Subclinical hypothyroidism, outcomes and management guidelines: a narrative review and update of recent literature

Abstract

Subclinical hypothyroidism (SCH) is diagnosed when serum thyroid stimulation hormone (thyrotropin; TSH) levels are above the reference range, accompanied by levels of free thyroxine within its reference range. The management of SCH remains a diagnostic and therapeutic challenge despite many years of research relating to its epidemiology, aetiology, effectiveness of treatment and safety. European Thyroid Association (ETA) guidelines for the management of SCH were published almost a decade ago. This narrative review summarizes the clinical literature relating to SCH and outcomes since the publication of these guidelines. Clinical evidence emerging during the previous decade generally supports the view that SCH is associated with adverse outcomes to an extent that is intermediate between euthyroidism and overt hypothyroidism although evidence that treatment with thyroid hormone replacement is beneficial is lacking. Accordingly, the rationale for the recommendations for intervention in the ETA guidelines based on the age of the patient, level of serum TSH, symptoms and comorbidities remains valid today.

Keywords:

• Subclinical hypothyroidism
• levothyroxine
• thyroid

Introduction

Thyroid homeostasis is governed by a complex and overlapping series of feedback systems, operating within and between the hypothalamus, the pituitary, the thyroid gland itself, and the many target organs for thyroid hormones throughout the body. Within this system, changes in the circulating level of thyroid hormones negatively affects the rate of secretion of thyrotropin (Thyroid Stimulating Hormone, TSH): an increase in thyroid hormones decreases TSH, and vice versa. Therapeutic intervention (with levothyroxine [LT4]) is uncontroversial for an individual with “overt” hypothyroidism, typically diagnosed on the basis of elevated serum TSH above the reference range with decreased free thyroxine (FT4) concentration below the reference range^1,2. Potential hazards of treatment (primarily related to sub-optimal therapy) are outweighed by the long-term clinical risks associated with uncontrolled overt hypothyroidism, which include heart disease or stroke, dyslipidaemia, goitre, infertility/sub-fertility, adverse pregnancy outcomes, and psychosocial disorders, in addition to symptoms causing impaired quality of life, such as fatigue^2–6. The use of LT4 to manage subclinical hypothyroidism (SCH), a more common condition affecting up to 10% of the general population, where the TSH level lies above the upper limit of the reference range but FT4 is normal^2,4, is more controversial⁷.

We review the prevalence and clinical consequences of SCH. The true clinical need for intervention to manage SCH will be determined ultimately by the effects of SCH (and LT4 treatment) on hard clinical outcomes. The principal guidance for the management of SCH from the European Thyroid Association (ETA) and the American Thyroid Association (ATA) dates back to 2012^4,8, and much research has been conducted since then. This article reviews the clinical evidence for an association of SCH with adverse clinical outcomes and the evidence base for the use of LT4 in selected adult patients with SCH, with a focus on research published within the last decade.

Overview of thyroid homeostasis and the management of hypothyroidism

Diagnosis and management

Small reductions in the secretion of thyroid hormones cause an increase in TSH that is much larger, by as much as 100-fold¹. In the periphery, specific deiodinases convert thyroxine to the active hormone triiodothyronine, and transform these hormones into other iodine-containing molecules which themselves may have biological activity^9,10. The relatively large increases in TSH associated with small reductions in thyroid hormones are easier and more accurate to interpret than the much smaller changes in the thyroid hormones themselves. Accordingly, the diagnosis of thyroid dysfunction in the routine care setting depends primarily on the level of serum TSH, compared with a reference range, derived from a population of (ideally local) people without thyroid dysfunction^2,4. The upper and lower bands of the reference range may vary slightly between different TSH assays, but are typically in the region of 0.5–4.5 mIU/L. For individuals who require intervention with thyroid hormone replacement to correct hypothyroidism, levothyroxine (LT4), a synthetic isomer of thyroxine, is the predominant treatment of choice: the dose of LT4 is titrated to maintain serum TSH level within its reference range².

Prevalence and clinical course of subclinical hypothyroidism

SCH is a relatively common disorder, and most population-based studies have demonstrated a prevalence of roughly 5–10% of the population^11–14. The prevalence of SCH may be decreasing in the general population, however, probably due to an increase in the number of people with relatively mild elevation of TSH who receive active treatment¹⁵. SCH is more common in women than in men, and also in older people. The latter finding may be confounded by observations that the average circulating level of TSH rises with age, increasing the likelihood that an older person’s TSH level might cross the upper boundary of the reference range, irrespective of underlying thyroid function, if an age-appropriate reference range for TSH is not used¹⁶. Individual variation in thyroid function, and potential for interference with the TSH result by comorbid conditions (including obesity), commonly used drugs, circulating macromolecules, ethnicity, iodine intake, pregnancy status, and time of day or year, suggest that minor fluctuations in the TSH level around the upper limit of the reference range should be interpreted with caution in the absence of clinical evidence of thyroid dysfunction^12,17–19. One study found that adjusting TSH reference ranges for subject-related factors such as gender, age and time of day or year would lead to a significant proportion of people being classed as having normal thyroid function and also lead to a substantial cost saving associated with avoidance of unnecessary treatment among people undergoing thyroid function tests²⁰. Observations such as these have led experts in the field to consider that the true prevalence of SCH may be overstated in some populations¹².

Higher levels of TSH in the setting of SCH have been associated with increased risk of progression to overt hypothyroidism and/or of reduced reversion to euthyroidism (Table 1)^21–26. The presence of anti-thyroid antibodies is a consistent predictor of progression from SCH to overt hypothyroidism (Table 1)^21,23,26.

Table 1. Overview of studies describing the clinical course of thyroid disease in subjects presenting with subclinical hypothyroidism.

Ref	N	Main findings
21	82	10-year incidence of overt HT according to baseline TSH was 0% (TSH 4–6 mIU/L), 42.8% (TSH >6–12 mIU/L), and 76.9% (TSH >12 mIU/L) (p < 0.001 for trend); the presence vs. absence of TPO Ab was associated with greater progression to overt HT (58.5% vs. 23.2%, p < 0.03)
22	3,996	Reversion to euthyroidism at 2 years was more common for TSH 4.5–6.9 vs. 7–9.9 or 10 mIU/L (46%, 10%, 7%, respectively, p < 0.001 among elderly subjects with SCH; the absence vs. presence of TPO Ab was associated with more frequent reversion to euthyroidism (48% vs. 15%, p < 0.001)
23	71	Progression to overt HT was higher for subjects with SCH (7.0%) vs, controls (1.6%) after adjustment for other factors (OR 4.56, p = 0.009); TPO Ab was also a significant multivariate predictor of progression
24	107	Increasing baseline TSH was associated with higher rates of conversion to overt HT (cases of overt HT/100 person-years were 1.76 (TSH 5–9.9 mIU/L), 19.67 (TSH 10–14.9 mIU/L), and 73.47 (TSH 15.0–19.9 mIU/L); TSH was the only significant multivariate predictor of progression
26	68	Optimal cut-off values for predicting overt HT were baseline TSH >2.5 mIU/L (vs. ≤4.0 mIU/L), TPO Abs >29 kIU/L (vs. 35 kIU/L), and Tg Abs >22 kIU/L (vs. 55 kIU/L); for women with TPO Ab or Tg Ab, the prevalence of overt HT after 14 years was 12.0% (baseline TSH ≤ mIU/L), 55.2% (TSH 2.5–4.0 mIU/L), 85.7% (TSH >4.0 mIU/L)

Abbreviations. Ab, antibodies; HT, hypothyroidism; Tg, thyroglobulin; OR, odds ratio; TPO, thyroid peroxidase; SCH, subclinical hypothyroidism; TSH, thyrotropin/thyroid stimulating hormone.

