3.4.4 Subclinical hypothyroidism

Subclinical hypothyroidism is defined biochemically as the association of a raised serum thyroid-stimulating hormone (TSH) concentration with normal circulating concentrations of free thyroxine (T₄) and free triiodothyronine (T₃). The term subclinical hypothyroidism implies that patients should be asymptomatic, although symptoms are difficult to assess, especially in patients in whom thyroid function tests have been checked because of nonspecific complaints such as tiredness. An expert panel has recently classified individuals with subclinical hypothyroidism into two groups (1): (1) those with mildly elevated serum TSH (typically TSH in the range 4.5–10.0 mU/l) and (2) those with more marked TSH elevation (serum TSH >10.0 mU/l).

Several population-based studies have examined the prevalence of subclinical hypothyroidism. Variation in results reflects the demographic characteristics of the populations studied, as well as the upper limit set for TSH measurements. Considerable debate has surrounded the setting of the upper limit of the reference range for TSH, with some arguing in favour of reduction in this upper limit to a value which would include a large proportion of the adult population. Meticulous studies in the USA and elsewhere have addressed this question, taking into account the influence of inclusion or exclusion of individuals with a personal or family history of thyroid disease or those with positive antithyroid antibodies. Evidence from one such study (NHANES III) of a very large ‘reference’ population without evidence of thyroid disease has suggested that 95% of adults have a serum TSH within the range 0.45–4.12 mU/l (2), determining that the widely applied upper limit of normal for serum TSH measurements of approximately 4.5–5.0 mU/l remains appropriate.

Using this biochemical definition for TSH elevation, most studies including all ages and both sexes have revealed a prevalence of subclinical hypothyroidism of around 5–10%, the diagnosis being more common in women and increasing with increasing age (see Chapter 3.1.7). The NHANES III study in the USA found subclinical hypothyroidism (TSH >4.6 mU/l) in 4.3% (2), while the Whickham survey in the UK reported TSH of more than 6.0 mU/l in 7.5% of females and 2.8% of males (3). TSH did not vary with age in males but increased markedly in women aged more than 45 years. In the large Colorado study of people attending health fairs, 9.5% had raised TSH, 75% of whom had mildly elevated TSH in the range 5.0–10.0 mU/l and 25% of them were taking thyroid hormones (4). Our own study of 1210 patients aged over 60 years who were recruited in primary care revealed a prevalence of subclinical hypothyroidism of 11.6% in women and 2.9% in men in that age group (5). Significant titres of antithyroid antibodies were found in 46% of those with a serum TSH between 5 and 10 mU/l, and in 81% of those with a serum TSH greater than 10 mU/l, providing supporting evidence for underlying autoimmune thyroid disease in the majority. However, a recent community screening study of older people in the same area revealed a lower population prevalence of subclinical hypothyroidism of 2.9%, perhaps reflecting more frequent testing of thyroid function and earlier treatment of raised TSH in the intervening years (6).

The commonest cause of subclinical hypothyroidism is autoimmune thyroiditis. Another major cause is previous treatment for hyperthyroidism (Box 3.4.4.1). It is well known that treatment of hyperthyroidism with radio-iodine results in thyroid failure in at least 50% of patients (depending upon the dose administered) (7), a rise in TSH being the earliest biochemical indicator. Partial thyroidectomy for hyperthyroidism or nodular goiter is associated with a similar risk of development of hypothyroidism, which is again first identified by a rise in serum TSH. In the early months after both radio-iodine treatment and partial thyroidectomy, subclinical hypothyroidism may be a transient phenomenon not always indicative of progressive or permanent thyroid failure. Graves’ disease is itself associated with the eventual development of hypothyroidism in 5–20% of patients (even in the absence of ablative thyroid treatment).

A further major category of patients with a biochemical diagnosis of subclinical hypothyroidism is that already treated with thyroxine for thyroid failure, a high serum TSH indicating that the dose prescribed is inadequate or compliance is poor. We found a raised serum TSH in approximately 25% of patients on T₄ identified in the community, with a close relationship evident between prescribed dose and TSH results, indicating that, at least in some patients (especially those prescribed T₄ in doses of 75 μg/day or less), the cause of subclinical hypothyroidism was inadequate dose prescription (8).

As well as those with a history of treatment for hyperthyroidism, other groups at particular risk of subclinical hypothyroidism include those with other autoimmune diseases such as type 1 diabetes mellitus and Addison’s disease. (Glucocorticoid deficiency may itself be associated with a rise in serum TSH which is corrected by steroid replacement alone and is not necessarily indicative of underlying thyroid disease.) Down’s syndrome is also associated with the development of both subclinical and overt thyroid failure of autoimmune aetiology. The risk of subclinical hypothyroidism during pregnancy is considerable in women identified in the first trimester as having positive antithyroid antibodies. This antibody status also represents a risk factor for the development of postpartum thyroiditis, subclinical or overt hypothyroidism being a feature of postpartum thyroiditis in about 75% of cases (9). While hypothyroidism may be a transient feature of postpartum thyroiditis, there is good evidence that the majority of affected women go on to develop permanent hypothyroidism after a period of months or years of follow-up (9). Subclinical hypothyroidism may also be a feature of thyroiditis which follows pregnancy loss, even of short duration.

