Thyroxine treatment in patients with symptoms of hypothyroidism but thyroid function tests within the reference range: randomised double blind placebo controlled crossover trialBMJ 2001; 323 doi: https://doi.org/10.1136/bmj.323.7318.891 (Published 20 October 2001) Cite this as: BMJ 2001;323:891
- M Anne Pollock, principal biochemist ()a,
- Alison Sturrock, senior house officerb,
- Karen Marshall, trainee clinical psychologistc,
- Kate M Davidson, research tutorc,
- Christopher J G Kelly, specialist registrarb,
- Alex D McMahon, consultant statisticiand,
- E Hamish McLaren, consultant physicianb
- a Department of Clinical Biochemistry, Stobhill Hospital, Glasgow G21 3UW
- b Department of Medicine, Stobhill Hospital
- c Department of Psychological Medicine, Gartnavel Royal Hospital, Glasgow G12 0XH
- d Robertson Centre for Biostatistics, University of Glasgow, Glasgow G12 8QQ
- Correspondence to: M A Pollock
- Accepted 8 May 2001
Objectives: To determine whether thyroxine treatment is effective in patients with symptoms of hypothyroidism but with thyroid function tests within the reference range, and to investigate the effect of thyroxine treatment on psychological and physical wellbeing in healthy participants.
Design: Randomised double blind placebo controlled crossover trial.
Setting: Outpatient clinic in a general hospital.
Participants: 25 patients with symptoms of hypothyroidism who had thyroid function tests within the reference range, and 19 controls.
Methods: Participants were given thyroxine 100 µg or placebo to take once a day for 12 weeks. Washout period was six weeks. They were then given the other to take once a day for 12 weeks. All participants were assessed physiologically and psychologically at baseline and on completion of each phase.
Main outcome measures: Thyroid function tests, measures of cognitive function and of psychological and physical wellbeing.
Results: 22 patients and 19 healthy controls completed the study. At baseline, patients' scores on 9 out of 15 psychological measures were impaired when compared with controls. Patients showed a significantly greater response to placebo than controls in 3 out of 15 psychological measures. Healthy participants had significantly lower scores for vitality when taking thyroxine compared to placebo (mean (SD) 60 (17) v 73 (16), P<0.01). However, patients' scores from psychological tests when taking thyroxine were no different from those when taking placebo except for a poorer performance on one visual reproduction test when taking thyroxine. Serum concentrations of free thyroxine increased and those of thyroid stimulating hormone decreased in patients and controls while they were taking thyroxine, confirming compliance with treatment. Although serum concentrations of free triiodothyronine increased in patients and controls taking thyroxine, the difference between the response to placebo and to thyroxine was significant only in the controls.
Conclusions: Thyroxine was no more effective than placebo in improving cognitive function and psychological wellbeing in patients with symptoms of hypothyroidism but thyroid function tests within the reference range. Thyroxine did not improve cognitive function and psychological wellbeing in healthy participants.
What is already known on this topic
What is already known on this topic Recent anecdotal accounts suggest that patients with symptoms of hypothyroidism but who are biochemically euthyroid may benefit from thyroxine treatment
No controlled trials in this area have been reported
What this study adds
What this study adds This study suggests that thyroxine is no more effective than placebo in improving psychological and physical wellbeing in patients who show symptoms of being clinically hypothyroid but whose thyroid function tests are within the reference range
Thyroxine replacement did not improve psychological and physical wellbeing in healthy participants
The classic symptoms of hypothyroidism are wide ranging and non-specific, therefore biochemical testing has become the cornerstone of diagnosis in patients for whom there is a clinical suspicion of thyroid dysfunction. However, recent anecdotal evidence has suggested there may be some clinical benefit in giving thyroxine to patients with symptoms of hypothyroidism who have thyroid function tests within the reference range.1–3
After a series of reports in our local newspaper suggesting that such patients benefited from thyroxine therapy we treated two patients empirically with thyroxine, and they both reported symptomatic relief.4
To investigate this further, we conducted a double blind placebo controlled crossover trial of thyroxine in patients who had symptoms of hypothyroidism but whose thyroid function tests were within the reference range. A group of controls, who were similar in age and sex to the patient group, took part in a parallel trial. The same protocol was used for controls and patients to test the clinical belief that thyroxine treatment would have an effect on wellbeing even in participants without symptoms of hypothyroidism. We assessed response to thyroxine by using a battery of biochemical, physical, and psychological tests.
Patients were required to have had at least three of the following symptoms for six months: tiredness, lethargy, weight gain or inability to lose weight, intolerance to cold, hair loss, or dry skin or hair. We recruited patients either by referral from their general practitioner or hospital clinician, or through an article, published in a local newspaper, which described the trial and asked for volunteers. Controls were healthy volunteers recruited by personal contact with the investigators. All participants were required to have no current medical disorder, no history of thyroid disease, and recent thyroid function tests within the reference range.
