Anda di halaman 1dari 11

SCIENTIFIC REVIEW CLINICIANS CORNER

AND CLINICAL APPLICATIONS

Subclinical Thyroid Disease


Scientific Review and Guidelines for Diagnosis and Management
Martin I. Surks, MD Context Patients with serum thyroid-stimulating hormone (TSH) levels outside the
Eduardo Ortiz, MD, MPH reference range and levels of free thyroxine (FT4) and triiodothyronine (T3) within the
reference range are common in clinical practice. The necessity for further evaluation,
Gilbert H. Daniels, MD possible treatment, and the urgency of treatment have not been clearly established.
Clark T. Sawin, MD Objectives To define subclinical thyroid disease, review its epidemiology, recommend
Nananda F. Col, MD, MPP, MPH an appropriate evaluation, explore the risks and benefits of treatment and consequences
of nontreatment, and determine whether population-based screening is warranted.
Rhoda H. Cobin, MD
Data Sources MEDLINE, EMBASE, Biosis, the Agency for Healthcare Research and
Jayne A. Franklyn, MD Quality, National Guideline Clearing House, the Cochrane Database of Systematic Re-
Jerome M. Hershman, MD views and Controlled Trials Register, and several National Health Services (UK) data-
bases were searched for articles on subclinical thyroid disease published between 1995
Kenneth D. Burman, MD and 2002. Articles published before 1995 were recommended by expert consultants.
Margo A. Denke, MD Study Selection and Data Extraction A total of 195 English-language or trans-
Colum Gorman MD, PhD lated papers were reviewed. Editorials, individual case studies, studies enrolling fewer
than 10 patients, and nonsystematic reviews were excluded. Information related to
Richard S. Cooper, MD authorship, year of publication, number of subjects, study design, and results were
Neil J. Weissman, MD extracted and formed the basis for an evidence report, consisting of tables and sum-
maries of each subject area.

S
UBCLINICAL OR MILD THYROID
Data Synthesis The strength of the evidence that untreated subclinical thyroid dis-
disease is a common disorder,
ease is associated with clinical symptoms and adverse clinical outcomes was assessed
particularly in middle-aged and and recommendations for clinical practice developed. Data relating the progression of
elderly individuals.1 Greater subclinical to overt hypothyroidism were rated as good, but data relating treatment to
sensitivity of assays and more fre- prevention of progression were inadequate to determine a treatment benefit. Data
quent assessment of serum thyroid- relating a serum TSH level higher than 10 mIU/L to elevations in serum cholesterol
stimulating hormone (TSH) levels have were rated as fair but data relating to benefits of treatment were rated as insufficient.
resulted in more patients requiring in- All other associations of symptoms and benefit of treatment were rated as insufficient
terpretation of abnormal thyroid func- or absent. Data relating a serum TSH concentration lower than 0.1 mIU/L to the pres-
tion test results. However, contro- ence of atrial fibrillation and progression to overt hyperthyroidism were rated as good,
but no data supported treatment to prevent these outcomes. Data relating restora-
versy surrounds the definition, clinical
tion of the TSH level to within the reference range with improvements in bone min-
importance, and necessity for prompt eral density were rated as fair. Data addressing all other associations of subclinical hy-
diagnosis and treatment of subclinical perthyroid disease and adverse clinical outcomes or treatment benefits were rated as
thyroid disease. Previous review ar- insufficient or absent. Subclinical hypothyroid disease in pregnancy is a special case
ticles2-6 and position statements7,8 dif- and aggressive case finding and treatment in pregnant women can be justified.
fer in their conclusions and recommen- Conclusions Data supporting associations of subclinical thyroid disease with symp-
dations, often a consequence of toms or adverse clinical outcomes or benefits of treatment are few. The consequences
difficulties in interpreting inadequate of subclinical thyroid disease (serum TSH 0.1-0.45 mIU/L or 4.5-10.0 mIU/L) are mini-
and conflicting data. In the midst of this mal and we recommend against routine treatment of patients with TSH levels in these
uncertainty, clinicians still desire ex- ranges. There is insufficient evidence to support population-based screening. Aggres-
pert guidance for the diagnosis and sive case finding is appropriate in pregnant women, women older than 60 years, and
management of subclinical thyroid others at high risk for thyroid dysfunction.
JAMA. 2004;291:228-238 www.jama.com
disease.

See also p 239. Author Affiliations and Financial Disclosures are listed 10467-2490 (e-mail: msurks@westnet.com).
at the end of this article. Reprints: Society Services, The Endocrine Society,
CME available online at Corresponding Author: Martin I. Surks, MD, Mon- 8401 Connecticut Ave, Suite 900, Chevy Chase,
www.jama.com tefiore Medical Center, Medical Arts Pavilion, 3400 MD 20815 (e-mail: societyservices@endo-society
Bainbridge Ave, Second Floor, Bronx, NY .org).

228 JAMA, January 14, 2004Vol 291, No. 2 (Reprinted) 2004 American Medical Association. All rights reserved.

Downloaded from www.jama.com on December 13, 2006


SUBCLINICAL THYROID DISEASE

In an effort to address these contro- cieties served without compensation ture of study (eg, cohort, blinded,
versial issues, representatives of the whereas nonendocrinologist panelists randomized), and principal findings.
American Thyroid Association (ATA), received an appropriate honorarium. The complete report is available at http:
the American Association of Clinical En- //www.endo-society.org/education
docrinologists (AACE), and the Endo- METHODS /evidence-report.cfm.
crine Society formed a planning com- The Lewin Group, an independent con- On the first 112 days of the consen-
mittee for a consensus development sulting organization, was contracted to sus conference, 12 experts identified by
conference to review the literature and review the literature and summarize the the planning committee presented re-
attempt to formulate some recommen- evidence relating to the clinical ques- views of selected areas including epi-
dations to guide clinical practice. The tions. Relevant articles were identified demiology, laboratory testing, symp-
committee adopted an approach pat- by searching MEDLINE, EMBASE, Bio- toms, effects on bone, lipids, and
terned on the National Institutes of sis, the Agency for Healthcare Re- cardiovascular systems, screening, and
Health (NIH) consensus development search and Quality (AHRQ) National effects of treatment to the panel and au-
process. The planning committee drafted Guideline Clearinghouse, the Coch- dience. These expert presenters left the
a series of clinically relevant questions rane Database of Systematic Reviews, the conference at the end of the informa-
related to the diagnosis and manage- Cochrane Controlled Trials Register, and tion gathering session. Over the re-
ment of subclinical hypothyroidism and several National Health Services (UK) maining 112 days, the panel discussed
hyperthyroidism. These questions were databases, including the Database of Ab- the information presented and the data
What is the definition of subclini- stracts of Reviews of Effectiveness, the abstracted from the literature review to
cal thyroid disease? Economic Evaluation Database, and the address the questions posed by the plan-
What is the epidemiology of sub- database of the International Network ning committee.
clinical thyroid disease? of Agencies for Health Technology As- The panel assessed the data for qual-
What are the consequences of un- sessment. Key search terms were sub- ity, scope, and relevance. Using criteria
treated subclinical thyroid disease? How clinical (text word) or subclinic and hy- adopted from the US Preventive Ser-
should it be evaluated? pothyroidism or thyroid deficien or vices Task Force (USPSTF),9 the panel
What are the risks and benefits of thyroid insufficien; subclinical (text rated the strength of the available evi-
treatment for subclinical thyroid dis- word) or subclinic and hyperthyroid- dence as either good, fair, or insuffi-
ease? ism or thyrotoxicosis or overactive thy- cient as it related to the association of
Is screening for subclinical thy- roid. The following areas were evalu- thyroid status or benefits of treatment
roid disease warranted? ated (key words in parentheses): to specified outcomes (BOX 1). Given the
The questions were presented to a epidemiology (etiology or ethnology or paucity of randomized controlled trials
panel of 13 experts selected by the plan- epidemiology or mortality), screening (RCTs), the panel relied on the avail-
ning committee who were either se- (screening or thyroid function tests), treat- able published evidence as well as that
nior endocrinologists not known to ment (therapy or treatment or radio- presented during the expert presenta-
publish or be advocates in this area or therapy or surgery or complication or tions, particularly for data related to
experts in other relevant fields. Mem- hormon), consequences of no treat- clinical outcomes. When evidence was
bers of the planning committee were ment (complications or mortality), and not available, was contradictory, or was
not members of the panel. Eight of the economics (cost or costs or cost analysis judged to be insufficient, the panelists
panelists were experts in thyroid dis- or cost-benefit or cost effective). relied on their experience, judgment,
ease and the remaining 5 had exper- All English-language research ar- and interpretation of the available lit-
tise in cardiology, epidemiology, bio- ticles or translations published on the erature in formulating recommenda-
statistics, evidence-based medicine, topic from 1995 to July 2002 were re- tions for clinical practice. Differences of
health services research, general inter- viewed, as well as 21 relevant articles opinion were settled by a majority vote
nal medicine, and clinical nutrition. and 4 abstracts published before 1995 after extensive discussion. The recom-
The conference was held Septem- that were identified by the planning mendations for clinical practice were de-
ber 21-23, 2002. The meeting was open committee. Excluded were editorials, veloped on the basis of the evidence
to the public and attended by mem- individual case studies, studies enroll- evaluations during the conference de-
bers of the 3 sponsoring societies. Con- ing fewer than 10 patients, and many liberations.
tinuing medical education credit was nonsystematic reviews. The final count Each clinical practice recommenda-
provided through the Endocrine Soci- was 195 articles, including the earlier tion was rated by individual members
ety. All potential conflicts of interest relevant publications identified by the of the panel for strength of supporting
were obtained from panelists and speak- planning committee. The report con- evidence (good, fair, insufficient, or
ers and printed in the conference ma- sisted of tables and summaries of each based on expert opinion) (BOX 2). Panel
terials. Endocrinologists who were subject area indicating the authors, year members were asked to indicate their
members of any of the 3 organizing so- of publication, numbers of subjects, na- level of agreement (none, minimal,
2004 American Medical Association. All rights reserved. (Reprinted) JAMA, January 14, 2004Vol 291, No. 2 229

