What TSH level is dangerously high? What the thyroid guidelines actually say

A high TSH means the pituitary is pushing a thyroid that is not keeping up. How hard it is pushing is useful information, but it is not a verdict. We searched the 2012 AACE and ATA hypothyroidism guideline and the 2014 ATA treatment guideline in full for any critical, alert or emergency TSH value. Neither has one.

The 2012 guideline goes further and rules the question out of its own scope. It covers people whose thyroid status has been stable for at least several weeks, and states plainly: they do not deal with myxedema coma, the life-threatening form of hypothyroidism. The sites on page one that print a single dangerous number, some saying 10 and some saying 20, are not quoting either guideline, because neither contains one.

What the guidelines do provide is more useful: one line where risk measurably changes, and a description of the situations that are urgent regardless of the number.

Recommendation 15 of the AACE and ATA guideline reads: patients whose serum TSH levels exceed 10 mIU/L are at increased risk for heart failure and cardiovascular mortality, and should be considered for treatment with L-thyroxine. Below 10, the guideline asks for an individual decision, weighing symptoms, TPO antibodies and cardiovascular disease or its risk factors. The guideline's narrative describes general agreement that primary hypothyroidism with a TSH above 10 should be treated.

Large cohort data line up with that recommendation. Two of the biggest pooled analyses, run by the Thyroid Studies Collaboration across tens of thousands of adults, compared people with a raised TSH and normal thyroxine against people with a normal TSH, and split their results by TSH band.

Read the table by its confidence intervals. Between 4.5 and 9.9, every interval in the table crosses 1.0, so neither study could separate coronary or heart failure events in those groups from people with a normal TSH. From 10 upward, both intervals sit clear of 1.0. In absolute terms, Rodondi's group with TSH 10 to 19.9 had 38.4 coronary events per 1,000 person-years against 20.3 in people with a normal TSH. Total mortality was not raised in that analysis. Coronary death was, at a hazard ratio of 1.42 in the 7.0 to 9.9 band and 1.58 in the 10 to 19.9 band.

The 2014 ATA treatment guideline reads the evidence the same way, citing a meta-analysis in which increased cardiovascular risk was primarily observed with TSH above 10, with potential effects above 7.

Many pages quote 20 as the dangerous level. In the primary literature, 20 appears in two places, and neither is a danger threshold.

First, the cohort studies above defined subclinical hypothyroidism as TSH 4.5 to 19.9 with normal thyroxine. Twenty is where their study population stopped, not where risk was shown to start.

Second, the AACE and ATA guideline notes that TSH may rise above normal, but generally below 20 mIU/L, during recovery from a non-thyroid illness. A TSH of 14 measured a week after a hospital stay can be part of that recovery rise, which is one reason the guideline recommends TSH testing in hospitalized patients only when there is real suspicion of thyroid disease, and why its definition of subclinical hypothyroidism applies only when thyroid function has been stable for weeks or more and there is no recent or ongoing severe illness.

A low free T4. The guideline defines overt hypothyroidism as an elevated TSH, usually above 10 mIU/L, together with a subnormal free T4. The free T4 is what separates a thyroid that is struggling from one that is failing. A high TSH on its own does not tell you which you have: in the Framingham study, 39 percent of adults over 60 with a TSH above 10 had a subnormal T4, which means most did not. If your result shows a high TSH and no free T4, the free T4 is the missing half of the answer.

How the person is doing. The authors of a 2014 diagnostic study describe myxedema coma as a decompensated state of extreme hypothyroidism with a high mortality rate if untreated. It is diagnosed clinically. A scoring system published in Endocrine Practice in 2014 adds up changes in body temperature regulation, the central nervous system, the heart, the gut and metabolism, plus whether a precipitating event is present. A score of 60 or more was potentially diagnostic, and 45 to 59 flagged patients at risk.

The 2014 ATA guideline is explicit on this point: the degree of TSH elevation may not be an accurate indicator of the severity of the hypothyroidism, based on the wide range of TSH values seen in myxedema coma. Older patients may show a smaller TSH rise, and when the cause is a pituitary that makes too little TSH, the TSH can look misleadingly unremarkable. The same guideline says treatment should start on clinical suspicion alone and should not wait for blood results.

Several things raise TSH without meaning the thyroid is failing, and the AACE and ATA guideline lists them:

• Interfering antibodies. Heterophilic antibodies, rheumatoid factor and antibodies against animal proteins (most often mouse) can produce falsely high TSH values. The 2016 ATA guideline says this should be ruled out by repeating the TSH in another assay.
• Recovery from illness. TSH may rise, generally below 20, as the body recovers.
• Pituitary tumors that do not make hormones. These can cause mild TSH elevations, generally not above 6 or 7, from a less active form of TSH, while the thyroid is actually under-stimulated.
• Adrenal insufficiency. TSH can rise and then fall back with steroid replacement.
• Age. The upper end of the normal TSH spread rises about 0.3 mIU/L for each decade after the thirties. We cover the age figures in the TSH normal range by age.

Biotin, the hair and nail supplement, is often blamed for odd thyroid results, but it pushes TSH the other way. The 2016 ATA guideline explains that excess biotin makes common TSH assays read falsely low and free T4 assays read falsely high. It does not explain a high TSH.

Place your number first. Our TSH of 10 or higher page covers the band above the guideline's line, and the subclinical zone page covers the range between your lab's upper limit and 10. The TSH reference page shows all four bands with their source and date, and the lab result lookup will place a specific value for you.

Then bring three questions to your clinician. What is my free T4? Have I been ill, started a new medicine or been in hospital in the past few weeks? And if my TSH is above 10 with a normal free T4, what do my antibodies and heart risk say about treating it? The guideline puts the yearly progression to overt hypothyroidism at 4.3 percent with positive thyroid antibodies and 2.6 percent without, so antibody status changes the conversation.

If your TSH is low rather than high, the reasoning runs differently and the danger signs are different; see what TSH level is dangerously low.