A study of 358,957 men found that testosterone therapy is safe when prescribed for the right reason. For the one in three men taking it without a diagnosis, the cardiovascular risk is dramatically higher.
The testosterone therapy debate has been one of medicine’s most contested arguments for over a decade. Early observational studies raised alarming signals about cardiovascular risk. Industry-funded randomized trials in men with confirmed hypogonadism generally found no excess risk of major cardiac events. Independent meta-analyses found signals pointing in different directions depending on which studies they included. The FDA mandated label changes. Clinicians argued in journals. Patients were confused.
A study published in eBioMedicine, part of the Lancet portfolio, has produced what may be the most important clarification yet in this debate. It did not find that testosterone is safe. It did not find that testosterone is dangerous. It found that those two statements apply to two different groups of men who have been treated as one group throughout most of the existing literature.
The research drew on electronic health record data from 123 healthcare organizations across multiple countries, identifying 358,957 men between the ages of 30 and 75 who had started testosterone therapy. A three-year washout period excluded anyone with prior testosterone use or major cardiovascular events. Then the researchers did something deceptively simple: they checked whether each man had clinical evidence of hypogonadism before starting treatment.
Hypogonadism is the condition testosterone therapy was developed and approved to treat. It is defined by documented low testosterone levels combined with symptoms that are attributable to that deficiency. Clinical guidelines from endocrinology and urology societies require both biochemical confirmation and symptomatic evidence before prescribing testosterone. The guidelines exist precisely because testosterone is not a general wellness supplement. It is a hormone replacement therapy for a specific medical condition.
Among the 358,957 men identified, 35.4%, roughly one in three, had no documented evidence of hypogonadism before starting treatment. They were taking testosterone without a confirmed diagnosis. After propensity-score matching and up to 10 years of follow-up, what happened to these two groups diverged dramatically.
What the numbers showed
The primary outcome in the study was major adverse cardiovascular events, defined as a composite of heart attack, ischaemic stroke, and cardiac arrest. Men who initiated testosterone therapy without evidence of hypogonadism experienced this composite event at a rate of 16.53% over the follow-up period. Men who had been properly diagnosed with hypogonadism before starting therapy experienced it at 11.83%.
The hazard ratio for MACE was 1.51, meaning the off-label group faced a 51% higher risk of a major cardiovascular event compared to the appropriately diagnosed group.
All-cause mortality told a starker story. Men without evidence of hypogonadism had a hazard ratio of 1.90 for all-cause death. They were 90% more likely to die during the follow-up period than men who had been properly diagnosed before starting testosterone.
The findings extended across individual cardiovascular endpoints. Men without evidence of hypogonadism had 23% higher risk of ischaemic stroke, 41% higher risk of cardiac arrest, and 32% higher risk of heart failure compared to their properly diagnosed counterparts.
“The central message is clear: approximately one third of men receiving testosterone therapy lacked evidence of hypogonadism, and this prescribing pattern was associated with a clinically meaningful excess risk of major adverse cardiovascular and cerebrovascular events over long-term follow-up,” the researchers wrote.
Why the conflicting literature suddenly makes sense
The testosterone safety debate has been difficult to resolve partly because different studies were measuring fundamentally different things without knowing it. Clinical trials typically enroll men with confirmed hypogonadism who are carefully monitored under guideline-based protocols. Large observational studies capture whoever is actually being prescribed testosterone in real-world clinical practice, which, as this study demonstrates, includes a substantial proportion of men who do not meet diagnostic criteria.
When a clinical trial in properly diagnosed, carefully monitored hypogonadal men finds no elevated cardiovascular risk, and an observational study in a mixed population of diagnosed and undiagnosed testosterone users finds elevated cardiovascular risk, they are not measuring the same intervention in the same population. They are measuring two different things. The apparent conflict dissolves once the populations are separated.
This is exactly what this study showed when it compared the two groups directly. The men with confirmed hypogonadism showed a cardiovascular risk profile consistent with what clinical trials have documented. The men without confirmed hypogonadism showed substantially elevated risk, which explains why broader observational studies have found alarming signals.
Lead author Hatim Kerniss and colleagues were explicit about this interpretation: “The cardiovascular safety of testosterone therapy appears to be context-dependent and less favorable when treatment is initiated without evidence of hypogonadism, with clinically relevant heterogeneity across race and ethnicity, supporting biologically informed prescribing and cardiovascular risk surveillance.”
Who the undiagnosed group actually is
The one in three men taking testosterone without a confirmed diagnosis are not a fringe population. They represent a well-documented and growing pattern of testosterone prescribing that has expanded dramatically over the past two decades.
Direct-to-consumer advertising for testosterone products, including campaigns around Low T clinics, contributed to a surge in prescribing that extended well beyond men who met clinical diagnostic criteria. Men presenting with nonspecific symptoms such as fatigue, reduced energy, decreased libido, mood changes, and difficulty concentrating were frequently offered testosterone without the biochemical testing and clinical evaluation required to confirm hypogonadism. Some sought testosterone through online prescription services, concierge medicine, or sports medicine practices with looser diagnostic thresholds.