Downloaded from www.jama.com on December 13, 2006


SUBCLINICAL THYROID DISEASE

cally important that the normal refer-


Box 1. Strength of the Overall Evidence ence range for TSH be standardized and
Good
that laboratories engage in appropriate
quality control procedures to ensure that
Evidence includes consistent results from well-designed, well-conducted studies
in representative populations that directly assess effects on health outcomes. the results they report are accurate and
reproducible.10,11 The TSH method used
Fair should have a functional sensitivity of
Evidence is sufficient to determine effects on health outcomes, but the strength of at least 0.02 mIU/L and the functional
the evidence is limited by the number, quality, or consistency of the individual sensitivity should be independently es-
studies; generalizability to routine practice; or indirect nature of the evidence on
tablished by each laboratory.
health outcomes.
What Is the Definition of Subclini-
Insufficient cal Hypothyroidism? Subclinical hypo-
Evidence is insufficient to assess the effects on health outcomes because of lim- thyroidism is defined as a serum TSH
ited number or power of studies, important flaws in their design or conduct, gaps concentration above the statistically de-
in the chain of evidence, or lack of information on important health outcomes. fined upper limit of the reference range
Source: Adapted from the US Preventive Services Task Force, Agency for Healthcare Re- when serum free T4 (FT4) concentra-
search and Quality.9
tion is within its reference range.12 Other
causes of an elevated serum TSH must
be excluded, for example: recent adjust-
ments in levothyroxine dosage with fail-
Box 2. Strength of Panelists Recommendations Based on Available ure to reach a steady state,13 particu-
Evidence larly in poorly compliant patients;
Rating
transient increase in serum TSH in hos-
pitalized patients during recovery from
A: Strongly recommends. The recommendation is based on good evidence that
the service or intervention can improve important health outcomes. severe illness14,15 or during recovery from
B: Recommends. The recommendation is based on fair evidence that the service destructive thyroiditis, including post-
or intervention can improve important health outcomes. viral subacute thyroiditis and postpar-
C: Recommends. The recommendation is based on expert opinion. tum thyroiditis; untreated primary ad-
D: Recommends against. The recommendation is based on expert opinion. renal insufficiency16,17; patients receiving
E: Recommends against. The recommendation is based on fair evidence that the recombinant human TSH injections18;
service or intervention does not improve important health outcomes or that and the presence of heterophilic anti-
harms outweigh benefits. bodies against mouse proteins, which
F: Strongly recommends against. The recommendation is based on good evi- cause falsely high TSH concentrations
dence that the service or intervention does not improve important health out-
in some assays.19-21 Although central hy-
comes or that harms outweigh benefits.
I: Recommends neither for nor against. The panel concludes that the evidence is
pothyroidism (usually hypothalamic)
insufficient to recommend for or against providing the service or intervention may cause a mildly elevated serum TSH
because evidence is lacking that the service or intervention improves impor- concentration (due to a circulating bio-
tant health outcomes, the evidence is of poor quality, or the evidence is con- inactive TSH molecule),22 the serum FT4
flicting. As a result, the balance of benefits and harms cannot be determined. concentration is generally clearly low in
Source: Adapted from the US Preventive Services Task Force, Agency for Healthcare Re- these patients.
search and Quality.9 Serum TSH concentrations in a
healthy population have a skewed dis-
tribution with a tail toward higher
TSH concentrations. Because of the rela-
moderate, or strong) with each recom- 2 of the recommendations were sup- tively high prevalence of subclinical hy-
mendation. Panel members submitted ported unanimously. pothyroidism in the general popula-
their assessments of the strength of evi- tion, it is likely that some of the skew
dence and their support for the recom- RESULTS in the upper limits of normal is a re-
mendations during their review of the Subclinical Thyroid Disease: sult of inclusion of patients with sub-
draft manuscript. A summary of the Questions and Recommendations clinical disease.
panelists assessment of the strength of Subclinical thyroid disease is, by its very The third National Health and Nu-
evidence and his/her degree of sup- nature, a laboratory diagnosis. Patients trition Examination Survey (NHANES
port for each recommendation is avail- with subclinical disease have few or no III)23 examined serum TSH values in a
able at http://www.endo-society.org definitive clinical signs or symptoms of disease-free subset (n=13344) of an
/education/evidence-report.cfm. All but thyroid dysfunction. Thus, it is criti- ethnically diverse reference popula-
230 JAMA, January 14, 2004Vol 291, No. 2 (Reprinted) 2004 American Medical Association. All rights reserved.

Downloaded from www.jama.com on December 13, 2006


SUBCLINICAL THYROID DISEASE

tion, aged 12 years and older (exclud- serum FT4 and T3 concentrations are tus, a family history of thyroid disease,
ing pregnant women, individuals tak- within their reference ranges.12 Other or previous head and neck cancer treated
ing estrogens, androgens, or lithium, causes of a low serum TSH must be ex- with external beam radiation all raise the
and those with detectable antithyroid cluded. Subclinical hyperthyroidism may likelihood of subclinical hypothyroid-
antibodies to thyroid peroxidase [TPO] result from endogenous overproduc- ism. About 20% of patients taking thy-
or laboratory evidence of hypothyroid- tion of thyroid hormone or intended, or roid medications (not otherwise speci-
ism or hyperthyroidism). In this se- inadvertent, overadministration of thy- fied) have subclinical hypothyroidism.1
lected population, the reference range roid hormone. Among other causes of a Of patients with subclinical hypothy-
of TSH concentration (2.5th-97.5th per- low serum TSH concentration with nor- roidism, approximately 2% to 5% per
centile) was 0.45 to 4.12 mIU/L, and the mal concentrations of FT4 are delayed re- year will progress to overt hypothyroid-
geometric mean TSH concentration was covery of the pituitary TSH-producing ism. Overt hypothyroidism is generally
1.4 mIU/L. The reference range varied cells during or after therapy for hyper- defined as a low serum FT4 concentra-
as a function of age, sex, and ethnic thyroidism,25 normal pregnancy,26 vari- tion with elevated serum TSH concen-
group, but because the differences are ous nonthyroidal illnesses (euthyroid tration,34,37 but in some cases individu-
relatively small, it is not considered nec- sick syndrome),27,28 or the administra- als with hypothyroid symptoms and
essary to adjust the reference range for tion of dopamine,29 glucocorticoids,30,31 high TSH (10 mIU/L) with low nor-
these factors in clinical practice. and possibly dobutamine.32 Although mal FT4 have been among those de-
Some investigators suggest that the subnormal serum TSH concentrations fined as having overt hypothyroid-
upper limit of normal for serum TSH are common in a variety of severe non- ism. 2 4 The rate of progression is
concentration should be 2.5 mIU/L11 in thyroidal illnesses, undetectable serum proportional to the baseline serum TSH
a population rigorously screened to ex- TSH concentrations (0.01 mIU/L) are concentration and is higher in individu-
clude thyroid disease or drugs that in- rare unless patients are receiving con- als with antithyroid antibodies.24 In in-
fluence thyroid function. In support of comitant glucocorticoids (usually in high dividuals not taking thyroid hormone,
this position is a higher rate of progres- doses) or dopamine. Although patients serum TSH returns to normal after 1 year
sion to overt hypothyroidism and a with pituitary or hypothalamic failure of follow-up in approximately 5% but re-
higher prevalence of antithyroid anti- (including anorexia nervosa) fre- mains elevated in the remainder.34
bodies in individuals with serum TSH quently have subnormal serum TSH con- Subclinical hyperthyroidism is much
higher than 2.5 mIU/L compared with centrations, the FT4 is also usually sub- less common than subclinical hypothy-
those with serum TSH between 0.5 and normal.12 When serum FT4 is in the roidism. When the lower limit of TSH is
2.5 mIU/L.24 Although a serum TSH con- normal range, it is almost invariably in less than 0.4 mIU/L, 3.2% of the popu-
centration higher than 2.5 but less than the lower part of the range in those with lation is defined as having subclinical hy-
4.5 mIU/L may identify some individu- nonthyroidal illness in contrast to the perthyroidism.23 If patients with known
als with the earliest stage of hypothy- high normal FT4 concentration of typi- thyroid disease are excluded, the preva-
roidism and those suspect for Hashi- cal subclinical hyperthyroidism. lence decreases to 2%. Subclinical hy-
moto thyroiditis, there is no evidence for What Is the Epidemiology of Sub- perthyroid disease is more common in
associated adverse consequences. Fur- clinical Thyroid Disease? The preva- women than men, in blacks than whites,
thermore, serum TSH concentrations be- lence of subclinical hypothyroidism in in the elderly,34 and in patients with low
tween 2.5 and 4.5 mIU/L may be due to the US adult population is about 4% to iodine intake.38 The presence of goiter,
minor technical problems in the TSH as- 8.5% in those without known thyroid personal history of previous thyroid dis-
say, circulating abnormal TSH iso- disease.1,23 The prevalence increases with ease, family history of thyroid disease,
forms, or heterophilic antibodies; nor- age, and in women older than 60 years, atrial fibrillation, or ingestion of iodine-
mal individuals with serum TSH subclinical hypothyroidism is present in containing drugs such as amiodarone all
concentrations in this range would be up to 20%.1,33,34 The data are less consis- make subclinical hyperthyroidism more
misidentified as having hypothyroid- tent in men; in those older than 65 years, likely. If the diagnosis is limited to only
ism. Given these concerns as well as the the prevalence increases and ap- those with a serum TSH level lower than
pulsatile nature and continuous distri- proaches that of women in some, but not 0.1 mIU/L, the prevalence of subclini-
bution of serum TSH concentrations, the all, studies.23 In patients found to have cal hyperthyroidism decreases to 0.7%.23
panel defined the reference range of nor- an elevated TSH level, approximately However, subclinical hyperthyroidism is
mal serum TSH concentration as 0.45 75% have values lower than 10 mIU/L.1 common in individuals treated with le-
to 4.5 mIU/L. The prevalence of subclinical hypothy- vothyroxine, being present in 14% to
What Is the Definition of Subclini- roidism in blacks is one third that in 21% of such patients.39,40
cal Hyperthyroidism? Subclinical hyper- whites,23 and a similar low prevalence is Overt hyperthyroidism is defined as
thyroidism is defined as a serum TSH con- seen in some populations with iodine de- a serum TSH level lower than 0.1 mIU/L
centration below the statistically defined ficiency.35,36 Factors such as previous hy- with serum FT4, T3, or FT3 concentra-
lower limit of the reference range when perthyroidism, type 1 diabetes melli- tions above the normal reference range.
2004 American Medical Association. All rights reserved. (Reprinted) JAMA, January 14, 2004Vol 291, No. 2 231

