The Levothyroxine Information Gap

Levothyroxine is not a drug you can take casually with breakfast. It is a narrow-therapeutic-index medicine: small differences in the amount absorbed can move the body between under- and over-replacement. The label says so directly:

“SYNTHROID has a narrow therapeutic index. Over- or undertreatment with SYNTHROID may have negative effects on growth and development, cardiovascular function, bone metabolism, reproductive function, cognitive function, emotional state, gastrointestinal function, and glucose and lipid metabolism.”

The same warning appears across the class. And the absorption is not merely “take it on an empty stomach” simple. The 2025 European Thyroid Association guidance summarises the physiology:

“The optimal absorption of LT4 is under fasting conditions, with 62–82% of the dose absorbed during the first 3 h after ingestion… Some types of foods and dietary supplements have been reported to reduce LT4 absorption, including soy, coffee, calcium, and grapefruit.”

That list is the practical core. Coffee, calcium, iron, soy, grapefruit, fibre, proton-pump inhibitors, antacids, bile-acid sequestrants and a long list of other medicines all change how much levothyroxine reaches the blood. The ATA’s 2014 guideline noted the same, including a case series of patients whose TSH rose when they took their tablet with espresso coffee — and, remarkably, that the effect of a multivitamin “does not appear to have been studied”, even though its calcium and ferrous content would be expected to impair absorption.

Malabsorption is real. So is pseudomalabsorption.

Patients who remain hypothyroid on a high dose are often assumed to be non-adherent. Sometimes they are. But the two possibilities are distinguishable: a levothyroxine absorption test can separate true malabsorption from “pseudomalabsorption” — intentional or unintentional non-adherence — and the ETA guidance recommends the test when clinically appropriate after other causes have been excluded. The existence of the test matters because it tells patients something the label does not: your experience of not absorbing the drug is a recognised clinical problem, not a character flaw.

When patients say “my thyroid is treated but I still feel unwell”, they are describing something the literature knows about. The 2022 Nature Reviews Endocrinology review puts the strict figure at 10–15% of levothyroxine-treated patients with persistent symptoms and dissatisfaction despite treatment. The ETA’s 2025 guidance, looking at real-world data, notes that almost half of treated patients have a TSH outside the reference range in the observational studies cited — low or high, over- or under-replacement. And the classic community study by Canaris and colleagues found that 40% of patients taking thyroid medication had an abnormal TSH.

Why does the distinction matter? Because “1 in 4” can be made to sound like one thing when it is really the overlap of several: patients with a dose that is too low, patients with a dose that is too high, patients whose absorption is impaired by food or another medicine, patients whose timing is wrong, and patients who take the tablet exactly as prescribed and still feel unwell. The persistent-symptom group — the one for whom “the drug is working on paper but not in the body” is the defining problem — is the 10–15% group. That is still millions of people worldwide, and it is not trivial.

“Approximately 10–15% of individuals with hypothyroidism treated with levothyroxine experience persistent symptoms and dissatisfaction with therapy (that might or might not be due to their hypothyroidism), which can lead to overtreatment.”

The symptoms in this group are not exotic: tiredness, low mood, poor concentration, apathy, weight change, and a sense of not being oneself. They overlap almost completely with depression. That overlap is where the next chapter begins.

The 2025 European Thyroid Association guideline is unusually direct about timing. It does not simply say “take it in the morning”. It says the timing should fit the patient’s life, and it gives two acceptable windows:

“The task force recommends that the timing of levothyroxine administration should be in keeping with the patient’s lifestyle. The interval between drug and food or drink intake must be not less than 30 min in any case. However, to obtain an optimal therapeutic efficacy, levothyroxine should be consistently taken either 60 min before breakfast or at bedtime (3 or more hours after the evening meal).”

The ATA’s 2014 guideline said the same thing eleven years earlier, but as a weak recommendation: the same two windows, the same 3-hour rule for bedtime, lower confidence. That difference matters for how the option is described — but not for whether it is permitted. Both guidelines permit it. The label does not mention it.

The trial evidence is genuinely mixed. Some crossover studies have found that bedtime dosing lowers TSH compared with morning dosing; others have found the two equivalent; a pragmatic study in older adults found no difference. The most defensible reading is not “bedtime is better” but “bedtime is a legitimate option that is equivalent in many studies and better in some, and its main advantage is that it can fit a life that morning fasting cannot.” That is exactly the kind of option a patient should be told about.

A trial specifically recruiting patients with persistent symptoms, with patient-reported outcomes as well as TSH, would test the benefit for the group this page is concerned with. That uncertainty is a reason to explain the option and monitor it, not to keep it obscure. A supervised timing change requires no new thyroid medicine; its suitability depends on the patient’s meals, other medicines and clinical circumstances.

In the United States, levothyroxine monotherapy accounted for an average 88.3% of new thyroid-hormone prescriptions in the 2010–2020 claims study, and 87.2% in 2020. That dominance coexisted with a substantial trend: desiccated thyroid’s share of new prescriptions nearly doubled, from 5.4% in 2010 to 10.2% in 2020. In Europe, the THESIS survey reported that 98.3% of specialists started with levothyroxine, while 39.7% (2,041/5,146) would choose levothyroxine plus liothyronine for persistent symptoms on monotherapy. Japan’s published symptom-specific percentage has an arithmetic inconsistency, shown in the table below. Across national surveys, willingness to consider combination therapy varies widely with the country and question; several surveys also found less willingness to choose it for oneself than to prescribe it for a symptomatic patient.

