Evaluation of thyroid function monitoring in people treated with lithium: Advice based on real‐world data

Blood test monitoring is essential for the management of lithium treatment and National Institute for Health and Care Excellence guidance recommends 6‐monthly serum testing of thyroid function. We examined conformity to these guidelines and the impact of monitoring outside these intervals.


| INTRODUC TI ON
Lithium is a common pharmacological intervention for the treatment of bipolar and mania or hypomania and is advocated by UK and international clinical practice guidelines. [1][2][3][4] Its use has been shown to decrease the risk of suicide attempts and halve mortality rates. 5 Despite the efficacy of lithium, it is associated with a range of adverse effects including reduced glomerular filtration rate (GFR), hyperparathyroidism and thyroid dysfunction. 6 Thyroid abnormalities can include both hyper and hypothyroidism, with the latter being more common. 7,8 Certain risk factors have been shown to increase the likelihood of developing hypothyroidism, including females >60 years old, elevated serum lithium concentrations and duration of lithium exposure. 8 There are several proposed mechanisms for lithium-induced thyroid hormone abnormalities including reduced synthesis of thyroid hormones, altered iodine uptake, structural thyroglobulin alterations, reduced hepatic deiodination and decreased clearance of free thyroxine. 9 These aberrations are slightly mitigated by the fact that they may be reversible after discontinuation of the drug meaning life-long thyroxine replacement may not be required. 10 In relation to these potentially adverse effects, blood test monitoring is an essential component of the management of lithium treatment. The UK National Institute for Health and Care Excellence (NICE) recommends 3-monthly monitoring of serum lithium levels for most patients and 6-monthly monitoring of thyroid function, estimated GFR and calcium to identify early end-organ abnormalities. 1 A recent study has demonstrated that adherence to the guidance on serum lithium levels is variable; in patients with serum lithium results within the therapeutic range only 25% of test requests for serum lithium levels occurred at 3-monthly intervals. 11 This is important as regular monitoring of serum lithium levels decreases psychiatric hospital admissions. 12 While there is clearly a case for ongoing monitoring, the evidence base underpinning these suggested testing frequencies is less clear, suggesting that it is based more on expert opinion than formal studies. Indeed, a recent study has additionally suggested that the NICE recommendation of 3-monthly lithium monitoring could be extended to 6-monthly intervals for patients who have 12 months of lithium results within the 0.40-0.79 mmol/L range, irrespective of age. 13 The adherence to the recommended 6-monthly monitoring of thyroid function remains unknown. Biochemical monitoring of thyroid function is particularly important as the symptoms of hypothyroidism and bipolar disorder may overlap making it difficult to identify hypothyroidism based on clinical presentation. 6 Poor adherence to national guidance on testing intervals for biochemical monitoring is common and has been shown for a variety of similar situations where long-term monitoring is required, including assessment of glycaemic control in people with diabetes mellitus, 14 post-partum screening in women with gestational diabetes 15 and monitoring of thyroid function in people with hypothyroidism on thyroxine replacement therapy. 16 The COVID-19 pandemic has provided additional challenges for conditions which require regular blood test monitoring; this has been demonstrated for HbA1c where an estimated 1.41 million diabetes monitoring tests have been delayed or missed across the UK. 17 In this study, we investigate the monitoring frequency of thyroid function testing in people on lithium treatment and explore its clinical impact.

| Data collection
The cohort included all patients (n = 582) who had serum lithium concentration measured between 1 January 2012 and 31 December 2018 at the University Hospitals of North Midlands NHS Trust (UHNM) Department of Clinical Biochemistry, as described previously. 11 All thyroid-stimulating hormone (TSH) and free T4 (fT4) tests requested from this patient cohort for the period 1 January 2009 to 31 December 2020 were then extracted from the Laboratory Information Management System (LIMS; 10,636 requests; Figure 1).
Reference ranges were as follows: TSH before 1 November 2014 was 0.3-5 mIU/L, TSH after 1 November 2014 was 0.1-5 mIU/L, fT4 before 12 September 2018 was 8-19 pmol/L and fT4 after 12 September 2018 was 12-23 pmol/L. The study period for each patient began from the date of their first lithium test and continued until 31 December 2020. Patients were identified as developing subclinical hypothyroidism (TSH above reference range, fT4 within reference range), overt hypothyroidism (TSH above reference range, fT4 below reference range), subclinical hyperthyroidism (TSH below reference range, fT4 within reference range) or overt hyperthyroidism (TSH below reference range, fT4 above reference range) upon the first abnormal result.
In order to exclude those with a possible pre-existing thyroid disorder, individuals were removed from the data set if they had an abnormal TSH during the 2 years before the first lithium request (assumed start date of lithium treatment; Figure 1). Individuals were additionally removed if there were less than 2 years of run-in data before the first lithium request, as it could not be determined whether the patient had a pre-existing thyroid disorder. Results were compared to the reference range in place at the time of the request. Thyroid results were removed from the data set for patients before they commenced lithium therapy and after the first abnormal TSH result.
As this study is an audit of practice against guidelines provided by NICE, it did not require ethical committee approval. All data extracted from the LIMS and used in the statistical analysis were anonymized.

