TSH and free T4: the test that can reverse the meaning of a testosterone result

The test that can make a testosterone result mean the opposite of what it appears to mean.

Why it is on the panel at all

Two reasons, and only one of them is the obvious one.

The obvious reason is that hypothyroidism and testosterone deficiency produce the same complaints. Fatigue, low libido, low mood, poor concentration, weight gain — the lists are nearly identical, and neither can be distinguished from the other at the bedside. In the Colorado Thyroid Disease Prevalence Study of 25,862 people, symptoms were more common in hypothyroid than euthyroid participants, but the authors stated plainly that “individual symptom sensitivities were low.” A single symptom is close to worthless. A cluster of newly changed symptoms carries real weight: reporting seven or more changed symptoms gave a likelihood ratio of 8.7, and two or fewer a likelihood ratio of 0.5. That is the same statistical shape as the symptom set used for testosterone deficiency, which is exactly why the two get confused.

The less obvious reason is the important one. Thyroid disease changes SHBG, and SHBG changes what a total testosterone means. A man with untreated thyrotoxicosis can present with a total testosterone that looks high while his bioavailable testosterone is genuinely low. A man with untreated hypothyroidism can present with a total testosterone that looks low and normalises on levothyroxine alone. In both cases, reading the testosterone without knowing the thyroid status gives the wrong answer.

Be honest about the yield, though. A man is roughly six times less likely than a woman to develop spontaneous hypothyroidism — 0.6 versus 3.5 per 1,000 per year in the Whickham survey’s twenty-year follow-up — and Whickham described the incidence of hyperthyroidism in men as “negligible.” We could find no study reporting what proportion of men evaluated for testosterone deficiency turn out to have thyroid disease instead, and we are not going to invent one. The reason to check TSH is not that the yield is high. It is that the miss is consequential, the test is cheap, and thyroid status changes how the testosterone result should be read.

What thyrotoxicosis does — the pattern most likely to mislead

The cleanest evidence is experimental. Seven healthy lean men were given low-dose T3 for 63 days in a metabolic unit:

Measure	Baseline	After T3	Change
Total testosterone	18.5 nmol/L	26.3 nmol/L	+42%
SHBG	18.0 nmol/L	44.9 nmol/L	+150%
Free testosterone	142.8 pmol/L	137.3 pmol/L	no change

Induce thyroid hormone excess in a normal man and his total testosterone rises by over 40% while his free testosterone does not move at all. Small study, seven men — but it is an interventional design in healthy volunteers, which is why it carries weight out of proportion to its size.

In clinical Graves’ disease the magnitudes are larger and the direction of the free fraction becomes clinically important. In 25 men with active Graves’ disease compared with 10 normal men: total testosterone 9.3 versus 5.4 ng/mL, SHBG 102.3 versus 19.0 nmol/L — more than five times the control mean — and bioavailable testosterone 1.7 versus 3.1 ng/mL, that is, 45% lower. The authors’ own conclusion was that bioavailable testosterone identified hypoandrogenism in these men despite their high total testosterone. Semen abnormalities were near-universal.

Where the sources disagree, and they do. Two other within-subject series found calculated free testosterone essentially unchanged in hyperthyroid men, while the non-SHBG-bound fraction fell. A fourth found the free androgen index reduced. The likely reconciliation is methodological — bioavailable testosterone includes the albumin-bound fraction and falls as SHBG sequesters more, while the true free fraction is better defended — but that is our interpretation, not a finding, and you should know that the underlying studies do not all point the same way.

The reassuring part is consistent across all of them: it reverses. Hormonal and seminal abnormalities corrected when men became euthyroid after radioiodine; SHBG fell on carbimazole to values not significantly different from euthyroid controls; and in six men restudied after seven to nineteen months of euthyroidism, about 85% of semen abnormalities had normalised.

Evidence: Established for the SHBG mechanism and its reversal. Mixed on what happens to true free testosterone.

What hypothyroidism does

The opposite, and with a different pituitary signature. Hypothyroidism lowers SHBG, and it produces a hypogonadotropic pattern — low testosterone with LH and FSH that are not elevated, which is the same pattern produced by obesity, opioids, glucocorticoids and acute illness.