Downloaded from www.jama.com on December 13, 2006


SUBCLINICAL THYROID DISEASE

sis and MI with a risk comparable to that


Table 1. Quality of Evidence on the Strength of Association and Risks/Benefits of Treatment
of Subclinical Hypothyroidism associated with diabetes mellitus, hyper-
Strength of Association Benefits of Treatment cholesterolemia, and smoking. How-
ever, a longitudinal component of this
Serum TSH Serum TSH Serum TSH Serum TSH study did not confirm an increased risk
Clinical Condition 4.5-10 mIU/L 10 mIU/L 4.5-10 mIU/L 10 mIU/L
Progression to overt Good Good * *
of MI in patients with subclinical hypo-
hypothyroidism thyroidism, and a 10-year cohort study50
Adverse cardiac end points Insufficient Insufficient No evidence No evidence found no such association. No random-
Elevations in serum total Insufficient Fair Insufficient Insufficient ized studies have assessed the impact of
and LDL cholesterol levothyroxine replacement on impor-
Cardiac dysfunction Insufficient Insufficient Insufficient tant clinical cardiac end points. Many
Systemic hypothyroid symptoms None Insufficient Insufficient Insufficient small interventional trials demonstrate
Neuropsychiatric symptoms None Insufficient Insufficient Insufficient
improvement in cardiac function, but
Abbreviations: LDL, low-density lipoprotein; TSH, thyroid-stimulating hormone.
*Thyroid hormone therapy normalizes serum TSH at any TSH concentration. Overt hypothyroidism occurs earlier in these changes are of uncertain clinical
untreated patients with serum TSH 10 mIU/L than in those with serum TSH between 4.5 and 10 mIU/L. importance. Although results of several
Data did not distinguish between serum TSH concentrations between 4.5 and 10 mIU/L and 10 mIU/L.
studies suggest that thyroid hormone
therapy will reduce total and LDL cho-
Few persons with a serum TSH be- those with a serum TSH higher than 10 lesterol levels in individuals with sub-
tween 0.1 and 0.45 mIU/L progress to mIU/L, when such data were avail- clinical hypothyroidism,51 this finding
overt hyperthyroidism,24,34 whereas 1% able. The panel examined the quality has not been confirmed in RCTs.56,62
to 2% per year of those with serum TSH of the evidence for the strength of an SYSTEMIC SYMPTOMS. Baseline data
lower than 0.1 mIU/L develop overt hy- association with certain adverse con- from an RCT55 illustrated an increased
perthyroidism.34 Serum TSH normal- sequences of subclinical hypothyroid prevalence of hypothyroid symptoms
izes in many of these individuals over disease and the quality of the evidence among individuals with subclinical hy-
time.41-43 Patients with large nodular addressing the risks and benefits of pothyroidism. One health fairbased
thyroids and subnormal serum TSH treatment (Table 1). cross-sectional study of 25862 partici-
concentrations may be at particular risk C ARDIAC D YSFUNCTION AND A D- pants1 reported more hypothyroid symp-
for developing overt hyperthyroidism VERSE EVENTS. Numerous small studies toms in individuals with subclinical hy-
when exposed to high concentrations reveal subtle decreases in myocardial pothyroidism than in euthyroid
of iodine.44 contractility detected by echocardiogra- individuals but fewer symptoms than in
phy in patients with subclinical hypo- overtly hypothyroid individuals. This
Subclinical Hypothyroidism: thyroidism.48 However, limitations in study was not population-based and did
Questions and Recommendations study design, including lack of blind- not distinguish between untreated sub-
What Are the Consequences of Un- ing, inclusion of patients with previous clinical hypothyroidism and under-
treated Subclinical Hypothyroidism? overt hyperthyroidism, and possible se- treated overt hypothyroidism. Other
Possible consequences of subclinical hy- lection bias in control group patients, cross-sectional58,59 and case-control stud-
pothyroidism include cardiac dysfunc- temper the conclusion that clinically im- ies53 did not confirm these observa-
tion45-48 or adverse cardiac end points portant reductions in cardiac contractil- tions, but they were conducted among
(including atherosclerotic disease and ity can be expected in subclinical hypo- selected or referred populations often in-
cardiovascular mortality),49,50 eleva- thyroidism. In addition, these studies volving elderly hospitalized patients.
tion in total and low-density lipopro- included individuals with high serum TREATMENT. A double-blind RCT re-
tein (LDL) cholesterol,51,52 systemic hy- TSH without further stratification. It re- ported significant improvement in
pothyroid symptoms1,53-60 or mains to be determined whether a con- symptomatic patients with subclinical
neuropsychiatric symptoms,1,56,57 and tinuum of cardiac dysfunction can be ex- hypothyroidism treated with levothy-
progression to overt, symptomatic hy- pected across the spectrum of subclinical roxine compared with placebo.55 How-
pothyroidism24,61 (TABLE 1). hypothyroidism. ever, study patients were primarily
Assessment of Evidence. The litera- Whether untreated subclinical hypo- treated hyperthyroid patients, includ-
ture on subclinical hypothyroidism of- thyroidism affects important cardiovas- ing those with serum TSH concentra-
ten arbitrarily separates patients into 2 cular outcomes such as angina pecto- tions in the 40 to 50 mIU/L range. A sec-
groups determined by the degree of se- ris, myocardial infarction (MI), and ond clinical trial reporting symptomatic
rum TSH elevation. To allow ease of cardiovascular death remains an unan- improvement used a crossover design
comparison with the published litera- swered question. One large, cross- but did not ensure that the treated pa-
ture, the panel assessed the evidence for sectional epidemiologic study49 con- tients were euthyroid rather than sub-
individuals with serum TSH concen- cluded that subclinical hypothyroidism clinically hyperthyroid.60 One study that
trations between 4.5 and 10 mIU/L and was a risk factor for aortic atherosclero- showed improvement in neuromuscu-
232 JAMA, January 14, 2004Vol 291, No. 2 (Reprinted) 2004 American Medical Association. All rights reserved.

