Introduction and Aim
The association between thyroid dysfunction and mortality is controversial. Moreover, the impact of duration of thyroid dysfunction is unclarified. Our aim was to investigate the correlation between biochemically assessed thyroid function as well as dysfunction duration and mortality.
Register-based follow-up study of 239,768 individuals with a serum TSH measurement from hospitals and/or general practice in Funen, Denmark. Measurements were performed at a single laboratory from January 1st 1995 to January 1st 2011. Cox regression was used for mortality analyses and Charlson Comorbidity Index (CCI) was used as comorbidity score.
Hazard ratios (HR) with 95% confidence intervals (CI) for mortality with decreased (<0.3 mIU/L) or elevated (>4.0 mIU/L) levels of TSH were 2.22; 2.14–2.30; P<0.0001 and 1.28; 1.22–1.35; P<0.0001, respectively. Adjusting for age, gender, CCI and diagnostic setting attenuated the risk estimates (HR 1.23; 95% CI: 1.19–1.28; P<0.0001, mean follow-up time 7.7 years, and HR 1.07; 95% CI: 1.02–1.13; P = 0.004, mean follow-up time 7.2 years) for decreased and elevated values of TSH, respectively. Mortality risk increased by a factor 1.09; 95% CI: 1.08–1.10; P<0.0001 or by a factor 1.03; 95% CI: 1.02–1.04; P<0.0001 for each six months a patient suffered from decreased or elevated TSH, respectively. Subdividing according to degree of thyroid dysfunction, overt hyperthyroidism (HRovert 1.12; 95% CI: 1.06–1.19; P<0.0001), subclinical hyperthyroidism (HRsubclinical 1.09; 95% CI: 1.02–1.17; P = 0.02) and overt hypothyroidism (HRovert 1.57; 95% CI: 1.34–1.83; P<0.0001), but not subclinical hypothyroidism (HRsubclinical 1.03; 95% CI: 0.97–1.09; P = 0.4) were associated with increased mortality.
Conclusions and Relevance
In a large-scale, population-based cohort with long-term follow-up (median 7.4 years), overt and subclinical hyperthyroidism and overt but not subclinical hypothyroidism were associated with increased mortality. Excess mortality with increasing duration of decreased or elevated serum TSH suggests the importance of timely intervention in individuals with thyroid dysfunction.
Citation: Laulund AS, Nybo M, Brix TH, Abrahamsen B, Jørgensen HL, Hegedüs L (2014) Duration of Thyroid Dysfunction Correlates with All-Cause Mortality. The OPENTHYRO Register Cohort. PLoS ONE 9(10): e110437. https://doi.org/10.1371/journal.pone.0110437
Editor: Mohammad Ebrahim Khamseh, Institute of Endocrinology and Metabolism, Islamic Republic of Iran
Received: May 8, 2014; Accepted: September 20, 2014; Published: October 23, 2014
Copyright: © 2014 Laulund et al. This is an open-access article distributed under the terms of the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original author and source are credited.
Data Availability: The authors confirm that, for approved reasons, some access restrictions apply to the data underlying the findings. Data is stored by Statistics Denmark (DS). DS provides access to authorized researchers and institutions for this project through password-protected remote desktop access and by special approval since the data is theoretically personally identifiable, despite being anonymized. Interested researchers will be able to gain access to the data from Statistics Denmark upon request to Statistics Denmark. Information is available on the Statistics Denmark webpage http://www.dst.dk/da/TilSalg/Forskningsservice/Dataadgang.aspx and the email contact is firstname.lastname@example.org.
Funding: The authors confirm that the study was funded by Novo Nordisk Foundation. The funder had no role in study design, data collection and analysis, decision to publish, or preparation of the manuscript.
Competing interests: The authors have the following interests. Bo Abrahamsen has the following interests: Grants from or conducted trials for Novartis, Nycomed/Takeda and Amgen; advisory board member Nycomed/Takeda, Merck and Amgen; and speakers fees from Nycomed/Takeda, Amgen, Merck, Eli Lilly. Laszlo Hegedüs is a consultant for and is supported by an unrestricted grant from the Novo Nordisk Foundation. There are no patents, products in development or marketed products to declare. This does not alter the authors' adherence to all the PLOS ONE policies on sharing data and materials, as detailed online in the guide for authors.
The thyroid hormones thyroxine (T4) and triiodothyronine (T3) are essential in the regulation of a number of vital pathways such as lipolysis, gluconeogenesis and oxidative phosphorylation . Thyroid dysfunction, as reflected by decreased as well as elevated TSH, has been linked with an increased risk of developing a number of clinical conditions, such as cardiovascular diseases , , diabetes –, various malignant conditions , lung diseases , and psychiatric diseases ,  – both before and after the diagnosis of thyroid dysfunction. Whether this increased burden of morbidity also manifests as increased mortality in patients with thyroid dysfunction is controversial. During the past decade, several meta-analyses – have dealt with the possible influence of thyroid dysfunction on mortality, but due to considerable heterogeneity in terms of design (definition and severity of thyroid dysfunction, inclusion and exclusion criteria, treatment type and number of individuals included and inconsistent control for comorbidity) it is not possible to draw any firm conclusions. Recently, two large-scale register-based studies from Denmark, where comorbidity was controlled for in a standardized manner, have shown an increased mortality in overt hypo- and hyperthyroidism , . However, neither study provided information on biochemical thyroid data permitting an evaluation of the influence of subclinical thyroid dysfunction. Nor did these studies take the duration of thyroid dysfunction into consideration. These shortcomings constitute essential limitations when investigating the consequences of thyroid dysfunction.