A further well-documented cause of subclinical hypothyroidism is radiotherapy to the head and neck (which is itself associated with the development of positive antithyroid antibodies). Nonthyroidal illness may be associated with a transient and typically modest increase in serum TSH, especially in the recovery phase from illness, although in most instances, even in patients with subclinical hypothyroidism diagnosed in hospital, an underlying ‘thyroid’ cause can be identified. Therapy with drugs such as lithium can induce subclinical hypothyroidism, as can administration of iodine-containing compounds such as radiographic contrast agents. Treatment with the iodine-containing antiarrhythmic drug amiodarone frequently leads to a modest elevation in serum TSH early in treatment, reflecting inhibition of thyroid hormone release, as well as a later increased risk of overt thyroid failure which is first identified by a sustained and progressive rise in serum TSH. Even the use of topical iodine-containing antiseptics can result in thyroid dysfunction, subclinical hypothyroidism being identified in one study in 20.8% of iodine-exposed infants (10).

The natural history of subclinical hypothyroidism depends upon the underlying cause of the biochemical disturbance and the population studied. One large follow-up study has shown that in those with modest elevation of serum TSH (5.5–10.0 mU/l) the TSH measurement returns spontaneously to the reference range in more than 60% of cases during 5 years of follow-up (11). Transient cases may occur in the early weeks or months after recovery from nonthyroidal illness, in the first 6 months after partial thyroidectomy or radio-iodine, or after iodine exposure (e.g. after starting amiodarone). Our own study of people over the age of 60 in the community revealed that the finding of a raised serum TSH identified on screening disappeared in 5.5% after a period of 12 months, while the biochemical abnormality remained stable in 76.7% and relatively few (17.8%) progressed to overt hypothyroidism (the latter defined biochemically as elevation in serum TSH in association with a serum free T₄ below the reference range) (5). Follow-up for 20 years of the Whickham cohort in the north-east of England revealed an annual rate of progression of subclinical to overt hypothyroidism of 2.6% if thyroid antibodies were negative, but a rate of progression of 4.3% if antibodies to thyroid peroxidase were present (12). The risk of development of hypothyroidism in that population was greater if serum TSH was within the upper half rather than the lower half of the typical reference range, fuelling debate regarding the ‘true’ upper limit of normal for TSH.

Given the prevalence of this biochemical diagnosis, much attention has focused upon the effects of mild thyroid hormone deficiency upon symptoms, quality of life, and cognitive function. Because of possible effects on the lipid profile, recent studies have focused on the cardiovascular system and effects on vascular morbidity and mortality. Epidemiological studies are beginning to provide insight into the question of whether subclinical hypothyroidism is associated with adverse outcomes and therefore should be treated with thyroxine replacement.

Studies addressing the relationship between symptoms suggestive of thyroid hormone deficiency and the finding of subclinical hypothyroidism have produced conflicting results. The Colorado health fair study revealed a slight increase in the mean number of reported symptoms in those with high TSH compared with euthyroid controls (13.8% vs 12.1%, p <0.05) (4); however, in another study, a combination of symptoms and signs was not predictive of subclinical hypothyroidism in a geriatric population. Similarly, in a cross-sectional study of women aged 18–75 years, subclinical hypothyroidism was not associated with poorer wellbeing or quality of life (13). Results are also conflicting with regard to any association with depression or decline in cognitive function. Nearly all large studies have failed to find an association with symptoms of depression or impaired cognitive function. For example, in our own study of 5865 subjects aged over 65 years, of whom 168 had subclinical hypothyroidism, we found no association with tests of cognitive function, anxiety, or depression (14).

Several placebo-controlled trials have examined the question of whether T₄ replacement leads to improvement in such measures. Once more results are conflicting, probably reflecting small sample sizes and sometimes short duration of therapy and failure to achieve stable euthyroidism in the treatment group. One of the larger studies of 66 women with a mean TSH of 11.7 mU/l demonstrated no difference between T₄-treated and placebo groups after 48 weeks therapy, although some improvement in symptoms was seen in those with TSH of more than 12.0 mU/l (15). Other studies of 89 subjects (mean TSH 5.57 mU/l) (16) and 100 subjects (17), found no significant effects of T₄ treatment on various tests of cognitive function, quality of life, and depression scores.