The single most useful reference here is small and old. Ten men with primary hypothyroidism were measured before and after levothyroxine:

Free testosterone rose from 161 to 315 pmol/L — approximately double — on thyroid hormone alone, with no androgen given at all. Gonadotropins were not elevated. Hyperprolactinaemia was not present in most of these men, though prolactin fell as free testosterone rose.

State the design honestly: ten men, uncontrolled, before-and-after. There is no randomised trial of levothyroxine with testosterone as an endpoint in hypothyroid men. This is the best primary evidence that exists, and it is not strong evidence — but it is the right shape, and it is corroborated by review-level statements that free testosterone is reduced in primary hypothyroidism and normalises on replacement.

There is one further wrinkle worth knowing. Hypothyroidism can raise prolactin through TRH stimulation, and prolactin normalises when TSH does. But the study most often cited for this was about 85% female, and its male subgroups were 8 and 16 men — far too small to generate a male figure. In men specifically, prolactin is not consistently elevated except in longstanding severe primary hypothyroidism. The TRH-to-prolactin mechanism is real; it is not the usual explanation for a hypothyroid man’s low testosterone.

Evidence: Limited on magnitude, given n=10 and no controlled trial. Moderate on direction, which is consistent across sources.

What the guidelines actually say — and where they are silent

This is worth stating precisely, because it is routinely overstated in both directions.

The Endocrine Society’s 2018 guideline names no thyroid test anywhere. Thyroid disease appears only in Table 2, as a modifier of SHBG: hypothyroidism among the conditions that decrease SHBG, hyperthyroidism among those that increase it. The instruction attached to that table is to measure free testosterone when SHBG is likely altered. For secondary hypogonadism the named further tests are prolactin, iron saturation, and in some cases pituitary function testing and MRI — a TSH could arguably fall inside “pituitary function testing” when panhypopituitarism is suspected, but that is an inference, not a guideline statement.

The AUA’s 2018 guideline mentions neither thyroid nor SHBG at all. Its named endocrine adjuncts are LH, prolactin, estradiol if breast symptoms, FSH if fertility is a concern, hemoglobin and hematocrit, and PSA over 40.

The EAU guideline mentions thyroid hormone only as an exogenous drug that raises SHBG — a narrower framing than the Endocrine Society’s, since it does not name thyroid disease as a modifier.

So all three converge on the same position, and it is not the position usually attributed to them. Thyroid status matters as an SHBG modifier that changes how a total testosterone should be read. None of the three makes TSH a routine part of the testosterone workup. Anyone who tells you the guidelines require a thyroid panel alongside a testosterone panel is describing a reasonable clinical practice, not a guideline.

The reference interval is a blunt instrument, and here is the honest version of why

Three separate problems, and they compound.

The upper limit is genuinely disputed. Two papers published back to back in the same 2005 issue of the Journal of Clinical Endocrinology and Metabolism remain the canonical statements. The case for lowering the limit to 2.5: more than 95% of normal individuals sit below 2.5 mIU/L, and older reference populations were contaminated with undiagnosed thyroid disease. The case against: in the NHANES III reference group of 14,333 people with no known thyroid disease and no antibodies, an additional 9.7% — about 20.6 million Americans — sit in the 2.5 to 4.5 band, and most people who do have thyroid autoimmunity have a TSH below 2.5. Both readings use the same data. This is a genuine interpretive disagreement, not a factual one.

TSH drifts upward with age, physiologically. In people without thyroid disease, the proportion with TSH above 2.5 rises from 10.6% at ages 20 to 29 to 40% at 80 and over, and the 97.5th centile rises from 3.56 to 7.49 mIU/L across the same span. Seventy percent of older people with a TSH above 4.5 are inside their own age-specific range. This matters here more than on a general thyroid page, because the men being evaluated for testosterone deficiency are precisely the middle-aged and older men in whom the drift occurs.

Assays disagree by more than the disputed band. Before international harmonisation, the 97.5th percentile differed by 2.13 mIU/L across fourteen commercial platforms — larger than the entire 2.5-to-4.5 argument. A man could be “subclinically hypothyroid” at one laboratory and unambiguously normal at another, with identical serum.

