone, 2,603 placebo.
• Eligibility: ages 45–80, one or more symptoms of hypogonadism, and two fasting morning testosterone concentrations below 300 ng/dL drawn at least 48 hours apart, plus pre-existing cardiovascular disease or high risk of it.
• Intervention: daily testosterone gel or matching placebo gel.
• Ascertainment: at every visit participants were asked whether they had fractured since the previous visit. Where they had, medical records were obtained and adjudicated — the events were verified, not self-reported.
• Median follow-up: 3.19 years.
• Analysis: time-to-event.

Adjudication matters here. A softer design could be waved away as reporting noise; this one cannot.

What it found

Clinical fracture: 91 of 2,601 (3.50%) in the testosterone group versus 64 of 2,603 (2.46%) in the placebo group. Hazard ratio 1.43 (95% CI, 1.04–1.97).[1]

Two details decide how much weight this carries.

First, the confidence interval excludes 1.0. The lower bound is 1.04. This is not a numerical wobble around no-effect; on its own terms it is a statistically significant difference in the direction of more fractures on treatment.

Second, the paper reports that fracture incidence also appeared higher in the testosterone group for all other fracture endpoints — not one endpoint out of many, which is the pattern you would expect from chance alone.

What it does not establish

Being accurate about an unwelcome finding also means not overstating it.

1. This was a pre-specified sub-study of a cardiovascular safety trial, not a dedicated fracture trial. It was powered for MACE. Fracture was not the endpoint the sample size was built around.
2. The population is specific. These were middle-aged and older men with hypogonadism and cardiovascular disease or elevated cardiovascular risk. They are not a general osteoporosis population, and they are not young men.
3. The mechanism is unexplained. Nothing in the trial identifies why fractures ran higher. Proposed explanations — more physical activity on treatment, more falls, changes in bone microarchitecture not captured by density — are hypotheses, not findings.
4. It does not overturn the density data. Testosterone still raises BMD. What the trial shows is that the density gain did not translate into fewer fractures in this population over roughly three years.

Why it gets reported backwards

The result is counterintuitive, and the summary sentence is easy to invert if read quickly. "Testosterone treatment did not result in a lower incidence of clinical fracture" is a negative finding phrased negatively — one careless paraphrase turns it into a positive claim about bone protection.

The stakes of that error are not academic. A man deciding whether to start therapy, or a clinician weighing it against a bisphosphonate, needs the direction of the effect stated correctly. Presented as a bone benefit, this finding recommends exactly the wrong thing.

Keen's position is to state it as the trial reported it. The same finding is set out in the starting TRT after 50 guide and in where testosterone therapy actually stands now.

Bottom line

TRAVERSE tested whether testosterone reduces fractures and found it did not. The point estimate ran the other way, the confidence interval excluded no-effect, and the direction held across every fracture endpoint. Testosterone should not be prescribed for fracture prevention, and anyone presenting this trial as evidence of bone protection has the result inverted.

If bone is the clinical concern, that is a conversation about bone-specific therapy and about what the labs actually need to cover — not a reason to start or continue testosterone.

References

1 source

1. Snyder PJ, Bauer DC, Ellenberg SS, et al. Testosterone Treatment and Fractures in Men with Hypogonadism. N Engl J Med. 2024;390(3):203–211. doi:10.1056/NEJMoa2308836 · PMID 38231621