The impact of the management strategies for patients with subclinical hypothyroidism on long-term clinical outcomes: An umbrella review.

AIM: This umbrella review summarises and compares synthesised evidence on the impact of subclinical hypothyroidism and its management on long-term clinical outcomes.
METHODS: We conducted comprehensive searches on MEDLINE, EMBASE, Scopus, Web of Science, Cochrane Database of Systematic Reviews, JBI Evidence Synthesis, the PROSPERO register, Epistemonikos Database and PDQ Evidence from inception to February and July 2021 using keywords on subclinical hypothyroidism, treatment with levothyroxine, monitoring and primary outcomes (all-cause mortality, cardiovascular events, stroke, frailty fractures and quality of life). Only systematic reviews and meta-analyses on adult patient populations were considered. Study selection, data extraction and quality appraisal using AMSTAR-2 were done independently by two reviewers and discrepancies were resolved through discussion. Overlap across the selected reviews was also assessed, followed by a narrative synthesis of findings.
RESULTS: A total of 763 studies were identified from literature searches; 20 reviews met inclusion criteria. Methodological quality ratings were high (n = 8), moderate (n = 7), and low (n = 5), but no reviews were excluded on this basis. Though there was slight overlap across all reviews, some pairwise comparisons had high corrected covered area scores. Compared to euthyroidism, untreated subclinical hypothyroidism was associated with a higher risk of cardiovascular events or death if Thyroid Stimulating Hormone was above 10mIU/L at baseline. Treatment was associated with a lower risk of death from all causes for patients younger than 70 years and possibly better cognitive and quality of life scores than untreated individuals. Evidence on the risk of strokes and fractures was inconclusive.
CONCLUSION: In the long term, treatment of subclinical hypothyroidism may be beneficial for some patient groups. However, the findings of this review are negatively impacted by the relative sparseness and poor quality of available evidence. Additional large and adequately powered studies are needed to investigate this topic further. SYSTEMATIC REVIEW REGISTRATION: PROSPERO (CRD42021235172).

Introduction

Subclinical hypothyroidism (SCH) is characterised by elevated Thyroid Stimulating Hormone (TSH) levels in contrast to free thyroid hormone–usually thyroxine/T4 –within the reference range [1-3]. The leading cause of SCH is Hashimoto’s thyroiditis, a chronic autoimmune disorder that affects more women than men [2]. Prevalence varies worldwide but has been estimated to be between 4% to 9%, increasing with age to more than 20% for women over 60 years old [2, 4]. This broad range can also be explained by differences in race, dietary intake of iodine and diagnostic cut-offs for SCH [2]. Patients usually exhibit few, if any, symptoms, so it is common for SCH to be detected incidentally from a routine blood test panel [5, 6].

The reference ranges used for measurements of thyroid hormones and TSH vary between laboratories because they are highly dependent on the reference population [7]. Nonetheless, a distinction is sometimes made between mild and severe SCH with a TSH measurement of 10mIU/L as the cut-off [8, 9]. Approximately 60% of cases with mild SCH revert to normal TSH levels over time [3, 9]. Furthermore, depending on the initial severity of their condition, female SCH patients and those that are antithyroid peroxidase antibody-positive are more likely to develop overt hypothyroidism [4]. This progression occurs in around 2–4% of cases per year [2, 3].

Measurement of TSH and thyroid hormone levels is achieved through thyroid function tests (TFTs) which are frequently ordered unnecessarily without medical indications [10-12]. One of the main pitfalls that could likely result from inappropriate TFTs is that more asymptomatic patients are diagnosed as having SCH. Following diagnosis, there are two options for the management of SCH, thyroid replacement therapy with levothyroxine or follow-up without prescribing medication [2, 3]. Even so, a reasonable expectation for the latter is that treatment would be initiated in the event of a patient’s worsening state, provided that the progression can be attributed to SCH. Regardless of the strategy, patients require periodic blood tests to monitor TSH levels for increasing severity or improvement of SCH [1, 3, 13].

The management of SCH is controversial–there has been no definitive evidence on the benefits of replacing thyroid hormones, especially the long-term clinical consequences. This is partly because few adequately powered randomised trials have investigated this topic [4, 14], the most notable being the Thyroid Hormone Replacement for Untreated Older Adults with Subclinical Hypothyroidism (TRUST) trial [15]. The TRUST Study Group reported that in their trial with 737 adults over 65 years old, levothyroxine treatment did not improve patient symptoms nor lower the risk of cardiovascular events and fractures [15]. On the other hand, other smaller trials and observational studies have linked treatment of SCH to improved patient outcomes [2, 4, 8]. Also, current UK clinical guidelines for the management of SCH differ in their recommendations for treatment thresholds and exclusions [14, 16]; these differences can be directly ascribed to inconsistencies in the existing evidence base.

Based on a systematic review by Feller et al. [17], a clinical guideline panel found no evidence to recommend thyroid hormone replacement for SCH patients, except for those with TSH levels above 20mIU/L and women that are pregnant or trying to conceive [14]. For the outcomes they considered, for example, quality of life, cognitive function and cardiovascular events, the panel found no vital difference between treated and untreated groups, irrespective of patient age. Moreover, they noted the issue of practicality regarding medication–patients require long-term treatment and follow-up and even risk developing hyperthyroidism in case of overuse [14]. In contrast, the latest NICE guidance recommendation is for physicians to consider treating adults with TSH of ≥ 10 mIU/L to improve SCH patient outcomes [16]. The reviewing committee found little evidence on SCH treatment but emphasised that additional factors–such as the presence of symptoms–should be considered, over and above TSH levels [13].

Therefore, an umbrella review was performed to collate and compare existing literature on the long-term effects of SCH treatment and follow-up with no medication. Umbrella reviews–also called overviews, meta-reviews or reviews of reviews [18]–are ideally suited to exploring discrepancies in the literature by allowing for a broader scope of inquiry than a typical systematic review [18, 19]. The review questions of interest were: (i) what is the impact of levothyroxine treatment on patient outcomes in subclinical hypothyroidism? and (ii) what is the impact of monitoring without treatment on clinical outcomes for patients with subclinical hypothyroidism? Rather than restrict the focus of this overview to a direct comparison of these management strategies, we sought also to identify what is known for either option.

Methods

To a large extent, umbrella reviews are conducted similarly to typical systematic reviews. However, the critical difference between these methods is that the former use existing systematic reviews and meta-analyses as the units of synthesis [18]. These will subsequently be referred to as ’primary reviews’ in this paper, in contrast to ’primary studies’, the empirical studies included in the systematic reviews. The protocol for this overview was registered on PROSPERO (CRD42021235172) and the methods followed have previously been described in detail [20]; there were no deviations from the registered protocol. The reporting of this overview follows a checklist developed for overviews of systematic reviews based on recommendations from existing guidelines [21].

Search strategy

Comprehensive searches were performed on multiple databases from inception to February 2021, namely MEDLINE, EMBASE, Scopus, Web of Science, Cochrane Database of Systematic Reviews, JBI Evidence Synthesis, the PROSPERO register, Epistemonikos Database and PDQ Evidence. The key search terms used included ’subclinical hypothyroidism’, ’monitoring’, ’treatment’ and ’levothyroxine’, in both free text and subject headings; the search syntax was modified to match the different databases. The searches were updated in July 2021, during the latter stages of data extraction, to identify any systematic reviews and meta-analyses that had been published as the review was in progress. The MEDLINE search strategy is provided in S1 Appendix. No additional date or language filters were applied. Grey literature searches were also performed, and the reference lists of eligible studies were scanned to identify other potentially relevant primary reviews.

Eligibility criteria

The following eligibility criteria were considered to determine inclusion in this umbrella review. Only systematic reviews and meta-analyses on SCH, either in part or as a whole, were considered, irrespective of whether the primary review included randomised trials or observational studies. Generally, all primary reviews had to report on the clinical outcomes of adult patients (>18 years old) with SCH, regardless of the diagnostic thresholds that were initially applied, for example, reference ranges for TSH and thyroid hormones. Patients with overt hypothyroidism were excluded, as were children and pregnant women, whose thyroid hormone requirements differ from the rest of the population.

The two eligible interventions were: (i) treatment with levothyroxine, and (ii) patient follow-up without medication. Another essential requirement was that the treatment status of patients was reported in the systematic reviews, such that it would be possible to distinguish between treated and untreated groups. There were no additional restrictions on study comparators and settings.

The primary outcomes of interest were all-cause mortality, defined as the death of patients with SCH, irrespective of the cause, at least 12 months from baseline or the start of follow-up; cardio- and cerebrovascular outcomes such as heart failure, arrhythmias, stroke, peripheral vascular disease, coronary heart disease; quality of life as measured using suitable instruments (or otherwise described as ’symptoms’ particularly in older publications); and, frailty fractures, defined as fractures resulting from low-impact trauma, usually due to pre-existing disease. Other long-term clinical outcomes reported in the included systematic reviews, for example, cognitive function, were considered secondary outcomes.

