Comparative Effectiveness of Generic vs Brand-Name Levothyroxine in Achieving Normal Thyrotropin Levels

This cohort study compares the effectiveness of generic vs brand-name levothyroxine in achieving and maintaining normal thyrotropin levels among new users.


Introduction
Approximately 20 million Americans have overt or subclinical hypothyroidism and may require thyroid hormone replacement with levothyroxine preparations. 1,2 In the United States, several generic and brand-name levothyroxine formulations are available for use. Generic levothyroxine preparations are less expensive and have been rated as bioequivalent by the US Food and Drug Administration (FDA) to their brand-name reference-listed drugs. 3 However, generic levothyroxine has been less widely prescribed than other generic pharmaceutical products. 4 Endocrinologists, in particular, are more than 3 times as likely to prescribe brand-name vs generic levothyroxine as initial therapy compared with general practitioners. 5 In a survey of 880 members of several endocrine societies, half of the respondents indicated that they preferred brand-name to generic levothyroxine. 6 Despite the entrenched preference for brand-name levothyroxine, it is still unclear whether brand-name levothyroxine is more effective in achieving normal and stable thyrotropin levels compared with generic levothyroxine.
To assess the comparative effectiveness of generic vs brand-name levothyroxine in patients with hypothyroidism, we used a national administrative claims database to identify patients who newly initiated use of levothyroxine. Our goal was to examine the association of generic vs brandname levothyroxine formulation use with subsequent thyrotropin levels, the widely accepted laboratory test of thyroid status. A better understanding of the comparative effectiveness of generic compared with brand-name levothyroxine preparations may help physicians optimize their decisions about levothyroxine treatment. In turn, the findings may affect the management of hypothyroidism in millions of patients in the United States and the costs of levothyroxine use.

Study Design and Data Source
This cohort study adhered to the International Society for Pharmacoeconomics and Outcomes Research (ISPOR) reporting guideline for defining, reporting, and interpreting nonrandomized studies of treatment effects using secondary data sources. 7 We conducted a retrospective analysis of deidentified administrative claims data linked with laboratory results from a large database, OptumLabs Data Warehouse, which includes privately insured patients and Medicare Advantage enrollees throughout the United States. 8 The database contains longitudinal health information on enrollees and patients, representing a diverse mix of ages, ethnicities, and geographic regions across the United States. 9 The health plan provides comprehensive full insurance coverage for physician, hospital, and prescription drug services and drug doses. Pharmacy claims include information on medications dispensed, including amount and dates of prescriptions. Laboratory data, available for a subset of the cohort on the basis of data sharing agreements, include test names, logical observation identifier names, and test results. Study data were accessed using techniques compliant with the Health Insurance Portability and Accountability Act of 1996. Because this study involved analysis of preexisting, deidentified data, the Mayo Clinic institutional review board declared it exempt from board approval and informed consent.

