Explainer · Heart & Metabolic
Thyroid tablets moved almost nothing in older adults with a borderline result, across 286 people
Trials already showed no benefit to symptoms. This looked underneath, at cholesterol and 167 blood measures, on the theory that something subtle was being missed. Almost nothing moved.
- A further analysis of two randomized trials, covering 286 adults aged 65 and over.
- All had subclinical hypothyroidism: a raised thyroid signal with normal thyroid hormone.
- Levothyroxine produced no meaningful change in cholesterol or related blood fats over a year.
- A hint of benefit appeared in the 27 people whose thyroid signal was highest.
- This is about a borderline result in older adults, not about treating an underactive thyroid.
Somewhere in the filing system of most older adults’ medical records is a mildly abnormal thyroid result. The pituitary is telling the thyroid to work harder than usual, but the thyroid hormone itself is still in range, and there are no symptoms worth the name. Whether to treat that pattern has been argued over for thirty years, and millions of prescriptions have been written while the argument continued.
Randomized trials have already given an answer most prescribers found unsatisfying: treating it did not make people feel better. This analysis, in The Journal of Clinical Endocrinology and Metabolism, went looking for the effect those trials might have missed, in the blood rather than the symptoms, across 286 older adults. It found almost nothing.
What subclinical hypothyroidism actually is
It is a result, not a disease, and the name oversells it. MedlinePlus describes hypothyroidism proper as a condition in which the thyroid does not make enough thyroid hormones, producing tiredness, weight gain, cold intolerance and slowed thinking.
The subclinical version has the raised signal without the low hormone. In older adults it is common, it often resolves on its own, and the slightly higher readings that trigger it may partly be normal aging rather than a failing gland. That is the context for asking whether a daily tablet helps.
Why the thyroid analysis looked at blood fats
Because the obvious outcomes had already come back empty. The authors state it directly: randomized controlled trials found no cardioprotective effects of levothyroxine therapy in older adults with subclinical hypothyroidism.
One response to a null result is that the measurement was too blunt. Symptoms and events are noisy; blood chemistry is not. So the team went back to two double-blind trials and examined what the treatment did to seven clinically relevant lipid measures and 167 standardized metabolomic measures from nuclear magnetic resonance, which is about as sensitive a net as this question can be given.
Among 286 included participants, with a median age of 75, 142 were randomized to levothyroxine, and the comparison ran over twelve months.
What the thyroid treatment did to the numbers
Essentially nothing. Levothyroxine showed no effects on total cholesterol, non-HDL cholesterol, LDL cholesterol or triglycerides, with the confidence ranges sitting close to zero rather than merely failing to reach a threshold.
That is the informative version of a negative result. A wide, uncertain range would mean the study could not see. Narrow ranges centred near zero mean the effect, if any, is too small to matter.
One subgroup hinted otherwise. In participants with baseline thyroid stimulating hormone of 10 or above, a group of just 27 people, potentially beneficial changes appeared but did not hold up once the analysis corrected for the number of measures being tested.
Why the small thyroid subgroup is a hypothesis, not a finding
Twenty-seven people, 174 outcomes, and a correction that the result does not survive: those three facts together describe exactly the kind of signal that vanishes when somebody tries to reproduce it.
It is still the most interesting part of the paper, because it matches where the clinical guidelines already draw their line. Most recommend considering treatment when the thyroid signal is well above the threshold rather than merely at it, and this is consistent with that, without confirming it.
The honest description is that the people with the most abnormal results are the ones most likely to benefit, which nobody doubted, and that this analysis was not designed to establish it.
What two older-adult trials cannot settle
This was a post-hoc analysis, meaning the question was asked after the trials were finished and the data were in. That is a legitimate way to learn something and a poor way to prove it, because the outcomes were not what the trials were built to detect.
The participants were also all 65 and over, which is the group where the argument is sharpest, and nothing here applies to a 35-year-old with the same blood result, in whom the calculation is different.
Twelve months is also short for cholesterol changes to translate into anything that happens to a person, and the trials measured no heart attacks or strokes here.
What to do with a borderline thyroid result
Not much, in most cases, and that is the point. The accumulating evidence in older adults says a mildly raised thyroid signal with normal hormone is usually a number to watch rather than a problem to medicate, and this analysis closes off one of the last plausible arguments for treating it anyway.
What it does not mean is that thyroid treatment is useless. For genuine hypothyroidism, levothyroxine is one of the more clearly effective drugs in medicine. The question here is narrower: whether to start it on the strength of a borderline result in someone who feels fine, and the answer keeps coming back the same way. Anyone already taking it should raise the question with their prescriber rather than acting on an article.
People also ask
What did the study find?
Among 286 participants (48% women; median age 75), 142 were randomized to levothyroxine. Overall it showed no effects on apolipoprotein B, total cholesterol, non-HDL cholesterol, LDL cholesterol or triglycerides at 12 months. In the 27 participants whose baseline thyroid stimulating hormone was 10 mIU/L or above, potentially beneficial changes appeared but did not survive correction for multiple testing.
What is subclinical hypothyroidism?
A blood result rather than an illness: the pituitary signal telling the thyroid to work harder is raised, while the thyroid hormone itself is still in the normal range. Many people with it have no symptoms, and in older adults it is common and often drifts back on its own.
Why look at cholesterol rather than how people feel?
Because earlier randomized trials found no benefit to symptoms or quality of life, and one explanation offered was that the trials measured the wrong things. Blood fats are more sensitive than a questionnaire, so if a real effect existed it should show there.
What about the people with the highest thyroid signal?
There were only 27 of them, and the apparent benefit did not survive the statistical correction applied when many outcomes are tested at once. It is a hypothesis for a future trial, not a result.
Does this mean thyroid tablets do not work?
No. Levothyroxine works well for genuine hypothyroidism, where the thyroid hormone itself is low. This is only about the borderline pattern, and only in older adults.
Should anyone stop their thyroid medicine?
No. Stopping or starting levothyroxine is a decision for the clinician who ordered the test and knows the rest of the picture. This is general information rather than medical advice.