Letters

Thyroid function tests

Tests must still be done in possible thyroid dysfunction

Thyroid stimulating hormone outside the normal range has important implications

Accurate diagnosis depends on both clinical judgment and results of tests

Doing more tests is not always better

Thyroid function testing means different things to different people

Author's reply

Tests must still be done in possible thyroid dysfunction

EDITORThe article by O'Reilly on reassessment of thyroid function tests raises some important questions but is misleading in several respects.¹ Clinical features must of course be given full consideration in the assessment of possible thyroid dysfunction, but appropriate tests must still be done.

The symptoms of both hyperthyroidism and hypothyroidism are non-specific and can be mimicked by other conditions. Thus the practice of prescribing thyroid treatment on a clinical basis alone without biochemical confirmation carries potential risks. The statement that "the clinical features of hypothyroidism . . . have been relegated to the status of historical curiosities" is absurd. What the doctor aims to do is not simply to categorise a patient into hypothyroidism, hyperthyroidism, or the subclinical variants but rather to make a full diagnostic assessment, of which thyroid function tests are one important facet. Surprisingly, O'Reilly makes no mention of autoantibody tests, which are also helpful in assessing thyroid disease.

With regard to hyperthyroidism, a reduced thyroid stimulating hormone concentration is not in fact diagnostic. Clinical assessment is imperative, and before thyrotoxicosis is diagnosed the thyroxine (and in some cases triiodothyronine) concentration should be checked. The practice of using results of thyroid stimulating hormone tests alone to indicate hyperthyroidism is to be deplored and has led to a mistaken diagnosis in several cases subsequently shown to be cases of hypopituitarism.

O'Reilly mentions the use of thyroid stimulating hormone for screening purposes; the figures quoted for misleading results in the general population are interesting but date from 10 or more years ago. Thyroid stimulating hormone assays have considerably improved since then, and thus these numbers may not now be relevant.

O'Reilly is probably correct in claiming that too many indiscriminate requests for thyroid stimulating hormone tests are made. In some situations, however, notably in pregnancy, thyroid tests are not performed frequently enough. Recent studies have shown that raised maternal thyroid stimulating hormone concentrations or low thyroxine concentrations, or both, in early pregnancy are associated with impaired neuropsychological development of the child. ^{2 3} There should be greater awareness of this and of the possibility of hypothyroidism in early pregnancy. All women known to be hypothyroid should be advised to increase their dose of thyroxine as soon as pregnancy is diagnosed, and the adequacy of the dose should be monitored by measurement of thyroid stimulating hormone concentration.

In conclusion, thyroid stimulating hormone assays are not infallible and must always be interpreted in the light of clinical features, effects of drug treatment, thyroxine concentrations, and antibody status.

P Kendall-Taylor, president, British Thyroid Association. 
Department of Endocrinology, University of Newcastle on Tyne, Newcastle upon Tyne NE2 4HH Pat.Kendall-Taylor{at}ncl.ac.uk

1.	O'Reilly DStJ. Thyroid function testsa time for reassessment. BMJ 2000; 320: 1332-1334[Free Full Text]. (13 May.)
2.	Pop VJ, Kuijpens JL, van Baar AL, Verkerk G, van Son MM, de Vijlder JJ, et al. Low maternal free thyroxine concentrations during early pregnancy are associated with impaired psychomotor development in infancy. Clin Endocr 1999; 50: 149-155.
3.	Haddow JE, Palomaki GE, Allan WC, Williams JR, Knight GJ, Gagnon J, et al. Maternal thyroid deficiency during pregnancy and subsequent neuropsychological development of the child. N Engl J Med 1999; 341: 549-555[Abstract/Free Full Text].

Thyroid stimulating hormone outside the normal range has important implications

EDITORAlthough O'Reilly suggests that the reference interval for thyroid stimulating hormone could be extended to 21.5 mU/l,¹ this assertion is unjustified by its numerous published reference intervals. As an example we quote our recent experience measuring thyroid stimulating hormone with the Roche Elecsys 2010 analyser in excess serum received in the laboratory for other investigations on 324 patients undergoing either elective surgery or cholesterol screening. Patients taking thyroxine were excluded. We found results ranged from <0.005 to >100 mU/l, but the central 95% portion was 0.5 to 5.8 mU/l. False positive and negative results are possible with any test but can be minimised if the confidence intervals for the limits of the reference interval are calculated, which for our population are 0.22-0.61 and 5.2-6.3 for the lower and upper limit respectively.

