Abstract
Context
Levothyroxine is one of the most prescribed medications in the United States.
Objective
This study explores the appropriateness of levothyroxine prescriptions.
Methods
A retrospective multicenter study was conducted on adult patients who were prescribed levothyroxine for the first time between 2017 and 2020 at three academic centers in the United States. We classified each case of levothyroxine initiation into one of three mutually exclusive categories: appropriate (clinically supported), indeterminate (clinically unclear), or nonevidence based (NEB, not clinically supported).
Results
A total of 977 participants were included. The mean age was 55 years (SD 19), there was female (69%) and White race predominance (84%), and 44% had possible hypothyroid symptoms. Nearly half of the levothyroxine prescriptions were considered NEB (528, 54%), followed by appropriate (307, 31%) and indeterminate (118, 12%). The most common reason for NEB prescription was an index thyrotropin (TSH) value of less than 10 mIU/L without previous TSH or thyroxine values (131/528, 25%), for appropriate prescription, was overt hypothyroidism (163/307, 53%), and for an indeterminate prescription was a nonconfirmed subclinical hypothyroidism with TSH greater than or equal to 10 mIU/L (no confirmatory testing) (51/118, 43%). In multivariable analysis, being female (odds ratio [OR]: 1.3; 95% CI, 1.0-1.7) and prescription by a primary care provider (OR: 1.5; 95% CI, 1.2-2.0) were associated with NEB prescriptions.
Conclusion
There is a considerable proportion of NEB levothyroxine prescriptions. These results call for additional research to replicate these findings and to explore the perspective of those prescribing and receiving levothyroxine.
Levothyroxine, a synthetic form of thyroxine (T4), is used to treat hypothyroidism by effectively replacing endogenous thyroid hormone and restoring normal thyroid hormone levels (1). It is one of the most commonly prescribed drugs in the United States (2), with approximately 7% of the population estimated to have an active prescription for levothyroxine (3). In addition, several prior studies suggest that the use of levothyroxine has increased over time. For instance, Ross et al (4) reported an increase in annual pharmacy fills of levothyroxine from around 9 million in 2007 to 12 million in 2016 in the United States. This pattern of levothyroxine-prescribing appears to be at odds with the epidemiology of overt hypothyroidism, a condition with a low and relatively stable prevalence rate of around 0.2% to 2%, as estimated by laboratory data (5, 6). In addition, although substantial increases in the incidence of thyroid cancer diagnosis and treatment in recent years could also be contributing to rising numbers of levothyroxine prescriptions (7), the observed trend is not proportional.
The discrepancy between high levothyroxine use and low rates of overt hypothyroidism could suggest that patients are receiving levothyroxine treatment for other indications, such as subclinical hypothyroidism (SCH). In contrast to overt hypothyroidism, SCH is much more common, affecting up to 12% to 18% of the population (5, 8, 9). In addition, studies suggest that the number of people diagnosed with SCH and initiating levothyroxine treatment has increased over time (10, 11). This increase in the diagnosis and treatment of SCH coincides with new evidence suggesting that levothyroxine therapy has little or no benefit for patients with SCH, particularly among patients older than 65 years who are also prone to complications from overtreatment (12-15). If levothyroxine is mostly used to treat patients with SCH, it would suggest that some prescriptions are nonevidence based (NEB). Similarly, a population-based study in the United States showed that about one-third of patients initiated levothyroxine for normal thyroid function (16), suggesting that factors other than thyroid hormone levels (eg, symptom management, pregnancy) may be drivers of levothyroxine prescriptions.
Understanding the factors associated with the appropriateness of levothyroxine prescribing is important because levothyroxine use may burden patients and society. Levothyroxine treatment requires modification of daily habits (eg, administration 30-60 minutes before a meal, monitoring of effects, frequent dose titration, and clinic and laboratory visits) as well as substantial financial costs to the patient (1). Compared with patients without hypothyroidism, patients with hypothyroidism are estimated to spend almost double in annual direct medical costs (17). In addition, the estimated total cost of care is $3166 per year in patients with no dose adjustment, and up to $8220 per year for patients requiring 3 or more dose adjustments (18). When levothyroxine prescriptions are appropriate and evidence based, the benefits of levothyroxine use outweigh its burden; on the contrary, when the prescription is NEB, the burden might be unnecessary and harmful.
The purpose of this study was to explore the drivers of levothyroxine prescriptions among patients at 3 large health care systems in the United States by answering the following questions: 1) what types of clinicians are prescribing levothyroxine and to what types of patients? 2) why is it being prescribed? 3) what proportion of levothyroxine prescriptions are potentially NEB based on current guidelines? and 4) which factors are most strongly associated with NEB prescriptions?
Materials and Methods
Study Design, Setting, and Patient Population
In this retrospective chart review study, we included adult patients who received their first outpatient levothyroxine prescription between January 1, 2017 and December 31, 2019, at the University of Arkansas for Medical Sciences (UAMS; Little Rock, Arkansas, USA) and Mayo Clinic (MCR; Rochester, Minnesota, USA), and between January 1, 2017 and December 21, 2020, at the University of Florida (UF; Gainesville, Florida, USA). We included all the cases at UAMS and UF. At MCR, we randomly screened a cohort of 500 patients from 1419 eligible; of those, the 467 cases who were deemed to be new prescriptions were included.
