Patient Context and Thyrotropin Levels Are Important When Considering Treatment of Subclinical Hypothyroidism

General Considerations About Clinical Practice Guidelines

Clinical practice guidelines (CPGs) are intended to provide recommendations based on the best available scientific evidence and thereby promote the standardization of medical practice along scientific principles, ultimately striving to improve patient outcomes (1). However, CPGs may be subjected to conscious and unconscious biases of authors (2). Furthermore, a popular methodological framework used in developing evidence-based CPGs has been questioned with respect to its theoretical basis in assigning the strength of recommendations as well as its lack of clarity in operationalizing and integrating the criteria/components (3). Recently, Djulbegovic et al. reported a lack of significant association between the “certainty of evidence” and the strength of CPG recommendations “against” the use of health care interventions (4). As such, there may be special challenges in interpreting and applying strong CPG recommendations against the use of health care interventions.

An important consideration in applying CPGs is patient-specific contextual factors. Mercuri et al. have reported that health care decisions are often influenced by the patient's circumstance (defined by personal, situational, or societal factors) (1). Some potentially relevant patient-specific factors could include: the severity/risk of disease, the presence of symptoms, relevant comorbidities, the attitude of patients/families/health care professionals to the health care options, access and affordability of health care, and other social factors (such as employment or other relevant life circumstances). The limitations in interpreting evidence and consideration of patient contextual factors are the focus of this editorial, as applied to the example of a recent CPG on the management of subclinical hypothyroidism (SCH) (5).

By a traditional definition, patients with SCH exhibit thyrotropin (TSH) levels that are above the normal reference range and normal free thyroxine (fT4) levels; the term “subclinical” implies an absence of clear clinical symptoms or signs of hypothyroidism.

In a recent Rapid Recommendation published in the British Medical Journal (BMJ), Bekkering et al. reported “a strong recommendation against thyroid hormone treatment in adults with SCH (defined as elevated thyroid stimulating hormone [TSH] levels and normal fT4 levels)” (5). The authors indicated that this recommendation was not applicable to women who are trying to become pregnant, individuals with TSH values >20 mIU/L, those with severe symptoms, or young adults (e.g., ≤30 years of age) (5). Rapid Recommendation CPGs are intended to promptly translate clinical research into evidence-based practice (6). The recommendation from Bekkering et al. (5) was based on a systematic review and meta-analysis previously reported in the Journal of the American Medical Association (JAMA) by Feller et al. (7). Feller et al. reported no significant benefit of levothyroxine treatment for multiple outcomes (e.g., quality of life [QoL]/mood-related outcomes, cognitive function, systolic blood pressure, and body mass index) in relatively short-term randomized controlled trials (3–18 months in duration) (7). A variety of relevant clinical, methodological, and contextual factors are discussed herein.

Consideration of Severity of Baseline TSH Elevation

In considering the baseline TSH values of individuals with SCH enrolled in levothyroxine treatment trials, both Feller et al. (7) and Bekkering et al. (5) reported that the baseline TSH values of individuals enrolled in randomized controlled trials of levothyroxine treatment of SCH ranged from 4.4 to 12.8 mIU/L, with only two trials including patients with TSH values >10 mIU/L. Given the paucity of patients with TSH values >10 mIU/L studied in these trials, Feller et al. acknowledged that the results of the meta-analyses “may not be generalizable to people with subclinical hypothyroidism and a thyrotropin level higher than 10 mIU/L” (7). However, Bekkering et al. used the same evidence to formulate a CPG treatment recommendation in individuals with TSH elevations up to 20 mIU/L (5).

