Abstract
Background:
Subclinical hypothyroidism has been linked to adverse renal outcomes, but it is unknown whether treatment of subclinical hypothyroidism reduces the risk of these adverse outcomes. We sought to assess whether treatment of subclinical hypothyroidism with levothyroxine is associated with a decreased risk of developing adverse renal outcomes.
Methods:
We conducted a population-based cohort study using a prevalent new-user design. Using the Clinical Practice Research Datalink in the United Kingdom, we assembled a study cohort of individuals aged 18 years and older with newly diagnosed subclinical hypothyroidism between 1998 and 2018. We matched individuals with subclinical hypothyroidism treated with levothyroxine to nontreated individuals with subclinical hypothyroidism based on age, sex, calendar time, duration of subclinical hypothyroidism, and time-conditional propensity score. We included several confounders, including data on demographics, clinical diagnoses, biochemical measures, lifestyle, and prescriptions, in the time-conditional propensity score estimate. The primary outcome was a composite end point of end-stage renal disease (requiring dialysis) and lowering of the estimated glomerular filtration rate of 50% or greater.
Results:
We matched 76 697 individuals with subclinical hypothyroidism treated with levothyroxine to 76 697 nontreated individuals. The mean age of the study cohort was 63 years, and 77% of patients were female. During a median follow-up time of 1.6 years (levothyroxine treated: 1.0, interquartile range [IQR] 0.4–2.7, yr; nontreated: 2.5, IQR 0.9–5.7, yr), 2426 adverse renal events occurred among nontreated individuals (incidence rate 8.35, 95% confidence interval [CI] 8.02–8.69, per 1000 person-years), and 1255 adverse renal events occurred among levothyroxine-treated individuals (incidence rate 7.93, 95% CI 7.50–8.39, per 1000 person-years). Treatment with levothyroxine was not associated with the risk of adverse renal outcomes among people with subclinical hypothyroidism (adjusted hazard ratio 0.97, 95% CI 0.90–1.04).
Interpretation:
In this population-based cohort study, there was no evidence that levothyroxine treatment of subclinical hypothyroidism reduced the risk of adverse renal outcomes. Future studies in different populations and with longer duration of follow-up are needed to confirm these findings.
Subclinical hypothyroidism has a global prevalence of 5%–10%,1,2 with a higher prevalence (about 18%) among people with chronic kidney disease.3 Studies have shown that the prevalence of subclinical hypothyroidism increases with worsening of renal function.4 Moreover, there is an association between higher thyrotropin levels and lower estimated glomerular filtration rate (eGFR), after accounting for differences in age, sex, and comorbidities.5 Animal studies have shown that hypothyroidism affects renal tubules, including the proximal tubule, medullary thick ascending limb, and the convoluted tubules, which were found to be shorter and thinner in rats with pharmacologically induced hypothyroidism.6,7 Animal studies have also shown that hypothyroidism results in decreased transporting functions in ionic transporters, cotransporters, exchangers, and channels.8 Although there is a paucity of research assessing the effect of hypothyroidism on kidney function, the authors of a retrospective study involving 16 individuals with known hypothyroidism who later developed renal disorders noted that all of these patients had various forms of glomerulonephritis, which resulted in reduced renal function.9,10 Consistent with this finding, a recent cross-sectional study in a Japanese cohort showed that higher thyrotropin levels were positively correlated with prevalence of chronic kidney disease.11
Current guidelines recommend treatment of subclinical hypothyroidism with levothyroxine when thyrotropin levels are greater than 10 mIU/L given that prior studies have shown that there is an increased risk of cardiovascular disease and cardiovascular-related death among people with subclinical hypothyroidism with thyrotropin levels greater than 10 mIU/L.12 However, controversy remains as to whether treating subclinical hypothyroidism when the thyrotropin level is above normal range but less than 10 mIU/L provides any health benefits. Given the potential link between thyroid hormone levels and the development of chronic kidney disease, we sought to assess whether treatment of subclinical hypothyroidism with levothyroxine is associated with a decreased risk of developing adverse renal outcomes.
Methods
This population-based cohort study was conducted using data from the Clinical Practice Research Datalink (CPRD) Aurum, which is a large clinical database from the United Kingdom.
Data sources
The CPRD contains the records of approximately 60 million patients who are followed by general practitioners in the UK, including data on demographics, clinical diagnoses, clinical measurements, lifestyle variables, and prescriptions.13 Diagnoses are documented using Read Codes and Systematized Nomenclature of Medicine Clinical Terms. We linked CPRD to the Hospital Episode Statistics database, which contains information on hospital admissions and is restricted to 85% of English practices in the UK. Finally, we linked CPRD to the Office for National Statistics database, which contains the electronic death certificate with the cause of death for all citizens in the UK.
Study population
We assembled a base cohort consisting of people aged 18 years and older, identified with subclinical hypothyroidism defined as having a thyrotropin level between 5 mIU/L and 10 mIU/L from Apr. 1, 1998, to Dec. 31, 2018. We first selected individuals with at least 2 thyrotropin readings between 5 mIU/L and 10 mIU/L on different dates (i.e., not necessarily consecutive) within 1 year of each other during the study period. Base cohort entry was determined by the date of the second measured thyrotropin level between 5 mIU/L and 10 mIU/L.
