Abstract
Context:
Whether over- or under-replacement with thyroid hormone is associated with incident heart failure remains unclear.
Objective:
Elucidate the relationship between thyroid hormone over- and under-replacement and incident heart failure.
Design:
Retrospective cohort study of 641,504 adults who initiated thyroid hormone treatment between 2004 and 2017.
Setting:
Veterans Health Administration.
Participants:
Two cohorts of adults aged ≥18 years treated with thyroid hormone were separately studied: 638,323 adults with at least two outpatient thyroid stimulating hormone (TSH) measurements, and 338,249 adults with at least two outpatient free thyroxine (FT4) measurements, between thyroid hormone initiation and incident heart failure or study conclusion.
Exposure:
Time-varying TSH and FT4.
Main Outcome Measure:
Incident heart failure.
Results:
Overall, 564,152/641,504 (87.9%) patients were men. Median age was 66 years. Incident heart failure occurred in 81,286/641,504 (12.7%) patients. Following adjustment for age, sex, and cardiovascular risk factors (e.g., smoking, hypertension), under-replacement (e.g., TSH>20 mIU/L, adjusted odds ratio [AOR], 1.93, 95% confidence interval [CI], 1.84–2.02; FT4<0.7 ng/dL, AOR 1.08; 95% CI, 1.04–1.12) and over-replacement (e.g., TSH<0.1 mIU/L, AOR, 1.06; 95% CI, 1.01–1.12; FT4>1.9 ng/dL, AOR, 1.23; 95% CI, 1.14–1.32) were associated with increased incident heart failure compared to euthyroidism. Incident heart failure risk was cumulative over time with both thyroid hormone over- and under-replacement, but association was strongest with under-replacement (e.g., TSH>5.5 mIU/L associated with a 5.8-fold increase in incident heart failure over 5 years).
Conclusions:
Our findings suggest that thyroid hormone treatment intensity may be a modifiable risk factor for heart failure.
Keywords: thyroid hormone replacement, heart failure, overtreatment, undertreatment
INTRODUCTION
Heart failure afflicts more than six million adults in the United States (U.S.) and the prevalence is increasing. Health care expenditures related to heart failure exceed $30 billion annually and are expected to approach $70 billion by 2030 (1). The most common causes of heart failure include ischemic heart disease, hypertension and valvular heart disease (2). Importantly, heart failure may also be a clinical manifestation of several endocrine disorders, including hypothyroidism and hyperthyroidism (2). Thyroid hormone plays a vital role in regulating cardiovascular hemodynamics through effects on heart rate, myocardial contractility, cardiac output and systemic vascular resistance (3). Although the relationship between endogenous thyroid dysfunction and cardiac performance is well-established (4), few studies have attempted to evaluate the effects of over- or under-replacement with exogenous thyroid hormone on important cardiovascular endpoints (5–9).
Levothyroxine, the most commonly used synthetic form of thyroid hormone, accounts for ~90% of thyroid hormone prescriptions in the U.S. (10). It is also consistently among the most frequently prescribed medications in the U.S. (11,12), with the number of annual prescriptions continuing to increase (13). Prior studies have demonstrated a high frequency of over- and under-replacement in thyroid hormone users, with rates surpassing 40% (14–16). This has implications for patients, as there is growing evidence that thyroid hormone over- and under-replacement may be associated with increased risk of cardiovascular disease, stroke, cardiovascular mortality, and all-cause mortality, with serum thyroid stimulating hormone (TSH) and free thyroxine (FT4) levels potentially being independent predictors of cardiovascular risk (5,6,8,9,17).
Despite some recent studies elucidating a possible association between thyroid function tests outside the reference range in thyroid hormone users and worse cardiovascular outcomes, data on the association between thyroid hormone over- and under-replacement and incident heart failure are scarce. The objective of this study was to examine the association between over- and under-replacement with thyroid hormone and incident heart failure, using a nationwide, population-based cohort of adults treated with thyroid hormone. Our hypothesis was that, after adjusting for demographic and cardiovascular risk factors, both over- and under-replacement would be associated with increased incident heart failure.
METHODS
Data Source and Study Population
The data source has been described in detail in our prior manuscripts (5,9). This was a retrospective, population-based cohort study using data from the Veterans Health Administration (VHA) (18). The VHA is the largest integrated health system in the U.S. providing healthcare to millions of veterans (19). The study period was January 1, 2004 to December 31, 2017. Patient-level data from the VHA Corporate Data Warehouse database, a national centralized data repository, were deidentified. This database includes data such as diagnoses, laboratory data, pharmacy prescription fills, health factors, and demographic information (20–22). The study was conducted in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) reporting guideline (23). This study was exempt from the University of Michigan Institutional Review Board and approved by the Ann Arbor Veterans Affairs Institutional Review Board, which included a waiver of informed consent as we used deidentified data.
