Taking levothyroxine, values fine, symptoms remain: what may be behind it

You have been taking your tablet for months. Perhaps for years. In the morning, on an empty stomach, dutifully. The report arrives: TSH inside the normal range. All good, says the piece of paper.

And there you sit with your tiredness, with the fog in your head, with the words that will not come to you. The value is fine, after all. So presumably something is wrong with you.

In my consulting room this is the most frequent sentence on the subject of the thyroid. Many people know this pattern, and many already have a pile of unremarkable reports at home. I want to show you what may be behind it, in the order in which it is worth looking. First the banal, then the fine detail, then the part that may have nothing to do with the thyroid at all.

And one sentence up front, which I will repeat several times: nothing about a thyroid medication is changed on your own. This text is material for a conversation, not for a decision at the kitchen table.

When the values are good and you are not

The first question almost everyone asks is: am I alone with this? No. It is described, counted and studied, by the professional societies themselves. It is only that the numbers circulating online measure very different things. So let me sort them once.

The most robust number comes from England, and it is so good because nobody knew what it was about.

Next to that stand two numbers that sound higher: the survey of the American Thyroid Association with 12,146 responses found a median satisfaction of 5 out of 10 points [Survey, n=12,146], and a survey of the British Thyroid Foundation found 77.6 percent dissatisfied [Survey, n=969]. Both describe who fills in an online survey run by a patient organisation: self selection, not a population sample. Even so, the British one contains a finding that matters more to me than any percentage.

And now the number that sets the frame. Gregory Canaris and colleagues surveyed 76 people with newly diagnosed hypothyroidism and 147 matched controls. Those affected reported 30.2 percent of the classic symptoms, the healthy controls 16.5 percent [Cross-sectional, n=223]. Tiredness, feeling cold, dry skin, forgetfulness: all of that has a background frequency even without a thyroid. If part of it remains after good adjustment, that is to be expected and is not yet a treatment failure.

A Danish follow up study of 78 people fits with this. Before treatment started, all 13 thyroid specific scales were impaired; after six weeks nine had clearly improved, and after six months an additional five out of eight general scales had as well. Full recovery was not reached [Cohort, prospective, n=78]. Saying both things at once is the honest answer. And one finding disarms a common self doubt: in 141 well adjusted people, weaker performance showed up in complex attention and verbal memory, without TSH or antibodies explaining it [Cohort, n=141]. If words are missing for you, that is a described pattern, and the TSH says nothing about it.

A note on scope: this article is about people who are already taking levothyroxine. If you have symptoms but no diagnosis and no medication, you will find the context in Normal values and still symptoms and in Functional hypothyroidism.

And now you know why the figure of 10 to 15 percent, which appears everywhere online, does not appear that way in any primary source. It is several numbers that have melted into one.

First the obvious: does the tablet arrive at all?

Imagine you send off the same parcel every morning. Whether it arrives depends on whether the route is clear and whether somebody takes something out along the way.

Levothyroxine behaves exactly like that. The active substance is taken up in the upper small intestine, and this uptake is easily disturbed: it needs an acidic stomach, a clear route, and it does not like company. That is why this section comes so early. Here there is the most to gain and the easiest thing to change.

The step the guideline names first

The 2025 European guideline on levothyroxine monotherapy does something remarkable: before it talks about absorption problems, it names adherence. If someone needs an unusually high daily dose per kilogram, the first thing to check is whether the tablet is really being taken every day [Guideline]. That is not an accusation. A tablet that has to be taken on an empty stomach collides with every hectic morning. It is more honest to raise it openly.

The interfering factors and their numbers

Gaps according to the ETA guideline 2025 [Guideline], effect sizes from the individual studies cited below. For magnesium the transfer is mechanistically plausible, but there is no dedicated study of comparable strength.

With calcium and iron the mechanism is the same: both are divalent cations, and both can bind the hormone molecule in the acidic stomach into a poorly soluble complex. In the 1992 iron study, mixing them even produced a violet precipitate [In vitro]. When twenty stably adjusted people took calcium carbonate together with their levothyroxine for three months, the mean TSH rose from 1.6 to 2.7 mIU/l and then fell back to 1.4 [Cohort, n=20]. With ferrous sulphate taken at the same time, it rose from 1.6 to 5.4 mU/l over twelve weeks [Clinical study without control group, n=14]. That study is small and has no control group, and simultaneous intake is the worst possible case. Which is precisely why it shows the order of magnitude so clearly. Why iron sometimes fails to arrive for entirely different reasons is covered in Iron, inflammation and the hepcidin block.