Each institution's ethics committee approved the study, and data transfer/use agreements were obtained.
Data Collection
Using a piloted extraction form developed a priori, electronic medical records (EMRs) were reviewed to extract the following elements: sociodemographic data, date and dosage of first levothyroxine prescription, factors that influence the initiation of levothyroxine (symptoms, comorbidities, family history of thyroid disorders, clinician specialty), documented reason for levothyroxine prescription (pregnancy or planning pregnancy, central hypothyroidism, after radioactive iodine [RAI], thyroid surgery, side effect of medications, symptoms, abnormal laboratory values, or others). We also extracted the results of any thyroid function tests (index TFTs) preceding the initial levothyroxine prescription, including thyrotropin (TSH), free T4 (FT4) or total T4, thyroid peroxidase (TPO) antibody tests. In cases where the patient had multiple TFTs, we extracted the TFTs dated closest to the levothyroxine prescription. In addition, we extracted whether previous TFTs were available up to 12 months before the index TFTs. Finally, for patients who had both TSH and FT4 or total T4 levels available at the time of levothyroxine prescription, we further classified them into the following categories based on the institution's reference range: overt hypothyroidism (high TSH and low T4); mild SCH (high TSH but <10 mLU/L and normal T4); moderate to severe SCH (TSH ≥ 10 mLU/L and normal T4); normal thyroid levels (normal TSH and normal T4); and other categories as needed for specific clinical profiles outside the previously listed combinations (eg, normal TSH with low T4; low TSH and low T4 without documentation of central hypothyroidism; hyperthyroidism). Finally, we used the nomenclature of “confirmed SCH” when there was at least one elevated TSH value before the index TSH; and “nonconfirmed SCH” when the previous TSH value was normal, or not obtained.
Based on the most recent American Thyroid Association guideline recommendation and research team consensus, we classified each case of levothyroxine initiation into one of three mutually exclusive categories: appropriate, indeterminate, or NEB (Supplement 1 (19)). Indications for appropriate levothyroxine prescription included patients with 1) overt hypothyroidism (high TSH and low T4), 2) confirmed moderate to severe SCH with index TSH of 10 mIU/L or greater, 3) confirmed mild SCH with index TSH less than 10 mIU/L in a patient younger than 65 years with at least one hypothyroid symptom, 4) prescription due to pregnancy or planning pregnancy, after RAI or thyroid surgery, or central hypothyroidism—appropriateness of these conditions were determined based on the diagnosis and documentation within the clinical note, rather than relying on TFT results. Indeterminate cases included 1) confirmed mild SCH with index TSH less than 10 mIU/L and age 65 years or older and symptoms, 2) confirmed mild SCH with index TSH less than 10 mIU/L (≥2 consecutive elevated TSH values documented), younger than 65 years and without symptoms, 3) nonconfirmed SCH with index TSH of 10 mIU/L or greater (only one documented elevated TSH value), 4) patients with index TSH of 10 mIU/L or greater without available T4 value, and with previous elevated TSH value. Likely NEB cases included 1) confirmed mild SCH with index TSH less than 10 mIU/L, and age 65 years or older without symptoms, 2) nonconfirmed mild SCH with index TSH less than 10 mIU/L, and younger than 65 years with and without symptoms, 3) nonconfirmed mild SCH with index TSH less than 10 mIU/L, and age 65 years or older with or without symptoms, 4) index TSH less than 10 mIU/L without available T4 value, and with preceding normal TSH value or no previous TSH value, 5) normal TSH and T4, and 6) other (eg, normal TSH and low T4 without evidence of central hypothyroidism; low TSH and high T4). To account for differences in expert recommendations for the appropriateness of levothyroxine prescription, we also classified the cases using a TSH cutoff of 7 mIU/L instead of 10 mIU/L.
During our investigation, we located the clinical note in which the levothyroxine prescription was documented. In this note, our reviewers diligently searched for any symptoms associated with hypothyroidism. The specific list of symptoms was predetermined and obtained through a thorough literature review (20), as well as the collective expertise of our research group.
To ensure precise and high-quality data extraction, we implemented initial rounds of duplication, covering around 10% of each institution's overall volume. The data extraction was carried out by specific teams: I.A., A.G.B., J.H., and S.C.M. extracted information at MCR; R.A.A. and J.S.R. at the U.F.; and S.M. and M.J. at the UAMS. During this process, any disagreements that arose were thoroughly reviewed and resolved collaboratively as a group. Subsequently, the remaining data extraction was performed individually.
Statistical Analysis
Categorical variables were analyzed as frequencies (percentages) and continuous variables as means (SD). Differences between categorical variables were assessed using the Fisher exact test and between continuous variables using the Welch unequal variances 2-sample t test. We performed a bivariate and multivariable logistic regression analysis to examine clinician and patient factors associated with NEB levothyroxine prescription for the main classification system (based on TSH cutoff of 10 mIU/L), and sensitivity analysis for the alternative classification system (based on TSH cutoff of 7 mIU/L). Those variables that were statistically significant in bivariate analyses, and deemed to be clinically relevant, were included in the model. TSH values were excluded from the model as this was the primary method for determining appropriateness of levothyroxine prescription. Noteworthy, the results were not stratified by the study sites as it was precluded by institutional review board approvals and data transfer agreements among the participant institutions. Results were reported as odds ratios (OR) and 95% CI. A P value of less than .05 was considered statistically significant, and all testing was 2-sided. All statistical analyses were performed using JMP 16.0.
