What does this mildly elevated TSH mean anyway ?
Allen Brantley, M.D.
Internal Medicine Resident Grand Rounds
January 12, 1999
MH is a 47 yowf who initially presented with c/o palpitations for the past 3-4 months.
She denied any chest pain or SOB. Further history revealed occasional nervousness as
well as a change in the patients bowel habits from “normal stools” to frequent loose
PMH: G2 P2
Fam Hx: No h/o early MI or CAD. ? h/o thyroid disease
Meds: None, including stimulants, decongestants, herbal drugs, or any alternative
PE: Vitals: Temp 98.6 BP 130/78 Pulse 108
Gen: Pleasant, NAD
HEENT: No proptosis or lid lag
Neck: Mild, diffuse enlargement of the thyroid gland, No defined nodules
CV: Tachycardic with regular rhythm, without M/G/R
Abd: Soft, NT
Ext: No edema
Neuro: Nonfocal. Reflexes symmetric and 3+
LABS: TSH < 0.5 mU/L(NL 0.5-5.0 mU/L), Total T4 15 g/dL(NL 5-11 g/dL)
Diagnosis: Hyperthyroidism, probable Graves’ disease
RP is a 58 yowf who presented with c/o fatigue, weakness, and weight gain of 30 pounds
over the last 4-6 months. Further history indicated cold intolerance, some difficulty with
concentration, and dry skin.
PMH: Tension headaches, Seasonal allergies with associated rhinitis
PSH: s/p C-section
FamHx: No history of thyroid disease
Meds: OTC pain relievers and Claritin D PRN
PE: Vitals: Temp 98.3 BP 142/89 Pulse 84
Weight 168 lbs.(130 lbs. approx. 8 months earlier)
Gen: Overweight wf, in NAD
HEENT: OP without lesions, tongue normal.
Neck: No JVD, no LAD. Thyroid gland not palpable.
CV: RRR without murmurs, gallops or rubs.
Abd: Positive bowel sounds, obese, soft, NT
Ext: 1+ pitting edema
Neuro: Nonfocal. Reflexes symmetric, but with slow relaxation phase.
Skin: dry and cool
Labs: TSH 23 mU/L, Total T4 2.1 g/dL, Antithyroid antibodies are positive
Diagnosis: Overt Hypothyroidism, probable Hashimoto’s disease
Subclinical hypothyroidism (SH) may be defined as a state in which an
asymptomatic person has a normal serum free T4 concentration in the face of an increased
serum TSH.1 Others define SH by an elevated TSH with normal T4, FT4 and T3 levels
(without mention of the presence or absence of symptoms).2 In contrast to SH, overt
hypothyroidism (OH) is suggested by an elevated TSH (usually a larger elevation than in
SH) and a decreased thyroxine level, often in the presence of clinical symptoms.3
The mean incidence of spontaneous OH is 3.5/1000/year in women and
0.6/1000/year in men, with an average age of diagnosis of 58-59.4 These numbers
indicate the well established gender differences in OH. The incidence of OH increases
with age, especially in women, with women ages 75-80 having an incidence of 14
The prevalence of TSH > 10 mU/L is estimated to be 7% in females and 3% in
males ages 60-89 years.5 The prevalence of increased TSH has been addressed in four
large population studies.1 The results of all four studies (see Table 11) show the
prevalence of increased TSH to be higher in women than in men and that prevalence
increased with age in both sexes. Within these four studies, fewer than 2% of those over
60 years of age had a decreased serum T4. This suggests that these values given for the
prevalence of elevated TSH, also approach the true prevalence for SH since
approximately 98% of the subjects had an increased TSH with a normal serum T41
Therefore the prevalence of SH can be estimated for females greater than 60 to be 11.6-
13.6 % and for males greater than 60 to be 2.9-5.7 %.