Study selection

References retrieved from the searches were imported into Covidence (www.covidence.org/) and initially screened in duplicate for eligibility by title and abstract. After that, the full texts of selected primary reviews were obtained and read independently by pairs of reviewers who assessed each paper against the selection criteria. When needed, primary review authors were contacted to provide additional information.

Data extraction

A data extraction form was developed on Covidence and piloted by two reviewers. This form was used to extract information on citation details, primary study selection criteria, search parameters, selection and quality assessment methods and primary review findings relating to the outcomes of interest for this umbrella review. Where provided, effect estimates were extracted alongside their 95% confidence intervals and the treatment status of the assessed group(s). This process was done by two reviewers working independently, and discrepancies in the extracted data were resolved through discussion to reach a consensus.

Quality appraisal

The quality of the selected systematic reviews was independently assessed by two reviewers using the Assessment of Multiple Systematic Reviews (AMSTAR-2) tool [22], an instrument with 16 questions on the methodological quality of systematic reviews. These questions include whether a comprehensive literature search was conducted, justification for excluding studies, and risk of bias assessments for included studies (S2 Appendix). Overall ratings and their meanings are shown in Table 1. AMSTAR-2 was chosen over the risk of bias in systematic reviews (ROBIS) tool [23] because while both assess strongly related aspects (methodological quality vs risk of bias), the former is advantageous for inter-rater reliability and usability [24-26]. Disagreements between the reviewers were similarly resolved through discussion.

Table 1

AMSTAR-2 overall confidence ratings (from Shea et al. [22]).

Rating	Interpretation
High	≤1 non-critical weakness
Moderate	>1 non-critical weakness
Low	1 critical flaw +/- non-critical weaknesses
Critically low	>1 critical flaw +/- non-critical weaknesses

It has been suggested that GRADE criteria can be applied to systematic reviews [27]. However, this approach was initially designed for empirical studies, hence the paucity of relevant guidance on how best to achieve this [18, 28]. Therefore, we did not perform any secondary GRADE assessments on the included primary reviews but extracted any reported quality ratings.

Assessing overlap

One of the unique challenges in conducting an umbrella review is overlap–the inclusion of the same primary study or trial in more than one selected systematic review or meta-analysis [29, 30]. Any subsequent synthesis of more than one of these primary reviews would result in ’double-counting’ and biased findings because the contribution of a subset of the data would have been multiplied by some factor [30, 31]. The proposed methods for dealing with overlap are: (i) selecting only the most recent systematic review or the one with the largest number of studies, (ii) selecting only the primary review of the highest quality, or (iii) including all primary reviews but evaluating the amount of overlap [18, 30, 31].

To assess overlap between the included primary reviews, we calculated the corrected covered area (CCA) using the formula described by Pieper et al. [29]: where N–number of included primary studies in selected reviews

r–number of index publications

c–number of included primary reviews

A matrix of the included systematic reviews and their primary studies was created to identify the numerators and denominators shown above. CCA is interpreted in banded thresholds: 5% or less indicates slight overlap, 6% to 10% shows moderate overlap, 11% to 15% for high overlap and values greater than 15% indicate very high overlap [29].

Synthesis of results

A narrative synthesis of results was performed due to high levels of study overlap and considerable heterogeneity in primary review inclusion criteria and reported outcomes. Summaries of the included primary reviews are presented below in tabular form alongside corresponding effect estimates such as odds and hazard ratios (where reported). Of note, no further re-analysis of empirical study data was performed, as previously stated in the umbrella review protocol [20]. The extracted data were grouped according to the clinical outcome of interest, regardless of the degree of overlap among sets of studies. Moreover, because there is currently no agreed-upon solution for the issue of low-quality systematic reviews in overviews [18, 26], all selected papers were included in the narrative synthesis.

Results

In total, 763 records were retrieved from the initial and updated searches. After screening by title, abstract and full-text, 20 syntheses were selected for inclusion in this umbrella review. Notably, two otherwise eligible primary reviews were excluded based on all their included studies having been used in later publications by the same authors [32, 33]. Authors were unable to provide further information for two other publications [34, 35]. One item of grey literature, a systematic evidence review commissioned by a government agency for healthcare research, was included [36]. A list of the systematic reviews that were excluded after reading full texts is provided in S3 Appendix. The PRISMA flowchart [37] showing the stages of study selection is presented in Fig 1.

Fig 1

PRISMA flowchart showing study selection (adapted from Page et al. [37]).

The characteristics of the selected primary reviews are shown in Table 2. Of the 20 included syntheses, four were systematic reviews [36, 38–40], five were published as combined systematic reviews and meta-analyses [17, 41–44], six were labelled meta-analyses [45-50], and five were individual participant data analyses [51-55]. The majority were published earlier than the TRUST trial [15], with only seven primary reviews published during or after 2017 [17, 38, 41–43, 50, 51]. Most of the primary reviews synthesised observational data, although three papers only included RCTs [17, 36, 40]. Generally, SCH was defined using similar TSH thresholds (> 4.5 mIU/L) and normal T4, but some studies subdivided this further into degrees of SCH or TSH elevation e.g., mild vs moderate (Table 2).

Table 2

Characteristics of the included systematic reviews and meta-analyses.