Results
Indications and General Characteristics of Patients and Clinicians Prescribing Levothyroxine
We included 977 participants (467 from MCR, 236 from UF, and 274 from UAMS) with the following characteristics: mean age, 55 (SD 19) years, female 69%, White 84%, obese (body mass index ≥30) 42%, with one or more comorbidities (69%) or symptoms (44%). The most common comorbidities were depression (19%) and dyslipidemia (15%), and the most common symptoms were fatigue (22%) and cold intolerance (6%). The median index TSH value before levothyroxine prescription was 7.1 mIU/L (interquartile range [IQR] 5.1-11.5 mIU/L). Levothyroxine was mostly prescribed by primary care clinicians (57%), followed by nonendocrine specialists (24%), and endocrinologists (19%). Finally, the median dose of levothyroxine was 50 mcg daily (IQR 25-75 mcg/daily). Table 1 shows additional demographics.
Table 1.
Baseline patient characteristics
| Characteristics | No. | (%) |
|---|---|---|
| Center | ||
| MCR | 467 | (47.8%) |
| UAMS | 274 | (28.0%) |
| UF | 236 | (24.1%) |
| Age, y | ||
| Mean (SD) | 55 | (19) |
| <65 y | 605 | (61.9%) |
| ≥65 y | 372 | (38.0%) |
| Sex | ||
| Female | 670 | (68.5%) |
| Male | 307 | (31.4%) |
| Race | ||
| White | 828 | (84.7%) |
| Black or African American | 85 | (8.7%) |
| Asian | 30 | (3.0%) |
| Unknown/Not reported | 23 | (2.3%) |
| Other | 11 | (1.1%) |
| Education | ||
| Unknown | 586 | (59.9%) |
| > High school | 248 | (25.3%) |
| ≤ High school | 143 | (15.6%) |
| BMI | ||
| Obese | 419 | (42.9%) |
| Normal | 296 | (30.3%) |
| Overweight | 261 | (26.7%) |
| Clinical setting | ||
| PCP | 561 | (57.4%) |
| Endocrinology | 182 | (18.6%) |
| Oncology | 99 | (10.1%) |
| Other | 94 | (9.6%) |
| Gynecology | 41 | (4.2%) |
| TSH (mlU/L), median (IQR) | 7.1 | (5.1-11.6) |
| FT4 (ng/dL), median (IQR) | 0.9 | (0.7-1.1) |
| No. of symptoms | ||
| 0 | 550 | (56.2%) |
| 1-2 | 299 | (30.6%) |
| >2 | 128 | (13.1%) |
| Type of symptoms | ||
| No symptoms | 550 | (56.2%) |
| Fatigue | 205 | (20.9%) |
| Cold intolerance | 58 | (5.9%) |
| Depression | 47 | (4.8%) |
| Weight gain | 38 | (3.8%) |
| Other | 36 | (3.6%) |
| Constipation | 32 | (3.2%) |
| Hair loss | 11 | (1.1%) |
| Comorbidities | ||
| No comorbidities | 307 | (31.4%) |
| Depression | 185 | (18.9%) |
| Dyslipidemia | 151 | (15.4%) |
| Arrhythmia | 95 | (9.7%) |
| Psychiatric illness | 31 | (3.1%) |
| Thyroid nodule | 29 | (2.9%) |
| Heart failure | 25 | (2.5%) |
| History of infertility/miscarriages | 12 | (1.2%) |
| Myocardial infarction | 4 | (0.4%) |
| Other | 138 | (14.1%) |
Abbreviations: BMI, body mass index; CVD, cardiovascular disease; FT4, free T4; IQR, interquartile range; MCR, Mayo Clinic Rochester; PCP, primary care provider; SCH, subclinical hypothyroidism; TSH, thyrotropin; UAMS, University of Arkansas Medical Center; UF, University of Florida.
Proportion and Factors Associated With Nonevidence-Based Levothyroxine Prescriptions
Based on our classification plan, nearly half of the levothyroxine prescriptions were considered NEB (528, 54%), followed by appropriate (307, 31%) and indeterminate (118, 12%). The most common reason for NEB prescription was an index TSH value of less than 10 mIU/L with previous normal TSH (112/528, 21%), or without previous TSH value (131/528, 25%), and without available T4 value. For appropriate prescription, the most common reason was overt hypothyroidism (163/307, 53%), and for indeterminate prescription was a nonconfirmed SCH with TSH of 10 mIU/L or greater (51/118, 43%) (Table 2). In bivariate analyses (Table 3), patients receiving NEB levothyroxine prescriptions were more likely to be female than male (377, 71.4% vs 274, 64.5%; P = .02), receive their prescription from a primary care clinician than other specialty (331, 62.7% vs 218, 51.3%; P < .01), and have at least one symptom compared with no symptoms (242, 45.8% vs 167, 39.3%; P = .04). In multivariable analysis (Fig. 1), factors associated with receipt of an NEB levothyroxine prescription were a prescription from primary care (OR = 1.5; 95% CI, 1.1-2.9) and TPO antibody negativity or not available (OR = 1.4; 95% CI, 1.0-2.0).