The result is a population of testosterone users that looks medically similar to properly diagnosed users but differs in one critical way: they did not have the specific hormonal deficiency that testosterone therapy was designed to address. In those men, the drug may be adding exogenous testosterone to a system that is not producing too little of its own. The physiological consequences of that scenario differ fundamentally from replacing a genuine deficiency.
The study found relevant heterogeneity in the cardiovascular risk elevation across racial and ethnic groups, with directionally consistent but varying magnitudes of excess risk across Black, White, Hispanic, and Asian men in the cohort. This variation suggests that the cardiovascular consequences of off-label testosterone use interact with underlying biology in ways that are not yet fully understood.
What the findings mean for men currently taking testosterone
The study’s implications are direct for any man who is currently taking or considering testosterone therapy and is uncertain whether he has actually been diagnosed with hypogonadism.
Men who have undergone formal testing, received a diagnosis of hypogonadism based on documented low testosterone levels and attributable symptoms, and are being monitored according to clinical guidelines are in the category that the study found to carry cardiovascular risk comparable to controlled clinical trial populations.
Men who began testosterone after reporting nonspecific symptoms such as fatigue or low energy, who have not had their testosterone levels formally measured and confirmed as deficient, who obtained prescriptions through online platforms or Low T clinics without comprehensive diagnostic evaluation, or who are using testosterone for performance, anti-aging, or general wellness purposes are in the category the study found to carry 50% higher cardiovascular risk and 90% higher mortality risk.
The researchers recommend that clinicians confirm hypogonadism with appropriate biochemical and clinical evaluation before initiating testosterone therapy and monitor cardiovascular risk throughout treatment, particularly in men who may have started therapy outside guideline-based indications.
“The findings reinforce the importance of confirming hypogonadism before starting therapy and monitoring cardiovascular risk over time,” the researchers concluded.
How the study controlled for confounding
A study comparing off-label and on-label users faces a fundamental challenge: the two groups may differ in ways other than their hypogonadism diagnosis that could explain the cardiovascular difference. Healthier men may be more likely to receive proper diagnostic workups. Men with more cardiovascular risk factors may be more likely to seek testosterone for energy and performance reasons without formal evaluation.
The researchers used 1:1 propensity-score matching to account for these differences, matching each off-label user to a properly diagnosed user on the basis of age, race, ethnicity, body mass index, baseline comorbidities including cardiovascular risk factors, diabetes, obesity, and dyslipidemia, baseline testosterone level where available, healthcare utilization patterns, and year of therapy initiation.
After matching, 113,554 pairs were followed for up to 10 years. The three-year pre-index washout period eliminated men with prior testosterone use or established cardiovascular disease, reducing the possibility that pre-existing conditions were driving the outcome differences.
A negative control outcome analysis tested whether the two groups differed on a condition that testosterone therapy biologically should not affect. If the matched groups were truly comparable, the negative control outcome should show no difference. If it showed a difference, that would suggest residual confounding was inflating the cardiovascular findings.
The negative control outcome showed no significant difference between groups, providing additional confidence that the cardiovascular risk elevation in the off-label group reflects something specific to the prescribing context rather than pre-existing differences between the populations.
What the study cannot establish
The study is observational and relies on electronic health record data, which captures what was documented rather than what was actually assessed. The absence of documented hypogonadism in a man’s record could reflect genuine lack of diagnosis, incomplete documentation of testing that was done, or variation in how different healthcare systems record diagnostic information.
The cause of the elevated cardiovascular risk in the off-label group cannot be identified from this data. Adding exogenous testosterone to a system that is not producing too little of its own could suppress endogenous production, alter hormonal balance, affect red blood cell production, or trigger other physiological changes that differ from the effects seen in genuinely hypogonadal men. The biological mechanism explaining the difference in cardiovascular outcomes remains to be established.
The results were consistent across racial and ethnic groups in direction but varied in magnitude, which the researchers describe as clinically relevant heterogeneity warranting further investigation.
What the study establishes with unusual statistical force, across 358,957 men followed for up to 10 years across 123 healthcare organizations, is that the population taking testosterone therapy is not homogeneous. It includes a large group of men who are taking the drug without confirmed hypogonadism, and that group is experiencing substantially worse long-term cardiovascular outcomes than the group for whom the therapy was designed and validated.
The study, “Off-label testosterone therapy is associated with higher long-term cardiovascular risk in men”, was authored by Hatim Kerniss, Philip Curman, Henning Olbrich, and colleagues across multiple international institutions, and published July 9, 2026 in eBioMedicine.
Source: eBioMedicine / The Lancet portfolio. DOI: 10.1016/j.ebiom.2026.106373