Because of limited resources, the number of patients was restricted to the first 25 who met the criteria. Three patients withdrew at an early stage: one was anxious about the thyroxine dosage, one was ill, and the third failed to attend for unknown reasons. We did not enter one interviewed patient into the study because she was unwilling to have a thyrotrophin releasing hormone test. Unfortunately, we were unable to recruit sufficient controls to match the patients strictly for age and sex. However, the control group was similar to the patient group (see table A on the BMJ's website). No controls were ineligible and all completed the study. The local ethical committee approved the study and we obtained informed consent from all participants.
The study was a randomised double blind placebo controlled crossover trial. The two treatment periods of twelve weeks each were separated by a washout period of six weeks. Half of the participants were given thyroxine for the first treatment period and placebo for the second treatment period; half were given placebo first and thyroxine second. The first 20 participants received the treatment in 100 µg capsules; thereafter 100 µg tablets were supplied. In each case a visually identical placebo was used. A 14 week supply of tablets, to be taken once a day, was provided for each phase and participants were asked to bring the remaining tablets to their assessment to assess compliance.
Randomisation was by toss of a coin in batches of four. Controls and patients attended their baseline assessment simultaneously, which led to unequal sequence groups. Two of the patients who withdrew had been assigned to the thyroxine-placebo group. The code was broken to two investigators (MAP and KMD) after each participant had completed the trial.
Serum thyroid stimulating hormone, free thyroxine, free triiodothyronine, cholesterol, and prolactin were measured at each visit. At baseline, serum ferritin and antithyroid peroxidase antibodies were measured and a thyrotrophin releasing hormone test was performed. Physiological and psychological assessments were performed at baseline and on completion of each phase.
At the end of the trial participants were asked to identify which treatment they thought they had received in each phase.
Serum thyroid stimulating hormone, free thyroxine, cholesterol, and prolactin were analysed at the time the blood was collected. All blood samples were stored at −80°C and free triiodothyronine, ferritin, and antithyroid peroxidase antibodies were analysed in single batches to minimise interassay variation. Serum thyroid stimulating hormone, free thyroxine, free triiodothyronine, prolactin, and ferritin were measured by fluorescent microparticle enhanced immunoassay (Abbott Laboratories Ltd, Maidenhead, UK). An increment in thyroid stimulating hormone of >25 mU/l constituted an abnormal result in the thyrotrophin releasing hormone test. Antithyroid peroxidase antibodies were measured by a solid phase chemiluminescent enzyme immunoassay (DPC, Llanberis, UK). Cholesterol was measured on a multichannel discrete analyser (Olympus Diagnostica, Hamburg, Germany) using a cholesterol oxidase method (Randox Laboratories Ltd, Co. Antrim, UK). Interassay coefficients of variation were <15% for thyroid stimulating hormone, <9% for free thyroxine, <8.5% for free triiodothyronine, <6% for prolactin, and <2% for cholesterol.
Physical and psychological evaluation
We recorded supine blood pressure and pulse of participants using a Criticare (CE1023, Waukesha, WI, USA) non-invasive blood pressure monitor after they had rested for five minutes. We measured their weight on SECA scales (Hamburg, Germany).
We measured cognitive functioning with logical memory, verbal paired associates, visual reproduction, and digit span tests from the revised Weschler memory scale.5 These tests assess attention and concentration, visual memory, and verbal memory. Psychomotor speed, attention, and sequencing were assessed by the trail making test.6
Psychological and physical wellbeing were measured by using two questionnaires. The hospital anxiety and depression scale assessed emotional disorder7 and the SF-36 health survey measured five health concepts.8 Two patients failed to complete one page of the SF-36 health survey. We calculated the missing values by taking the means of their scores on their remaining two visits.
Baseline characteristics were summarised by means (SD) or counts and percentages as appropriate. For each variable, participants were grouped into their sequence group (thyroxine-placebo or placebo-thyroxine) to create data summaries in relation to treatment.
In each set of participants, the effect of treatment on each variable was studied by using analysis of variance models in relation to patient, period, and treatment. The within patient treatment differences were adjusted by period and 95% confidence limits created. Some of the results were not normally distributed; therefore we repeated all analyses using non-parametric tests and rank data.
The placebo effect was calculated by using participants who received placebo as their first treatment. We calculated changes from baseline by subtracting the baseline value from placebo response.
All participants had thyroid function results within the reference range, with the exception of one patient who had a concentration of thyroid stimulating hormone of 5.8 mU/l (see table A on BMJ's website). Concentrations of the hormone in this patient, however, normalised to 4.5 mU/l when taking placebo. Other biochemical tests for the same patient were within the reference range. Three participants—the patient described above, one control, and one patient who failed to complete the study—had exaggerated responses to thyrotrophin releasing hormone and raised concentrations of antithyroid peroxidase antibodies (range 262-1656 U/ml). One patient and one control had raised concentrations of antithyroid peroxidase antibodies (211 U/ml and 76 U/ml, respectively) but normal thyrotrophin releasing hormone responses.
Patients' scores on psychological testing were significantly impaired for logical memory I and II, trail making test, hospital anxiety and depression scale, and all components of the SF-36 health survey in comparison to healthy participants (see table B on BMJ's website).