Table 2.
Classification of appropriate prescription (analysis for groups based on thyrotropin cutoff of 10 mIU/L)
| Category | N = 977 | (%) |
|---|---|---|
| Appropriate | 307 | (31.4%) |
| Overt hypothyroidism | 163/307 | (53.0%) |
| Confirmed SCH with TSH ≥10 mIU/L | 37/307 | (12.0%) |
| Confirmed SCH with TSH <10 mIU/L, <65 y, and symptomatic | 32/307 | (10.4%) |
| After surgery, RAI, pregnancy, planning pregnancy, central hypothyroidism | 75/307 | (24.4%) |
| Indeterminate | 118 | (12.0%) |
| Nonconfirmed SCH with TSH ≥10 mIU/L | 51/118 | (43.2%) |
| Confirmed SCH with TSH <10 mIU/L, <65 y, and asymptomatic | 25/118 | (21.1%) |
| Confirmed SCH with TSH <10 mIU/L, ≥65 y, and symptomatic | 17/118 | (14.4%) |
| Only index TSH ≥10 mIU/L with previous elevated TSH | 25/118 | (21.1%) |
| Nonevidence based | 528 | (54.1%) |
| Nonconfirmed SCH with TSH <10 mIU/L, <65 y, and symptomatic | 85/528 | (16.0%) |
| Nonconfirmed SCH with TSH <10 mIU/L, <65 y, and asymptomatic | 58/528 | (10.9%) |
| Nonconfirmed SCH with TSH <10 mIU/L, ≥65 y, and symptomatic | 28/528 | (5.2%) |
| Confirmed SCH with TSH <10 mIU/L, ≥65 y, and asymptomatic | 30/528 | (5.6%) |
| Nonconfirmed SCH with TSH <10 mIU/L, ≥65 y, and asymptomatic | 22/528 | (4.1%) |
| Only index TSH with previous normal TSH values | 112/528 | (21.1%) |
| Only index TSH with no previous TSH values | 131/528 | (24.7%) |
| Euthyroid | 39/528 | (7.3%) |
| Other | 24/528 | (4.5%) |
| Missing values | 24 | (2.4%) |
Abbreviations: RAI, radioactive iodine; SCH, subclinical hypothyroidism; TSH, thyrotropin.
Table 3.
Factors associated with nonevidence-based prescription (bivariate analysis based on thyrotropin cutoff of 10 mIU/L)
| Factor | Appropriate(n = 307) | Indeterminate(n = 118) | NEB(n = 528) | Appropriate plus indeterminate (n = 425) | P a | ||||
|---|---|---|---|---|---|---|---|---|---|
| No. | (%) | No. | (%) | No. | (%) | No. | (%) | ||
| Age, y | |||||||||
| <65 | 189 | 61.5% | 69 | 36.7% | 331 | 62.6% | 258 | 60.7% | .53 |
| ≥65 | 118 | 38.4% | 49 | 41.5% | 197 | 37.3% | 167 | 39.2% | |
| Sex | |||||||||
| Female | 200 | 65.1% | 74 | 62.7% | 377 | 71.4% | 274 | 64.4% | .02 |
| Male | 107 | 34.8% | 44 | 23.4% | 151 | 28.6% | 151 | 35.5% | |
| Race | |||||||||
| White | 259 | 84.3% | 100 | 53.1% | 448 | 84.8% | 359 | 84.4% | .87 |
| Other/Unknown/NR | 48 | 15.6% | 18 | 15.2% | 80 | 15.1% | 66 | 15.5% | |
| Setting | |||||||||
| PCP | 144 | 46.9% | 74 | 62.7% | 331 | 62.6% | 218 | 51.2% | <.01 |
| Other | 163 | 53.0% | 44 | 23.4% | 197 | 37.3% | 207 | 48.7% | |
| Symptoms | |||||||||
| No symptom | 190 | 61.8% | 68 | 57.6% | 286 | 54.1% | 258 | 60.7% | .04 |
| Symptom | 177 | 57.6% | 50 | 26.6% | 242 | 45.8% | 167 | 39.2% | |
| Weight | |||||||||
| Overweight or Obese | 223 | 72.6% | 70 | 59.3% | 369 | 69.8% | 293 | 68.9% | .79 |
| Normal | 83 | 27.0% | 48 | 25.5% | 159 | 30.1% | 131 | 30.8% | |
| TPOAb | |||||||||
| Negative or not available | 244 | 79.4% | 90 | 76.2% | 422 | 79.9% | 334 | 78.5% | .61 |
| Positive | 63 | 20.5% | 28 | 14.8% | 106 | 20.0% | 91 | 21.4% | |
| Comorbidities | |||||||||
| ≥1 | 207 | 67.4% | 92 | 48.9% | 369 | 69.8% | 299 | 70.3% | .87 |
| None | 100 | 32.5% | 26 | 22.0% | 159 | 30.1% | 126 | 29.6% | |
Abbreviations: NEB, nonevidence based; NR, not reported; PCP, primary care provider; TPOAb, thyroid peroxidase antibody.
a P value of comparison between NEB vs appropriate and indeterminate.
