Concomitant use of levothyroxine and interacting medications in US ambulatory care visits

Abstract

Background

Levothyroxine is the 3rd most commonly prescribed medication in the United States. It is a narrow therapeutic index medication and thus can be impacted by drug-drug interactions, which are primarily available over-the-counter (OTC). The prevalence and associated factors with concomitant interacting drugs with levothyroxine is limited since OTC products are not routinely captured in many drug databases.

Objective(s)

This study aimed to characterize the concomitant use of levothyroxine with interacting drugs at ambulatory care visits in the United States.

Design

A cross-sectional analysis of the National Ambulatory Medical Care Survey (NAMCS) from 2006 to 2018 was completed.

Setting and participants

Ambulatory care visits in the United States involving adult patients with a levothyroxine prescription were included in the analysis.

Outcome measures

The primary outcome was initiation or continuation of a selected concomitant interacting drug which impacts levothyroxine absorption (e.g., proton pump inhibitor (PPIs)) in a patient visit in conjunction with levothyroxine.

Results

The authors analyzed 372,942,000 visits (weighted from a sample of 14,880) with a reported levothyroxine prescription. Concomitant use of interacting drugs with levothyroxine occurred in 24.4% of visits in which 80% of interacting drugs were PPIs. Age 35–49 (aOR 1.59), 50–64 (aOR 2.27), and ≥65 (aOR 2.87) compared to 18–34, female (aOR 1.37) vs. males, and visits in 2014 or later (aOR 1.27) vs. 2006–2009 were associated with increased odds of concomitant interacting drug use in multivariable analysis.

Conclusion

At ambulatory care visits between 2006 and 2018, concomitant use of levothyroxine and interacting drugs impacted one-quarter of patient visits. Increased age, females, and visits later in the study period were associated with increased odds for concomitant interacting drugs. Additional work is needed to identify downstream consequences of concomitant use.

INTRODUCTION

Levothyroxine (LT4), a synthetic thyroid hormone, is in the top three of most commonly prescribed medication in the United States.¹ Over 100 million prescriptions are dispensed each year, equating to an estimated 11.4% of the adult population in 2018.^1,2 LT4 is an effective treatment for patients with overt hypothyroidism; however, it has a narrow therapeutic range, suggesting that minor changes in serum levels may result in significant changes in thyroid status, and have a profound impact on treatment outcomes (e.g. symptoms) as well as adverse effects (e.g. bone health).³ Although patients commonly achieve steady levothyroxine doses, compliance with levothyroxine treatment, aging, weight changes, and pregnancy can lead to dose adjustments.^4,5

Additionally, concomitant food intake, timing of administration, or co-administration of LT4 with other medications can also be associated with variations in thyroid hormone levels.⁶ A patient’s LT4 requirement can be affected by many agents through alterations in any of the following: LT4 absorption, thyroid hormone synthesis or release, transport, metabolism, or thyroid stimulating hormone (TSH) secretion.⁵ Depending on the interfering medication used, the patient may present with symptoms of hypothyroidism (e.g., fatigue, weight gain, cold intolerance, constipation), requiring higher LT4 doses, or hyperthyroidism (e.g., tachycardia, increased appetite, unintentional weight loss, irritability), requiring lower LT4 doses.^7,8 In fact, concomitant use of LT4 with proton-pump inhibitors (PPIs), bile acid sequestrants, minerals (calcium, aluminum, iron), binding agents (lanthanum, sevelamer) or select antibiotics (rifampin, ciprofloxacin) are considered to be clinically significant.⁶ Inconsistent utilization, or abrupt discontinuation, of these agents may result in variable LT4 serum concentration levels thus complicating the management of hypothyroidism.

Previous evaluations of concomitant use of medications affecting levothyroxine bioavailability have shown that numerous prescription agents, namely PPIs, decrease the absorption of LT4 and consequently lead to an increased use (i.e. higher dose, more prescriptions dispensed) of LT4 to compensate.^6,9 Although prescription claims databases are ideal for evaluating drug-drug interactions, they may be limited in evaluating LT4 interactions as a large portion of interacting drugs that can interfere with LT4 can be purchased over-the-counter (OTC) and will not be typically captured (e.g. calcium carbonate, omeprazole).¹¹ Data sources that can objectively capture complete medication therapy information gathered during patient visits including OTC products are necessary to answer this research question. Therefore, we aim to fill this gap by using the National Ambulatory Medical Care Survey (NAMCS) database to evaluate utilization trends in medications that interact with LT4 and associated predictors in US ambulatory care visits.

