Elevated TSH with Normal FT4: Should You Treat Subclinical Hypothyroidism?

Subclinical hypothyroidism is a laboratory state rather than a disease diagnosis in the classical sense: the concentration of TSH (the pituitary hormone that stimulates the thyroid) exceeds the upper limit of the reference range while free thyroxine (FT4) is within the normal range. When FT4 is low, we speak of overt hypothyroidism. So the word “subclinical” does not mean “without symptoms”, only “without a low hormone level in the blood” — some people report fatigue, cold intolerance or dry skin, but these symptoms are non-specific and also occur in people with normal TSH.

The 2013 guideline of the European Thyroid Association (ETA) divides this condition into two categories: mildly raised TSH (about 4.0–10.0 mU/L) and clearly raised TSH (above 10 mU/L). The distinction matters because the two groups differ in risk and in the rationale for treatment. The vast majority of cases fall into the first category — people with TSH around 4.5–7 mIU/L, in whom the shift is small and variable over time.

The mechanism is simple in outline: the pituitary responds to even a small drop in thyroid hormone production with a strong rise in TSH secretion (the relationship is log-linear, so a small change in FT4 can mean a several-fold change in TSH). This makes TSH a sensitive early signal, but also one prone to fluctuation that does not always mean permanent damage to the thyroid. In countries with adequate iodine intake, the most common cause of a persistently raised TSH is autoimmune Hashimoto’s thyroiditis, which has its own entry and article on our site.

The most important practical rule is very simple: a single raised TSH is not a diagnosis. The ETA guideline recommends that an initially raised TSH with a normal FT4 be verified by repeat measurement of TSH and FT4, together with anti-TPO antibodies, preferably after 2–3 months. The reason is that TSH shows considerable biological variability (it depends on time of day, sleep, acute illness and nutritional state), and during recovery from an infection or a transient thyroiditis it can be temporarily elevated.

The practical conclusion: if the first result is only slightly raised and the patient has no severe symptoms, a reasonable strategy is to repeat the test after 2–3 months under comparable conditions (in the morning, at a similar time of day, with no change in supplements — especially biotin, whose effect on thyroid results we describe separately in our article on biotin and the TSH result). Only a persistently raised TSH on two measurements provides a basis for further decisions.

TRUST remains the largest and most reliable piece of evidence on this topic, because it had a placebo group, double blinding and measured what matters most to the patient: symptoms and tiredness, not just the number on the printout. The authors conclude that levothyroxine provided no apparent benefit in older people with subclinical hypothyroidism. It is worth noting that the drug did lower TSH, so this was not ineffective dosing — it was a lack of clinical benefit despite biochemical improvement.

An honest presentation of the results also requires saying what they do not cover. TRUST enrolled only people aged at least 65, with a mean age of 74; mean baseline TSH was 6.4 mIU/L, so the trial mainly concerned mild elevation, and conclusions for TSH above 10 are limited. The result does not automatically transfer to a thirty-year-old woman with TSH 9, to someone with TSH 15 or to a woman planning a pregnancy. The 2018 meta-analysis, for its part, excluded pregnant women, and the trials it included were heterogeneous (I-squared for quality of life about 67%).

The second important limit concerns the endpoints. TRUST and the meta-analysis assessed symptoms and quality of life after a year or less. They do not answer whether treatment over many years reduces cardiovascular risk, slows progression to overt hypothyroidism or prevents goitre growth. These questions remain largely open, and some observational data suggest that the level of TSH matters — which we discuss later in this article.

Finally, the trial does not mean that levothyroxine will help no patient. A zero average effect in a group does not rule out a small subgroup benefiting — but the results suggest that we cannot identify that subgroup in advance from TSH alone. That is why the ETA guideline allows, in younger symptomatic patients, a so-called therapeutic trial with a clearly defined time for review, as described in the section on practical treatment rules.

The standard upper limit of TSH in many laboratories is around 4.0–4.5 mIU/L and is applied regardless of age. Yet the distribution of TSH in the population shifts towards higher values with age — and this happens also in people without thyroid disease and without antibodies. This means that a result “out of range” at the age of 80 may be a physiological phenomenon rather than a disease.

The ETA guideline points in the same direction: it recommends considering age-specific TSH ranges when diagnosing subclinical hypothyroidism in older people, and in the oldest (over 80–85 years) with TSH up to 10 mU/L it suggests a watchful-waiting strategy and, in general, avoiding hormonal treatment. For younger adults the situation is the opposite: the lower part of the reference range is the norm for their age, so a moderately raised TSH more often than in seniors reflects a genuine process in the thyroid.

Antibodies against thyroid peroxidase (anti-TPO) are the single most important factor predicting whether a mildly raised TSH will persist and whether overt hypothyroidism will develop in the future. In the van der Spoel analysis, absence of anti-TPO was one of the factors favouring spontaneous normalisation. Positive antibodies are, however, not an automatic indication for treatment — rather, they say that results should be monitored, less often but consistently. We discuss the meaning of the antibodies themselves, including in people with normal TSH, in a separate article on raised anti-TPO.

The conclusion cuts both ways. On the one hand, at TSH 4.5–6.9 mIU/L there is no increased risk of coronary events, which fits well with the lack of benefit in treatment trials. On the other hand, at TSH above 10 mIU/L the risk rises, and this is also the range in which progression to overt hypothyroidism is most common. That is why the 10 mIU/L threshold appears in guidelines as a turning point: ETA recommends levothyroxine even without symptoms in younger patients (below about 65–70 years) with TSH above 10 mU/L. It should be repeated, however, that the evidence for benefit of treatment in this group comes mainly from observational data and expert opinion rather than large randomised trials.

Everything written above about the lack of benefit from treatment applies to non-pregnant adults. In pregnancy the mother’s thyroid supplies the foetus with hormones in the first trimester, and the demand for thyroid hormones rises. That is why a raised TSH in a pregnant woman or one trying to conceive is assessed differently: against pregnancy-specific and laboratory-specific ranges, often taking the presence of anti-TPO into account.

The 2017 guidelines of the American Thyroid Association (ATA) contain detailed recommendations on interpreting and treating abnormal TSH in pregnancy, and an updated version was published in 2026, incorporating new data from recent years — although the authors stress that much of the evidence remains of low to moderate quality. Specific thresholds and regimens change, so a woman who is pregnant or planning pregnancy and has a raised TSH should discuss the result with the endocrinologist or obstetrician in charge of her care rather than apply the general rules from this article. Delaying a consultation in this situation is not a good idea.

If treatment is started after all, according to the ETA guideline TSH is checked about 2 months after starting and after any dose adjustment, with the aim of a stable TSH in the lower half of the reference range (about 0.4–2.5 mU/L), and then monitored at least once a year. In older people the dose is usually started lower (in TRUST 25–50 µg) to avoid overdosing and the associated risks to the heart and bones. How to take the drug — with regard to coffee, breakfast and supplements — is covered separately in our article on levothyroxine.

A raised TSH with a normal FT4 is rarely an emergency, but a few situations call for a quicker consultation than the usual check in a few months.

Subclinical hypothyroidism is an example of a topic in which medicine has moved from “treat the result” to “treat the patient”. The best available data do not say the result is unimportant — they say that a number above the range alone is rarely sufficient reason to begin lifelong therapy. What matters more is whether the result is reproducible, how high it is, the patient’s age, the antibodies and pregnancy plans. For the patient this means a calmer approach to the first result, but also consistency in follow-up — because ignoring a TSH above 10, or a raised TSH in pregnancy, would be a mistake in the opposite direction.