The stomach as doorkeeper

Now the part that is thought of less often. The uptake of levothyroxine needs an acidic stomach. If it produces less acid, less arrives.

From a functional perspective the stomach is interesting because many routes converge there, explored further in Low stomach acid, the overlooked cause. And an acid blocker is not prescribed without reason. The point is not to question it, but that the gap is known and that measurements follow a new prescription [Guideline].

Coeliac disease, lactose and the number that works in both directions

A Jordanian cross-sectional analysis of 914 people with autoimmune hypothyroidism estimated a coeliac rate of about 5.7 percent [Cross-sectional, n=914]. Usable as an order of magnitude, not as a precise figure. Incidentally, gluten appears here purely as an absorption topic, not as a dietary concept; what else may help in Hashimoto is covered in Nutrition in Hashimoto and in Gluten and gliadin without coeliac disease.

For lactose intolerance the data are weaker: in a Turkish study the TSH fell clearly after eight weeks of a low lactose diet, and did not in the comparison group [Cohort, n=83]. Small, not randomised, unblinded. The European guideline names about 31 percent higher requirement for this, and about 27 percent after a gastric bypass [Guideline]. In a case series of 17 people after bariatric surgery, the TSH fell clearly through the switch to a liquid formulation alone [Case Series, n=17]. Where these topics converge in the gut is covered in Iron deficiency and absorption problems in the gut.

Morning, evening, and what that really shifts

In a crossover trial, 65 people went through three regimens for eight weeks each: fasting in the morning, at bedtime, with breakfast. The mean TSH values were 1.06, 2.19 and 2.93 mIU/l [RCT, n=65]. In everyday practice that decides whether someone counts as well adjusted.

The European guideline draws a pragmatic conclusion from this: at least 30 minutes before breakfast, optimally 60 minutes, alternatively at least three hours after the evening meal. And the timing should fit the everyday life of the person being treated [Guideline]. A gap you keep every day is worth more than a theoretically better one you forget twice a week. A change of intake time is not something to switch quietly on your own, though. It can shift the TSH measurably, so it belongs discussed beforehand and measured again afterwards.

Two more points: liquid and soft capsule formulations depend less on how acidic the stomach happens to be, and with a repeatedly raised TSH on tablets a change of formulation may be considered. And whoever is well adjusted is best off staying with the same preparation; if a switch is necessary, free T4 and TSH should be checked after six weeks [Guideline].

And now you know why two people on the same dose can have different values without anything being different about their thyroid.

The reference range is not the target range, and what the studies give us

Perhaps you have read this sentence: the TSH should not simply lie inside the normal range, it should be below 2.5. The idea promises a simple lever and sounds like established knowledge in many forums. That very assumption has been tested twice under double blind conditions.

In the first, 56 people with primary hypothyroidism received three dose levels in random order, with mean TSH values of 2.8, 1.0 and 0.3 mU/l. In wellbeing, mood, symptom questionnaire, quality of life and cognitive tests, no effect and no significant preference appeared [RCT, n=56, double blind]. The second shifted the TSH more markedly.

In the consulting room this often draws an objection that I understand well: but I did notice that I felt better on the higher dose. I am not disputing that impression. The study only says that it could not be reproduced under blinding. That is a difference between two routes to knowledge, not a statement about you.

And now the other direction. The British professional societies draw a different practical conclusion from the same data: they recommend first making full use of the levothyroxine dose, with a target TSH of 0.3 to 2.0 mU/l over three to six months, before it is judged at all whether there is a response [Guideline]. That is not a contradiction of the studies, it is a different question. The studies ask about the group difference on average. The guideline asks how we make sure not to stop too early with this one person.

One sentence absolutely belongs here, because otherwise this section reads like a set of instructions: making full use of the dose is a medically guided process, with a measurement before and one afterwards. Anyone who sees their TSH at 3.1 and derives a dose change of their own from it is adjusting a long term medication blind, and levothyroxine only shows its consequences after weeks. Changes to the dose are made exclusively in agreement with your treating practice.

That leaves the question of the individual set point. Every person has their own TSH range before an illness, considerably narrower than the reference range of the population. A review by the leading author group names exactly that as a promising research direction [Review]. And here is where what studies can show today ends: so far there is no method to reconstruct the personal set point after a diagnosis. That is a hypothesis, not a measurement. Everything further on which values are measured and what they can do is in Thyroid values: which ones really count.

And now you know why the figure of 2.5 turns up so often and is still not a proven threshold.