Overall, eligible patients were those aged 18 years and older without a diagnosis of pre-existing heart failure, who were initiated on thyroid hormone during the study period (N = 641,504). Continuous thyroid hormone use throughout the study was ascertained through pharmacy prescription fill dates. Of the eligible patients, 638,323 patients who had at least two outpatient TSH measurements between the beginning of the study period and incident heart failure or between the beginning and the end of the study period were included in the initial analyses. Additionally, we evaluated a separate cohort of 338,249 patients, aged 18 years and older, with at least two outpatient FT4 measurements between the beginning of the study period and either incident heart failure or study completion. Notably, the vast majority of patients in the FT4 cohort were also in the TSH cohort (i.e., 99.2% of the individuals with at least two FT4 measurements also had at least two TSH measurements). Patients with a history of thyroid cancer, who may sometimes have a TSH goal below the normal range to reduce recurrence risk (n = 18,890), and individuals on medications that may interfere with thyroid function tests such as amiodarone and lithium (n = 94,148) were excluded prior to reaching the final analytic samples (Figure 1). Thyroid function tests from inpatient admissions were not included as abnormalities are often attributable to non-thyroidal illness. Finally, TSH and FT4 measurements that were taken after an incident event (heart failure) were excluded from the analyses.
Figure 1. Study Flow Diagram.
This diagram depicts how final analytic samples were reached.
Measures
Study Outcomes
The study outcome was incident heart failure as determined by International Statistical Classification of Diseases and Related Health Problems, Ninth Revision (ICD-9) code 428.xx or International Statistical Classification of Diseases and Related Health Problems, Tenth Revision (ICD-10) code I50.xx. The incident diagnosis of heart failure must have occurred after the initiation of thyroid hormone replacement and before the end of the study period (December 31, 2017). Data were censored at death or last follow-up.
Exposure Variables
The exposure variables included time-varying TSH and FT4 values, which were obtained from the patients’ laboratory records and entered into longitudinal datasets. Separate datasets were created for TSH and FT4, and TSH values were log-transformed prior to analyses given non-normal distribution. The reference ranges for TSH and FT4 at the Ann Arbor VA (0.5–5.5 mIU/L and 0.7–1.9 ng/ml, respectively) were used in all analyses.
Covariates
The study included both fixed and time-varying covariates. Fixed covariates included patient age, sex, race, ethnicity and smoking status. Age was analyzed as a continuous variable, and to improve the clinical applicability of our findings, separately as a categorical variable using clinically relevant cutpoints (18–49, 50–64, 65–74, 75–84, and ≥85 years). Sex, race, and ethnicity were obtained from the VHA Corporate Data Warehouse database at the time of study entry. Sex was recorded as male or female. Race was based on self-report and categories included Alaskan Native or American Indian, Asian, Black, Native Hawaiian or Pacific Islander, White, multiracial, or unknown. Because the vast majority of individuals identified as Black or White, all other racial groups were collapsed and classified as “other” in all analyses. Ethnicity was self-reported and defined as either non-Hispanic, Hispanic, or unknown. Smoking status was classified as never smoker, current or former smoker, or unknown.
Time-varying covariates included hypertension (ICD9: 401.0, 401.1, 401.9, ICD10: I10, I15), diabetes mellitus (ICD9 250.xx, ICD10 E08-E13), hyperlipidemia (ICD9: 272.0–272.4, ICD10: E78.x), and pre-existing cardiovascular disease (ICD9 410xx-414.xx, 434.91, 434.11, ICD10 I00-I99, CPT4 92982, 92984, 92995, 92996, 92980, 92981, 92980, 92982, 92984, 33510–33515).
Statistical Analysis
Data were analyzed from August 2023 to October 2024. Bivariate associations between individual factors and incident heart failure as the outcome were examined. Descriptive data were generated separately for the TSH and FT4 cohorts. Then, to determine correlates for incident heart failure, separate multivariable generalized linear mixed-effects regression models were performed for each cohort. To capture the potential variability in thyroid function tests during the study period, the exposure variables (TSH and FT4) were treated as time-varying covariates. Residual-type random component and a variance component covariance structure for the mixed models between multiple measurements from each patient were used. For both models, fixed covariates included age, sex, race, ethnicity, and smoking status. Time-varying covariates included hypertension, diabetes, hyperlipidemia, and prior history of cardiovascular disease.
We performed separate analyses with incident heart failure as the outcome variable; first using TSH as the exposure variable and then separately using FT4 as the exposure variable. Both TSH and FT4 were analyzed as continuous and categorical variables. When analyzing TSH and FT4 as categorical variables, the multiple measurements were aggregated by defining annual time periods from the initiation of thyroid hormone therapy, as previously done by our team (9). For each annual time period (which was defined as a calendar year), the multiple serum TSH measurements were combined using the geometric mean, while the multiple FT4 measurements were combined using the arithmetic mean (6). For the TSH cohort, under-replacement was defined as a TSH 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, >20 mIU/L), and over-replacement was defined as a TSH value lower than 0.5 mIU/L (categorized as <0.1 mIU/L and 0.1–0.5 mIU/L). For the FT4 cohort, under-replacement was defined by FT4 levels lower than 0.7 ng/dl and over-replacement was defined as FT4 levels higher than 1.9 ng/dl. The analyses were conducted with the annually aggregated serum values. No observations were excluded due to missing data.
Separate time-to-event models regressing the risk of incident heart failure (outcome) based on TSH and free T4 levels used as categorical variables were also conducted, creating trajectories of cumulative risk over time.