Study	Design	Study aim	Included studies	Definition of SCH	SCH patients (%)	Summary of findings	AMSTAR-2 Overall confidence
Baumgartner et al. (2017) [51]	IPD	To examine the risk of AF in individuals with thyroid function within the normal range and SCH	11 cohort studies (IPD)	TSH level between 4.5 and 19.9 mIU/L with fT4 levels in the reference range	1958 (6.5)	The reviewers found no link between SCH and the risk of AF; this was the same for individuals with TSH levels within the normal range.	High
Blum et al. (2015) [45]	MA	To assess the association of subclinical thyroid dysfunction with fractures	13 cohort studies	TSH level of 4.50 to 19.99 mIU/L with normal FT4 levels	4092 (5.8)	There was no observed association between SCH and fracture risk.	Moderate
Chaker et al. (2015) [52]	IPD	To evaluate the association between SCH and stroke	17 cohort studies	TSH levels of 4.5 to 19.9 mIU/L with normal T4 levels	3451 (7.3)	There was no overall increase in the risk of stroke events and fatal stroke in patients with SCH than euthyroid patients, except for patients younger than 65 years.	Moderate
Collet et al. (2014) [53]	IPD	To compare the risks of CHD mortality and events associated with SCH by thyroid antibody status	6 cohort studies	TSH 4.5 to 19.9 mIU/L and normal T4 level	1691 (4.4)	Thyroid antibodies were found to have no effect on CHD events and mortality though SCH patients with higher TSH levels were generally at higher risk of developing these outcomes.	Moderate
Dhital et al. (2017) [41]	SR + MA	To look at the association between thyroid function profile and outcomes after acute ischemic stroke	12 cohort studies	Elevated TSH and normal fT4 (study-specific cut-offs)	Unclear	SCH was associated with better functional outcomes after acute ischemic stroke, but this depended on the initial levels of free T3.	Low
Feller et al. (2018) [17]	SR + MA	To examine the association of THT with quality of life and thyroid-related symptoms in adults with SCH	21 RCTS	Thyrotropin and free thyroxine levels above and within centre-specific reference ranges, respectively	2192 (100)	There was no association between treatment of SCH and improving thyroid-related symptoms and quality of life (primary outcomes) or cognitive function, depressive symptoms and the other secondary outcomes.	High
Gencer et al. (2012) [54]	IPD	To clarify the association between subclinical thyroid dysfunction and HF events	6 cohort studies	TSH level of 4.5 to 19.9 mIU/L with normal FT4 levels	2068 (8.1)	Patients with TSH levels higher than 10mIU/L faced a significantly higher risk of HF events.	Moderate
Helfand (2004) [36]	SR	To evaluate the benefits of screening for subclinical thyroid dysfunction	8 RCTs	Elevated TSH and normal T4	Unclear	Evidence of an association between treatment and reduced symptoms was demonstrated only for SCH patients with TSH >10 mIU/L and those with a history of Graves’ disease.	Low
Peng et al. (2021) [42]	SR + MA	To investigate whether THT is associated with decreased mortality in adults with SCH	2 RCTs and 5 cohort studies	Grade 1 (TSH level 5.0–10 mIU/L); Grade 2 (TSH level >10 mIU/L) with free thyroxine level within the reference range	21055*3 (100)	Treatment was found to benefit SCH patients younger than 65 years; all-cause mortality decreased by 50%, and cardiovascular mortality decreased by 46%. However, the same did not apply to patients older than 65 years. There was also no overall benefit of treatment on mortality.	High
Razvi et al. (2008) [46]	MA	To examine the influence of age and gender on IHD and mortality in SCH	15 cohort studies	Mild SCH—TSH levels < 10 mIU/L	2,531 (8.7)	The overall incidence of IHD and mortality was not significantly higher for patients with SCH, but IHD prevalence was found to be significantly elevated for patients younger than 65 years.	High
Reyes Domingo et al. (2019) [38]	SR	To synthesize the evidence on the effects of screening and subsequent treatment for thyroid dysfunction	5 RCTs and 3 cohort studies*2	Study-specific	Unclear	Evidence was found linking treatment for SCH with reduced all-cause mortality for patients younger than 65 years, but it was determined to be of low quality.	High
Rodondi et al. (2006) [47]	MA	To determine whether SCH is associated with an increased risk for CHD	5 cohort, 6 cross-sectional and 3 case-control studies	Elevated TSH and a normal T4 (no pre-specified cut-offs)	1409 (10.8)	Compared to euthyroid patients, CHD was 1.6 times more likely in patients with SCH; this association was constant throughout the included studies but less pronounced in the prospective cohorts.	High
Rodondi et al. (2010) [55]	IPD	To assess the risks of CHD and total mortality for adults with SCH	11 cohort studies	Serum TSH level of 4.5 mIU/L or greater to less than 20 mIU/L, with a normal T4 concentration	3450 (6.2)	SCH patients with TSH levels higher than 10mIU/L had a significantly higher risk of CHD events and mortality than euthyroid patients.	High
Rugge et al. (2015) [39]	SR	To assess the benefits and harms of screening and treatment of subclinical and undiagnosed overt hypothyroidism and hyperthyroidism in adults*	13 RCTs and 1 cohort study	4.5–10.0 mIU/L (mildly elevated) or ≥10 mIU/L (markedly elevated) TSH levels with normal thyroxine	Unclear	Reviewers found a potential association between SCH and cardiovascular disease but inconclusive evidence that treatment would be beneficial; SCH treatment was also not associated with improved cognitive function or quality of life.	Moderate
Singh et al. (2008) [48]	MA	To compare the relative risk for incident CHD events, cardiovascular-related and total mortality associated with subclinical thyroid abnormalities	6 cohort studies	Serum TSH above 4.0–5.0 mIU/L with normal free T4 (range 0.7–1.8 ng/dL)	1365 (10.2)	SCH was linked to a significant risk of CHD at baseline and both CHD and cardiovascular mortality during follow-up. On the other hand, all-cause mortality was not found to be increased with SCH.	Low
Sun et al. (2017) [43]	SR + MA	To explore the relationship between subclinical thyroid dysfunction and the risk of cardiovascular outcomes	16 cohort studies	TSH levels >3.6 to 6 mIU/L (study-specific)	5178 (7.2)	There was a significantly higher risk of CHD and cardiovascular mortality for SCH patients younger than 65 years, but the same effect was not observed for patients older than 80 years. A slightly higher risk of AF and HF was also associated with SCH.	Moderate
Villar et al. (2007) [40]	SR	To assess the effects of thyroid hormone replacement for SCH	12 RCTs	TSH level above the upper limit of the reference range with normal values of total T4 or free T4 (FT4), with or without T3 or free T3 (FT3) measurements	350 (100)	It was not possible to assess the benefits of SCH treatment on reducing cardiovascular mortality. However, there was also no significant impact of levothyroxine on health-related quality of life and symptoms.	High
Wirth et al. (2014) [44]	SR + MA	To assess the risk for hip and non-spine fractures associated with subclinical thyroid dysfunction	7 cohort studies	TSH level greater than 4.5 to 20.0 mIU/L and an FT4 level in the reference range	Unclear	No association between SCH and fracture risk was found, but the reviewers could not assess the effects of treatment vs no treatment due to insufficient data.	Moderate
Yan et al. (2016) [49]	MA	To identify the relationship between subclinical thyroid dysfunction and the risk of fracture	5 cohort studies	TSH level greater than 4.0 to 5.5 mIU/L (study-specific)	2580 (0.9)	A link between SCH and higher fracture risk was not found, but the reviewers acknowledge that they had limited data.	Low
Yang et al. (2019) [50]	MA	To assess the association between subclinical thyroid dysfunction and the clinical outcomes of HF patients	14 cohort studies	Elevated TSH values in the presence of normal FT4 values	2308 (10.9)	Both adjusted and unadjusted analyses showed a significantly higher risk of all-cause mortality and cardiovascular death associated with SCH for patients with heart failure.	Low

THT—Thyroid Hormone Therapy; SR–Systematic Review; SR + MA–Systematic Review and Meta-Analysis; MA–Meta-analysis; IPD–Individual Participant Data analysis; SCH–Subclinical Hypothyroidism; CHD–Coronary Heart Disease; AF–Atrial Fibrillation; RCT–Randomised Controlled Trial; HF–Heart Failure; IHD–Ischaemic Heart Disease; Thyroxine–T4, fT4, thyroid hormone; Thyrotropin–Thyroid Stimulating Hormone (TSH)

*this was an update to Helfand et al. [36], but because the searches did not overlap, this was considered a separate review.

*2only for the relevant research question on clinical outcomes for SCH.

*3the authors report potential overlap between the studies; hence the estimate may be incorrect.

Coverage of the primary outcomes was good, given that all the outcomes of interest were reported in at least two publications. However, it is crucial to note that most relevant results were obtained via subgroup or sensitivity analyses in the primary reviews. As such, they were not necessarily representative of the overall findings shown in Table 2.

All-cause mortality

Seven publications reported findings on all-cause mortality; of these, three primary reviews compared rates between treated and untreated patients [38, 39, 42], three compared untreated and euthyroid individuals [43, 46, 48], and one compared both treated and untreated SCH groups with euthyroid participants [50]. There was no statistically significant difference in the overall numbers of deaths from all causes for patients with SCH between those who were and were not on treatment (Table 3).

Table 3

Review findings on all-cause mortality.

Study	Outcome	Treatment status	Comparator	Effect estimate (95% CI)
Peng et al. (2021) [42]	All-cause mortality	Treated	Untreated	RR 0.95 (0.75–1.22)
	All-cause mortality; age <65–70 years			RR 0.50 (0.29–0.85)
	All-cause mortality; age > = 65–70 years			RR 1.08 (0.91–1.28)
Reyes Domingo et al. (2019) [38]	All-cause mortality; adults (>18 years)	Treated	Untreated	HR 1.91 (0.65–5.60)
	All-cause mortality; adults (<65 or <70 years)			IRR 0.63 (0.40–0.99
				HR 0.36 (0.19–0.66)
	All-cause mortality; adults (>65 years)			HR 1.91 (0.65–5.60)
	All-cause mortality; females			IRR 0.99 (0.85–1.16)
				1.08 (0.80–1.48)
	All-cause mortality; males			IRR 1.24 (0.89–1.16)
				1.43 (0.87–2.34)
Rugge et al. (2015) [39]	All-cause mortality; 40–70 years	Treated	Untreated	HR 0.36 (0.19–0.66)
	All-cause mortality; >70 years			HR 0.71 (0.56–1.08)
Yang et al. (2019) [50]	All-cause mortality	Untreated	Euthyroid	HR 1.48 (1.29–1.70)
		Treated	Euthyroid	HR 1.48 (1.14–1.94)
Razvi et al. (2008) [46]	IHD/all-cause mortality; <65 years	Untreated	Euthyroid	OR 1.32 (0.95–1.83)
	IHD/all-cause mortality; > 65 years			OR 0.87 (0.51–1.45)
Sun et al. (2017) [43]	Total mortality	Untreated	Euthyroid	RR 1.01 (0.90–1.15)
Singh et al. (2008) [48]	All-cause mortality	Untreated	Euthyroid	RR 1.115 (0.990–1.255)

HR–Hazard Ratio; RR–Relative Risk; IRR–Incident rate Ratio; IHD–Ischemic Heart Disease; OR–Odds Ratio.

Taking age into account, lower estimates of all-cause mortality were reported for patients younger than 70 years on treatment (RR 0.50, 95% CI 0.29 to 0.85 [42]; HR 0.36, 95% CI 0.19 to 0.66 [38, 39]). However, these estimates were based on one study in Rugge et al. [39], and another paper rated the same evidence as being of very low certainty [38]. On the other hand, older patient groups demonstrated no significant association between levothyroxine treatment and all-cause mortality (Table 3).

