The European Male Ageing Study showed that the effects of age and of adiposity on the axis differ. Age was associated mainly with a rise in luteinizing hormone and sex hormone-binding globulin, a pattern of primary testicular impairment, whereas obesity was associated with low total and free testosterone without a compensatory rise in luteinizing hormone (Wu et al., 2008). In the same cohort, men who gained weight during follow-up lost testosterone and men who lost weight regained it (Camacho et al., 2013). Chronic disease, opioids and glucocorticoids act in the same direction. Much of what is called age-related hypogonadism is thus better described as the hormonal footprint of poor health.

Symptoms alone do not identify the condition. In the European cohort only three sexual symptoms, poor morning erection, low sexual desire and erectile dysfunction, were consistently related to low testosterone, and when these were combined with biochemical thresholds the syndrome was present in about two percent of men aged 40 to 79 (Wu et al., 2010). The Endocrine Society accordingly recommends a diagnosis only in men with symptoms and signs of deficiency and unequivocally and consistently low concentrations, measured fasting in the morning on at least two occasions, and it advises against screening the general population (Bhasin et al., 2018). Assay variability long hampered the choice of thresholds, and a harmonized reference range from four cohorts now provides a common footing (Travison et al., 2017).

The clearest benefit is sexual. Men assigned to testosterone reported more sexual activity, greater desire and better erectile function than men assigned to placebo (Snyder et al., 2016). The effect was statistically robust and of moderate size, smaller for erectile function than that expected from a phosphodiesterase inhibitor. It is on this evidence that the American College of Physicians suggests discussing treatment only with men who have age-related low testosterone and sexual dysfunction and who wish to improve sexual function, and suggests against starting it to improve energy, vitality, physical function or cognition (Qaseem et al., 2020).

Results for the outcomes that most often prompt a prescription were disappointing. Among men enrolled for slow walking, testosterone did not significantly increase the proportion who improved their six-minute walk distance by the prespecified margin, although an effect emerged when all participants were pooled. Fatigue scores did not improve by the primary criterion, and there were small favorable changes in mood and depressive symptoms (Snyder et al., 2016). Testosterone reliably increases muscle mass; whether that translates into function that matters to an older man, such as rising from a chair, climbing stairs or avoiding falls, is much less certain. The trials were also short, and a year may be too brief for changes in muscle to alter mobility in either direction.

Three further effects are well documented. Volumetric bone density and estimated bone strength rose, more in the spine than the hip (Snyder et al., 2017). Hemoglobin increased, and in men with otherwise unexplained anemia it was frequently restored to normal (Roy et al., 2017). Memory and other cognitive functions did not change (Resnick et al., 2017). In the Australian T4DM trial, men with impaired glucose tolerance or newly diagnosed type 2 diabetes who received injectable testosterone undecanoate on top of a lifestyle program were less likely to have diabetes at two years than those who received the program with placebo (Wittert et al., 2021). That finding is important, but it does not show that testosterone is preferable to more effective weight loss.

The alarm was raised by the Testosterone in Older Men with Mobility Limitations trial, which was stopped early because cardiovascular adverse events were more frequent in the testosterone group (Basaria et al., 2010). The participants were frail, the doses were relatively high and the events were heterogeneous and not a prespecified endpoint. Two observational analyses then reported more myocardial infarction and death among treated men (Vigen et al., 2013; Finkle et al., 2014); both attracted methodological criticism, and both influenced regulators. Within the Testosterone Trials, coronary computed tomography showed a greater increase in noncalcified plaque volume with treatment, a finding of uncertain clinical meaning (Budoff et al., 2017).

TRAVERSE was designed to answer the question directly. It enrolled men aged 45 to 80 with hypogonadal symptoms, two low testosterone measurements and either established cardiovascular disease or high risk of it, and it found testosterone gel noninferior to placebo for a composite of cardiovascular death, nonfatal myocardial infarction and nonfatal stroke (Lincoff et al., 2023). An individual participant meta-analysis of earlier trials had reached a compatible conclusion for the short to medium term (Hudson et al., 2022). The reassurance is real but bounded. Adherence in TRAVERSE was incomplete, average exposure was around two years, and the achieved testosterone concentrations were modest. The trial also recorded higher incidences of atrial fibrillation, acute kidney injury and pulmonary embolism in the testosterone group. These were secondary safety observations and not tests of a prior hypothesis, but they are biologically plausible, given the effect of testosterone on hematocrit, and they are a reason for continued caution in men with a history of thrombosis or arrhythmia.

Prostate events in TRAVERSE were infrequent and did not differ between groups, including high-grade cancer (Bhasin et al., 2023). Men at raised prostate risk were excluded at entry and monitoring was structured, so the result applies to carefully selected and supervised patients. Erythrocytosis is the adverse effect seen most often in every trial; guidelines require hematocrit to be checked before and during treatment and treatment to be withheld or reduced when it exceeds the stated limit (Bhasin et al., 2018). The most surprising safety finding was skeletal. Despite the gain in bone density seen earlier, clinical fractures in TRAVERSE were more frequent with testosterone than with placebo (Snyder et al., 2024). No mechanism has been established, and the result is a caution against inferring fracture protection from densitometry.