All statistical analyses were conducted using SAS version 9.4_M6 (SAS Institute Inc., Cary, NC, U.S.A.). For all analyses, 95% confidence intervals and p-value <0.05 were used to determine statistical significance where appropriate. Generalized residual-based diagnostics using default plots from the PROC GLIMMIX diagnostic functions were utilized to assess model adequacy. Forest-plot of primary, model-based results, was created with ggplot (v3.5.1) in R (v4.4.1).
RESULTS
Salient demographic and clinical data for the study population are shown in Table 1. The combined cohort included 641,504 patients. The majority of patients were male (564,152 [87.9%]), White (509,642 [79.4%]), and non-Hispanic (555,138 [86.5%]). Additionally, 519,424 (81.0%) had hypertension, 523,905 (81.7%) had hyperlipidemia, and 423,085 (66.0%) were current or former smokers. The median age was 66 years (IQR, 56–76; range 18–110 years). Median follow-up was 59 months (IQR, 25–108). A total of 81,286 (12.7%) patients were diagnosed with incident heart failure during the study period. Patient sex, age, race, ethnicity, smoking status, hypertension, hyperlipidemia, diabetes, prior history of cardiovascular disease (excluding a prior diagnosis of heart failure), and TSH and FT4 levels were significantly associated with incident heart failure in univariate analyses.
Table 1.
Characteristics of Patients on Thyroid Hormone Therapy
| Patient Characteristics | Overall Cohort (n=641,504) N (%) |
Cohort with at least 2 TSH measurements (n=638,323) N (%) |
Cohort with at least 2 FT4 measurements (n=338,249)a N (%) |
|---|---|---|---|
| Sex | |||
| Female | 77,352 (12.1) | 77,026 (12.1) | 47,164 (13.9) |
| Male | 564,152 (87.9) | 561,297 (87.9) | 291,085 (86.1) |
| Mean, SD / Median, IQR | 65.1, 14.3 / 66, 56–76 | 65.1, 14.4 / 66, 56–76 | 63.1, 14.4 / 63, 54–74 |
| 18–49 | 88,683 (13.8) | 88,376 (13.85) | 56,180 (16.61) |
| 50–64 | 211,940 (33.0) | 211,029 (33.06) | 124,148 (36.70) |
| 65–74 | 152,634 (23.8) | 151,921 (23.8) | 76,378 (22.6) |
| 5–84 | 143,135 (22.3) | 142,238 (22.3) | 63,200 (18.7) |
| ≥85 | 45,112 (7.0) | 44,759 (7.0) | 18,343 (5.4) |
| Race | |||
| White | 509,642 (79.4) | 507,448 (79.5) | 270,701 (80.0) |
| Black | 44,804 (7.0) | 44,408 (7.0) | 28,136 (8.3) |
| Other | 87,058 (13.6) | 86,467 (13.5) | 39,412 (11.7) |
| Ethnicity | |||
| Not Hispanic | 555,138 (86.5) | 552,489 (86.6) | 297,462 (87.9) |
| Hispanic | 34,956 (5.5) | 34,850 (5.5) | 19,061 (5.6) |
| Unknown | 51,410 (8.0) | 50,984 (8.0) | 21,726 (6.4) |
| Smoking status | |||
| Never smoker | 85,451 (13.3) | 85,158 (13.3) | 52,498 (15.5) |
| Current/Former smoker | 423,085 (66.0) | 420,992 (66.0) | 223,447 (66.1) |
| Unknown | 132,968 (20.7) | 132,173 (20.7) | 62,304 (18.4) |
| Hypertension | 519,424 (81.0) | 516,909 (82.0) | 273,577 (80.9) |
| Diabetes mellitus | 259,162 (40.4) | 257,935 (40.4) | 138,524 (41.0) |
| Hyperlipidemia | 523,905 (81.7) | 521,603 (81.7) | 279,836 (82.7) |
| Prior history of cardiovascular disease | 70,050 (10.9) | 69,731 (10.9) | 40,919 (12.1) |
| Prior history of atrial fibrillation | 91,834 (14.3) | 91,427 (14.3) | 47,718 (14.1) |
| Serum TSHb | 2.90 (<0.01–125) | 2.87 (<0.01–125) | --------- |
| Serum Free T4c | 1.02 (0.004–20) | -------- | 1.05 (0.004–20) |
| Incident Event | N (%) | N (%) | N (%) |
| Heart Failure | 81,286 (12.7) | 80,734 (12.7) | 43,317 (12.8) |
99.24% of individuals who had at least two free T4 measurements, also had at least two TSH values.
Serum TSH in mIU/L, median and range.
Serum FT4 in ng/dL, median and range.
Abbreviations: TSH, thyroid stimulating hormone; FT4, free thyroxine
Table 2 shows the patient characteristics associated with incident heart failure for the TSH and FT4 cohorts in adjusted multivariable analyses with age as a categorical variable and TSH and FT4 as continuous variables. When adjusting for relevant demographic and cardiovascular risk factors, thyroid hormone under-replacement was independently associated with an increased risk of incident heart failure (Adjusted Odds Ratio [AOR], 1.04; 95% CI, 1.03–1.05) compared to euthyroidism. Similar findings were noted when age was analyzed as a continuous variable (Table 3).
