Testosterone, Thyroid, and the Space Between Normal and Optimal

A man of forty-six arrives with a lab printout and a proposal. His total testosterone came back at 340 ng/dL — about 11.8 nmol/L — flagged in bold, and a clinic has offered him a prescription on the strength of it. He is tired, his training has stalled, and he has gained nine kilograms in four years. The number is real. The conclusion drawn from it — that low testosterone is his diagnosis — is very likely wrong.

Testosterone and thyroid panels are among the most frequently ordered and most frequently misread tests in preventive medicine. The errors run both ways: men medicated for a value that sleep and weight loss would have corrected, and patients told their thyroid is normal while a treatable pattern sits in plain sight. The failures are structural rather than careless. They follow from how reference ranges are built, how these hormones move across a day, and how readily one measurement is mistaken for a diagnosis.

What a reference range actually tells you

A reference range is a statistical description of a population, not a statement about health. It captures the central 95 percent of people who were tested — a group that includes the obese, the sleep-deprived and the subclinically ill, because those people have blood tests too. Age adjustment compounds the problem: when ranges are stratified by decade, whatever is common at that age becomes normal by construction. That is defensible if the decline is physiological, and unhelpful when it is being driven by weight gain, untreated sleep apnea or alcohol, as it frequently is.

Pulsatility does the rest. Testosterone is secreted in pulses, peaks in the early morning, and falls measurably after a poor night, an acute illness or a large meal. A mid-afternoon sample drawn after four hours of sleep is not a measurement of a man's androgen status; it is one unrepresentative moment.

What low testosterone actually presents as

Hypogonadism requires both consistent symptoms and reproducibly low measurements. The relatively specific features are sexual and physical: reduced libido, loss of spontaneous morning erections, erectile dysfunction, small or shrinking testes, gynaecomastia, reduced bone density. The non-specific ones — fatigue, low mood, poor concentration, muscle loss and fat gain — are what bring most men in, and they are shared with depression, sleep apnea, anemia, thyroid disease and ordinary overwork. A man with fatigue alone, whose libido and morning erections are intact, has an unlikely diagnosis whatever the number says.

Confirmation is a sequence, not a single test. Roughly a third of men with one low reading are back within range when the sample is repeated properly.

Low testosterone is more often a symptom than a diagnosis

In most men presenting in midlife with a low or borderline total testosterone, the axis is intact and something else is suppressing it. Obesity is the largest single contributor: adipose tissue converts testosterone to estradiol, insulin resistance lowers SHBG and with it the total, and central adiposity suppresses hypothalamic output. Weight loss raises testosterone roughly in proportion to the weight lost. Others recur constantly:

• Obstructive sleep apnea and chronic short sleep — apnea depresses testosterone independently of weight, and even a week of restricted sleep lowers daytime levels in healthy young men.
• Alcohol, at intakes patients routinely describe as moderate.
• Opioids, which suppress the axis profoundly and reliably — along with glucocorticoids and anything that raises prolactin.
• Training volume outrunning energy intake — the low energy-availability picture familiar from endurance athletes, and not confined to women.
• Anabolic steroid use, current or past — common, under-declared, and capable of leaving the axis suppressed long after the last cycle.

A low testosterone level is a finding, not a diagnosis. In a large proportion of men it is the body reporting accurately on sleep, weight, alcohol or training load. Prescribing testosterone silences the report without addressing what generated it — and it is the only item on that list that cannot easily be undone.

Hence the order of operations. A reversible driver should be addressed first, or at least alongside — not deferred behind a prescription that suppresses the very axis you were trying to assess. This is not an argument for delay: severe, confirmed, symptomatic hypogonadism should be treated. It is an argument against treating the number in isolation.

What treatment does, and what it costs

For men with symptomatic, biochemically confirmed hypogonadism, testosterone replacement works. Sexual function improves most reliably; lean mass rises and fat mass falls modestly, bone density improves, and mood and energy often lift. Effects on cognition, physical performance and glucose metabolism are smaller and less consistent. In men with borderline levels and non-specific symptoms the effect sizes shrink and the placebo response is large — precisely the group most likely to be offered treatment commercially.

On cardiovascular safety the field spent a decade in genuine uncertainty. A large randomised trial designed to settle it, in middle-aged and older men with hypogonadism and established or high cardiovascular risk, found no excess of major adverse cardiac events against placebo. That is meaningful reassurance on the central question. It was not entirely clean: higher rates of atrial fibrillation, pulmonary embolism and acute kidney injury were seen on treatment, and those belong in the conversation rather than in a footnote.