Only one of four comparisons of all-cause mortality between untreated and euthyroid study participants was statistically significant (HR 1.48, 95% CI 1.29 to 1.70) [50]. The same review found that death was more likely among SCH patients on treatment than in euthyroid persons (HR 1.48, 95% CI 1.14 to 1.94) [50]. However, the population of interest for this review all had heart failure, thereby limiting the generalisability of these findings to other SCH patients.

Cardiovascular outcomes

Cardiovascular outcomes were the most extensively reported outcomes of interest across the included reviews (n = 13) as shown in Table 4. No difference was found in the number of incident atrial fibrillation events between untreated persons and euthyroid controls, irrespective of age and TSH level [51]. Similarly, the difference between treated and untreated SCH patients was not statistically significant, though notably, the evidence was rated as being of very low to moderate certainty [38].

Table 4

Reported primary review findings on cardiovascular outcomes.

Study	Outcome	Treatment status	Comparator	Effect estimates (95% CI)
Baumgartner et al. (2017) [51]	Atrial fibrillation	Untreated	Euthyroid (TSH 3.50–4.49 mIU/L)	For TSH 4.5–6.9 mIU/L: HR 0.87 (0.66–1.16)
				For TSH 7.0–9.9 mIU/L: HR 1.22 (0.78–1.92)
				For TSH 10.0–19.9 mIU/L: HR 1.56 (0.84–2.90)
Reyes Domingo et al. (2019) [38]	Atrial fibrillation; adults (>18y)	Treated	Untreated	HR 0.80 (0.35–1.80)
	Atrial fibrillation; adults (<65 or <70)			HR 0.76 (0.26–1.73)
	Atrial fibrillation; adults (>65y)			HR 0.80 (0.35–1.80)
Collet et al. (2014) [53]	CHD events	Untreated	Euthyroid	SH With -ve TPOAb HR 1.25 (1.06–1.47)
				SH With +ve TPOAb HR 1.12 (0.88–1.41)
				SH with TSH ≥10.0 mIU/L and neg. TPOAb HR 3.76 (1.77–8.01)
				SH with TSH ≥10.0 mIU/L and pos. TPOAb HR 1.19 (0.61–2.32)
Rodondi et al. (2006) [47]	CHD	Untreated	Euthyroid	OR 2.06 (1.36–3.14)
Rodondi et al. (2010) [55]	CHD	Untreated	Euthyroid	For TSH 4.5–19.99 mIU/L: HR 1.17 (0.91–1.50)
				For TSH 10–19.99 mIU/L: HR 2.17 (1.19–3.93)
Sun et al. (2017) [43]	CHD	Untreated	Euthyroid	RR 1.02 (0.92–1.14)
Singh et al. (2008) [48]	CHD (during follow-up)	Untreated	Euthyroid	RR 1.188 (1.024–1.379)
Gencer et al. (2012) [54]	Heart failure events; TSH 4.5–19.9 mIU/L	Untreated	Euthyroid	HR 1.26 (0.93–1.69)
	Heart failure events; TSH 10.0–19.9 mIU/L			HR 2.37 (1.59–3.54)
Razvi et al. (2008) [46]	IHD incidence; < 65 yrs	Untreated	Euthyroid	OR 1.58 (1.07–2.35)
	IHD incidence; > 65 yrs			N/P
Rugge et al. (2015) [39]	IHD; 40–70 yrs	Treated	Untreated	HR 0.61 (0.39–0.95)
	IHD; >70 yrs			HR 0.99 (0.59–1.33)
Reyes Domingo et al. (2019) [38]	Fatal and non-fatal cardiovascular events (not AF); adults (>18y)	Treated	Untreated	HR 0.89 (0.47–1.69)
	Fatal and non-fatal cardiovascular events (not AF); adults (<65 or <70)			HR 0.61 (0.39–0.95)
				HR 1.03 (0.51–2.13)
				IRR 1.11 (0.61–2.02)
	Fatal and non-fatal cardiovascular events (not AF); adults (>65y)			HR 0.89 (0.47–1.69)
	Fatal and non-fatal cardiovascular events (not AF); females			IRR 0.99 (0.70–1.38)
				0.99 (0.70–1.40)
	Fatal and non-fatal cardiovascular events (not AF); males			IRR 1.41 (0.83–2.40)
				1.36 (0.79–2.35)

CHD–Coronary heart disease; HR–Hazard Ratio; RR–Relative Risk; IRR–Incident rate Ratio; IHD–Ischemic Heart Disease; OR–Odds Ratio.

Compared to euthyroidism, untreated SCH was significantly associated with a higher likelihood of CHD and heart failure if patients had TSH levels above 10mIU/L (HR 2.17, 95% CI 1.19 to 3.93) [55], (HR 2.37, 95% CI 1.59 to 3.54) [54] or were thyroid peroxidase antibody-negative (HR 1.25, 95% CI 1.06 to 1.47); HR (3.76, 95% CI 1.77 to 8.01) [53]. It was also reported that untreated SCH was associated with higher odds of ischaemic heart disease (OR 1.58, 95% CI 1.07 to 2.35) [46] and a higher risk of developing coronary heart disease during follow-up (RR 1.188, 95% CI 1.024 to 1.379) [48] than euthyroid participants. However, one primary review found that incident CHD was not associated with untreated SCH [43]–this difference may have resulted from the reviewers’ decision to restrict the inclusion of primary studies based on quality appraisal scores.

SCH patients receiving treatment and younger than 70 years were significantly less likely to develop IHD than untreated individuals (HR 0.61, 95% CI 0.39 to 0.95) [38]. However, these findings were based on a single empirical study in both primary reviews, in which it was reported that the GRADE rating for this evidence was very low. A similar association was not found for patients older than 70 years nor subgroups based on sex [38].

We distinguished between all-cause mortality and cardiovascular mortality (n = 8), which primary review authors defined as deaths arising from cardiovascular diseases (Table 5). Considering treated vs untreated SCH, an association between treatment and cardiovascular death was found only for adult patients younger than 65–70 years [38, 39, 42]. Stratifying the results by sex did not yield statistically significant findings. In addition, whereas the risk of cardiovascular mortality was found to be higher for untreated SCH patients compared to euthyroid controls in four primary reviews [48, 50, 53, 55], Sun et al. [43] reported a lower nonsignificant estimate (RR 0.86, 95% CI 0.56 to 1.32) [43]. A possible reason for this difference, despite very high overlap between pairs of these primary reviews, could be discrepancies in the determination of treatment status. Furthermore, Sun et al. [43] rated the quality of evidence for cardiovascular mortality in their primary review as low because of high heterogeneity.

Table 5

Reported primary review findings on cardiovascular mortality.

Study	Outcome	Treatment status	Comparator	Effect estimates (95% CI)
Peng et al. (2021) [42]	Cardiovascular mortality	Treated	Untreated	RR 0.99 (0.82–1.20)
	Cardiovascular mortality; age <65–70 years			RR 0.54 (0.37–0.80)
	Cardiovascular mortality; age > = 65–70 years			RR 1.05 (0.87–1.27)
Reyes Domingo et al. (2019) [38]	Cardiovascular deaths; adults (>18y)	Treated	Untreated	OR 2.01 (0.18–22.27)
	Cardiovascular deaths; adults (<65 or <70)			HR 0.54 (0.37–0.92) IRR 0.55 (0.25–1.20)
	Cardiovascular deaths; adults (>65y)			OR 2.01 (0.18–22.27)
	Cardiovascular deaths; females			IRR 0.96 (0.77–1.21)
	Cardiovascular deaths; males			IRR 1.32 (0.83–2.08)
Rugge et al. (2015) [39]	Cardiovascular deaths (40–70 years)	Treated	Untreated	HR 0.54 (0.37–0.92)
Collet et al. (2014) [53]	CHD mortality	Untreated	Euthyroid	SH With -ve TPOAb HR 1.34 (1.07–1.69)
				SH With +ve TPOAb HR 1.28 (0.94–1.72)
				SH with TSH ≥10.0 mIU/L and negative TPOAb HR 1.95 (0.76–4.98)
				SH with ≥10.0 mIU/L and positive TPOAb HR 1.92 (1.09–3.36)
Rodondi et al. (2010) [55]	CHD mortality	Untreated	Euthyroid	For TSH 4.5–19.99 mIU/L: HR 1.25 (1.04–1.51)
				For TSH 10–19.99 mIU/L: HR 1.85 (1.13–3.05)
Sun et al. (2017) [43]	Cardiovascular mortality	Untreated	Euthyroid	RR 0.86 (0.56–1.32)
Singh et al. (2008) [48]	Cardiovascular mortality	Untreated	Euthyroid	RR 1.278 (1.023–1.597)
Yang et al. (2019) [50]	Cardiac death and/or hospitalization	Untreated	Euthyroid	HR 1.32 (1.08–1.60)
		Treated		HR 1.36 (1.12–1.66)

CHD–Coronary Heart Disease; HR–Hazard Ratio; RR–Relative Risk; IRR–Incident rate Ratio; OR–Odds Ratio.