Thus, the objective of the present long-term, adequately powered follow-up study was to investigate the consequences of biochemically assessed thyroid dysfunction, as evaluated by a single first measurement of serum TSH and duration of thyroid dysfunction with respect to mortality. The study covers the whole spectrum of thyroid dysfunction from subclinical to overt hypo- and hyperthyroidism, including account of comorbidity and whether the test was obtained in a hospital setting or at a general practitioner (GP).
Subjects and Methods
The Danish Civil Registration System (DCRS) provides all Danish residents with a unique social security number. The DCRS contains e.g. personal information on birthday, year, gender, place of birth, hospital records and citizenship(s) . The social security number allows for individual record linkage between all Danish health registers , . The biochemical data is stored in the OPENTHYRO database, which retrieved information on blood samples analyzed at the Department of Clinical Biochemistry and Pharmacology at Odense University Hospital, including all tests ordered by GPs on the island of Funen. All biochemical data were linked to social security numbers and cross-linked with the National Patient Registry (DNPR) and the National Database of Reimbursed Prescriptions (DNDRP), which contain information about comorbidity from all somatic hospital admissions (both in- and outpatient contacts)  and records of all medications dispensed from Danish pharmacies, respectively . Patient information was anonymized and de-identified prior to analysis.
The initial extraction from OPENTHYRO includes all subjects with at least one TSH measurement between January 1st 1995 and January 1st 2011 (n = 275467). In order to investigate only incident cases a washout period of one year (1995) was used (excluding 5644 subjects). Patients who before the initial TSH measurement had received radioactive iodine, had had thyroid surgery, had received treatment with anti-thyroid medication (carbimazole, propylthiouracil or methimazole) or thyroid hormone, were excluded (n = 8819); as were patients diagnosed with any pituitary disease prior to the initial TSH measurement (n = 753). Patients under the age of 18 years (n = 20260) and those lost to follow-up or emigration were also excluded, leaving 239768 subjects.
Thyroid function phenotypes
For dose-response analyses by severity of thyroid dysfunction, serum TSH was divided into subgroups. The index date and the thyroid function phenotypes were based on the first serum TSH measurement. Euthyroidism was defined as a TSH value between 0.3–4.0 mlU/L. Decreased TSH covers subjects with a TSH below 0.3 mIU/L. Overt hyperthyroidism was defined as TSH<0.3 mlU/L and T4>135.0 nmol/L or T3>2.2 nmol/L; subclinical hyperthyroidism as a TSH<0.3 mlU/L and T4<135.0 nmol/L and T3<2.2 nmol/L. Elevated TSH covers subjects with a TSH above 4.0 mIU/L. Overt hypothyroidism was defined as a TSH>4.0 mlU/L and a T4<60.0 nmol/L and, finally, subclinical hypothyroidism was defined as a TSH>4.0 mlU/L and a normal T4 (60.0–135.0 nmol/L).
All serum TSH determinations were done in the same laboratory, which served all hospitals and general practitioners in the region. Until 2006, TSH values were measured using a time-resolved fluoroimmunoassay based on a direct sandwich technique with three mouse anti-human monoclonal antibodies using the AutoDELFIA equipment (Wallac, Turku, Finland). The limit of detection (LOD) was 0.005 mIU/L, while within-run imprecision had a coefficient of variation (CV) of 2.7% at 0.89 mIU/L and 1.3% at 17.6 mIU/L . From 2006, measurements were performed with a solid-phase, two-site chemiluminescent immunometric assay on Immulite 2000 equipment (Siemens, Erlangen, Germany). The analytical LOD was 0.004 mIU/L. Within-run CV was 5.3% at 0.32 mIU/L and 3.9% at 3.3 mIU/L. To assure compatibility between the two analyses, method comparison was performed in 120 patient samples showing comparable means (range 0.008–49 mIU/L) and a regression coefficient of 0.991. In addition, an external quality control program assured comparability (Ringversuche, RfB, Bonn, Germany).
Concentrations of T4 and T3 were analyzed using time-resolved fluoroimmunoassays based on competitive binding to T3- or T4-specific antibodies, respectively. Analyses were performed using the AutoDELFIA equipment (Wallac, Turku, Finland). LOD was 0.3 nmol/L (T3) and 5.0 nmol/L (T4). CV was 3.2% at 1.37 nmol/L (T3) and 2.8% at 79.0 nmol/L (T4).
The Charlson Comorbidity Index (CCI) is based on 19 comorbid conditions (myocardial infarction, congestive heart failure, peripheral vascular disease, cerebrovascular disease, dementia, chronic pulmonary disease, connective tissue disease, ulcer disease, diabetes, hemiplegia, moderate or severe renal disease, diabetes with end organ damage, any tumor, leukemia, lymphoma, mild liver disease, moderate or severe liver disease, metastatic solid tumor and AIDS) . All the disease categories were identified from DNPR. The CCI was applied to hospital contacts in the time period from January 1st 1977 (establishment of DNPR) until end of the study (November 30th 2012).
