A testosterone study answers a narrow question: what happened to a specific group of men, on a specific product, for the outcomes the researchers chose to measure. Who was enrolled and what was measured tell you how far a result reaches. TRAVERSE, for example, enrolled 5,246 men aged 45 to 80 with existing or high cardiovascular risk and tested a daily 1.62% testosterone gel, not an injection.
Start with who was enrolled
The men in a trial set the limits of its findings. Here are 3 large trials side by side.
| Question | What the source says |
|---|---|
| Who was in TRAVERSE? | 5,246 men aged 45 to 80 with existing or high risk of cardiovascular disease, symptoms of hypogonadism and 2 fasting testosterone levels below 300 ng/dL. Daily 1.62% testosterone gel or placebo gel. |
| Who was in the Testosterone Trials? | 790 men aged 65 or older with testosterone below 275 ng/dL and symptoms suggesting low testosterone. Testosterone gel or placebo gel for 1 year. |
| Who was in T4DM? | 1,007 men aged 50 to 74 with waist 95 cm or more, testosterone 14.0 nmol/L or lower, and impaired glucose tolerance or new type 2 diabetes, all in a lifestyle program. Testosterone undecanoate 1000 mg injections or placebo for 2 years. |
None of these 3 trials tested testosterone cypionate. A result in older men, men with heart risk or men with diabetes risk does not by itself show what happens in younger or healthier men.
Find the main question the trial was built to answer
Each trial names a primary endpoint. In TRAVERSE it was cardiovascular death, nonfatal heart attack or nonfatal stroke. It occurred in 7.0% on testosterone and 7.3% on placebo (hazard ratio 0.96, 95% CI 0.78 to 1.17), meeting noninferiority. Mean treatment was 21.7 months and mean follow-up 33.0 months.
Other findings still count. TRAVERSE observed a higher incidence of atrial fibrillation, acute kidney injury and pulmonary embolism with testosterone. The prescribing information for Aveed, a long-acting testosterone injection, summarizes TRAVERSE as showing more nonfatal arrhythmias warranting intervention (5.2% vs 3.3%), atrial fibrillation (3.5% vs 2.4%), acute kidney injury (2.3% vs 1.5%) and bone fracture (3.5% vs 2.5%) with the gel than with placebo. The same document reports that about 61% of all patients discontinued gel or placebo, a detail worth keeping in mind when reading the results.
Weigh benefits and risks from the same trial
In T4DM, at 2 years 12% of the testosterone group versus 21% of the placebo group had a 2-hour glucose of 11.1 mmol/L or higher, indicating type 2 diabetes (relative risk 0.59, 95% CI 0.43 to 0.80). In the same trial, a hematocrit safety trigger above 54% occurred in 22% on testosterone versus 1% on placebo, and prespecified serious adverse events occurred in 10.9% versus 7.4%. The authors said hematocrit increases might be treatment limiting.
In the Testosterone Trials, testosterone significantly increased sexual activity, sexual desire and erectile function and slightly improved mood and depressive symptoms, but had no significant benefit on vitality. The authors said the trial was too small to draw conclusions about risks. A cardiovascular substudy (170 men; 138 analyzed) found a significantly greater increase in noncalcified coronary plaque volume with testosterone gel than placebo (estimated difference 41 mm3, 95% CI 14 to 67). That measured plaque, not heart attacks.
Know what a review can and cannot add
Reviews pool many studies but share their limits. The ACP evidence review covered 38 RCTs of at least 6 months plus 20 observational studies. It found small improvements in sexual functioning and quality of life, and little to no effect on physical functioning, depressive symptoms, energy and vitality, or cognition. It found no trials powered to assess cardiovascular events or prostate cancer, and judged long-term efficacy and safety to be unknown.
The ACP 2020 guideline suggests clinicians discuss starting testosterone in men with age-related low testosterone and sexual dysfunction who want better sexual function, and suggests not starting it to improve energy, vitality, physical function or cognition. Both are conditional recommendations based on low-certainty evidence. A guideline is advice to clinicians, not a law.
What this does not tell you
These examples do not tell you whether testosterone fits your situation. Gel and undecanoate trials do not show how testosterone cypionate performs, and results in men over 65 or men with diabetes risk do not show what happens in other groups. A single trial also cannot answer questions it was not built or sized to answer. A licensed provider weighs studies like these against a person's own history and lab results.