Association of Thyroid Hormone Treatment Intensity With Cardiovascular Mortality (2024)

Key Points

Question Is there an association between the intensity of thyroid hormone treatment and cardiovascular mortality?

Findings In this population-based cohort study of 705 307 adults who received thyroid hormone treatment, 10.8% died of cardiovascular causes. Both exogenous hyperthyroidism and exogenous hypothyroidism were associated with increased risk of cardiovascular mortality after adjusting for a comprehensive set of demographic and traditional cardiovascular risk factors.

Meaning These findings suggest that the intensity of thyroid hormone treatment may be a modifiable risk factor for cardiovascular mortality.

Importance Cardiovascular disease is the leading cause of death in the United States. Synthetic thyroid hormones are among the 3 most commonly prescribed medications, yet studies evaluating the association between the intensity of thyroid hormone treatment and cardiovascular mortality are scarce.

Objective To evaluate the association between thyroid hormone treatment intensity and cardiovascular mortality.

Design, Setting, and Participants This retrospective cohort study used data on 705 307 adults who received thyroid hormone treatment from the Veterans Health Administration Corporate Data Warehouse between January 1, 2004, and December 31, 2017, with a median follow-up of 4 years (IQR, 2-9 years). Two cohorts were studied: 701 929 adults aged 18 years or older who initiated thyroid hormone treatment with at least 2 thyrotropin measurements between treatment initiation and either death or the end of the study period, and, separately, 373 981 patients with at least 2 free thyroxine (FT₄) measurements. Data were merged with the National Death Index for mortality ascertainment and cause of death, and analysis was conducted from March 25 to September 2, 2020.

Exposures Time-varying serum thyrotropin and FT₄ levels (euthyroidism: thyrotropin level, 0.5-5.5 mIU/L; FT₄ level, 0.7-1.9 ng/dL; exogenous hyperthyroidism: thyrotropin level, <0.5 mIU/L; FT₄ level, >1.9 ng/dL; exogenous hypothyroidism: thyrotropin level, >5.5 mIU/L; FT₄ level, <0.7 ng/dL).

Main Outcomes and Measures Cardiovascular mortality (ie, death from cardiovascular causes, including myocardial infarction, heart failure, or stroke). Survival analyses were performed using Cox proportional hazards regression models using serum thyrotropin and FT₄ levels as time-varying covariates.

Results Of the 705 307 patients in the study, 625 444 (88.7%) were men, and the median age was 67 years (IQR, 57-78 years; range, 18-110 years). Overall, 75 963 patients (10.8%) died of cardiovascular causes. After adjusting for age, sex, traditional cardiovascular risk factors (eg, hypertension, smoking, and previous cardiovascular disease or arrhythmia), patients with exogenous hyperthyroidism (eg, thyrotropin levels, <0.1 mIU/L: adjusted hazard ratio [AHR], 1.39; 95% CI, 1.32-1.47; FT₄ levels, >1.9 ng/dL: AHR, 1.29; 95% CI, 1.20-1.40) and patients with exogenous hypothyroidism (eg, thyrotropin levels, >20 mIU/L: AHR, 2.67; 95% CI, 2.55-2.80; FT₄ levels, <0.7 ng/dL: AHR, 1.56; 95% CI, 1.50-1.63) had increased risk of cardiovascular mortality compared with individuals with euthyroidism.

Conclusions and Relevance This study suggests that both exogenous hyperthyroidism and exogenous hypothyroidism were associated with increased risk of cardiovascular mortality. These findings emphasize the importance of maintaining euthyroidism to decrease cardiovascular risk and death among patients receiving thyroid hormone treatment.

Despite widespread efforts at prevention and advances in diagnosis and treatment, cardiovascular disease remains the leading cause of death in the United States and affects nearly 50% of individuals in the US aged 20 years or older.^1,2 In addition, it is estimated that the annual financial burden of heart disease in the United States is more than $200 billion.¹ Although many cardiovascular risk factors are known (eg, hypertension, diabetes, and smoking), the persistent public health impact of cardiovascular disease mandates a more complete understanding of novel risk factors.^1,3 A recent study has shown that the intensity of thyroid hormone treatment is a modifiable risk factor for incident atrial fibrillation and stroke⁴; however, its association with cardiovascular mortality remains unclear.

