|Home | About | Journals | Submit | Contact Us | Français|
JAMA 2004;291: 228–38
In the past decade a number of review articles addressing subclinical thyroid disease have been published. They have in turn focused on evolving issues regarding definition, diagnosis and management of this common condition. It is important to note that the literature in this area is still inadequate and consensus statements may often be an expert panel opinion rather than strictly evidence based.
Whilst the American literature has been the most prolific in recent times in this area,1–6 articles have also been published by British7 and Australian groups.8 Many of these focus on overt thyroid disease, rather than the more difficult area of subclinical thyroid disease.
Perhaps the most ambitious attempt to address the contentious issues of subclinical thyroid disease in a non-biased and systematic way was undertaken recently by The (American) Endocrine Society, the American Thyroid Association (ATA) and the American Association of Clinical Endocrinologists (AACE). These societies co-sponsored a Consensus Development Conference in 2002 and contracted an independent consulting firm to review and summarise existing published evidence (195 articles in total). Principal conclusions of the panel were published earlier this year9 and the complete report is available at http://www.endo-society.org/education/evidence-report.cfm.
The planning committee drafted a series of clinically relevant questions related to the diagnosis and management of subclinical thyroid disease. The five questions were:
Subclinical hypothyroidism is a more common entity than subclinical hyperthyroidism and as such this article will focus on hypothyroidism only. This is arguably also the area of most uncertainty. Highlights of the panel’s conclusions for subclinical hypothyroidism are summarised below.
Patients with subclinical thyroid disease have few or no symptoms or signs of thyroid dysfunction and thus by its very nature subclinical thyroid disease is a laboratory diagnosis. Subclinical hypothyroidism is defined as a serum thyroid stimulating hormone (TSH) above the defined upper limit of the reference range, with a serum free thyroxine (T4) within the reference range. Other causes of a raised TSH, a past history of thyroid disease and patients on T4 hormone treatment need to be excluded. It is therefore critically important that the reference limits for TSH be standardised. The TSH method used should have a high functional sensitivity (at least 0.02 mU/L), although this is of most importance for the diagnosis of subclinical hyperthyroidism.
The panel concluded that the reference range for TSH should be based on the third National Health and Nutrition Examination Survey (NHANES III).10 In this population the range (2.5th–97.5th percentile) was 0.45–4.12 mU/L and varied with age, sex, and ethnic background, although these differences were small. The mean concentration was 1.4 mU/L consistent with a skewed distribution and a tail toward higher TSH concentrations. The panel concluded the upper limit of the range should not be reduced as suggested by some associations11 and a range of 0.45–4.5 mU/L should ultimately be adopted.
Whilst the normal reference interval was felt to be adequately defined, the TSH range defining subclinical hypothyroidism remained elusive. An upper limit of 10 mU/L has been quoted in the literature.2,8 Perhaps this is because of patients found to have an elevated TSH level, the majority (approximately 75%) have values lower than 10 mU/L.12 However, the highest TSH quoted in the cross-sectional prevalence studies examined by the panel was 7.0 mU/L.13
This is clearly likely to be affected by the TSH range used to define the problem. Certainly different studies reviewed by the panel utilised different ranges. The prevalence of subclinical hypothyroidism in the United States adult population is 4–8.5%,10,12 although this figure increases with age, may differ among ethnic groups and less consistent data is available among men.10
The progression to overt hypothyroidism is approximately 2–5% per year. The rate of progression is proportional to baseline TSH concentration and is higher in individuals with antithyroid antibodies.14 Despite the impact of thyroid antibody positivity on the epidemiology of subclinical hypothyroidism, the panel did not advise the use of anti-thyroid peroxidase (TPO) antibodies. This differs from recommendations by the Royal College of Physicians (RCP),7 the AACE,1 the ATA6 and Australian thyroidologists.8
In addition to progression to overt hypothyroidism a number of possible consequences of subclinical hypothyroidism exist. The panel attempted to define the nature of association of adverse cardiac events/cardiac dysfunction, effects on lipids, neuropsychiatric symptoms and systemic hypothyroid symptoms. An attempt was made to stratify for TSH (i.e. TSH less than or greater than 10 mU/L), although available data did not always distinguish between serum TSH concentrations. In general there was insufficient or no evidence to support an association between subclinical hypothyroidism and these clinical conditions. The exception was dyslipidaemia, and only in patients with a TSH >10 mU/L.13,15,16 Interestingly although some studies assessing T4 replacement suggest that dyslipidaemia can be improved with treatment,17 others have failed to confirm these findings.18,19
The panel concluded that there was insufficient evidence for a clinically significant relationship between subclinical hypothyroidism and adverse cardiac events or cardiac dysfunction. Numerous small studies have demonstrated subtle echocardiographic changes,20 however these studies all had significant limitations. It is of note that a single large cross-sectional study concluded that subclinical hypothyroidism was a risk factor for atherosclerosis and myocardial infarction. However, a longitudinal component of the same study failed to confirm any association, although event rates were low in both the euthyroid and subclinical hypothyroid groups.21 Whilst small interventional treatment trials have shown improvement in cardiac function, these are of uncertain significance, and there have been no randomised studies assessing T4 replacement therapy on important clinical endpoints.