Only one primary review considered the relationship between thyroid peroxidase antibody status and cardiovascular mortality. Collet et al. [53] found that untreated thyroid antibody-negative SCH was significantly associated with cardiovascular mortality (HR 1.34 95% CI 1.07 to 1.69), but the same did not apply for antibody-positive SCH (HR 1.28, 95% CI 0.94 to 1.72) [53], except for patients that also had TSH levels above 10mIU/L (HR 1.92, 95% CI 1.09 to 3.36) [53]. Finally, compared to euthyroid controls, death and hospitalisation due to cardiovascular causes were more likely to occur among treated SCH patients with heart failure [50].

Stroke

There were no direct comparisons of the risk of stroke between treated and untreated patients with SCH. One primary review compared untreated individuals with SCH with euthyroid controls, and no significant difference was found in either the incidence of strokes or deaths arising from strokes [52]. Dhital et al. [41] found that functional outcomes for untreated SCH (based on the modified Rankin scale) were twice as likely to be better than those for euthyroid controls 1 and 3 months after acute ischemic stroke (Table 6).

Table 6

Primary review findings on stroke.

Study	Outcome	Treatment status	Comparator	Effect estimates (95% CI)
Chaker et al. (2015) [52]	Stroke events	Untreated	Euthyroid	HR 0.96 (0.70–1.31)
	Fatal stroke			HR 1.27 (0.74–2.16)
Dhital et al. (2017) [41]	Stroke–modified Rankin scale	Untreated	Euthyroid	OR after 1 month 2.58 [1.13–5.91]
				OR after 3 months 2.28 [1.33–3.91]
	Stroke–mortality after 3 months			OR 0.20, (0.04–1.12)

HR–Hazard Ratio; OR–Odds Ratio.

Fractures

No primary reviews that compared the risk or likelihood of fractures between euthyroid individuals and SCH patients that did or did not receive treatment found a significant difference (Table 7). This result was similar across various types of fractures, for example, hip fractures (HR 1.02, 95% CI 0.87 to 1.19 [45]; HR 1.10, 95% CI 0.81 to 1.50 [44]) and spine fracture (HR 1.16, 95% CI 0.66 to 2.04) [45].

Table 7

Primary review findings on fractures.

Study	Outcome	Treatment status	Comparator	Effect estimates (95% CI)
Blum et al. (2015) [45]	Hip fracture	Untreated	Euthyroid	HR 1.02 (0.87–1.19)
	Any fracture			HR 1.11 (0.94–1.30)
	Non-spine fracture			HR 1.13 (0.93–1.38)
	Spine fracture			HR 1.16 (0.66–2.04)
Reyes Domingo et al. (2019) [38]	Fractures; adults (all >65)	Treated	Untreated	HR 1.06 (0.41–2.76)
Yan et al. (2016) [49]	Fractures (any)	Untreated	Euthyroid	RR 1.25 (0.85–1.84)
		Treated		RR 1.22 (0.61–2.47)
Wirth et al. (2014) [44]	Hip fractures	Untreated	Euthyroid	HR 1.10 (0.81–1.50)
	Non-spine fractures			HR 1.11 (0.60–2.05)

HR–Hazard Ratio; RR–Relative Risk.

Quality of life and presence of symptoms

For the 5 studies that explicitly reported quality of life outcomes, no statistically significant differences were found between patients who did and did not receive treatment for SCH (Table 8). Similarly, thyroid-related symptoms, fatigue, mental and general well-being scores were not significantly associated with treatment status. However, Helfand [36] reported that specific subgroups–patients with TSH values greater than 10 mIU/L and those with a history of Graves’ disease seemed to benefit from treatment. Graves’ disease is an autoimmune thyroid disorder treated with antithyroid medication, radiotherapy or surgery [56]. Nonetheless, it is noted that the single study that this finding was based upon was a small trial of 33 participants, all of whom had previously treated Graves’ disease [36].

Table 8

Primary review findings on quality of life and symptoms.

Study	Outcome	Treatment status	Comparator	Effect estimates (95% CI)
Feller et al. (2018) [17]	General QoL	Treated	Untreated	SMD -0.11 (-0.25–0.03)
Reyes Domingo et al. (2019) [38]	Thyroid QoL—less than 12 mo	Treated	Untreated	MD 0.0 (-2.0–2.1)
	Thyroid QoL—more than 12 mo			MD 1.0 (-1.9–3.9)
				-0.5 (-2.2–1.3)
Rugge et al. (2015) [39]	Quality of life	Treated	Untreated	Multiple
Rugge et al. (2015) [39]	Thyroid-related symptoms	Treated	Untreated	SMD 0.01 (-0.12–0.14)
	Fatigue and tiredness			SMD -0.01 (-0.16–0.15)
	Depressive symptoms			SMD -0.10 (-0.34–0.13)
Helfand (2004) [36]	Symptoms	Treated	Untreated	Multiple
Villar et al. (2007) [40]	Symptoms, mood and quality of life	Treated	Untreated	Multiple
Reyes Domingo et al. (2019) [38]	Fatigue/tiredness—less than 12 mo	Treated	Untreated	MD 0.4 (-2.1–2.9)
	Fatigue/tiredness—more than 12 mo			MD -3.5 (-7.0–0.0)
	Mental well-being			Multiple
	Physical well-being			MD -0.1 (-0.3–1.0)
				-0.1 (-0.3–1.0)
	General well-being			Multiple

SMD–Standardised Mean difference; MD–Mean Difference.

Secondary outcomes

Some of the included papers (n = 4) reported on cognitive function (Table 9), which was assessed using various tools such as the Letter-Digit Coding Test and Mini-Mental State Examination. All of the primary reviews found no significant difference in cognitive function between treated and untreated groups [17, 38, 39] except Villar et al. [40]. However, this result was based on only one included study with an unclear risk of bias assessment.

Table 9

Primary review findings on cognitive function.

Study	Outcome	Treatment status	Comparator	Effect estimates (95% CI)
Feller et al. (2018) [17]	Cognitive function	Treated	Untreated	Difference 1.01 (95% CI −0.56 to 2.46)
Reyes Domingo et al. (2019) [38]	Cognitive function	Treated	Untreated	Multiple (no difference)
Villar et al. (2007) [40]	Cognitive function	Treated	Untreated	MD 2.4 (0.3–4.5)
Rugge et al. (2015) [39]	Cognitive function	Treated	Untreated	Multiple (no difference)

MD–Mean Difference.

Overlap

The extent of overlap in this umbrella review is shown in Fig 2, an intersection heatmap of the calculated CCA between pairs of the 20 included primary reviews. As shown, only one of the included evidence syntheses [41] had a unique set of primary publications compared to all the other primary reviews. Overall, excluding the diagonal, the pairwise comparisons showed slight (66.7%), moderate (10%), high (2.6%) and very high (20.8%) overlap. However, it should be noted that these values were obtained with no consideration of the specific outcomes reported in each of the primary reviews and therefore require cautious interpretation. That is because, although calculating CCA involved mapping all the primary studies included in each publication, not all provided findings relevant to this umbrella review. The fundamental reason for this complexity is that we were only interested in estimates reported with the participants’ corresponding treatment status. On balance, CCA for the entire overview was calculated to be 5.12%, the higher limit for slight overlap [29].

Fig 2

Heatmap showing pairwise calculated CCA.

Overall confidence in review findings was found to be high for eight primary reviews [17, 38, 40, 42, 46, 47, 51, 55], moderate for seven primary reviews [39, 43–45, 52–54] and low for five primary reviews [36, 41, 48–50]. Three syntheses did not include a meta-analysis and therefore could not be assessed for questions 11, 12 and 15 [36, 38, 39]. The breakdown of checklist questions is shown in 0 10.

Table 10

Results of the AMSTAR-2 assessments.