Clinical complications associated with subclinical hypothyroidism: updating the evidence base since the international guidelines

Approach

The approach in this section will be to consider the implications of research on several clinically important potential adverse clinical conditions associated with SCH that has been published since the publication of the ETA guideline for the management of SCH in 2013⁴. In each case, a brief overview of the scope of the evidence base provided in the 2013 guidance is presented in Table 2⁴ (information on the management of SCH in pregnancy was from a separate ETA guideline published in 2014⁵). The information presented below is a summary of a literature review of clinical evidence published over the last decade (see details of the search strategy in the “Supporting Information” section at the end of the article) for several important therapeutic areas. A brief summary of changes in the field since the publication of the 2013 guidelines is given in italic at the end of each section.

Table 2. Overview of evidence for associations between subclinical hypothyroidism (SCH) and adverse clinical outcomes in adult patients, presented in the 2013–4 European Thyroid Association (ETA) guidelines for the management of SCH^4,5.

Clinical issue	Overview of the extent of evidence presented for an aetiological relationship with SCH
Dyslipidaemia	Observational studies published before 2013 reported inconsistent associations between the level of TSH and effects on the lipid profile. Overall, the evidence at the time supported a modest, though clinically significant, adverse effect of increased TSH on the lipid profile, with elevations of total and LDL-cholesterol. There was a suggestion of more marked adverse effects on lipids effects in women and in older subjects. Randomized evaluations of LT4, and meta-analyses of these studies, supported a reduction in total- and LDL-cholesterol following LT4 therapy, but these effects may have been more marked where TSH was particularly high, possible weakening the association between dyslipidaemia and SCH (as opposed to overt hypothyroidism), as seen in most patients with the condition.
Obesity	Longitudinal studies demonstrated an association between increasing levels of TSH and higher body weight, including following careful adjustment for potential confounders of this relationship, such as age, smoking and menopausal status. Importantly, the relationship was bidirectional – increased TSH was associated with weight gain, and reduction in the TSH level was associated with weight loss.
ASCVD and vascular function	Three observational studies associated SCH with increased aortic or carotid IMT, a validated measure of atherosclerosis, and one of these associated SCH with an increased risk of MI. Three further studies demonstrated an association of SCH with endothelial dysfunction. RCTs demonstrated LT4-mediated improvements in endothelial function (2 studies) IMT (1 study) and cardiac dynamics (2 studies).
Ischaemic heart disease or stroke	Evidence for association of TSH with CHD events was consistent, especially for TSH >10 mIU/L. RCTs (mostly small) of LT4 and observational data suggested a possible, but modest, benefit for LT4 on CV risk. One observational study suggested a lack of association between SCH and stroke.
Diabetes	SCH and overt hypothyroidism coexist frequently with type 1 (autoimmune) diabetes. Hypothyroidism increases insulin resistance which may exacerbate or precipitate type 2 diabetes and metabolic syndrome (see also the sections on obesity and dyslipidaemia).
Heart failure and LV function	Several studies associated SCH with systolic and diastolic LV dysfunction. Some (not all) observational studies suggested a increased risk of heart failure events with increased TSH (including in patients with SCH), but much of this evidence related to TSH >10 mIU/L.
Bone health	The ETA guidelines did not discuss effects of SCH on bone health in adults.
Cancer	Good evidence to support an association between TSH levels (including within the normal range) and thyroid cancer. Encouraging (but not definitive) cross-sectional evidence for reduction of thyroid cancer risk with LT4 in people with nodular goitre.
Pregnancy outcomes^a	Although overt hypothyroidism clearly promotes adverse pregnancy outcomes (including pre-eclampsia, pregnancy loss, devastating developmental impairment of offspring), associations between SCH and adverse pregnancy outcomes (and the effects of LT4) were less clear. However, several studies associated SCH with pregnancy loss, gestational hypertension and pre-eclampsia. SCH was also associated with gestational diabetes. Data on SCH and other pregnancy complications, and on associations between SCH in pregnancy and neuropsychological impairment in offspring, were conflicting.
Psychological outcomes	Multiple placebo-controlled or observational studies demonstrated improvements in mood or psychological function when LT4 was used to control TSH; conversely, mood/psychological function deteriorated when LT4 was withdrawn. This was not demonstrated for TSH <10 mIU/L, however. Heterogeneous study populations and study designs seem to have complicated the interpretation of these data.

References supporting the conclusions from the 2013 guidance can be found within the original article and are not reproduced here for conciseness. Abbreviations. ASCVD, atherosclerotic CV disease; CV, cardiovascular; IMT, intima-media thickness; LV, left ventricle (or ventricular); RCT, randomized, controlled trial; SCH, subclinical hypothyroidism.

^aAll material is from the 2013 ETA guideline for the management of SCH⁴ except for pregnancy outcomes, which was abstracted from the 2014 ETA guideline for the management of SCH in pregnancy and children⁵.

Cardiovascular risk factors

Dyslipidaemia

Perturbation of the lipid profile has been a longstanding, and one of the most consistent, findings in people with SCH, although it remains unclear whether LT4 treatment is beneficial²⁷. In a study from 2015, a population of 136 normotensive patients with early type 2 diabetes, early diabetic nephropathy and TPOAb-positive SCH (TSH 4.0–7.0 mIU/L) were randomized to LT4 or placebo for 48 weeks²⁸. Treatment with LT4 vs. placebo induced modest but potentially clinically significant reductions in total cholesterol (mean treatment difference −0.37 mmol/L [–14 mg/dL]) and LDL-cholesterol (mean treatment difference −0.13 mmol/L [–5 mg/dL]).

A meta-analysis published in 2017 (12 studies, 940 patients) of the effects of LT4 on the lipid profile in people with SCH found an average reduction of −0.29 mmol/L (–11 mg/dL) in total cholesterol, and of −0.22 mmol/L (9 mg/dL) in LDL-cholesterol, with no effect on HDL-cholesterol or triglycerides (TG)²⁹. These changes were similar overall and in patients with TSH <10 mIU/L at baseline. Other meta-analyses from 2017³⁰ and 2014³¹ reported similar results. A 15-month trial in 369 patients with SCH demonstrated improved LDL-cholesterol in patients receiving LT4 vs. controls (no treatment)³². In addition, LT4 treatment was associated with reduced atherogenic remnant lipoproteins, associated with increased hepatic lipase activity, in a cross-sectional study involving 37 women with SCH (TSH ≥4.5 mIU/L) compared with age- and body mass index (BMI)-matched 29 euthyroid women³³.

Other meta-analysis of studies in people with polycystic ovary disease (PCOS) described adverse effects of SCH on HDL-cholesterol and TG, with no significant effects on total- or LDL-cholesterol³⁴, or deterioration of TG, total-, LDL- and HDL-cholesterol^35,36. These meta-analyses also demonstrated higher insulin resistance in people with SCH, which is interesting as insulin resistance is a central pathogenic feature of PCOS³⁷. Conversely, the risk of having SCH was increased by 2.9-fold (95% confidence intervals [CI] 1.82–9.92) for women with vs. without PCOS, and by 3.6-fold (95% CI 2.25–5.73) in studies where only women with TSH >4 mIU/L were included³⁸.

Overall, new data from a randomized trial and, especially, meta-analyses support earlier findings of disturbance in the lipid profile in people with SCH. These studies also suggest improvements in the lipid profile during treatment with LT4 which are modest but could nevertheless contribute to a reduction in the risk of adverse cardiovascular outcomes.

