Testosterone Therapy
The fear that testosterone causes heart attacks has collapsed under one large trial. What remains is narrower than the clinics selling "optimization" suggest: testosterone reliably treats genuine, symptomatic deficiency, but it is not a longevity drug or a tune-up for healthy men.
If you suspect low testosterone, fix sleep, apnea, weight and alcohol first, and get two fasting morning blood tests before anyone writes a prescription. If you are genuinely deficient and still symptomatic, treatment is reasonable — aimed at the middle of the normal range, not the top, with your red-cell count and blood pressure checked for as long as you take it.
Who actually has low testosterone
Testosterone falls slowly with age, by about 1.6% a year within the same man in a long-running study of 1,709 Massachusetts men aged 40–70.[1] In a similar Baltimore study of 890 men, levels low enough to count as deficient were present in about 20% of men over 60, 30% over 70 and 50% over 80.[2]
Where "low" starts depends on whose rulebook you use:
- American urologists: under 300 ng/dL (10.4 nmol/L) on two separate morning draws, plus symptoms.[3]
- The Endocrine Society: under 264 ng/dL (9.2 nmol/L), slightly stricter.[4]
- European urologists: 12 nmol/L (about 346 ng/dL) or less, the most inclusive.[5]
So a man at 320 ng/dL (11.1 nmol/L) is normal in America and deficient in Europe.
The symptoms — fatigue, low libido, poor recovery, flat mood, lost muscle — are vague and shared with poor sleep, alcohol, untreated sleep apnea, depression, overtraining and several common medications. That makes the test essential, and it is often skipped: among 61,474 insured US men aged 40 and over who started testosterone between 2001 and 2010, 26.6% had no testosterone level drawn beforehand, and only 61% had a baseline prostate-specific antigen (PSA) test.[6]
Large pooled cohort studies do link low natural testosterone to higher mortality — but the excess was seen in men below about 7.4 nmol/L (213 ng/dL), well under the usual diagnostic thresholds, and an association with low natural levels is not evidence that treatment helps.[7]
Fix these first
A meaningful fraction of men who present as "low T" probably recover into the normal range once the things suppressing their own production are corrected. The Endocrine Society puts this ahead of prescribing; the American urologists only conditionally recommend counselling about it and do not require it first.[8]
- Sleep and sleep apnea. One week of five hours in bed cut daytime testosterone by 10–15% in ten healthy young men.[9] Untreated obstructive sleep apnea is a major reversible suppressor and causes the same fatigue and low libido testosterone gets prescribed for. Get screened before starting — see sleep and breathing.
- Weight and exercise. 150–300 minutes a week of moderate aerobic activity plus resistance training cuts visceral fat and improves sexual function; the Endocrine Society puts weight loss ahead of hormone replacement in obese and metabolically unwell men.[10] See resistance training.
- Stress, alcohol and general health. Chronic stress hormones and heavy drinking both suppress production. Massachusetts men with no chronic illness, no prescription drugs, no obesity and no heavy drinking had levels 10–15% higher than their peers of the same age.[11]
- Medications. Opioids, steroid tablets and several other common drugs suppress testosterone on their own — have the medication list reviewed before treating.
- Hormone-disrupting chemicals. Phthalates, bisphenols and parabens interfere with androgen signalling, and exposure is quickly changeable: when 103 young French women swapped their personal-care products for five days, urinary levels of several of these chemicals fell by roughly 30–64%.[12] That measured exposure in women, not testosterone in men. See environmental toxins.
- Varicocele repair, where a clinical varicocele (enlarged scrotal veins) is present, raised testosterone by about 82 ng/dL (2.8 nmol/L) across 48 before-and-after studies,[13] and by about 114 ng/dL (4.0 nmol/L) in men who started out deficient.[14] These were not randomised trials, and the men were mostly in their late twenties to mid-forties being seen for infertility or sexual problems — this does not transfer to a 60-year-old without a varicocele.
