Association between thyroid dysfunction and venous thromboembolism: A systematic review and meta-analysis

Abstract

Background: 

Thyroid dysfunction plays an important role in the development of cardiovascular disease. However, its relationship with venous thromboembolism (VTE) remains unclear. We performed a meta-analysis of published cohort and case-control studies to investigate the association between thyroid dysfunction and VTE comprehensively.

Methods: 

Three reviewers independently searched EMbase, PubMed, China national knowledge infrastructure, and Cochrane Library databases for relevant articles from the time of database establishment to 01 October 2022 and identified all studies on thyroid dysfunction and VTE as studies of interest. Of the 2418 publications retrieved, we identified 10 articles with 15 studies that met our selection criteria. Pooled ORs and 95% confidence intervals were calculated using fixed- or random-effect models.

Results: 

We pooled 8 studies by a fixed-effect model, which suggested an increased risk of VTE in patients with (subclinical) hyperthyroidism (OR 1.33, 95% CI: 1.29–1.38). In the other 7 studies on patients with (subclinical) hypothyroidism, the risk was similarly increased when pooled by a random-effect model (OR 1.52, 95% CI: 1.23–1.89). After sensitivity analysis and risk of bias analysis, the risk of VTE was still increased in both (subclinical) hyperthyroidism (OR 1.322, 95% CI: 1.278–1.368) and (subclinical) hypothyroidism (OR 1.74, 95% CI: 1.41–2.16).

Conclusion: 

Patients with thyroid dysfunction have an increased risk of VTE. Therefore, it is recommended to perform thyroid function screening routinely in patients at high risk of VTE.

1. Introduction

Venous thromboembolism (VTE) includes deep vein thrombosis, and pulmonary embolism. Tumors, surgery, immobilization, advanced age, obesity, smoking habit, respiratory failure, rheumatic immune disease, and other factors leading to blood stasis, hypercoagulable state, and endothelial cell injury can increase the risk of VTE.^[1] Deep vein thrombosis and pulmonary embolism are usually associated with poor prognosis in cardiovascular inpatients, and early diagnosis and intervention of VTE have become the most common and important measures to prevent in-hospital death.^[2]

Thyroid dysfunction, including (subclinical) hyperthyroidism and (subclinical) hypothyroidism, is a very common endocrine disorder that is usually defined according to thyroid-stimulating hormone, triiodothyronine, free thyroxine, free triiodothyronine, and free thyroxine levels.^[3] Thyroid dysfunction is often accompanied by abnormalities in the coagulation and fibrinolysis systems.^[4,5] It is reported that hypercoagulable and hypofibrinolytic states usually characterize hyperthyroidism, while hypothyroidism shows hypocoagulable and hyperfibrinolytic states. Abnormalities in coagulation and fibrinolytic factors have been shown to contribute to the development of thrombosis. However, it is unclear whether the presence of thyroid dysfunction increases the risk of thrombosis. In some clinical observational studies, patients with (subclinical) hyperthyroidism and (subclinical) hypothyroidism were found to have an increased risk of VTE formation. Thyroid diseases with opposite trends in hormone levels may require a more refined assessment of patient coagulation and VTE risk during clinical diagnosis and treatment. Therefore, there is an urgent need to know the risk of VTE in patients with thyroid dysfunction.

This study aimed to assess the risk of VTE in patients with thyroid dysfunction compared with a euthyroid population and to comprehensively investigate whether patients with thyroid dysfunction have an increased risk of VTE by identifying all available cohort and case-control studies and summarizing the results.

2. Methods

Our study complied with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses.^[6] A protocol was registered with PROSPERO, registration number CRD42022369938.

2.1. Literature retrieval

Three reviewers (Y.W., C.D., and C.G.) independently searched PubMed, EMBASE, China national knowledge infrastructure, and the Cochrane Library databases from the time of database establishment to 01 October 2022. The search terms were derived from terms related to “(subclinical) hypothyroidism,” “(subclinical) hyperthyroidism,” “venous thromboembolism,” “pulmonary embolism,” and “deep vein thrombosis” without language restrictions. The retrieval strategy is shown in Supplemental Digital Content (Table S1, https://links.lww.com/MD/I672).