Downloaded from www.jama.com on December 13, 2006


SUBCLINICAL THYROID DISEASE

lar symptoms and dysfunction when pa- cal action is always either indicated or efit. Still, the panel considers the likeli-
tients were treated with levothyroxine contraindicated. As the serum TSH con- hood of improvement small, and it must
was not an RCT. 54 The 2 RCTs re- centration increases above 10 mIU/L, be balanced against the inconvenience,
stricted to individuals with TSH levels however, the basis for initiating treat- expense, and potential risks of therapy.
lower than 10 mIU/L found no im- ment is more compelling. Clinical con- Physicians and patients must under-
provement in symptoms with levothy- text is particularly important. This opin- stand that there is insufficient evidence
roxine therapy.56,57 ion reflects clinical experience and to expect therapeutic benefit in patients
judgment as well as the literature that in this group and that distinguishing a
How Should Subclinical suggests improvement in symptoms55 true therapeutic effect from a placebo
Hypothyroidism Be Evaluated? and possible lowering of LDL choles- effect in an individual patient is diffi-
If the serum TSH concentration is high terol.56 There are no studies that dem- cult. Still, the possibility that some pa-
and serum FT4 concentration has not onstrate decreased morbidity or mortal- tients may benefit cannot be ruled out.
been measured, the TSH measure- ity with treatment. The potential risks of Physicians and patients should under-
ment should be repeated along with an therapy are limited to the development stand the natural history of subclinical
FT4 measurement at a minimum of 2 of subclinical hyperthyroidism, which hypothyroidism and the small but defi-
weeks, but no longer than 3 months, may occur in 14% to 21% of individuals nite risk of progression to overt hypo-
after the initial assessment. The panel treated with levothyroxine.39,40 thyroidism. The special case of preg-
recommends thyroid hormone therapy Subclinical Hypothyroidism With Se- nancy or the planned pregnancy in
in individuals with elevated serum TSH rum TSH of 4.5 to 10 mIU/L. Although women with subclinical hypothyroid-
concentrations whose FT4 concentra- some studies suggest an association be- ism is discussed below.
tion is below the reference range (0.8- tween subclinical hypothyroidism and Subclinical Hypothyroidism With Se-
2.0 ng/dL [10.3-25.7 pmol/L]). systemic hypothyroid symptoms1,55 or rum TSH Higher Than 10 mIU/L. Levo-
If a high serum TSH concentration cardiac dysfunction, 4 8 others do thyroxine therapy is reasonable for pa-
is confirmed on repeat testing and se- not.53,58,59 No population-based studies tients with subclinical hypothyroidism
rum FT4 is within the reference range, examined symptoms in patients with se- and serum TSH higher than 10 mIU/L.
the patient should be evaluated for signs rum TSH concentrations between 4.5 The rate of progression is 5% in com-
and symptoms of hypothyroidism, pre- and 10 mIU/L. The likelihood of pro- parison with patients with lower lev-
vious treatment for hyperthyroidism gression to overt hypothyroidism ap- els of TSH, and treatment may poten-
(radioiodine, partial thyroidectomy), pears to be higher than for those with tially prevent the manifestations and
thyroid gland enlargement, or family TSH levels lower than 4.5 mIU/L14 (Table consequences of hypothyroidism in
history of thyroid disease. Lipid pro- 1). Although early levothyroxine therapy those patients who do progress. Still,
files should be reviewed. Women who does not alter the natural history of the the evidence that therapy will reduce
are pregnant or hope to become preg- disease, it may prevent symptoms and total and LDL cholesterol levels and im-
nant in the near future deserve special signs of overt disease in those who do prove symptoms in these patients is in-
consideration. progress. The available data do not con- conclusive (TABLE 2).
The evidence was insufficient to rec- firm clear-cut benefits for early therapy Subclinical Hypothyroidism During
ommend either for or against routine compared with treatment when symp- Pregnancy. The panel gave a rating of
measurement of anti-TPO antibodies in toms or overt hypothyroidism de- fair to the evidence of an association
patients with subclinical hypothyroid- velop56,57 (Table 1). Therefore, the panel between subclinical hypothyroidism and
ism. The presence of anti-TPO antibod- does not recommend routine levothy- adverse outcomes of pregnancy for either
ies identifies an autoimmune etiology for roxine treatment for patients with TSH the fetus or the mother. However, the
thyroid dysfunction and predicts a levels between 4.5 and 10 mIU/L, but panel made the following recommen-
higher risk of developing overt hypo- thyroid function tests should be re- dation: a TSH level might be obtained
thyroidism (4.3% per year vs 2.6% per peated at 6- to 12-month intervals to in pregnant women and women who
year in antibody-negative individuals).24 monitor for improvement or worsen- wish to become pregnant if they have a
Still, antibody presence or absence does ing in TSH level. family or personal history of thyroid dis-
not change the diagnosis of subclinical The panel realizes that some individu- ease, physical findings or symptoms sug-
hypothyroidism (which is based on se- als with TSH levels between 4.5 and 10 gestive of goiter or hypothyroidism, type
rum TSH measurements) or the ex- mIU/L have symptoms compatible with 1 diabetes mellitus, or a personal his-
pected efficacy of treatment. hypothyroidism. Clinicians and pa- tory of autoimmune disorders. For
What Are the Risks and Benefits of tients may decide on a several-month trial women who already take levothyrox-
Treating Subclinical Hypothyroid- of levothyroxine, while monitoring for ine but whose TSH level is in the sub-
ism? Among patients with untreated sub- improvement in hypothyroid-type symp- clinical hypothyroidism range, compli-
clinical hypothyroidism, there is no toms. Continuation of therapy should be ance and appropriateness of dose should
single level of serum TSH at which clini- predicated on clear symptomatic ben- be assessed.
2004 American Medical Association. All rights reserved. (Reprinted) JAMA, January 14, 2004Vol 291, No. 2 233

Downloaded from www.jama.com on December 13, 2006


SUBCLINICAL THYROID DISEASE

to the approach taken in many reported


Table 2. Quality of Evidence on the Strength of Association and Benefits of Treatment of
Subclinical Hyperthyroidism studies, the panel classified patients with
Strength of Association Benefits of Treatment subclinical hyperthyroidism into 2 cat-
egories: those with mildly low but de-
Serum TSH Serum TSH Serum TSH Serum TSH tectable serum TSH (0.1-0.45 mIU/L)
Clinical Condition 0.1-0.45 mIU/L 0.1 mIU/L 0.1-0.45 mIU/L 0.1 mIU/L
Progression to overt Insufficient Good None None
and those with a clearly low serum TSH
hyperthyroidism (0.1 mIU/L). In all clinical settings,
Adverse cardiac end points apart Fair * None None causes of subnormal serum TSH con-
from atrial fibrillation centration other than subclinical hyper-
Atrial fibrillation Insufficient Good None None thyroidism must be excluded.
Cardiac dysfunction Insufficient Fair * Insufficient Cardiac Dysfunction, Arrhythmias, and
Systemic hyperthyroid and Insufficient Insufficient None Insufficient Adverse Events. Exogenous and endog-
neuropsychiatric symptoms
Reduced bone mineral density None Fair None Fair
enous subclinical hyperthyroidism have
Fractures None Insufficient None None
been reported to increase heart rate, left
Abbreviation: TSH, thyroid-stimulating hormone. ventricular (LV) mass, and cardiac con-
*Data did not distinguish between serum TSH concentrations of 0.1 to 0.45 mIU/L and TSH lower than 0.1 mIU/L. tractility, to cause diastolic dysfunction
Noted particularly in postmenopausal women and women with previous history of overt hyperthyroidism.
(delayed relaxation), and atrial arrhyth-
mias, but not to increase the prevalence
Pregnant women or women of child- in the lower half of the reference range of ventricular arrhythmias.48 The panel
bearing potential planning to become is controversial because there are no data interpreted reported echocardio-
pregnant who are found to have el- demonstrating improved clinical out- graphic changes in cardiac function in
evated serum TSH should be treated with comes with this strategy. Nevertheless, persons with subclinical hyperthyroid-
levothyroxine to restore the serum TSH when the serum TSH is in the upper half ism vs comparison groups to be small
concentration to the reference range. This of the reference range and levothyroxine- and of uncertain clinical importance.
recommendation is based on the pos- treated patients continue to note symp- Only 3 small and unblinded studies have
sible association between high TSH and toms suggestive of hypothyroidism, it is assessed cardiac function in patients with
either increased fetal wastage or subse- reasonable to increase the levothyrox- endogenous subclinical hyperthyroid-
quent neuropsychological complica- ine dosage to bring the serum TSH into ism.71,73,74 Few studies included strati-
tions occurring in the offspring due to the lower portion of the reference range. fied analyses for patients whose TSH was
thyroid insufficiency.63 Although there The rapidity of the dosage adjustment in the 0.1 to 0.45 mIU/L range. One study
are no published intervention trials as- depends on the patients age and medi- reported increased all-cause (up to 2.2-
sessing the benefits of thyroid hormone cal comorbidities. Minimal TSH eleva- fold) and cardiovascular mortality (up to
replacement in this special population, tions may not require dosage adjust- 3-fold) in individuals older than 60 years
the potential benefit-risk ratio of levo- ment in patients who feel well, with endogenous subclinical hyperthy-
thyroxine therapy justifies its use. It is particularly those with arrhythmias or roidism and serum TSH concentration
important to note that the requirement other cardiac disorders. lower than 0.5 mIU/L.50
for levothyroxine in treated hypothy- One study reported a 3-fold in-
roid women frequently increases dur- Subclinical Hyperthyroidism: creased risk of atrial fibrillation over 10
ing pregnancy. Therefore, serum TSH Questions and Recommendations years in men and women at least 60 years
concentration should be monitored ev- What Are the Consequences of Un- of age with serum TSH of 0.1 mIU/L or
ery 6 to 8 weeks during pregnancy and treated Subclinical Hyperthyroidism? lower with endogenous and exogenous
the levothyroxine dose modified as Assessment of Evidence. The panel evalu- subclinical hyperthyroidism. Although
needed. The risks of appropriately man- ated the strength of the evidence for the some believe the risk of atrial fibrilla-
aged levothyroxine therapy in preg- association of untreated subclinical hy- tion is also increased in patients with a
nancy are minimal. Continuation of le- perthyroidism and the following clini- serum TSH level of 0.1 to 0.4 mIU/L, evi-
vothyroxine treatment post partum is cal outcomes: progression to overt hy- dence for this is limited, whereas evi-
beyond the scope of this discussion. perthyroidism,1,64-67 adverse cardiac end dence for an increased risk of atrial fi-
Subclinical Hypothyroidism in Treated points,50 atrial fibrillation,43,68,69 cardiac brillation when the TSH value is lower
Overt Hypothyroid Individuals. When dysfunction,48,70-76 systemic and neuro- than 0.1 mIU/L is solid.43 A second study
subclinical hypothyroidism is noted in psychiatric symptoms,42,71,72,77,78 re- found a 5-fold increased risk of atrial fi-
levothyroxine-treated patients with overt duced bone mineral density,79-87 and frac- brillation in individuals with endoge-
hypothyroidism, the dosage of levothy- tures 88,89 (Table 2). The panel also nous subclinical hyperthyroidism (age
roxine should be adjusted to bring the assessed the strength of the association not specified) and TSH lower than 0.4
serum TSH into the reference range. between the TSH level and the risks and mIU/L compared with euthyroid indi-
Whether the target TSH level should be benefits of treatment (Table 2). Similar viduals aged at least 45 years.69 A third
234 JAMA, January 14, 2004Vol 291, No. 2 (Reprinted) 2004 American Medical Association. All rights reserved.