Obesity

Abdominal obesity, measured as increased waist circumference, is one of the five diagnostic criteria for the metabolic syndrome. An updated meta-analysis of studies published up to 2020 found a significant association between SCH and different versions of the metabolic syndrome, and between SCH and 4/5 of its components³⁹. The odds ratio (OR) for abdominal obesity in subjects with or without SCH was 1.21 (95% CI 1.07–1.37). Other meta-analyses (2016) also associated SCH with the metabolic syndrome (although not with all definitions of the metabolic syndrome)^40,41. Treatment of people with SCH with LT4 did not reduce BMI significantly after 6 or 12 months of treatment, according to a systematic review and meta-analysis from 2018⁴². It is interesting that treatment of SCH with LT4 has not been shown to reduce body weight significantly whereas serum TSH levels in individuals with obesity undergoing bariatric surgery demonstrate a sustained and significant reduction, especially where TSH was raised before surgery^43–46.

Little has changed in our understanding of the relationships between SCH and obesity per se since the publication of the European guideline on SCH in 2013, beyond further associations of SCH with the metabolic syndrome and abdominal obesity. No new evidence has emerged supporting a benefit for LT4 treatment on adiposity.

Vascular dysfunction

The 2013 guideline for the management of SCH described several reports of associations between SCH and structural and functional abnormalities of the myocardium and vasculature; new evidence from randomized or observational studies in discussed below and summarized in Table 3. Observational data from 32 children with SCH and 32 age- and gender-matched controls suggested that the level of TSH and accumulation of epicardial fat acted as drivers of impaired vascular function⁴⁷. A meta-analysis from 2018 (27 case-control studies including 1,065 SCH cases) reported reduced endothelial function in the SCH group⁴⁸.

Table 3. Summaries of randomized, controlled trials or observational studies that evaluated the effects of SCH and its treatment on classical cardiovascular risk factors.

Ref/y	Risk factor or outcome	Design	N^a	Overview of key findings
50	Vascular dysfunction	RCT^b	184	18 months of treatment with LT4 vs. placebo did not reduce mean CIMT (mean treatment difference 0.02 mm [95%CI, −0.01 to 0.06], p = 0.30); no difference between groups for proportion of carotid plaque, plaque thickness, and no signal for treatment effect in subgroups defined by age, baseline TSH, or presence/absence of CVD
66	AMI	RCT	95	1 year of LT4 vs. placebo did not affect LVEF significantly vs. placebo (mean difference 0.76% [95%CI −0.93 to 2.46], p = 0.37) in a population with SCH admitted for AMI; secondary endpoints also did not demonstrate differences between treatment groups (LV volumes, infarct size, AE, patient-reported outcomes [health status, HRqOL, depression])
47	Vascular dysfunction	Obs	32^c	Children with SCH (autoimmune thyroiditis) vs. controls had higher atherogenic index (AI), levels of hs-CRP EFT and lower endothelial function (flow-mediated vasodilatation); epicardial fat thickness correlated with TSH, hs-CRP, AI, and endothelial dysfunction
51	Vascular dysfunction	Obs	18^c	CIMT was higher for people with SCH vs. healthy controls; six months of treatment with LT4 was associated with a significant reduction in CIMT after 3 and 6 months of treatment
68	Blood pressure	Obs	3,078	No difference in office BP according to presence vs. absence of SCH but measures of ambulatory BP were higher for SCH vs. euthyroid, with more in the SCH group not displaying a night-time dip in BP; SCH was associated with sustained and masked hypertension.
57	CHF	Obs	54	Subjects with vs. without SCH had lower E/A ratio, higher E/e’ sep. ratio, higher myocardial performance index (MPI), lower global longitudinal strain (GLS), and lower S wave (tissue Doppler imaging); 5 months of LT4 treatment were associated with higher ejection fraction, lower E/e’ sep. ratio, lower MPI, and improved GLS vs. baseline TSH levels correlated negatively with ejection fraction, E/A ratio, GLS, and S/TDI and correlated positively with E/e’ sep., irrespective of thyroid status
58	CHF	Obs	80	LV systolic and diastolic function and RV function were impaired in women with SCH vs. 35 control subjects; LV mass remained higher for SCH vs. controls after LT4 therapy
59	CHF	Obs	38	Lower phosphocreatine:ATP ratio was associated independently with SCH implying impaired cardiac bioenergentics; LT4 treatment improved this parameter so that it did not differ significantly from controls
60	CHF	Obs	20	Cardiac magnetic resonance (CMR) myocardial longitudinal relaxation time (T1) mapping demonstrated diffuse myocardial injury and deficits correlated positively with TSH and negatively with FT4; LV diastolic function was poorer in subjects with vs. without SCH
63	CHF	Obs	1,100	SCH was associated with lower O2 consumption and higher pulmonary artery pressure (no difference for LV function) in patients with pre-existing CHF; SCH significantly predicted increased risk of MACE and mortality
76	MACE	Obs	3,021	Subjects with SCH and TSH in the highest quartile were at increased risk of CV events and death, especially for subjects with high CV risk according to risk factor scoring
77	MACE	Obs	344^d	Elevated TSH predicted increased risk of CV events, but adding TSH to conventional risk factor scoring had little effect on risk prediction
78	MACE	Obs	936	SCH vs. euthyroidism in patients who had undergone coronary revascularization independently predicted increased risk of revascularization, cardiac death and a composite event occurred more frequently in the SCH group than in the euthyroidism; TSH levels predicted the risk of a CV event

^aNumber of subjects enrolled. ^bRCT nested within the Thyroid Hormone Replacement for Subclinical Hypothyroidism trial. ^cPlus the same number of matched control subjects. ^d344 with SCH plus 2,624 euthyroid controls.

Abbreviations. AE, adverse event; ASCVD, atherosclerotic CV disease; CV, cardiovascular; IMTm intima-media thickness; LV, left ventricle (or ventricular); RCT, randomized, controlled trial; AMI, acute myocardial infarction; CHF, congestive heart failure; CIMT, carotid intima-media thickness; CV, cardiovascular; HRqOL, health-related quality of life; LT4, levothyroxine; LV, left ventricular; LVEF, left ventricular ejection fraction; MACE, major adverse cardiovascular event; Obs, observational study; RCT, randomized, controlled trial; RV, right ventricular; SCH, subclinical hypothyroidism; TSH, thyrotropin/thyroid stimulating hormone. Type 2 diabetes and obesity are not included here as were only meta-analyses were published post-2014 in these areas.

Carotid intima-media-thickness (CIMT) is a validated surrogate measure of the overall burden of atherosclerosis⁴⁹. Eighteen months of LT4 treatment did not reduce CIMT progression vs. placebo in a randomized trial in 185 older (average age 74 years) subjects with SCH (TSH 4.6–19.99 mIU/L) in a randomized trial⁵⁰. A 6-month prospective observational study found significant progression of CIMT in 18 patients with SCH (with median TSH 6.15 mIU/L) compared with 18 healthy controls; moreover, CIMT regressed during treatment with LT4 in the SCH group⁵¹.

The results of meta-analyses^48,52, together with an additional meta-analysis (2013, 3,602 patients with SCH in 8 observational studies⁵³), supported an adverse effect of SCH on CIMT progression. Further meta-analyses (2017, 3 studies, 117 subjects⁵⁴ and 2017, 12 studies, 543 subjects⁵⁵) reported regression of CIMT with LT4 treatment in people with SCH, with larger effects occurring after longer treatment (>6 months) in the latter study. A meta-analysis (2019, 10 studies involving 760 subjects) demonstrated significant association of SCH with endothelial dysfunction (decreased flow-mediated dilatation)⁵².