We then excluded individuals with less than 12 months of recorded medical history in the CPRD before base cohort entry, those with date inconsistencies, those with a diagnostic code for end-stage renal disease in the year before cohort entry, and those without follow-up. We excluded individuals who received prescriptions for thionamides (i.e., methimazole or propylthiouracil), levothyroxine, and amiodarone during the 12 months before base cohort entry. Women who were pregnant at any time during the year before base cohort entry were also excluded as they can have postpartum thyroiditis, which may result in temporary subclinical hypothyroidism or hypothyroidism.14 Finally, we excluded individuals with a history of pituitary disorder at any time before base cohort entry, as this condition renders the thyrotropin unreliable in determining thyroid hormone levels.15
Study design
We used a prevalent new-user design with time-conditional propensity score matching to adjust for confounding.16,17 Individuals in the base cohort who were initiated on levothyroxine were used to create time-based exposure sets including all individuals not given levothyroxine up to this point in time, with the same calendar time, age (± 3 yr), sex, and duration of subclinical hypothyroidism (± 30 d), and with a general practitioner visit within 30 calendar days of the date of levothyroxine initiation for the treated individual. A visit with the general practitioner was defined by the date when the patient had medical and prescription codes on the same day. Conditional logistic regression, with models conditioned on the exposure set number, was used to estimate the time-conditional propensity scores including covariates. To verify the positivity assumption, we ensured there was propensity score overlap between treated and nontreated individuals (i.e., the propensity score of each levothyroxine-treated individual was within bounds of the propensity score distribution for nontreated individuals of each exposure set), and exposure sets that had violations were excluded. To create our study cohort, we matched each levothyroxine-treated individual used to define the exposure set to a nontreated individual (1:1) in the exposure set using nearest neighbour matching without replacement based on their time-conditional propensity scores (Appendix 1, Supplementary Figure 1, available at www.cmaj.ca/lookup/doi/10.1503/cmaj.250356/tab-related-content). Study cohort entry was defined as the date of levothyroxine initiation and as the corresponding physician visit for those who were not initiated on levothyroxine. Individuals were followed until the occurrence of the outcome of interest; change in exposure status (i.e., treatment crossover with discontinuation of levothyroxine, defined by a treatment gap of > 30 days between the end of duration of one prescription and the next one, or the initiation of levothyroxine in individuals who were not initially treated); death; end of registration with the general practice; end of the study period (Dec. 31, 2019); a thyrotropin level greater than 10 mIU/L on 2 consecutive tests; initiation of methimazole, propylthiouracil, or amiodarone; pregnancy; or diagnosis of a pituitary disorder, whichever occurred first.
Exposure
We used an as-treated approach in which individuals in the study cohort were followed until treatment crossover.
Outcome
Our primary outcome was adverse renal events, defined as a composite end point that included end-stage renal disease (i.e., requiring dialysis) and decrease of eGFR of 50% or greater. End-stage renal disease was identified using relevant diagnostic codes in the CPRD and International Statistical Classification of Diseases and Related Health Problems, 10th revision codes in the Hospital Episode Statistics. Estimated glomerular filtration rate that decreased by 50% or greater during follow-up compared with the baseline eGFR at study cohort entry was considered an event.
Covariates
The time-conditional propensity score included prespecified potential confounders. We also adjusted for quintiles of the English Indices of Deprivation 2010 as a proxy for socioeconomic status. We used multiple imputation to estimate missing values for ethnicity, eGFR, body mass index, and smoking.18–20
Statistical analysis
We estimated descriptive statistics using standardized mean differences to summarize patient characteristics of each exposure group before and after time-conditional propensity score matching. The incidence rates for each outcome were estimated along with their 95% confidence interval (CI) based on the Poisson distribution. We used age-stratified Cox proportional hazards models (< 70 yr v. ≥ 70 yr) to estimate the hazard ratio (HR) and 95% CI for the primary outcome of adverse renal events associated with levothyroxine compared with no treatment among matched individuals with subclinical hypothyroidism. We used a robust sandwich estimator to account for the same individual contributing person-time to the exposed (i.e., treated with levothyroxine) and unexposed (i.e., no treatment) groups.21 We imputed missing data using chained equations to create 5 imputed data sets and combined results using Rubin’s rules.19
Four secondary analyses were conducted. We conducted subgroup analyses by age (< 70 yr and ≥ 70 yr), sex, history of cardiovascular disease (i.e., myocardial infarction, unstable angina, chronic stable angina, coronary artery disease, peripheral arterial disease, and cerebrovascular disease), and prior diagnosis of chronic renal insufficiency. To assess the robustness of our findings, we conducted 6 sensitivity analyses. First, to assess exposure misclassification, we conducted the primary analysis with grace periods (i.e., the time allowed between the end of one prescription and the start of the next for a patient to be considered continuing treatment) of 0 days and 90 days. Second, we conducted a modified as-treated analysis in which the follow-up time was limited to a maximum of 1 year. Third, we conducted a modified intention-to-treat (ITT) analysis in which levothyroxine-treated individuals were not censored at discontinuation of treatment and nontreated individuals who were initiated on levothyroxine were censored to avoid double counting person-time, given that we used a prevalent new-user design. In the modified ITT analysis, we also limited the maximum follow-up time to be 1 year. Fourth, we conducted our primary analysis using a base cohort of people with subclinical hypothyroidism defined as having 2 consecutive thyrotropin levels between 5 mIU/L and 10 mIU/L within 1 year. Fifth, we repeated the primary analysis using a base cohort composed of individuals with subclinical hypothyroidism defined as having at least 2 thyrotropin levels between 5 mIU/L and 10 mIU/L and a normal free thyroxine level (7.9–24 pmol/L). Finally, we used inverse probability of censoring weighting to account for potential informative censoring, using 5 censoring models: thyrotropin levels rising greater than 10 mIU/L, death, levothyroxine discontinuation, levothyroxine initiation among people who were not treated, and administrative censoring (i.e., end of study period or end of patient’s registration with the practice or date of the last collection of the data from a practice).