Downloaded from www.jama.com on December 13, 2006


SUBCLINICAL THYROID DISEASE

study of individuals with endogenous sity (BMD) during prolonged subclini- total T3 or FT3 levels to exclude cen-
subclinical hyperthyroidism (mean age, cal hyperthyroidism.83,84 These analy- tral hypothyroidism or nonthyroidal ill-
65 years) and serum TSH lower than 0.1 ses concluded that exogenous ness. Clinical circumstances dictate
mIU/L reported a 2.8-fold increased risk subclinical hyperthyroidism resulted in when the retesting should occur. For
of atrial fibrillation over 2 years com- a significant loss of BMD among post- patients with atrial fibrillation, car-
pared with aged-matched euthyroid con- menopausal women but not among pre- diac disease, or other serious medical
trols.68 No studies demonstrated an in- menopausal women. Considering the conditions, repeat testing within 2
creased incidence of arterial embolism in data from individual studies, one pro- weeks is prudent. When these factors
patients with subclinical hyperthyroid- spective study found no association be- are absent, repeat testing is recom-
ism. However, atrial fibrillation due to tween low serum TSH and accelerated mended within 3 months.
overt hyperthyroidism is a known risk loss in BMD.85 One study reported no If the repeat serum TSH concentra-
factor for arterial embolism.90,91 increased fracture risk in levothyroxine- tion remains higher than 0.1 but lower
Treatment. Several studies have as- treated patients with serum TSH lower than 0.45 mIU/L, with normal FT4 and
sessed the effects of treatment on car- than 0.05 mIU/L or in those with se- T3 concentrations, and the patient has
diac function. Successful treatment of rum TSH between 0.05 and 4.0 mIU/L.93 no signs or symptoms of cardiac dis-
endogenous subclinical hyperthyroid- However, another prospective study re- ease, atrial fibrillation, or other arrhyth-
ism decreased the heart rate and car- ported an increased risk of hip and mias, retesting should occur at 3- to 12-
diac output and increased the systemic spine fracture in levothyroxine- month intervals, until either serum TSH
vascular resistance in an unblinded study treated women older than 65 years normalizes or the clinician and patient
of 6 patients.74 There is limited evi- whose serum TSH was 0.1 mIU/L or are confident that the condition is stable.
dence that treatment of subclinical hy- lower, but this study did not distin- Patients with known nodular thyroid
perthyroidism facilitates spontaneous re- guish between overt and subclinical hy- disease may develop overt hyperthy-
version or cardioversion of atrial perthyroidism.88 The risk of fractures roidism when exposed to excess iodine
fibrillation to normal sinus rhythm.92 was not increased in women with se- (eg, radiographic contrast agents) and
Among patients with exogenous sub- rum TSH between 0.1 and 0.5 mIU/L require special consideration.44
clinical hyperthyroidism, decreasing the when adjustment was made for prior Individuals With a Serum TSH Lower
levothyroxine dose normalized the heart hyperthyroidism. Overt thyrotoxico- Than 0.1 mIU/L. If serum TSH concen-
rate and resulted in a nonsignificant re- sis increased the risk of fracture in tration is lower than 0.1 mIU/L, the
duction in LV ejection fraction, 75 most 88,89 but not all 94 studies. Pro- panel recommends repeating the mea-
whereas -blockers decreased atrial pre- longed subclinical hyperthyroidism surement, along with an FT4 and a total
mature beats and LV mass index and im- prior to overt hyperthyroidism may T3 or FT3, within 4 weeks of the initial
proved diastolic filling.76 contribute to the increased risk of frac- measurement. If the patient has signs
Systemic and Neuropsychiatric Symp- ture in patients with thyrotoxicosis.89 or symptoms of cardiac disease, atrial
toms. Several relatively small case- Treatment. Treatment of hyperthy- fibrillation or other arrhythmia, or
control, cross-sectional, and cohort stud- roidism to restore the TSH level to within medical issues requiring urgent diag-
ies found more hyperthyroid-type signs the reference range preserves BMD, but nosis and treatment, these tests should
and symptoms in individuals with sub- normalization of bone turnover may be be performed within a shorter inter-
clinical hyperthyroidism (compared delayed for up to 1 year.86,87 Two stud- val, particularly if there are signs or
with euthyroid individuals) but fewer ies of endogenous subclinical hyperthy- symptoms of hyperthyroidism.
than in individuals with overt hyper- roidism (TSH 0.2 mIU/L81 and TSH Endogenous Subclinical Hyperthyroid-
thyroidism.42,71,72,77 However, some of 0.1 mIU/L 82 ) in postmenopausal ism (TSH Lower Than 0.45 mIU/L). The
these studies involved patients se- women demonstrated significant con- panel recommends further evaluation
lected from hospital clinics or elderly in- tinued bone loss in untreated patients to establish the etiology of the low se-
patients. The only large, population- compared with bone stabilization in rum TSH. A radioactive iodine uptake
based study (N=6884) of an unselected, treated patients. Only one82 of these stud- measurement and scan can distin-
healthy cohort found no association be- ies was randomized and neither in- guish between destructive thyroiditis
tween those with TSH lower than 0.21 cluded a placebo group. and hyperthyroidism due to Graves dis-
mIU/L (who were not taking levothy- How Should Subclinical Hyperthy- ease or nodular goiter.
roxine) and physical or psychological roidism Be Evaluated? Individuals With What Are the Risks and Benefits of
symptoms of hyperthyroidism; nor were Serum TSH 0.1 to 0.45 mIU/L Not Treatment of Subclinical Hyperthy-
differences in concentration, depres- Treated With Levothyroxine. If serum roidism? The risks of treatment of sub-
sion, or anxiety detected by means of TSH is reported to be between 0.1 and clinical hyperthyroidism with antithy-
validated instruments.78 0.45 mIU/L, the measurement should roid drugs are potential allergic
Skeletal System. Two meta-analyses be repeated for confirmation. The panel reactions including agranulocytosis. Ra-
reported declines in bone mineral den- recommends measuring FT4 and either dioactive iodine therapy commonly
2004 American Medical Association. All rights reserved. (Reprinted) JAMA, January 14, 2004Vol 291, No. 2 235