The more recent studies described here continue to support the associations between SCH and vascular dysfunction. Interestingly, the newer studies report a beneficial effect of LT4 treatment on the progression of atherosclerosis, as measured by CIMT progression. Reductions in CIMT progression have been shown to correlated with reduced risk of adverse cardiovascular outcomes in a large meta-analysis⁴⁹. Accordingly, these data support the hypothesis that intervention with LT4 in SCH might improve cardiovascular outcomes⁵⁶, and the current evidence base for this is summarized later in this article.

Heart failure and cardiac dysfunction

Reports of cardiac abnormalities in people with SCH were also cited in the 2013 management guideline, and this has been an active area of research during the following decade. Reports of new data in this area (randomized or observational studies) are summarized below and in Table 3. A population of 54 subjects with SCH and TSH 4.2–10 mIU/L had adverse measures of multiple systolic and diastolic LV function, measured using echocardiography, compared with 30 euthyroid control subjects⁵⁷. These measurements correlated with the TSH level in both the SCH and control groups. In addition, 5 months of LT4 treatment improved systolic and diastolic function in the SCH group. Another observational study, in 45 women with SCH and 35 age-matched euthyroid control women, found that the SCH vs. control group displayed deficits in right and left ventricular function, and systolic and diastolic function, which were partially reversed during 1 year of LT4 treatment⁵⁸. Six months of LT4 treatment improved markers of myocardial mitochondrial function (phosphocreatine [PCr]:ATP ratio) in another study⁵⁹. The magnitude of deficits in PCr:ATP vs. control subjects in this study correlated positively with the TSH level. Magnetic resonance imaging (MRI) revealed diffuse areas of myocardial injury in people with SCH that correlated positively with TSH and negatively with FT4, especially where TSH was ≥10 mIU/L⁶⁰.

Meta-analytic data (2013, 14 cross-sectional studies with 675 participants) reported that various parameters of diastolic LV function were impaired in people with SCH vs. euthyroid controls⁶¹. A very recent meta-analysis (2022, 11 studies involving 294 patients) confirmed this finding, and also found that aspects of LV function improved from baseline after LT4 therapy (cardiac output, LV ejection fraction [LVEF] and E/A ratio), although there was no change in cardiac morphology associated with LT4 treatment⁶².

Two observational studies evaluated the relationship between SCH and the prognosis of pre-existing heart failure. Firstly, follow-up of 1,100 consecutive patients with heart failure at baseline for an average of about 3 years revealed a significant association between SCH, reduced cardiopulmonary fitness and increased pulmonary artery pressure⁶³. The second study, involving 4 years of follow-up of 1,365 patients with heart failure, demonstrated an increased risk of a composite outcome of requirement for a ventricular assist device, heart transplantation, or death for people with SCH vs. euthyroid status⁶⁴. A meta-analysis of studies in patients with comorbid SCH and heart failure with reduced LVEF (HFrEF; 2019, 14 studies, 21,221 patients) found that SCH predicted significantly increased risk of death (by 45%), and of cardiac death/hospitalization (by 31%)⁶⁵.

Randomization of 95 patients with SCH (median TSH 5.7 mIU/L) to LT4 (titrated to achieve TSH in the lower half of the reference range) for 1 year following acute myocardial infarction (MI) had no effect on LVEF, or other measures of cardiac structure and function, in a randomized trial⁶⁶. Similar negative findings for LT4 vs. placebo emerged from a randomized sub-study of the TRUST trial in patients aged ≥65 years with “mild SCH” (baseline TSH 6.35 mIU/L)⁶⁷.

Observational studies continue to associate SCH strongly with heart failure and cardiac dysfunction. Intervention with LT4 post-MI or in an elderly population did not affect measures of cardiac performance, although we await a randomized, controlled clinical evaluation of LT4 in patients with pre-existing heart failure and SCH.

Other cardiovascular risk factors

The sections immediately above cite a series of (mainly observational) studies that associate SCH with vascular and/or cardiac dysfunction, so that the effects of SCH and LT4 on cardiovascular outcomes are of particular interest. This section considers the effects of SCH and/or LT4 on cardiovascular risk (Table 3).

Observational data from 3,078 subjects with or without SCH demonstrated no significant differences between office systolic or diastolic blood pressure (SBP and DBP) values⁶⁸. However, ambulatory BP was higher in the SCH group, and SCH was associated statistically with sustained and masked hypertension. Meta-analyses have shown that SCH was associated significantly with hypertension especially in middle-aged, but not in older women⁶⁹, or that associations between SCH and BP were small and uncertain⁷⁰. An additional meta-analysis of randomized trials found that LT4 treatment reduced SBP by a modest, but clinically significant, −2.48 mmHg (–4.63 to −0.33)⁷¹. A larger reduction in BP was found by pooling prospective observational studies (–4.80 [–6.50 to −3.09]/2.74 (–4.06 to −1.43] mmHg).

Table 4 summarizes the effects of SCH and/or LT4 on additional cardiovascular risk factors^72–74. SCH has been associated with reduced fibroblast growth factor-15⁷², a shift towards a more hypercoagulable state^73,74 (largely corrected by LT4 therapy in one study⁷³) and increased circulating homocysteine⁷⁵.

Table 4. SCH and cardiovascular risk factors other than obesity, diabetes, chronic kidney disease or high blood pressure.

Ref/y	Risk factor	Design	N^a	Overview of key findings
72/2016	FGF-19	Obs	99	Significantly decreased fibroblast growth factor-19 (FGF-19) levels, an endocrine factor associated with metabolic regulation, in SCH or overt hypothyroidism vs. euthyroid controls in a cross-sectional study. The presence vs. absence of isolated anti-thyroid antibodies did not affect serum FGF.
73/2015	Haemostasis	Obs	82	SCH was associated with a prothrombotic state vs. healthy controls (high Factor VII, PAI-1, tPA, mean platelet volume, higher platelet aggregability, and low D-dimer). Six months of LT4 improved all of these parameters significantly, except for a non-significant trend to improvement in D-dimer.
74/2022	Haemostasis	MA	12/1,325	People with SCH had a hypercoagulable and hypofibrinolytic haemostatic profile with increased tPA and PAI-1, compared with a control group. No difference for activated partial thromboplastin time (APTT) or D-dimer between groups
75/2020	Homocysteine	MA	12/684	Slight but significant increase in plasma homocysteine, an amino acid associated with higher cardiovascular risk, for people with SCH vs. controls aged 18–65 years.

^aNumber of subjects enrolled (no. of trials/patients for meta-analyses). Abbreviations. MA, meta-analysis; Obs, observational study.

A 12-year observational study in 3,021 subjects in Korea reported an approximate doubling of the risk of cardiovascular and all-cause mortality in those with TSH >6.7 mIU/L vs. euthyroid status. Effects were more pronounced in younger patients and those at elevated CV risk⁷⁶. Another observational study showed that the incidence of major adverse cardiovascular events (MACE) in people with SCH was higher than that predicted by risk scoring among 344 people with SCH and 2,624 euthyroid controls, followed for 10 years⁷⁷. An observational study in 936 patients undergoing percutaneous coronary intervention for ischaemic heart disease found a 52% adjusted increase in the risk of a composite adverse cardiac outcome (cardiac death, non-fatal MI or repeat revascularization) for 100 with SCH vs. the remainder of the population⁷⁸. An observational study in 1,100 people with heart failure showed that those with SCH demonstrated a higher risk of cardiac events or death vs. euthyroid subjects⁶³.