Ethics approval
The study protocol was approved by the CPRD Research Data Governance (protocol no. 20_055) and the Biomedical Research Ethics Committee of the Jewish General Hospital, Montréal, Canada (no. 2021-2190). We reported our findings in accordance with the Reporting of Studies Conducted Using Observational Routinely-collected Data guidelines.22
Results
The base cohort included 176 813 individuals with subclinical hypothyroidism. From the base cohort, we identified 90 997 individuals who initiated levothyroxine during follow-up, and 90 600 of these individuals had at least 1 individual as a comparator in their exposure set. Before imputation and time-conditional propensity score matching, we compared these 90 600 levothyroxine-treated individuals with 90 600 nontreated individuals, with 1 individual randomly selected from each exposure set (Figure 1). People treated with levothyroxine had a higher mean thyrotropin level at cohort entry than nontreated individuals (mean thyrotropin level 6.8 mIU/L v. 6.5 mIU/L); were less likely to have diabetes; had a lower Charlson Comorbidity Index score; were less likely to be prescribed β-blockers, antihypertensive medications, lipid-lowering medications, antidiabetic medications such as metformin and sulfonylureas, antipsychotics, antidepressants, nonsteroidal inflammatory drugs, and opioids; and had fewer physician visits in the year before study cohort entry (Appendix 1, Supplementary Table 1). After imputation and matching on time-conditional propensity score, the size of the 5 imputed study cohorts ranged from 153 296 to 153 394 individuals. The following findings presented came from the first imputed cohort. A total of 76 697 individuals who were treated with levothyroxine were matched to 76 697 individuals who were not treated. The covariates were well balanced between the 2 exposure groups with mean standardized differences of less than 0.1 (Table 1). The mean age was 63 (standard deviation 16.1) years, and approximately 77% were female. The median follow-up time was 1.0 (interquartile range [IQR] 0.4–2.7) year among people treated with levothyroxine and 2.5 (IQR 0.9–5.7) years among people who were not treated. During an overall median follow-up time of 1.6 years, 1255 adverse renal events occurred among people treated with levothyroxine (incidence rate 7.93, 95% CI 7.50–8.39 per 1000 person-years) and 2426 adverse renal events among people who were not treated (incidence rate 8.35, 95% CI 8.02–8.69 per 1000 person-years). The average thyrotropin level among people with subclinical hypothyroidism at the end of follow-up who received levothyroxine was 4.24 mIU/L, and the average thyrotropin level was 6.88 mIU/L among individuals who were not treated (Appendix 1, Supplementary Table 2). Levothyroxine treatment was not associated with the risk of adverse renal outcomes among people with subclinical hypothyroidism compared with no treatment (adjusted HR 0.97, 95% CI 0.90–1.04) (Table 2).
Figure 1:
Flowchart showing assembly of study cohort of individuals with subclinical hypothyroidism using the first imputed data set. *A total of 13 499 were excluded because their data did not meet the quality standards of the CPRD (e.g., missing sex); 4489 were excluded from practices with problematic data, identified by CPRD. Note: CPRD = Clinical Practice Research Datalink, HES = Hospital Episode Statistics, ONS = Office for National Statistics, TCPS = time-conditional propensity score.
Table 1:
Baseline characteristics of levothyroxine-treated and untreated individuals with subclinical hypothyroidism after time-conditional propensity score matching in the first imputed study cohort
| Characteristic | No. (%) of patients* | Standardized mean difference | |
|---|---|---|---|
| Levothyroxine treatment n = 76 697 |
No levothyroxine treatment n = 76 697 |
||
| Age, yr, mean ± SD | 62.6 ± 16.1 | 62.7 ± 16.1 | 0.00 |
| Age group, yr | |||
| < 40 | 6781 (8.8) | 6751 (8.8) | 0.00 |
| 40–69 | 41 266 (53.8) | 41 130 (53.6) | 0.00 |
| ≥ 70 | 28 650 (37.4) | 28 816 (37.6) | 0.00 |
| Sex | |||
| Female | 58 732 (76.6) | 58 732 (76.6) | 0.00 |
| Male | 17 965 (23.4) | 17 965 (23.4) | 0.00 |
| Ethnicity† | |||
| White | 69 150 (90.2) | 68 417 (89.2) | 0.03 |
| Other | 7547 (9.8) | 8280 (10.8) | 0.03 |
| Thyrotropin, mIU/L,† mean ± SD | 6.7 ± 1.0 | 6.6 ± 1.0 | 0.15 |
| Calendar year of cohort entry date† | |||
| 1998–2001 | 3013 (3.9) | 3005 (3.9) | 0.00 |
| 2002–2005 | 11 677 (15.2) | 11 640 (15.2) | 0.00 |