The heart question is settled; other side effects are not
For a decade testosterone carried a 2015 FDA warning, added to every label, about a possible increase in heart attack and stroke. One randomised trial settled it. TRAVERSE assigned 5,204 men aged 45–80 with confirmed deficiency and either heart disease or high cardiovascular risk to testosterone gel or placebo for an average of 21.7 months. Heart attack, stroke or cardiovascular death struck 182 men on testosterone (7.0%) and 190 on placebo (7.3%) — no detectable difference.[15] Nor was there more prostate cancer or worse urinary symptoms, but men with a PSA above 3.0 ng/mL, severe urinary symptoms or a prostate nodule were excluded, so the reassurance doesn't extend to them.[16]
The trial was not entirely clean. On testosterone compared with placebo:
- Atrial fibrillation was 3.5% against 2.4%, about one extra case per hundred men — a safety signal rather than proof, since the trial was not designed to detect it.[17]
- Acute kidney injury (2.3% against 1.5%) and lung clots (0.9% against 0.5%) were also more common; in men with heart disease or high risk, so read them as an upper bound.[18]
- Fractures were 3.5% against 2.5% over about three years, roughly one extra per hundred men, most after falls.[19] In older men with low testosterone it can raise spine density, but raising bone density is not the same as preventing fractures. See bone density.
- Blood pressure rose. In February 2025 the FDA ordered the cardiovascular-risk warning off testosterone labels but extended a blood-pressure warning to every product, after required monitoring studies found systolic rises of about 2 mmHg on gel and about 2–5 mmHg on the oral capsules; against placebo in TRAVERSE the gap was 1.5 mmHg.[20]
That February 2025 action did not turn testosterone into an anti-aging drug. Boxed warnings remain in the class — gels still carry one about children absorbing testosterone through skin contact — and the labels still say safety and effectiveness in "age-related hypogonadism" have not been established; testosterone is approved only for deficiency with an identifiable cause. In April 2026 the FDA invited manufacturers to apply for a new indication — low libido in men with low testosterone and no identifiable cause — but nothing has been approved, and any approval "would require demonstration of substantial evidence of effectiveness".[21] In July 2026 the Endocrine Society replied that its 2018 guideline stands and that symptoms alone are not diagnostic.[22]
Why more is not better
A 2026 genetic study points the other way from TRAVERSE. It used inherited gene variants as a stand-in for lifelong testosterone exposure, built from more than 425,000 UK Biobank adults and tested against coronary disease in over 1.16 million people. Men genetically inclined to higher testosterone had about 17% higher odds of coronary artery disease (odds ratio 1.17, 95% CI 1.07–1.27 — the range sits wholly above 1, so the link is unlikely to be neutral). There was no such link in women.[23] Adjusting for blood pressure substantially weakened it, which suggests decades of slightly higher blood pressure are the main route. The effect is small: the authors calculate a trial would need more than 6,400 men to detect it.
The two results answer different questions. Restoring normal levels for a few years in a deficient older man is not the same thing as carrying higher-than-average testosterone from puberty onwards.[24] The genetic signal doesn't undo the safety result for genuine deficiency; it argues against "optimizing" men who are already normal. Pushing a 55-year-old toward the levels of a 25-year-old buys subjective gains at the price of a higher blood-pressure load for decades.
Why "77% fewer deaths" is too good to be true
A 2024 registry of 737 men with adult-onset deficiency, followed for a median of 114 months, reported that men on long-acting injected testosterone had a 77% lower death rate than untreated men (hazard ratio 0.23, 95% CI 0.14–0.40), after adjusting for age, waist, blood sugar, cholesterol, blood pressure, smoking and diabetes.[25]
That would beat statins, quitting smoking and exercise — from a hormone injection. It is implausible, and the design explains why. This was a non-randomised, unblinded, single-centre registry: the treated men chose treatment, stayed on it and kept turning up for eleven years. People who stick with any therapy, even a placebo, die less. And you cannot be counted as a treated survivor until you have survived long enough to be treated. Statistical adjustment removes neither bias. The site's glucosamine article shows how a far smaller effect can be produced by selection alone.
The randomised evidence decides it: TRAVERSE found no mortality benefit. What survives from the registry is only its shape — the apparent advantage was confined to the men with the worst cardiometabolic health, the same pattern seen with the Ozempic-class drugs. Testosterone is a treatment for deficiency, not a life-extension lever.