Downloaded from www.jama.com on December 13, 2006


SUBCLINICAL THYROID DISEASE

causes hypothyroidism and may cause Endogenous Subclinical Hyperthyroid- lated to the test being applied to deter-
exacerbation of hyperthyroidism or ism (Serum TSH Lower Than 0.1 mIU/L). mine the persons likelihood of having
Graves eye disease.95 Subclinical hyperthyroidism due to a particular disease or condition.
Exogenous Subclinical Hyperthyroid- destructive thyroiditis (including post- Thyroid dysfunction is more preva-
ism With TSH 0.1 to 0.45 mIU/L. When viral subacute thyroiditis and postpar- lent in certain population groups, in-
the serum TSH concentration is be- tum thyroiditis) resolves spontane- cluding women older than 60 years, per-
tween 0.1 and 0.45 mIU/L in a levothy- ously. Treatment, apart from sons with previous radiation treatment
roxine-treated individual, the indica- symptomatic therapy (eg, -blockers), is of the thyroid gland (radioactive iodine
tion for thyroid hormone therapy should usually not required. or therapeutic external beam radiation),
be reviewed. Many patients with thy- The panel recommends that treat- those who have had previous thyroid sur-
roid cancer and some patients with thy- ment be considered for subclinical hy- gery or thyroid dysfunction, and those
roid nodules require TSH suppression, perthyroidism (TSH 0.1 mIU/L) due who have type 1 diabetes mellitus, a per-
and the target TSH level should be re- to Graves or nodular thyroid disease. sonal history of autoimmune disease, a
viewed by the treating endocrinologist The panel recognizes the paucity of in- family history of thyroid disease, or atrial
or other physician. When levothyrox- tervention trials, apart from those dem- fibrillation. The panel recommends ag-
ine is prescribed for hypothyroidism in onstrating stabilization of bone den- gressive case finding in these high-risk
the absence of thyroid nodules or thy- sity. However, the panel was concerned groups. The panel also endorses thy-
roid cancer, the panel recommends de- about the risk of atrial fibrillation and/or roid function testing (serum TSH mea-
creasing the dosage of levothyroxine to bone loss, particularly in the elderly. Spe- surement) for patients seeking medical
allow serum TSH to increase toward the cifically, treatment should be consid- care who have signs or symptoms sug-
reference range. This dosage adjust- ered for patients who are older than 60 gestive of thyroid dysfunction8 or those
ment may be particularly important years and for those with or at increased being evaluated for palpable thyroid ab-
when the serum TSH is in the lower part risk for heart disease, osteopenia, or os- normalities.
of the range (Table 2). teoporosis (including estrogen- The panel recommends against popu-
Exogenous Subclinical Hyperthyroid- deficient women), or for those with lation-based screening for thyroid dis-
ism With TSH Lower Than 0.1 mIU/L. symptoms suggestive of hyperthyroid- ease. Case ascertainment in certain high-
When the serum TSH concentration is ism. Younger individuals with subclini- risk groups is encouraged. The panel
lower than 0.1 mIU/L in a levothyroxine- cal hyperthyroidism and serum TSH per- finds the evidence insufficient to recom-
treated individual, the indication for thy- sistently (months) lower than 0.1 mIU/L mend for or against routine determina-
roid hormone therapy should be may be offered therapy or follow-up de- tion of TSH levels (screening) in preg-
reviewed. For patients with thyroid can- pending on individual considerations. nant women or women planning to
cer and thyroid nodules, the target serum Is Screening for Subclinical Thy- become pregnant. It is reasonable to con-
TSH value should be reviewed by the roid Disease Warranted? The ratio- sider serum TSH measurement for
patients endocrinologist or treating phy- nale for population screening hinges on women with a family history of thyroid
sician. When levothyroxine is pre- the high prevalence of subclinical thy- disease, prior thyroid dysfunction, symp-
scribed for hypothyroidism in the absence roid dysfunction in the adult popula- toms or physical findings suggestive of
of thyroid nodules or thyroid cancer, the tion and on the potential health ben- hypothyroidism or hyperthyroidism, an
panel recommends decreasing the dos- efits and risks of detecting and treating abnormal thyroid gland on examina-
age of levothyroxine to allow serum TSH these diseases. We used the US Preven- tion, type 1 diabetes mellitus, or a per-
to increase toward the reference range. tive Services Task Force criteria96 for rec- sonal history of an autoimmune disorder.
Endogenous Subclinical Hyperthyroid- ommending a screening test, which re-
ism (Serum TSH 0.1-0.45 mIU/L). The quires evidence of effectiveness of early CONCLUSION
panel recommends against routine detection. One of the most important cri- Our review of the literature revealed a
treatment for all patients whose TSH is teria for recommending a screening test striking paucity of evidence bearing on
mildly decreased (0.1-0.45 mIU/L). The is that screening asymptomatic per- the major clinical questions examined.
panel found insufficient evidence to es- sons and treating them for the condi- Our recommendations are based on the
tablish a clear association between this tion should result in improved measur- existing evidence and the panels clini-
mild degree of hyperthyroidism and ad- able and important health outcomes cal experience, but they are limited by
verse clinical outcomes, including atrial when compared with persons who are the paucity of definitive data. Well-
fibrillation. However, because of a pos- not screened and who present with signs conceived and executed intervention
sible association with increased cardio- or symptoms of the disease. An alterna- trials are needed to bring definitive data
vascular mortality,50 clinicians might tive to population screening is aggres- to light on these questions. Until such
consider treatment of elderly individu- sive case finding, defined as the appli- data are available, clinical judgment and
als, despite the absence of supportive cation of a test to a person presenting patients preferences remain para-
data from intervention trials. to a clinician for a reason usually unre- mount. Although the panel recom-
236 JAMA, January 14, 2004Vol 291, No. 2 (Reprinted) 2004 American Medical Association. All rights reserved.

Downloaded from www.jama.com on December 13, 2006


SUBCLINICAL THYROID DISEASE

mended against population screening for ment of AHRQ or the federal government. Neither do 19. Brennan MD, Klee GG, Preissner CM, Hay ID. Het-
statements made in this article reflect official policy erophilic serum antibodies: a cause for falsely el-
subclinical thyroid disease, clinicians are of the American Thyroid Association, the American As- evated serum thyrotropin levels. Mayo Clin Proc. 1987;
encouraged to make individual patient sociation of Clinical Endocrinologists, and the Endo- 62:894-898.
crine Society organizers of the consensus develop- 20. Wood JM, Gordon DL, Rudinger AN, Brooks MM.
assessments when determining the need ment conference. Artifactual elevation of thyroid-stimulating hor-
for testing and treatment. Previous Presentation: Presented in part at the Ameri- mone. Am J Med. 1991;90:261-262.
can Thyroid Association annual meeting, October 9-13, 21. Ward G, McKinnon L, Badrick T, Hickman PE. Het-
2002, Los Angeles, Calif. erophilic antibodies remain a problem for the immu-
Author Affiliations: Departments of Medicine and Pa- Acknowledgment: The panel thanks David S. Coo- noassay laboratory. Am J Clin Pathol. 1997;108:417-
thology, Montefiore Medical Center and the Albert per, MD, for help in the organization of the Consen- 421.
Einstein College of Medicine, Bronx, NY (Dr Surks); sus Guidelines for the Diagnosis and Management of 22. Beck-Peccoz P, Amr S, Menezes-Ferreira MM, Fa-
Center for Primary Care, Prevention, and Clinical Part- Subclinical Thyroid Disease and Patricia A. Stephens, glia G, Weintraub BD. Decreased receptor binding of
nerships, Agency for Healthcare Research and Qual- PhD, for editorial assistance. biologically inactive thyrotropin in central hypothyroid-
ity, Rockville, Md (Dr Ortiz); Thyroid Unit and De- ism: effect of treatment with thyrotropin-releasing hor-
partment of Medicine, Massachusetts General Hospital, mone. N Engl J Med. 1985;312:1085-1090.
Harvard Medical School, Boston (Dr Daniels); Veter- 23. Hollowell JG, Staehling NW, Flanders WD, et al.
REFERENCES
ans Administration, Washington, DC (Dr Sawin); Har- Serum TSH, T4, and thyroid antibodies in the United