2.2. Inclusion and exclusion criteria

Inclusion criteria: Case-control or cohort design studies. Data on the association between thyroid dysfunction and VTE were reported. VTE was defined based on the definitive diagnosis (by lower-limb venous compression ultrasonography, computed tomography, ventilation-perfusion scan, angiography, or autopsy), or VTE’s diagnostic code could be extracted from medical records. Thyroid dysfunction was defined based on thyroid-stimulating hormone, free triiodothyronine, free thyroxine, or patients with clear medical records on thyroid dysfunction (hyperthyroidism, hypothyroidism, subclinical hyperthyroidism, subclinical hypothyroidism).

Exclusion criteria: Duplicate, non-English or non-Chinese, non-human studies, case reports, reviews, meta-analyses, or conference reports. If there were multiple publications in the same cohort, only the publication with the latest statistical data would be used. Cases complicated with other diseases such as tumors. Literature without full-text access.

2.3. Data extraction and quality assessment

Two authors (Y.W. and C.D.) independently screened abstracts and performed a full-text review to check eligibility for inclusion. A third independent author (C.G.) was consulted to reach a consensus when there was controversy. The following data were extracted from the included studies: author, year, country, study design (case-control study or cohort study), baseline characteristics of patients, sample size, and risk of VTE (HR, RR, OR). OR was calculated from the original data if not provided directly. Study quality was assessed by Newcastle-Ottawa Score (NOS),^[7] and studies with NOS scores≥7 were considered of high quality.

2.4. Data pooling and analysis

The association between thyroid dysfunction and VTE was quantified with OR.^[8] These data were also pooled together by random-effects or fixed-effects meta-analysis to measure the effect of thyroid dysfunction on VTE risk. Integrated analysis of each study was assessed by forest plots. Statistical heterogeneity of the included studies was evaluated using the I² statistic. A random-effects model was used when heterogeneity was significant (P < .1 or I² > 50%); otherwise, a fixed-effects model^[9] was employed. A sensitivity test was performed to evaluate whether a single study could significantly affect the pooled results. Funnel plot, Begg and Egger tests were used to evaluate the publication bias. If there was significant publication bias, the trim-and-fill method was introduced to evaluate the difference in the results.^[10] All analyses mentioned above were performed in STATA statistical software, version 15.1.

3. Results

3.1. Literature retrieval

We retrieved 2418 articles according to the retrieval strategy, of which 289 were duplicate studies. Two thousand and eighty articles were excluded based on the information given in the title and abstract. For the remaining 32 articles, 10 articles with 15 studies were finally included after reviewing the full text for a detailed evaluation. The screening process is illustrated in Figure 1.

Figure 1.: Flow chart of study selection.

3.2. Basic characteristics of included studies

Fifteen studies in 10 articles with a total of 941,731,087 participants were included. Eight studies reported the risk of VTE in patients with (subclinical) hyperthyroidism, of which 7 were cohort studies^[11–16] and 1 was a case-control study.^[17] Regarding study location, 6 studies were conducted in Europe,^{[11–13,15,17]} 1 in North America^[16] and 1 in China.^[14] One study involved women only.^[16] The risk of VTE in patients with (subclinical) hypothyroidism was reported in the other 7 studies included, of which 6 were cohort studies^[13,18–20] and 1 was a case-control study.^[17] Five studies were conducted in Europe^{[13,17,18,20]} and 2 studies were conducted in China.^[19] Three studies were conducted only in women,^[17,19,20] 2 studies only in men,^[19,20] and the remaining studies involved both genders. The basic characteristics of the included studies are shown in Table 1.

Table 1 - Characteristics of studies included in this meta-analysis.