Prevalence of Increased Serum Thyrotropin Concentrations
Source No. of Sex Age, years Prevalence, %
Tunbridge et al. 1494 F >18 7.5
1977 1285 M >18 2.8
Bagchi et al. 694 F >55 8.5
1990 274 M >55 4.4
Sawin et al. 1260 F >60 13.6
1985 879 M >60 5.7
Parle et al. 683 F >60 11.6
1991 510 M >60 2.9
SH is the most common condition found by screening with 5-10% of adult women
having an elevated TSH.3 SH presents an interesting challenge to the primary care
physician or any health care provider who utilizes thyroid function tests because SH
involves mild elevations in TSH in frequently asymptomatic persons. So the question
becomes: What does this mildly elevated TSH mean anyway ?
Causes of Hypothyroidism:
Subclinical hypothyroidism is caused by the same disorders that cause overt
hypothyroidism. Most common among these is autoimmune thyroiditis - Hashimoto’s
disease - which is often associated with increased titers of antithyroid antibodies. These
include antithyroid microsomal antibodies (antithyroid peroxidase) and antithyroglobulin
antibodies. Antithyroid antibodies have important implications within the diagnosis of
SH in terms of the progression of SH to OH. This will be discussed in more detail later.
Table 2 below lists many of the common causes of hypothyroidism.5
Causes of Hypothyroidism
Chronic autoimmune thyroiditis
Treated Graves’ disease
Radioactive iodine therapy
Head and neck surgery
Radiation therapy to the head, neck, or chest area
Medications: lithium, iodine, amiodarone
Secondary hypothyroidism - hypopituitarism
In the hypothyroid patient, it is very important to take an accurate drug history.
Multiple drugs containing iodine, such as SSKI, amiodarone and radiocontrast
substances can increase TSH. Lithium and antithyroid drugs, such as methimazole and
propylthiouracil, can also have an effect on TSH. Insufficient thyroxine supplementation
in the hypothyroid patient will cause an elevated TSH.6
Potential Pathology/ Complications of Subclinical
Why should the physician care about a mildly elevated TSH in an otherwise
asymptomatic patient ? This is an excellent question, and one that is not easily answered.
Essentially there are three potential complications of SH. Two of these include
progression of SH to OH, and hyperlipidemia. The third involves the presence of various
symptoms, which may be related to the patient’s SH. (Certain authors include
hypothyroid-like symptoms in their definition of SH.)
Progression of Subclinical Hypothyroidism to Overt Hypothyroidism
First we will look at the progression of SH to OH. To begin, what is the course of
an initially elevated TSH as it is followed over time ? If one rechecks the TSH several
months later, one of three results occurs. The first is that the repeat TSH is normal. This
usually occurs as a result of the initial TSH being a lab error or can occur in the setting of
an episode of silent thyroiditis involving a transient hypothyroid phase. The second result
involves a similar TSH value, indicating no change in the SH. The third result is an
increase in the TSH (at times a marked increase), often with an increase in the patients
symptoms, suggesting progression to OH.5
The presence of antithyroid antibodies (ATA) is a strong indicator of who will
progress to OH.3 Females with SH who are positive for ATA have a relative risk (RR) of
38 for developing OH. This is compared to a RR of only 8 for similar women who are
negative for ATA. Females who have a normal TSH, but are positive for ATA also have
a RR of 8 for developing OH.4 One source suggests that patient’s with a TSH > 6 mU/L
and positive ATA, will progress to OH at a rate up to 10% per year.6
Table 3 shows the relationship between TSH and the percentage of patients who
demonstrate ATA.2 As the TSH increases, the percentage of patients positive for ATA
FT4 and TMAb abnormalities associated with elevated TSH
TSH FT4 -~ 0.6 ng/dL TMAb -> 1:400
(mlU/L) Number Number (%) Number (%)
5.1-7.4 1202 2 (0.2) 342 (28.5)
7.5-9.9 366 3 (0.8) 145 (39.6)
10.0-14.9 253 2 (0.8) 140 (55.3)
15.0-19.9 64 4 (6.3) 37 (57~.8)
->20 108 25 (23.1) 68 (63.0)
Total 1993 36 (1.8) 732 (36.7)
TSH = serum thyrotropin; FT,, = free thyroxine; TMAb = thyroid microsomal antibody
Older age and higher TSH level also increase a patient’s risk of conversion to
OH.3 In other words, younger women and women with lower TSH levels ( 6 to 9 mU/L)
have a lower risk for OH.3
Helfand and Redfern3 estimated the number needed to treat (NNT) to prevent one
asymptomatic patient from progressing to OH is 4.3 to 14.3. This implies that 77-93% of
the patients studied would take thyroid hormone replacement 5 years without evidence of
The most important issue, regarding the progression of SH to OH, is does
treatment actually improve outcome ? Unfortunately, no studies have been done looking
specifically at the progression of SH to OH. No studies have disclosed the intensity of
symptoms, the degree of disability, or the morbidity associated with being newly, overtly
hypothyroid.3 So while a NNT in the range above might seem reasonable, one must
wonder if the patient will benefit from early treatment, particularly if there are no
Close follow-up is an alternative to treatment. Since we can identify those
patients at greater risk for progression to OH based on their age, ATA status, and TSH
level, it makes sense that closer follow-up may be helpful. More routine examination and
screening should allow for earlier detection of OH and therefore a decrease in any
associated morbidity. Helfand and Redfern3 define a TSH level of greater than or equal
to 10 mU/L as being “markedly elevated” and seem to imply that this number may be a
useful cutoff, since progression to OH is this group is both common and rapid (usually in
the first few years after the TSH is obtained).