Table 2.
Patient Characteristics Associated with Incident Heart Failure
| TSH Cohorta | FT4 Cohortb | |||
|---|---|---|---|---|
| Patient Characteristics | N (%) patients with heart failure | Adjusted OR (95% CI) | N (%) patients with heart failure | Adjusted OR (95% CI) |
| Thyroid function test | -- | 1.04 (1.03–1.05) | -- | 1.00 (0.99–1.01) |
| Sex | ||||
| Male | 75,904 (13.5) | Ref | 40,318 (13.9) | Ref |
| Female | 4,830 (6.3) | 0.78 (0.76–0.81) | 2,999 (6.4) | 0.78 (0.75–0.81) |
| Age | ||||
| 18–49 | 3,354 (3.8) | Ref | 2,302 (4.1) | Ref |
| 50–64 | 23,935 (11.3) | 1.94 (1.87–2.02) | 14,482 (11.7) | 1.88 (1.80–1.97) |
| 65–74 | 21,113 (13.9) | 2.56 (2.47–2.66) | 11,296 (14.8) | 2.41 (2.30–2.52) |
| 75–84 | 24,578 (17.3) | 3.56 (3.43–3.70) | 11,770 (18.6) | 3.32 (3.17–3.48) |
| ≥85 | 7,754 (17.3) | 5.58 (5.35–5.82) | 3,467 (18.9) | 4.90 (4.64–5.19) |
| Race | ||||
| White | 65,983 (13.0) | Ref | 35,615 (13.2) | Ref |
| Black | 5,562 (12.5) | 1.15 (1.12–1.18) | 3,535 (12.6) | 1.03 (0.99–1.06) |
| Other | 1,630 (11.2) | 0.95 (0.91–0.99) | 874 (11.1) | 0.89 (0.83–0.95) |
| Ethnicity | ||||
| Not Hispanic | 71,429 (12.9) | Ref | 38,715 (13.0) | Ref |
| Hispanic | 4,097 (11.8) | 0.81 (0.78–0.84) | 2,354 (12.4) | 0.96 (0.92–0.99) |
| Unknown | 5,208 (10.2) | 0.98 (0.92–1.03) | 2,248 (10.4) | 1.01 (0.94–1.09) |
| Smoking status | ||||
| Never | 10,711 (12.6) | Ref | 6,456 (12.3) | Ref |
| Current/Former | 56,121 (13.3) | 1.11 (1.08–1.13) | 30,041 (13.4) | 1.17 (1.13–1.20) |
| Unknown | 13,902 (10.5) | 1.02 (1.00–1.05) | 6,820 (11.0) | 1.03 (1.00–1.07) |
| Hypertension | ||||
| No | 6,085 (5.0) | Ref | 2,986 (4.6) | Ref |
| Yes | 74,649 (14.4) | 2.32 (2.25–2.38) | 40,331 (14.7) | 2.65 (2.55–2.75) |
| Diabetes mellitus | ||||
| No | 37,557 (9.9) | Ref | 19,736 (9.9) | Ref |
| Yes | 43,177 (16.7) | 1.78 (1.76–1.81) | 23,581 (17.0) | 1.81 (1.77–1.84) |
| Hyperlipidemia | ||||
| No | 12,107 (10.4) | Ref | 5,891 (10.1) | Ref |
| Yes | 68,627 (13.2) | 1.08 (1.06–1.10) | 37,426 (13.4) | 1.20 (1.16–1.23) |
| Prior history of cardiovascular disease | ||||
| No | 57,398 (10.5) | Ref | 30,782 (10.6) | Ref |
| Yes | 23,336 (25.5) | 3.71 (3.65–3.77) | 12,535 (26.3) | 3.68 (3.60–3.76) |
Serum TSH in mIU/L, median and range; TSH was log-transformed and used as a time-varying covariate.
Serum FT4 in ng/dL, median and range; FT4 was used as a time-varying covariate.
Abbreviations: TSH, thyroid stimulating hormone; FT4, free thyroxine
Table 3.