Overt hypothyroidism results in reductions in the synthesis and degradation of lipids, but the latter effect predominates so that hypothyroidism results in increases in total and low-density lipoprotein (LDL) cholesterol, as well as marked changes in other lipoprotein and apolipoprotein concentrations. Lipid changes in subclinical hypothyroidism are considerably less marked. Cross-sectional studies comparing subjects with subclinical hypothyroidism and euthyroid controls have shown that subclinical hypothyroidism is associated with variable and inconsistent increases in total cholesterol, LDL cholesterol, and an inconsistent decrease in high-density lipoprotein cholesterol, findings compatible with an increase in atherogenic risk. For example, in the NHANES III cohort, mean total cholesterol (but not LDL) levels were higher in subclinical hypothyroid subjects than euthyroid controls, a finding lost in terms of statistical significance when adjusted for factors such as age and use of lipid-lowering agents (18). Overall, it has been estimated that 0.5 mmol/l total cholesterol might be accounted for by subclinical hypothyroidism. Unsurprisingly, meta-analyses of intervention studies with T₄ have shown only minor effects on the lipid profile, and the most recent meta-analysis revealed reductions of 0.2–0.3 mmol/l in total and LDL cholesterol values after T₄ treatment, with no associated change in triglycerides (19). Generally, more marked changes in cholesterol are seen in those with higher baseline values and in those with higher TSH.

Influences of subclinical hypothyroidism upon the vascular system have been studied in some detail, the most consistent findings being left ventricular diastolic dysfunction in association with an increase in systemic vascular resistance and arterial thickness. Generally, these haemodynamic changes are thought to be corrected by T₄ replacement. These, together with lipid findings, have prompted epidemiological studies of vascular morbidity and mortality, with inconsistent results. In the 20-year follow-up of the Whickham cohort from the north-east of England, there was no association with a diagnosis of autoimmune thyroid disease and a diagnosis of ischaemic heart disease (12). In contrast, in the Rotterdam cohort of women over 55 years there was an association between subclinical hypothyroidism and atherosclerosis (defined as aortic calcification on lateral radiograph) and with a history of myocardial infarction, although no association with incident ischaemic heart disease (20). In our own study of 1200 subjects aged more than 60 years followed for 10 years, we found no association of subclinical hypothyroidism with circulatory mortality (although 40% had commenced T₄ therapy during follow-up) (21). Intriguingly, in the Leiden study of those aged more than 85 years, raised TSH was associated with increased longevity and decreased risk of death from cardiovascular disease (22). The longitudinal Cardiovascular Health Study in the USA found no association between subclinical hypothyroidism and the incidences of cardiovascular or cerebrovascular diseases, nor all-cause mortality (23); however, an association between a serum TSH of more than 10.0 mU/l and heart failure events has recently been described in the same cohort (24). A recent meta-analysis has examined the possible association between subclinical hypothyroidism and vascular or all-cause mortality. It was concluded that, at present, the evidence for association is weak (25).

Subclinical hypothyroidism is a common condition, especially among specific patient groups including elderly patients, those with a past or family history of thyroid disease, those with other autoimmune diseases such as type 1 diabetes, and those receiving therapy with drugs such as amiodarone and lithium. The marked prevalence of this disorder has led to debate regarding the appropriateness of population screening (i.e. routine testing of asymptomatic individuals). This debate centres on the lack of evidence that treatment of subclinical (as opposed to overt) hypothyroidism has a beneficial effect in terms of patient wellbeing and/or long-term morbidity, e.g. due to cardiovascular disease, and takes into account the variable natural history of the disorder in different patient groups and the potential influence upon patient wellbeing of the knowledge that they have an abnormal test result. While opposing views have been expressed, a consensus statement from UK experts in thyroid disease has suggested that general testing of the population is at present unjustified, even in those aged over 60 years and those with a family history of thyroid disease; these views are in accord with a US consensus panel (1) and the US Preventative Task Force. Groups in whom screening is considered appropriate include those with a past history of treatment for hyperthyroidism with radio-iodine or surgery and, perhaps, those with type 1 diabetes (especially in pregnancy), as well as those receiving lithium or amiodarone (see also Chapter 3.1.7).

Once the diagnosis of subclinical hypothyroidism has been made, either as a result of routine testing of a particular patient group or prompted by nonspecific symptoms such as tiredness or weight gain, the question arises as to if and when to treat with thyroxine replacement therapy. Given the relative paucity of evidence that treatment of subclinical hypothyroidism results in benefit in terms of symptoms or long-term outcome, the debate continues. The association between serum TSH values of more than 10 mU/l and ‘adverse’ findings, such as faster progression to overt hypothyroidism, hyperlipidaemia, and perhaps vascular morbidity, leads many experts to treat with thyroxine in this group. It is much less clear that those with modestly elevated serum TSH (<10 mU/l) should be treated. The US consensus panel of experts concluded that there was insufficient evidence to warrant treatment of those with mildly elevated TSH (who should have repeat testing at 6–12 monthly intervals) but that those with TSH of more than 10 mU/l should be considered for treatment (1). An exception is in pregnancy where most experts would recommend thyroxine treatment for even modest elevations of serum TSH in view of possible adverse outcomes, such as pregnancy loss and slightly impaired neurodevelopment in offspring (see Chapter 3.4.5).