The project is approved by the Danish Data Protection Agency and by Statistics Denmark to access health records. OPEN is an approved research institution permitted to access data hosted by Statistics Denmark (project 704047).
Data processing was done in SPSS version 19.0 (2010, IBM Corp. IBM SPSS Statistics for Windows, Armonk, NY, USA) and SAS version 9.3 (2011, SAS Institute Inc., Cary, NC, USA) through VPN access to Statistics Denmark.
The primary outcome was mortality. Differences between baseline values of the outcome groups were analyzed using parametric or categorical statistical methods depending on the nature of the data.
Multivariate Cox proportional hazards models were used to compute hazard ratios (HR) for all-cause mortality. HRs were adjusted for age, gender, CCI and treatment setting (general practice or hospital clinic). The validity of the proportional hazards assumption was evaluated by inspection of Schoenfeld residuals versus time. No significant associations between the residuals and time were present. Significant differences between the groups were defined as a P-value below 0.05 using two-tailed tests. In addition, we performed a sub-analysis, where TSH was used as a continuous variable in individuals with TSH within the normal range. A dynamic model was developed to encompass changes in thyroid dysfunction over time. A time-dependent cumulative covariate to capture the number of six month periods during which the mean TSH level had been decreased or elevated, was designed. This took into account all episodes of abnormal TSH in both study groups rather than the baseline classification alone. Any six month periods in which no further TSH measurement had been performed were considered periods of normal TSH for the purpose of this conservative sub-analysis.
A total of 239768 subjects (56.3% female and 43.7% male) with at least one serum TSH measurement were identified from the OPENTHYRO database (Table 1). Of the study population, 3.5% were hypothyroid, 92.7% were euthyroid, and 3.8% were hyperthyroid. After subdivision according to thyroid dysfunction phenotype, we found 2.0% of the cohort to be overtly hyperthyroid, 1.0% to be subclinically hyperthyroid, 0.2% to be overtly hypothyroid and 2.7% to be subclinically hypothyroid.
Demographic characteristics are summarized in table 1. The mean age was significantly higher in the group with decreased TSH (61.7 years) when compared with the euthyroid (50.8 years) and the hypothyroid groups (53.6 years). The hyperthyroid individuals were more likely to have a higher burden of comorbidity as reflected by a higher mean CCI (0.88; SD 1.47). Diagnostic setting influenced the burden of comorbidity, as individuals with samples from GP had a mean CCI of 0.37 (SD 0.82), while individuals diagnosed in a hospital setting had 0.81 (SD 1.41), P = 0.0001. During a median follow-up time of 7.4 years, 37964 subjects (18.5% of those with elevated TSH, 15.0% of the normal TSH group and 33.6% of those with decreased TSH) died. Within the group with decreased TSH, mortality was significantly higher for males than for females.
Looking at mortality for the group with high serum TSH, males and females separately, we found a HR 1.20; 95% Cl: 1.11–1.31; P<0.0001) in the male population, after adjusting for the previously mentioned factors. No significant association was found in the female population with high TSH (HR 1.01; 95% Cl: 0.95–1.08; P = 0.64). When investigating the male population with low TSH, we found a significant association with mortality (HR 1.21; 95% CI: 1.16–1.27; P<0.0001). The association was also found to be significant in the low TSH female group with a HR 1.25; 95% CI: 1.17–1.34; P<0.0001.
Mortality in individuals with decreased TSH versus euthyroid individuals
The hazard ratio (with 95% confidence interval) for all-cause mortality in subjects with decreased TSH was 2.31; 95% CI: 2.24–2.39; P<0.0001. After adjusting for confounders the HR was HR 1.23; 95% CI: 1.19–1.28; P<0.0001 (Figure 1, panel A).
Decreased TSH (<0.3 mlU/L) (panel A) and elevated TSH (>4 mlU/L) (panel B) as predictors of mortality. The figure illustrates the crude hazard ratios and the adjusted hazard ratios from the Cox regression analysis with stepwise addition of the covariates. Population N = 239768. ** P<0.0001.
Dose-response analysis. Cumulative exposure to decreased TSH
A Cox proportional hazards model in which a time-dependent covariate had been defined was used to capture the cumulative duration of decreased TSH. For each six month time intervals in which the mean serum TSH value was below 0.3 mIU/L, mortality risk increased by a factor 1.09; 95% CI: 1.08–1.10; P<0.0001. This corresponds to a 137% increase in mortality risk per five years with decreased TSH.
Degree of hyperthyroidism and mortality
Subdividing into overt and subclinical hyperthyroidism showed significantly increased mortality in both conditions (HRovert 1.12; 95% CI: 1.06–1.19; P<0.0001, and HRsubclinical 1.09; 95% CI: 1.02–1.17; P = 0.012) compared to euthyroid individuals (Table 2).
Impact of diagnostic setting (GP or hospital)
Individuals with decreased TSH diagnosed in GP as well as individuals diagnosed in a hospital setting had an increased HR for all-cause mortality (Table 3). The association persisted for individuals with decreased TSH and without comorbidities, whether diagnosed in GP or in a hospital (HRGP, CCI = 0 1.31; 95% CI: 1.17–1.48; P<0.0001, and HRhospital, CCI = 0 1.20; 95% CI: 1.12–1.29; P<0.0001).