| 2006–2009 | 17 101 (22.3) | 17 112 (22.3) | 0.00 |
| 2010–2013 | 20 078 (26.2) | 20 078 (26.2) | 0.00 |
| 2014–2017 | 17 997 (23.5) | 18 004 (23.5) | 0.00 |
| 2018–2019 | 6831 (8.9) | 6858 (8.9) | 0.00 |
| Quintiles of Index of Multiple Deprivation† | |||
| 1 (most deprived) | 13 073 (17.0) | 12 869 (16.8) | 0.01 |
| 2 | 13 183 (17.2) | 13 036 (17.0) | 0.01 |
| 3 | 16 085 (21.0) | 15 925 (20.8) | 0.01 |
| 4 | 16 677 (21.7) | 16 764 (21.9) | 0.00 |
| 5 (least deprived) | 17 679 (23.1) | 18 103 (23.6) | 0.01 |
| Comorbidities‡ | |||
| Diabetes | 9721 (12.7) | 10 528 (13.7) | 0.03 |
| Renal insufficiency | 10 056 (13.1) | 10 356 (13.5) | 0.01 |
| Cancer | 5421 (7.1) | 5398 (7.0) | 0.00 |
| COPD | 3755 (4.9) | 3785 (4.9) | 0.00 |
| Dementia | 2232 (2.9) | 2336 (3.0) | 0.01 |
| Liver disease | 150 (0.2) | 147 (0.2) | 0.00 |
| Obstructive sleep apnea | 322 (0.4) | 312 (0.4) | 0.00 |
| Rheumatoid disease | 983 (1.3) | 1030 (1.3) | 0.01 |
| Myocardial infarction | 3377 (4.4) | 3453 (4.5) | 0.00 |
| Unstable angina | 489 (0.6) | 513 (0.7) | 0.00 |
| Angina | 5152 (6.7) | 5295 (6.9) | 0.01 |
| Coronary artery disease, | 6154 (8.0) | 6268 (8.2) | 0.01 |
| Peripheral arterial disease | 1680 (2.2) | 1724 (2.2) | 0.00 |
| Atrial fibrillation | 2130 (2.8) | 2278 (3.0) | 0.01 |
| Heart failure | 3474 (4.5) | 3664 (4.8) | 0.01 |
| Cerebral vascular accident or transient ischemic attack | 5611 (7.3) | 5860 (7.6) | 0.01 |
| Osteoporosis | 2682 (3.5) | 2734 (3.6) | 0.00 |
| Falls | 6296 (8.2) | 6218 (8.1) | 0.00 |
| Charlson Comorbidity Index,† mean ± SD | 0.9 ± 1.4 | 0.9 ± 1.4 | 0.02 |
| Charlson Comorbidity Index grouping‡ | |||
| 0 | 43 250 (56.4) | 42 629 (55.6) | 0.02 |
| 1 | 15 048 (19.6) | 14 983 (19.5) | 0.00 |
| 2 | 8781 (11.4) | 8837 (11.5) | 0.00 |
| 3 | 5381 (7.0) | 5703 (7.4) | 0.02 |
| ≥ 4 | 4237 (5.5) | 4545 (5.9) | 0.02 |
| Medications§ | |||
| Oral contraceptives | 4687 (6.1) | 4661 (6.1) | 0.00 |
| Hormone replacement therapy | 4695 (6.1) | 4429 (5.8) | 0.01 |
| β-adrenoceptor blockers | 13 893 (18.1) | 14 262 (18.6) | 0.01 |
| Nitrates | 4288 (5.6) | 4330 (5.6) | 0.00 |
| Antihypertensives | 24 801 (32.3) | 25 402 (33.1) | 0.02 |
| Lipid-lowering therapy | 24 335 (31.7) | 24 851 (32.4) | 0.01 |
| Anticoagulants | 3480 (4.5) | 3722 (4.9) | 0.01 |
| Antiplatelets | 16 052 (20.9) | 16 273 (21.2) | 0.01 |
| Diuretics | 10 443 (13.6) | 10 641 (13.9) | 0.01 |
| Acarbose | 41 (0.1) | 48 (0.1) | 0.00 |
| Metformin | 6773 (8.8) | 7440 (9.7) | 0.03 |
| Sulfonylureas | 3337 (4.4) | 3680 (4.8) | 0.02 |
| Meglitinides | 58 (0.1) | 63 (0.1) | 0.00 |
| Dipeptidyl peptidase-4 inhibitors | 891 (1.2) | 1011 (1.3) | 0.01 |
| GLP-1 receptor agonists | 203 (0.3) | 230 (0.3) | 0.01 |
| Thiazolidinedione | 764 (1.0) | 861 (1.1) | 0.01 |
| SGLT2 inhibitors | 145 (0.2) | 163 (0.2) | 0.01 |
| Insulin | 2626 (3.4) | 2754 (3.6) | 0.01 |
| Antipsychotics | 2060 (2.7) | 2305 (3.0) | 0.02 |
| Antiarrhythmics | 4283 (5.6) | 4411 (5.8) | 0.01 |
| Antidepressants | 17749 (23.1) | 17253 (22.5) | 0.02 |
| NSAIDs | 29 453 (38.4) | 29 200 (38.1) | 0.01 |
| Opioids | 7475 (9.7) | 7305 (9.5) | 0.01 |
| No. of physician visits in the year before cohort entry§ | |||
| Mean ± SD | 8.6 ± 7.2 | 8.2 ± 7.2 | 0.06 |
| Median (IQR) | 6.0 (4.0–11.0) | 6.0 (3.0–11.0) | |
| 0–1 | 3274 (4.3) | 4009 (5.2) | 0.05 |
| 2 | 6038 (7.9) | 7240 (9.4) | 0.06 |
| ≥ 3 | 67 385 (87.9) | 65 448 (85.3) | 0.07 |
| Body mass index‡ | |||
| Underweight (< 18.5) | 1068 (1.4) | 1108 (1.4) | 0.00 |
| Normal (18.5 to < 25) | 22 636 (29.5) | 22 627 (29.5) | 0.00 |
| Overweight (25 to < 30) | 27 230 (35.5) | 27 162 (35.4) | 0.00 |
| Obese (≥ 30) | 25 763 (33.6) | 25 800 (33.6) | 0.00 |
| Smoking‡ | |||
| Not currently | 47 156 (61.5) | 47 155 (61.5) | 0.00 |
| Currently | 29 541 (38.5) | 29 542 (38.5) | 0.00 |
| eGFR, mL/min/1.73 m2‡ | |||
| Mean ± SD | 74.1 ± 20.9 | 74. 1 ± 21.1 | 0.00 |
| < 15 | 100 (0.1) | 129 (0.2) | 0.01 |
| 15–29 | 1179 (1.5) | 1118 (1.5) | 0.01 |
| 30–44 | 4839 (6.3) | 4866 (6.3) | 0.00 |
| 45–59 | 11 999 (15.6) | 12 134 (15.8) | 0.00 |
| 60–89 | 39 449 (51.4) | 39 142 (51.0) | 0.01 |
| ≥ 90 | 19 131 (24.9) | 19 308 (25.2) | 0.01 |
| Excessive alcohol use¶ | 2819 (3.7) | 2949 (3.8) | 0.01 |
Note: COPD = chronic obstructive pulmonary disease, eGFR = estimated glomerular filtration rate, GLP-1 = glucagon-like peptide-1, IQR = interquartile range, NSAID = nonsteroidal anti-inflammatory drug, SD = standard deviation, SGLT = sodium glucose cotransporter.