METHODS

Data source

We conducted a cross-sectional study of the NAMCS database, from 2006 to 2018 (2017 was excluded as data are not available). The NAMCS is a national sample survey conducted annually by the National Center for Health Statistics (NCHS) under Centers for Disease Control and Prevention (CDC), reflecting data on non-federally employed office-based physician visits.¹² Data is collected from physicians involved in direct patient care and excludes those in the specialties of radiology, pathology, and anesthesiology. Physicians are randomly selected to participate and confirmed as eligible by a U.S. Census Bureau field representative. Once consent is obtained, information collected from physician visits includes patient demographics, conditions treated, a medication profile, and services provided.

The sampling design applied by the NAMCS involves multiple stages including geographically defined area (primary sampling units, PSU), physicians within PSUs, and patient visits within practices. PSUs cover all 50 states, including the District of Columbia, and are selected with a probability proportional to their size. Physician sampling is performed through the utilization of master files maintained by the American Medical Association (AMA) and American Osteopathic Association (AOA), for which inclusion criteria are applied (i.e., office-based, involvement in direct patient care, specialty, non-federally employed). Ultimately, a “patient visit weight” is generated that allows the user to obtain national estimates of all ambulatory care visits in a given time period.¹³

Study Population

All visits, from 2006 to 2018 (2017 unavailable), in patients aged ≥18 in which levothyroxine was initiated or continued were included in this study. Due to an increase in the maximum number of drug mentions per single patient visit (8 from 2006– 2011, 10 from 2012–2013, and 30 starting in 2014), levothyroxine use was only assessed among the first 8 drugs (inclusive of OTC medications) documented to minimize the influence of differential capture.¹⁴

Outcomes

The primary outcome was the occurrence of concomitant levothyroxine use with absorption-impacting interacting medications. To ensure consistency over the study period, only the first 8 drugs documented were assessed each year.

We used Lexicomp Online classification to identify medications with interactions considered significant with “good” or “very good” strength of evidence.¹⁵ Interacting medications considered “poor” and “controversial” were not considered. These interacting medications included minerals (calcium acetate/carbonate/citrate, aluminum hydroxide, magnesium hydroxide, iron sulfate, chromium), PPIs (dexlansoprazole, esomeprazole, lansoprazole, omeprazole, pantoprazole), binding agents (lanthanum, sevelamer), antibiotics (ciprofloxacin, rifampin), antiepileptics (carbamazepine, phenobarbital, phenytoin) cholestyramine, sucralfate, orlistat, and raloxifene. Drug entry codes used in our analyses are referenced in Appendix Table 1.

Covariates

The covariates assessed were patient age, race, sex, and diagnosis of the following comorbidities consistently captured in NAMCS: arthritis, asthma or chronic obstructive pulmonary disorder (COPD), cancer, chronic renal failure (CRF), depression, diabetes, hyperlipidemia, hypertension, obesity, and osteoporosis (Appendix Table 2). Other covariates assessed included payment source, geographic region, and whether the patient was new or established at the practice.

Statistical Analysis

To produce national estimates from sample data, each record in the NAMCS micro-data file is assigned an inflation factor, of which data analyzed were weighted to. Survey analysis procedures, including SAS PROC SURVEYFREQ and PROC SURVEYLOGISTIC, were employed to account for the complex survey design. Multivariable logistic regression was utilized, while adjusting for the selected covariates, to assess the prevalence of levothyroxine use with a known interacting medication in a single ambulatory care office visit.

This study was considered exempt by the University of Florida Institutional Review Board. All statistical analyses were performed using SAS version 9.4 (SAS Institute, Inc, Cary, NC).

RESULTS

Within the included years, levothyroxine was initiated or continued in an estimated 372,942,000 office-based weighted visits. A majority of the patients were white (79.5%), female (78.1%), aged ≥65 years (51.5%), and were previously established at the office (88.9%; Appendix Table 3). The most common comorbidities amongst the study population were hypertension (48.9%) and hyperlipidemia (36.4%).

