The most hyped molecule in medicine
Testosterone may be the most oversold molecule in medicine. The distance between what the trials measured and what men are promised is itself measurable, and much of it sits in the placebo arms of those trials.
Testosterone is a real hormone with real effects, and the size of those effects has been measured repeatedly in randomised trials. Much of what is claimed for it in wellness marketing, and at standalone hormone clinics, rests on effects of a size that the placebo arms of those same trials also produced, rather than on the effect sizes the trials attributed to the hormone. A similar body of claim has grown up around testosterone prescribed to women, covered separately in testosterone in women. What follows is that distance, stated in the numbers the trials reported.
The placebo arm sets the baseline
Most discussion of testosterone quotes the difference between drug and placebo and stops there. The improvement inside the placebo group matters just as much, because it marks the part of what a treated man feels that the hormone does not account for.
| Placebo arm, measured outcome | Result |
|---|---|
| FACIT-Fatigue score at three months, Testosterone Trials | improved 7.2 points |
| SF-36 vitality score, same trial | improved 5.9 points |
| Men meeting the pre-specified threshold for a meaningful fatigue response | 138 of 207 men, 66.7%, against a design expectation of about 20% |
| Diary count of actual sexual activity | +0.1, then −0.1 |
| Men who believed they had benefited after two years of placebo injections, T4DM | 49% |
None of this means those men were imagining things. Fatigue improves when symptoms are taken seriously, when a man is enrolled in a trial and given reason to expect improvement, and that improvement is real without being evidence that a hormone produced it. The informative pattern lies inside the placebo arm itself: two self-rated scores moved substantially while the count of what the men actually did moved by a tenth of an encounter and then back again. Self-report moves; behaviour moves much less, and that contrast accounts for most of the gap between the marketing and the literature.
Most of the symptom list does not track testosterone
The European Male Ageing Study tested a long list of candidate symptoms against measured hormone levels to establish which of them travelled with low testosterone. In 3,369 men aged 40 to 79, with testosterone measured by mass spectrometry, three of 32 candidate symptoms were independently associated with a low level, and all three were sexual: fewer morning erections, reduced sexual thoughts, and erectile dysfunction. Fatigue, low mood and poor concentration were not among them.
Two qualifications sit inside that finding. Those three sexual symptoms are not specific to low testosterone either, occurring in roughly 25 to 35% of men with entirely normal levels, so their presence does not establish a diagnosis. And in EMAS itself, 42.9% of men classified as hypogonadal recovered to normal levels without any treatment, and their symptoms “did not show significant concurrent improvement” — the number changed while the men did not.
A single blood test is not a diagnosis
Testosterone is not a stable quantity. It falls after a meal, falls during acute illness, falls with poor sleep, and varies substantially within the same man from one week to the next. A single result describes a morning rather than a man.
The clearest figure on this comes from the Testosterone Trials screening data. Of 2,163 men whose first measurement qualified as low, 1,490 still qualified on a second draw, so roughly 31% no longer met the criteria with no treatment at all, simply by being measured again. A figure above 50% is sometimes quoted for the same phenomenon, but no primary study reports it, and the familiar 30% traces to a paper reporting statistical variance rather than a measured reclassification rate.
Diagnosis therefore rests on two morning, fasted measurements taken between 7 and 10 AM, together with LH, FSH and SHBG to establish where in the axis the problem sits. A diagnosis built on one afternoon draw is not a diagnosis.
Two measurements establish only that the level is genuinely low. Where in the system the problem sits is a separate question, and LH and FSH are what answer it: a low testosterone with a raised LH points to the testes themselves, while a low testosterone with an LH that is low or unremarkable points upwards, to the pituitary and hypothalamus. Those two situations have different causes and different treatments. SHBG belongs in the same first set of tests because it determines how much of a total testosterone result is available to tissue at all, so two men with the same total figure are not necessarily in the same position.
How often the diagnostic workup is done
Prescribing practice can be described from recorded data rather than from impression. In a review of 111,631 men who received testosterone in the Veterans Affairs system, 3.1% had the complete recommended workup, and 16.5% had no testosterone measured at all before starting. In a separate analysis, 93.7% of men prescribed testosterone had no documented disorder of the hypothalamic-pituitary-testicular axis.
Those figures describe what was documented, and the consequence is the point. Treatment started without two morning measurements, without LH and FSH, and without an identified cause rests on a symptom list that the EMAS data indicate mostly does not track the hormone, and it will then be judged against a placebo response that improved two-thirds of men. Both halves of that sentence matter: the reason for starting is weaker than it appears, and the evidence that it worked is harder to read than it appears.
A man who improves in those circumstances has no way to tell which of several explanations applies to him, and neither has his clinician, because the comparison that would settle it was never set up. That is the practical cost of skipping the workup, and it is separate from any question of whether the treatment itself was reasonable.
The claim with real biology behind it
One widely repeated statement runs in the opposite direction and is wrong. Androgen receptors are said to saturate at around 250 ng/dL, so that raising a level from 500 to 900 ng/dL adds nothing. The saturation model was proposed for the prostate, with prostate growth and PSA as its endpoints, and it was never proposed for muscle, energy or libido.