Figure 1.
Forest plots for multivariable logistic regression analysis of factors associated with the nonevidence-based use of levothyroxine. Left panel, TSH cutoff of 10 mIU/L. Right panel, TSH cutoff of 7 mIU/L. OR, odds ratio; PCP, primary care provider; TPOAb, thyroid peroxidase antibody; TSH, thyrotropin.
In sensitivity analysis, when a TSH cutoff of 7 mIU/L rather than 10 mIU/L was used as the threshold for appropriateness, fewer prescriptions were classified as NEB (444, 45%), followed by appropriate (339, 35%) and indeterminate (171, 17%) (Table 4).
Table 4.
Classification of appropriate prescription (sensitivity analysis for groups based on thyrotropin cutoff of 7 mIU/L)
| Category | N = 977 | (%) |
|---|---|---|
| Appropriate | 339 | (34.7%) |
| Overt hypothyroidism | 163/339 | (48.0%) |
| Confirmed SCH with TSH ≥7 mIU/L | 77/339 | (22.7%) |
| Confirmed SCH with TSH <7 mIU/L, <65 y, and symptomatic | 24/339 | (7.0%) |
| After surgery, RAI, pregnant, planning pregnancy, central hypothyroidism | 75/339 | (22.1%) |
| Indeterminate | 171 | (17.5%) |
| Nonconfirmed SCH with TSH ≥7 mIU/L | 119/171 | (69.5%) |
| Confirmed SCH with TSH <7 mIU/L, <65 y, and asymptomatic | 16/171 | (9.3%) |
| Confirmed SCH with TSH <7 mIU/L, ≥65 y, and symptomatic | 11/171 | (6.4%) |
| Only index TSH ≥7 mIU/L with previous elevated TSH | 25/171 | (14.6%) |
| Nonevidence based | 444 | (45.4%) |
| Nonconfirmed SCH with TSH <7 mIU/L, <65 y, and symptomatic | 61/444 | (13.7%) |
| Nonconfirmed SCH with TSH <7 mIU/L, <65 y, and asymptomatic | 39/444 | (8.7%) |
| Nonconfirmed SCH with TSH <7 mIU/L, ≥65 y, and symptomatic | 13/444 | (2.9%) |
| Confirmed SCH with TSH <7 mIU/L, ≥65 y, and asymptomatic | 13/444 | (2.9%) |
| Nonconfirmed SCH with TSH <7 mIU/L, ≥65 y, and asymptomatic | 12/444 | (2.7%) |
| Only index TSH with previous normal TSH values | 112/444 | (25.2%) |
| Only index TSH with no previous TSH values | 131/444 | (29.5%) |
| Euthyroid | 39/444 | (8.7%) |
| Other | 24/444 | (5.4%) |
| Missing values | 24 | (2.4%) |
Abbreviations: RAI, radioactive iodine; SCH, subclinical hypothyroidism; TSH, thyrotropin.
Discussion
In this retrospective observational study based on the review of EMRs of patients from 3 large academic medical centers in the United States, we described patient characteristics and clinical features associated with receiving a levothyroxine prescription for the first time. In our sample, most patients started on levothyroxine were White, middle-aged, female, and overweight or obese, with at least one comorbidity. Almost half of the patients had at least one symptom of hypothyroidism documented in their chart. Most patients had mildly elevated TSH levels recorded before the levothyroxine prescription was issued (median: 7.1 mIU/L), and the average prescribed dose of levothyroxine was 50 mcg/daily. In addition, we found that approximately 8 out of 10 levothyroxine prescriptions were issued by primary care or nonendocrine clinicians. Finally, almost half of the levothyroxine prescriptions were considered NEB under our classification plan. The independent factors associated with NEB use were receiving a prescription from a primary care provider and being female.
A previous retrospective, longitudinal study using a national administrative claims database linked to laboratory results for commercially insured and Medicare Advantage enrollees found that the number of levothyroxine prescriptions in the United States is increasing and that the population receiving these prescriptions reflects that of the present study (16). This study also showed that mild SCH triggered most prescriptions, yet this study, due to a lack of data on patient symptoms, could not determine the appropriateness of the prescription. Thus, to our knowledge, our study is the first to present information about the possible number of NEB prescriptions in a selected sample of patients cared for in tertiary care centers in the United States. To classify for appropriateness, we opted to operationalize and adapt recommendations of levothyroxine use given by guidelines (1, 21, 22), other groups of experts (23, 24), and the research team opinion. Interestingly, despite having different TSH cutoffs, the 2 frameworks used consistently showed that almost half of the prescriptions were NEB. For both frameworks, the most common clinical indication preceding an NEB levothyroxine prescription was a TSH value of less than 10 mIU/L without confirmation or T4 value determination. Although it is not clear why some clinicians opt to prescribe levothyroxine solely based on a single mildly elevated TSH level without checking T4 levels, it is possible that this NEB use of levothyroxine is due to a misinterpretation of thyroid test results. Perhaps there is a perceived limited value of T4 testing among clinicians since guidelines recommend using TSH (and not T4 values) for thyroid dysfunction screening; therefore, clinicians may decide to use levothyroxine regardless of T4 levels. However, this practice has the potential to misdiagnose many patients and carries considerable implications for their care. Once the prescription is written, patients assume the diagnosis is accurate, leading them to believe they have hypothyroidism and may need levothyroxine therapy for the rest of their lives. This can create the wrong expectation that levothyroxine will resolve all their “hypothyroid symptoms,” even if those symptoms are not necessarily related to thyroid function.