After subdividing the samples collected from GP according to biochemical thyroid phenotypes, the association with excess mortality persisted (HRovert 1.16; 95% CI: 1.04–1.31; P<0.0001, and HRsubclinical 1.28; 95% CI: 1.12–1.46; P = 0.0003). Essentially, similar results were found among individuals without comorbidity (HRovert, CCI = 0 1.23; 95% CI: 1.05–1.43; P = 0.01, and HRsubclinical CCI = 0, 1.31; 95% CI: 1.08–1.60; P = 0.006). In the hospital setting, only overt hyperthyroidism was associated with a statistically significant excess mortality in both the overall group and in the group with no registered comorbidities (HRovert 1.09; 95% CI: 1.02–1.17; P = 0.01, and HRovert, CCI = 0 1.16; 95% CI: 1.05–1.29; P = 0.005). In contrast, subclinical hyperthyroidism diagnosed in a hospital setting was not found associated with excess mortality compared to euthyroid individuals (HRsubclinical 1.03; 95% CI: 0.95–1.12; P = 0.5, and HRsubclinical, CCI = 0 1.11; 95% CI: 0.97–1.28; P = 0.1).
Mortality in individuals with elevated TSH versus euthyroid individuals
As evident from figure 1, panel B, subjects with elevated TSH had an increased mortality. The HR for all-cause mortality in subjects with elevated TSH was 1.08; 95% CI: 1.04–1.12; P<0.0001, in the unadjusted setting. After adjusting for confounders, the HR was 1.07; 95% CI: 1.03–1.11; P = 0.001.
Dose-response analysis. Cumulative exposure to elevated TSH
For each six month time interval in which the mean TSH value was above 4.0 mIU/L, mortality risk increased by a factor 1.03; 95% CI: 1.02–1.04; P<0.0001. This corresponds to a 34% increase in mortality risk per five years with elevated TSH.
Degree of hypothyroidism and mortality
Subdividing hypothyroidism into overt and subclinical hypothyroidism showed excess mortality in the overtly (HRovert 1.57; 95% CI: 1.34–1.83; P<0.0001), but not in the subclinically hypothyroid population (HRsubclinical 1.03; 95% CI: 0.97–1.09; NS) (Table 2).
Impact of diagnostic setting (GP or hospital)
Subjects diagnosed with elevated TSH in GP did not have a statistically significant excess mortality (HRGP 1.01; 95% CI: 0.90–1.14; NS). In contrast, subjects diagnosed in a hospital setting had an increased risk of all-cause mortality (HRhospital 1.09; 95% CI: 1.03–1.15; P = 0.003), which disappeared completely when looking at the population without comorbidities (HRhospital, CCI = 0 1.00; 95% CI: 0.90–1.10; NS).
In the hospital setting we observed a significant excess mortality for the group with overt hypothyroidism (HRovert 1.60; 95% CI: 1.35–1.90; P<0.0001). However, this association disappeared when looking at the population with no registered comorbidities (HR 1.20; 95% CI: 0.88–1.63; NS). In GP neither subclinical (HRsubclinical, CCI = 0 0.90; 95% CI: 0.74–1.08; NS) nor overt hypothyroidism (HRovert, CCI = 0 1.32; 95% CI: 0.80–2.20; NS) was associated with excess mortality (Table 3).
In the following, we will discuss the association between abnormal serum TSH, the duration of this, and all-cause mortality. Furthermore, we will argue that there are separate effects of the biochemically defined four thyroid phenotypes on mortality. Finally, that whether the serum TSH abnormality was diagnosed in general practice or in a hospital setting matters.
Hyperthyroidism, duration and mortality
In a register-based health care study with biochemical data from 239768 subjects without previously known thyroid disease, we demonstrate that mortality is significantly increased in subjects with a single first TSH measurement below the reference range.
Intuitively, comorbidity has an impact on all-cause mortality, but may also facilitate the development of decreased TSH, attributable e.g. to treatment with iodine-containing drugs and/or non-thyroidal illness. Therefore, it is essential to assess the burden and timing of comorbidity before estimating the independent consequence of a decreased TSH on mortality. Previous studies are inconsistent in respect to comorbidity adjustment. Some have adjusted for cardiovascular diseases , , some for cancer  and one study accounted for comorbidity from a population-based healthcare register . Similar to the latter study by Brandt and coworkers, we identified comorbidity from a health-related register and thus adjusted for a wide range of comorbidities. When using this approach, the adjusted mortality risk was increased by 23%, which is of the same magnitude as demonstrated in previous studies , , , , , –. In accordance with our previous study , subjects with hyperthyroidism and without comorbidity and irrespectively of diagnostic setting had an increased all-cause mortality, suggesting that decreased TSH per se is associated with excess mortality, independent of preexisting comorbidity.
Besides replicating the previous findings, we performed a cumulated dose-dependent mortality analysis. For each six months during which the mean TSH was decreased, mortality risk increased by 9%. This corresponds to a 137% increase in mortality risk per five years with decreased TSH, even after adjusting for confounders. Exact details concerning the timing of onset of disease are unattainable to us, but by using the index TSH measurement, and by only including incident cases, we sought to limit this issue. Our findings emphasize the need for timely and efficient intervention in order to minimize the duration of exposure to abnormal TSH.