vard Medical School, Brigham and Womens Hospi- 1. Canaris GJ, Manowitz NR, Mayor G, Ridgway EC. States population (1988 to 1994): National Health and
tal, Division of Womens Health and Department of The Colorado Thyroid Disease Prevalence Study. Arch Nutrition Examination Survey (NHANES III). J Clin En-
Medicine, Boston (Dr Col); Department of Medicine, Intern Med. 2000;160:526-534. docrinol Metab. 2002;87:489-499.
Mount Sinai School of Medicine, New York, NY (Dr 2. Cooper DS. Clinical practice: subclinical hypothy- 24. Vanderpump MP, Tunbridge WM, French JM, et
Cobin); Division of Medical Sciences, University of Bir- roidism. N Engl J Med. 2001;345:260-265. al. The incidence of thyroid disorders in the commu-
mingham, Birmingham, England (Dr Franklyn); En- 3. Toft AD. Clinical practice: subclinical hyperthyroid- nity: a twenty-year follow-up of the Whickham Sur-
docrinology and Diabetes Division, West Los Ange- ism. N Engl J Med. 2001;345:512-516. vey. Clin Endocrinol (Oxf ). 1995;43:55-68.
les VA Medical Center, University of California at Los 4. Surks MI, Ocampo E. Subclinical thyroid disease.
25. Uy HL, Reasner CA, Samuels MH. Pattern of re-
Angeles School of Medicine (Dr Hershman); Endo- Am J Med. 1996;100:217-223.
covery of the hypothalamic-pituitary thyroid axis fol-
crine Section, Washington Hospital Center, Washing- 5. McDermott MT, Ridgway EC. Clinical perspec-
lowing radioiodine therapy in patients with Graves
ton, DC (Dr Burman); Department of Medicine, Uni- tive: subclinical hypothyroidism is mild thyroid failure
disease. Am J Med. 1995;99:173-179.
versity of Texas Southwestern Medical Center at Dallas and should be treated. J Clin Endocrinol Metab. 2001;
26. Glinoer D, De Nayer P, Bourdoux P, et al. Regu-
(Dr Denke); Research Development, Mayo Founda- 86:4585-4590.
lation of maternal thyroid during pregnancy. J Clin En-
tion, Rochester, Minn (Dr Gorman); Epidemiology, 6. Chu JW, Crapo LM. Clinical perspective: the treat-
docrinol Metab. 1990;71:276-282.
Loyola University Medical School, Chicago, Ill (Dr Coo- ment of subclinical hypothyroidism is seldom neces-
27. Wehmann RE, Gregerman RI, Burns WH, Saral
per); and Cardiac Ultrasound, Washington Hospital sary. J Clin Endocrinol Metab. 2001;86:4591-4599.
R, Santos GW. Suppression of thyrotropin in the low-
Center, and Department of Medicine, Georgetown 7. Baskin HJ. American Association of Clinical Endo-
thyroxine state of severe nonthyroidal illness. N Engl
University Medical College, Washington, DC (Dr crinologists guidelines for clinical practice for the evalu-
Weissman). Dr Denke is now with the University of ation and treatment of hyperthyroidism and hypo- J Med. 1985;312:546-552.
Texas, San Antonio. thyroidism. Endocr Practice. 2002;8:458-467. 28. Franklyn JA, Black EG, Betteridge J, Sheppard MC.
Financial Disclosure: Dr Burman has served on the 8. Ladenson PW, Singer PA, Ain KB, et al. American Thy- Comparison of second and third generation methods
speakers bureau for Abbott Laboratories, as a con- roid Association guidelines for detection of thyroid dys- for measurement of serum thyrotropin in patients with
sultant for New River Pharmaceuticals, and as an ad- function. Arch Intern Med. 2000;160:1573-1575. overt hyperthyroidism, patients receiving thyroxine
visor to Norovax Inc. 9. US Preventive Services Task Force, Agency for therapy and those with non-thyroidal illness. J Clin
Author Contributions: Study concept and design: Healthcare Research and Quality Web site. Third Endocrinol Metab. 1994;78:1368-1371.
Surks, Ortiz, Daniels, Sawin, Cobin, Franklyn, Hersh- USPSTF strength of overall evidence. Available at: http: 29. Van den Berghe G, de Zegher F, Lauwers P. Do-
man, Burman, Cooper, Weissman. //www.ahrq.gov/clinic/3rduspstf/prostatecr/prostaterr pamine and the sick euthyroid syndrome in critical ill-
Acquisition of data: Surks, Ortiz, Daniels, Col. .htm#overall. Accessed December 9, 2002. ness. Clin Endocrinol (Oxf ). 1994;41:731-737.
Analysis and interpretation of data: Surks, Ortiz, 10. Spencer CA, Takeuchi M, Kazarosyan M. Current 30. Benker G, Raida M, Olbricht T, Wagner R, Rein-
Daniels, Sawin, Col, Cobin, Franklyn, Hershman, status and performance goals for serum thyrotropin hardt W, Reinwein D. TSH secretion in Cushings syn-
Burman, Denke, Gorman. (TSH) assays. Clin Chem. 1996;42:2051-2052. drome: relation to glucocorticoid excess, diabetes, goi-
Drafting of the manuscript: Surks, Ortiz, Daniels, Sawin, 11. National Academy of Clinical Biochemistry Web ter, and the sick euthyroid syndrome. Clin Endocrinol
Col, Cobin, Franklyn, Burman, Denke, Gorman, site. NACB laboratory medicine practice guidelines. (Oxf ). 1990;33:777-786.
Cooper. Available at: http://www.nacb.org/lmpg/main.stm. 31. Hangaard J, Andersen M, Grodum E, Koldkjaer
Critical revision of the manuscript for important in- Accessed December 9, 2002. O, Hagen C. Pulsatile thyrotropin secretion in pa-
tellectual content: Surks, Ortiz, Daniels, Sawin, Col, 12. Ross DS. Serum thyroid-stimulating hormone mea- tients with Addisons disease during variable gluco-
Cobin, Franklyn, Hershman, Burman, Denke, Gorman, surement for assessment of thyroid function and dis- corticoid therapy. J Clin Endocrinol Metab. 1996;81:
Weissman. ease. Endocrinol Metab Clin North Am. 2001;30:245- 2502-2507.
Statistical expertise: Surks, Sawin, Col. 264. 32. Lee E, Chen P, Rao H, Lee J, Burmeister LA. Effect
Administrative, technical, or material support: Surks, 13. Surks MI, Oppenheimer JH. Metabolism of phe- of acute high dose dobutamine administration on se-
Ortiz, Franklyn, Burman, Gorman. nolic- and tyrosyl-ring labeled L-thyroxine in human rum thyrotropin (TSH). Clin Endocrinol (Oxf ). 1999;
Study supervision: Surks, Sawin, Cobin, Burman, beings and rats. J Clin Endocrinol Metab. 1971;33: 50:487-492.
Gorman, Weissman. 612-618. 33. Sawin CT, Castelli WP, Hershman JM, et al. The
Funding/Support: Abbott Laboratories was the first 14. Wong ET, Bradley SG, Schultz AL. Elevations of aging thyroid: thyroid deficiency in the Framingham
and principal sponsor. Other supporters were Diag- thyroid-stimulating hormone during acute nonthy- Study. Arch Intern Med. 1985;145:1386-1388.
nostic Products Corporation, Laboratory Corpora- roidal illness. Arch Intern Med. 1981;141:873-875. 34. Parle JV, Franklyn JA, Cross KW, Jones SC, Shep-
tion of America, Tosoh Medics Inc, ARUP Laborato- 15. Bhakri HL, Fisher R, Khadri A, MacMahon DG. pard MC. Prevalence and follow-up of abnormal thy-
ries, and Dade Behring Inc. Funds were donated as Longitudinal study of thyroid function in acutely ill el- rotrophin (TSH) concentrations in the elderly in the United
unrestricted educational grants to the Endocrine So- derly patients using a sensitive TSH assaydefer test- Kingdom. Clin Endocrinol (Oxf ). 1991;34:77-83.
ciety, which coordinated the meeting. ing until recovery. Gerontology. 1990;36:140-144. 35. Robuschi G, Safran M, Braverman LE, et al. Hy-
Role of the Sponsors: The sponsors had no role in plan- 16. Gharib H, Hodgson SF, Gastineau CF, Scholz DA, pothyroidism in the elderly. Endocr Rev. 1987;8:142-
ning the meeting, selection of participants, data col- Smith LA. Reversible hypothyroidism in Addisons dis- 153.
lection or analysis, or manuscript preparation. None ease. Lancet. 1972;2:734-736. 36. Kung AW, Janus ED. Thyroid dysfunction in am-
of the authors had a financial relationship with the 17. Ismail AA, Burr WA, Walker PL. Acute changes bulatory elderly Chinese subjects in an area of bor-
sponsors. Some did give lectures and received hono- in serum thyrotropin in treated Addisons disease. Clin derline iodine intake. Thyroid. 1996;6:111-114.
raria that were occasionally provided by unrestricted Endocrinol (Oxf ). 1989;30:225-230. 37. Sawin CT. Subclinical hypothyroidism in older per-
educational grants from the sponsors. 18. Ladenson PW, Braverman LE, Mazzaferri EL, et al. sons. Clin Geriatr Med. 1995;11:231-238.
Disclaimer: Dr Ortiz is a federal government em- Comparison of administration of recombinant human 38. Laurberg P, Pedersen KM, Vestergaard H, et al.
ployee with the Agency for Healthcare Research and thyrotropin with withdrawal of thyroid hormone for ra- High incidence of multinodular toxic goiter in the el-
Quality (AHRQ). Statements made in this publica- dioactive iodine scanning in patients with thyroid car- derly population in a low iodine intake area vs high
tion do not represent the official policy or endorse- cinoma. N Engl J Med. 1997;337:888-896. incidence of Graves disease in the young in a high