Another important clinical profile associated with NEB levothyroxine prescription was the treatment of nonconfirmed mild SCH (only one measurement of TSH). TFTs fluctuate, and it is likely that many patients in these categories were indeed euthyroid. For instance, in a population-based study in more than 420 000 presumably healthy people, approximately 12 500 (3%) had a TSH elevation between 5.5 and 10 mIU/L (25). In 62% of them, without intervention, the TSH levels returned to normal within the following 5 years. Given this TSH variability, guidelines recommend that patients with possible SCH should have their TFTs repeated to confirm the diagnosis (1, 21, 22). In addition to the lack of confirmatory testing, these patients might not benefit from treatment given the lack of evidence of levothyroxine effectiveness in mild SCH (12, 26). A systematic review of 21 randomized trials enrolling 2192 patients (age predominantly >65 years) with mostly mild SCH showed that when levothyroxine was compared to placebo, or there was no intervention, there were little to no differences between the 2 groups in terms of quality of life, thyroid-related symptoms, depressive symptoms, fatigue, or cognitive function (21). These randomized trials have included mostly women (46%-100%) with mean age of 50 years (range, 32-74 years), mean TSH at baseline of 6.6 mIU/L (range, 4.4-12.8 mIU/L), and mild to moderate burden of symptoms. In addition, one trial found that treatment with SCH did not affect the incidence of cardiovascular events within 1 year after initiation of therapy (13).
Additional research (eg, interviews with patients and clinicians, or conversation analysis) should explore the reasons behind the apparent NEB levothyroxine prescriptions. However, our results suggest a few clues that could already affect the development of interventions to decrease NEB prescriptions. In adjusted analyses, we found that the variable most strongly associated with NEB levothyroxine use was the type of clinician. Primary care clinicians were more likely to prescribe levothyroxine NEB when compared with endocrinologists and other specialties. Primary care clinicians may not receive the most up-to-date information about recommendations for levothyroxine use, as the hypothyroidism guidelines may mostly be known by endocrinologists. This finding highlights the need to broaden the reach, scope, and dissemination of thyroid-related guidelines to affect the clinicians who prescribe levothyroxine most often. Furthermore, in this study, we found that women were more likely to receive more NEB levothyroxine prescriptions than men despite adjustment for symptoms. The reason for this finding is unclear, and additional research should uncover factors associated with prescription patterns in women. Finally, in our sensitivity analysis, we noted that a negative, unchecked, or unavailable TPO antibody titer was associated with more NEB prescription; However, our confidence in this association’s strength is weak due to the high number of patients with unavailable TPO antibody titers. These missing data do not allow us to conclude inferences from our results, and the findings are, therefore, exploratory and require further confirmation.
Our study has several limitations. Our study included patients treated in tertiary care centers, and it is unclear if these results apply to a population seen in secondary or primary care centers, which likely represents most people receiving levothyroxine. In addition, this is a retrospective study limited by selection and ascertainment bias, and the reasons for levothyroxine prescription are based on the available EMR data. Furthermore, the development of a classification framework to determine NEB prescriptions, although guided by guidelines and expert recommendations, required the use of the research team judgment as new and unexpected reasons for levothyroxine prescription emerged during the conduct of the study (for instance, the use of levothyroxine preceded by a single TSH measurement only). During the development of our classification frameworks and the subsequent analyses, we opted for a more conservative approach that favored allocating categories to appropriate or indeterminate, and having the 2 categories merged together when comparing with the NEB category. We acknowledge that alternative classification frameworks could have been used to determine the appropriateness of levothyroxine use; for instance, a recent guideline recommended against levothyroxine use in SCH regardless of symptoms, TSH values, and age (21). Following this guideline, the estimates of NEB levothyroxine prescriptions in our sample would have been much higher. Finally, we must acknowledge the possibility that certain laboratory records, especially those from outside laboratories, may not have been available in a digital format and thus were not accessible for data extraction. While we believe this practice was uncommon among this cohort of patients, it is essential to recognize that as a retrospective chart review, we lacked the means to mitigate this potential limitation, which could potentially lead to the misallocation of cases.
Conclusion
Our analysis suggests there is a considerable proportion of NEB levothyroxine prescriptions and calls for additional research to replicate these findings and explore the perspective of those prescribing and receiving levothyroxine, including other factors that may explain the NEB levothyroxine prescription.