| Assessment of clinical outcomes
For patients with an abnormal TSH within the first 3 or 6 months of commencing lithium treatment, the clinical details accompanying thyroid hormone and antibody requests until they commenced thyroxine replacement or the end of the study period were obtained from the LIMS. The following information was documented: date of thyroid request, type of thyroid disorder indicated by results, thyroxine treatment and anti-thyroid antibody (thyroid peroxidase and TSH receptor antibody) results. Four patients were excluded from the clinical assessment: one was deceased 1 month after commencing lithium, and three had incomplete records. At UHNM, for all requests for thyroid function, the user is prompted to state whether the patient is on thyroxine. Clinical outcome data were charted in a timeline for each patient using Microsoft Excel 2010.

| Patient demographics
The final data set contained 1811 requests from 275 patients ( Figure 1). The average age for the cohort was 52.4 years and 57% were females (Table 1). During the study period, a total of 58 patients developed an abnormal TSH; 46 demonstrated subclinical hypothyroidism, six subclinical hyperthyroidism, three overt hypothyroidism and three overt hyperthyroidism. Of the patients who developed subclinical hypothyroidism, 32 (70%) were female and 6 (13%) progressed to overt hypothyroidism during the study period.

F I G U R E 1 Study inclusion and exclusion criteria. *six patients
were excluded as no thyroid results were available for the study period after the patient was initiated on lithium. †for Kaplan-Meier analysis a further nine patients were excluded (seven were lost to follow-up after the initial thyroid result and two had an abnormal TSH accompanying their first lithium)

| Lithium testing interval
The adequacy of thyroid monitoring for patients on lithium was determined by measuring the interval between TSH requests for each patient. The number of requests associated with each testing interval is shown in Figure 2. The modal testing interval was 13 weeks, and 265 requests (17.3% of the total) were made at intervals between 11 and 15 weeks. A secondary peak at an interval of approximately 26 weeks was also observed; the number of requests with an interval between 24 and 28 weeks was 147 (9.6% of the total). The data set includes 150 specimens (9.8% of the total) on 89 patients taken during the post-COVID-19 impact period (after 23 March 2020 17 ).
The median test request interval in this group was 21.5 weeks (interquartile range 12-36 weeks) compared with 17 weeks (interquartile range 9-32 weeks) in those collected during the pre-COVID-19 impact period (p = 0.025).

| Identification of patients who developed an abnormal TSH within the first 3 months of initiating lithium
To determine whether the testing interval led to early detection of developed an abnormal TSH. The annual incidence rate for thyroid abnormalities at the end of years 1, 2 and 3 was 12.8%, 6.4% and 8.2% respectively.

| Relation between clinical outcomes and earlier testing
It was next investigated whether early detection of thyroid abnor-  In all these cases, there were multiple occasions of subclinical hypothyroidism before the initiation of thyroxine replacement therapy.
The average peak pre-treatment TSH result was 9.2 mIU/L for patients who later received thyroxine and 7.0 mIU/L for those who did not ( Table 2)

| DISCUSS ION
This study represents the first time that adherence to the NICE recommendation that thyroid function is monitored at 6-monthly intervals in people on lithium has been investigated. 1 We have demonstrated these tests are over-requested, which may lead to overdiagnosis of transient thyroid abnormalities which do not require treatment.