And the fact that cuts the other way

There is a legitimate argument that a normal TSH is not necessarily normal for you, and it deserves fair presentation because it is the strongest version of the case.

Sixteen healthy men were sampled monthly for a year. Every man varied around a distinct individual set point, and the width of an individual’s 95% interval was about half that of the group. The index of individuality for TSH was 0.49 — meaning the population range is a poor instrument for judging one person. A man whose personal set point is 0.8 could double his TSH to 1.6 and still sit comfortably in the middle of “normal.” The authors’ conclusion, quoted: “a test result within laboratory reference limits is not necessarily normal for an individual.”

That is a real and under-used fact. But read what it actually licenses. The same paper found that the difference needed to be 95% confident of real change was, on average, only 0.75 mIU/L — which cuts both ways. Small changes can be meaningful, and a single value cannot be interpreted as a deviation from a set point you have never measured. It is an argument for serial measurement in the same person on the same assay. It is not an argument for treating a first-ever TSH of 3.2.

Evidence: Established.

Treating a mildly raised TSH: where the honest answer is negative

This is the question that actually arises — a man with fatigue, a TSH of 5, and a normal free T4 — and the evidence quality here is high.

TRUST randomised 737 adults aged 65 and over with persisting subclinical hypothyroidism (TSH 4.60 to 19.99, normal free T4) to levothyroxine or placebo, double-blind, with mock dose adjustment. The drug worked biochemically: TSH fell from 6.40 to 3.63 versus 5.48 on placebo. On the two co-primary symptom outcomes, against a minimum clinically important difference of 9 points, the between-group differences were 0.0 (95% CI −2.0 to 2.1) for hypothyroid symptoms and 0.4 (−2.1 to 2.9) for tiredness. No benefit on any secondary outcome.

Read the confidence intervals rather than the point estimates. TRUST did not merely fail to find a benefit; it excluded a clinically important one.

The obvious objection — but what about the men who feel genuinely terrible? — was answered by a prespecified secondary analysis. Among the participants with the highest symptom burden, the levothyroxine group improved by 12.3 points on symptoms and 8.9 on tiredness. The placebo group improved by 10.4 and 10.9. The adjusted between-group differences were −2.0 (p=0.27) and 0.0 (p=0.99), with no interaction by baseline symptom burden.

That row is the most important one on this page for this conversation. The most symptomatic patients improved substantially — above the threshold for a meaningful change — on placebo. The drug added nothing. Regression to the mean, natural fluctuation and expectation effects are large in symptom scores, which is precisely why uncontrolled before-and-after experience is so persuasive and so unreliable.

Across all adult ages, a meta-analysis of 21 randomised trials in 2,192 adults found general quality of life SMD −0.11 (95% CI −0.25 to 0.03) and thyroid-related symptoms SMD 0.01 (−0.12 to 0.14), at moderate-to-high GRADE quality. The point estimate for quality of life slightly favours placebo.

What this does not say. It does not say levothyroxine is useless. Overt hypothyroidism — a raised TSH with a low free T4 — is a treatment indication and is not in dispute anywhere in this literature. It does not address pregnancy, fertility, or a TSH above 20. And TRUST enrolled people aged 65 and over; no trial reports a young-male subgroup.

Evidence: Strong (null result) for treating subclinical hypothyroidism to improve fatigue or quality of life.

The practical algorithm

If a man’s total testosterone is unexpectedly high for his clinical picture, thyrotoxicosis belongs on the list alongside the commoner SHBG-raising states — age, HIV, cirrhosis or hepatitis, anticonvulsants, exogenous estrogen.

If a man’s total testosterone is low with LH and FSH that are not elevated, untreated hypothyroidism belongs on the differential of reversible causes, alongside obesity, opioids, glucocorticoids, sleep apnea and acute illness.

Whenever SHBG is likely to be abnormal, measure free or bioavailable testosterone rather than relying on total. This is the one point every major guideline does support.

And correct the thyroid disease first, then re-measure. One group recommended postponing specific treatment for erectile dysfunction until at least six months after euthyroidism is achieved; the same logic applies with considerably more force before committing a man to lifelong testosterone therapy.

Related

Total testosterone · SHBG · Free testosterone · Prolactin · When testosterone isn’t the answer · The Lab Library

Questions patients ask