Among visits in which levothyroxine was initiated or continued, 91,153,000 weighted visits (24.4%) included the concomitant use of a known interacting medication. When assessed on a yearly basis, the proportion of visits with concomitant use of a known interacting medication was fairly stable between 2006 and 2016 (ranging from 21.2% to 25.8%) with an increase in 2018 (32.4%, Figure 1.) PPIs accounted for 79.6% (n=72,585,000, Appendix Table 4) of the interacting medications and the proportion of visits with concomitant use of PPIs appeared to be proportional to the concomitant use of known interacting medications over the study period (Appendix Figure 1). The second most known interacting medications were minerals (e.g., aluminum hydroxide, calcium acetate, calcium carbonate, calcium citrate, chromium, iron sulfate, magnesium hydroxide), which accounted for 13.8% of visits with an interacting medication (n=12,616,000). The remaining interacting medications are described in Appendix Table 4.

Figure 1:

Unadjusted annual utilization rates of drug-interacting medications with levothyroxine at all visits 2006–2016, 2018. Error bars indicate 95% confidence intervals.

In the multivariable logistic regression, levothyroxine users who concomitantly used drug-interacting medications differed from those who did not by age, sex, and year of visit (Table 1). Levothyroxine users aged 35–49 (adjusted odds ratio (aOR) 1.59 [95% CI 1.03–2.44]), aged 50–64 (aOR 2.27 [95% CI 1.48–3.47]), and aged ≥65 (aOR 2.87 [95% CI 1.86–4.43]) were more likely to concomitantly use a drug-interacting medication compared with users aged 18–34 years. Additionally, female LT4 users were more likely to concomitantly use a drug-interacting medication when compared to male LT4 users (aOR 1.37 [95% CI 1.13–1.65]). Finally, visits during 2014–2016 and 2018 were more likely to concomitantly use a drug-interacting medication compared to visits during 2006–2009 (aOR 1.27 [95% CI 1.05–1.53]). No other covariates analyzed were significant.

Table 1:

Factors associated with visits in levothyroxine users with drug-drug interactions