A meta-analysis suggested that the risk of major adverse cardiovascular events was higher for men vs. women with SCH, but only after at least 10 years of follow-up⁷⁹. SCH was significantly associated with cardiovascular disease and mortality, but the effect on mortality was seen only patients aged ≤65 years and in patients with elevated baseline cardiovascular risk, according to a large meta-analysis (35 studies with 555,530 subjects), reminiscent of the results of the large observational study, described above⁸⁰. Another meta-analysis, in this case of 5 observational and 2 randomized studies in a total of 21,055 subjects found that treatment with LT4 in patients aged ≤65 years reduced all-cause mortality (RR 0.50 [95%CI: 0.29–0.85, p = 0.011) and cardiovascular mortality (RR 0.54 [95%CI: 0.37–0.80], p = 0.002)⁸¹. In older patients in this study, there was no reduction with LT4 for all-cause mortality (RR 1.08 [95%CI: 0.91–1.28, p = 0.363) or cardiovascular mortality (RR1.05 [95%CI: 0.87–1.27], p = 0.611)⁸¹.

Another large meta-analysis (2014, 6 studies, 38,274 participants for CHD mortality and 4 studies, 33,394 participants for CHD events) found no effect of anti-peroxidase antibody status on the risk of adverse cardiovascular during follow-up of up to 460,333 person-years⁸².

Forty-three of 229 patients with minor stroke or transient ischaemic attack were found to have SCH in a single-centre observational study and the SCH group bore a higher prevalence of white matter lesions, cerebral microbleeds and cerebral small vessel disease⁸³. A meta-analysis (2014, 5 studies, 10,018 subjects) found no significant association between subclinical hypothyroidism and stroke (HR 1.08 [0.87–1.34])⁸⁴.

There is now strong evidence from observational and meta-analytic studies that SCH is associated with increased cardiovascular risk and increased risk of adverse cardiovascular outcomes. The role of modest elevations of TSH in promoting adverse cardiovascular outcomes remains unclear, and needs to be studied in populations with varying levels of cardiovascular risk due to risk factors that are modifiable and unmodifiable (especially age). There remains no compelling evidence that intervention with LT4 reduces this risk, however, and randomized trials in this area are clearly needed.

Type 2 diabetes

Three meta-analyses relating to diabetes and SCH have been published. The first (2022, 18 studies, 61,178 subjects), used individual patient-level data to demonstrate an essentially identical risk of incident diabetes in patients with vs. without SCH (age- and gender-adjusted OR 1.03 (95% CI 0.82–1.30)⁸⁵. An analysis published in 2015 and incorporating 61 studies, found that the presence of type 2 diabetes increased the risk of concurrent SCH (OR 1.93 [1.66–2.24])⁸⁶. Conversely, the presence of SCH appeared to exacerbate the severity of diabetes, as shown by an increased risk of diabetes complications, namely for diabetic nephropathy (OR 1.74 [1.34–2.28]), for diabetic retinopathy (OR 1.42 [1.21–1.67]), peripheral arterial disease (OR 1.85 [1.35–2.54]), and for diabetic peripheral neuropathy (OR 1.87 [1.06–3.28). The third meta-analysis (2015, 8 studies in 3,631 patients) also reported an increased risk of diabetic retinopathy in people with SCH (OR 2.13 [1.41–3.23])⁸⁷. No randomized controlled trials of LT4 therapy assessing the impact of treatment on diabetes-related complications have been conducted in people with SCH and diabetes.

The data associating SCH with type 2 diabetes are conflicting. The strong design of the patient-level meta-analysis (that found no association between SCH and type 2 diabetes risk) and the lack of new data on LT4 treatment in this population suggest that consideration of the risk of type 2 diabetes is not currently an important factor guiding the management of people with SCH.

Cancer

Thyroid cancer was considered as a possible complication of thyroid disease in the 2013 guideline, but there was no discussion of the risk of any cancer, or of specific types of cancer. An observational study in 622 patients reported an increased incidence of non-medullary thyroid cancer for TSH >1.64 mIU/L (OR 2.57 [1.41–4.70]) among patients with one or more thyroid nodules^88,89. However, a prospective, large (164,596 initially cancer-free subjects followed for a total of 2,277,750 person-years), cohort study in the Republic of Korea found a significant association between low TSH (hyperthyroidism) and thyroid cancer⁹⁰. The relationship between thyroid status and thyroid cancer thus remains poorly defined.

With regard to other types of cancer, a large observational study in 115,746 adults in Taiwan associated SCH significantly with overall cancer incidence (relative risk [RR] 1.51 [1.06–2.15]), as well as cancers of the mouth (RR 1.51 [1.06, 2.15]), gut (RR 1.64 [1.00, 2.70]), and for bone, skin and breast (pooled RR 2.79 [1.01–7.70])⁹¹. Another case control study, based on 20,990 patients with colorectal cancer and 82,054 control patients from a UK primary care database (The Health Improvement Network) reported an increased risk of colorectal cancer in patients with untreated subclinical or overt hypothyroidism vs. euthyroid state (OR 1.16 [1.08–1.24]) and a decreased colorectal cancer risk for LT4-treated subclinical or overt hypothyroidism vs. euthyroid (OR 0.92 [0.86–0.98])⁹². In addition, the magnitude of the apparent protective effect increased in line with the duration of thyroid hormone replacement. Unfortunately, this publication did not differentiate between the effects of LT4 in patients with SCH and overt hypothyroidism.

Observational data from 1,587 subjects found no association between TSH and cancer death (relatively hazard ratio 0.96 [0.85–1.07] per mIU/L) or between SCH vs. euthyroid and cancer death (relative hazard 0.80 [0.29–2.23])⁹³. Higher TSH was associated with poor outcomes in hepatocellular carcinoma in another observational study of 838 patients with this disorder⁹⁴.

Associations between SCH and cancer risk remain speculative and controversial, as no randomized trial has confirmed the effects of elevated TSH or LT4 treatment on cancer incidence or outcomes. The associations described above may have resulted from confounding by an unknown factor or been mediated by conditions such as obesity or ageing.

Pregnancy

Numerous publications have appeared since 2013 concerning possible links between SCH and adverse pregnancy outcomes. Table 5(a) provides and overview of the main results of randomized or observational trials that reported pregnancy outcome or effects on pregnancy rates^95–104 and systematic reviews/meta-analyses are summarized in Table 5(b)^105–117.

Table 5. Publications on pregnancy outcomes in people with subclinical hypothyroidism (SCH) that have appeared since the publication of the 2014 European Thyroid Association guideline for the management of SCH during pregnancy.

