Borderline thyroid results: what they mean and when treatment helps

How thyroid blood tests work

The pituitary gland makes TSH, which tells the thyroid to work. When the thyroid underperforms, TSH rises to push it harder, so a raised TSH is usually the first sign of an underactive thyroid. Free T4 is the main hormone the thyroid makes. Testing free T3 is mostly useful when TSH is low, to look for an overactive thyroid.

What "borderline" means

A raised TSH with a normal free T4 is called subclinical hypothyroidism. It is common, especially in women and in older people, and TSH tends to drift up with age, so the usual reference range fits older people less well. It is often temporary: TSH can rise during recovery from an illness and settle without treatment, which is why NICE advises repeating the test before deciding anything.

Do antibodies matter?

Thyroid peroxidase (TPO) antibodies show that the thyroid is under autoimmune attack, the usual cause of an underactive thyroid in the UK. In the Whickham study, which followed a County Durham population for 20 years, women with both a raised TSH and positive antibodies were far more likely to develop an underactive thyroid than women with neither. Antibodies help predict the future; on their own they are not a reason to start treatment. NICE suggests measuring TPO antibodies once when the TSH is raised; repeating them adds nothing. Thyroglobulin antibodies are mainly used in thyroid cancer follow-up and don't help with this question.

When treatment helps

NICE gives a broad framework. When TSH is 10 or more on two tests three months apart, treatment is usually considered. Below 10, a six-month trial of levothyroxine can be considered for adults under 65 who have symptoms, stopped if it doesn't help. Otherwise the result is monitored, more often if there are signs of underlying thyroid disease such as positive antibodies.

The trials explain why the decision needs thought. Across 21 trials involving more than 2,000 adults with subclinical hypothyroidism, levothyroxine made no meaningful difference to quality of life or tiredness. In the TRUST trial, adults aged 65 and over who took levothyroxine for about a year felt no better than those on placebo. Some individuals do feel better, and a time-limited trial with a clear review point is a reasonable way to find out. Over-treatment carries risks of its own, including an irregular heart rhythm and bone loss.

Pregnancy, and planning a pregnancy, are different: the thresholds are lower and specialist advice is worth having early.

When the tests are normal but you feel unwell

Tiredness, weight gain, low mood, poor concentration and thinning hair are common, and they are real. When TSH and free T4 are both normal, the thyroid is very unlikely to be the cause. In a trial of people with these symptoms and normal thyroid tests, levothyroxine was no better than placebo.

That does not end the enquiry. When I assess this, I look for the uncommon situations in which normal-looking thyroid tests mislead, such as a pituitary problem or interference with the test, and for other causes. Worth considering: low iron stores or anaemia, low B12, coeliac disease, diabetes, sleep apnoea, low mood, perimenopause and side effects of medicines. Low vitamin D is often tested, though its link with tiredness is weaker than is usually claimed. Rarely, low cortisol is the cause.

Tests and treatments to be wary of

Some online services offer reverse T3 measurements, body temperature charts or day-long saliva cortisol profiles, with diagnoses such as "Wilson's syndrome" or "adrenal fatigue". None of these has been shown to identify a real condition, and the treatments sold for them, including high doses of thyroid hormone and animal gland extracts, can cause harm. "Thyroid support" supplements have been found to contain thyroid hormone that wasn't on the label. If you take anything like this, please bring it to the appointment: it can change your results.

If you are already on levothyroxine and still unwell

This is covered in more detail on the thyroid disorders page. In short, the diagnosis, the absorption of the tablets and other possible causes are usually looked at first, because fine-tuning the dose within the normal range has not been shown to help, and European guidance advises against it. The place of T3 in this situation has not been fully established, and so far it has not been shown to be better than levothyroxine alone for most people.

Genetic "poor converter" tests (DIO2)

Some people have had a genetic test for a common variant in the DIO2 gene, which helps turn T4 into T3 inside cells, and have been told they are "poor converters". Between one in ten and one in six people carry two copies of this variant. Several studies, including large UK and Norwegian studies, have linked it to slightly lower well-being in people taking levothyroxine, and one trial suggested carriers might feel better on combination treatment. Other studies have not found this, and no trial has yet tested whether choosing treatment by genotype helps. Neither UK nor European guidance recommends the test. I do not offer it, and I will not use it to decide treatment until there is good evidence that doing so improves how people feel. If you have had the test, please bring the result and I am happy to discuss it: it is useful background, and your symptoms are assessed in the same careful way whatever it shows.

Results that can mislead

Biotin, sold in hair, skin and nail supplements, interferes with some thyroid tests and can produce results that look like an overactive thyroid. Tell whoever takes the blood if you take it; labs usually advise stopping it for a few days beforehand. A test taken during or just after an illness may also be temporarily abnormal.

When a specialist opinion helps

When results and symptoms don't fit, when you are pregnant or planning to be, when you remain unwell on treatment, or when you would like a considered decision about whether to treat at all. In clinic I go through your results as a series, look for other causes, and agree a plan with you, including when to recheck.