Study Population
We identified 2 cohorts. Cohort 1 consisted of adult patients (aged Ն18 years) who newly filled prescriptions for either generic or brand-name levothyroxine preparations from January 1, 2008, to October 1, 2017. Patients were required to have at least 365 days of continuous medical and pharmacy benefits coverage before treatment initiation (index date) and 90 days after initiation. We applied a number of cohort exclusions. We excluded patients who filled any thyroid preparation within the 365 days before the index data (ie, only new users were included). We also excluded patients who switched from generic to brand-name or from brand-name to generic levothyroxine within 90 days (3 months) of treatment initiation, and patients who used other forms of thyroid hormone replacement therapy at any point during follow-up, including thyroid extracts or triiodothyronine therapy (liothyronine, thyroid desiccated/extracts, Cytomel [Pfizer Inc], Armour Thyroid [Allergan], and Nature-Thyroid [RLC Labs]). We also excluded patients who were not adherent to levothyroxine therapy. Adherence was measured by the proportion of days covered at 90 days, and patients were considered adherent if the proportion of days covered was greater than 80. 10,11 We also excluded patients taking any levothyroxine dose exceeding 200 μg/d owing to difficulty assessing the full prescribed dose (maximum dose in a single tablet is 200 μg) and because doses greater than 200 μg might suggest levothyroxine malabsorption issues. To assess thyroid status, we further limited our study to patients who had linked thyrotropin results at both baseline (within 90 days before drug initiation) and follow-up (within 6-12 weeks after initiation). Only patients with baseline thyrotropin values ranging from 4.5 to 19.9 mIU/L were included because those with thyrotropin values greater than 20.0 mIU/L may take longer than 3 months to achieve the normal range of levels. Finally, we excluded patients with conditions that require specific thyrotropin targets that may fall outside of the reference range, such as pregnancy, thyroid cancer, and hypopituitarism, or patients exposed to medications known to affect thyroid hormone levels during the baseline and follow-up periods (a list of medications and International Classification of Diseases codes for excluded conditions is found in eTable 1 in the Supplement). 12 Enrollment of the study population is depicted in Figure 1.
Cohort 2 was a subset of cohort 1. We included only patients from cohort 1 who had a normal follow-up thyrotropin level within 3 months and a subsequent follow-up thyrotropin level measured 6 to 12 weeks after the thyrotropin test with normal findings. Given that these patients had a previous normal thyrotropin value, the individuals were likely to be receiving the same dose of levothyroxine. We also excluded patients who switched from generic to brand-name or from brandname to generic levothyroxine within 180 days (6 months) of treatment initiation.

Exposures
We characterized whether the index pharmacy fill was for a generic or a brand-name thyroid hormone drug. We used First Databank to categorize each fill as brand-name or generic. First Databank categorizes pharmacy products as generic if they are sold under a generic pharmacy label, which includes authorized generic products. The brand-name products included Synthroid (AbbVie;

Study Outcomes
We examined the effectiveness of levothyroxine based on attained thyrotropin levels measured in the outpatient setting. For assessment of effectiveness, we examined the proportion of individuals who initiated use of either generic or brand-name levothyroxine and who attained a normal thyrotropin level within 3 months, clinically meaningful abnormal thyrotropin level within 3 months, and stable thyrotropin level(s) within 3 months after thyrotropin fell into the normal range (Figure 2). We defined a normal thyrotropin level as ranging from 0.3 to 4.4 mIU/L, a range that includes as much as 95% of the healthy population. 13 Given that thyrotropin values are only reliable as treatment markers at least 6 weeks after initiation of therapy, 12 we used only the first available thyrotropin value from 6 weeks to 3 months after initiation of therapy to assess treatment effectiveness. Using thyrotropin values beyond 3 months of therapy may not be an accurate surrogate for effectiveness because the values might reflect changes in doses rather than the comparative effectiveness between brand-name and generic levothyroxine. Moreover, levothyroxine dose adjustments may not be reliably obtained retrospectively from the OptumLabs Data Warehouse, because physician-directed medication adjustments may involve pill splitting and alternate-day dosing. Abnormal thyrotropin values were categorized as clinically meaningful when thyrotropin levels were suppressed (<0.1 mIU/L) or elevated (>10 mIU/L). We defined stable thyrotropin levels as a normal thyrotropin level within 3 months after thyrotropin levels fell into the normal range. When there were multiple thyrotropin level measurements during this period of time, we used the one closest to the first thyrotropin measurement ( Figure 2).

Statistical Analysis
Data were analyzed from August 13, 2018, to October 25, 2019. We used propensity scores to minimize confounding. Patients receiving generic levothyroxine were matched 1:1 with patients receiving brand-name levothyroxine using nearest-neighbor matching with a caliper of 0.2. We developed a propensity score model with the binary outcome of initiating treatment with brandname levothyroxine. The model included demographics, comorbidities, and baseline thyrotropin values shown in Table 1 (cohort 1) and Table 2 (cohort 2). Variables within these models were selected based on clinical relevance and evidence from prior studies. For each cohort, we evaluated the balance among the treatment groups by comparing standardized mean differences of baseline covariates between the groups. A baseline characteristic was considered balanced if the maximum standardized mean difference was less than 10%. Using this matched cohort, we then tested for differences between patients treated with generic vs brand-name levothyroxine using χ 2 tests for   categorical outcomes. We used SAS software, version 9.4 (SAS Institute, Inc), for all statistical analyses. We considered a 2-sided P < .05 to be statistically significant.