Unless stated otherwise.
Measured at study cohort entry.
Measured in the 5 years before or at study cohort.
Measured in the year before study cohort entry.
Diagnosis of alcohol abuse or alcohol use disorder.
Table 2:
Risk of adverse renal outcomes among people with subclinical hypothyroidism treated with levothyroxine, compared with no treatment*
| Variable | No. of patients | Events | Person-years | Incidence rate (95% CI)† | Matched HR (95% CI)‡ | Matched and adjusted HR (95% CI)§ |
|---|---|---|---|---|---|---|
| No treatment | 76 697 | 2426 | 290 672.8 | 8.35 (8.02–8.69) | Ref. | Ref. |
| Levothyroxine treatment | 76 697 | 1255 | 158 167.2 | 7.93 (7.50–8.39) | 0.93 (0.87–1.00) | 0.97 (0.90–1.04) |
Note: CI = confidence interval, HR = hazard ratio, Ref. = reference category.
The number of patients, events, person-years, and the incidence rate presented here came from the first imputed matched cohort.
Per 1000 person-years.
Individuals were matched on calendar time, subclinical hypothyroidism duration, sex, age, and time-conditional propensity scores. To account for differences of baseline hazard of major adverse renal events, we ran age-stratified Cox models (< 70 yr v. ≥ 70 yr).
In addition to matching on calendar time, subclinical hypothyroidism duration, sex, age, and time-conditional propensity scores, analyses were adjusted for time-conditional propensity scores. To account for differences of baseline hazard of major adverse renal events, we ran age-stratified Cox models (< 70 yr v. ≥ 70 yr).
In the secondary analyses, the association between levothyroxine treatment and the risk of adverse renal outcomes compared with no treatment did not differ meaningfully in subgroup analyses by age, sex, prior history of cardiovascular disease, and prior history of renal insufficiency (Table 3). Our primary result remained consistent in our sensitivity analyses (Appendix 1, Supplementary Table 3).
Table 3:
Subgroup analyses assessing the risk of adverse renal events among people with subclinical hypothyroidism treated with levothyroxine compared with no treatment by age, sex, history of cardiovascular disease, and history of chronic kidney disease*
| Variable | No. of patients | Events | Person-years | Incidence rate (95% CI)† | Matched HR (95% CI)‡ | Matched and adjusted HR (95% CI)§ |
|---|---|---|---|---|---|---|
| Age, yr | ||||||
| < 70 | ||||||
| No treatment | 47 735 | 762 | 190 478.4 | 4.00 (3.72–4.29) | Ref. | Ref. |
| Levothyroxine treatment | 47 735 | 304 | 96 502.3 | 3.15 (2.81–3.52) | 0.87 | 0.92 (0.80–1.05)¶ |
| ≥ 70 | ||||||
| No treatment | 29 098 | 1626 | 100 553.5 | 16.17 (15.39–16.98) | Ref. | Ref. |
| Levothyroxine treatment | 29 098 | 963 | 61 650.2 | 15.62 (14.65–16.64) | 0.98 | 1.00 (0.91–1.09) |
| Sex | ||||||
| Male | ||||||
| No treatment | 17 965 | 795 | 66 481.8 | 11.96 (11.14–12.82) | Ref. | Ref. |
| Levothyroxine treatment | 17 965 | 404 | 33 836.8 | 11.94 (10.80–13.16) | 0.96 | 0.99 (0.88–1.12) |
| Female | ||||||
| No treatment | 58 732 | 1631 | 224 191.0 | 7.28 (6.93–7.64) | Ref. | Ref. |
| Levothyroxine treatment | 58 732 | 851 | 124 330.4 | 6.84 (6.39–7.32) | 0.93 | 0.97 (0.89–1.06) |
| History of cardiovascular disease | ||||||
| Prior history of cardiovascular disease | ||||||
| No treatment | 13 957 | 898 | 48 101.9 | 18.67 (17.47–19.93) | Ref. | Ref. |
| Levothyroxine treatment | 13 562 | 485 | 27 423.8 | 17.69 (16.15–19.33) | 0.94 | 0.95 (0.85–1.06) |
| No prior history of cardiovascular disease | ||||||
| No treatment | 62 740 | 1528 | 242 570.9 | 6.30 (5.99–6.62) | Ref. | Ref. |
| Levothyroxine treatment | 63 135 | 770 | 130 743.4 | 5.89 (5.48–6.32) | 0.94 | 0.98 (0.90–1.08) |
| History of renal insufficiency | ||||||
| Prior history of renal insufficiency | ||||||
| No treatment | 10 356 | 807 | 31 694.7 | 25.46 (23.74–27.28) | Ref. | Ref. |
| Levothyroxine treatment | 10 056 | 512 | 19 217.9 | 26.64 (24.38–29.05) | 0.99 | 1.03 (0.92–1.16) |
| No prior history of renal insufficiency | ||||||
| No treatment | 66 341 | 1619 | 258 978.2 | 6.25 (5.95–6.56) | Ref. | Ref. |
| Levothyroxine treatment | 66 641 | 743 | 138 949.3 | 5.35 (4.97–5.75) | 0.90 | 0.93 (0.85–1.02) |
Note: CI = confidence interval, HR = hazard ratio, Ref. = reference category.