*Japan source discrepancy. The published paper prints 15.2% alongside 41/207 for this response. Those values do not agree arithmetically. Both are disclosed here; the exact percentage remains unresolved rather than being silently recalculated.

The values come from different surveys with different questions, response rates and definitions. They are shown to make the range visible, not to be averaged. “Would consider” is a clinician’s stated willingness, not a patient’s preference; the self-versus-patient columns are the clearest sign that clinicians are more willing to prescribe a trial for a patient who is unwell than to choose it for themselves. THESIS (Attanasio 2024) reports 40.0% in its abstract; 39.7% is the full-text figure among the 5,146 respondents who answered this question.

Why does the gap persist? Partly because the evidence is genuinely uncertain: the meta-analyses do not show an average benefit, and guidelines are right to say so. Partly because a trial is work: monitoring, titration, stopping rules, and the possibility of over-replacement. Partly because the consultation is short and the default is strong. The gap is not evidence of indifference. It is evidence of a default — and defaults are exactly what information and choice can change.

Trust and communication matter more than the pill

The patient surveys add a finding that complicates the simple story. In the international E-MPATHY survey (3,218 respondents with treatment data), satisfaction was not associated with the type of thyroid hormone treatment; trust and communication were important. In Mitchell and colleagues’ UK survey of 969 patients, 77.6% were dissatisfied. In the multivariate analysis, satisfaction was higher when the GP was the one prescribing DTE, LT4+LT3 or LT3 (OR 3.10, 95% CI 1.83–5.15) and when patients had been fully informed by their GP (OR 3.43); quality of life was lower when patients had to source those treatments without the GP's help. The type of thyroid hormone itself was not independently associated with satisfaction or quality of life. A forum analysis of 673 posts from patients using desiccated thyroid found that social networks (53%) and the internet (38%) were the most-cited sources of information, clinicians only 9%; 54% had initiated the request themselves, and 31% had bought thyroid extract outside the United States.

That is the agency gap in numbers: patients are not primarily learning about their options from their clinicians. They are learning from each other, and then asking. Sometimes that works. Sometimes it leads them to unregulated products or to a trial that was never monitored. The answer is not to shut down the forums; it is to put the information — and the options — in front of patients and clinicians before the search begins.

The gap is not a mystery that needs a new molecule. It is a set of answerable questions:

1. A chart audit or registry analysis. Among patients on levothyroxine, how many have an abnormal TSH, documented persistent symptoms, an antidepressant prescription, and no documented discussion of timing or absorption? That is the size of the problem — and it can be measured from records that already exist.
2. A pragmatic randomised trial of timing in the refractory group. Bedtime versus morning, same dose, with TSH, adherence, fatigue and thyroid-specific quality of life at 6–12 weeks. The intervention costs nothing; the trial should specifically recruit the persistent-symptom group and measure how patients feel.
3. A properly designed combination-therapy trial. The 2021 consensus already set the specification: thyroid-specific quality of life as the primary outcome, adequate levothyroxine dosing before randomisation, a time-limited trial with a stopping rule, and preference as a secondary outcome. That is a feasible trial design worth funding.
4. An rT3 study with a control group and an outcome. Patients with elevated rT3 and persistent symptoms, randomised to a T3-containing regimen versus continued levothyroxine, with rT3, free T3 and fatigue measured. The 2025 cross-sectional study called for exactly this.
5. A patient-reported survey. How many patients were told about timing? How many asked about combination? How many were on an antidepressant before their thyroid was ever reviewed? The site can ask these questions anonymously, with the caveats that survey data are self-reported and not a substitute for a trial.

None of these is glamorous. All of them are cheap compared with another decade of the same argument. The strongest version of the patient case is not “we know combination therapy works” — it is “we do not know, and we could”.

How the evidence was selected

Sources were gathered as full texts where possible, read against their abstracts, and checked for the specific number or statement used. Preference was given to randomised trials and meta-analyses for efficacy and preference questions, to national register studies for prevalence and antidepressant use, to current guidelines for recommendations, and to the product labels themselves for what patients are told. Patient surveys are labelled as self-selected; case reports are labelled as case reports; mechanistic hypotheses are labelled as hypotheses.

Key sources

Limits

• This is a narrative synthesis, not a systematic review. The search was purposive, not exhaustive; no formal risk-of-bias assessment was performed.
• Several sources are self-selected patient surveys; they describe the experience of the people who responded, not the population.
• Some studies are cited from their abstracts or from secondary summaries where the full text was not available; those are marked in the working dossier. Numbers that could not be verified against a full text are not used as primary claims on this page.
• The label analysis covers the current US prescribing information reviewed for this page; wording varies by product and revision, and patient leaflets may differ from the prescribing information.
• “Would consider combination” survey numbers are not directly comparable across countries; they are shown to make the range visible, not to be pooled.
• Where the page discusses depression and antidepressants, the association is not causal, and the page does not recommend stopping or changing any medicine.

Corrections

If a number on this page conflicts with the primary source, the primary source wins. Corrections will be recorded here and on the site’s corrections page. Source-review notes are maintained separately; the links above let readers check the cited publications.

This page is not medical advice. It is a patient-researcher’s synthesis of published evidence, written to make options and uncertainties visible. Decisions about levothyroxine, combination therapy, antidepressants or any other treatment belong with you and a clinician who knows your history. The author is not a physician.