The route from T4 to T3, kept short

T4 is the storage hormone, T3 the active one. The body converts one into the other, in liver, kidney, muscle and in the tissue itself. I am keeping this section short, because the depth is in From T4 to T3: the conversion. Here comes only what concerns people on levothyroxine.

Similar in a population representative US analysis: on levothyroxine, T4 values were higher and T3 values lower, with a T3 to T4 ratio about 15 to 20 percent lower [Cross-sectional, n=9,981]. The observation is strong, but it carries no statement about cause. A cross-sectional study shows a state, not the direction of causation.

The matter of the DIO2 gene

In 2009 a group around Vijay Panicker studied gene variants of the deiodinases in 552 participants of a British trial. The rarer CC genotype of variant rs225014 was present in 16 percent, had worse baseline values on T4 alone and improved more strongly on T4 plus T3, by 2.3 questionnaire points, without any influence on hormone levels in blood [Genetic analysis within an RCT, n=552]. The authors themselves called for replication.

That replication came, and it turned out differently. In the most thorough three arm trial to date, from 2021, the Thr92Ala polymorphism of the DIO2 gene did not influence the results [RCT, n=75]. And the joint consensus document of the ATA, BTA and ETA explicitly asks for deiodinase variants to be examined in future studies first [Guideline].

I find one rarely cited finding more interesting: in 141 people on levothyroxine, two of three variants of the brain specific hormone transporter OATP1C1 were associated with fatigue and depressive symptoms, but not with neurocognitive tests and not with a preference for a combination [Genetic analysis within an RCT, n=141]. If genetics play a part here, then perhaps not only at the enzymes, but already at the door to the brain. A plausible trail with thin human evidence, and that is exactly how it belongs treated. Reverse T3 has its own discussion in Reverse T3: what the value can do.

And now you know why the DIO2 story is big online and small in the guidelines.

T4 plus T3: the studies, the preference and the contradiction between them

This section has to hold a tension rather than dissolve it. It all started in 1999: Robertas Bunevicius and colleagues had 33 people go through two phases of five weeks each, once on their usual T4 dose, once with part of it replaced by T3. Of 17 scores, six came out better on the combination, and of 15 visual analogue scales, ten did [RCT, n=33, double blind]. Small, short, clearly positive. And never replicated with that clarity again.

After that came a long series of larger studies with a consistent result. A systematic review from 2005 summarised nine controlled trials, and only in one did the combination show advantages [Systematic Review, k=9]. A meta-analysis a year later was more explicit.

That could settle the matter. But it does not. The most recent meta-analysis, from 2021, included 18 studies and found no difference in clinical status, quality of life, psychological distress and fatigue. In the same analysis, 43 percent preferred the combination and 23 percent the monotherapy [Meta-analysis, k=18]. That is the core of the debate: the same studies, two answers, depending on what you measure.

This is exactly where two studies become important that barely appear in the German language internet. The first is Dutch, from 2005: 141 people were randomised to levothyroxine, a 10 to 1 combination and a 5 to 1 combination. The study medication was preferred by 29.2, 41.3 and 52.2 percent, no secondary endpoint confirmed the trend, and the end of study TSH was 0.64, 0.35 and 0.07 mU/l [RCT, n=141, double blind]. Preference rises neatly with the T3 share, and the TSH falls just as neatly along with it. Whether people preferred the combination or a slight overdose cannot be separated from this.

The second is Danish, from 2009, and it anticipated that objection: there the T4 dose was readjusted so that the TSH stayed unchanged between groups. In seven of eleven scores, differences appeared in favour of the combination; 49 percent preferred it, 15 percent the monotherapy [RCT, n=59, double blind, crossover]. Two good studies, two directions. This question is open.

The study that comes closest to the matter

The same pattern turns up with desiccated thyroid extract: in a crossover trial with 70 people, no differences were found in symptoms and cognition, and at the end 48.6 percent preferred the extract and 18.6 percent the levothyroxine [RCT, n=70, double blind]. More on this group of preparations is in Desiccated thyroid extract (NDT). Here it was only about the study data, because they belong to the same body of evidence.

So that this section is not read wrongly: these are study data, not a recommendation and not a source of supply. T3 containing preparations and thyroid extracts are prescription medicines with a narrow margin of control, and they are prescribed medically, monitored medically and reviewed medically again. None of it begins or ends on your own. If these data interest you, that is a good reason to put them on the table in your consultation.

And now you know why studies and personal reports diverge so often here. They do not measure the same thing, and they do not mean the same people.

What the professional societies actually say

Perhaps you have heard that the professional societies are against combination therapy. Or the opposite, that they recommended it long ago. Neither gets it right.