(a) Randomized and observational studies
Ref/y	N^a	Design	Overview of key findings
95/2022	1,736	RCT	LT4 treatment increased the live birth rate and reduced spontaneous pregnancy loss in women who were TPOAb + or who had SCH; no effect of LT4 on women without recurrent pregnancy loss irrespective of TPOAb or thyroid status.
96/2018	573 neonates	RCT	No difference in neonatal TSH according to LT4 use by mothers with hypothyroxinaemia or SCH (pre-specified secondary analysis of data from 2 prior RCTs)
97/2017	677	RCT	No effect of LT4 on neurocognitive outcome of children of mothers treated with LT4 or placebo during gestational weeks 8–20.
98/2016	1,671	RCT	Women screened for thyroid dysfunction in pregnancy reported fewer miscarriages or foetal macrosomia, and more Caesarean sections, than those who were not.
99/2021	1,014	Obs	No significant effect of LT4 vs. control on live birth rates in pregnant women with SCH and a history of pregnancy loss: OR 0.95 (0.23–3.83).
100/2018	534	Obs	No effect of SCH on the risk of subsequent PCOS in obese women: OR 0.984 (0.581–1.667).
101/2017	270	Obs	No difference in clinical pregnancy rate, miscarriage, or live births for groups with TSH in the upper vs. lower part of the reference range in a prospective cohort study of women undergoing first assisted reproduction.
102/2017	5,405	Obs	LT4-treated women with SCH in a large US claims database had lower median TSH (median 3.3 mIU/L) vs. non-treated women with SCH (4.8 mIU/L). The LT4 group had lower ORs for pregnancy loss (0.62 [0.48–0.82]) but higher ORs for preterm delivery (1.60 [1.14–2.24]), gestational diabetes (1.37 [1.05–1.79]), and pre-eclampsia (1.61 [1.10–2.37]); the benefit for pregnancy loss was seen only for pre-treatment TSH of 4.1–10 mIU/L and not for 2.5–4 mIU/L.
103/2016	1,193	Obs	No difference for TSH ≥2.5 mIU/L vs. <2.5 mIU/L in risk ratios for pregnancy loss (1.07 [0.81–1.41]), live births (0.97 [0.88–1.07]) or time to pregnancy (1.09 [0.90–1.31]) in a prospective cohort study
104/2013	286	Obs	No difference in live birth rate between women with SCH or no thyroid dysfunction or between treated or untreated SCH in a cohort of women with a history of pregnancy loss.
(b) Systematic reviews and meta-analyses
Ref/y	N^a	Overview of key findings
105/2022	22/108,831	Pregnant women with SCH had increased risk of hypertensive disorder of pregnancy vs. euthyroid pregnant women (OR 1.54 [1.21–1.96]), with risk increasing with TSH >4 IU/m.
106/2022	6/7,955	Study of women diagnosed with SCH according to 2017 US guidance.^6 Treatment with LT4 vs. a control group was associated with lower OR for pregnancy loss pregnant women with SCH treated with LT4 had a lower risk of pregnancy loss (0.55 [0.43–0.71), preterm birth (0.63 [0.41–0.98]) and gestational hypertension (0.78 [0.63–0.97]).
107/2021	13/7,342	Pregnant women with SCH treated with LT4 vs. controls had lower RR for pregnancy loss (0.79 [0.67–0.93]) and neonatal death (0.35; [0.17–0.72]); no difference for other pregnancy outcomes or cognitive status of offspring.
108/2020	17/–^b	No influence of SCH or LT4 on recurrent pregnancy loss. Thyroid autoimmunity predicts recurrent pregnancy loss.
109/2020	13/11,503	Lower OR for pregnancy loss (0.78 [0.66–0.94) and higher OR for live birth (2.72 [1.44–5.11]), but also higher OR for preterm labour (1.82 [1.14–2.91]) for LT4 treatment vs. placebo in pregnant women with SCH.
110/2019	4/820	No clear conclusions could be drawn concerning the effects of LT4 on fertility outcomes in women with SCH (with or without thyroid autoimmunity) undergoing assisted reproduction (Cochrane review).
111/2019	11/56,055	Increased risk of gestational diabetes in women with SCH and thyroid autoimmunity (OR 3.22 [1.72–6.03]).
112/2018	4787	Decreased miscarriage rate for LT4 vs. placebo -treated women undergoing assisted reproduction who had SCH and thyroid autoimmunity (RR 0.51 [0.32–0.82]); the effect was not significant in women with SCH or thyroid autoimmunity. No effect on live births, preterm births, or clinical pregnancy rate.
113/2018	3/1,837	No effect of LT4 vs. no LT4 on a range of outcomes including miscarriage, gestational hypertension, pre-eclampsia, preterm delivery, mode of delivery, neonatal intensive care unit admission, birth weight, gestational age at delivery, childhood IQ and neurodevelopmental scores.
114/2018	15/1,896	Maternal SCH in pregnancy increased the risk of foetal growth restriction (RR 2.4 [91.56–3.7]) and preterm delivery premature delivery (RR 1.96 [1.34–2.88]). Significant effects of maternal SCH were also reported for foetal distress and postnatal development of the offspring.
115/2017	9/17,931	Higher RR for miscarriage for pregnant women with untreated SCH vs. euthyroid (1.90 [1.59–2.27]) but not for LT4-treated SCH vs. euthyroid (1.14 [0.82–1.58]). Also higher RR for miscarriage for presence vs. absence of thyroid autoimmunity in women with SCH (2.47 [1.77–3.45]) and for SCH without thyroid autoimmunity vs. euthyroid (1.45 [1.07–1.96]).
116/2016	18/3,995	Higher RRs for pregnancy loss (2.01 [1.66–2.44]), placental abruption (2.14 [1.23–3.70]), premature rupture of membranes (RR 1.43 [1.04–1.95]), and neonatal death (2.58 [1.41–4.73]) for presence vs. absence of SCH.
117/2013	3 220	LT4 vs. no LT4 was associated higher RR for delivery rate (2.76 [1.20–6.44]), and lower RR for miscarriage (0.45 [0.24–0.82]), with no effect on clinical pregnancy rate (1.75 [0.90–3.38]) in women undergoing assisted reproduction.

^aNumber of subjects enrolled (no. of trials/patients for meta-analyses). ^bNot given.

Abbreviations. Obs, observational study; OR, odds ratio; PCOS, polycystic ovary syndrome; RCT, randomized, controlled trial; RR, relative risk. Figures in parentheses with ORs or RRs are 95% confidence intervals.

A randomized trial suggested improved pregnancy outcomes with LT4 treatment in women with a history of recurrent pregnancy loss if they had SCH or thyroid autoimmunity⁹⁵. However, another randomized trial found no effect of LT4 on pregnancy or neonatal outcomes, nor on the neurocognitive development of offspring at ages up to five years⁹⁷. Maternal LT4 use did not affect neonatal TSH levels in another randomized study⁹⁶. The results of observational studies reported conflicting findings regarding the effect of SCH itself^100,104, variations of TSH within the normal range^101,103, or LT4 on pregnancy outcomes^99,102,104. One meta-analysis supported an association between SCH during pregnancy and developmental issues in offspring¹¹⁴, but another meta-analysis did not¹¹³.

Meta-analyses in the last 10 years (Table 5(b)) generally supported an adverse effect of SCH on pregnancy outcomes^{105,111,115,116}, although some did not¹⁰⁸. Treatment with LT4 vs. a non-LT4 control group improved pregnancy outcomes in several meta-analyses, especially for the outcome of miscarriage/pregnancy loss^{106,107,109,115,117}, although again this finding was not universal^108,113. Some of these analyses suggested benefit of LT4 treatment in women undergoing assisted reproduction, although a Cochrane review from 2019 was unable to reach a conclusion on this point to due insufficient quality of evidence¹¹⁰. One meta-analysis suggest an increased risk of preterm labour with LT4 (along with improved miscarriage and birth rates)¹⁰⁹, though this was not seen elsewhere^106,111.

The large volume of research conducted into the management of SCH in women during pregnancy has not yielded greater clarity on the management of this population. Currently, the evidence base does not support intervention during pregnancy to improve pregnancy outcomes, although further data from populations with a history of pregnancy loss are required.

Psychological dysfunction and cognition

Psychological health is a key determinant of overall quality of life that was not addressed in the 2013 guideline for the management of SCH. Recent (2019–2021) meta-analyses suggest that SCH increases the risk of depression^118–120. One of these analyses found a clearly stronger association between depression and overt hypothyroidism (OR 1.77 [95% CI 1.13–2.77]) than with SCH (OR 1.13 [95% CI 1.01–1.28])¹²⁰. Cognitive dysfunction is another possible complication of hypothyroidism¹²¹: meta-analysis either did not link SCH with cognitive decline^122,123, or found such an association only in people aged <75 years or in those with higher TSH¹²⁴. A study using functional MRI demonstrated alterations of neural networks consistent with observations of reduced working memory in people with SCH compared with euthyroid controls¹²⁵.