JAMA Network Open | Diabetes and Endocrinology
The comparative effectiveness of generic and brand-name levothyroxine among new users could be influenced by the presence of endogenous production of thyroid hormone (patients with a functional thyroid gland). Therefore, we examined whether outcomes differed among patients with and without endogenous thyroid hormone production. We categorized patients as not having endogenous thyroid production if they had a history of total thyroidectomy or if they received thyroid hormone doses greater than 100 μg/d. Because the mean full replacement dose of levothyroxine in adults is approximately 1.6 μg/kg per day, a dose of greater than 100 μg/d is likely to reflect full replacement.

Sensitivity Analysis
Propensity score matching shrinks the sample to produce groups of the same size. Therefore, we used an alternative propensity score method, inverse probability of treatment weighting (IPTW), to preserve the data lost during matching to confirm our findings. Inverse probability of treatment weighting is an alternative method of incorporating propensity scores to isolate treatment effects used in observational studies. Contrary to propensity score matching, in which the resulting conditional model for the outcomes is fitted though matching, IPTW uses weighting. 14,15 95% CI, 3.4%-5.0%) and 88 (3.9%; 95% CI, 3.1%-4.7%) had a clinically meaningful abnormal thyrotropin value of less than 0.1 mIU/L or greater than 10 mIU/L in the generic and brand-name cohorts, respectively (P = .65).

Subgroup Analyses
Within cohort 1, only 30 (0.7%) of patients taking levothyroxine had a history of total thyroidectomy in the year before the index date; thus, a subgroup analysis on this subset of patients was not

JAMA Network Open | Diabetes and Endocrinology
Generic vs Brand-Name Levothyroxine for Hypothyroidism possible. An analysis of the subset of patients taking thyroid hormone doses greater than 100 μg/d (n = 117) showed that normal thyrotropin values were attained in 25 of 52 (48.1%) vs 42 of 65 (64.6%) of generic vs brand-name levothyroxine users, respectively (P = .07). Within cohort 2, a subgroup analysis was not possible because of the low number of patients requiring doses of at least 100 μg/d.

Sensitivity Analyses
Results using IPTW were consistent with the results of propensity score-matching analysis. Using the entire population with IPTW analyses (eTable 2 in the Supplement), the proportion of patients who achieved a normal thyrotropin value within 3 months of filling their first levothyroxine prescription was similar for patients who received generic vs brand-name levothyroxine (

