Jiunn-Horng Kang, Department of Physical Medicine and Rehabilitation, Taipei Medical University Hospital, Taipei, Taiwan. Tel.: + 886 2 2737 2181; fax: + 886 2 2378 9788. E-mail: email@example.com
Summary Background: Although studies have indicated that hyperthyroidism is associated with hypercoagulability, most such studies have focused only on examining the incidence of venous thrombosis. As far as we know, no study has attempted to explore the risk of pulmonary embolism (PE) among patients with hyperthyroidism. Objective: Using a nationwide population-based dataset, this study was aimed at estimating the risk of PE among hyperthyroidism patients during a 5-year period, as compared with non-hyperthyroidism patients during the same period. Methods: Data sourced from the Taiwan Longitudinal Health Insurance Database were analyzed. The study included 8903 patients with hyperthyroidism as a study cohort and 44 515 randomly selected patients without hyperthyroidism as a comparison cohort. Stratified Cox proportional hazard regressions were used to compute the 5-year PE-free survival rate between these two cohorts. Results: Of the total of 53 418 patients, 41 patients (0.08%) were identified as having PE during the follow-up period, 14 from the study cohort (0.16% of the hyperthyroidism patients) and 27 comparison patients (0.06% of patients from the comparison cohort). After adjustment for geographic region, monthly income, hypertension, diabetes, hyperlipidemia, peripheral vascular disease, coronary heart disease, cancer, recent surgery, recent fracture, pregnancy and the use of anticoagulants, the risk of having PE during the 5-year follow-up period was 2.31 times greater (95% confidence interval 1.20–4.45, P =0.012) for patients with hyperthyroidism than for patients in the comparison cohort. Conclusion: We found an increased risk of PE in patients with hyperthyroidism. Clinicians should be aware of this increased risk.
Hyperthyroidism is a common disorder, with an incidence of 0.5–2.5% worldwide, and women are much more likely to develop this disorder than men [1–3]. Hyperthyroidism has widespread effects on the heart and cardiovascular system; it can increase the heart rate and blood pressure , resulting in increased risk of atrial fibrillation and ventricular dysrhythmias . In addition, a previous study reported that thyroid dysfunction may modify the physiologic processes of primary and secondary hemostasis, and lead to bleeding or thrombosis [6–8].
Previous work has shown a hypercoagulable state in patients with hyperthyroidism [7–10]. Furthermore, data have indicated that hyperthyroidism influences endothelial function by upregulating the levels of adhesion molecules (vascular cell adhesion molecule-1) and endothelial marker proteins [tissue-type plasminogen activator (t-PA) and plasminogen activator inhibitor-1 (PAI-1)] . Although these studies have all indicated that hyperthyroidism is associated with hypercoagulability and endothelial dysfunction, most of them have only focused on examining the incidence of venous thrombosis. Very few have attempted to explore the risk of pulmonary embolism (PE) among patients with hyperthyroidism. Furthermore, most studies on the relationship between hyperthyroidism and PE have only been case reports or have lacked adjustments for potential confounders, so their findings may not be generalized [11–13]. To fill this gap, using a nationwide population-based dataset, this study aimed to estimate the risk of PE among hyperthyroidism patients aged 18 years or more during a 5-year period, as compared with non-hyperthyroidism patients during the same period.
This study analyzed data sourced from the Longitudinal Health Insurance Database (LHID) provided by the Taiwan National Health Research Institute. Taiwan began its National Health Insurance (NHI) program in 1995; this provides very low out-of-pocket payment medical care services for all citizens in Taiwan. As of 2007, 22.6 million (98.4%) of Taiwan’s 22.96 million people were enrolled in the NHI program. The LHID includes all original claims data for 1 000 000 beneficiaries, randomly sampled from the year 2000 Registry for Beneficiaries of the NHI program. The Taiwan National Health Research Institute reported that there were no statistically significant differences in age, gender or healthcare costs between the sample group in the LHID and all beneficiaries under the NHI program. Hundreds of researchers have used the LHID for published studies.
As the LHID consists of de-identified secondary data released to the public for research purposes, this study was exempt from full review by the Institutional Review Board.