	Levothyroxine Users Alone		Levothyroxine Users with DDI		Unadjusted Analysis OR (95%CI)*	Multivariable Analysis OR (95%CI)*
	Unweighted (n=11,374) n(%)	Weighted* (n=281,789,000) n(%)	Unweighted (n=3,506) n(%)	Weighted* (n=91,153,000) n(%)
Age 18–34	895 (7.9)	17,517,000 (6.2)	121 (3.5)	2,161,000 (2.3)	-	-
35–49	2,215 (19.5)	45,184,960 (16.0)	461 (13.1)	8,972,000 (9.8)	1.61 (1.05–2.48)	1.59 (1.03–2.44)
50–64	3,708 (32.6)	82,761,000 (29.4)	1,140 (32.5)	24,372,000 (26.7)	2.39 (1.57–3.64)	2.27 (1.48–3.47)
≥65	4,556 (40.1)	136,325,000(48.4)	1,784 (50.9)	55,648,000 (61.0)	3.31 (2.24–4.89)	2.87 (1.86–4.43)
Race White	8,603 (75.6)	225,752,000 (80.1)	2,762 (78.8)	74,765,000 (82.0)	-	-
Black	819 (7.2)	18,178,000 (6.5)	223 (6.4)	5,363,000 (5.9)	0.89(0.59–1.35)	0.92 (0.63–1.34)
Hispanic	1,520 (13.4)	25,029,000 (8.0)	416 (11.9)	6,833,000 (7.5)	0.82 (0.63–1.08)	0.90 (0.69–1.17)
Other	432 (3.8)	12,830,000 (4.6	105 (3.0)	4,191,000 (4.6)	0.99 (0.66–1.47)	1.08 (0.71–1.64)
Sex Male	2,481 (21.8)	64,404,000 (22.9)	722 (20.6)	17,403,000 (19.0)	-	-
Female	8,893 (78.2)	217,385,000 (77.1)	2,784 (79.4)	73,750,000 (80.9)	1.26 (1.04–1.51)	1.37 (1.13–1.65)
Year 2006–2009	3,517 (30.9)	119,473,000 (42.3)	1,037 (29.6)	34,550,000 (37.9)	-	-
2010–2013	4,366 (38.4)	70,250,000 (24.9)	1,336 (38.1)	21,776,000 (23.9)	1.07 (0.90–1.27)	1.09 (0.92–1.29)
2014–2016, 2018	3,491 (30.7)	92,067,000 (32.7)	1,133 (32.3)	34,826,000 (38.2)	1.31 (1.09–1.57)	1.27 (1.05–1.53)
Comorbidities Arthritis	114 (1.0)	3,075,000 (1.1)	48 (1.4)	1,605,000 (1.8)	1.63 (0.95–2.77)	1.38 (0.81–2.33)
Asthma/COPD	1,292 (11.4)	31,567,000 (11.2)	462 (13.2)	13,162,000 (14.4)	1.34 (1.07–1.68)	1.23 (0.98–1.54)
Cancer	973 (8.6)	29,513,000 (10.5)	376 (10.7)	12,292,000 (13.5)	1.33 (1.07–1.65)	1.19 (0.95–1.49)
CRF	164 (1.4)	6,601,000 (2.3)	73 (2.1)	3,420,000 (3.8)	1.63 (1.06–2.50)	1.16 (0.73–1.83)
Diabetes	2,346 (20.6)	56,489,000 (20.0)	766 (21.8)	19,886,000 (21.8)	1.11 (0.93-1.33)	0.97 (0.81–1.15)
Depression	1,907 (16.8)	40,689,000 (14.4)	655 (18.7)	13,686,000 (15.0)	1.11 (0.93–1.33)	1.06 (0.86–1.31)
Hyperlipidemia	3,780 (33.2)	98,897,000 (35.1)	1,329 (37.9)	36,876,000 (40.5)	1.05 (0.85–1.29)	1.08 (0.91–1.28)
Hypertension	5,121 (45.0)	133,082,000 (47.2)	1,845 (52.6)	49,348,000 (54.1)	1.26 (1.06–1.49)	1.07 (0.91–1.26)
Obesity	1,630 (14.3)	32,741,000 (11.6)	526 (15.0)	12,428,000 (13.6)	1.32 (1.14–1.52)	1.29 (1.00–1.66)
Osteoporosis	485 (4.3)	18,620,000 (6.6)	245 (7.0)	7,189,000 (7.9)	1.21 (0.91–1.61)	0.94 (0.70–1.26)
Payer Private	3,812 (33.5)	121,353,000 (43.1)	988 (28.2)	30,433,000 (33.4)	-	-
Medicare	4,268 (37.5)	123,101,000 (43.7)	1,724 (49.2)	49,494,000 (54.3)	1.60 (1.36–1.89)	1.16 (0.92–1.46)
Medicaid, CHIP	1,397 (12.3)	14,733,000 (5.2)	401 (11.4)	3,835,000(4.2)	1.04 (0.74–1.45)	1.04 (0.73–1.48)
Worker’s comp	21 (0.2)	594,000 (0.2)	6 (0.2)	72,000 (<0.1)	0.49 (0.13–1.88)	0.66 (0.18–2.45)
Self-pay	905 (8.0)	7,197,000 (2.6)	161 (4.6)	1,659,000 (1.8)	0.92 (0.56–1.50)	0.90 (0.55–1.47)
Other	971 (8.5)	14,811,000 (5.3)	226 (6.4)	5,659,000 (6.2)	1.52 (1.12–2.08)	1.31 (0.98–1.75)
Patient New	1,342 (11.8)	31,707,000 (11.3)	383 (10.9)	9,837,000 (10.8)	-	-
Established	10,032 (88.2)	250,082,000 (88.7)	3,123 (89.1)	81,316,000(89.2)	1.05 (0.78–1.40)	1.00 (0.76–1.31)

Abbreviations: COPD= chronic obstructive pulmonary disorder, CRF= chronic renal failure, CHIP= Children’s Health Insurance Program, DDI= drug-drug interaction

^*Patient visit weight applied using NAMCS sampling design

DISCUSSION

Co-utilization of LT4 and interacting medications in a single office visit ranged from a minimum of 21.18% in 2012 to a maximum of 32.38% in 2018. Of the interacting medication classes, PPIs accounted for an overwhelming majority and appear to drive the trend of unadjusted utilization rates with levothyroxine.

A similar retrospective cohort study utilizing claims data was conducted by Livecchi et al in patients ≥ 65 years-old (n=538,137) prescribed thyroid hormone therapy between 2004 and 2017 at the Veterans Health System.¹⁶ Through use of the Veterans Health Administration (VHA) Corporate Data Warehouse (CDW), it was determined that 31.4% of patients were on at least one interacting medication while also on LT4.¹⁶ The interacting medications in this study (prednisone, prednisolone, amiodarone, phenytoin, carbamazepine, phenobarbital, tamoxifen) were selected due to their potential effect on thyroid hormone transport and metabolism.¹⁶ Although the rate of co-utilization reported was similar to our study, the extent of comparison is limited due to the difference in cohort age (i.e., ≥ 18 years-old vs ≥ 65 years-old) and inability of the VHA CDW to capture OTC medication use. Furthermore, our most commonly utilized interacting medications, PPIs and minerals, were not evaluated in this study.