Review	Question																Overall confidence in results
	1	2	3	4	5	6	7	8	9	10	11	12	13	14	15	16
Baumgartner et al. (2017) [51]	Y	Y	Y	Y	Y	Y	Y	Y	Y	N	Y	Y	Y	Y	Y	Y	High
Blum et al. (2015) [45]	Y	N	N	Y	Y	N	N	Y	Y	Y	Y	Y	Y	Y	Y	Y	Moderate
Chaker et al. (2015) [52]	Y	N	Y	PY	Y	Y	Y	Y	Y	N	Y	Y	N	Y	Y	Y	Moderate
Collet et al. (2014) [53]	Y	N	N	PY	Y	N	N	PY	N	N	Y	N	N	N	Y	Y	Moderate
Dhital et al. (2017) [41]	Y	N	N	PY	Y	Y	N	Y	PY	N	Y	Y	Y	N	N	Y	Low
Feller et al. (2018) [17]	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	N	Y	High
Gencer et al. (2012) [54]	Y	PY	Y	PY	Y	N	N	Y	Y	Y	Y	Y	Y	Y	Y	Y	Moderate
Helfand (2004) [36]	Y	PY	Y	Y	N	N	N	Y	PY	N	NMA	NMA	N	N	NMA	N	Low
Peng et al. (2021) [42]	Y	Y	N	PY	Y	Y	N	Y	Y	Y	Y	Y	Y	Y	Y	Y	High
Razvi et al. (2008) [46]	Y	N	Y	PY	Y	Y	PY	Y	Y	N	Y	Y	Y	Y	Y	Y	High
Reyes Domingo et al. (2019) [38]	Y	Y	Y	Y	Y	Y	Y	Y	Y	N	NMA	NMA	Y	Y	NMA	Y	High
Rodondi et al. (2006) [47]	Y	N	Y	N	Y	Y	Y	N	Y	N	Y	Y	Y	Y	Y	Y	High
Rodondi et al. (2010) [55]	Y	PY	Y	PY	Y	N	N	Y	Y	Y	Y	Y	Y	Y	Y	Y	High
Rugge et al. (2015) [39]	N	PY	Y	Y	Y	Y	N	N	PY	N	NMA	NMA	Y	N	NMA	Y	Moderate
Singh et al. (2008) [48]	Y	N	Y	PY	N	N	N	Y	N	N	Y	N	N	N	N	N	Low
Sun et al. (2017) [43]	Y	N	Y	PY	Y	Y	N	PY	Y	N	Y	Y	Y	Y	N	Y	Moderate
Villar et al. (2007) [40]	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	Y	N	Y	High
Wirth et al. (2014) [44]	Y	PY	N	Y	Y	Y	Y	Y	Y	N	Y	Y	Y	Y	Y	Y	Moderate
Yan et al. (2016) [49]	Y	N	N	PY	Y	Y	N	Y	Y	N	Y	Y	N	Y	N	Y	Low
Yang et al. (2019) [50]	Y	N	N	PY	Y	Y	N	N	N	N	Y	N	N	Y	Y	Y	Low

Y- Yes; N–No; PY–Partial Yes; NMA–No Meta-Analysis.

Discussion

This umbrella review on the impact of the management of SCH on clinical outcomes covers evidence from 20 selected systematic reviews and meta-analyses of RCTs and observational studies. Across the outcomes of interest, the synthesised literature found can be summarised as follows. We found that the treatment of SCH may be associated with a reduced likelihood of death from all causes for patients under 70 years old. On the other hand, the relationship between SCH treatment status and the risk of death compared to the euthyroid population remains unclear. Increased risk of all-cause mortality for untreated SCH was reported in only one primary review [50], for which inclusion was restricted to patients with comorbid heart failure.

We also found that compared to euthyroidism, untreated SCH patients with very high TSH (>10mIU/L) may be at greater risk of cardiovascular events and death from cardiovascular disease [54, 55]. The same effects were observed for thyroid peroxidase antibody-negative patients [53]. Even so, there was discordance in findings between the primary reviews; whereas seven primary reviews reported a higher risk of CHD and cardiovascular mortality for untreated SCH patients than euthyroid persons [46–48, 50, 53–55], one primary review did not [43]. We rated the latter as having more than one non-critical weakness according to the AMSTAR-2 checklist; the others were either ’high’ or ’moderate’ in overall confidence in their results (Table 10). A high degree of overlap was calculated between the studies reporting cardiovascular outcomes, as high as 38%. Therefore, it was not easy to ascertain the precise source of the difference in results.

It was not possible to investigate the impact of treatment on the risk of stroke because the only available comparisons were of untreated SCH and euthyroidism. The finding that untreated patients had better functional outcomes one month following stroke was reported only in one low-quality study [41] and is, therefore, inconclusive. In a similar vein, there was insufficient evidence of the impact of treatment or no treatment of SCH on fracture risk. Overall confidence in the results of three out of the four primary reviews [38, 44, 45] was rated as ’high’ or ’moderate’. However, none of the effect estimates was statistically significant, so it is also not possible to make conclusions on this relationship based on the quantity of evidence.

Reported findings on quality of life and the presence of symptoms between treated and untreated SCH patients were mainly of no statistical significance. As such, we cannot definitively state whether levothyroxine treatment improves or worsens these outcomes. Nonetheless, medication potentially benefits two patient groups–patients with severe SCH from the start and those that previously received treatment for autoimmune hyperthyroidism/Graves’ disease [36].

The secondary outcome reported in the included primary reviews, cognitive function, was only compared between treated and untreated SCH patients. Given that the majority of findings were similar, it may be said that among patients with SCH, levothyroxine may have no significant impact on cognitive function, notwithstanding the type of assessment tool used [17, 38–40]. Crucially, however, two points must be emphasised. First, that the amount of evidence in favour of this statement is notably low, considering that few primary reviews that reported on cognitive function. Second, that only the primary outcomes were included in the literature searches, so the findings in this review cannot accurately reflect the body of evidence regarding the relationship between SCH and cognitive function.

With reference to the number of primary reviews, the volume of evidence was discernibly skewed in favour of cardiovascular outcomes (n = 13) rather than all-cause mortality (n = 7), stroke (n = 2), fractures (n = 4), quality of life (n = 5) and cognitive function (n = 4). This observation can be explained as having arisen from the umbrella review selection process, but the relatively broad inclusion criteria make it less probable. Instead, two alternatives are suggested; either that less research has been performed on the other clinical outcomes of interest or that the evidence may not have already been synthesised due to high between-study heterogeneity, for example, in outcome definitions and measurements. Additional factors, such as the comparative ease of measuring certain outcomes over others, may also influence which types of studies are performed. However, it is not possible to conclusively account for this asymmetry of evidence from this overview alone.

Generally, it cannot be ignored that most of our findings were based on empirical studies of poor quality, as reported by the authors of the primary reviews. Equally important were the critical flaws we found in the methodological quality of five of the selected primary reviews [36, 41, 48–50] consequently rated as ’low’ in overall confidence in their results. Upon inspection, there was no clear boundary of review quality based on the type of empirical research that was initially selected. For example, all the syntheses that included only RCTs did not consistently get higher AMSTAR-2 ratings than those of only observational studies. As such, it can be argued that cohort studies have an essential role in filling the gap left by insufficient randomised trials on this topic.

It should be noted that there was a tendency for papers with the lowest ratings on the AMSTAR-2 checklist to have little overlap of empirical studies with other higher-rated primary reviews. This could be explained by differences in the types of outcomes reported in these syntheses; for instance, one would expect minimal overlap between fractures and cardiovascular mortality. Collectively, the reviews included in this overview had slight overlap, but as Hennesy and Johnson [31] contend, such an observation can be attributed to the breadth of the literature. This is especially true if only a small set of identical studies is shared across the included syntheses, or the overlap is highly outcome-dependent [31]. In these cases, the overall CCA would obscure the true level of overlap.

Strengths and limitations

To the authors’ knowledge, this is the first umbrella review on this topic. This overview was conducted in a systematic manner and comprehensive searches were performed to identify the synthesised literature on the impact of the management of SCH on long-term clinical outcomes. The database searches–including grey literature, to minimise the effects of publication bias [57]–were updated in the course of the review. Screening, data extraction and quality appraisal were all done in duplicate. Furthermore, the intended aim of the umbrella review to compare the synthesised literature on this topic was achieved, even though a secondary meta-analysis was not feasible.

Nonetheless, it is crucial to consider the limitations of this review which relied exclusively on the availability, methods and quality of existing systematic reviews and meta-analyses. Of note, it was not possible to re-analyse and pool all primary review findings due to the variety of selection criteria and outcome definitions. Combining the findings of the included reviews in spite of these differences–and potential confounders–would result in biased and misleading inferences [58].

Also, an inherent limitation of the umbrella review methodology is the limited capacity to conduct detailed evaluations of empirical studies when dealing with synthesised literature. This was particularly challenging when evaluating overlap across the included reviews, as it may have been influenced by factors such as study scope and eligibility criteria. On the other hand, because this type of review was performed, it was possible to examine a wide variety of outcomes for SCH and treatment status within our specific resource constraints. The comprehensive nature of umbrella reviews has been recommended for controversial topics [19]. Furthermore, in this overview, we included IPD meta-analyses, which have been described as beneficial for analysing long-term patient outcomes [59].