The study design was a retrospective case–cohort study. All patients who visited ambulatory care centers with a principal diagnosis of hyperthyroidism (ICD-9-CM code 242, thyrotoxicosis with or without goiter) between 1 January 2001 and 31 December 2003 were selected as the study cohort (n =12 246). In order to increase the validity of hyperthyroidism diagnosis, we only included those patients who had at least two consensus diagnosed episodes of hyperthyroidism during the study period and who still had a hyperthyroidism diagnosis after receiving the blood test for the level of thyroid-stimulating hormone. We then excluded patients who were under 18 years old, to restrict our study sample to the adult population (n =618). We also excluded patients who had a hyperthyroidism diagnosis prior to 2001 (n =2720), in order to increase the likelihood of including only new cases. However, we cannot rule out patients who had been diagnosed with hyperthyroidism prior to 1995, as the NHI program in Taiwan was initiated that year, and the LHID only allows us to trace the use of medical services from 1996 to 2008. In this study, the first visit for the treatment of hyperthyroidism was assigned as the index ambulatory care visit. We also excluded patients who had diagnoses of PE prior to their index ambulatory care visits (n =5). The resulting study cohort included 8903 patients with hyperthyroidism.
We selected the comparison cohort from the remaining patients in the LHID. We excluded patients who were under 18 years old. We also excluded patients who had had a hyperthyroidism diagnosis between 1996 and 2008. We then randomly selected 44 515 subjects from the registry of beneficiaries (five for every patient with hyperthyroidism) matched with the study cohort patients in terms of age (18–34, 35–44, 45–54, 55–64 and > 64 years), gender and the year of index ambulatory care visit. We assigned their first ambulatory visits occurring in the index year as the index ambulatory care visits. We also confirmed that all selected patients in the comparison cohort had no diagnoses of PE prior to their index ambulatory care visits.
Variables of interest
The outcome variable was whether or not a patient had PE during the 5-year follow-up period. We individually tracked each patient in this study for 5 years from their index ambulatory visit to identify those who had PE (ICD-9-CM codes 415.1, 415.11 and 415.19). In addition, cases were censored if individuals died from non-PE causes during the 5-year follow-up period (4086 patients died, comprising 643 from the study cohort and 3443 from the control cohort, P = 0.09).
We also took potential confounders into consideration in the regression modeling. These confounders included sociodemographic characteristics (age, sex and the geographic location of the community in which the patient resided) and comorbid medical disorders, including hypertension, diabetes, peripheral vascular disease, hyperlipidemia, coronary heart disease (CHD) and cancer, at baseline. The criteria for identifying these comorbid medical disorders were the diagnosis codes that were either recorded in the inpatient setting or appeared in two or more ambulatory care claims coded 6 months before and after the index ambulatory care visits. In addition, we added the variables of recent surgery (defined as receiving general anesthesia, epidural anesthesia or spinal anesthesia), recent fracture, pregnancy and the use of anticoagulants within 6 months before PE, taking them into consideration in the regression modeling, because of their strong associations with PE.
In this study, we performed all analyses with the sas statistical package (SAS System for Windows, Version 8.2; SAS, Cary, NC, USA). We used Pearson chi-squared tests to examine differences in sociodemographic characteristics and comorbid medical disorders. In addition, the Kaplan–Meier method and log-rank test were used to compare 5-year PE-free survival rates and to examine differences in the risk of PE between the two cohorts during the follow-up period. Finally, we used stratified Cox proportional hazard regressions (stratified on age, gender and the year of index ambulatory care visit) to compute the 5-year PE-free survival rate for these two cohorts, after adjusting for the potential confounders mentioned above. Our data met the proportionality assumption; survival curves for two strata (hyperthyroidism patients and patients in the comparison cohort) had hazard functions that were proportional over time and met the proportionality assumption. A two-tailed level of 0.05 was considered to be significant in this study.
The mean age for the total sample was 40.6 years [standard deviation (SD) 15.3 years), and the mean ages for patients with and without hyperthyroidism were 40.7 and 40.5 years, respectively (P =0.357). Table 1 shows the distributions of sociodemographic characteristics and comorbid medical disorders for patients with and without hyperthyroidism. In total, about 40% of the sample patients were between 18 and 34 years old and 77% were female. After matching for age and sex, Table 1 shows that patients with hyperthyroidism had a greater tendency to have hypertension (P < 0.001), CHD (P < 0.001), diabetes (P < 0.001), peripheral vascular disease (P < 0.001) and hyperlipidemia (P <0.001) at the time of the index ambulatory care visits than patients without hyperthyroidism. In addition, patients with hyperthyroidism were more likely to reside in the northern part of Taiwan than patients in the comparison cohort.