Degree of hyperthyroidism and mortality
Previous studies found ambiguous results of an association between subclinical hyperthyroidism and mortality , –. Here, we demonstrate a 9% excess mortality associated with subclinical hyperthyroidism. However, this estimate is influenced by the diagnostic setting as patients in GP had a 28% excess mortality compared to euthyroid individuals. On the other hand, despite a higher CCI in a hospital setting, subjects with subclinical hyperthyroidism did not have significant excess mortality (HR 1.03; 95% CI 0.95–1.12; NS). Whether this is due to a more aggressive treatment in a hospital setting compared to a GP setting is unaccounted for. The excess mortality associated with hyperthyroidism has been attributed to systemic effects of thyroid hormones such as cardiac function changes , ,  or arrhythmias . This issue has not been addressed in the present work, since data on such systemic changes were not accessible.
Intuitively, more pronounced hyperthyroidism as seen in the overt phenotype should be linked with a poorer prognosis and more severe clinical manifestations, at least when left untreated –. In the present study we found an adjusted 12% and 9% mortality increase in the overtly and subclinically hyperthyroid group, respectively. Based on these findings, our results are in line with a dose-response relationship between thyroid dysfunction and mortality.
Hypothyroidism, duration and mortality
Considerable heterogeneity with respect for comorbidity, is evident between the studies addressing whether there is a link between elevated TSH and mortality , , , , –. In our study, after adjusting for comorbidity, we found a 7% excess mortality in those with a first TSH measurement above the reference range. Patients diagnosed in a hospital setting, as compared to GP, had a slightly poorer prognosis (HR 1.09; 95% CI 1.03–1.15; P = 0.0028). We speculate that this may be due to additional unmeasured comorbidities or frailty , . Accordingly, excess mortality was not present in the population without comorbidity (HR 1.00; 95% CI: 0.90–1.10; NS), which is in line with that found in another large scale survey .
Our cumulated dose-dependent mortality analysis resulted in a 34% increase in mortality risk per 5 years with elevated TSH, even after adjusting for confounders. Similarly to the results from the hyperthyroid dose-dependent mortality analysis, this result favors early and efficient correction of thyroid dysfunction to reduce all-cause mortality risk.
Degree of hypothyroidism and mortality
We did not find evidence of increased mortality in subjects with subclinical hypothyroidism, which is in accordance with the majority of the meta-analyses within this field , , , , . Overt hypothyroidism is a more severe form of hypothyroidism than subclinical hypothyroidism . Accordingly, this phenotype had a considerably poorer prognosis (57% increased mortality) as compared to euthyroidism, even after adjustment for age, gender, CCI and whether the individual was seen in GP or at hospital. Overall, our data from the hypothyroid individuals suggests that the excess mortality is driven by findings in the clinically more exposed hospital population.
Hitherto, the relationship between the severity of hypothyroidism and all-cause mortality has not been systematically assessed , . This is due to considerable heterogeneity between the studies addressing overt hypothyroidism and mortality , , , –, –. The heterogeneity is based on lack of all-cause mortality data , –, differences in inclusion criteria and differences in confounder adjustment. Most studies adjusted for age and gender, but some also adjusted for atherogenic risk factors ,  or comorbidities , . Furthermore, one study did not differentiate between overt and subclinical disease . Three studies , ,  found excess mortality in patients with overt hypothyroidism compared to euthyroid individuals. Seven studies found no excess mortality in patients with overt hypothyroidism , , , –. In the present study no association was found with mortality in the GP setting (HR 1.32; 95% CI: 0.90–1.94; NS), and in the hospital setting only the overall group had an excess mortality (HR 1.60; 95% CI: 1.35–1.90; P = 0.0001). This correlation with mortality resolved in the group without comorbidity.
Strengths and limitations
The major strengths of our study include high statistical power, standardized and validated procedures for evaluating the degree of comorbidity, the population-based setting and the long period of follow-up. Other strengths include the analysis of TSH measurements in a single laboratory. Although all TSH measurements were collected in a well-defined geographical area (Funen, Denmark) over a long period of time, this is not to be confused with a random population sample, since the initial investigation of thyroid function was most probably done on a clinical indication. This could obscure possible year-long duration of abnormal TSH at the time of detection.
While we adjusted for most of the established risk factors, covariates such as alcohol intake , cigarette smoking , body mass index ,  and level of education , also important risk factors for mortality and thyroid dysfunction, were unavailable to us. Furthermore, the categorization of subjects was based on a single baseline measurement of TSH. This could lead to misclassification in a small number of subjects because single measurements might not reflect a steady biochemical status but may be influenced also by the circadian variability of the hormone and presence of non-thyroidal illness . If this bias (type 1 error) is present it would increase the false positives. However, similar associations were present in the sensitivity analyses where multiple TSH assessments over time were used to classify abnormal TSH states.
In addition, our study did not take into account that some individuals may change from hyper- to hypothyroidism, or vice versa after the initial TSH measurement. The direction and magnitude of this potential bias is unknown. Lastly, it was not possible to distinguish the two major clinical etiologies of thyrotoxicosis. The majority are caused by Graves' disease or toxic nodular goiter. Since the two diseases present with considerable epidemiological and clinical differences , , it is forthright to assume differences in mortality pattern. Nyrienda and coworkers  found no association between mortality and Graves' disease, nor did they find a correlation between mortality and toxic nodular goiter. In contrast, Metso and coworkers  found an increased relative risk for mortality in toxic nodular goiter patients, but not in patients with Graves' disease, whereas we have recently shown that both Graves' disease and toxic nodular goiter are associated with excess all-cause mortality, at least in a hospital setting .
