Primary Care Corner with Geoffrey Modest MD: ?Thyroid meds for subclinical hypothyroidism in older adults

​by Dr Geoffrey Modest

A randomized controlled trial assessed the effect of levothyroxine therapy in older adults with subclinical hypothyroidism, finding no clear benefit (see DOI: 10.1056/NEJMoa1603825 )


— 737 adults at least 65-years-old who had persistent subclinical hypothyroidism were randomized to levothyroxine starting at a dose of 50 µg a day, or 25 µg if their body weight were <50 kg or had coronary heart disease, with subsequent dose adjustment to achieve a TSH between 0.4 and 4.6, versus placebo.

— Subclinical hypothyroidism was defined as TSH of 4.60-19.99, with a free thyroxine level that was within the normal reference range.

— Mean age 74, 54% women, 98% white, 97% in non-sheltered community housing, 50% with hypertension/15% diabetes/14% ischemic heart disease/13% osteoporosis/9% smokers, mean baselineTSH level was 6.4

— Primary outcomes were changes in the Hypothyroid Symptoms score and in the Tiredness score on the thyroid related quality-of-life questionnaire, at one year (range of each score was 0-100, with higher scores indicating more symptoms, and the minimum clinically important difference being 9 points for each scale). Baseline Hypothyroid Symptom score was 17, Tiredness score was 26

— Secondary outcomes included changes in generic health-related quality-of-life, comprehensive thyroid related quality-of-life, hand grip strength, executive cognitive function (as assessed with the letter-digit coding test which indicates the speed of processing according to the number of correct responses in matching 9 letters with 9 digits in 90 seconds), blood pressure, weight, BMI, waist circumference, activities of daily living, instrumental activities of daily living, fatal/nonfatal cardiovascular events.


— Mean TSH level decreased from 6.4 to 5.5 in the placebo group as compared to 3.6 in the levothyroxine group, with median dose of 50 µg of levothyroxine. This was achieved within 6-8 weeks after starting the medication.

— There was no difference in the mean change at one year in the Hypothyroid Symptom score (0.2 for each group).

— There was no significant difference in the change in the Tiredness score (3.2 in those on levothyroxine, 3.8 in those on placebo)

— There was no benefit in any of the secondary outcome measures

— There were extended outcomes assessed for about half the patients, at a median of 24.2 months, also finding no difference in the primary or secondary outcomes

— Adverse events: no difference



— Subclinical hypothyroidism is an important issue for a few reasons:

— It is very common, between 8 and 18% of adults > 65yo

— Thyroid hormone acts throughout the body with receptors pretty much everywhere, affecting cognition, skeletal muscle function, vascular tree and heart, skeletal muscle, bone, etc, etc

— There are epidemiologic data suggesting that patients with subclinical hypothyroidism are at increased risk of coronary heart disease and perhaps heart failure. Data on total mortality are mixed.

— To me, there is a fundamental contradiction in the term “subclinical hypothyroidism”, since the normal limits of free T4 level reflects the bell-shaped curve of the community lab values, whereas TSH reflects the individual person’s response to their own circulating hormone levels. And patients may not be asymptomatic (ie “subclinical”). Subclinical hypothyroidism therefore, I think, just reflects a low level of hypothyroidism, such that depression of T4 levels still remains within the community norm, but still could have effects on that individual’s body.

— Other studies have found that treating subclinical hypothyroidism has shown improvements in the Tiredness score. These studies have been small and somewhat underpowered, and often with younger patients.

— About half of the patients with subclinical hypothyroidism will progress to overt hypothyroidism with a low serum thyroxine level over 10 to 20 years, with an annual progression rate of 2 to 4%. However some also have spontaneous recovery, less likely in those that are anti-TPO antibody positive

— Some limitations of the study which limit its generalizability:

​– The study tested a pretty uniform demographic (white people with stable housing and not a lot of medical comorbidities)

— The median achieved TSH level was 3.6, and some people believe that a more reasonable target is between 0.4 and 2.5 (i.e., it is possible that there would have been a measurable effect if they had achieved the lower and perhaps optimal TSH concentration)

— Very few people had TSH levels > 10, with a mean only slightly above the normal range, meaning that most of the patients had really mild hypothyroidism, which has a lower likelihood of progressing further or being symptomatic

— Hypothyroid symptom levels at trial entry were also quite low to begin with

— The trial was underpowered to detect an effect on cardiovascular events or mortality

— They did not measure thyroid antibody levels (which do predict to some extent which patients are more likely to progress to hypothyroidism)

— They did not find any difference in the speed of information processing, which has been found to be slowed in persons with subclinical hypothyroidism. However they did not assess other measures of cognitive function, though these are typically pretty blunt instruments (MMSE, MOCA, etc) and might not pick up very subtle though potentially important changes for the person and family/supports. But treating the subclinical hypothyroidism might still make a real difference for the individual, especially in the long-term (and a 65 year old in otherwise good health has a 20ish year life expectancy)

So, what is one to do with older patients who have subclinical hypothyroidism? The answer is not entirely clear, and this study really only adds the finding that short-term treatment of essentially asymptomatic patients with minimal laboratory abnormalities suggesting hypothyroidism does not seem to be effective. And I am particularly concerned about the potential links with atherosclerosis and cognitive decline. ​My sense is that it does seem reasonable to treat people with higher TSH levels (e.g.>10) since they have a higher likelihood of progressing to overt hypothyroidism. In those with lower TSH levels, it might be reasonable to check anti-TPO levels and treat those patients. It also might be reasonable to treat those who are more symptomatic than in this trial. But, overall, if one elects not to treat, it does make sense to follow these patients closely to see if they progress to more thyroid dysfunction. But one concern I have is that the usual symptoms of hypothyroidism, on the one hand, are pretty nonspecific, and, on the other hand, in many cases reveal themselves so slowly over time that patients may accommodate to them and not even notice them (though their treatment might still positively affect their quality-of-life).


for review and critique of thyroid screeninng guidelines, see here