Thyroid hormone treatment is widespread, with levothyroxine prescriptions consistently among the top 3 of all prescription medications in the United States in the past decade.^5-7 However, up to 50% of patients who receive thyroid hormone treatment may exhibit exogenous hyperthyroidism or hypothyroidism (ie, have thyrotropin levels below or above the reference range, respectively).^8,9 The associations of long-term exogenous hyperthyroidism and hypothyroidism with clinical outcomes, including cardiovascular risk and all-cause mortality, have recently been investigated.^4,10-13 Previous studies have shown that serum thyrotropin concentrations outside the euthyroid range correlated with increased cardiovascular risk and all-cause mortality among patients who received thyroid hormone treatment for hypothyroidism^11,13,14; however, studies focusing specifically on the association between the intensity of thyroid hormone treatment and cardiovascular mortality are lacking.

The objective of this study was to evaluate the association between the intensity of thyroid hormone treatment and cardiovascular mortality using a nationwide, population-based cohort of adults receiving thyroid hormone treatment. We hypothesized that both exogenous hyperthyroidism and hypothyroidism would be associated with increased cardiovascular mortality, even when adjusting for age, sex, and traditional cardiovascular risk factors, such as hypertension and smoking.

We conducted a population-based, retrospective cohort study between January 1, 2004, and December 31, 2017, using data from the Veterans Health Administration.¹⁵ This large, integrated health care system provides care to more than 9 million US veterans annually.¹⁶ Deidentified patient-level data were obtained using the Veterans Health Administration Corporate Data Warehouse database, which is a national centralized data repository for the Veterans Health Administration. This database provides clinical and administrative information, including diagnoses, laboratory data, pharmacy prescription fills, health factors, and demographic information.^17-19 These data were then linked to the National Death Index for mortality ascertainment and to identify cause of death.²⁰ The study followed the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline.²¹ This study was exempt from the University of Michigan institutional review board and approved by the Ann Arbor Veteran Affairs institutional review board, which included a waiver of informed consent as data were deidentified.

The study population included 705 307 patients aged 18 years or older who initiated thyroid hormone treatment during the study period. Of these patients, 701 929 had at least 2 outpatient measurements of serum thyrotropin between initiation of thyroid hormone treatment and either death or the end of the study. We also studied patients aged 18 years or older receiving thyroid hormone treatment who had at least 2 outpatient free thyroxine (FT₄) measurements between initiation of thyroid hormone treatment and death or study conclusion (n = 373 981). Patients with a history of thyroid cancer (n = 15 090) were excluded because lower thyrotropin levels are often targeted to reduce the risk of disease recurrence. In addition, patients prescribed lithium (n = 23 715) or amiodarone (n = 70 358) were excluded because these medications have a known association with abnormal thyroid function test results. Patients who did not have a documented date of birth (n = 17) were also excluded prior to attaining the final analytic sample.

The study outcome was cardiovascular mortality as determined by cause of death from cardiovascular diseases based on International Statistical Classification of Diseases and Related Health Problems, Tenth Revision (ICD-10) codes (including codes I00-I99). The incident event (ie, death from cardiovascular causes) occurred after initiation of thyroid hormone treatment and through December 31, 2017, with a median follow-up of 4 years (IQR, 2-9 years).

The exposure variables were time-varying serum thyrotropin and FT₄ levels. Serum thyrotropin and FT₄ levels were compiled using the patients’ laboratory records and assembled into 2 separate longitudinal data sets. Prior to analyses, thyrotropin levels were log transformed owing to nonnormal distribution, as has been done in prior studies cited in the literature.^4,10,22 Although there is some variation by laboratory, the reference ranges for thyrotropin and FT₄ levels at the Ann Arbor VA laboratory were used for analyses (thyrotropin level, 0.5-5.5 mIU/L; FT₄ level, 0.7-1.9 ng/mL [to convert to picomoles per liter, multiply by 12.87]). Of note, 99.1% of patients (370 603 of 373 981) who had at least 2 FT₄ measurements also had at least 2 thyrotropin measurements.