Author, year	Country	Study design	Types of thyroid dysfunction	Thyroid dysfunction confirmation	Population (n)	Events (n)	Age		Gender		Multi-variates analysis	NOS
							Cases	Comparators	Male	Female
Danescu, 2009	USA	Retrospective cohort	Hypothyroidism	ICD-9-CM	927,651,000	10,497,000	NA	NA	373,581,000	554,070,000	NA	5
Dekkers 2017	Denmark	Retrospective cohort	Hyperthyroidism	ICD-8,10	932,913	85,856	NA	NA	168,820	764,093	Age, sex, calendar years and other co-morbidities	8
Hizkiyahu, 2022	Canada	Retrospective cohort	Hyperthyroidism	ICD-9-CM	9,096,788	NA	NA	NA	–	9,096,788	Age, race, type of health insurance, obesity, smoking and other co-morbidities	8
Lerstad, 2016	Norway	Prospective cohort	(Subclinical) hyperthyroidism	TSH < 0.5 mlU/L	17,061	262	62 ± 8	61 ± 12	7453	9608	BMI and smoking	8
			(Subclinical) hypothyroidism	TSH > 5.0 mlU/L	17,491	271	65 ± 10	61 ± 12	7664	9827	BMI and smoking
Lin, 2010	China	Retrospective cohort	Hyperthyroidism	ICD-9-CM	53,418	8903	40.7 ± 14.9	40.5 ± 15.5	12,312	41,106	Age, sex, oral anticoagulant use, monthly income, geographic region, and other co-morbidities	8
Ramagopalan, 2011(England)	UK	Retrospective cohort	Hyperthyroidism	ICD-7,8,9-CM	3,707,315	91,913	NA	NA	2,187,316	1,519,999	Age, sex, calendar years and region of residence	7
Ramagopalan, 2011(ORLS)	UK	Retrospective cohort	Hyperthyroidism	ICD-7,8,9-CM	187,609	NA	NA	NA	101,309	86,300	Age, sex, calendar years and region of residence	7
Segna, 2016	Switzerland	Prospective cohort	(Subclinical) hypothyroidism	TSH 4.5–19.99 mlU/L	535	52	Median: 75	Median: 74	325	210	Age, sex, previous venous thromboembolism, active cancer	7
Wei, 2020	China		Hypothyroidism	ICD-9-CM	32,454	–	48.9 ± 15.4	48.9 ± 15.4	5835	26,619	Age, sex, income level and other co-morbidities	8
Zaane, 2010	Netherlands	Case-control	(Subclinical) hyperthyroidism/hypothyroidism	Medical records	569	190	Median: 57	Median: 56	240	329	NA	7
Zoller, 2012	Sweden	Retrospective cohort	Hyperthyroidism	ICD-7,8,9,10-CM	50,954	NA	NA	NA	147,802	174,191	Age, sex, period and other co-morbidities	7

BMI = body mass index, NA = not available, NOS = Newcastle-Ottawa Score

3.3. Quality evaluation

Newcastle-Ottawa Score scores were employed to assess the included studies, with the majority of studies (13 studies in 9 articles) being high-quality studies (score ≥ 7.0) and only 2 studies in 1 article^[20] considered of moderate quality (NOS score 5). Quality scores for the included studies are presented in Supplemental Digital Content (Table S2A,B, https://links.lww.com/MD/I673).

3.4. Data analysis

3.4.1. Pooled results of (subclinical) hyperthyroidism.

In the 8 studies involving (subclinical) hyperthyroidism, we pooled the VTE risk using a fixed-effects model to compare (subclinical) hyperthyroid patients with euthyroid populations: OR 1.33 (95% CI: 1.29–1.38, I² = 27.8%, P = .206) (Fig. 2). A sensitivity analysis of the included studies, ignoring 1 study at a time and recalculating the pooled OR for the other studies, indicated that none had a large impact on the results (Fig. 3), suggesting that the pooled results were relatively robust. Then the funnel plot was introduced to detect publication bias, which was not symmetrical (Fig. 4), and there may be publication bias that impacted the results. Further Begg and Egger tests were performed, and the Egger test also suggested publication bias (Begg test = 0.174, Egger test = 0.007). On this basis, we tried to use the trim-and-fill method to recalculate the results, and 5 missing studies were added to make the funnel plot symmetrical (see Fig. 5). After pooling the 5 theoretical studies, the OR was 1.322 (95% CI, 1.278–1.368), which did not substantially change from the previous result.