Other clinical conditions where evidence is insufficient is regarding neuropsychiatric and systemic hypothyroid symptoms. A large cross-sectional study confirmed that participants with subclinical hypothyroidism reported more symptoms than euthyroid individuals, but fewer symptoms than overtly hypothyroid participants and did not distinguish between untreated subclinical hypothyroidism and undertreated overt hypothyroidism.12 Other studies have failed to confirm this association. Studies were criticised for being referred or selected populations, or hospitalised patients. Randomised control trials restricted to patients with a TSH less than 10 mU/L assessing the effects of T4 replacement found no improvement in symptoms.18,22
Guidelines for the evaluation of patients with subclinical hypothyroidism were published by members of the panel.23 The suggested algorithm includes repeating the TSH and T4 levels within 2–12 weeks, ongoing monitoring of patients in whom the TSH has normalised and further clinical assessment in those patients with a persistently abnormal TSH. If the T4 is low then treatment was advocated irrespective of the TSH result. If the TSH was greater than 10 mU/L then a trial of treatment was thought reasonable even if the T4 was normal, especially in women contemplating pregnancy. Where the TSH was less than 10 mU/L a trial of treatment could be considered if there were signs or symptoms consistent with hypothyroidism. Monitoring of untreated patients at 6–12 monthly intervals was advocated. The algorithm did not include the evaluation of antibody status.
Benefits of treatment need to be weighed against any potential risks. Replacement T4 is the corner stone of therapy. The arguments against treatment are its expense and the likelihood that some patients will not benefit. There is also a danger of over-treatment, which could cause iatrogenic hyperthyroidism. Subclinical hyperthyroidism has been associated with negative outcomes including osteopaenia and atrial fibrillation.24 One study found that 21% of participants on T4 therapy had a suppressed TSH suggestive of overtreatment.12 It is worth noting that once on T4 therapy the optimal TSH level has yet to be defined. The panel and the ATA advocate treating to a TSH of 0.5–2.0 mU/L on the basis that this is the mean normal TSH level.10
Initiation of early therapy with T4 (TSH less than 10 mU/L) is not clearly associated with benefit18,22 and as such the panel did not support the use of T4 therapy in this setting, although it stated that a trial of T4 might be reasonable in “symptomatic” patients. However T4 therapy is reasonable for patients with a TSH greater than 10 mU/L because the rate of progression to overt hypothyroidism is up to 5% per year in comparison with lower TSH levels.14
Special consideration should be given to pregnant women. An elevated TSH in this group of women should be treated with T4, as there is a possible association between high maternal TSH and adverse neuropsychological foetal outcomes.25 There are no intervention trials assessing the benefits of T4 therapy in this group of women.
Screening remains a controversial area with lack of consensus amongst physician groups. For example whilst the ATA recommends screening all adults over the age of 35 yrs every 5 years,3 the RCP believe it unjustified in the general population but acknowledges that exceptions need to be made for various high risk groups.7 Overall the panel recommended against population screening, but argued for case ascertainment in certain high risk groups, similar to recent recommendations by the US Preventive Service Task Force.26 The panel believed that special consideration could be given to pregnant women, but only if they were at increased risk of thyroid disease (personal or family history of thyroid or autoimmune disease). Routine screening of pregnant women after consultation between physician and patient has however been recommended by the AACE.27 It is noteworthy that controversy remains over what is the best screening test in this situation (TSH or T4).28
Of particular interest to laboratories, is that having defined subclinical hypothyroidism as a laboratory rather than a clinical diagnosis, the panel made little reference to the role of the laboratory and its interaction with the physician. A recent publication from a British group sheds some light on this.29 The hospitals serviced by the laboratories had adopted the RCP guidelines7 which advise routine testing of TPO antibodies. The authors found in order to better advise clinicians and to achieve compliance with accepted protocols, automatic cascade testing for TPO antibodies as well as appropriate interpretive reports should be an integral part of the investigation of subclinical hypothyroidism. Thus, although guidelines may be generally accepted, they may not necessarily be put into clinical practice.
The recommendations of the consensus panel have been previously criticised30 and it has been suggested that they may be no more valid than those previously published by other professional groups. The panel’s conclusions are however, based on a thorough examination of available evidence, although unfortunately there is a paucity of evidence addressing the important clinical questions posed.
Until there is widespread consensus, and appropriate large, well executed trials upon which to base recommendations, the panel encouraged clinicians to continue to make individual patient assessments when determining the need for testing and treatment of thyroid dysfunction.