Abbreviations
- EMR
electronic medical record
- FT4
free thyroxine
- IQR
interquartile range
- MCR
Mayo Clinic
- NEB
nonevidence based
- OR
odds ratio
- RAI
radioactive iodine
- SCH
subclinical hypothyroidism
- T4
thyroxine
- TFT
thyroid function test
- TPO
thyroid peroxidase
- TSH
thyrotropin
- UAMS
University of Arkansas for Medical Sciences
- UF
University of Florida
Contributor Information
Ivan Nicolas Ayala, Knowledge and Evaluation Research Unit, Department of Medicine, Mayo Clinic, Rochester, MN 55902, USA.
Cristian Soto Jacome, Knowledge and Evaluation Research Unit, Department of Medicine, Mayo Clinic, Rochester, MN 55902, USA; Division of Endocrinology, Diabetes, Metabolism and Nutrition, Department of Medicine, Mayo Clinic, Rochester, MN 55902, USA.
David Toro-Tobon, Division of Endocrinology, Diabetes, Metabolism and Nutrition, Department of Medicine, Mayo Clinic, Rochester, MN 55902, USA.
Elizabeth Golembiewski, Knowledge and Evaluation Research Unit, Department of Medicine, Mayo Clinic, Rochester, MN 55902, USA.
Andrea Garcia-Bautista, Knowledge and Evaluation Research Unit, Department of Medicine, Mayo Clinic, Rochester, MN 55902, USA.
Jessica Hidalgo, Knowledge and Evaluation Research Unit, Department of Medicine, Mayo Clinic, Rochester, MN 55902, USA.
Sandra Cordova-Madera, Department of Medicine, MetroWest Medical Center, Framingham, MA 01702, USA.
Raghda Al Anbari, Division of Endocrinology, Department of Medicine, University of Florida, Gainesville, FL 32610, USA.
Jessica Sohn R, Division of Endocrinology, Department of Medicine, University of Florida, Gainesville, FL 32610, USA.
Naykky Singh Ospina, Division of Endocrinology, Department of Medicine, University of Florida, Gainesville, FL 32610, USA.
Spyridoula Maraka, Division of Endocrinology and Metabolism, Department of Internal Medicine, University of Arkansas for Medical Sciences, Little Rock, AR 72205, USA; Section of Endocrinology, Medicine Service, Central Arkansas Veterans Healthcare System, Little Rock, AR 72205, USA.
Marina Joseph, Department of Internal Medicine, University of Arkansas for Medical Sciences, Little Rock, AR 72205, USA.
Juan P Brito, Knowledge and Evaluation Research Unit, Department of Medicine, Mayo Clinic, Rochester, MN 55902, USA; Division of Endocrinology, Diabetes, Metabolism and Nutrition, Department of Medicine, Mayo Clinic, Rochester, MN 55902, USA.
Funding
S.M. was supported by the Arkansas Biosciences Institute, the Office of Health Services Research and Development Service of the US Department of Veterans Affairs (under merit review award No. 1I21HX003268-01A1), and the National Center for Advancing Translational Sciences of the National Institutes of Health (under award No. UL1 TR003107). N.S.O. was supported by the National Cancer Institute of the National Institutes of Health (under award No. K08CA248972). The content is solely the responsibility of the authors and does not necessarily represent the official views of the National Institutes of Health, US Department of Veterans Affairs, or the US government.
Author Contributions
S.M., J.P.B., and N.S.O. conceived and designed the study with input from all the coauthors. I.A., A.E.G.B., J.H., S.C.M., R.A.A., J.S.R., and M.J. performed the data collection. J.P.B. conducted the statistical analysis with input from S.M., N.S.O., D.T.T., C.S.J., and E.G. All the coauthors contributed to critically appraising and reviewing the results and drafting the manuscript. All the authors reviewed and agreed on the final version of the article.
Disclosures
S.M. serves as a member of the American Thyroid Association Thyroid & Pregnancy Guidelines Task Force. The other authors have nothing to disclose.
Data Availability
The data sets used and/or analyzed during the current study are available from the corresponding author on reasonable request.