A small, randomized trial (N = 60 subjects with SCH) showed that randomization to LT4 was followed by an improved score for the validated Beck Depression Inventory, while randomization to placebo was not¹²⁶. The improvement was driven almost exclusively by changes in somatic subscale score, rather than in the affective domain. A pre-specified secondary outcome analysis in 427 older participants with SCH enrolled in a randomized controlled trial (TRUST) evaluated the effect of LT4 on depressive symptoms¹²⁷. There were no differences at 12 months between the LT4 and placebo groups in Geriatric Depression Scale (GDS-15) scores. The results of a previous meta-analysis of effects on this parameter remained negative after adding these data. In another ancillary study to a previous randomized trial, in 245 pregnant women with SCH, LT4 given at 11–20 weeks’ gestation did not improve another measure of depression, the Center for Epidemiological Studies-Depression scale (CES-D), either in the 3rd trimester or at 1 year postpartum¹²⁸. Finally, treatment with LT4 for 6 months abolished the deficit in working memory seen in an fMRI study (see above)¹²⁶.

The effect of SCH on psychological well being is probably modest, compared with overt hypothyroidism, and the effects of LT4 on this outcome are again mixed.

Bone health and fractures

Fractures are an important consequence of hyperthyroidism, rather than hypothyroidism, but it is possible in principle that LT4 could adversely affect bone health, and this subject is addressed here for completeness. LT4 treatment did not affect bone mass, bone mineral density or bone mineral content in older subjects with SCH in the TRUST study, consistent with effects on other endpoints, described above^129,130. Meta-analyses have demonstrated no adverse effects of SCH on bone¹³¹, a modest increase in fracture risk¹³², or no increase in fracture risk^133–135. In one study an association between SCH and fracture risk became non-significant when lower quality studies were excluded¹³².

There is no consistent association of SCH, or guideline-driven application of LT4 therapy, with bone health.

Renal function

Prospective follow-up of 480 euthyroid patients and 89 patients with SCH for 26 months revealed adverse renal outcomes in the SCH group, with increased HRs for a composite outcome incorporating doubling of serum creatinine (SCr), end-stage kidney disease (ESKD), or death (1.61 [1.16–2.23]), or doubling of SCr or ESKD (1.53 [1.07–2.20])¹³⁶. SCH also increased the risk of chronic kidney disease (CKD) in people with type 2 diabetes (OR 1.22 [1.09–1.36]), as shown by a case-control study in 3,815 patients¹³⁷. The risk of CKD was higher for TSH ≥3 mIU/L and the relationship between TSH and CKD was linear, with each 1 mIU/L increment of TSH associated with an OR for CKD of 1.09 (1.03–1.16). In another study, SCH was associated with a higher risk of acute kidney injury among 1,593 patients undergoing PCI for ST-elevation MI; changes in serum creatinine were also larger for the SCH group (0.19 mg/dL) vs. controls (0.08 mg/dL; p = 0.04)¹³⁸.

LT4 reduced the rate of albuminuria in a 48-week randomized trial in patients with type 2 diabetes, diabetic nephropathy, and SCH with thyroid autoimmunity²⁸. Elsewhere, a randomized evaluation of LT4 enrolled 140 patients with type 2 diabetes and early diabetic nephropathy, of whom 92 also had SCH. The SCH group were randomized to LT4 vs. placebo, while the euthyroid group served as controls¹³⁹. The rate of albuminuria, serum malondialdehyde (MDA), superoxide dismutase (SOD) activity, and urine 8-hydroxyl deoxyguanosine (8-OHdG) were higher in the SCH group. Albuminuria, MDA and 8-OHdG were no longer significantly different from the control group after 6 months of LT4 treatment, and SOD activity was also reduced. Taken together, these results suggested that LT4 treatment improved renal function (reduced albuminuria) and reduced oxidative stress.

An observational study from the USA used Veterans Health Administration claims data to evaluate the effects of 2 years of LT4 treatment on renal outcomes over a period of 2 years¹⁴⁰. Although there was no difference in eGFR or progression between CKD stages between groups, the LT4-treated group tended to progress less often to higher CKD stages and spent less time in hospital for CKD-related reasons.

Renal dysfunction may represent a complication of hypothyroidism, including SCH. The observation in two randomized trials (supported to some extent by observational data) of improved renal function with LT4 in populations at risk of chronic kidney disease is intriguing and may be a previously underappreciated benefit of LT4 treatment in this population.

Where are we today with intervention for subclinical hypothyroidism?

Reflections on the current evidence base

Recent evidence confirms that subjects with SCH with marginal increases in serum TSH and no thyroid autoimmunity are at low risk of progression to overt hypothyroidism. In addition, new evidence has appeared linking SCH (or TSH values consistent with SCH for most patients) with adverse outcomes considered by the 2013/4 ETA guidelines (Table 2)^4,5, namely dyslipidaemia, obesity, vascular dysfunction (including endothelial dysfunction and atherogenesis), adverse cardiovascular outcomes (in younger subjects, see below), diabetes, heart failure, cancer, adverse pregnancy outcomes, and neuropsychological issues. In addition, evidence has been published associating SCH with novel cardiovascular factors and, importantly, CKD, which are areas considered by European guidelines^4,5.

The mechanism of the effect of SCH (if any) on cardiovascular outcomes remains unclear. Hypothyroidism per se appears to affect classical CV risk factors negatively and one of the observational studies described in Table 3 suggest exacerbation of atherosclerosis (increased CIMT)⁵¹, yet the single recent randomized trial that measured CIMT in a population with SCH found no effect of intervention with LT4 on this parameter⁵⁰. The effects of thyroid homones on the heart are complex and multifactorial and it appears likely that multiple mechanisms may be involved^141,142. One expert review, summarizing the results of the many meta-analyses that have appeared in this field, noted the common finding of associations between SCH and adverse cardiovascular outcomes¹⁴³. This author considered that the adverse effects of SCH on the cardiovascular system may be mediated via other cardiovascular risk factors, and that SCH may act as a “triggering event”, where cardiovascular risk is already high through the presence of multiple other cardiovascular risk factors¹⁴³. Obesity is a well-described CV risk factor: exacerbation of overweight/obesity would be expected to promote more adverse cardiovascular outcomes¹⁴⁴. However, the relationship between adiposity and thyroid homeostasis is bidirectional, and it is not possible to attribute changes in body weight solely or mainly to changes in thyroid hormone status^18,145. A recent study found that a large proportion of subjects with morbid obesity and a diagnosis of SCH had elevations of TSH that were not driven by thyroid autoimmunity¹⁴⁴. This raises the question of whether a modest elevation of TSH alone should support a diagnosis of SCH in people with severe obesity^144–146.

Nevertheless, there remain important limitations in this evidence base. The most pronounced associations between SCH and adverse outcomes remains for those with TSH >10 mIU/L, although there is some newer evidence for harm associated with SCH at lower levels of TSH. Furthermore, definitive evidence confirming a causal relationship between SCH and these important clinical conditions is still lacking due to lack of randomized controlled trials with adequate power to detect a signal of efficacy. In addition, a number of newer studies are meta-analyses, which also include the same, older, evidence used to inform the ETA guidance.