F I G U R E 4
Timeline of thyroid disorder-related events in patients with an abnormal TSH at 3 and 6 months after starting lithium. Vertical grey lines represent 13-week intervals. The chart shows events occurring up to 4 years after initiating lithium therapy

| Interval for thyroid function testing
Treatment with lithium is known to increase the risk of thyroid abnormalities, with a particular risk of hypothyroidism. 7,8 For this reason, NICE recommends biochemical monitoring of thyroid function every 6 months for patients treated with lithium. 1 Previous work has demonstrated that therapeutic drug monitoring of serum lithium is not consistent with the 3-monthly guidance, 11 and we hypothesized that this would extend to biochemical monitoring of thyroid function.
In support of this, we found that <10% of thyroid requests were made at the recommended 6-monthly intervals. Instead, the most common requesting interval for thyroid function testing was 3 months, indicating that thyroid function tests are overrequested relative to guidance. This interval corresponds to the recommended serum lithium monitoring frequency, suggesting that lithium and thyroid function tests may be co-requested during routine serum lithium testing. Although the monitoring frequency for thyroid function tests for patients on lithium remains previously unknown, this has been investigated for hypothyroid patients on levothyroxine replacement. 16 Here, it has been shown that thyroid monitoring is not compliant with guidance and can be subject to both under and over-requesting depending on the nature of the thyroid results. This study implies that our findings are consistent with a wider issue of appropriately timed thyroid monitoring.
The COVID-19 pandemic has had a negative impact on conditions that require blood tests for monitoring and diagnosis. This has been recently demonstrated for diabetes mellitus, where an estimated 1.41 million diabetes mellitus monitoring tests have been delayed or missed across the UK. 17 The impact of the pandemic on monitoring of lithium complications has not previously been investigated and our study has demonstrated that the pandemic has increased the thyroid function testing interval in patients on lithium by approximately 1 month. Counter-intuitively, as thyroid function tests are over-requested in these patients, this delay may indicate closer adherence to the 6-monthly monitoring guidance. One of the potential implications of this finding is that therapeutic drug monitoring of serum lithium levels has also been affected by the pandemic, which may result in increased psychiatric hospital admissions. 12

| Identification of patients with abnormal thyroid function within 3 months of commencing lithium
The fact that the peak monitoring interval for TSH was approximately 3 months, led us to investigate how many patients were identified as having an abnormal TSH earlier than if following the 6-monthly guidance. We hypothesized that more frequent monitoring may lead to early detection and treatment of thyroid disorders, particularly in patients pre-disposed to a thyroid disorder. To explore TA B L E 2 Thyroid results accompanying patients who developed an abnormal TSH within 3 and 6 months after starting lithium this, we used Kaplan-Meier curves to visualize the rate of thyroid disorder development. We observed that >90% of patients who developed thyroid dysfunction did so within 3 years of commencing lithium. The rate of thyroid disorder development was similar to previous studies. 8,18 It has been recently suggested that the serum lithium testing interval could be extended to 6 months for patients on stable lithium therapy. 13 The apparent decrease in detection rates after the first 12 months of initiating lithium therapy could similarly imply that the testing interval for thyroid testing could increase for some patients, further work to establish this is on-going.
A small number of patients were identified who developed thyroid abnormalities within 3 months which would not have been detected until 6 months if the interval recommended by the guideline was being followed. These patients were characterized and the clinical outcomes were assessed to determine whether there was a benefit in early detection.

| Relation between clinical outcomes and test interval
Of the patients who were identified within 3 months of starting In keeping with other studies, we found that the most common thyroid abnormality in patients on lithium was subclinical hypothyroidism. 7,10 Multiple factors dictate whether subclinical hypothyroidism benefits from thyroxine replacement, including magnitude and duration of TSH abnormality, development of anti-thyroid antibodies and presence of symptoms. 19 In this data set, the level of TSH did not predict whether treatment was initiated, with levels as high as 53.5 mIU/L not receiving replacement.

| Strengths/Limitations
Anti-thyroid antibodies were only measured in a small number of patients despite the fact that these tests may particularly benefit female patients who are more likely to develop anti-TPO antibodies. 18 A limitation of this study is that it was not possible to determine whether the development of symptoms contributed to either thyroxine replacement or a shorter interval between thyroid hormone measurements. In addition, detailed analysis of clinical outcomes was only available for 12 patients. Nevertheless, we feel that the results of our analysis make a relevant contribution to the literature in this area.

| CON CLUS ION
Our data suggest that thyroid function tests are typically requested too frequently in patients on lithium, potentially resulting in overdiagnosis of thyroid abnormalities. There was no evidence that early detection of abnormal thyroid results leads to earlier treatment and the initial thyroid dysfunction spontaneously resolves in half of the cases without intervention. More than 90% of patients who developed thyroid dysfunction did so within 3 years of commencing lithium. This work supports the 6-monthly testing interval recommended by NICE, 1 and highlights poor adherence to this important guideline.

CO N FLI C T O F I NTE R E S T
All authors declare no conflict of interest exists.

DATA AVA I L A B I L I T Y S TAT E M E N T
The data that support the findings of this study are available from HD on reasonable request.