Does Treatment Improve Symptoms in Patients with Subclinical
Many patients, when screened, have at least one symptom that could be related to
subclinical hypothyroidism. Common symptoms include muscle cramps, dry skin, cold
intolerance, constipation, poor energy levels, fatigue and mental slowness.3 Three small
trials discussed below have looked at the relationship between SH and symptoms and the
benefit (or lack of) of treatment.
The first study by Cooper et al.7 was a double-blind, placebo-controlled trial of 33
patients (only one male) most of who previously had been treated for Graves’ disease and
now had SH. Patients were given a questionnaire on hypothyroid symptoms, specifically,
muscle cramps, dry skin, cold intolerance, constipation, poor energy, and easy fatigability.
Twenty euthyroid controls were also given the same questionnaire. It is important to note
that the euthyroid controls had one fewer symptom on average than did those with SH.
Each patient scored each symptom on a scale of -2 to +2 and symptoms for each patient
were added to give a total score.
After one year of placebo treatment, the placebo group had a mean symptom score
of -1.2 (Note: a mean score of 0 would indicate no change in symptoms for the group as
a whole) indicating an overall tendency for symptoms to remain essentially unchanged or
to worsen in some patients. The individual range was -8 to +4 suggesting that some
patients did worsen; however, others improved based on the positive scores, e.g. the “+4”
score. Within the treatment group, the mean symptom score was +2.1 with individual
scores ranging from -5 to +9. The trend in this group was for symptoms to improve,
although it is clear some patients worsened. Overall this study suggests that those treated
with thyroxine tend to have improvement, while those treated with placebo tend to have
either no change in their symptoms or a slight worsening of them.
The second study by Nystrom et al.8 involved a total of 20 women with SH.
These patients did not have a prior history of thyroid disease. The study was a double-
blind, cross-over, 12 month study comparing placebo to thyroxine treatment. Three
patients did not complete the study, one because of geographic relocation and two
because of nervousness and a sense of tachycardia, although none of the three had
objective signs of overtreatment.
All patients were given a questionnaire regarding symptoms (similar to the first
study) and also underwent various psychometric testing which included reaction time,
memory testing, and figure identification. Patients were also asked subjectively whether
they had feelings of improvement. The study protocol dictated that for treatment to be
deemed “beneficial” a patient must score better in at least 2 of the 3 psychometric tests
(listed above), and have a subjective improvement in well being. Based on these criteria,
only 4 of 17 (24%) women benefited from treatment. However 9 of 17 (53%)
experienced a subjective improvement.
The third study by Jaeschke et al.9 involves 37 patients (most female) over the age
of 55 with SH (TSH level greater than 6). It is a randomized, double-blind, placebo-
controlled trial. Outcome measures included: (1)chronic thyroid questionnaire - related to
physical complaints, mood and emotions, energy and well-being, and cognitive functions,
(2) symptom questionnaire, (3) sickness impact profile, and (4) a battery of cognitive
function testing. Patients in the treatment group were treated with thyroxine and the dose
was titrated to suppress the TSH level below 5.5.