The dose-response trials show muscle behaving in the opposite way. Across achieved concentrations of 253, 306, 542, 1,345 and 2,370 ng/dL, fat-free mass rose at the top three doses with no plateau.
| Achieved concentration | Change in fat-free mass |
|---|---|
| 542 ng/dL | +3.4 kg |
| 1,345 ng/dL | +5.2 kg |
| 2,370 ng/dL | +7.9 kg |
Every gram of lean mass was gained above 250 ng/dL. Sexual desire separated between nearly every adjacent dose pair, so 805 ng/dL genuinely outperformed 470 ng/dL, and the largest concentration-response analysis found no threshold level for any outcome.
A mid-normal target is nonetheless the usual choice, and the reason is not that a higher level does nothing. Haematocrit rises with dose and rises more steeply in older men, so the same target is a different proposition in two different men depending on age, weight and the preparation used. Suppression of sperm production has an ED50 of about 50 mg per week, below the 125 mg per week needed to gain muscle, which means fertility is lost before any measurable physique change arrives. The argument for aiming mid-normal is that a higher level does something and brings consequences along with it, not that it does nothing.
Low testosterone is usually a signal rather than the disease
In most adult men a low testosterone level is downstream of something else: visceral fat, obstructive sleep apnoea, opioids, alcohol, chronic undersleeping, acute illness, or simply the week in which the blood was drawn. Those causes are reversible to varying degrees, and in many men one of them is the reason the level is low rather than an incidental finding beside it.
Exogenous testosterone does not correct that process. It replaces the output while the upstream cause continues, and it suppresses the body’s own production loop in the bargain. That trade is sometimes exactly the right one, and it should be made in the knowledge of what is being traded.
Which cause is operating in a given man, whether it can be addressed, and whether replacement is the right trade for him are the substance of an individual assessment at the practice.
Questions patients ask
I felt dramatically better within two weeks, so the testosterone is clearly working.
Possibly, although two weeks is also what the placebo arms produce.
What the evidence showsIn the Testosterone Trials, placebo-treated men improved 7.2 points on the FACIT-Fatigue scale at three months, and 66.7% met the response threshold against a design expectation of about 20%. In T4DM, 49% of men who had received two years of placebo injections believed they had benefited. Over the same period the placebo group’s diary count of sexual activity moved +0.1 and then −0.1.
What remains uncertainHow much of an individual man’s improvement is drug, placebo, regression to the mean, or the training and sleep changes that usually begin at the same time. No study has separated them.
Bottom lineFeeling better is real and worth having, and on its own it is not evidence about the hormone.
Strong
My symptoms match low testosterone, so my testosterone must be the problem.
For fatigue, mood and concentration, the data indicate otherwise.
What the evidence showsIn 3,369 men, three of 32 candidate symptoms were independently associated with low testosterone, and all three were sexual; non-sexual symptoms were not. Those three symptoms occur in 25 to 35% of men with normal levels. And when men in that study recovered normal testosterone without treatment, their symptoms did not improve alongside the number.
What remains uncertainWhether a subgroup exists in whom fatigue is genuinely androgen-driven. If it does, nobody has identified it.
Bottom lineA symptom list is a reason to investigate rather than a reason to treat.
Strong
My doctor checked my testosterone once and said it was low, so that settles it.
One draw misclassifies about a third of men.
What the evidence showsOf 2,163 men whose first testosterone measurement qualified as low, 1,490 still qualified on a second draw, so 31% no longer qualified, untreated. Testosterone falls after meals, during illness and with poor sleep, and varies week to week in the same man.
What remains uncertainHow many men currently on lifelong therapy would not have qualified on a second morning draw.
Bottom lineTwo morning fasted draws, plus LH, FSH and SHBG, before any conclusion about diagnosis.
Strong
Pushing my level to 900 will get me more than 600.
More muscle, yes, and more of the consequences as well.
What the evidence showsFat-free mass rose by 3.4, 5.2 and 7.9 kg at achieved concentrations of 542, 1,345 and 2,370 ng/dL, with no plateau, and sexual desire separated between nearly every adjacent dose. Haematocrit also rises with dose and more steeply in older men, and sperm suppression has an ED50 of about 50 mg per week, below the 125 mg per week needed for muscle gain.
What remains uncertainWhether any long-term benefit of a high-normal target offsets the monitoring it requires. No trial has compared target levels against each other for outcomes that matter to patients.
Bottom lineA higher level is not inert, and the reason to aim mid-normal is what comes with it rather than the absence of an effect.
Strong
Low testosterone is a disease in its own right, so correcting the number treats it.
In most men the low number is a consequence of something else.
What the evidence showsLow testosterone in adult men is commonly downstream of visceral fat, obstructive sleep apnoea, opioids, alcohol, chronic undersleeping or acute illness, and several of those are reversible. In EMAS, 42.9% of men classified as hypogonadal returned to normal levels without any treatment. Replacement raises the level while suppressing the body’s own production and leaving the upstream cause in place.
What remains uncertainHow often treating the upstream cause restores testosterone far enough to make replacement unnecessary, which has not been quantified in a trial.
Bottom lineThe cause of a low level determines whether replacing it is the right treatment.
Moderate
Ready for testosterone care built on a diagnosis?
The first step is a full endocrine evaluation, not a prescription. We see patients across San Diego County and welcome referrals from other physicians.
Our endocrinologists evaluate low testosterone in La Jolla, La Mesa and Poway: low testosterone care at Diabetes & Endocrine Specialists.


