Thyroid function tests and hypothyroidism

BMJ 2003; 326 doi: https://doi.org/10.1136/bmj.326.7384.295 (Published 08 February 2003) Cite this as: BMJ 2003;326:295

Measurement of serum TSH alone may not always reflect thyroid status

  1. Anthony D Toft (anthony.toft{at}luht.scot.nhs.uk), consultant physician,
  2. Geoffrey J Beckett, reader
  1. Endocrine Clinic, Royal Infirmary, Edinburgh EH3 9YW
  2. University Department of Clinical Biochemistry, Royal Infirmary

    Paper p 311

    It is extraordinary that more than 100 years since the first description of the treatment of hypothyroidism and the current availability of refined diagnostic tests, debate is continuing about its diagnosis and management. Symptoms of thyroid failure are often non-specific, such as weight gain, low mood, and fatigue. Some patients seeking an explanation for feeling “below par” are disappointed when thyroid function tests are normal. Unable to accept that there may be psychosocial reasons for their symptoms, a vociferous minority believe that hypothyroidism may exist with normal serum concentrations of both thyroxine (T4) and thyroid stimulating hormone (TSH).

    Their hypothesis is that a doctor cannot know whether a concentration of free T4 or TSH within wide reference ranges is normal for that individual. Such an argument, supported by some misguided medical practitioners to justify prescribing various combinations of thyroid hormones, does not appreciate the sensitivity of the pituitary thyrotroph, which modifies the synthesis and secretion of TSH in response to minor changes in thyroid hormone concentrations within their reference ranges. For example, a reduction in free T4 from 20 pmol/l to 15 pmol/l is likely to cause a rise in serum TSH to above the upper limit of the reference range, and a similar incremental rise in free T4 to suppress thyrotroph secretion, with a resultant serum TSH concentration of less than 0.05 mU/l.1 In effect, any significant deviation from the set point for serum thyroid hormone concentrations, which is remarkably constant from day to day in healthy people, will trigger changes in serum TSH.

    The finding of raised or undetectable serum TSH with thyroid hormone concentrations within their reference ranges is not usually associated with symptoms, hence the basis for the unsatisfactory terms subclinical hypothyroidism and hyperthyroidism. It is better to consider them as the mildest forms of thyroid failure and thyrotoxicosis, respectively, particularly as a variable proportion of patients with subclinical hypothyroidism benefit from replacement therapy with thyroxine,2 and endogenous subclinical hyperthyroidism is a recognised risk factor for atrial fibrillation and osteoporosis.3

    In contrast, patients with non-specific symptoms of hypothyroidism and unequivocally normal T4 and TSH concentrations do not benefit from treatment with thyroxine.4 In the paper by Meier et al in this issue we are reminded that in severe primary hypothyroidism with serum TSH greater than 20 mU/l the correlation between TSH concentrations and other end organ responses to a low serum T4 is poor (p 311).5 This must not be interpreted as thyrotroph insensitivity but exhaustion after prolonged stimulation6; an analogous apparent loss of sensitivity occurs after treatment of hyperthyroidism as the suppressed thyrotroph requires several weeks to recover its responsiveness to falling serum thyroid hormone concentrations.

    It is the exquisite sensitivity of the thyrotroph that led to the use of serum TSH measurements as a first line test of thyroid function; a normal TSH indicated euthyroidism whereas only a raised or suppressed concentration prompted the measurement of T3 or T4 or both, to assess the degree of hypothyroidism or hyperthyroidism.7 This approach has been strongly championed by some laboratories to contain costs, but they may provide misleading information. For example, a normal TSH may be recorded in patients with profound hypothyroidism secondary to pituitary or hypothalamic disease,8 a remediable condition that may have serious consequences if not recognised; and rarely hyperthyroidism may be associated with a normal TSH due to pituitary tumour, thyroid hormone resistance, or assay interference.9

    There is also the difficulty of interpreting a serum concentration of TSH in isolation. A concentration at or near the upper limit of the reference range, particularly if associated with a normal free T4, may indicate underlying autoimmune thyroid disease. A consensus exists for early treatment of such patients with thyroxine if anti-thyroid peroxidase antibodies are present in the serum, not because any immediate benefit may be expected but because the risk of overt thyroid failure in future years is high,10 and it makes sense to anticipate morbidity rather than risk loss to follow up.

    The other difficulty in interpreting serum TSH concentrations is to decide what value should be aimed for in patients taking thyroxine replacement. It is not sufficient to satisfy the recommendations of the American Thyroid Association11 by simply restoring both serum T4 and TSH concentrations to normal, as in our experience most patients feel well only with a dose resulting in a high normal free T4 and low normal TSH concentration, and those patients with continuing symptoms despite “adequate” doses of thyroxine12 may be slightly under-replaced. Some patients achieve a sense of wellbeing only if free T4 is slightly elevated and TSH low or undetectable.13 The evidence that this exogenous form of subclinical hyperthyroidism is harmful is lacking in comparison to the endogenous variety associated with nodular goitre,3 and it is not unreasonable to allow these patients to take a higher dose if T3 is unequivocally normal.

    Although the potential improvement in the wellbeing of patients with hypothyroidism while taking a combination of T3 and T4 is of great interest,14 the greatest advantage will be the security of a normal TSH while taking physiological replacement, removing the anxiety about whether a little too much thyroxine alone is harmful. Of course, we are perhaps naive in thinking that patients with autoimmune thyroid disease who continue to complain of non-specific symptoms despite restoration to normal of TSH and T4 concentrations can be improved by tinkering with the dose and form of thyroid hormone used for treatment. It is just possible that these symptoms arise from the chronic inflammatory basis of the underlying thyroid disease, but that story is largely unwritten.


    • Competing interests None declared.


    1. 1.
    2. 2.
    3. 3.
    4. 4.
    5. 5.
    6. 6.
    7. 7.
    8. 8.
    9. 9.
    10. 10.
    11. 11.
    12. 12.
    13. 13.
    14. 14.
    View Abstract

    Log in

    Log in through your institution


    * For online subscription