Patient Characteristics Associated with Incident Heart Failure (Age as a Continuous Variable)
| Serum TSHa | Serum Free T4b | |||
|---|---|---|---|---|
| Patient Characteristics | N (%) patients with heart failure | Adjusted OR (95% CI) | N (%) patients with heart failure | Adjusted OR (95% CI) |
| Thyroid function test | -- | 1.04 (1.03–1.05) | -- | 1.00 (0.99–1.01) |
| Sex | ||||
| Male | 75,904 (13.5) | Ref | 40,318 (13.9) | Ref |
| Female | 4,830 (6.3) | 0.78 (0.76–0.81) | 2,999 (6.4) | 0.78 (0.75–0.81) |
| Age, Mean (SD) | 70.0 (11.6) | 1.04 (1.04–1.04) | 68.5 (11.8) | 1.04 (1.04–1.04) |
| Race | ||||
| White | 65,983 (13.0) | Ref | 35,615 (13.2) | Ref |
| Black | 5,562 (12.5) | 1.15 (1.12–1.18) | 3,535 (12.6) | 1.03 (0.99–1.06) |
| Other | 1,630 (11.2) | 0.95 (0.91–0.99) | 874 (11.1) | 0.89 (0.83–0.95) |
| Ethnicity | ||||
| Not Hispanic | 71,429 (12.9) | Ref | 38,715 (13.02) | Ref |
| Hispanic | 4,097 (11.8) | 0.81 (0.78–0.84) | 2,354 (12.4) | 0.96 (0.92–0.99) |
| Unknown | 5,208 (10.2) | 0.98 (0.92–1.03) | 2,248 (10.4) | 1.01 (0.94–1.09) |
| Smoking status | ||||
| Never | 10,711 (12.6) | Ref | 6,456 (12.30) | Ref |
| Current/Former | 56,121 (13.3) | 1.11 (1.08–1.13) | 30,041 (13.4) | 1.17 (1.13–1.20) |
| Unknown | 13,902 (10.5) | 1.02 (1.00–1.05) | 6,820 (11.0) | 1.03 (1.00–1.07) |
| Hypertension | ||||
| No | 6,085 (5.0) | Ref | 2,986 (4.6) | Ref |
| Yes | 74,649 (14.4) | 2.32 (2.25–2.38) | 40,331 (14.7) | 2.65 (2.55–2.75) |
| Diabetes mellitus | ||||
| No | 37,557 (9.9) | Ref | 19,736 (9.9) | Ref |
| Yes | 43,177 (16.7) | 1.78 (1.76–1.81) | 23,581 (17.0) | 1.81 (1.77–1.84) |
| Hyperlipidemia | ||||
| No | 12,107 (10.4) | Ref | 5,891 (10.1) | Ref |
| Yes | 68,627 (13.2) | 1.08 (1.06–1.10) | 37,426 (13.4) | 1.20 (1.16–1.23) |
| Prior history of cardiovascular disease | ||||
| No | 57,398 (10.5) | Ref | 30,782 (10.6) | Ref |
| Yes | 23,336 (25.5) | 3.71 (3.65–3.77) | 12,535 (26.3) | 3.68 (3.60–3.76) |
Serum TSH in mIU/L, median and range; TSH was log-transformed and used as a time-varying covariate.
Serum FT4 in ng/dL, median and range; FT4 was used as a time-varying covariate.
Abbreviations: TSH, thyroid stimulating hormone; FT4, free thyroxine
Figure 2 is a forest plot depicting the association between TSH and FT4 levels and incident heart failure in adjusted multivariable analyses when TSH and FT4 were analyzed as categorical variables. In these models, both under-replacement and over-replacement were associated with increased odds of incident heart failure compared to euthyroidism. This relationship held across all categories of under-replacement based on TSH values, but the odds were greatest in the most severely under-replaced patients (e.g., TSH > 5.5 - < 7.5 mIU/L, AOR, 1.24, 95% CI, 1.21–1.27; TSH 7.5- < 10 mIU/L, AOR, 1.40, 95% CI, 1.35–1.46; TSH 10–20 mIU/L, AOR, 1.57, 95% CI, 1.51–1.63; TSH > 20 mIU/L, AOR 1.93, 95% CI, 1.84–2.02). The same association was seen when thyroid hormone under-replacement was defined by a low FT4 (FT4 < 0.7 ng/dl, AOR, 1.08, 95% CI, 1.04–1.12). Mild over-replacement (TSH 0.1–0.5 mIU/L) was not associated with increased odds of incident heart failure (AOR 0.94, 95% CI, 0.91–0.97), whereas more significant over-replacement TSH <0.1 mIU/L) was significantly associated with increased odds of incident heart failure (AOR, 1.06, 95% CI, 1.01–1.12). Thyroid hormone over-replacement indicated by an elevated FT4 was similarly associated with increased odds of incident heart failure (AOR, 1.23, 95% CI, 1.14–1.32).
Figure 2. Association of TSH and FT4 with Incident Heart Failure.
This forest plot demonstrates the association between TSH and FT4 levels and incident heart failure, after adjusting for demographic and cardiovascular risk factors. A greater degree of thyroid hormone under-replacement is associated with greater odds of incident heart failure.
As shown in Figure 3, the risk of incident heart failure was cumulative over time for both thyroid hormone under-replacement and over-replacement, with under-replacement demonstrating a stronger association with incident heart failure. Specifically, when evaluating cumulative risk of incident heart failure by TSH category, there was a 5.8-fold increase in incident heart failure following 5 years of having a TSH value higher than 5.5 mIU/L (thyroid hormone under-replacement) and a 1.6-fold increase in incident heart failure following 5 years of having a TSH less than 0.1 mIU/L (thyroid hormone over-replacement) (Figure 3A). When evaluating cumulative risk of incident heart failure by FT4 category, there was a 1.3-fold increase in incident heart failure following 5 years of having a FT4 less than 0.7 ng/dL L (thyroid hormone under-replacement) and a 2.8-fold increase in incident heart failure following 5 years of having a FT4 higher than 1.9 ng/dL (thyroid hormone over-replacement) (Figure 3B). Notably, a TSH of 0.1–0.5 mIU/L was associated with a cumulative 0.7-fold decreased risk in incident heart failure.