Previous studies on the impact of thyroid dysfunction on mortality are characterized by a number of shortcomings, the most important being lack of statistical power, which complicates and potentially invalidates interpretation of the data . Since it has been estimated that at least 900 participants are needed in a Cox power calculation to demonstrate an HR increase in mortality of 20% in individuals with thyroid dysfunction , it is evident that most studies are underpowered. Even with this very large amount of data, our power for subgroup analyses was limited because of the small number of deaths in these subgroups.
In addition, the majority of the previous studies did not have access to biochemical data and hence had to depend on hospital diagnoses and prescriptions of thyroid hormone and/or antithyroid drugs. Importantly, the present study included diagnosed as well as un-diagnosed individuals and combined raw/unprocessed data and national health register data, which allowed for a systematic and comprehensive collection of comorbidity and mortality data. Furthermore, availability of the setting in which the TSH sample was taken allowed us to investigate the prognostic implications of an abnormal TSH level in relation to whether it was obtained by a GP or in a hospital setting. Thus, our results can be generalized far beyond that limited to hospitalized elderly individuals with a high degree of morbidity.
In this study a single abnormal serum TSH measurement and an increasing duration of decreased or elevated serum TSH were strong risk factors for mortality. When combining the TSH measurement with T3 and T4 measurements, overt hyper-, subclinical hyper- and overt hypothyroidism were all found to be strong indicators of excess mortality. Taken together, these findings suggest the importance of timely intervention in individuals with thyroid dysfunction.
Conceived and designed the experiments: ASL MN BA HLJ THB LH. Performed the experiments: ASL MN BA HLJ THB LH. Analyzed the data: ASL MN BA HLJ THB LH. Contributed reagents/materials/analysis tools: ASL MN BA HLJ THB LH. Wrote the paper: ASL MN BA HLJ THB LH.
- 1. Brown-Borg HM (2007) Hormonal regulation of longevity in mammals. Ageing Res Rev 6: 28–45.
- 2. Osman F, Franklyn JA, Holder RL, Sheppard MC, Gammage MD (2007) Cardiovascular manifestations of hyperthyroidism before and after antithyroid therapy: a matched case-control study. J Am Coll Cardiol 49: 71–81.
- 3. Flynn RW, Macdonald TM, Jung RT, Morris AD, Leese GP (2006) Mortality and vascular outcomes in patients treated for thyroid dysfunction. J Clin Endocrinol Metab 91: 2159–2164.
- 4. Brandt F, Thvilum M, Almind D, Christensen K, Green A, et al. (2013) Morbidity before and after the diagnosis of hyperthyroidism: a nationwide register-based study. PLoS One 8: e66711.
- 5. Nyirenda MJ, Clark DN, Finlayson AR, Read J, Elders A, et al. (2005) Thyroid disease and increased cardiovascular risk. Thyroid 15: 718–724.
- 6. Goldman MB, Monson RR, Maloof F (1990) Cancer mortality in women with thyroid disease. Cancer Res 50: 2283–2289.
- 7. Milla CE, Zirbes J (2012) Pulmonary complications of endocrine and metabolic disorders. Paediatr Respir Rev 13: 23–28.
- 8. Thvilum M, Brandt F, Almind D, Christensen K, Brix TH, et al.. (2014) Increased Psychiatric Morbidity Before and After the Diagnosis of Hypothyroidism: A Nationwide Register Study. Thyroid. In press.
- 9. Brandt F, Thvilum M, Almind D, Christensen K, Green A, et al. (2014) Hyperthyroidism and psychiatric morbidity: evidence from a Danish nationwide register study. Eur J Endocrinol 170: 341–348.
- 10. Ochs N, Auer R, Bauer DC, Nanchen D, Gussekloo J, et al. (2008) Meta-analysis: subclinical thyroid dysfunction and the risk for coronary heart disease and mortality. Ann Intern Med 148: 832–845.
- 11. Thvilum M, Brandt F, Brix TH, Hegedus L (2012) A review of the evidence for and against increased mortality in hypothyroidism. Nat Rev Endocrinol 8: 417–424.
- 12. Brandt F, Green A, Hegedus L, Brix TH (2011) A critical review and meta-analysis of the association between overt hyperthyroidism and mortality. Eur J Endocrinol 165: 491–497.
- 13. Bauer DC, Rodondi N, Stone KL, Hillier TA (2007) Study of Osteoporotic Fractures Research Group: Universities of California PM, et al (2007) Thyroid hormone use, hyperthyroidism and mortality in older women. Am J Med 120: 343–349.
- 14. Singh S, Duggal J, Molnar J, Maldonado F, Barsano CP, et al. (2008) Impact of subclinical thyroid disorders on coronary heart disease, cardiovascular and all-cause mortality: a meta-analysis. Int J Cardiol 125: 41–48.
- 15. Yang LB, Jiang DQ, Qi WB, Zhang T, Feng YL, et al. (2012) Subclinical hyperthyroidism and the risk of cardiovascular events and all-cause mortality: an updated meta-analysis of cohort studies. Eur J Endocrinol 167: 75–84.