What it doesn't do
Prevent diabetes. In TRAVERSE's men with prediabetes, testosterone did not significantly slow progression to diabetes (for example 10.1% against 14.6% at two years, overall not significant), improve blood sugar or reverse existing diabetes, and the authors concluded that testosterone alone should not be used to prevent or treat diabetes.[26] Levels sat in the lower half of the target,[27] so a benefit at higher levels can't be fully ruled out. The "testosterone prevents diabetes" claim traces to a different trial, T4DM: 1,007 Australian men aged 50–74 with a large waist and poor glucose control but no true deficiency, all in a lifestyle programme. Diabetes developed in 12% on injected testosterone against 21% on placebo.[28] But hematocrit crossed the 54% safety trigger in 22% of treated men against 1%, it has never been replicated, and the drug's maker helped fund it. Metformin and the Ozempic-class drugs are better-evidenced ways to prevent diabetes.
Sharpen your mind. A 2025 review reported moderate cognitive gains, but its abstract and results disagree with each other and it showed significant publication bias.[29] The trials say otherwise. A year of testosterone gel in 493 men over 65 with memory complaints and low testosterone did nothing for memory, and every other cognitive score nominally favoured placebo.[30] TRAVERSE also found no effect on cognition, though it did find modest improvements in mood and energy.[31] For thinking and memory, the dementia-prevention levers do far more.
Lift depression. Reviews report improvement in treatment-resistant depression,[32] but pooling the individual records from 17 placebo-controlled trials (median age 67) found that a year of testosterone improved erections and deficiency symptoms but not depressive symptoms.[33] Those men were not selected for depression, so the question is not closed for men who are clinically depressed.
Turn back your biological clock. The often-cited TRIIM pilot — nine men, no control group — used growth hormone, DHEA and metformin, not testosterone, and reported epigenetic age about 1.5 years below baseline after a year.[34] Its larger follow-up has posted no results.[35] In population data, men with a higher testosterone-to-estradiol balance look slightly younger on one DNA-methylation aging clock (not on six others) and on a methylation marker of clotting risk,[36] and in a small uncontrolled study of gender-affirming therapy, testosterone held trans men's methylation-estimated pace of aging steady while slightly shortening estimated telomere length.[37] None of these tested testosterone treatment against biological aging in a controlled trial; these are associations, not a treatment. See hallmarks of aging.
Testosterone for women
Women make testosterone too, and testosterone for women has one evidence-based use: low sexual desire that causes distress after menopause, diagnosed by a full clinical assessment, never by a blood level.[38] Claims of more energy and better quality of life rest on uncontrolled, unblinded studies, such as a survey of women who had paid for treatment through a telehealth platform.[39] In one chart review of 100 women given pellets, 58 had side effects and two developed breast cancer during therapy.[40]
Short trials through the skin recorded no cardiovascular deaths, though the certainty about long-term safety is low.[41] Acne and unwanted hair growth are dose-related. The US has no approved female product; the global consensus calls an approved male product at doses that keep levels in the female range, regularly checked, reasonable, and advises against compounded "bioidentical" testosterone wherever an approved product, including a male one, is available, and against pellets, which produce levels well above the female range.[42]
Doing it properly
Before starting (on top of the general midlife lab panel):
- Two fasting morning total testosterone tests — levels peak after waking, and a non-fasting draw can read falsely low.
- Sex hormone-binding globulin and calculated free testosterone when the total is borderline or the binding protein is likely off, which covers most obese, diabetic and older men.
- Hematocrit (the share of blood that is red cells): if it is above 50%, hold off until the cause is found. Also LH, FSH and prolactin; estradiol if there are breast symptoms; and PSA in men over 40.
- Metabolic panel, lipids including apoB, and HbA1c.
Choosing a form. Sources: [43], [44], and the current FDA prescribing information for the oral and under-the-skin products.
| Form | How often | Main trade-off |
|---|---|---|
| Cypionate / enanthate injection into muscle | Every 7–14 days | Cheap and easy to adjust; sharp peaks and troughs, mood and libido swings, and the highest rate of thickened blood |
| Long-acting undecanoate injection | At start, at 4 weeks, then every 10 weeks | Most convenient; you must stay 30 minutes after each shot to watch for oil reaching the lungs and allergic reactions |
| Oral undecanoate capsules | Twice daily with food | No needles or skin transfer; must be taken with a meal; carries the blood-pressure warning; label asks for hematocrit checks about every 3 months. Not the older oral androgens, which should never be prescribed |
| Weekly under-the-skin auto-injector | Weekly, self-given | Smaller swings than two-weekly injections; label asks for hematocrit checks about every 3 months |
| Gel | Daily | Mimics the daily rhythm; can transfer to women and children through skin contact |
| Pellets under the skin | Every 3–4 months | Nothing to remember; minor procedure; hard to adjust if side effects appear |
Targets and follow-up.