The concept of subclinical hypothyroidism is based on the log-linear feedback loop between thyroxine and thyroid stimulating hormone: for one unit change in thyroxine there is a 10 unit change in thyroid stimulating hormone. The prediction of disease by the measurement of an intermediate marker is now well established in the absence of clinical symptomsfor example, calcium concentration, cholesterol concentration, and blood pressure. Since the review by one of us (APW) that seems to have provoked some of the statements made by O'Reilly it has been shown that changes in endothelial function and cholesterol concentration are apparent in subclinical hypothyroidism and even in people whose thyroid stimulating hormone is greater than 2 mU/l. ^{3 4} This provides a physiological basis for the Rotterdam study, which clearly shows that subclinical hypothyroidism is associated with an increased risk of ischaemic heart disease.⁵

O'Reilly is also concerned that non-specialists cannot understand from the review² the implications of a high normal thyroid stimulating hormone (>2 mU/l) compared with one which is outside the reference interval. We trust the BMJ readership more. The fact remains that a value outside the reference interval is not simply a minor variation but is important both in terms of predicting future hypothyroidism and in causing biological effects. ^{3 4}

Finally, the danger we perceive in O'Reilly's article is that it may encourage the mistaken belief that hypothyroidism can be diagnosed clinically. At least let us be clear that symptoms and signs are inadequate for diagnostic purposes and thyroxine is not indicated unless hypothyroidism (clinical or subclinical) is confirmed biochemically.

Alun Price, chief medical laboratory scientific officer, clinical chemistry. 
A P Weetman, honorary consultant physician. 
Northern General Hospital, Sheffield S5 7AU

1.	O'Reilly DStJ. Thyroid function testsa time for reassessment. BMJ 2000; 320: 1332-1334. (13 May.)
2.	Weetman AP. Hypothyroidism: screening and subclinical disease. BMJ 1997; 314: 1175-1178[Free Full Text].
3.	Hak EA, Pols HAP, Visser T J, Drexhage HA, Hofman W, Jacqueline CM. Subclinical hypothyroidism is an independent risk factor for atherosclerosis and myocardial infarction in elderly women: the Rotterdam study. Ann Intern Med 2000; 132: 270-278[Abstract/Free Full Text].
4.	Lekakis J, Papamichael C, Alevizaki M, Piperingos G, Marafelia P, Mantzos J, et al. Flow-mediated, endothelium-dependent vasodilatation is impaired in subjects with hypothyroidism, borderline hypothyroidsm and high-normal serum thyrotropin (THS) values. Thyroid 1997;7(3).
5.	Michalopoulou G, Alevizaki M, Piperingos G, Mitsibounas D, Mantzos E, Adamopoulos P, et al. High serum cholesterol levels in persons with 'high-normal' TSH levels: should one extend the definition of subclinical hypothyroidism? Eur J Endocrinol 1998; 138: 141-145[Abstract].

Accurate diagnosis depends on both clinical judgment and results of tests

EDITORO'Reilly is correct in highlighting the difficulties in interpreting the results of thyroid function tests but overstates his view that the clinical aspects of thyroid disease have been downgraded.¹ Furthermore, he misrepresents the data presented in two of our papers. We did not consider abnormal thyroid stimulating hormone concentrations in isolation when arriving at a diagnosis of subclinical hyperthyroidism. The patients with suppressed serum thyroid stimulating hormone and normal thyroid hormone concentrations on whom we reported also had clinical evidence of thyroid disease.² The finding of an abnormal serum thyroid stimulating hormone in some patients taking thyroxine is an indication for adjustment of the dose; we did not imply that these patients were treated with an optimal dose of thyroxine.³

The sensitivity of pituitary thyrotrophs to minor changes in thyroid hormone concentrations is such that biochemical evidence of thyroid disease will be apparent before clinical features develop. Thus a patient with an incidental finding of a serum thyroid stimulating hormone concentration of 7.0 mU/l (normal <5.0 mU/l) is unlikely to be clinically hypothyroid but may well have a goitre of Hashimoto's thyroiditis on examination and antibodies in the serum directed against thyroid peroxidase. Treatment with thyroxine would be an acknowledgment not of hypothyroidism but of the well recognised progression of thyroid failure in future years, as shown by the second Whickham surveyin other words, prevention is better than cure.

The finding of a suppressed serum thyroid stimulating hormone concentration, with or without a raised free thyroxine concentration, would perhaps indicate examination for the presence of a nodular goitre. There is then the potential for isotope scanning and measurement of thyroid stimulating hormone receptor antibodies to determine whether thyroid disease is present.

Some doctors argue that clinical judgment is more sensitive than biochemical tests of thyroid function and justify the use of thyroxine to treat non-specific symptoms in patients with normal biochemistry. The only controlled trial of thyroxine in such patients, however, showed no benefit,⁴ which is good evidence for the robustness of the currently available tests.