2012: the European guideline names the problem itself

In 2012 the European Thyroid Association published a guideline on the use of T4 plus T3. It contains the sentence this article is built on: in 5 to 10 percent of people treated with levothyroxine who have a normal TSH, symptoms persist [Guideline]. As explanations it names four lines: the awareness of being chronically ill, accompanying autoimmune conditions, thyroid autoimmunity itself, and the fact that giving T4 does not in all cases reproduce the natural T4 and T3 concentrations in serum and tissues.

On the combination it said: the evidence is not sufficient, monotherapy remains the standard. And then comes the clause that is often left out. The combination may be considered as an experimental approach, in adherent people with persisting symptoms despite a TSH inside the reference range, after support in dealing with chronicity and after ruling out accompanying autoimmune conditions, initiated only by appropriately qualified physicians [Guideline]. Why the immune system attacks the thyroid is covered in Hashimoto and the immune system.

2014: the ATA turns it into a question to be examined

In 2014 the American Thyroid Association issued an extensive guideline on the treatment of hypothyroidism and formulated 24 questions in it, among them explicitly the sources of dissatisfaction with levothyroxine and the evidence on alternatives [Guideline]. So the professional society does not dispute the problem. It turns it into a question to be examined.

2021: the joint document declares equipoise

2023: the British order of steps

The most practical document comes from the British Thyroid Association and the Society for Endocrinology. It answers not with yes or no, but with an order of steps [Guideline]. First: secure the diagnosis. Anyone with persisting symptoms but no clear evidence of overt hypothyroidism should first have a medically guided trial without hormone replacement. Second: make full use of monotherapy, target TSH 0.3 to 2.0 mU/l over three to six months. Third: in confirmed overt hypothyroidism, with persisting symptoms and accompanying conditions ruled out, a trial with a combination may be justified. Fourth: decide together.

With that goes the honest postscript from the same paper: individual clinicians need not feel obliged to start liothyronine or to continue a medication begun elsewhere. That is the tone I would wish for. It takes both sides seriously.

The German point of reference

In Germany the primary care S2k guideline on the raised TSH value comes into it (AWMF register number 053-046, versions 2023 and 2024) [Guideline]. It sets the threshold by age: for 18 to 70 years and in pregnancy above 4.0 mU/l, for 70 to 80 years above 5.0, and from 80 years above 6.0 mU/l. In subclinical hypothyroidism up to 10 mU/l it does not generally provide for hormone treatment in people without symptoms; from 10 mU/l up to age 75 it provides for treatment, and above that age, with a TSH below 20 mU/l, generally also for withholding treatment.

And now you know why the same body of data can lead to different practice recommendations in two countries, without anyone having read the studies wrongly.

When it is too much rather than too little

Now the uncomfortable section. The thought that symptoms can also come from a dose that is too high barely appears in German language guides. It belongs here though, for two reasons.

First, the threshold at which treatment is started at all has shifted: a British analysis of 52,298 first prescriptions between 2001 and 2009 showed that the median TSH at the start of treatment fell over the years [Cohort, n=52,298]. Second, the European guideline says it directly: worldwide, almost half of those treated appear to be over treated or under treated [Guideline]. Together this means that some people with symptoms on levothyroxine may be dosed too high, while others might, by today's German criteria, not have had an indication for treatment at all.

These are not side effects to be sat out. Cardiac arrhythmias, a TSH suppressed over years and bone density all count in the long term view. I write this without drama: a reason for an appointment, not for panic. The other direction is just as serious. Anyone gaining a lot of weight, feeling ever colder, becoming very slow or noticing swelling also belongs examined at short notice.

For the rare but serious extremes at both ends there are separate names, and they belong named once. A thyroid storm with high fever, a very fast or irregular pulse, vomiting and confusion is an emergency. A myxoedema coma with severe hypothermia, extreme slowing and increasing clouding of consciousness is too. Both are very rare under an ongoing, monitored replacement, and I do not write this to frighten anyone. I write it so that you know: if such signs appear, the right place is the emergency department and not waiting until the next appointment.

Pregnancy and older age

In pregnancy different rules apply, and that is not a formality. The hormone requirement rises, different target values apply, with closer monitoring intervals and a different framework of care. The German primary care guideline names a range of 0.4 to 4.0 mU/l for pregnant women on treatment and at least one check per trimester [Guideline]. If you are pregnant or would like to become pregnant, the thyroid belongs on the table early and explicitly. That is a subject for the practice looking after you, not for a blog article.

In older age the target range shifts in the other direction. The European guideline considers a more relaxed target range from the age of 70 appropriate [Guideline], and the German one raises the thresholds with age. Not a lapse in care, but the recognition that a slightly higher TSH in advanced age carries a different risk than an overly tight setting.