Our findings are complemented by an observational study in Italian general practice conducted by Trifiro et al, in which patients with hypothyroidism (n=5,426) on LT4 were analyzed for variation in thyroid-stimulating hormone (TSH) level, number of levothyroxine prescriptions, and dose in the presence of potential interacting medications (PPIs, minerals, antiepileptics, sucralfate, orlistat, estrogens, ferrous sulfate, raloxifene, cholestyramine, sevelamer).⁹ They found that the number of levothyroxine prescriptions increased by 6% after the initiation of a potential drug-interacting medication compared to the time period before (adjusted IRR 1.06, [95 % CI 1.05–1.07]).⁹ A trend of dosage increase was also observed, likely due to clinicians’ attempt to counter the TSH imbalance.⁹ Additionally, approximately 70% of the cohort were also utilizing PPIs, which is similar to the findings of our study.⁹

In light of known PPI overprescribing, efforts to deprescribe these agents should consider the potential effect on hypothyroidism control.¹⁷ Clinical vigilance and frequent therapeutic re-evaluation in LT4 users is needed to prevent downstream effects (e.g., uncontrolled hypothyroidism, acute symptoms, long term cardiac and bone health complications, cost and patient burden associated with dose adjustment). ^{3 4,5 7,8}

Predictors of LT4 use with interfering medications in our study population included female sex, older age, and year of office-based visit. The Veterans Health System study shared the same predictors of sex and age, in addition to Black race, Hispanic ethnicity, and presence of comorbidities.¹⁶ These findings suggest that older patients are at higher risk for concomitant use of interacting medications given their likelihood to be diagnosed with more comorbidities, see more prescribers, and experience polypharmacy.^18,19,20

The management of interfering medications can pose further challenges. Since these interacting medications are commonly used and are often clinically necessary, clinicians have a primary responsibility to identify and counsel patients on the potential effects of these medications, adjust levothyroxine dosing appropriately, and consider frequent monitoring.

Patients can also contribute to safe medication management by providing a full medication list at each visit, including both initiation and discontinuation, so that drug-drug interactions can be adequately assessed. This is not always the case though, especially in patients with extensive medication lists, where errors of omission on a medication reconciliation are more likely to occur.¹⁸ Designing systems that can help identify concomitant use of interfering medications, to assure proper utilization and clinical response, is necessary.

Although the identification of concomitant LT4 use with interacting medications is a primary concern, the nature by which this usage occurs should also be considered. Abrupt initiation or discontinuation of interacting medications can significantly impact thyroid levels given that levothyroxine has a narrow range of therapeutic efficacy. A result of the use of concomitantly interacting medications is an increased complexity in hypothyroidism management, likely requiring a dose alteration to correct the T4 imbalance. This may lead to an increase in healthcare utilization including additional visits to a provider, associated labs, and any other diagnostic tools needed to address the clinical presentation. These additional costs incurred can be avoided through proactive, rather than reactive actions by both the patient and the healthcare team.

Strengths of our analysis include the utilization of a dataset that represents a national sample of ambulatory care visits in the United States for over 10 years. Moreover, this dataset captures medications that are purchased over the counter (OTC) which may not be captured in other data sources (e.g., claims data). However, there are also several limitations. Since the NAMCS provides only cross-sectional records of visits, patients are unable to be followed over time to assess for changes in prescribing and subsequent health outcomes. Pertinent information including dose, duration of therapy, and timing of medication initiation or discontinuation would be extremely valuable, but is inherently unavailable through this database. Another limitation is including only the first 8 medications documented in the analysis, as it is unclear whether or not there is a prioritization of medication input order. Furthermore, NAMCS users are unable to determine whether OTC products are used routinely or as needed, as well as the timing of administration, which is relevant to LT4 absorption.⁶ We also limited our analysis to levothyroxine and did not incorporate other thyroid preparations; however, levothyroxine comprises nearly 80% of all thyroid preparations used in the United States.² Further research that captures downstream effects of LT4 use with interfering medications, including changes in prescribing patterns and hypothyroidism management over time, is necessary.

CONCLUSION

During the study period, nearly one-quarter of ambulatory care visits in which levothyroxine was initiated or continued included the concomitant use of a known interacting medication. PPIs accounted for the majority (79.6%) of interacting medication classes, followed by minerals (13.8%), which can both be obtained without a prescription. Clinicians should be vigilant when making therapeutic adjustments involving levothyroxine, ensuring that both OTC and prescription drug usage are carefully considered.