Fixed covariates included patient sex, age, race, ethnicity, and smoking status. Data on sex were obtained from the Veterans Health Administration Corporate Data Warehouse database at study entry and recorded as male or female. Age was analyzed both as a continuous variable and as a categorical variable using clinically meaningful categories (18-49, 50-64, 65-74, 75-84, and ≥85 years). Race was self-reported as Alaska Native or American Indian, Asian, Black, Native Hawaiian or Pacific Islander, White, multiracial, or unknown. Because the number of patients who identified as Alaska Native or American Indian, Asian, Native Hawaiian or Pacific Islander, or multiracial was small, these groups were collapsed and classified as “other” for analyses. Ethnicity was self-described as Hispanic, non-Hispanic, or unknown. Smoking status was determined at the time of initiation of thyroid hormone treatment and was recorded as never smoker, current or former smoker, or unknown. Time-varying covariates included hypertension, hyperlipidemia, diabetes, prior history of cardiovascular disease (coronary artery disease, ischemic heart disease, heart failure, or stroke), and prior history of cardiac arrhythmia and were determined by using International Classification of Diseases, Ninth Revision; International Statistical Classification of Diseases and Related Health Problems, Tenth Revision; and Current Procedure Terminology, Fourth Edition codes.

Data were analyzed from March 25 to September 2, 2020. Descriptive information was tabulated for both the thyrotropin and FT₄ cohorts. Univariate associations between individual factors and the outcome were examined. We then performed separate survival analyses using Cox proportional hazards regression models to determine correlates for the study outcome (ie, cardiovascular mortality). In these models, the exposure variables (thyrotropin and FT₄ levels) were treated as time-varying covariates, which allowed us to better account for variability in these measures during the study period. Serum thyrotropin and FT₄ levels were treated as categorical variables. Exogenous hyperthyroidism was defined by thyrotropin levels lower than 0.5 mIU/L (categorized as <0.1 mIU/L and 0.1-0.5 mIU/L) or by FT₄ levels higher than 1.9 ng/dL; euthyroidism was defined by thyrotropin levels from 0.5 to 5.5 mIU/L and FT₄ levels from 0.7 to 1.9 ng/dL; and exogenous hypothyroidism was defined by thyrotropin levels higher than 5.5 mIU/L (categorized as >5.5 to <7.5 mIU/L, 7.5 to <10 mIU/L, 10-20 mIU/L, and >20 mIU/L) or by FT₄ levels lower than 0.7 ng/dL. When multiple measurements were present in an annual time period (defined as a calendar year), which occurred for 5.9% of the thyrotropin cohort (41 414 of 701 929) and 2.7% of the FT₄ cohort (10 097 of 373 981), the geometric mean was calculated for the thyrotropin cohort, and the arithmetic mean was calculated for the FT₄ cohort; these values were used for analyses. For both models, additional fixed covariates included patient sex, age, race, ethnicity, and smoking status, and additional time-varying covariates included hypertension, hyperlipidemia, diabetes, prior history of cardiovascular disease, and prior history of cardiac arrhythmia. No observations were excluded from the statistical analyses owing to missing information.

All statistical analyses were conducted using SAS, version 7.15 HF8 (SAS Institute Inc). A 95% CI was used to determine statistical significance, and P < .05 from 2-sided tests was considered statistically significant for all analyses. Model adequacy was assessed using generalized residuals-based diagnostics in SAS PROC PHREG.

Table 1 provides key demographic, comorbidity, and mortality data for the 705 307 patients receiving thyroid hormone treatment. Most patients were male (625 444 [88.7%]), White (559 173 [79.3%]), non-Hispanic (609 537 [86.4%]), had a history of hypertension (582 061 [82.5%]) or hyperlipidemia (582 389 [82.6%]), and were current or former smokers (467 606 [66.3%]). The median age was 67 years (IQR, 57-78 years; range, 18-110 years). During the study period, 224 943 patients (31.9%) died of any cause, and 75 963 patients (10.8%) died of cardiovascular disease.