Figure 3. Cumulative Risk of Incident Heart Failure by TSH and FT4 Category.
This figure shows that the risk of incident heart failure was cumulative over time for both thyroid hormone under-replacement and over-replacement, with under-replacement demonstrating a stronger association with incident heart failure.
DISCUSSION
In this population-based cohort study of adults using data from the largest integrated healthcare system in the United States, both over- and under-replacement with thyroid hormone were associated with an increased risk of incident heart failure compared to euthyroidism. Importantly, this association persisted after adjusting for salient demographic and traditional cardiovascular risk factors and the risk was cumulative over time. Furthermore, the degree of under-replacement strongly correlated with the risk of incident heart failure, such that the higher the TSH the higher the odds of incident heart failure.
Few prior studies examined the relationship between thyroid hormone treatment intensity and heart failure. Thayakaran et al. evaluated a cohort of 162,369 patients with hypothyroidism treated with thyroid hormone and found an association between TSH greater than 10 mIU/L and increased risk of incident heart failure (24). Notably, that study also found an association between TSH values below the reference range and decreased risk of incident heart failure. In contrast, our study found that while mild over-replacement was associated with decreased risk of incident heart failure, more significant over-replacement (TSH <0.1 mIU/L) was associated with an increased risk. This finding was further supported by the association of elevated FT4 with increased risk of incident heart failure. A single institution study retrospectively examined the association between under- and overtreatment of hypothyroidism on hospitalization outcomes in 231 patients with acute decompensated heart failure (7). In this study, patients who were markedly undertreated or overtreated had a significant increase in 90-day mortality. Contrary to this study, we utilized a national integrated health care system, which allowed for a much larger cohort size, and we evaluated the risk of incident heart failure rather than hospital outcomes in those with preexisting heart failure.
It is possible that known effects of endogenous hypothyroidism on the cardiovascular system, including impaired endothelial function, increased oxidative stress and impaired cardiac muscle relaxation may be extrapolated to explain the association between under-replacement with thyroid hormone and incident heart failure. A study by Rodondi et al. that included patients aged 65 years and older with subclinical hypothyroidism, demonstrated that in those with the highest TSH values (≥10.0 mIU/L), echocardiographic measurements showed higher early diastolic filling velocity which may be associated with higher risk of incident heart failure (25). In another study of patients with congenital hypothyroidism on lifelong levothyroxine therapy, there was evidence of left ventricular diastolic dysfunction, impaired exercise tolerance, and increased intima media thickening compared to healthy controls (26). Additionally, the number of episodes in which the TSH was outside the normal reference range for the study was an independent predictor of diastolic filling and cardiopulmonary performance indexes (26).
Our results add to a growing body of evidence suggesting that thyroid hormone treatment intensity may be a modifiable risk factor for important cardiovascular outcomes including atrial fibrillation (9), stroke (9), cardiovascular mortality (5), and as demonstrated in this study, incident heart failure (24). Given the widespread use of thyroid hormone replacement, these findings have implications for thyroid hormone management, emphasizing the importance of both proper patient selection for thyroid hormone initiation and appropriate laboratory monitoring to avoid inappropriate treatment. Individuals with persistently high or low serum TSH levels while on thyroid hormone replacement therapy should be comprehensively evaluated for factors affecting thyroid hormone absorption and availability, proper dosage and administration, adherence with treatment, and concurrent use of medications and comorbid conditions that may result in thyroid function abnormalities.
STRENGHTS AND LIMITATIONS
Our study has several strengths. First, this is a large, population-based study. Second, using data from a large, integrated health care system, allowed for robust analyses while adjusting for a comprehensive set of cardiovascular risk factors including smoking status, hypertension, diabetes, and hyperlipidemia, and for demographic data including age, sex, race, and ethnicity. Third, we analyzed serum TSH and FT4 as time-varying covariates which allowed us to better capture the effect of the variability of these laboratory measures over the course of the study period. Fourth, we reduced the risk of confounding by excluding individuals with a history of thyroid cancer or with prescriptions for medications known to affect thyroid function, and through the use of an internal comparator. Fifth, we evaluated the association between under- and over-replacement with thyroid hormone and the cumulative risk of incident heart failure over time, uniquely emphasizing the potential importance of the degree and duration of thyroid hormone under- and over-replacement.
Our study also has limitations. This is a retrospective observational study and findings cannot be used to establish a causal relationship between thyroid hormone treatment intensity and incident heart failure. Although we accounted for many potential sources of confounding, we were unable to account for certain known cardiovascular risk factors, such as obesity, due to missing data. However, we adjusted for a broad list of factors representing downstream effects of obesity, such as hypertension and diabetes, partly mitigating this limitation. Also owing to missing data, we were not able to ascertain the diagnosis of incident heart failure by echocardiograms. Additionally, because of reliance on diagnosis codes, we could not assess the degree to which cardiovascular risk factors, such as hypertension, diabetes and hyperlipidemia, were adequately managed. While we excluded individuals taking amiodarone and lithium as those are commonly used medications, we recognize that other medications, such as tyrosine kinase inhibitors or checkpoint inhibitor immunotherapy, as well as supplements such as biotin, can interfere with thyroid hormone metabolism and thyroid function tests. Furthermore, we acknowledge that FT3 was not included in the analyses, as typically this is not a marker that is tested in clinically evaluating adequate thyroid hormone replacement in patients. Finally, while the study population is enriched in individuals at high risk for heart failure, studies using VHA data underrepresent the U.S. population distribution of women and non-White racial and ethnic groups. However, because of the large sample size, our study included more than 70,000 females, more than 40,000 Black and more than 30,000 Hispanic patients.