- 16. Thvilum M, Brandt F, Almind D, Christensen K, Hegedus L, et al. (2013) Excess mortality in patients diagnosed with hypothyroidism: a nationwide cohort study of singletons and twins. J Clin Endocrinol Metab 98: 1069–1075.
- 17. Brandt F, Almind D, Christensen K, Green A, Brix TH, et al. (2012) Excess mortality in hyperthyroidism: the influence of preexisting comorbidity and genetic confounding: a danish nationwide register-based cohort study of twins and singletons. J Clin Endocrinol Metab 97: 4123–4129.
- 18. Pedersen CB (2011) The Danish Civil Registration System. Scand J Public Health 39: 22–25.
- 19. Sortso C, Thygesen LC, Bronnum-Hansen H (2011) Database on Danish population-based registers for public health and welfare research. Scand J Public Health 39: 17–19.
- 20. Thygesen LC, Ersboll AK (2011) Danish population-based registers for public health and health-related welfare research: introduction to the supplement. Scand J Public Health 39: 8–10.
- 21. Thygesen LC, Daasnes C, Thaulow I, Bronnum-Hansen H (2011) Introduction to Danish (nationwide) registers on health and social issues: structure, access, legislation, and archiving. Scand J Public Health 39: 12–16.
- 22. Johannesdottir SA, Horvath-Puho E, Ehrenstein V, Schmidt M, Pedersen L, et al. (2012) Existing data sources for clinical epidemiology: The Danish National Database of Reimbursed Prescriptions. Clin Epidemiol 4: 303–313.
- 23. Jensen E, Hyltoft Petersen P, Blaabjerg O, Hansen PS, Brix TH, et al. (2004) Establishment of a serum thyroid stimulating hormone (TSH) reference interval in healthy adults. The importance of environmental factors, including thyroid antibodies. Clin Chem Lab Med 42: 824–832.
- 24. Charlson ME, Pompei P, Ales KL, MacKenzie CR (1987) A new method of classifying prognostic comorbidity in longitudinal studies: development and validation. J Chronic Dis 40: 373–383.
- 25. Metso S, Jaatinen P, Huhtala H, Auvinen A, Oksala H, et al. (2007) Increased cardiovascular and cancer mortality after radioiodine treatment for hyperthyroidism. J Clin Endocrinol Metab 92: 2190–2196.
- 26. Hall P, Lundell G, Holm LE (1993) Mortality in patients treated for hyperthyroidism with iodine-131. Acta Endocrinol (Copenh) 128: 230–234.
- 27. Franklyn JA, Maisonneuve P, Sheppard MC, Betteridge J, Boyle P (1998) Mortality after the treatment of hyperthyroidism with radioactive iodine. N Engl J Med 338: 712–718.
- 28. Volzke H, Schwahn C, Wallaschofski H, Dorr M (2007) Review: The association of thyroid dysfunction with all-cause and circulatory mortality: is there a causal relationship? J Clin Endocrinol Metab 92: 2421–2429.
- 29. Haentjens P, Van Meerhaeghe A, Poppe K, Velkeniers B (2008) Subclinical thyroid dysfunction and mortality: an estimate of relative and absolute excess all-cause mortality based on time-to-event data from cohort studies. Eur J Endocrinol 159: 329–341.
- 30. Collet TH, Gussekloo J, Bauer DC, den Elzen WP, Cappola AR, et al. (2012) Subclinical hyperthyroidism and the risk of coronary heart disease and mortality. Arch Intern Med 172: 799–809.
- 31. Surks MI, Ortiz E, Daniels GH, Sawin CT, Col NF, et al. (2004) Subclinical thyroid disease: scientific review and guidelines for diagnosis and management. JAMA 291: 228–238.
- 32. Biondi B, Cooper DS (2008) The clinical significance of subclinical thyroid dysfunction. Endocr Rev 29: 76–131.
- 33. Biondi B, Palmieri EA, Lombardi G, Fazio S (2002) Effects of subclinical thyroid dysfunction on the heart. Ann Intern Med 137: 904–914.
- 34. Biondi B (2012) Mechanisms in endocrinology: Heart failure and thyroid dysfunction. Eur J Endocrinol 167: 609–618.
- 35. Coceani M, Iervasi G, Pingitore A, Carpeggiani C, L'Abbate A (2009) Thyroid hormone and coronary artery disease: from clinical correlations to prognostic implications. Clin Cardiol 32: 380–385.
- 36. Mingote E, Merono T, Rujelman R, Marquez A, Fossati P, et al. (2012) High TSH and low T4 as prognostic markers in older patients. Hormones (Athens) 11: 350–355.
- 37. Iglesias P, Ridruejo E, Munoz A, Prado F, Macias MC, et al. (2013) Thyroid function tests and mortality in aged hospitalized patients: a 7-year prospective observational study. J Clin Endocrinol Metab 98: 4683–4690.
- 38. Maldonado LS, Murata GH, Hershman JM, Braunstein GD (1992) Do thyroid function tests independently predict survival in the critically ill? Thyroid 2: 119–123.
- 39. McQuade C, Skugor M, Brennan DM, Hoar B, Stevenson C, et al. (2011) Hypothyroidism and moderate subclinical hypothyroidism are associated with increased all-cause mortality independent of coronary heart disease risk factors: a PreCIS database study. Thyroid 21: 837–843.