The frequency distributions of the number of thyrotropin measurements and the number of FT₄ measurements overall and in association with the number of years, as well as the mean number of thyrotropin measurements and FT₄ measurements per patient per year, are shown in eTable 1 and eFigures 1 and 2 in the Supplement. In univariate analyses, patient sex, age, race, ethnicity, smoking status, hypertension, hyperlipidemia, diabetes, prior history of cardiovascular disease, prior history of cardiac arrhythmia, and serum thyrotropin and FT₄ levels were significantly associated with cardiovascular mortality.

Table 2 shows results from survival analyses using Cox proportional hazards regression models demonstrating patient characteristics associated with cardiovascular mortality for the thyrotropin and FT₄ cohorts. When adjusting for age, sex, and other relevant demographic and traditional cardiovascular risk factors (such as hypertension and smoking), patients with exogenous hyperthyroidism (eg, thyrotropin levels <0.1 mIU/L: adjusted hazard ratio [AHR], 1.39; 95% CI, 1.32-1.47; FT₄ levels >1.9 ng/dL: AHR, 1.29; 95% CI, 1.20-1.40) and patients with exogenous hypothyroidism (eg, thyrotropin levels >20 mIU/L: AHR, 2.67; 95% CI, 2.55-2.80; FT₄ levels <0.7 ng/dL: AHR, 1.56; 95% CI, 1.50-1.63) had an increased risk of cardiovascular mortality compared with individuals with euthyroidism. Furthermore, the risk of cardiovascular mortality increased progressively with lower and higher thyrotropin levels compared with euthyroidism. Similar findings were obtained when age was treated as a continuous variable (eTable 2 in the Supplement).

The forest plot in the Figure illustrates the association between serum thyrotropin and FT₄ levels with cardiovascular mortality after adjustment for relevant demographic and cardiovascular risk factors. Cardiovascular mortality was higher among patients with exogenous hyperthyroidism (thyrotropin levels <0.1 mIU/L: AHR, 1.39; 95% CI, 1.32-1.47; thyrotropin levels of 0.1 to <0.5 mIU/L: AHR, 1.13; 95% CI, 1.09-1.17; FT₄ levels >1.9 ng/dL: AHR, 1.29; 95% CI, 1.20-1.40) and those with exogenous hypothyroidism (thyrotropin levels from >5.5 to <7.5 mIU/L: AHR, 1.42; 95% CI, 1.38-1.46; thyrotropin levels from 7.5 to <10 mIU/L: AHR, 1.76; 95% CI, 1.70-1.82; thyrotropin levels of 10-20 mIU/L: AHR, 2.13; 95% CI, 2.05-2.21; thyrotropin levels >20 mIU/L: AHR, 2.67; 95% CI, 2.55-2.80; FT₄ levels <0.7 ng/dL: AHR, 1.56; 95% CI, 1.50-1.63), with risk increasing with higher serum thyrotropin levels compared with individuals with euthyroidism.

In this population-based study of a large cohort of adults receiving thyroid hormone treatment, we found that the intensity of thyroid hormone treatment was associated with cardiovascular mortality. Both exogenous hyperthyroidism and hypothyroidism were associated with increased risk of cardiovascular mortality, even when adjusting for relevant demographic and traditional cardiovascular risk factors as well as previous history of cardiovascular disease and/or arrhythmia. In addition, our findings suggest that the risk of cardiovascular mortality is directly associated with the degree of thyrotropin abnormality outside the euthyroid range, such that thyrotropin levels lower than 0.1 mIU/L and higher than 20 mIU/L were associated with the highest increased risk. We observed that the AHRs increased more rapidly for older age categories in the thyrotropin cohort compared with the FT₄ cohort. This finding may suggest that the thyrotropin level is more strongly associated with cardiovascular risk than is the FT₄ level in older adults. From a clinical perspective, older adults, and particularly the oldest old (aged ≥85 years), appear to be the most vulnerable, with increased risk of cardiovascular mortality with both exogenous hyperthyroidism and hypothyroidism.

