Thyroid function tests are among the most ordered assays in any laboratory, and among the most likely to produce a result that does not match the patient in front of you.
When that happens, the answer is usually not an unusual thyroid disorder. It is one of a small number of interferences, and they are worth knowing by name.
## Read TSH and free T4 together
TSH is exquisitely sensitive. A small fall in thyroid hormone produces a large rise in TSH, which is why TSH alone is a good screening test in a stable outpatient.
But TSH alone is a poor test in several important situations, and reading it with free T4 is what protects you.
The classic patterns first. High TSH with low free T4 is primary hypothyroidism. Low TSH with high free T4 is primary hyperthyroidism. High TSH with normal free T4 is subclinical hypothyroidism, and low TSH with normal free T4 is subclinical hyperthyroidism.
The pattern that should stop you is **low or normal TSH with low free T4**. That is not primary thyroid disease. It suggests either central hypothyroidism — a pituitary or hypothalamic problem — or non-thyroidal illness. A TSH-only screening strategy misses it entirely.
## Non-thyroidal illness
Any significantly unwell patient develops abnormal thyroid results. T3 falls first, then free T4, and TSH can go in either direction. It is a physiological response rather than thyroid disease, and it resolves with recovery.
The practical implication is straightforward and frequently ignored: **do not test thyroid function in an acutely ill inpatient unless you specifically suspect thyroid disease as the cause of the illness.** You will get abnormal results that mean nothing and generate treatment that helps nobody. Wait until recovery.
## Interferences worth knowing
**Biotin.** High-dose biotin supplements — sold widely for hair and nails — interfere with many immunoassays because the assays themselves use biotin-streptavidin chemistry. The result is characteristic: a low TSH with high free T4, mimicking hyperthyroidism convincingly in a euthyroid person. Ask about supplements, and advise stopping biotin for a few days before retesting. This one has led to patients being treated for thyrotoxicosis they did not have.
**Heterophile antibodies**, including human anti-mouse antibodies, interfere unpredictably and produce results that do not fit any physiological pattern. If a result is internally inconsistent, ask the laboratory to repeat it on a different platform or after a blocking step.
**Macro-TSH**, where TSH is bound in a large complex, produces a persistently raised TSH in a patient who is entirely well and has a normal free T4.
**Drugs.** Amiodarone can cause either hypo- or hyperthyroidism. Steroids and dopamine suppress TSH. Oestrogen raises thyroid binding globulin, which affects total but not free hormone measurements — one reason free T4 is preferred over total T4.
## Pregnancy
Trimester-specific reference intervals are needed and frequently not used. hCG has weak TSH-like activity, so TSH falls physiologically in the first trimester, and applying a standard adult range to a pregnant woman will suggest hyperthyroidism where none exists.
If you report thyroid function on pregnant patients, get the trimester ranges onto the report.
## For the laboratory
Report TSH and free T4 together as a panel rather than reflexing free T4 only when TSH is abnormal — the reflex approach saves money and misses central hypothyroidism.
State the platform on the report, because thyroid assays are not standardised across manufacturers and a change of laboratory produces apparent changes that are methodological. Immunoassay systems such as the Beckman Coulter Access 2 are widely used; the point is that serial results should stay on one of them.
And when a clinician rings about a result that does not fit, the first two questions are worth having ready: is the patient acutely unwell, and are they taking biotin.