The consequence most consistently under-discussed is fertility. Exogenous testosterone suppresses LH and FSH; intratesticular testosterone falls to a fraction of what spermatogenesis requires, and sperm counts drop, often to zero. Recovery after stopping is usual but neither guaranteed nor quick — months to more than a year, and less reliable with older age and longer treatment. Any man who might want children should hear this before the first prescription, along with the alternatives: cryopreservation, or axis-preserving approaches such as hCG. Treatment also commits him to monitoring:

• Symptoms and a trough level, to confirm the dose produces something the patient can feel and not only a satisfying number.
• Hematocrit, which rises on therapy; above 54 percent calls for dose reduction, a change of formulation, or interruption.
• Prostate assessment and PSA to age-appropriate practice, with a baseline before starting — and blood pressure, which some formulations raise.
• Duration. The axis is suppressed throughout treatment, so stopping is not a neutral act; men often feel worse afterwards, for a time, than they did before they began.

The thyroid: two numbers, carefully read

Thyroid testing is simpler than it is usually made to look. TSH is a logarithmically amplified readout of how the pituitary perceives thyroid hormone at tissue level — small changes in free T4 produce large changes in TSH — which is what makes it the sensitive first test. Adding free T4 resolves nearly everything else. Free T3 fluctuates with nutrition and illness and rarely changes management; reverse T3 has no validated role in outpatient diagnosis, whatever is claimed for it.

Two things change the interpretation. Thyroid antibodies — principally TPO — establish an autoimmune cause, predict progression from a borderline TSH to overt disease, and matter considerably in pregnancy and fertility; they are the most useful single addition when a TSH sits in the grey zone. Interference is equally real: high-dose biotin, common in hair and nail supplements, distorts many immunoassays and can mimic hyperthyroidism in a well person. Acute illness perturbs these tests too, and TSH drifts upward with age.

Subclinical hypothyroidism, and why "tired" is a poor reason to treat

A raised TSH with a normal free T4 is common, and a meaningful proportion of these values normalise when repeated some weeks later — which is why repeating before acting prevents a great many unnecessary lifelong prescriptions. Where the TSH is persistently above about 10 mIU/L, treatment is generally warranted. Between the upper limit of normal and 10 the decision is individual: positive TPO antibodies, a rising TSH, younger age, pregnancy or planned pregnancy, or a goitre argue for treating, while an isolated mildly raised value in an older adult argues for monitoring. Randomised trials of levothyroxine in older adults with mildly elevated TSH have not shown improvement in tiredness or quality of life.

Fatigue deserves particular scepticism as an indication. It is among the least specific symptoms in medicine, and its common causes — fragmented sleep, sleep apnea, iron deficiency, depression, alcohol, medication effects, deconditioning, simply doing too much — are far more prevalent than thyroid disease. Treating a borderline TSH in a tired patient often produces a normal number and an unchanged patient, and overtreatment carries its own risks: atrial fibrillation and accelerated bone loss in older adults. A minority do remain symptomatic on levothyroxine at an adequate dose; that phenomenon is real and unexplained, and trials of combined T4 and T3 have not shown consistent benefit.

Menopause, briefly and honestly

No area of hormonal medicine has been more distorted by the reception of a single trial. The initial reporting of the Women's Health Initiative in 2002 produced an abrupt worldwide collapse in the use of menopausal hormone therapy. Later analyses, stratified by age and by time since the final menstrual period, told a different story: for women beginning therapy near the menopause and generally under sixty, the balance of benefit and risk is considerably more favourable than the original headline implied.

Systemic therapy remains the most effective treatment available for vasomotor symptoms, and local vaginal estrogen — with minimal systemic absorption — is a straightforward and under-used treatment for genitourinary symptoms. The risks are real and depend on preparation, route and duration: the venous thromboembolism signal is largely a feature of oral estrogen, and the breast cancer signal is modest and tied mainly to combined estrogen-progestogen therapy. This is a decision with genuine trade-offs, not an anti-ageing therapy, and it deserves a conversation with someone who knows the current evidence rather than the 2002 summary of it.

What to ask your physician

• If a hormone level is called low, ask when the sample was taken and whether it has been repeated. For testosterone: morning, fasting, two separate days, before any prescription.
• Ask what else could be producing the number — weight, sleep and possible apnea, alcohol, medications, training load, anabolic steroid use past or present — and which of those is being addressed.
• Ask for LH, SHBG and free testosterone before treatment rather than after. Once therapy begins, what those tests would have told you is no longer obtainable.
• If you might want children, raise fertility before the first prescription, not after.
• For thyroid, ask for TSH with free T4, and TPO antibodies if the TSH is borderline. Before starting levothyroxine for a mildly raised value, ask for a repeat — and ask what specifically is expected to improve, and by when.

Hormones are a system with feedback loops, daily rhythms and a long list of upstream inputs, and a laboratory value is one observation of that system at one moment. The question worth answering is not whether a number falls inside a printed range, but whether the whole pattern — symptoms, timing, upstream drivers, the rest of the panel — describes someone who has a hormonal problem. Treat the person and the pattern. Never the single number.