- Aim for the middle third of the normal range, about 450–600 ng/dL (15.6–20.8 nmol/L) — the urologists' own target, and in line with the genetic blood-pressure finding.[45]
- Recheck levels 2–4 weeks into a gel, or mid-cycle after 3–4 injection cycles.
- Then every 6–12 months: testosterone, hematocrit, PSA, lipids, blood pressure — hematocrit about every 3 months on the oral capsules and the auto-injector, as their labels direct.
- After three to six months, if your level is normal but you feel no better, discuss stopping.
Who should be careful
- Men who want children. Testosterone shuts down the body's own hormone signals, halts sperm production and shrinks the testicles. Fertility needs a different approach — hCG, clomiphene, an aromatase inhibitor or a combination. These are not equivalent: only hCG carries an FDA label in men (for deficiency caused by low pituitary hormone signalling), the others are off-label, and enclomiphene is approved by neither the FDA nor the European regulator and reaches patients only as a compounded product.[46]
- Everyone on treatment: thick blood. Raised hematocrit is the most consistent side effect and raises the risk of clots. A hematocrit of 54% or above is the bright line: lower the dose first; a longer interval or a switch to gel is reasonable, since thickened blood runs at 19% on muscle injections, 12.5% on pellets and 5.4% on gels. If it stays high on low or normal testosterone, that needs a hematologist, not do-it-yourself blood removal.[47]
- Men with atrial fibrillation, high blood pressure, kidney disease, a past clot, or fall and fracture risk. All five turned up as signals in TRAVERSE.
- Men with a raised PSA, severe urinary symptoms or a prostate nodule. They were excluded from the reassuring prostate data; PSA monitoring stays standard over 40.
- Women and children in close contact with a gel user. Gel can transfer through skin contact.
Further reading
- Lincoff AM et al. Cardiovascular Safety of Testosterone-Replacement Therapy (TRAVERSE). NEJM 2023.[48]
- Snyder PJ et al. Testosterone Treatment and Fractures in Men with Hypogonadism. NEJM 2024.[49]
- Bhasin S et al. Effect of Testosterone on Progression From Prediabetes to Diabetes in Men With Hypogonadism: A Substudy of the TRAVERSE Trial. JAMA Intern Med 2024 — no significant effect on diabetes progression or glycemic control.[50]
- Morbey EJ et al. Higher Circulating Testosterone Linked to Higher CAD Risk in Men: Mendelian Randomization and Survival Analyses. JCEM 2026;111(4):e1029–e1036.[51]
- Mauvais-Jarvis F. Testosterone and Cardiovascular Disease: The Enduring Myth. JCEM 2026.[52]
- Mann A et al. Testosterone replacement therapy: association with mortality in high-risk patient subgroups. Andrology 2024.[53]
- Resnick SM et al. Testosterone Treatment and Cognitive Function in Older Men With Low Testosterone and Age-Associated Memory Impairment. JAMA 2017.[54]
- Hudson J et al. Symptomatic benefits of testosterone treatment in patient subgroups: a systematic review, individual participant data meta-analysis, and aggregate data meta-analysis. Lancet Healthy Longev 2023.[55]
- Bhasin S et al. Testosterone Therapy in Men With Hypogonadism: An Endocrine Society Clinical Practice Guideline. J Clin Endocrinol Metab 2018.[56]
- Evaluation and Management of Testosterone Deficiency. American Urological Association guideline, 2018 (validity reviewed 2024).[57]
- FDA issues class-wide labeling changes for testosterone products — cardiovascular-risk warning removed, blood-pressure warning extended class-wide, age-related-hypogonadism Limitation of Use retained. 28 February 2025.[58]
- Davis SR et al. Global Consensus Position Statement on the Use of Testosterone Therapy in Women. JCEM 2019.[59]
- Feldman HA et al. Age trends in the level of serum testosterone and other hormones in middle-aged men: longitudinal results from the Massachusetts Male Aging Study. J Clin Endocrinol Metab 2002.[60]