In an effort to resolve an apparent discrepancy in interpretation of the baseline thyroid biochemistry and inferred generalizability of evidence from SCH trials, we reviewed all trials included in the review by Feller et al. (7) and performed meta-analyses of baseline TSH values, grouping trials according to respective examined outcomes. We abstracted all the baseline mean TSH values, their standard deviations (or calculated the standard deviations based on reported data), and trial sample sizes. TSH data were preferentially abstracted for entire study populations, but treatment/control subgroup data were used, if that was all that was reported in the original trials. We then performed random effects meta-analyses of the baseline TSH values from the respective trials (CMA meta-analysis version 2), according to each clinical outcome reported by Feller et al. (7) and Bekkering et al. (5). We determined that the mean pooled baseline TSH values of patients included in the respective outcome meta-analyses ranged from 6.4 to 7.3 mIU/L, with the upper limit of the 95% confidence intervals ranging from 6.9 to 8.8 mIU/L and statistically significant heterogeneity in all meta-analyses (Table 1). Thus, overall, there was a relatively mild degree of TSH elevation (TSH ≤10 mIU/L) in patients included in the trials, although these findings were limited by statistical heterogeneity (p < 0.05 as measured by the Cochrane's Q-test). We agree with Feller et al. (7) that it is inappropriate to generalize the findings of SCH trials to individuals with TSH values ≥10 mIU/L.

One of the key factors that Bekkering et al. reported in supporting a strong recommendation against levothyroxine treatment in SCH was the “burden of lifelong management” and “uncertainty on potential harms” of levothyroxine treatment (5). Important harms were not identified in the antecedent systematic review of (relatively short-term) randomized trials (7). The CPG authors did not report any formal systematic review of long-term observational studies in SCH. TSH elevations above 10 mIU/L may have some prognostic implications as such values have been associated with increased age- and sex-adjusted risk for coronary heart disease events (8), coronary heart disease mortality (8), and heart failure events (9), all in large-scale epidemiological studies. Although not specifically considered as a harm by Bekkering et al. (5), progression to overt hypothyroidism (TSH elevation >20 mIU/L or low free thyroid hormone levels and known associated morbidity), would be a risk in untreated patients. Moreover, in long-term prospective studies, the risk of progression to overt hypothyroidism has been shown to significantly vary according to the degree of the baseline TSH elevation (10 –12), with greater TSH elevations being associated with higher progression risk. For example, Díez et al. reported that in 107 untreated individuals with SCH who were followed every 6 months over a mean period of 32 months, overt hypothyroidism was diagnosed in 66.7% (24/36) of patients with TSH values of 10 to 19.9 mIU/L compared with the rate of 5.6% (4/71) in patients with TSH values of 5–9.9 mIU/L (10). The risk of progression to overt hypothyroidism was significantly associated with baseline TSH value, after adjusting for sex, age, symptoms, goiter, and presence of thyroid peroxidase antibodies (10). These data suggest that important subgroup considerations exist, specifically relating to whether the TSH value is above or below the threshold of ∼10 mIU//L. It is important to note that an inferred TSH treatment threshold of >20 mIU/L reported by Bekkering et al. (5) is higher than the 10 mIU/L threshold recommended in established CPGs on SCH (13,14). We have concerns about generalizing the results of data derived (on treatment benefit/harms or harms of no treatment) from populations at lower risk of disease progression to overt hypothyroidism (i.e., those with TSH values <10 mIU/L) to that of higher risk individuals (i.e., TSH ≥10 mIU/L), without sufficiently reviewing the available observational evidence in the higher risk group.

Consideration of Mild TSH Elevation in the Elderly

Another special consideration is mild TSH elevations in the elderly. Both American and European Thyroid Association guidelines on thyroid hormone replacement have highlighted that in asymptomatic elderly individuals, mild TSH elevations (i.e., TSH elevations <7–10 mIU/L) should not be considered as an indicator of thyroid disease (14,15). Treatment of asymptomatic elderly individuals with mild TSH elevations is not warranted; moreover, it may be inappropriate to label such individuals with a thyroid disease (SCH). The elderly may be particularly vulnerable to the consequences of overdiagnosis of SCH, particularly complications of overtreatment with levothyroxine and inadvertent thyroid hormone excess (e.g., risks of fractures and atrial fibrillation).

Consideration of Symptoms and QoL

Symptoms that could be suggestive of hypothyroidism are common (e.g., fatigue, weight gain, or others) and can be distressing to patients. Feller et al. reported that only about one-third of trials (7/21) included in the systematic review reported on hypothyroid symptoms at baseline, and for those trials, symptoms were considered mild to moderate in severity (7). Furthermore, 75–80% of the overall effect size for general QoL and thyroid-related symptoms in the meta-analysis by Feller et al. (7) was derived from a single study, that is, the TRUST trial (16), which recruited largely asymptomatic older individuals with very mild SCH (median TSH 5.7 mIU/L). The baseline QoL scores of the participants in TRUST, using the thyroid-specific QoL ThyPRO questionnaire, were similar to those from the “healthy” control group with very few symptoms (expected low score group) in the original THYPRO validation study (17). A lack of treatment effect in asymptomatic individuals with very mild SCH does not preclude a beneficial effect in symptomatic individuals with more severe SCH, that is, a TSH value >10 mIU/L.