- 40. Petersen K, Bengtsson C, Lapidus L, Lindstedt G, Nystrom E (1990) Morbidity, mortality, and quality of life for patients treated with levothyroxine. Arch Intern Med 150: 2077–2081.
- 41. Radacsi A, Kovacs G, Bernard W, Feldkamp J, Horster FA, et al. (2003) Mortality rate of chronically ill geriatric patients with subnormal serum thyrotropin concentration: a 2-yr follow-up study. Endocrine 21: 133–136.
- 42. Cappola AR, Fried LP, Arnold AM, Danese MD, Kuller LH, et al. (2006) Thyroid status, cardiovascular risk, and mortality in older adults. JAMA 295: 1033–1041.
- 43. Alevizaki M, Synetou M, Xynos K, Alevizaki CC, Vemmos KN (2006) Hypothyroidism as a protective factor in acute stroke patients. Clin Endocrinol (Oxf) 65: 369–372.
- 44. Gussekloo J, van Exel E, de Craen AJ, Meinders AE, Frolich M, et al. (2004) Thyroid status, disability and cognitive function, and survival in old age. JAMA 292: 2591–2599.
- 45. McGue M, Osler M, Christensen K (2010) Causal Inference and Observational Research: The Utility of Twins. Perspect Psychol Sci 5: 546–556.
- 46. Rodondi N, Newman AB, Vittinghoff E, de Rekeneire N, Satterfield S, et al. (2005) Subclinical hypothyroidism and the risk of heart failure, other cardiovascular events, and death. Arch Intern Med 165: 2460–2466.
- 47. Klein I, Danzi S (2007) Thyroid disease and the heart. Circulation 116: 1725–1735.
- 48. Parle JV, Maisonneuve P, Sheppard MC, Boyle P, Franklyn JA (2001) Prediction of all-cause and cardiovascular mortality in elderly people from one low serum thyrotropin result: a 10-year cohort study. Lancet 358: 861–865.
- 49. Vanderpump MP, Tunbridge WM, French JM, Appleton D, Bates D, et al. (1996) The development of ischemic heart disease in relation to autoimmune thyroid disease in a 20-year follow-up study of an English community. Thyroid 6: 155–160.
- 50. Slag MF, Morley JE, Elson MK, Crowson TW, Nuttall FQ, et al. (1981) Hypothyroxinemia in critically ill patients as a predictor of high mortality. JAMA 245: 43–45.
- 51. Iglesias P, Munoz A, Prado F, Guerrero MT, Macias MC, et al. (2009) Alterations in thyroid function tests in aged hospitalized patients: prevalence, aetiology and clinical outcome. Clin Endocrinol (Oxf) 70: 961–967.
- 52. Franklyn JA (1999) Thyroid disease and its treatment: short- and long-term consequences. J R Coll Physicians Lond 33: 564–567.
- 53. Osman F, Gammage MD, Franklyn JA (2001) Thyroid disease and its treatment: short-term and long-term cardiovascular consequences. Curr Opin Pharmacol 1: 626–631.
- 54. van den Beld AW, Visser TJ, Feelders RA, Grobbee DE, Lamberts SW (2005) Thyroid hormone concentrations, disease, physical function, and mortality in elderly men. J Clin Endocrinol Metab 90: 6403–6409.
- 55. Hegedus L, Rasmussen N, Ravn V, Kastrup J, Krogsgaard K, et al. (1988) Independent effects of liver disease and chronic alcoholism on thyroid function and size: the possibility of a toxic effect of alcohol on the thyroid gland. Metabolism 37: 229–233.
- 56. Brix TH, Hansen PS, Kyvik KO, Hegedus L (2000) Cigarette smoking and risk of clinically overt thyroid disease: a population-based twin case-control study. Arch Intern Med 160: 661–666.
- 57. Solanki A, Bansal S, Jindal S, Saxena V, Shukla US (2013) Relationship of serum thyroid stimulating hormone with body mass index in healthy adults. Indian J Endocrinol Metab 17: S167–169.
- 58. Stoving RK, Bennedbaek FN, Hegedus L, Hagen C (2001) Evidence of diffuse atrophy of the thyroid gland in patients with anorexia nervosa. Int J Eat Disord 29: 230–235.
- 59. Ceresini G, Lauretani F, Maggio M, Ceda GP, Morganti S, et al. (2009) Thyroid function abnormalities and cognitive impairment in elderly people: results of the Invecchiare in Chianti study. J Am Geriatr Soc 57: 89–93.
- 60. Laurberg P, Pedersen KM, Vestergaard H, Sigurdsson G (1991) High incidence of multinodular toxic goitre in the elderly population in a low iodine intake area vs. high incidence of Graves' disease in the young in a high iodine intake area: comparative surveys of thyrotoxicosis epidemiology in East-Jutland Denmark and Iceland. J Intern Med 229: 415–420.
- 61. Laurberg P, Cerqueira C, Ovesen L, Rasmussen LB, Perrild H, et al. (2010) Iodine intake as a determinant of thyroid disorders in populations. Best Pract Res Clin Endocrinol Metab 24: 13–27.
- 62. Brandt F, Thvilum M, Almind D, Christensen K, Green A, et al. (2013) Graves' disease and toxic nodular goiter are both associated with increased mortality but differ with respect to the cause of death: a Danish population-based register study. Thyroid 23: 408–413.