The TSH Range Debate: Why Optimal Differs from Reference

A patient with fatigue, cold hands, thinning hair and a foggy head gets a TSH of 3.8 and is told her thyroid is normal. Somebody with an identical result and no symptoms is told the same thing. Both statements are true against the reference range, and neither is a useful answer — because that range was never designed to identify who is doing well. It describes what is common in a population containing a great many people with undiagnosed thyroid disease, and understanding how it got so wide explains most of the argument about where the upper limit belongs.

What TSH is actually measuring

The first thing to be clear about is that TSH is not a thyroid hormone. It is a pituitary hormone, and it measures the pituitary's opinion of how much thyroid hormone is reaching it. Low thyroid output means more TSH is sent to push the gland harder; adequate output means less. It is a feedback signal, read backwards.

One property of that signal makes it both powerful and easy to misread: the relationship between TSH and circulating thyroid hormone is logarithmic rather than proportional, so a small fall in free T4 produces a disproportionately large rise in TSH. That amplification is why TSH is the most sensitive single screening test, and why it moves before free T4 leaves its own range. It is also why TSH exaggerates — a value that has climbed noticeably may reflect a modest change in actual hormone availability. The full thyroid axis covers how the signal is generated. TSH tells you what the pituitary thinks, and the pituitary can be misled by illness, medication, an unusual set point, or the time of day the sample was drawn.

Reference against optimal

• Reference range: typically 0.5-5.0 mIU/L
• Optimal range as commonly proposed: 0.5-2.5 mIU/L
• Patients sitting between 2.5 and 5.0 may represent early hypothyroidism

How the range got so wide

Reference ranges are built by measuring a large sample of apparently healthy people and taking the central portion of the distribution. That works when the condition being screened for is rare, and badly when it is common — because the "healthy" reference population is quietly seeded with people who have the disease.

Autoimmune thyroid disease is common, and much of it is undiagnosed. When those people were included in the reference population, they pulled the upper limit up with them. As the prevalence of thyroid autoimmunity became clearer through population survey work, and as reference cohorts were re-derived after excluding people with positive thyroid antibodies, the upper limit came down (Hollowell et al., J Clin Endocrinol Metab 2002). This is the mechanical heart of the whole debate: the number at the top of the range is partly a measurement of how much undetected thyroid disease was in the room.

The AACE recommendation, and why it did not stick

The American Association of Clinical Endocrinologists recommended tightening the upper limit toward 3.0 mIU/L on the basis of that improved understanding (Garber et al., Endocrine Practice 2012). Implementation has been inconsistent, and different laboratories still report different limits — which means the same blood sample can be flagged abnormal by one laboratory and normal by another.

The resistance was not simply inertia. Narrowing the range reclassifies a large number of people as abnormal overnight, most of whom will never develop overt hypothyroidism and some of whom would be treated unnecessarily. That objection deserves stating plainly rather than dismissing.

Three complications the tightening argument usually skips

Anyone advocating a hard 2.5 cut-off should account for these, because they are well established and they cut against a single universal number.

TSH rises with age, normally. The distribution shifts upward across the decades even in people without thyroid antibodies, which means a value that is genuinely high for a thirty-year-old can be entirely unremarkable for someone in their eighties (Surks & Hollowell, J Clin Endocrinol Metab 2007). Applying a young adult's target to an older patient converts normal ageing into a diagnosis.

Each person has a narrow individual set point inside the wide population range. Repeated sampling in the same healthy individual produces a far tighter band than the population range implies (Andersen et al., J Clin Endocrinol Metab 2002). This is the most useful fact in the debate and the least often mentioned: a TSH that has moved from 1.1 to 2.9 over several years is a real change in that person even though every value was "normal," while someone else may have sat at 3.5 their whole adult life at no cost. A trend beats a threshold.

TSH is not a stable number. It follows a diurnal rhythm, peaking overnight and falling through the morning, so draw time alone can move a result meaningfully. Acute illness, recent weight loss, and a number of medications all shift it. A single value taken on a random afternoon is a snapshot of a moving quantity.

What upper-normal TSH usually means

TSH between 2.5 and 5.0 with a normal free T4 covers a genuinely mixed population, and separating them is the actual clinical work:

• Early autoimmune thyroid disease. The pituitary is compensating for a gland that is slowly losing capacity. Antibody testing is frequently positive, symptoms may be present, and this is the group that progresses. Thyroid antibodies is the test that identifies them.
• A person's normal set point. No antibodies, no symptoms, no trend — and no problem.
• Something else entirely. Recovering illness, a medication effect, iodine excess or deficiency, or simply the time of day.

Three pieces of information sort those apart, and none of them is another TSH in isolation: antibody status, the direction the value has moved over previous years, and whether symptoms are actually present and consistent.

When to look further

For a patient with an upper-normal TSH and symptoms that fit, a fuller evaluation is reasonable:

• Free T4 and free T3 — since the two are not interchangeable and TSH describes neither directly
• Reverse T3, particularly where chronic stress or illness is in the picture, as in reverse T3 and chronic stress
• Thyroid antibodies, TPO and thyroglobulin both
• A structured symptom assessment rather than an impression
• Repeat testing at a consistent time of day before drawing conclusions from any single value
• A discussion of a treatment trial in a symptomatic, antibody-positive patient

The half that gets skipped: iron, ferritin, B12 and vitamin D all produce fatigue and cognitive fog that reads exactly like hypothyroidism. Attributing those symptoms to a TSH of 3.4 without checking is how people end up on thyroid medication that never helps.

What treatment does, honestly

Two things pull in opposite directions here. Subclinical hypothyroidism has been associated with increased coronary heart disease risk in pooled cohort analysis, particularly at higher TSH values (Rodondi et al., JAMA 2010) — which is the case for not dismissing it. But the treatment evidence is less flattering than enthusiasm suggests. Randomised treatment of older adults with subclinical hypothyroidism has generally not produced meaningful improvement in symptoms or quality of life (Stott et al., N Engl J Med 2017). That does not settle the question for a symptomatic forty-year-old with positive antibodies and a rising trend — a different patient entirely — but it means the argument has to be made rather than assumed.

Where a trial is undertaken, set expectations first: which symptom is being targeted, over what period, and what would count as failure. Energy and cognition, if they respond, do so over weeks. Committing to a lifelong medication on a vague sense of improvement is worse than declining to start.

Bottom line

Reference TSH ranges are wider than optimal because they were derived from populations containing undiagnosed thyroid disease, which is why an upper-normal value may represent early hypothyroidism rather than health. But a single tight cut-off is not the answer either: TSH rises with age normally, each person has a narrow set point inside the wide population range, and the value swings with time of day and intercurrent illness. Compare a person against their own previous values, check antibodies, look at free T4 and free T3 rather than TSH alone, and rule out the non-thyroid causes of the same symptoms first. A trend with antibodies and symptoms behind it means something; a number in isolation does not.

Educational content, not medical advice. Laboratory interpretation and any treatment decision are made by a licensed physician after individual evaluation. Individual results vary.