Patient Perspectives, Values, and Preferences

Patient perspectives, values, and preferences are an important consideration, but the means by which guideline authors distill these complex concepts is not standardized and validated. Some approaches that have been described to incorporate patient preferences in CPG development include systematic reviews of related quantitative or qualitative literature with patient panel members serving as “complementary” sources of information (18).

To address patient perspectives, Bekkering et al. included two patient panelists who had been diagnosed with SCH (5) but did not report any specific literature review on patient experiences/values/preferences. The patient panelists indicated that “patients may feel anxious about deteriorating or developing overt hypothyroidism when no treatment was given,” and to address this “regular follow-up is very important” (5). However, the authors of the CPG did not provide specific recommendations on the frequency and nature of follow-up. The patient panelists also indicated that “it is difficult for patients to make a decision when feeling unwell” (5). In considering this issue further, Nexø et al. elicited three key themes in a qualitative study of patients with thyroid dysfunction: (a) a sense of a loss of control over physical and mental states, (b) ambiguous symptoms, such as fatigue/feeling drained, which were diffuse and merged with the strains of everyday life, and (c) negotiating sickness, particularly trying to validate the sickness in the context of biochemical findings, often without validation, understanding, or support of health care practitioners (itself resulting in negative emotional consequences) (19). The “loss of control” described by Nexø et al. (19) may reflect the challenges in treatment decision-making in symptomatic individuals reported by the patient panelists in the guideline from Bekkering et al. (5). The experience of such individuals is complex and may require clinical care and attention beyond that of a review of the biochemical results.

We were interested to obtain input on the BMJ guideline from Bekkering et al. (5) from an individual diagnosed with SCH and a TSH value >10 mIU/L. In the case of our patient representative, the potential for many years of nonspecific symptoms (e.g., fatigue) with the need for more frequent biochemical and clinical monitoring without treatment, as well as a potential association with increased cardiovascular risk, were all factors that influenced the decision to take levothyroxine treatment. This patient is followed yearly and is on a stable low dose of levothyroxine with a TSH level well in the normal range, with no complications. Clearly, every individual is different; however, it is important to not dismiss “nonspecific” symptoms as being not important to the patient. An understanding of the complex experience of being diagnosed with a “thyroid disease” is needed.

Incorporating patient perspectives, values, and preferences in CPGs is clearly challenging. Montori et al. have advised that “guideline panelists must recognize, with humility, the challenges they face in working often without access to informed patient preferences and acknowledge that their recommendations should rarely assume uniform patient values and contexts in favor of a particular course of action” (20).

Case Examples

To highlight the importance of clinical and other contextual factors, a selection of four cases is presented (Table 2). For each case, clinical, biochemical, and relevant contextual data are presented. The reader is encouraged to review the cases and deliberate whether one or more factors presented would alter the decision to offer treatment for SCH to the individual patient. What becomes evident in this exercise is that in appropriately applying recommendations of CPGs to individual patients, contextual factors often not addressed in the guideline by Bekkering et al. (5) need to be considered. Such factors become particularly important, when recommendations are relatively broad in scope and do not consider relevant subgroup effects.

In conclusion, we agree that routine levothyroxine treatment of mild TSH elevation in asymptomatic individuals is not indicated, in the absence of pregnancy or pregnancy consideration. However, we have concerns relating to the certainty of evidence in issuing a strong recommendation against offering thyroid hormone treatment to individuals with TSH values ≥10 mIU/L. Furthermore, we also believe in the importance of treating the patient, and not simply a biochemical abnormality (21). In considering disease management options for individuals with SCH, additional relevant contextual factors such as symptoms, comorbidities, safety/feasibility issues (for both treatment and nontreatment), and patient values and preferences need to be considered.