A consistent theme in the evidence summarized above is an apparent lack of effect of thyroid hormone replacement therapy in elderly patients, especially from the primary and secondary analyses from the randomized TRUST study. A recent randomized trial in older patients (>65 years) with SCH with minor symptoms score found that adopting a higher than usual target value for TSH for titrating LT4 therapy (8 mIU/L) did not result in any adverse changes in symptoms or quality of life, or on risk factors for adverse thyroid outcomes (cardiovascular biomarkers and bone health)^130,147,148. A small feasibility trial [The Study of Optimal Replacement of Thyroxine in the Elderly (SORTED)] indicated that lower doses of LT4 replacement in older individuals with hypothyroidism was not associated with any adverse impact on quality of life, cardiovascular or bone health markers¹⁴⁹. Today, taken together with the observation that TSH may increase naturally with age, these findings stress the need for age-appropriate cut-off values for TSH to diagnose SCH in this population, also as described above^16,150. A prospective pooled analysis of data from two clinical trials in patients aged >80 years did not support the use of LT4 in this population, due to a lack of effect on symptoms of hypothyroidism and well being¹⁵¹.

Accordingly, an elderly person with TSH marginally above the all-age cut-off, and little or no disturbance of FT4, may reflect the shifting set point for thyroid hormone homeostasis with age, rather than thyroid dysfunction requiring intervention. Nevertheless, a recent (2019) multinational survey of physicians who manage people with hypothyroidism found that a fifth would treat an older patient with high TSH levels consistent with SCH, but who did not have signs or symptoms of hypothyroidism¹⁵². Data from the USA suggest a similar picture¹⁵³. These findings suggest that further education of some physicians is needed here¹⁵².

At the other extreme, little evidence has emerged concerning the pathophysiology, clinical course or management of SCH in paediatric patients. SCH in children has been described as a relatively benign condition with a high likelihood of remission to euthyroidism, and intervention with LT4 is considered to be required only for paediatric patients with overt hypothyroidism^154,155. Moreover, TSH changes markedly during childhood, and adult TSH reference ranges may not apply to these patients¹⁵⁶.

The very nature of SCH as a disease entity remains controversial⁷. Recent expert opinion considered that intervention for LT4 should not be considered for people with SCH, unless they are ≤30 years old, pregnant or planning pregnancy with TSH >20 mIU/L, or people with severe symptoms, i.e. “almost all adults with SCH would not benefit from treatment with thyroid hormones”⁷. A rebuttal from the Society for Endocrinology and the British Thyroid Association considered that this consensus publication was itself not strongly evidence based, and called for more research on SCH, especially in younger subjects¹⁵⁷.

Overall, the recent research summarized above has deepened the level of evidence relating to SCH and outcomes but does not alter the conclusions of the 2013/4 ETA guidelines for the management of SCH in adults (summarized in Box 1). Recommendations such as focussing the need for LT4 treatment primarily on subjects with TSH >10 mIU/L (and considering LT4 for younger (≤65 years) while avoiding thyroid hormone treatment where possible for elderly patients with TSH <10 mIU/L) seem reasonable today. LT4 remains the mainstay of thyroid hormone replacement, titrated carefully to normalize TSH (preparations with a wide range of tablet strengths may be helpful when starting at very low LT4 doses, such as for patients with cardiac disease¹⁵⁸). There is no evidence for the use of triiodothyronine (T3) in thyroid hormone replacement, at least for now, although appropriately designed trials may establish a role for supplementation with triiodothyronine in selected subgroups in the future (reviewed elsewhere¹⁵⁹).

Box 1 Summary of European Thyroid Association guidance on the use of levothyroxine (LT4) within the management of subclinical hypothyroidism (SCH).

Non-pregnant adults with SCH:

• Consider LT4 for subjects aged ≤65 years with SCH and symptoms of hypothyroidism

• Withdraw LT4 if hypothyroidism symptoms persist for patients with SCH and TSH <10 mIU/L.

• Treat with LT4 irrespective of symptoms if TSH is >10 mIU/L irrespective of the presence or absence of symptoms.

• Administer LT4 to normalise TSH for people with SCH and diffuse or nodular goitre, or with persistent SCH following hemithyroidectomy.

• There is insufficient evidence for use of LT4 to control psychological symptoms or body weight.

• LT4 can improve (but rarely normalises) a dyslipidaemic lipid profile, especially where TSH is >10 mIU/L.

• Start with a low dose (25 µg) of LT4 for patients with serious comorbidity, e.g. cardiac disease.

Older people with SCH:

• Use age-specific TSH reference ranges for older people.

• Individualise the treatment approach.

• Accept a slightly higher TSH target when treating older people.

• Avoid LT4 wherever possible for very elderly (age 80–85 y) with SCH and TSH ≤10 mIU/L.

Pregnancy:

• Use trimester-specific TSH ranges if possible.

• Treat SCH arising before conception or during gestation with LT4, with a TSH target of <2.5 mIU/L.

• Post-partum, reduce the LT4 dose to the preconception dose (if LT4 was used then).

Compiled from information presented in references 4 and 5. Recommendations are paraphrased: always consult the full guidelines. Abbreviations. SCH, subclinical hypothyroidism; LT4, levothyroxine; TSH, thyroid stimulating hormone.

Strengths and limitations of our approach

Our review focussed on recent new research that has occurred since the publication of the European guidelines on the management of SCH. Inevitably, a number of these publications are systematic reviews/meta-analyses. These publications are based on evidence gathered before and after our 2014 cut-off. We do not perceive this as a limitation as this approach presents a more complete picture of the current evidence in this area. Concentrating on recent primary publications alongside these meta-analyses remains a strength of our approach. We describe our findings in the context of the guideline itself, which serves as an expert summary of evidence created before this time. Finally, this appraisal of contemporary literature is not a systematic review (another limitation) and we do not present formal grades of evidence; however, we have been careful to distinguish randomized trials from observation data, which serves this purpose.

Conclusions

Clinical evidence that has appeared during the previous decade generally supports the view that SCH promotes adverse outcomes to an extent that is intermediate between the euthyroid state and overt hypothyroidism. Accordingly, the rationale for the recommendations for intervention in the ETA guidelines published almost ten years ago remains generally sound today. Nevertheless, there remains a need for more evidence, especially from large randomized clinical trials.

Transparency

Declaration of funding

Merck (Crossref Funder ID: 10:13039/100009945) funded editorial assistance (see above) and expedited peer review at CMRO. No other funding applied.

Declaration of financial/other relationships

SR has received speaker fees from Merck plc, Abbott Pharmaceuticals India Ltd and Berlin Chemie plc, makers of levothyroxine. BU is a full-time employee of Merck Healthcare KGaA, Darmstadt, Germany. Peer reviewers on this manuscript have no relevant financial or other relationships to disclose.

Author contributions

Both authors provided material for inclusion in the article, contributed to the development of the article, and approved its submission for publication.

Supporting information

Search strategy: Clinical research data on SCH was identified by a PubMed search for articles with “subclinical hypothyroidism” or “subclinical thyroid dysfunction” in the title, limited to the previous 10 years and article types, “clinical trial”, randomized clinical trial”, “meta-analysis”, “systematic review”, and “observational study”. Reference lists of articles identified in the search, together with reference lists of recent expert reviews, provided additional material. Guidelines on the management of SCH were identified from the websites of leading expert societies or institutions. We did not review the precise diagnostic criteria for SCH from individual trials, as these by definition were conducted after the production of authoritative guidelines for the management of SCH.

Acknowledgements

A medical writer (Dr Mike Gwilt, GT Communications provided editorial assistance in writing of this manuscript, funded by Merck Healthcare KGaA, Darmstadt, Germany.