The number of patients, events, and person-years, and the incidence rate presented here came from the first imputed matched cohort.
Per 1000 person-years.
Individuals were matched on calendar time, subclinical hypothyroidism duration, sex, age, and time-conditional propensity scores. To account for differences of baseline hazard of major adverse renal events, we ran age-stratified Cox models (< 70 yr v. ≥ 70 yr).
In addition to matching on calendar time, subclinical hypothyroidism duration, sex, age, and time-conditional propensity scores, analyses were adjusted for time-conditional propensity scores. To account for differences of baseline hazard of major adverse renal events, we ran age-stratified Cox models (< 70 yr v. ≥ 70 yr).
The ratio of the hazard ratios by age: 0.92 (95% CI 0.77–1.07), sex: 1.03 (95% CI 0.97–1.08), history of cardiovascular disease: 0.97 (95% CI 0.92–1.01) and history of renal insufficiency: 1.11 (95% CI 1.07–1.16).
Interpretation
In this population-based cohort study, levothyroxine treatment was not associated with a reduced risk of adverse renal outcomes compared with no treatment among people with subclinical hypothyroidism. The risk of adverse renal outcomes associated with levothyroxine treatment did not differ meaningfully by age, sex, prior history of cardiovascular disease, and prior diagnosis of renal insufficiency in subgroup analyses. Results were consistent across several sensitivity analyses.
Renal and thyroid physiologies are interlinked. Hypothyroidism can affect renal structure by decreasing the kidney-to–body weight ratio, decreasing tubular mass, and altering glomerular structure.23 Thyroid hormone has also been shown to increase arterial relaxation and decrease arterial resistance, which may improve renal blood flow.24 Consistent with these findings, studies have shown that hypothyroidism is associated with an increased risk of renal insufficiency.4,25–27 Subclinical hypothyroidism is a milder form of hypothyroidism in which thyrotropin levels are elevated with normal free thyroxine levels. There continues to be controversy as to whether subclinical hypothyroidism requires treatment and whether there is any clinical benefit when levothyroxine treatment is given, as most patients with this condition are asymptomatic.15,28,29 Subclinical hypothyroidism is associated with increased risk of adverse renal outcomes and developing renal insufficiency in observational studies.30,31 However, there is a paucity of studies assessing whether levothyroxine treatment of subclinical hypothyroidism is associated with a decreased risk of adverse renal outcomes. Prior randomized controlled trials conducted among older individuals (age ≥ 65 yr) with subclinical hypothyroidism who were randomly assigned to levothyroxine versus placebo did not assess adverse renal outcomes.32,33
In a pilot randomized, double-blind, placebo-controlled trial, 32 patients with proteinuric chronic kidney disease and thyrotropin levels between 2.6 mIU/L and 9.9 mIU/L were randomly assigned to levothyroxine versus placebo for 12 weeks.34 In this study, people with levothyroxine treatment had decreased proteinuria (−1.1 [−4.1 to 0.9] g/d v. +0.2 [−0.4 to 2.1] g/d; p = 0.0011) and improved eGFR by 4.0 mL/min/1.73 m2 v. −2.0 mL/min/1.73 m2 (p = 0.049) after 12 weeks compared with people in the placebo arm. Similarly, a retrospective study involving 309 individuals with stage 2–4 chronic kidney disease diagnosed with subclinical hypothyroidism showed that individuals treated with levothyroxine had a lesser decline in eGFR than individuals not treated (levothyroxine-treated eGFR change: −2.11 ± 1.12 v. not treated −5.93 ± 1.65 mL/min/yr/1.73 m2; p = 0.04).35 This study also found that levothyroxine treatment was associated with a decreased risk of developing a 50% reduction in eGFR and end-stage renal disease (HR 0.28, 95% CI 0.12–0.68) over a mean follow-up time of 34.8 (standard deviation 24.3) months.35 However, this study did not account for the difference in the duration of subclinical hypothyroidism between levothyroxine-treated and nontreated groups before cohort entry or for confounding by indication in which individuals given treatment may be healthier and less likely to develop adverse renal events at baseline. This result contrasts with our finding of no association between levothyroxine treatment and the risk of adverse renal outcomes. The differences in our findings may be due to differences in methodology; in our study, we used a prevalent new-user design and conducted a time-conditional propensity score–matched study that accounted for a large number of potential confounders and additionally matched on age, sex, disease duration, and calendar time. Furthermore, the mean follow-up time in our study was shorter, which may result in fewer adverse renal events observed.
Limitations
The CPRD documents prescriptions but does not confirm adherence to treatment. However, we assessed exposure to levothyroxine using continued prescription of levothyroxine with a grace period of 30 days. People who continued to obtain their levothyroxine prescription within 30 days were likely to be adherent to treatment.
We were unable to identify people with an underlying autoimmune cause of subclinical hypothyroidism, which warrants treatment with levothyroxine. Thyroid autoimmune disease may be associated with autoimmune renal disorders such as glomerular disease.36 Further studies are needed to determine whether autoimmune thyroid disease is associated with renal function and whether levothyroxine treatment affects this association.