Nothing is to be gained by those who advocate the primacy of thyroid function tests or of clinical examination in the diagnosis of thyroid disease. The correct diagnosis will usually be made on the basis of both.

A D Toft, consultant physician, endocrine unit. 
G J Beckett, reader in clinical biochemistry. 
G.J.Beckett{at}ed.ac.uk Royal Infirmary of Edinburgh NHS Trust, Edinburgh EH3 9YW

1.	O'Reilly DStJ. Thyroid function testsa time for reassessment. BMJ 2000; 320: 1332-1334. (13 May.)
2.	Seth J, Kellet HA, Caldwell Sweeting VM, Beckett GJ, Gow SM, et al. A sensitive immunoradiometric assay for serum thyroid stimulating hormone: a replacement for thyrotrophin releasing hormone test? BMJ 1984; 289: 1334-1336.
3.	Wilkinson E, Rae PWH, Thomson KJT, Toft AD, Spencer CA, Beckett GJ. Chemiluminescent third generation assay (Amerlite thyroid stimulating hormone 30) of thyroid stimulating hormone in serum or plasma assessed. Clin Chem 1993; 39: 2166-2173.
4.	Pollock MA, Sturrock A, Marshall K, Davidson K, Kelly CJ, McMahon A, et al. Efficacy of thyroxine replacement in patients who feel clinically hypothyroid but are biochemically euthyroid. J Endocrinol 2000; 164(suppl): P329.

Doing more tests is not always better

EDITORO'Reilly's review of thyroid function tests is apposite at a time when most laboratories are experiencing an exponential increase in requests for these tests.¹ It is worrying that clinical diagnosis has been relegated to history and that assessment is based almost entirely on biochemical tests. The only comfort that laboratories can derive from the huge increases in workload and the cost of doing these tests is that doctors apparently have more confidence in laboratory results than in their own clinical assessment.

Thyroid function tests are notorious for producing misleading results in non-thyroidal illness, and yet there are few patients in medical and care of the elderly wards who do not have these tests. Routine preoperative laboratory testing is unnecessary except in specific clinical conditions.² Yet most patients who have an asymptomatic euthyroid goitre or are taking adequate thyroxine replacement have their thyroid function tested before elective non-thyroid surgery (even if results of tests were normal a few weeks or months before). Evidence based medicine has been slow to reach thyroid function testing.

While there are no data on the relative importance of biochemical thyroid function tests and clinical symptoms and signs in assessing thyroid dysfunction, laboratories would be well advised to consider ways of reducing unnecessary and excessive testing. We refuse samples on groups of patients who have (a) had these tests done within the previous month irrespective of the reason, (b) started thyroxine or had the dose changed within the previous six to eight weeks, or (c) had normal results on routine screening for primary thyroid disease within the previous year. This has led to a saving of over 100 requests a monthour current daily workloadwhich, if translated into money, is not an insignificant amount. This is of course only possible because we have a laboratory computer system that can easily identify these patients.

Unless we have data about thyroid function testing related to clinical outcomes it is difficult to justify the increasing trend in the use of these tests. Doing more biochemical tests only leads to more confusion, especially if results do not agree with clinical presentation. This is certainly an area where more is not necessarily better.

Sudha Bulusu, consultant chemical pathologist. 
Newham General Hospital, London E13 8SL newham.pathology{at}virgin.net

1.	O'Reilly DStJ. Thyroid function teststime for reassessment. BMJ 2000; 320: 1332-1334. (13 May.)
2.	Tabas GH, Vanek MS. Is "routine" laboratory testing a thing of the past? Postgrad Med 1999; 105: 213-220.

Thyroid function testing means different things to different people

EDITORO'Reilly suggests that the role of thyroid function tests should be reassessed.¹ In any reassessment it needs to be recognised that different specialties may have different requirements for thyroid function testing. Minor degrees of thyroid dysfunction that seem inconsequential in endocrinological settings may be important in groups who are differentially sensitive to changes in thyroid function, such as patients susceptible to mood disorders.

Thyroid dysfunction, including dysfunction classed as subclinical according to existing biochemical norms, is an important factor in the onset of depressive states ^{2 3} and resistance to antidepressant treatment⁴ and can aggravate mood instability in bipolar mood disorders.² Treatment with thyroid hormones or antithyroid treatment can be beneficial in these and related cases even when circulating thyroid hormone concentrations fall within the normal range. ^{2 4 5}

If thyroid stimulating hormone assays are not routinely performed in these groups, subclinical thyroid dysfunction is more likely to be unrecognised and untreated. In these groups at least, the difficulties identified by O'Reilly need to be addressed primarily by further biochemical research. Increased attention to clinical signs of thyroid dysfunction will be of limited help.