This study did find an improvement in the composite memory score of the
treatment group compared with the placebo group. No other outcome measure, including
the symptom questionnaire, demonstrated a trend in favor of treatment. Based on this
study, symptoms are not improved with treatment. The authors of this paper suggest
close follow-up and reservation of treatment for those who develop OH.
The main drawback for all three of these studies was their sample size. The study
by Cooper et al.7 suggests that treatment with thyroxine is beneficial with regard to
improvement of symptoms. The study by Nystrom et al.8 is more equivocal and the
results by Jaeschke et al.9 clearly suggest no improvement of symptoms with active
treatment. Table 4 below is a modification of table 3 found in Helfand and Redfern3, and
summarizes the findings of the above three studies.
Summary of trials considering clinical symptoms
Cooper et al.7 Nystrom et al.8 Jaeschke et al.9
Study Design Rand, DB, PC Rand, DB, Crossover Rand, DB,PC
Population Previous I131 therapy Females > 50 with SH Patients > 55 with SH
Age Range (Mean) 32-78 (53) 51-64 (58) 55-86 (68)
Women/ Men 32/1 17/0 28/9
Final Sample Size (N) 33 17 32
Mean TSH (mU/L) 10.9 7.7 9.9
% Improved 20 N/A ---
% Worsened 40 N/A ---
% Improved 47 53 ---
% Worsened 24 --- ---
Overall Result Favors Treatment One-half Improved with No difference with
The study by Cooper et al.7 brings up two interesting points. First, these patients
were treated with radioactive iodine as a treatment for hyperthyroidism, and may not
represent most patients who develop SH as a result of early thyroid failure. Secondly, the
mean TSH was higher in this study than in the other two. This suggests that patients in
this study may in fact have been closer to OH, i.e. that their symptoms were in fact due to
hypothyroidism, and therefore should be more responsive to hormone replacement.
Overall the data is less than compelling. Helfand and Redfern3 conclude that for
every 100 patients treated for symptomatic SH, between 0 and 28 may receive benefit.
They suggest performing future studies with patients found by screening, and also
comment that including antithyroid antibodies as a marker of disease (discussed above)
may allow for more definitive results.
Clearly there is a relationship between overt hypothyroidism and
hyercholesterolemia and treatment of OH results in a significant decrease in total
cholesterol levels. The relationship between hypercholesterolemia and SH is much less
Tanis et al.10, performed a meta-analysis looking at the effects of thyroid hormone
replacement on hypercholesterolemia (HC) in patients with SH and OH. Thirteen studies
looking at SH with a total of 278 patients were examined. Initial TSH levels were on
average 3.5 times the upper limit of normal (range listed 4.8-29, with most being greater
than 10). The precision weighted decrease in cholesterol was 0.4 mmol/l (95% CI 0.2-
0.6). Twelve of the 13 studies demonstrated a decrease in cholesterol levels with
treatment. The weighted decrease in the cholesterol level was 6%. The effect was
somewhat higher in a subset of studies with a pretreatment cholesterol > 6.5 mmol/L (for
reference: 6.2 mmol/L = 240 mg/dL). Results in this subset indicated a reduction of 0.6
mmol/L (95% CI 0.3-0.8). Twelve of the 13 studies had a post-treatment, mean
cholesterol greater than 200. This would seem to indicate that even after treatment many
patients still had hypercholesterolemia and therefore were still at significant risk for
cardiovascular events. It is difficult to know whether the observed 6% decrease in
cholesterol actually translates into decreased cardiovascular risk when compared with
As one would expect, the decrease in cholesterol levels in patients with OH was
much more dramatic. The estimated, weighted average decrease was 2.5 mmol/l (95% CI
2.2-2.8). The range of cholesterol reduction among 18 studies (all of patients with OH)
was 15-47%. According to this study, the mean plasma level of total cholesterol
normalized in nearly all studies with thyroid hormone therapy.