Table 1. Patients with hyperthyroidism and comparison group in Taiwan by sociodemographic characteristics, 2001–2003 (n = 53 418)
Patients with hyperthyroidism N = 8903
Comparison group N = 44 515
Age (years), mean ± standard deviation)
40.7 ± 14.9
40.5 ± 15.5
Age group (years)
Coronary heart disease
Peripheral vascular disease
Use of anticoagulants
NT$15 000–30 000
NT$30 001–50 000
> NT$50 000
The distributions of PE during the 5-year follow-up period for these two cohorts are presented in Table 2. Of the total of 53 418 patients, 41 (0.08%) were identified as having PE during the follow-up period, 14 from the study cohort (0.16% of the hyperthyroidism patients) and 27 comparison patients (0.06% of patients from the comparison cohort). Similarly, the log-rank test suggests that patients with hyperthyroidism had significantly lower 5-year PE-free survival rates than patients in the comparison cohort (log-rank test: 9.036; P =0.003). Figure 1 presents the 5-year PE-free survival rates between these two cohorts by Kaplan–Meier survival analysis. The mean time between ambulatory care visits and the onset of PE was 478 days (SD 494 days); the values for patients with and without hyperthyroidism were 423 and 901 days, respectively (P =0.006).
Table 2. Crude and adjusted hazard ratios for pulmonary embolism during the 5-year follow-up period for patients in Taiwan with hyperthyroidism and patients in the comparison group (n = 53 418)
Presence of pulmonary embolism
Patients with hyperthyroidism
CI, confidence interval; HR, hazard ratio. *Stratified Cox proportional hazard regressions (stratified by sex and age) were performed to adjust for patients’ hypertension, diabetes, hyperlipidemia, peripheral vascular disease, coronary heart disease, cancer, recent surgery, recent trauma, pregnancy, the use of oral anticogulants, monthly income and geographic region; †P < 0.01; ‡P < 0.05.
Five-year follow-up period
Crude HR (95% CI)
Adjusted* HR (95% CI)
Table 2 also shows the crude and adjusted hazard ratios for PE by cohort. Stratified Cox proportional hazard regressions (stratified by age, gender and the year of index ambulatory care visit) reveals that patients with hyperthyroidism were 2.60 times more likely [95% confidence interval (CI) 1.36–4.95, P =0.004] to have PE during the 5-year follow-up period than their counterparts in the comparison cohort. The risk of having PE during the 5-year follow-up period was 2.31 times greater (95% CI 1.20–4.45, P =0.012) for patients with hyperthyroidism than for patients in the comparison cohort, after adjustment for geographic region, monthly income, hypertension, diabetes, peripheral vascular disease, hyperlipidemia, CHD, cancer, recent surgery, recent fracture, pregnancy and the use of anticoagulants.
We performed further analysis regarding the treatment for hyperthyroidism in our study. Of the 8903 patients with hyperthyroidism, 3448 (38.7%) had previously been prescribed antithyroid medications for more than 30 days and nine (0.1%) had undergone thyroidectomy during the 5-year follow-up period. We found there were no significant differences in the numbers of PEs during the 5-year follow-up period among patients who had previously been prescribed antithyroid medications (four PEs in 3448 patients), who underwent thyroidectomy (no PEs in nine patients), and who had received no treatment (10 PEs in 5446 patients) (P =0.730).
To the best of our knowledge, this is the first nationwide population-based study to explore the association between hyperthyroidism and PE. We found that 0.16% and 0.06% of the patients with and without hyperthyroidism, respectively, had PE during the 5-year follow-up period. More specifically, patients with hyperthyroidism were 2.31 times more likely to have PE during the year to year follow-up period than those in the comparison cohort, after taking age, gender, geographic region, monthly income, hypertension, diabetes, peripheral vascular disease, hyperlipidemia, CHD, cancer, recent surgery, recent fracture, pregnancy and the use of anticoagulants into consideration.
Our study results are in agreement with previous observations, from several case reports, of PE and cerebral venous thrombosis in patients with hyperthyroidism [12,14]. A systematic analysis also supports an increased risk of venous thrombotic complications, including cerebral venous thrombosis (CVT), deep vein thrombosis (DVT) and PE in patients with hyperthyroidism . In contrast, one recent retrospective study based on National Hospital Discharge Survey data showed inconsistent results . Danescu et al. reported that among 633 000 US patients discharged from non-Federal hospitals with hyperthyroidism during the years 1979–2005, the risk of PE and DVT was not significantly higher than in all other patients with no thyroid dysfunction (relative risk 0.98, 95% CI 0.96–1.01). However, their study data were limited to inpatient records, and did not include ambulatory care claim records; it is possible that asymptomatic or mild cases of hyperthyroidism or PE were not diagnosed, leading to underestimation of the true incidence of PE in patients with hyperthyroidism.