And now you know why the question about too much and the question about too little belong in the same consultation.

What may have nothing to do with the thyroid at all

This is the most honest section of this article. It stands at the end because it puts everything before it in perspective.

Do you remember the number from the first section? People without hypothyroidism reported 16.5 percent of the classic symptoms. These symptoms are non specific, they are the common final path of many states. If you have a thyroid diagnosis, the temptation is great to sort everything into that box. That is understandable and may cost you the answer. From the perspective of clinical psychoneuroimmunology I therefore look at four systems at the same time: nervous system, immune system, metabolism and hormones. Exhaustion rarely arises in one of them alone.

Iron, even without anaemia

Vitamin B12, vitamin D and sleep

In a study of 116 people with hypothyroidism, 39.6 percent had low B12 levels [Cohort, n=116]. The limitation is written by the authors themselves: no control group, and among the people without a deficiency, 40 percent also reported an improvement on B12. The frequency figure is usable, the efficacy statement is not. With vitamin D it is similar: in a double blind trial of 120 young healthy people with tiredness and a pronounced deficiency, the exhaustion score fell more strongly after a single high dose than on placebo [RCT, n=120]. Four weeks, young healthy people, not directly transferable to people on levothyroxine. More in Vitamin D deficiency, sun and winter.

The point most often overlooked, in my view, is sleep. In the Lausanne population cohort HypnoLaus, 2,121 people were examined at home with polysomnography. Moderate to severe sleep disordered breathing was found in 23.4 percent of the women and 49.7 percent of the men [Cohort, n=2,121]. The numbers are high because modern measurement technology was used, and the authors themselves call for a revision of the definitions. The message stands: anyone exhausted for years who does not wake up rested belongs examined for sleep, independently of the thyroid, see Recognising sleep apnoea.

Perimenopause

And then there is an overlap almost too obvious to be noticed: the age group in which Hashimoto and replacement therapy are particularly common is the same one in which the hormonal transition begins.

In the American SWAN cohort, 3,289 women gave information on 58 symptoms over 16 years. Even before menopause, 10 percent were in the highly symptomatic and 16 percent in the moderately symptomatic class, and hot flushes clustered with sleep problems and tiredness without being a defining feature of the clusters [Cohort, n=3,289]. That is not an argument against the thyroid, it is an argument against single causation. Going deeper: Perimenopause, from when and Thyroid and female hormones.

Depression, exhaustion of another cause, adrenals

A narrative review by Petros Perros and colleagues sorts the attempts at explanation into four lines that do not exclude one another [Review]. First the tissue thesis: levothyroxine might leave an undersupply in individual tissues such as the brain while the hypothalamus is satisfied. Second, a somatic symptom disorder present at the same time. Third, autoimmune neuroinflammation. Fourth, accompanying conditions and psychosocial factors. The authors criticise that twenty years of research have flowed predominantly into the first hypothesis.

I write this point carefully, because it easily sounds hurtful. It is not meant that way. An exhaustion depression is not imagination and not a weakness of character, it is a condition in its own right, treatable, with its own biology. If you recognise yourself in Burnout and exhaustion depression, that is a trail and not a defeat.

To finish, I am leaving the area of robust studies, and I am saying so. Clinical tradition without a strong study basis: in functional medicine, chronic strain is discussed as an explanation in its own right for the same symptoms, often under terms that do not hold up scientifically. What can be said seriously is that ongoing stress reaches into sleep, the immune system and metabolism through the axis of hypothalamus, pituitary and adrenal gland. What is not serious is turning that into a diagnosis via a saliva test. The context is in Adrenal fatigue: what is behind it and in Cortisol and the HPA axis.

What follows from this in practice

I am deliberately not giving you a laboratory list to tick off and no dosages. What I can give you is an order of steps. It follows the size of the effects, not the popularity of the topics.

What I observe clinically, explicitly as an observation and not as a study result: in most people with this question, the answer does not lie in one single point but in two or three small ones at the same time. A little coffee too early, a little ferritin too low, a little too little sleep, a life stage with a heavy load. That sounds less spectacular than a single explanation. But it is the route that, in my experience, most often leads somewhere.

And now you know why the most honest part of this article is the last one and must not be a footnote.

Frequently asked questions about levothyroxine and persisting symptoms

These are the questions I meet most often on this topic, in the consulting room and by email.

The thyroid and the rest of the body

This topic hangs on more than one organ. Each of these routes can explain part of the exhaustion on its own.