CONCLUSIONS
In this large, population-based study of adult patients treated with thyroid hormone, we found an association between thyroid hormone treatment intensity and incident heart failure after adjusting for demographic and cardiovascular risk factors. The strongest association was with thyroid hormone under-replacement. Because the prevalence of thyroid hormone use is high among U.S. adults (13,27), the potential impact of this finding is significant. In addition to ensuring adequate monitoring and avoidance of thyroid hormone under-replacement, care must be taken to limit improper initiation and continuation of thyroid hormone that may lead to over-replacement, especially in older adults who may have higher baseline risk of cardiovascular disease and heart failure.
ACKNOWLEDGMENTS:
The authors would like to acknowledge Mr. David Reyes-Gastelum, MS, University of Michigan, for his help with the statistical analyses; he was compensated for his contributions. The authors would also like to acknowledge Ms. Brittany Gay, BA, University of Michigan, for her assistance with manuscript formatting and preparation for submission; she was compensated for her contribution.
Funding/Support:
This work was supported by a pilot grant co-funded by the Claude D. Pepper Older Americans Independence Center (OAIC), the Michigan Institute for Clinical and Health Research (MICHR) and the Michigan Biology of Cardiovascular Aging (M-BoCA) to Dr. Papaleontiou. Dr. Papaleontiou is funded by 4RF AG079833 from the National Institute on Aging of the National Institutes of Health.
Disclosure Summary:
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.
DATA AVAILABILITY:
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.
REFERENCES
- 1.Tsao CW, Aday AW, Almarzooq ZI, Alonso A, Beaton AZ, Bittencourt MS, Boehme AK, Buxton AE, Carson AP, Commodore-Mensah Y, Elkind MSV, Evenson KR, Eze-Nliam C, Ferguson JF, Generoso G, Ho JE, Kalani R, Khan SS, Kissela BM, Knutson KL, Levine DA, Lewis TT, Liu J, Loop MS, Ma J, Mussolino ME, Navaneethan SD, Perak AM, Poudel R, Rezk-Hanna M, Roth GA, Schroeder EB, Shah SH, Thacker EL, VanWagner LB, Virani SS, Voecks JH, Wang N-Y, Yaffe K, Martin SS. Heart Disease and Stroke Statistics—2022 Update: A Report From the American Heart Association. Circulation. 2022;145(8):e153–e639. [DOI] [PubMed] [Google Scholar]
- 2.Heidenreich PA, Bozkurt B, Aguilar D, Allen LA, Byun JJ, Colvin MM, Deswal A, Drazner MH, Dunlay SM, Evers LR, Fang JC, Fedson SE, Fonarow GC, Hayek SS, Hernandez AF, Khazanie P, Kittleson MM, Lee CS, Link MS, Milano CA, Nnacheta LC, Sandhu AT, Stevenson LW, Vardeny O, Vest AR, Yancy CW. 2022 AHA/ACC/HFSA Guideline for the Management of Heart Failure: A Report of the American College of Cardiology/American Heart Association Joint Committee on Clinical Practice Guidelines. Circulation. 2022;145(18):e895–e1032. [DOI] [PubMed] [Google Scholar]
- 3.Klein I, Ojamaa K. Thyroid hormone and the cardiovascular system. N Engl J Med. 2001;344(7):501–509. [DOI] [PubMed] [Google Scholar]
- 4.Biondi B Mechanisms in Endocrinology: Heart failure and thyroid dysfunction. European Journal of Endocrinology. 2012;167(5):609–618. [DOI] [PubMed] [Google Scholar]
- 5.Evron JM, Hummel SL, Reyes-Gastelum D, Haymart MR, Banerjee M, Papaleontiou M. Association of Thyroid Hormone Treatment Intensity With Cardiovascular Mortality Among US Veterans. JAMA Netw Open. 2022;5(5):e2211863. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Flynn RW, Bonellie SR, Jung RT, MacDonald TM, Morris AD, Leese GP. Serum thyroid-stimulating hormone concentration and morbidity from cardiovascular disease and fractures in patients on long-term thyroxine therapy. J Clin Endocrinol Metab. 2010;95(1):186–193. [DOI] [PubMed] [Google Scholar]
- 7.Kagansky D, Or K, Elkan M, Koren S, Koren R. The effect of over- and undertreatment of hypothyroidism on hospitalization outcomes of patients with decompensated heart failure. J Investig Med. 2023:10815589231162542. [DOI] [PubMed] [Google Scholar]
- 8.Lillevang-Johansen M, Abrahamsen B, Jorgensen HL, Brix TH, Hegedus L. Duration of over- and under-treatment of hypothyroidism is associated with increased cardiovascular risk. Eur J Endocrinol. 2019;180(6):407–416. [DOI] [PubMed] [Google Scholar]
- 9.Papaleontiou M, Levine DA, Reyes-Gastelum D, Hawley ST, Banerjee M, Haymart MR. Thyroid Hormone Therapy and Incident Stroke. J Clin Endocrinol Metab. 2021;106(10):e3890–e3900. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Ettleson MD, Ibarra S, Wan W, Peterson S, Laiteerapong N, Bianco AC. Demographic, Healthcare Access, and Dietary Factors Associated With Thyroid Hormone Treatments for Hypothyroidism. J Clin Endocrinol Metab. 2023;108(12):e1614–e1623. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.IQVIA. Medicine Use and Spending in the U.S.: a review of 2018 and outlook to 2023. https://www.iqvia.com/insights/the-iqvia-institute/reports/medicine-use-and-spending-in-the-us-a-review-of-2018-and-outlook-to-2023. Accessed September 15, 2024.