2004 American Medical Association. All rights reserved. (Reprinted) JAMA, January 14, 2004Vol 291, No. 2 237

Downloaded from www.jama.com on December 13, 2006


SUBCLINICAL THYROID DISEASE

iodine intake area: comparative surveys of thyrotoxi- 57. Kong WM, Sheikh MH, Lumb PJ, et al. A 6-month and somatic changes [in German]. Klin Wochenschr.
cosis epidemiology in East-Jutland Denmark and Ice- randomized trial of thyroxine treatment in women with 1987;65:264-273.
land. J Intern Med. 1991;229:415-420. mild subclinical hypothyroidism. Am J Med. 2002; 78. Schlote B, Schaaf L, Schmidt R, et al. Mental and
39. Ross DS, Daniels GH, Gouveia D. The use and limi- 112:348-354. physical state in subclinical hyperthyroidism: investi-
tations of a chemiluminescent thyrotropin assay as a 58. Bemben DA, Hamm RM, Morgan L, Winn P, Davis gations in a normal working population. Biol Psychia-
single thyroid function test in an outpatient endocrine A, Barton E. Thyroid disease in the elderly, part 2: pre- try. 1992;32:48-56.
clinic. J Clin Endocrinol Metab. 1990;71:764-769. dictability of subclinical hypothyroidism. J Fam Pract. 79. Foldes J, Tarjan G, Szathmari M, et al. Bone min-
40. Parle JV, Franklyn JA, Cross KW, Jones SR, Shep- 1994;38:583-588. eral density in patients with endogenous subclinical hy-
pard MC. Thyroxine prescription in the community: 59. Lindeman RD, Schade DS, LaRue A, et al. Sub- perthyroidism: is this thyroid status a risk factor for os-
serum thyroid stimulating hormone level assays as an clinical hypothyroidism in a biethnic urban commu- teoporosis? Clin Endocrinol (Oxf ). 1993;39:521-527.
indicator of undertreatment or overtreatment. Br J Gen nity. J Am Geriatr Soc. 1999;47:703-709. 80. Foldes J, Lakatos P, Zsadanyi J, Horvath C. De-
Pract. 1993;43:107-109. 60. Nystrom E, Caidahl K, Fager G, Wikkelso C, Lund- creased serum IGF-I and dehydroepiandrosterone sul-
41. Sawin CT, Geller A, Kaplan MM, et al. Low se- berg PA, Lindstedt G. A double-blind cross-over 12- phate may be risk factors for the development of re-
rum thyrotropin (thyroid-stimulating hormone) in older month study of L-thyroxine treatment of women with duced bone mass in postmenopausal women with
persons without hyperthyroidism. Arch Intern Med. subclinical hypothyroidism. Clin Endocrinol (Oxf ). endogenous subclinical hyperthyroidism. Eur J Endo-
1991;151:165-168. 1988;29:63-76. crinol. 1997;136:277-281.
42. Stott DJ, McLellan AR, Finlayson J, et al. Elderly 61. Huber G, Staub JJ, Meier C, et al. Prospective study 81. Faber J, Jensen IW, Petersen L, et al. Normaliza-
patients with suppressed serum TSH but normal free of the spontaneous course of subclinical hypothyroid- tion of serum thyrotrophin by means of radioiodine
thyroid hormone levels usually have mild thyroid over- ism: prognostic value of thyrotropin, thyroid reserve, treatment in subclinical hyperthyroidism: effect on bone
activity and are at increased risk of developing overt and thyroid antibodies. J Clin Endocrinol Metab. 2002; loss in postmenopausal women. Clin Endocrinol (Oxf ).
hyperthyroidism. QJM. 1991;78:77-84. 87:3221-3226. 1998;48:285-290.
43. Sawin CT, Geller A, Wolf PA, et al. Low serum 62. Caraccio N, Ferrannini E, Monzani F. Lipopro- 82. Mudde AH, Houben AJ, Nieuwenhuijzen Kruse-
thyrotropin concentrations as a risk factor for atrial fi- tein profile in subclinical hypothyroidism: response to man AC. Bone metabolism during antithyroid drug
brillation in older persons. N Engl J Med. 1994;331: levothyroxine replacement, a randomized placebo- treatment of endogenous subclinical hyperthyroid-
1249-1252. controlled study. J Clin Endocrinol Metab. 2002;87: ism. Clin Endocrinol (Oxf ). 1994;41:421-424.
44. Stanbury JB, Ermans AE, Bourdoux P, et al. Iodine- 1533-1538. 83. Faber J, Galloe AM. Changes in bone mass dur-
induced hyperthyroidism: occurrence and epidemiol- 63. Haddow JE, Palomaki GE, Allan WC, et al. Ma- ing prolonged subclinical hyperthyroidism due to L-
ogy. Thyroid. 1998;8:83-100. ternal thyroid deficiency during pregnancy and sub- thyroxine treatment: a meta-analysis. Eur J Endocri-
45. Monzani F, DiBello V, Caraccio N, et al. Effect of sequent neuropsychological development of the child. nol. 1994;130:350-356.
levothyroxine on cardiac function and structure. J Clin N Engl J Med. 1999;341:549-555. 84. Uzzan B, Campos J, Cucherat M, et al. Effects on
Endocrinol Metab. 2001;86:1110-1115. 64. Wiersinga WM. Subclinical hypothyroidism and bone mass of long term treatment with thyroid hor-
46. Kahaly GJ. Cardiovascular and atherogenic as- hyperthyroidism, I: prevalence and clinical relevance. mones: a meta-analysis. J Clin Endocrinol Metab. 1996;
pects of subclinical hypothyroidism. Thyroid. 2000; Neth J Med. 1995;46:197-204. 81:4278-4289.
10:665-679. 65. Haden ST, Marqusee E, Utiger RD. Subclinical hy- 85. Bauer DC, Nevitt MC, Ettinger B, Stone K. Low
47. Biondi B, Fazio S, Palmieri EA, et al. Left ventricu- perthyroidism. Endocrinologist. 1996;6:322-327. thyrotropin levels are not associated with bone loss
lar diastolic dysfunction in patients with subclinical hy- 66. Marqusee E, Haden ST, Utiger RD. Subclinical thy- in older women: a prospective study. J Clin Endocri-
pothyroidism. J Clin Endocrinol Metab. 1999;84: rotoxicosis. Endocrinol Metab Clin North Am. 1998; nol Metab. 1997;82:2931-2936.
2064-2067. 27:37-49. 86. Kumeda Y, Inaba M, Tahara H, et al. Persistent
48. Biondi B, Palmieri EA, Lombardi G, Fazio S. Ef- 67. Samuels MH. Subclinical thyroid disease in the el- increase in bone turnover in Graves patients with sub-
fects of subclinical thyroid dysfunction on the heart. derly. Thyroid. 1998;8:803-813. clinical hyperthyroidism. J Clin Endocrinol Metab.
Ann Intern Med. 2002;137:904-914. 68. Tenerz A, Forberg R, Jansson R. Is a more active 2000;85:4157-4161.
49. Hak AE, Pols HA, Visser TJ, Drexhage HA, Hof- attitude warranted in patients with subclinical thyro- 87. Pantazi H, Papapetrou PD. Changes in param-
man A, Witteman JC. Subclinical hypothyroidism is an toxicosis? J Intern Med. 1990;228:229-233. eters of bone and mineral metabolism during therapy
independent risk factor for atherosclerosis and myo- 69. Auer J, Scheiber P, Mische T, et al. Subclinical hy- for hyperthyroidism. J Clin Endocrinol Metab. 2000;
cardial infarction in elderly women: the Rotterdam perthyroidism as a risk factor for atrial fibrillation. Am 85:1099-1106.
Study. Ann Intern Med. 2000;132:270-278. Heart J. 2001;142:838-842. 88. Bauer DC, Ettinger B, Nevitt M, Stone KL, for the
50. Parle JV, Maisonneuve P, Sheppard MC, Boyle P, 70. Biondi B, Fazio S, Cuocolo A, et al. Impaired cardiac Study of Osteoporotic Fractures Research Group. Risk
Franklyn JA. Prediction of all-cause and cardiovascu- reserve and exercise capacity in patients receiving long- for fracture in women with low serum levels of thyroid-
lar mortality in elderly people from one low serum thy- term thyrotropin suppressive therapy with levothyrox- stimulating hormone. Ann Intern Med. 2001;134:
rotropin result: a 10-year cohort study. Lancet. 2001; ine. J Clin Endocrinol Metab. 1996;81:4224-4228. 561-568.
358:861-865. 71. Biondi B, Palmieri EA, Fazio S, et al. Endogenous 89. Vestergaard P, Mosekilde L. Fractures in pa-
51. Danese MD, Ladenson PW, Meinert CL, Powe NR. subclinical hyperthyroidism affects quality of life and tients with hyperthyroidism and hypothyroidism: a na-
Effect of thyroxine therapy on serum lipoproteins in cardiac morphology and function in young and middle- tionwide follow-up study in 16,249 patients. Thy-
patients with mild thyroid failure: a quantitative re- aged patients. J Clin Endocrinol Metab. 2000;85: roid. 2002;12:411-419.
view of the literature. J Clin Endocrinol Metab. 2000; 4701-4705. 90. Staffurth JS, Gibberd MC, Fui SN. Arterial embo-
85:2993-3001. 72. Shapiro LE, Sievert R, Ong L, et al. Minimal car- lism in thyrotoxicosis with atrial fibrillation. BMJ. 1977;
52. Kanaya A, Harris F, Volpato S, et al. Association diac effects in asymptomatic athyrotic patients chroni- 2:688-690.
between thyroid dysfunction and total cholesterol level cally treated with thyrotropin (TSH)-suppressive doses 91. Presti CF, Hart RG. Thyrotoxicosis, atrial fibrilla-
in an older biracial population. Arch Intern Med. 2002; of L-thyroxine (T4). J Clin Endocrinol Metab. 1997; tion, and embolism, revisited. Am Heart J. 1989;117:
162:773-779. 82:2592-2595. 976-977.
53. Zulewski H, Muller B, Exer P, Miserez AR, Staub 73. Boutin JM, Matte R, DAmour P, et al. Charac- 92. Forfar JC, Feek CM, Miller HC, Toft AD. Atrial fi-
JJ. Estimation of tissue hypothyroidism by a new clini- teristics of patients with normal T3 and T4 and a low brillation and isolated suppression of the pituitary-
cal score: evaluation of patients with various grades TSH response to TRH. Clin Endocrinol (Oxf ). 1986; thyroid axis: response to specific antithyroid therapy.
of hypothyroidism and controls. J Clin Endocrinol 25:579-588. Int J Cardiol. 1981;1:43-48.
Metab. 1997;82:771-776. 74. Faber J, Wiinberg N, Schiffer S, Mehlsen J. Hae- 93. Leese GP, Jung RT, Guthrie C, Waugh N, Brown-
54. Monzani F, Caraccio N, DelGuerra P, et al. Neuro- modynamic changes following treatment of subclini- ing MC. Morbidity in patients on L-thyroxine: a compari-
muscular symptoms and dysfunction in subclinical hypo- cal and overt hyperthyroidism. Eur J Endocrinol. 2001; son of those with a normal TSH to those with a suppressed
thyroid patients: beneficial effect of L-T4 replacement 145:391-396. TSH. Clin Endocrinol (Oxf ). 1992;37:500-503.
therapy. Clin Endocrinol (Oxf ). 1999;51:237-242. 75. Grund FM, Niewoehner CB. Hyperthyroxinemia 94. Hallengren B, Elmstahl B, Berglund J, et al. No increase
55. Cooper DS, Halpern R, Wood LC, Levin AA, Ridg- in patients receiving thyroid replacement therapy. Arch in fracture incidence in patients treated for thyrotoxi-
way EC. L-thyroxine therapy in subclinical hypothy- Intern Med. 1989;149:921-924. cosis in Malmo during 1970-74: a 20-year population-
roidism: a double-blind, placebo-controlled trial. Ann 76. Fazio S, Biondi B, Carella C, et al. Diastolic dys- based follow-up. J Intern Med. 1999;246:139-144.
Intern Med. 1984;101:18-24. function in patients on thyroid-stimulating hormone 95. Weetman AP. Graves disease. N Engl J Med.
56. Meier C, Staub JJ, Roth CB, et al. TSH-controlled suppressive therapy with levothyroxine: beneficial effect 2000;343:1236-1248.
L-thyroxine therapy reduces cholesterol levels and clini- of beta-blockade. J Clin Endocrinol Metab. 1995;80: 96. US Preventive Services Task Force. Guide to Clini-
cal symptoms in subclinical hypothyroidism: a double- 2222-2226. cal Preventive Services, Third Edition: Periodic Up-
blind, placebo-controlled trial (Basel Thyroid Study). 77. Rockel M, Teuber J, Kaumeier S, et al. Correla- dates. Rockville, Md: Agency for Healthcare Re-
J Clin Endocrinol Metab. 2001;86:4860-4866. tion of latent hyperthyroidism with psychological search and Quality; 2002. Publication 02-500.

238 JAMA, January 14, 2004Vol 291, No. 2 (Reprinted) 2004 American Medical Association. All rights reserved.

Downloaded from www.jama.com on December 13, 2006