Another limitation was scope mismatch between the umbrella review and the included primary reviews, for example, in cases where a selected systematic review included patients with subclinical hypo- and hyperthyroidism. This problem is commonly encountered in overviews [18], and we opted to include such papers for two key reasons. First, a preliminary literature search yielded few results with precisely the same research questions. Second, for an unresolved topic such as this, it was anticipated that the exclusion of these reviews would severely restrict this synthesis by omitting potentially relevant findings. Therefore, to limit this type of bias, inclusion in the umbrella review was based on the availability of results for which treatment status was explicitly stated.

It was also not possible to calculate overlap for the included primary reviews subdivided by their reported outcomes because most included both treated and untreated SCH patients. Consequently, assessing overlap in this way would require a detailed inspection of all their primary studies to identify the exact data sources for the respective subgroup analyses. These activities were considered to be burdensome and beyond the scope of this umbrella review, given that systematic reviews and meta-analyses were the principal units of analysis. Even so, to visualise overlap, we created a citation matrix and presented the results of the pairwise calculations, most of which were in the ’slight’ band.

Conclusion

Through this umbrella review, we systematically gathered the existing synthesised literature on the impact of the management of subclinical hypothyroidism on clinical outcomes. Our findings seem to indicate that treatment may be beneficial for SCH patients younger than 70 years due to the higher risk of all-cause mortality and cardiovascular events. In addition, untreated SCH patients with TSH levels above 10mIU/L may be at higher risk of developing cardiovascular diseases than the euthyroid population. However, more robust evidence is needed on stroke, fractures, quality of life and cognitive function in SCH. The main challenge in investigating long-term outcomes is the need for large, adequately powered and timed randomised trials. This overview further highlights this need, given that majority of the significant findings were based on very few empirical studies often deemed to be of poor quality by the primary reviewers. Future work in observational studies may also be instrumental in strengthening the evidence base.

MEDLINE search strategy.

(PDF)

Click here for additional data file.

AMSTAR-2 checklist.

(PDF)

Click here for additional data file.

Excluded papers.

(PDF)

Click here for additional data file.

PRISMA checklist.

(PDF)

Click here for additional data file.

PRISMA abstract checklist.

(PDF)

Click here for additional data file.

30 Mar 2022

A rebuttal letter that responds to each point raised by the academic editor and reviewer(s). You should upload this letter as a separate file labeled 'Response to Reviewers'.

A marked-up copy of your manuscript that highlights changes made to the original version. You should upload this as a separate file labeled 'Revised Manuscript with Track Changes'.

An unmarked version of your revised paper without tracked changes. You should upload this as a separate file labeled 'Manuscript'.

If you would like to make changes to your financial disclosure, please include your updated statement in your cover letter. Guidelines for resubmitting your figure files are available below the reviewer comments at the end of this letter.

If applicable, we recommend that you deposit your laboratory protocols in protocols.io to enhance the reproducibility of your results. Protocols.io assigns your protocol its own identifier (DOI) so that it can be cited independently in the future. For instructions see: https://journals.plos.org/plosone/s/submission-guidelines#loc-laboratory-protocols. Additionally, PLOS ONE offers an option for publishing peer-reviewed Lab Protocol articles, which describe protocols hosted on protocols.io. Read more information on sharing protocols at https://plos.org/protocols?utm_medium=editorial-email&utm_source=authorletters&utm_campaign=protocols.

We look forward to receiving your revised manuscript.

Kind regards,

Mahmoud Nassar, MD, PhD, MSc, MHA, MPA, CPHQ, SSBB

Academic Editor

PLOS ONE

Journal Requirements:

When submitting your revision, we need you to address these additional requirements.

1. Please ensure that your manuscript meets PLOS ONE's style requirements, including those for file naming. The PLOS ONE style templates can be found at

https://journals.plos.org/plosone/s/file?id=wjVg/PLOSOne_formatting_sample_main_body.pdf and

https://journals.plos.org/plosone/s/file?id=ba62/PLOSOne_formatting_sample_title_authors_affiliations.pdf

[Note: HTML markup is below. Please do not edit.]

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Yes

**********

2. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: Yes

**********

3. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

**********

4. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

5. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: Although a statistically sound and well thought out paper, the main conclusion, namely that treatment for SCH in patients less than 70 may be beneficial is based off of limited data. The paper does little to further the current knowledge base for SCH and does not change or modify the current thought process surrounding SCH. Its main conclusion was based off of one limited study and really does not warrant being made without further research into the subject. Although there is a large role for "negative studies" in the existing medical literature, a negative umbrella review does little to further critical thinking or existing treatment paradigms. It was a well thought out and reasoned paper which kept me engaged throughout, but does not warrant publication in a major journal.

Reviewer #2: Dear Author,

I appreciate your time and effort in researching this subject.

The effort and work shown here were impressive. I have some suggestions to improve it.

-Please use the PRISMA check list when preparing an abstract and manuscript (and attach them as a supplement)

Please mention the name of the databases, keywords for searching, and inclusion and exclusion criteria in the abstract.

Please specify the duration of mortality outcomes in the methods section.

-Including gray literature weakens the analysis, please clarify if any studies have been included.

Please cite the figures in your manuscript.

The included study was not statistically analyzed.

**********

6. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: Yes: Mahmoud Nassar, MD, PhD

[NOTE: If reviewer comments were submitted as an attachment file, they will be attached to this email and accessible via the submission site. Please log into your account, locate the manuscript record, and check for the action link "View Attachments". If this link does not appear, there are no attachment files.]

While revising your submission, please upload your figure files to the Preflight Analysis and Conversion Engine (PACE) digital diagnostic tool, https://pacev2.apexcovantage.com/. PACE helps ensure that figures meet PLOS requirements. To use PACE, you must first register as a user. Registration is free. Then, login and navigate to the UPLOAD tab, where you will find detailed instructions on how to use the tool. If you encounter any issues or have any questions when using PACE, please email PLOS at figures@plos.org. Please note that Supporting Information files do not need this step.

7 Apr 2022

Comments & Responses

Reviewer #1: Although a statistically sound and well thought out paper, the main conclusion, namely that treatment for SCH in patients less than 70 may be beneficial is based off of limited data. The paper does little to further the current knowledge base for SCH and does not change or modify the current thought process surrounding SCH. Its main conclusion was based off of one limited study and really does not warrant being made without further research into the subject. Although there is a large role for "negative studies" in the existing medical literature, a negative umbrella review does little to further critical thinking or existing treatment paradigms. It was a well thought out and reasoned paper which kept me engaged throughout, but does not warrant publication in a major journal.

Response:

We thank Reviewer #1 for this feedback. Though we appreciate their comments on the lack of novel insights generated from this tertiary review, we must emphasise that its primary aim was to compare and contrast existing systematic reviews on the impact of the management of subclinical hypothyroidism on long-term outcomes. This is unlike typical secondary reviews (i.e. systematic reviews or meta-analyses) of empirical studies, which are usually performed with the expectation of new findings arising from the synthesis of earlier studies. Evidence of this distinction for tertiary studies is provided below:

Umbrella reviews are conducted to provide an overall examination of the body of information that is available for a given topic, and to compare and contrast the results of published systematic reviews.2 The wide picture obtainable from the conduct of an umbrella review is ideal to highlight whether the evidence base around a topic is consistent or contradictory, and to explore the reasons for the findings. Furthermore, an umbrella review allows ready assessment of whether review authors addressing similar review questions independently observe similar results and arrive at generally similar conclusions. (Aromataris et al, 2015)

Source: Aromataris, E., Fernandez, R., Godfrey, C.M., Holly, C., Khalil, H. and Tungpunkom, P., 2015. Summarizing systematic reviews: methodological development, conduct and reporting of an umbrella review approach. JBI Evidence Implementation, 13(3), pp.132-140.

In light of the above explanation, we put forward that an umbrella review can neither be positive nor negative, as such a description does not align with the overall purpose of this kind of tertiary research.

Nonetheless, a statement has been added to the Strengths and Limitations section for added clarity:

“Furthermore, the intended aim of the umbrella review to compare the synthesised literature on this topic was achieved, even though a secondary meta-analysis was not feasible.” (Lines 550 - 552)

Also, the quality and availability of literature is acknowledged as a key limitation in the manuscript, including in the Conclusion:

“The main challenge in investigating long-term outcomes is the need for large, adequately powered and timed randomised trials. This overview further highlights this need, given that majority of the significant findings were based on very few empirical studies often deemed to be of poor quality by the primary reviewers. Future work in observational studies may also be instrumental in strengthening the evidence base.” (Lines 597 - 602)

Reviewer #2: Please use the PRISMA check list when preparing an abstract and manuscript (and attach them as a supplement).