Our follow-up time was relatively short (median 1.6 yr), which may underestimate the effect of levothyroxine on renal function among people with subclinical hypothyroidism. There was also a difference in the duration of follow-up among individuals treated with levothyroxine compared with nontreated individuals with subclinical hypothyroidism. In this real-world data, individuals treated with levothyroxine had a shorter follow-up time given that many of them stopped treatment (Appendix 1, Supplementary Table 4). Future studies with longer follow-up are needed.
Conclusion
In this population-based cohort study, levothyroxine treatment was not associated with the risk of adverse renal outcomes among people with subclinical hypothyroidism. Further studies in different populations and with longer follow-up are recommended to further assess the association between levothyroxine treatment of subclinical hypothyroidism and adverse renal outcomes.
Supplementary Information
Footnotes
Competing interests: Robert Platt has received consulting fees from Biogen, Boehringer Ingelheim, Merck, Nant Pharma, Pfizer, Vanda Pharmaceuticals, and Organon for work unrelated to this project. Dr. Platt is a study advisory committee member for Pfizer. Oriana Hoi Yun Yu received fees for a Novo Nordisk Advisory Board conducted in December 2021 for work unrelated to this project. No other competing interests were declared.
This article has been peer reviewed.
Contributors: Oriana Yu developed the study protocol (lead) and methodology (equal), wrote the first draft of the manuscript, and contributed to review and editing (equal). Christopher Filliter conducted statistical analyses (lead) and contributed to writing, review, and editing (equal). Kristian Filion contributed to the study design and to writing, review, and editing (equal). Robert Platt contributed to the study design and to writing, review, and editing (equal). Roland Grad contributed to the study design (reviewed the study protocol) and contributed to writing, review, and editing (equal). Mona Vahidi Rad was involved in defining exposure and outcomes for the study using the Clinical Practice Research Datalink, and contributed to the manuscript review. Christel Renoux contributed to the study design and methodology (equal), and to the writing, review, and editing. Oriana Yu and Christopher Filliter had full access to the data in the study. Oriana Yu takes responsibility for the integrity of the data and the accuracy of the data analysis. All authors provided final approval of the revised manuscript and agreed to be accountable for all aspects of the work.
Funding: This study was funded by a project grant from the Canadian Institutes of Health Research (project grant reference no. 169011). The sponsors did not influence the study design, data collection, analysis, interpretation of the results, writing of the manuscript, or decision to submit the manuscript for publication in any way.
Data sharing: This study used data from the Clinical Practice Research Datalink obtained under licence from the UK Medicines and Healthcare Products Regulatory Agency. The data are collected by the UK National Health Service from patients receiving clinical care in the United Kingdom. The interpretation and conclusions from the study findings were from the authors alone. Because of the UK Data Protection Act, the data cannot be shared in a public repository but can be requested via application to the Clinical Practice Research Datalink (https://www.cprd.com).
Disclosures: Robert Platt holds the inaugural Albert Boehringer I Chair in Pharmacoepidemiology at McGill University. Kristian Filion is supported by a William Dawson Scholar award from McGill University.
References
- 1.Surks MI, Ortiz E, Daniels GH, et al. Subclinical thyroid disease: scientific review and guidelines for diagnosis and management. JAMA 2004;291:228–38. [DOI] [PubMed] [Google Scholar]
- 2.Pearce SHS, Brabant G, Duntas LH, et al. 2013 ETA guideline: management of subclinical hypothyroidism. Eur Thyroid J 2013;2:215–28. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Chonchol M, Lippi G, Salvagno G, et al. Prevalence of subclinical hypothyroidism in patients with chronic kidney disease. Clin J Am Soc Nephrol 2008;3: 1296–300. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Lo JC, Chertow GM, Go AS, et al. Increased prevalence of subclinical and clinical hypothyroidism in persons with chronic kidney disease. Kidney Int 2005; 67:1047–52. [DOI] [PubMed] [Google Scholar]
- 5.Gopinath B, Harris DC, Wall JR, et al. Relationship between thyroid dysfunction and chronic kidney disease in community-dwelling older adults. Maturitas 2013;75:159–64. [DOI] [PubMed] [Google Scholar]
- 6.Bentley AG, Madsen KM, Davis RG, et al. Response of the medullary thick ascending limb to hypothyroidism in the rat. Am J Pathol 1985;120:215–21. [PMC free article] [PubMed] [Google Scholar]
- 7.Davis RG, Madsen KM, Fregly MJ, et al. Kidney structure in hypothyroidism. Am J Pathol 1983;113:41–9. [PMC free article] [PubMed] [Google Scholar]
- 8.Spahia N, Rroji M, Barbullushi M, et al. Subclinical hypothyroidism, kidney, and heart from normal to uremic milieu. Metab Syndr Relat Disord 2023;21:415–25. [DOI] [PubMed] [Google Scholar]