Three important factors have been highlighted as contributing to the pathophysiology of thrombosis, including hypercoagulation, endothelial dysfunction and local hemodynamic change (stasis and turbulence) . The patients with overt hyperthyroidism had shortened activated partial thromboplastin time, higher fibrinogen levels, and significant increases in the turnover of factor (F)II, FVII and FX [17, 18]. Similarly, patients with hyperthyroidism were found to have FIX, von Willebrand factor (VWF), antithrombin and PAI-1, as well as decreased levels of t-PA, suggesting reduced plasma fibrinolytic capacity, also predisposing them to a hypercoagulable state [19,20]. Furthermore, studies have demonstrated that, in hyperthyroid patients, the balance between t-PA and PAI-1 appears to favor PAI-1, resulting in impaired endothelial function [21–24]. Therefore, patients with hyperthyroidism may often have accompanying endothelial dysfunction, and hypercoagulable states, that contribute to the development of venous thrombosis and increased risk of PE.
It is worth noting that the abnormal coagulation profile could be varied according to the disease status of hyperthyroidism. Recently, one study has shown that elevated levels of free thyroxin in hyperthyroidism patients are significantly associated with an increased risk for venous thrombosis . Hyperthyroidic patients had increased FVIII activity and VWF antigen and ristocetin cofactor levels, which have been reported to become normalized after treatment with antithyroid drugs . These findings imply that the risk for venous thrombosis and associated events could vary according to the disease status and course of hyperthyroidism. In addition, successful treatment for hyperthyroidism and normalizing of thyroid status could be of benefit in preventing venous thrombosis and PE in these patients.
Our study demonstrated that patients with hyperthyroidism had significantly higher risks of having comorbidities such as hypertension, hyperlipidemia and diabetes. Although the pathomechanism is still not well known, altered glucose and lipid metabolism have been reported in patients with hyperthyroidism [16,27]; however, data on the clinical association between hyperthyroidism and dyslipidemia and diabetes are still lacking. These medical comorbidities are all associated with atherosclerosis and arterial thrombosis, which could indirectly predispose patients with hyperthyroidism to having cardiovascular events. Interestingly, increased risk for ischemic stroke in young patients with hyperthyroidism was recently demonstrated . Nevertheless, further study is advised to explore related issues.
A particular advantage of this study was the availability of nationwide population-based datasets, which enabled us to trace the use of medical services for nearly all patients with hyperthyroidism during the study period. Moreover, the large sample size provides ample statistical power to detect differences in the risk of PE between the hyperthyroidism and non-hyperthyroidism groups. However, some limitations should be noted. First, we identified the diagnoses of hyperthyroidism and PE from an administrative database, and these are generally less accurate than diagnoses taken directly from medical records. However, to avoid erroneous diagnoses, we selected only patients who had at least two consensus diagnosed episodes of hyperthyroidism during the study period, and who still had a hyperthyroidism diagnosis after undergoing the blood test for the level of thyroid-stimulating hormone.
Second, although we have adjusted for the influence of some potential confounders, information on protein C and protein S deficiency, oral contraceptive use and body mass index (BMI), all of which may contribute to PE , was not available through our datasets. However, none of the data available indicate that hyperthyroidism patients tend to use more oral contraceptives than the general population. Consequently, there is no reason to infer that this bias would be disproportionate between the hyperthyroidism and control groups. Similarly, the BMI of hyperthyroidism patients generally tends to be lower than that of the general population. Therefore, the risk for PE without adjusting for BMI could be underestimated, not overestimated, among patients with hyperthyroidism, so we believe that the conclusions in our study are not compromised severely. However, further investigation is still necessary to explore other potential causes in the pathogenesis of PE in hyperthyroidism.
Third, our data regarding antithyroid treatment and PE risk should be interpreted carefully, as the status of hyperthyroidism could not be determined in our study. In addition, disease status also varied over the follow-up period. Further research is needed in order to clarify the relationship between PE and a prior prescription for antihyroid medications for more than 30 days. The disease status of our patients could not be determined in our study. Finally, there may be a surveillance bias, in that patients with hyperthyroidism are more likely to receive frequent check-ups than comparison patients, thus increasing the probability of their PE being detected by a physician. However, almost all PE cases need immediate emergency care. Therefore, the possibility of surveillance bias does not compromise our findings.
In summary, this study found an increased risk of PE in patients with hyperthyroidism after adjustment for potential confounders. Physicians should be alert to the increased risk of venous thrombosis, particularly PE, among patients with hyperthyroidism. In circumstances where patients are at risk for venous thrombosis development, such as fracture, immobilization and major surgery, adequate prophylaxis programs to prevent venous thrombosis should be considered, particularly in patients with hyperthyroidism.
This study is based in part on data from the National Health Insurance Research Database provided by the Bureau of National Health Insurance, Department of Health, Taiwan and managed by the National Health Research Institutes. The interpretations and conclusions contained herein do not represent those of the Bureau of National Health Insurance, Department of Health, or the National Health Research Institutes.
Disclosure of Conflict of Interests
The authors state that they have no conflict of interest.