Martin Eales, consultant psychiatrist. 
Somerset Partnership NHS and Social Care Trust, Bridgwater, Somerset TA6 3LS mjeales{at}totalise.co.uk

1.	O'Reilly DStJ. Thyroid function teststime for a reassessment. BMJ 2000; 320: 1332-1334. (13 May.)
2.	Hendrick V, Altschuler L, Whybrow P. Psychoneuroendocrinology of mood disorders. The hypothalamic-pituitary-thyroid axis. Psychiatr Clin North Am 1998; 21: 277-292[CrossRef][Medline].
3.	Maes M, Meltzer HY, Cosyns P, Suy E, Schotte C. An evaluation of basal hypothalamic-pituitary-thyroid axis function in depression: results of a large-scaled and controlled trial. Psychoneuroendocrinol 1993; 18: 607-620[CrossRef][Medline].
4.	Howland RH. Thyroid dysfunction in refractory depression: implications for pathophysiology and treatment. J Clin Psychiatry 1993; 54: 47-54[Medline].
5.	Eales MJ, van der Merwe PL. Severe apathetic hyperthyroidism with normal thyroid hormone levels. Br J Psychiatry 1995; 167: 823-824.

Author's reply

EDITORI agree with Kendall-Taylor that thyroid stimulating hormone assays are not infallible and must be interpreted in the light of clinical features. I also agree with Toft and Beckett that there is nothing to be gained by those who advocate the primacy of thyroid function tests or clinical exmination and that the correct diagnosis will be made on the basis of both clinical examination and the results of tests.

I drew attention to the observation that the clinical features of thyroid dysfunction are now rarely discussed in the medical literature and that, as a consequence, the impression is given that they are of little importance. The clinical features are given in publications such as Thyroid Diseasethe Facts,¹ which I recommend to students and trainees. Sadly, some have the impression that because such publications were written primarily for patients the data they contain that are not available in the conventional medical literature are of little importance.

This view is re-enforced when they note that the clinical features of thyroid disorders have been downgraded in current textbooks. Anyone who consults the definitive Werner and Ingbar'sthe Thyroid² will note that the clinical features of hypothyroidism were given in the 5th edition and effectively abandoned in the 6th, 7th, and 8th editions. The Newcastle thyrotoxicosis index was given, shortly after it was published, in the 2nd edition of the textbook Fundamentals of Clinical Endocrinology³ and dropped from subsequent editions.

I do not advocate extending the reference range for thyroid stimulating hormone to 21.5 mU/l, as suggested by Price and Weetman. I was pointing out that from our first experiences with the measurement of thyroid stimulating hormone it was clear that there was a considerable difference between the reference range (often referred to as the normal range) and what could be considered diagnostic values.

I have no difficulty with cholesterol and triglyceride measurements being used for the diagnosis and monitoring of hyperlipidaemias. However, the statistically derived reference range for plasma cholesterol bears no relation to the cholesterol concentrations used, along with other variables, when establishing risk factor status for coronary heart disease or the therapeutic goals for treatment. Coronary heart disease is one of the major causes of death and morbidity. Yet the number of cholesterol measurements made in hospital laboratories in Scotland in 1999 was 72% of the number of thyrotrophin measurements; figures for England and Wales are unavailable. As Bulusu highlights, there seems to be some inappropriate requesting of thyroid function tests.

Toft and Beckett state that finding an abnormal serum thyroid stimulating hormone concentration in patients taking thyroxine is an indication for adjustment of the dose. Franklyn et al found that in 55 of 153 patients taking thyroxine the serum thyroid stimulating hormone concentration was below the functional sensitivity of the assay (that is, <0.03 mU/l), which is an order of magnitude below the lower end of the reference range.⁴ They stated that "An important finding from the present study was the observation that the serum thyrotropin values were undetectable, even in the most sensitive assays employed, in subjects receiving long term thyroxine therapy."⁴ In practice, to maintain the serum thyroid stimulating hormone concentration within the reference range for the population not taking thyroxine is an unachievable goal in some patients if one takes account of their clinical status.

I fully endorse the view that serum thyroid hormone measurements are essential in diagnosing hypothyroidism and hyperthyroidism. The reference range is so narrow that to diagnose hypothyroidism in patients who have a serum thyroid stimulating hormone concentration within the range, in the absence of hypothalamic-pituitary disease, is virtually untenable. This is in keeping with the findings of Pollock et al.⁵

Denis O'Reilly, consultant clinical biochemist. 
Department of Clinical Biochemistry, Glasgow Royal Infirmary, Glasgow G4 0SF Doreilly{at}clinmed.gla.ac.uk