- 9.Chang Y-C, Chang CH, Yeh Y-C, et al. Subclinical and overt hypothyroidism is associated with reduced glomerular filtration rate and proteinuria: a large cross-sectional population study. Sci Rep 2018;8:2031. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Singh U, Rai V, Singh R, et al. Renal biopsy findings in patients with hypothyroidism: report of 16 cases. J Clin Diagn Res 2016;10:EC27–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Toda A, Hara S, Kato M, et al. Association of thyrotropin concentration with chronic kidney disease in a Japanese general population cohort. Nephron 2019;142:91–7. [DOI] [PubMed] [Google Scholar]
- 12.Rodondi N, den Elzen WPJ, Bauer DC, et al.; Thyroid Studies Collaboration. Subclinical hypothyroidism and the risk of coronary heart disease and mortality. JAMA 2010;304:1365–74. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 13.Wolf A, Dedman D, Campbell J, et al. Data resource profile: Clinical Practice Research Datalink (CPRD) Aurum. Int J Epidemiol 2019;48:1740–40g. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 14.Stagnaro-Green A. Approach to the patient with postpartum thyroiditis. J Clin Endocrinol Metab 2012;97:334–42. [DOI] [PubMed] [Google Scholar]
- 15.Jonklaas J, Bianco AC, Bauer AJ, et al.; American Thyroid Association Task Force on Thyroid Hormone Replacement. Guidelines for the treatment of hypothyroidism: prepared by the American Thyroid Association Task Force on Thyroid Hormone Replacement. Thyroid 2014;24:1670–751. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 16.Suissa S, Dell’Aniello S, Renoux C. The prevalent new-user design for studies with no active comparator: the example of statins and cancer. Epidemiology 2023;34:681–9. [DOI] [PubMed] [Google Scholar]
- 17.Suissa S, Moodie EEM, Dell’Aniello S. Prevalent new-user cohort designs for comparative drug effect studies by time-conditional propensity scores. Pharmacoepidemiol Drug Saf 2017;26:459–68. [DOI] [PubMed] [Google Scholar]
- 18.Rubin D. Basic ideas of multiple imputation for nonresponse. Surv Methodol 1986;12:37–47. [Google Scholar]
- 19.Rubin DB, Schenker N. Multiple imputation in health-care databases: an overview and some applications. Stat Med 1991;10:585–98. [DOI] [PubMed] [Google Scholar]
- 20.McLennan D, Barnes H, Noble M, et al. The English indices of deprivation 2010. London (UK): Department for Communities and Local Government, Eland House; 2011:1–145. [Google Scholar]
- 21.Lin DY, Wei LJ. The robust inference for the Cox proportional hazards model. J Am Stat Assoc 1989;84:1074–8. [Google Scholar]
- 22.Benchimol EI, Smeeth L, Guttmann A, et al.; RECORD Working Committee. The reporting of studies conducted using observational routinely-collected health data (RECORD) statement. PLoS Med 2015;12:e1001885. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Tunbridge WM, Evered DC, Hall R, et al. The spectrum of thyroid disease in a community: the Whickham Survey. Clin Endocrinol (Oxf) 1977;7:481–93. [DOI] [PubMed] [Google Scholar]
- 24.Liu KL, Lo M, Canaple L, et al. Vascular function of the mesenteric artery isolated from thyroid hormone receptor-α knockout mice. J Vasc Res 2014;51:350–9. [DOI] [PubMed] [Google Scholar]
- 25.Enia G, Panuccio V, Cutrupi S, et al. Subclinical hypothyroidism is linked to micro-inflammation and predicts death in continuous ambulatory peritoneal dialysis. Nephrol Dial Transplant 2007;22:538–44. [DOI] [PubMed] [Google Scholar]
- 26.Chen H-S, Wu T-EJ, Jap T-S, et al. Subclinical hypothyroidism is a risk factor for nephropathy and cardiovascular diseases in type 2 diabetic patients. Diabet Med 2007;24:1336–44. [DOI] [PubMed] [Google Scholar]
- 27.Huang C-W, Li BH, Reynolds K, et al. Association between hypothyroidism and chronic kidney disease observed among an adult population 55 years and older. Medicine (Baltimore) 2020;99:e19569. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Bekkering GE, Agoritsas T, Lytvyn L, et al. Thyroid hormones treatment for subclinical hypothyroidism: a clinical practice guideline. BMJ 2019;365:l2006. [DOI] [PubMed] [Google Scholar]
- 29.Biondi B, Cappola AR, Cooper DS. Subclinical hypothyroidism: a review. JAMA 2019;322:153–60. [DOI] [PubMed] [Google Scholar]
- 30.Kim HJ, Park SJ, Park HK, et al. Subclinical thyroid dysfunction and chronic kidney disease: a nationwide population-based study. BMC Nephrol 2023;24:64. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 31.Tsuda S, Nakayama M, Matsukuma Y, et al. Subclinical hypothyroidism is independently associated with poor renal outcomes in patients with chronic kidney disease. Endocrine 2021;73:141–50. [DOI] [PubMed] [Google Scholar]
- 32.Stott DJ, Rodondi N, Kearney PM, et al.; TRUST Study Group. Thyroid hormone therapy for older adults with subclinical hypothyroidism. N Engl J Med 2017;376:2534–44. [DOI] [PubMed] [Google Scholar]
- 33.Mooijaart SP, Du Puy RS, Stott DJ, et al. Association between levothyroxine treatment and thyroid-related symptoms among adults aged 80 years and older with subclinical hypothyroidism. JAMA 2019;322:1977–86. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 34.Blackaller GN, Chávez-Iñiguez JS, Carreón-Bautista EE, et al. A pilot trial on the effect of levothyroxine on proteinuria in patients with advanced CKD. Kidney Int Rep 2020;6:110–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Shin DH, Lee MJ, Kim SJ, et al. Preservation of renal function by thyroid hormone replacement therapy in chronic kidney disease patients with subclinical hypothyroidism. J Clin Endocrinol Metab 2012;97:2732–40. [DOI] [PubMed] [Google Scholar]
- 36.Santoro D, Vadalà C, Siligato R, et al. Autoimmune thyroiditis and glomerulopathies. Front Endocrinol (Lausanne) 2017;8:119. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.