References
- 1. Jonklaas J, Bianco AC, Bauer AJ, et al. Guidelines for the treatment of hypothyroidism: prepared by the American thyroid association task force on thyroid hormone replacement. Thyroid. 2014;24(12):1670‐1751. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2. Fuentes AV, Pineda MD, Venkata KCN. Comprehension of top 200 prescribed drugs in the US as a resource for pharmacy teaching, training and practice. Pharmacy. 2018;6(2):43. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3. Kantor ED, Rehm CD, Haas JS, et al. Trends in prescription drug use among adults in the United States from 1999-2012. JAMA. 2015;314(17):1818‐1831. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4. Ross JS, Rohde S, Sangaralingham L, et al. Generic and brand-name thyroid hormone drug use among commercially insured and medicare beneficiaries, 2007 through 2016. J Clin Endocrinol Metab. 2019;104(6):2305‐2314. [DOI] [PubMed] [Google Scholar]
- 5. Wyne KL, Nair L, Schneiderman CP, et al. Hypothyroidism prevalence in the United States: A retrospective study combining national health and nutrition examination survey and claims data, 2009-2019. J Endocr Soc. 2022;7(1):bvac172. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6. Hollowell JG, Staehling NW, Flanders WD, et al. Serum TSH, T4, and thyroid antibodies in the United States population (1988 to 1994): National Health and Nutrition Examination Survey (NHANES III). J Clin Endocrinol Metab. 2002;87(2):489‐499. [DOI] [PubMed] [Google Scholar]
- 7. Brito JP, Al Nofal A, Montori VM, et al. The impact of subclinical disease and mechanism of detection on the rise in thyroid cancer incidence: A population-based study in olmsted county, Minnesota during 1935 through 2012. Thyroid. 2015;25(9):999‐1007. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8. Kim YA, Park YJ. Prevalence and risk factors of subclinical thyroid disease. Endocrinol Metab (Seoul). 2014;29(1):20‐29. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 9. Hennessey JV, Espaillat R. Subclinical hypothyroidism: a historical view and shifting prevalence. Int J Clin Pract. 2015;69(7):771‐782. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10. Taylor PN, Iqbal A, Minassian C, et al. Falling threshold for treatment of borderline elevated thyrotropin levels-balancing benefits and risks: evidence from a large community-based study. JAMA Intern Med. 2014;174(1):32‐39. [DOI] [PubMed] [Google Scholar]
- 11. Medici BB, Nygaard B, la Cour JL, et al. Changes in prescription routines for treating hypothyroidism between 2001 and 2015: an observational study of 929,684 primary care patients in Copenhagen. Thyroid. 2019;29(7):910‐919. [DOI] [PubMed] [Google Scholar]
- 12. Calissendorff J, Falhammar H. To treat or not to treat subclinical hypothyroidism, what is the evidence? Medicina (Kaunas). 2020;56(1):40. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13. Razvi S, Weaver JU, Butler TJ, et al. Levothyroxine treatment of subclinical hypothyroidism, fatal and nonfatal cardiovascular events, and mortality. Arch Intern Med. 2012;172(10):811‐817. [DOI] [PubMed] [Google Scholar]
- 14. Somwaru LL, Arnold AM, Joshi N, et al. High frequency of and factors associated with thyroid hormone over-replacement and under-replacement in men and women aged 65 and over. J Clin Endocrinol Metab. 2009;94(4):1342‐1345. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 15. Calsolaro V, Niccolai F, Pasqualetti G, et al. Overt and subclinical hypothyroidism in the elderly: when to treat? Front Endocrinol (Lausanne). 2019;10:177. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16. Brito JP, Ross JS, El Kawkgi OM, et al. Levothyroxine use in the United States, 2008-2018. JAMA Intern Med. 2021;181(10):1402‐1405. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17. Hepp Z, Lage MJ, Espaillat R, et al. The direct and indirect economic burden of hypothyroidism in the United States: a retrospective claims database study. J Med Econ. 2021;24(1):440‐446. [DOI] [PubMed] [Google Scholar]
- 18. Ernst FR, Barr P, Elmor R, et al. The economic impact of levothyroxine dose adjustments: the CONTROL HE study. Clin Drug Investig. 2017;37(1):71‐83. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19. Ayala IN, Soto Jacome C, Toro-Tobon D. Appropriateness of Levothyroxine Prescription: A Multicenter Retrospective Study. Supplemental table 1.; 2023. Available at: 10.6084/m9.figshare.24033891. [DOI] [PMC free article] [PubMed]
- 20. Carlé A, Pedersen IB, Knudsen N, et al. Hypothyroid symptoms and the likelihood of overt thyroid failure: a population-based case-control study. Eur J Endocrinol. 2014;171(5):593‐602. [DOI] [PubMed] [Google Scholar]
- 21. Bekkering GE, Agoritsas T, Lytvyn L, et al. Thyroid hormones treatment for subclinical hypothyroidism: a clinical practice guideline. BMJ. 2019;365:l2006. [DOI] [PubMed] [Google Scholar]
- 22. Garber JR, Cobin RH, Gharib H, et al. Clinical practice guidelines for hypothyroidism in adults: cosponsored by the American Association of Clinical Endocrinologists and the American Thyroid Association. Endocr Pract. 2012;18(6):988‐1028. [DOI] [PubMed] [Google Scholar]
- 23. Biondi B, Cappola AR. Subclinical hypothyroidism in older individuals. Lancet Diabetes Endocrinol. 2022;10(2):129‐141. [DOI] [PubMed] [Google Scholar]
- 24. Jonklaas J. Optimal thyroid hormone replacement. Endocr Rev. 2022;43(2):366‐404. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25. Meyerovitch J, Rotman-Pikielny P, Sherf M, et al. Serum thyrotropin measurements in the community: five-year follow-up in a large network of primary care physicians. Arch Intern Med. 2007;167(14):1533‐1538. [DOI] [PubMed] [Google Scholar]
- 26. Rodriguez-Gutierrez R, Maraka S, Ospina NS, et al. Levothyroxine overuse: time for an about face? Lancet Diabetes Endocrinol. 2017;5(4):246‐248. [DOI] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
The data sets used and/or analyzed during the current study are available from the corresponding author on reasonable request.