Response to thyroxine or placebo
In both groups the serum concentrations of thyroid stimulating hormone decreased and free thyroxine increased in response to thyroxine, confirming compliance with treatment. Although free triiodothyronine concentration increased in both groups when participants were taking thyroxine, the difference between the response to placebo and the response to thyroxine was only significant in the healthy group (tables 1 and 2). This finding was replicated by non-parametric analysis of the data. The response to thyroxine in patients with positive autoantibodies or abnormal thyrotrophin releasing hormone responses did not differ from that in other participants; numbers were too small for detailed analysis. No other significant biochemical changes were observed.
Physical health and psychological measures
We compared the differences in the scores at baseline and after placebo in participants taking placebo first. The patients showed a significant symptomatic improvement in the general health, role physical, and hospital anxiety and depression scale scores after placebo, compared with healthy participants (mean (95% confidence interval) 8 (2 to 15) v −8 (−14 to −3), 25 (4 to 46) v −9 (−32 to 13), and −7 (−11 to −3) v −1 (−2 to 0), respectively; table 3). We observed no changes in measures of cognitive function. There was no placebo effect with regard to psychological or cognitive function scores for the controls (table 3).
In the comparison of the response to thyroxine or placebo, the only difference in cognitive function tests was in the patient group, which showed a significant improvement in visual reproduction I score with placebo (32 (6) v 30 (7), P=0.016; table 1). The difference was less significant on non-parametric analysis (P=0.035) and was not replicated in the delayed recall test, which suggests that the finding may be spurious. No other tests of cognitive or psychological function showed differences (table 1). The vitality score for the healthy control group was significantly better with placebo compared with thyroxine (73 (16) v 60 (17), P=0.007; table 2), but otherwise no differences were observed.
No significant changes occurred in patients with respect to blood pressure, pulse rate, or weight during the study.
At the end of the study neither group was able to identify accurately which treatment period was thyroxine or placebo (table 4).
This is the first randomised double blind placebo controlled trial of thyroxine treatment in patients who have symptoms of hypothyroidism but are biochemically euthyroid.
Compliance was confirmed in both groups by the rise in free thyroxine and fall in thyroid stimulating hormone while participants were taking thyroxine. The lack of significant increase in free triiodothyronine in patients taking thyroxine might reflect impairment of the peripheral conversion of thyroxine to triiodothyronine. Although this finding requires further investigation, anecdotal evidence suggests that patients could benefit from thyroxine treatment alone.9
Comparison of thyroxine and placebo treatments
Psychological testing showed that patients differed from the controls at baseline. Cognitively, they scored worse on immediate and delayed verbal recall and had slower motor movements. They also perceived themselves to have poorer general health, more fatigue, increased problems with routine tasks and activities related to work, and higher levels of anxiety and depression. These findings may be consistent with a depressive illness, although no formal assessment was performed.
Controls showed no significant changes in psychological measurements after treatment with either thyroxine or placebo. This suggests that, contrary to widespread belief, thyroxine does not have a non-specific effect on wellbeing. In the participants who received placebo first, patients showed a small but significant improvement in general health, physical wellbeing, and anxiety and depression after placebo when compared with baseline. Thyroxine treatment, however, had no greater effect than placebo in this group of patients. This contrasts with previous studies in biochemically hypothyroid patients, where thyroxine treatment was associated with psychological improvement.10–12
Numbers in the study
The small number of participants in this preliminary study means that, although there was no significant difference between placebo and thyroxine in 13 of the 14 well validated psychological tests, the power of the study may not have been sufficient to eliminate definitively a possible biological effect of thyroxine. If this was the case we would have expected to see a trend in favour of thyroxine over placebo in the test results, especially as a recent open intervention study of thyroxine (mean dose 125 µg daily) reported self assessed improvements in energy and poor memory in 80% of 139 participants.9 Our study showed no discernible trend (table 1).
We can find no support for the hypothesis that people with symptoms of hypothyroidism but thyroid function tests within the reference range benefit from treatment with 100 µg thyroxine daily. However, our results require confirmation in a larger study. The improvement noted anecdotally and in open studies may be due to the placebo effect shown in our study.
We are grateful to the staff of the clinical biochemistry department at Stobhill Hospital for performing the biochemical analyses and to the staff of the pharmacy department for performing the randomisation and preparing the capsules. We thank Goldshield Pharmaceuticals for supplying the placebo tablets.
Contributors: MAP conceived the study and coordinated the laboratory component. MAP, EHMcL, CJGK and KMD designed the study. KMD and KM chose the cognitive function tests and questionnaires, and KM tested all participants. AS coordinated all contact with the study participants and undertook the clinical assessment of the participants at each visit. ADMcM performed the statistical analysis and provided further statistical advice. All authors were involved in writing the paper, with MAP providing coordination. EHMcL is guarantor for the study.
Funding MAP received a scientific development scholarship from the Association of Clinical Biochemists.
Competing interests None declared.
Details of baseline measurements and scores are available on the BMJ's website