Discussion
For adults with mild forms of thyroid dysfunction, defined as thyrotropin levels less than 20 mIU/L before initiation of treatment, consistent use of generic or brand-name levothyroxine formulations is associated with similar rates of achieving normal and stable thyrotropin levels, that is, laboratorybased outcomes that are considered clinically important. During the last decade, one-quarter of all levothyroxine prescriptions filled were still for brand-name levothyroxine products. 5  One driver of the continued preference for brand-name levothyroxine may be physicians' skepticism about the equivalent effectiveness of generic compared with brand-name levothyroxine.
After the FDA approved the use of several generic levothyroxine products, 3 endocrine professional societies released a joint position statement in 2004 raising questions about the FDA's method for determining bioequivalence. 16 The position statement specifically pointed out that the FDA's method does not consider the role of endogenously produced hormone and does not rely on the thyrotropin level. They further implied that these methodological flaws, along with the fact that levothyroxine has a relatively narrow therapeutic index, may lead to a significant difference in bioequivalence between generic and brand-name levothyroxine. In turn, this difference might affect thyrotropin levels and associated clinical outcomes. 16 In 2007, the FDA, in response to concerns expressed about levothyroxine products by health care professionals and patients, tightened the potency specifications for levothyroxine to meet a 95% to 105% potency specification until their expiration date. 17 Despite these changes, the concerns initially raised in 2004 by endocrine societies have continued to affect endocrine guideline recommendations and practice. For instance, the 2014 American Thyroid Association guideline specifically recommends avoidance of switches between levothyroxine products. 12 To maintain the same preparation, the guideline recommends that physicians either prescribe the same identifiable formulation of generic levothyroxine or a specific brand-name levothyroxine. The latter strategy is logistically easier, as it avoids substitution for another levothyroxine preparation at the level of the pharmacy, which is common when a generic prescription is used and can occur without physician approval.
Our results should reassure physicians and patients that generic levothyroxine as initial therapy for mild thyroid dysfunction is as effective as brand-name levothyroxine. Moreover, prescribing generic compared with brand-name levothyroxine may offer substantial financial benefits to patients and to the health care system. For instance, information available from the OptumLabs Data Warehouse shows that in 2017 the mean 30-day out-of-pocket cost was $6.20 for generic levothyroxine and $28.65 for brand-name levothyroxine across commercial, Medicare Advantage, and Medicare Part D insurance. However, our results do not address the potential effect of switches, either purposeful or passive, between formulations. These switches may occur between brandname and generic products or between products (either brand-name or generic) made by different levothyroxine manufacturers. Thus, our results do not directly address the issue of interchangeability of levothyroxine formulations. Additional research should compare the effectiveness of levothyroxine products among patients who switch therapy during the course of their treatment.
Our study included patients with baseline thyrotropin values ranging from 4.5 to 19.9 mIU/L, which reflect mild thyroid dysfunction. The most commonly prescribed levothyroxine dose in our cohort was no greater than 50 μg/d, which is unlikely to be a full thyroid hormone replacement dose.
Thus, patients in our study likely represent a cohort with some endogenous thyroid hormone production. However, our results are likely generalizable to most patients in the United States who initiate levothyroxine therapy; most patients have partial thyroid dysfunction (eg, subclinical hypothyroidism) and may be able to maintain some endogenous thyroid hormone production. 18 Patients with endogenous levothyroxine production might retain some ability to regulate thyroid hormone levels despite thyroid hormone replacement therapy. As a result, patients with residual endogenous levothyroxine production may have different thyrotropin values than patients without any endogenous levothyroxine receiving the same levothyroxine dose or formulation. 8,9 In our analysis, based on a small subset of patients, the proportion of patients achieving normal thyrotropin values within 3 months was higher among patients who initiated brand vs generic levothyroxine when doses greater than 100 μg/d were prescribed, but these findings were not statistically significant. In a small crossover study that randomized children to 8 weeks of brandname or generic levothyroxine therapy, 19 serum thyrotropin concentrations were significantly lower when patients used brand-name levothyroxine compared with the same AB-rated generic levothyroxine (Sandoz AG) but only in the subgroup of 20 patients with acquired hypothyroidism (ie, in those with no endogenous levothyroxine production). Clearly, more research is needed to understand the comparative effectiveness and safety of brand-name vs generic levothyroxine in patients without remaining endogenous thyroid function.

Limitations
Our study has several limitations that need to be considered. Given the observational nature of our study, residual confounding may still exist. As a result, unmeasured factors may have influenced our findings, and patients filling generic levothyroxine may be different from those filling brand levothyroxine prescriptions; these differences may have influenced the study outcomes. For instance, we could not account for all the factors that might affect gastrointestinal tract absorption of levothyroxine, such as concomitant intake of over-the-counter calcium preparations. 20 Our primary measure for the assessment of baseline thyroid hormone level status and for the laboratory outcomes was based on a thyrotropin level and not free thyroxine. We preferred this approach, given that thyrotropin is the most sensitive test to assess response to therapy. 12 We could not reliably ascertain levothyroxine dose adjustments from the OptumLabs Data Warehouse because physiciandirected medication adjustments may involve pill cutting and alternate-day dosing. Finally, our study sample includes mostly patients with commercial health insurance; the applicability of study findings to underinsured populations is unclear.