In this study, the weighted decrease in cholesterol among patients with SH was
6%. Danese et al.11 references studies involving SH which suggest a 6% to 13%
reduction in total cholesterol with thyroid replacement. Treatment appears to be more
beneficial in patients with total cholesterol concentrations > 6.2mmol/L or in patients
with TSH levels > 10 mU/L. Clearly hypercholesterolemia found in SH is very different
from that found in OH. Patients with OH who are treated with thyroid hormone often
have normalization of their total cholesterol. This implies that the HC is in fact
secondary to their overt thyroid disease. In patients with SH, the reduction in cholesterol
with thyroid supplementation is less dramatic suggesting that HC in this setting is not
purely a result of thyroid dysfunction.10 These finding are in congruence with the
observation that most all patients with OH have HC, while a much smaller fraction of
patients with SH have similar cholesterol problems.
We can put a cholesterol reduction of 6% in perspective by comparing this figure
to some of the more conventional and standard therapies for HC. Bile acid-binding
resins, Niacin, and HMG-CoA reductase inhibitors reduce total cholesterol by 15-25%,
15-25% and 20-30% respectively.12 Standard therapies then, would seem to provide
more effective cholesterol reduction than would thyroid hormone therapy in the setting of
SH. It is unclear whether any trials have been performed comparing thyroid hormone to
more standard therapy for HC, in the setting of patients who have both HC and SH.
Despite the lack of direct comparison between therapies, it would seem unwise given the
current data, to rely solely on thyroid hormone replacement to reduce HC in patients with
How Should We Screen for Subclinical Hypothyroidism ?
The sensitive TSH assays represent the “gold standard for evaluating thyroid
hormone replacement levels in patients with an intact HPT axis.” This is a general
consensus opinion supported by multiple studies.2 If the diagnosis is uncertain, it can be
verified by measuring a free or total thyroixine.2
Should We Screen for Subclinical Hypothyroidism or Thyroid
Disease in General ?
Needless to say, the opinions on screening for mild thyroid failure (SH) are varied
and currently there is little agreement on what should be done. The US Preventative
Services Task Force (1996) recommends against screening asymptomatic adults.11
Danese et al.11 recommend measurement of TSH in patients aged 35 and older during
routine health examinations. Hefland et al.13 (writing a position paper for the ACP)
expresses that “it is reasonable to screen women older than 50 years of age for
unsuspected but symptomatic thyroid disease.” Helfand et al. also recommend
measurement of a free thyroxine level if the TSH is undetectable or is 10mU/L or more.
As is common with other forms of screening, it is difficult to know what to do with all
these varied recommendations.
Should We Treat Subclinical Hypothyroidism ?
To determine if we should treat SH, let’s return to the former three potential
complications of SH. First, we will consider the conversion of SH to OH. Unfortunately,
no studies have been performed to assess the morbidity associated with being newly,
overtly hypothyroid. In an asymptomatic patient, it seems counterintuitive to treat that
patient for an unknown length of time for something that has not yet developed.
Secondly, let’s consider symptoms. The three studies above are really too small
to draw any definite conclusions. The benefit of treating patients with SH in an effort to
improve symptoms, often very nonspecific, is questionable at best. The differential
diagnosis for common patient complaints is vast, and one should not expect or be
comfortable with the idea that SH will often explain common symptomatology.
Finally, we return to HC. The average reductions in total cholesterol by the
treatment of SH are modest. This doesn’t imply that a patient’s cardiovascular risk might
not improve with treatment. However, treatment with thyroid replacement is not likely to
reduce a patient’s cardiac risk to the same degree that standard therapies potentially can.
There are several studies2,4,5 which either state or have algorithms suggesting
treatment if the TSH is > than 10mU/L, particularly if symptoms suggesting
hypothyroidism are present. The presence of antithyroid antibodies, independent of TSH
level, deserves at least strong consideration for treatment5, and according to the other two
studies2,4, is sufficient to warrant treatment.
According to Helfand et al.13, “the available evidence is not sufficient to
recommend for or against treatment of subclincial hypothyroidism.” This paper does
expose particular patient populations that either are more likely or less likely than the
population as a whole to respond to treatment. First, women greater than 50 years of age
with SH who have markedly elevated TSH levels > 10mU/L have the highest risk of
complications. Secondly, women with a TSH greater than 10mU/L and a cholesterol
concentration greater than or equal to 6.2 mmol/L, tend to experience greater reduction in
cholesterol levels with thyroid replacement. Thirdly, other groups including younger
women, men and those with only mild elevations of TSH have a lower risk for
complications and the benefits of treatment are even less certain. Within these groups,
there is no strong evidence that treatment is effective or ineffective in terms of relieving
How Should We Treat Subclincial Hypothyroidism ?