A few prior studies have evaluated the association between serum thyrotropin levels and cardiovascular outcomes among patients receiving thyroid hormone treatment.^4,10,11,13 Flynn et al¹⁰ used a time-weighted mean thyrotropin level to group patients and demonstrated increased risk of a composite outcome of cardiovascular admission or death with suppressed or elevated serum thyrotropin levels in a cohort of 17 684 patients treated with levothyroxine. In addition, Lillevang-Johansen et al¹¹ performed a case-control study of 20 487 patients with incident cardiovascular disease nested within a larger cohort of individuals with hypothyroidism and showed that, compared with matched controls, patients with treated hypothyroidism (n = 636) had increased odds of incident cardiovascular disease and all-cause mortality for each 6-month period of overtreatment or undertreatment. Finally, Thayakaran et al¹³ evaluated the association between thyrotropin concentration and incident cardiovascular outcomes, but not mortality, among patients with hypothyroidism (N = 160 439), adjusting for levothyroxine treatment. This study found that compared with a thyrotropin level of 2 to 2.5 mIU/L, a thyrotropin level higher than 10 mIU/L was associated with an increased risk of ischemic heart disease and heart failure. None of these studies evaluated the association of the intensity of thyroid hormone treatment with cardiovascular mortality, possibly because of inadequate power. Our study is novel because, unlike these prior studies, it specifically evaluated the association between cardiovascular mortality and serum thyrotropin or FT₄ level, while adjusting for a comprehensive list of possible confounders, exclusively in a large cohort of patients receiving thyroid hormone treatment.

Our study findings have the potential to affect how we think about the risks and benefits associated with thyroid hormone treatment, particularly for vulnerable populations, such as older adults or those with underlying cardiovascular disease. Because synthetic thyroid hormones have consistently been one of the top 3 most frequently prescribed medications in the United States in the past decade,^5-7 and because cardiovascular disease remains the leading cause of death, an association between the intensity of thyroid hormone treatment and cardiovascular death has far-reaching implications for both patients and physicians. Although the variability in thyrotropin and FT₄ levels and thyroid hormone dose adjustments are an inevitable reality for most patients, our study emphasizes the importance of regular monitoring of thyroid function test results and correction of both overtreatment and undertreatment with exogenous thyroid hormones to reduce patient harm, particularly for older adults who are at higher risk for adverse effects.^4,8,9,23

Our study has several strengths. First, it is a large, population-based study using data from the largest integrated health care system in the United States. Second, we used serum thyrotropin and FT₄ levels as time-varying covariates, which allowed us to incorporate all qualifying thyrotropin and FT₄ levels measured during the study period. Because prior studies have shown wide variability in thyroid function test results over time among patients receiving thyroid hormone treatment,^8,9,24,25 this method facilitates a more comprehensive evaluation of the association between thyroid function test results and cardiovascular mortality than would be possible using a single value at study entry in a cross-sectional design. Third, while the use of an internal comparator cannot completely eliminate risk of confounding, it avoids the inherent limitations of comparing an exposed group with the background population. Fourth, because the Veterans Health Administration Corporate Data Warehouse includes comprehensive information on patient demographic characteristics, comorbid conditions, and smoking status, we were able to account for most traditional cardiovascular risk factors.

Population-based studies using databases have some inherent limitations that merit consideration. Although we were able to account for most known cardiovascular risk factors, we were not able to control for other potential confounders, such as alcohol status and body mass index or rates of obesity, because these could not be accurately captured in the Veterans Health Administration data. Adjusting for a comprehensive list of cardiovascular risk factors, including hypertension, hyperlipidemia, diabetes, and prior history of cardiovascular disease, which represent some of the downstream effects of obesity, partially mitigates this limitation. In addition, we were unable to determine the degree to which risk factors for cardiovascular mortality, such as diabetes and hypertension, were appropriately treated. We also acknowledge that we were not able to account for all medications and supplements that could interfere with thyroid hormone metabolism and action and/or thyroid function test results. Furthermore, because cause of death in the National Death Index relies on death certificates, risk of misclassification is possible.²⁶ Because the Veterans Health Administration population is predominantly male, women are generally underrepresented in studies using this database. However, because the risk of cardiovascular disease is higher for men than for women²⁷ and because more than 70 000 women were included in this cohort, the results of this study are highly clinically relevant. Additionally, this was an observational study, and a causal relationship between the intensity of thyroid hormone treatment and cardiovascular mortality could not be definitively established.