Response:

We thank Reviewer #2 for this recommendation. An additional PRISMA checklist for abstracts has been filled and is attached as Supplementary File S5 (S5 Appendix).

Reviewer #2: Please mention the name of the databases, keywords for searching, and inclusion and exclusion criteria in the abstract.

Response:

We agree with this suggestion. This section of the abstract has been revised and now reads:

“We conducted comprehensive searches on MEDLINE, EMBASE, Scopus, Web of Science, Cochrane Database of Systematic Reviews, JBI Evidence Synthesis, the PROSPERO register, Epistemonikos Database and PDQ Evidence from inception to February and July 2021 using keywords on subclinical hypothyroidism, treatment with levothyroxine, monitoring and primary outcomes (all-cause mortality, cardiovascular events, stroke, frailty fractures and quality of life). Only systematic reviews and meta-analyses on adult patient populations were considered.” (Lines 27 - 39)

Reviewer #2: Please specify the duration of mortality outcomes in the methods section.

Response:

We thank Reviewer #2 for this suggestion. The methods section has been revised to include the duration:

“…all-cause mortality, defined as the death of patients with SCH, irrespective of the cause, at least 12 months from baseline or the start of follow-up…” (Lines 177 - 178)

Reviewer #2: Including gray literature weakens the analysis, please clarify if any studies have been included.

Response:

We thank Reviewer #2 for this comment. However, umbrella review guidance recommends the inclusion of grey literature for wider coverage, thereby minimising the effects of publication bias. The main example of this is the Joanna Briggs Institute Manual for Evidence Synthesis which states:

A comprehensive search for a JBI Umbrella Review should also encompass a search for grey literature or reports that are not commercially published. As decision makers are increasingly required to base their decisions on available evidence, more and more research syntheses are being commissioned by practitioners and health care policy makers in governments globally; as a result many reports available via government or organizational websites are syntheses of research evidence and may be eligible for inclusion in a JBI Umbrella Review. A JBI Umbrella Review should include a search of at least two or three relevant sources for “grey” reports. (Aromataris et al. 2020)

Source: Aromataris E, Fernandez R, Godfrey C, Holly C, Khalil H, Tungpunkom P. Chapter 10: Umbrella Reviews. In: Aromataris E, Munn Z (Editors). JBI Manual for Evidence Synthesis. JBI, 2020. Available from https://synthesismanual.jbi.global. https://doi.org/10.46658/JBIMES-20-11

Nonetheless, the first paragraph in the Results section has been amended to show that grey literature was included:

“One item of grey literature, a systematic evidence review commissioned by a government agency for healthcare research, was included [36].” (Lines 266 - 267)

Justification for including grey literature has also been added to the Strengths and Limitations section:

“The database searches – including grey literature, to minimise the effects of publication bias [57] – were updated in the course of the review.” (Lines 547 - 549)

Reviewer #2: Please cite the figures in your manuscript.

Response:

Additional in-text citations have been included in the Results section for all the sentences where effect estimates were copied from the outcome summary tables. (Lines 315 - 403)

Reviewer #2: The included study was not statistically analyzed.

Response:

We thank Reviewer #2 for this comment. The 20 studies included, though eligible for inclusion in this umbrella review, were too varied for a meaningful and non-misleading statistical analysis. This is a particular weakness of umbrella review methodology, given the level of overlap between some of the reviews. The process of unpicking data from all the individual empirical studies was beyond the scope of this tertiary review. A statement has been added to the Strengths and Limitations to highlight this reasoning:

“Combining the findings of the included reviews in spite of these differences – and potential confounders – would result in biased and misleading inferences [57].” (Lines 557 - 558)

Submitted filename: Response to Reviewers.docx

Click here for additional data file.

22 Apr 2022

The impact of the management strategies for patients with subclinical hypothyroidism on long-term clinical outcomes: An umbrella review

PONE-D-22-05735R1

Dear Dr. Bauer,

We’re pleased to inform you that your manuscript has been judged scientifically suitable for publication and will be formally accepted for publication once it meets all outstanding technical requirements.

Within one week, you’ll receive an e-mail detailing the required amendments. When these have been addressed, you’ll receive a formal acceptance letter and your manuscript will be scheduled for publication.

An invoice for payment will follow shortly after the formal acceptance. To ensure an efficient process, please log into Editorial Manager at http://www.editorialmanager.com/pone/, click the 'Update My Information' link at the top of the page, and double check that your user information is up-to-date. If you have any billing related questions, please contact our Author Billing department directly at authorbilling@plos.org.

If your institution or institutions have a press office, please notify them about your upcoming paper to help maximize its impact. If they’ll be preparing press materials, please inform our press team as soon as possible -- no later than 48 hours after receiving the formal acceptance. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information, please contact onepress@plos.org.

Kind regards,

Mahmoud Nassar, MD, PhD, MSc, MHA, MPA, CPHQ, SSBB

Academic Editor

PLOS ONE

Additional Editor Comments (optional):

Dear Author

Congratulations!

1. Please check proof of the article at least twice (best done by two different people) before finalizing.

2. Please share your study with academic colleagues, in medical conferences and in social media (LinkedIn, twitter), so that it is cited widely. If you are putting your study on twitter.

3. Please continue to contribute to the Journal.

Reviewers' comments:

Reviewer's Responses to Questions

Comments to the Author

1. If the authors have adequately addressed your comments raised in a previous round of review and you feel that this manuscript is now acceptable for publication, you may indicate that here to bypass the “Comments to the Author” section, enter your conflict of interest statement in the “Confidential to Editor” section, and submit your "Accept" recommendation.

Reviewer #1: All comments have been addressed

Reviewer #2: All comments have been addressed

**********

2. Is the manuscript technically sound, and do the data support the conclusions?

The manuscript must describe a technically sound piece of scientific research with data that supports the conclusions. Experiments must have been conducted rigorously, with appropriate controls, replication, and sample sizes. The conclusions must be drawn appropriately based on the data presented.

Reviewer #1: Yes

Reviewer #2: Yes

**********

3. Has the statistical analysis been performed appropriately and rigorously?

Reviewer #1: Yes

Reviewer #2: N/A

**********

4. Have the authors made all data underlying the findings in their manuscript fully available?

The PLOS Data policy requires authors to make all data underlying the findings described in their manuscript fully available without restriction, with rare exception (please refer to the Data Availability Statement in the manuscript PDF file). The data should be provided as part of the manuscript or its supporting information, or deposited to a public repository. For example, in addition to summary statistics, the data points behind means, medians and variance measures should be available. If there are restrictions on publicly sharing data—e.g. participant privacy or use of data from a third party—those must be specified.

Reviewer #1: Yes

Reviewer #2: Yes

**********

5. Is the manuscript presented in an intelligible fashion and written in standard English?

PLOS ONE does not copyedit accepted manuscripts, so the language in submitted articles must be clear, correct, and unambiguous. Any typographical or grammatical errors should be corrected at revision, so please note any specific errors here.

Reviewer #1: Yes

Reviewer #2: Yes

**********

6. Review Comments to the Author

Please use the space provided to explain your answers to the questions above. You may also include additional comments for the author, including concerns about dual publication, research ethics, or publication ethics. (Please upload your review as an attachment if it exceeds 20,000 characters)

Reviewer #1: The authors have addressed all pending comments and have met criteria for submission. I recommend publication.

Reviewer #2: Dear Author

I have reviewed the article again and I'd thank you for responding to all comments.

Best regards

**********

7. PLOS authors have the option to publish the peer review history of their article (what does this mean?). If published, this will include your full peer review and any attached files.

If you choose “no”, your identity will remain anonymous but your review may still be made public.

Do you want your identity to be public for this peer review? For information about this choice, including consent withdrawal, please see our Privacy Policy.

Reviewer #1: No

Reviewer #2: No

12 May 2022

PONE-D-22-05735R1

The impact of the management strategies for patients with subclinical hypothyroidism on long-term clinical outcomes: An umbrella review

Dear Dr. Bauer:

I'm pleased to inform you that your manuscript has been deemed suitable for publication in PLOS ONE. Congratulations! Your manuscript is now with our production department.

If your institution or institutions have a press office, please let them know about your upcoming paper now to help maximize its impact. If they'll be preparing press materials, please inform our press team within the next 48 hours. Your manuscript will remain under strict press embargo until 2 pm Eastern Time on the date of publication. For more information please contact onepress@plos.org.

If we can help with anything else, please email us at plosone@plos.org.

Thank you for submitting your work to PLOS ONE and supporting open access.

Kind regards,

PLOS ONE Editorial Office Staff

on behalf of

Dr. Mahmoud Nassar

Academic Editor

PLOS ONE