- 12.GoodRx. Top 10 Prescription Medications in the U.S. (November 2021). https://www.goodrx.com/drug-guide. Accessed September 15, 2024.
- 13.Kantor ED, Rehm CD, Haas JS, Chan AT, Giovannucci EL. 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]
- 14.Somwaru LL, Arnold AM, Joshi N, Fried LP, Cappola AR. 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.Mammen JS, McGready J, Oxman R, Chia CW, Ladenson PW, Simonsick EM. Thyroid Hormone Therapy and Risk of Thyrotoxicosis in Community-Resident Older Adults: Findings from the Baltimore Longitudinal Study of Aging. Thyroid. 2015;25(9):979–986. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Parle JV, Franklyn JA, Cross KW, Jones SR, Sheppard MC. Thyroxine prescription in the community: serum thyroid stimulating hormone level assays as an indicator of undertreatment or overtreatment. Br J Gen Pract. 1993;43(368):107–109. [PMC free article] [PubMed] [Google Scholar]
- 17.Lillevang-Johansen M, Abrahamsen B, Jorgensen HL, Brix TH, Hegedus L. Over- and Under-Treatment of Hypothyroidism Is Associated with Excess Mortality: A Register-Based Cohort Study. Thyroid. 2018;28(5):566–574. [DOI] [PubMed] [Google Scholar]
- 18.U.S. Department of Veterans Affairs. VA Informatics and Computing Infrastructure (VINCI). https://www.research.va.gov/programs/vinci/default.cfm. Accessed October 2, 2024.
- 19.Veterans Health Administration. https://www.va.gov/health/. Accessed October 5, 2024.
- 20.Fihn SD, Francis J, Clancy C, Nielson C, Nelson K, Rumsfeld J, Cullen T, Bates J, Graham GL. Insights from advanced analytics at the Veterans Health Administration. Health Aff (Millwood). 2014;33(7):1203–1211. [DOI] [PubMed] [Google Scholar]
- 21.Kilbourne AM, Elwy AR, Sales AE, Atkins D. Accelerating Research Impact in a Learning Health Care System: VA’s Quality Enhancement Research Initiative in the Choice Act Era. Med Care. 2016;55(7 Suppl 1):S4–S12. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Kizer KW, Demakis JG, Feussner JR. Reinventing VA health care: systematizing quality improvement and quality innovation. Med Care. 2000;38(6 Suppl 1):I7–16. [PubMed] [Google Scholar]
- 23.Ev Elm, Altman DG, Egger M, SJ Pocock, Gøtzsche PC, Vandenbroucke JP. The Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) Statement: guidelines for reporting observational studies. https://www.equator-network.org/reporting-guidelines/strobe/. Accessed October 5, 2024. [DOI] [PMC free article] [PubMed]
- 24.Thayakaran R, Adderley NJ, Sainsbury C, Torlinska B, Boelaert K, Šumilo D, Price M, Thomas GN, Toulis KA, Nirantharakumar K. Thyroid replacement therapy, thyroid stimulating hormone concentrations, and long term health outcomes in patients with hypothyroidism: longitudinal study. BMJ. 2019;366:l4892. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 25.Rodondi N, Bauer DC, Cappola AR, Cornuz J, Robbins J, Fried LP, Ladenson PW, Vittinghoff E, Gottdiener JS, Newman AB. Subclinical thyroid dysfunction, cardiac function, and the risk of heart failure. The Cardiovascular Health study. J Am Coll Cardiol. 2008;52(14):1152–1159. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 26.Salerno M, Oliviero U, Lettiero T, Guardasole V, Mattiacci DM, Saldamarco L, Capalbo D, Lucariello A, Sacca L, Cittadini A. Long-term cardiovascular effects of levothyroxine therapy in young adults with congenital hypothyroidism. J Clin Endocrinol Metab. 2008;93(7):2486–2491. [DOI] [PubMed] [Google Scholar]
- 27.Ross JS, Rohde S, Sangaralingham L, Brito JP, Choi L, Dutcher SK, Graham DJ, Jenkins MR, Lipska KJ, Mendoza M, Qiang Y, Wang Z, Wu Y, Yao X, Shah ND. 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]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Data Availability Statement
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.