Typical effective doses in SH may be as low as 25 to 50 ug/day. Dosage
adjustments can be made based on the TSH, keeping in mind that it takes 6 to 8 weeks for
the TSH to reach steady-state after a dosage change.5
Potential Complications of Treatment
There are essentially two potential complications from the overtreatment of
subclinical hypothyroidism. These are the same complications seen with the
overtreatment of OH, atrial fibrillation and osteoporosis.
Even mild hyperthyroidism can cause atrial fibrillation.14 Helfand and Redfern3
suggest that one additional case of atrial fibrillation may occur in every 114 patients
treated with thyroid hormone in amounts sufficient to suppress TSH. Woeber1
determined that a low TSH concentration conferred a 3-fold increased risk for the
development of atrial fibrillation. Keeping the TSH in the normal range can minimize the
risk for atrial fibrillation.
Osteoporosis can be a complication of thyroid hormone excess. Faber and
Galloe found that overtreatment leading to a reduced TSH resulted in annual bone loss
of 0.91% after 9.9 years when compared to controls. Ross16 suggests that short-term
treatment of SH in postmenopausal women with appropriate doses of thyroid replacement
is not associated with loss of bone density. Guo et al.17 concluded that bone turnover is
related to TSH and that a reduction in thyroid replacement dose can result in decreased
bone turnover and increased bone mineral density.
There is no evidence that treating SH to prevent OH actually reduces morbidity.
Treating symptoms in the context of SH may or may not be beneficial. The physician
should not use thyroid replacement in place of SSRI’s, anxiolytics, or other drugs
The benefits of thyroid replacement in patients who have SH and HC appear to be
minimal. If a patient clinically requires diet or lifestyle modification and/or standard
medication for HC, then he or she should receive it despite mild thyroid dysfunction.
Screening of men and women less than 50, is probably not useful unless the patient is
symptomatic without other explanation, or there is another clinical screening
indication such as unexplained weight gain.
Older age (> 50 years of age), female gender, TSH elevations greater than 10mU/L,
and the presence of antithyroid antibodies all confer greater chance of progression to
OH and to complications related to the thyroid.
In an equivocal situation, checking for the presence of (and titer of) antithyroid
antibodies may help the physician make prognosis and treatment decisions.
Among patients with SH, women older than 50 years of age with TSH levels greater
than 10, and women who have TSH levels greater than 10 as well as cholesterol
concentrations greater than or equal to 6.2 mmol/L are more likely to benefit from
Younger women, men and patients with a mildly elevated TSH level (6 to 9 mU/L)
are less likely to benefit from thyroid replacement.
Potential complications from the overtreatment of SH include atrial fibrillation and
osteoporosis. Both complications can be minimized by maintaining the TSH within
the normal range.
72 yowf presents with c/o fatigue, constipation, and dry skin. TSH is 13.4 mU/L, Total
T4 7.2 g/dL and total cholesterol is 270. Antithyroid antibodies are positive.
Treatment with thyroid replacement in this patient is reasonable and may be beneficial.
The patient is an older female with antithyroid antibodies, factors which increase her risk
of progression to OH. Additionally, she has a cholesterol > 6.5 mmol/dL and her
symptoms may represent early thyroid failure.
Conclusion: Favors Treatment.
37 yowf present with c/o depression, fatigue, and mental slowness. TSH 8.2 mU/L and
Total T4 is 6.9 g/dL. Antithyroid antibodies are negative.
Treatment with thyroid replacement in this patient is less reasonable. She is in a group of
patients who are less likely to benefit from treatment. She is young, with mildly elevated
TSH, and without antithyroid antibodies. Her risk of progression to OH is unlikely. If
her symptoms are felt to be secondary to clinical depression, then she should be treated
Conclusion: Favors No Treatment.