In this population-based cohort of patients receiving thyroid hormone treatment, we found that both exogenous hyperthyroidism and hypothyroidism were associated with an increased risk of cardiovascular mortality after adjusting for a comprehensive set of demographic and traditional cardiovascular risk factors. Cardiovascular disease remains the leading cause of death in the United States, and its economic impact is enormous. Identifying and addressing modifiable risk factors continues to be critically important to reducing the rates of cardiovascular disease and mortality. The emergence of the intensity of thyroid hormone treatment as a potential associated risk factor provides a highly relevant and easily modifiable clinical parameter for patients who receive thyroid hormone treatment.

Accepted for Publication: March 27, 2022.

Published: May 12, 2022. doi:10.1001/jamanetworkopen.2022.11863

Open Access: This is an open access article distributed under the terms of the CC-BY License. © 2022 Evron JM et al. JAMA Network Open.

Corresponding Author: Maria Papaleontiou, MD, Division of Metabolism, Endocrinology and Diabetes, Department of Internal Medicine, University of Michigan, North Campus Research Complex, 2800 Plymouth Rd, Bldg 16, Room 453S, Ann Arbor, MI 48109 (mpapaleo@med.umich.edu).

Author Contributions: Mr Reyes-Gastelum and Dr Papaleontiou had full access to all of the data in the study and take responsibility for the integrity of the data and the accuracy of the data analysis.

Concept and design: Evron, Papaleontiou.

Acquisition, analysis, or interpretation of data: All authors.

Drafting of the manuscript: Evron, Reyes-Gastelum, Banerjee, Papaleontiou.

Critical revision of the manuscript for important intellectual content: Evron, Hummel, Haymart, Banerjee, Papaleontiou.

Statistical analysis: Evron, Reyes-Gastelum, Banerjee.

Obtained funding: Papaleontiou.

Administrative, technical, or material support: Haymart, Papaleontiou.

Supervision: Papaleontiou.

Conflict of Interest Disclosures: Dr Hummel reported serving as a clinical trial site principal investigator for Pfizer, Corvia Medical, and Axon Therapeutics; serving as a clinical trial site coinvestigator for Novartis; and receiving grants from the Veterans Health Administration and grants from the National Institutes of Health outside the submitted work. No other disclosures were reported.

Funding/Support: This work was supported by the National Institute on Aging of the National Institutes of Health under award K08 AG049684 and by a pilot grant cofunded by the Claude D. Pepper Older Americans Independence Center, the Michigan Institute for Clinical and Health Research, and the Michigan Biology of Cardiovascular Aging (Dr Papaleontiou).

Role of the Funder/Sponsor: The funding sources had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; preparation, review, or approval of the manuscript; and decision to submit the manuscript for publication.

Additional Contributions: The authors would like to acknowledge Richard Evans, MS, US Department of Veterans Affairs, VA Ann Arbor, Jennifer Burns, MHSA, US Department of Veterans Affairs, VA Ann Arbor, and Wyndy Wiitala, PhD, VA Ann Arbor Healthcare System, for their help with the statistical analyses; they were not compensated for their contributions. The authors would also like to acknowledge Brittany Gay, BA, University of Michigan, for her assistance with manuscript formatting and preparation for submission; she was compensated for her contribution.

Additional Information: Restrictions apply to the availability of some or all data generated or analyzed during this study to preserve patient confidentiality or because they were used under license. The corresponding author will on request detail the restrictions and any conditions under which access to some data may be provided.