Figure 2.: Forest plot of (subclinical) hyperthyroidism and risk of VTE for all included studies (fixed effect model). VTE = venous thromboembolism.

Figure 3.: The sensitivity analysis of (subclinical) hyperthyroidism and risk of VTE for all included studies. VTE = venous thromboembolism.

Figure 4.: The funnel plot of (subclinical) hyperthyroidism and risk of VTE for all included studies. VTE = venous thromboembolism.

Figure 5.: Funnel plots following trim-and-fill.

3.4.2. Pooled results of (subclinical) hypothyroidism.

In the 7 studies involving (subclinical) hypothyroidism, the random-effects model was used to pool the VTE risk results: OR 1.52 (95% CI: 1.23–1.89, I² = 99.9%, P = .00) (Fig. 6). We found large heterogeneity in this pooled result, and sensitivity analysis suggested that the results were largely impacted by 2 studies conducted by Danesce, 2009^[20] (Fig. 7). Removing the 2 studies resulted in no significant change in results but led to a clear decrease in heterogeneity (including Danesce, 2009: OR 1.52 [95% CI: 1.23–1.89, I² = 99.9%, P = .00]; excluding: OR 1.74 [95% CI: 1.41–2.16, I² = 7.1%, P = .366]). We believe that Danesce’s study only selected diagnostic codes extracted from patient discharge information as raw data and was not adjusted by other confounding factors when calculating the RR value. The NOS score of the 2 studies involved in this article indicated that they were of moderate quality. More importantly, the 95% CI range of RR value in this study was narrow, giving the 2 studies heavier weight when the data were pooled. After excluding the 2 studies, the results suggested low heterogeneity. Therefore, Danesce’s^[20] studies were the main source of heterogeneity in the results. The funnel plot showed asymmetry (see Fig. 8), and further Begg and Egger tests indicated no publication bias (Begg test = 0.806, Egger test = 0.083).

Figure 6.: Forest plot of (subclinical) hypothyroidism and risk of VTE for all included studies(random effect model). VTE = venous thromboembolism.

Figure 7.: The sensitivity analysis of (subclinical) hypothyroidism and risk of VTE for all included studies. VTE = venous thromboembolism.

Figure 8.: The funnel plot of (subclinical) hypothyroidism and risk of VTE for all included studies. VTE = venous thromboembolism.

4. Discussion

This study is a meta-analysis aiming to evaluate the association between thyroid dysfunction and VTE. There was another meta-analysis investigating the risk of VTE in patients with hyperthyroidism^[21] but with some limitations: fewer studies (n = 5) were included, and the associated VTE risk in patients with (subclinical) hypothyroidism was not explored. Therefore, it is necessary to conduct a new meta-analysis to comprehensively assess the association between different types of thyroid dysfunction and VTE.

This meta-analysis found a significantly higher risk of VTE in patients with thyroid dysfunction, in both (subclinical) hyperthyroidism and (subclinical) hypothyroidism, compared with euthyroid individuals. However, the underlying mechanism is still unclear, especially in hypothyroidism, with conflicting study results.

In a meta-analysis investigating coagulation changes in hyperthyroid states, 29 articles, including 51 studies, were included, and the results showed that coagulation factors VIII, IX, von Willebrand factor (vWF), fibrinogen, and plasminogen activator inhibitor-1 (PAI-1) levels were increased in (subclinical) hyperthyroid patients.^[22] This may be caused by high levels of thyroid hormone acting directly on hepatocytes and endothelial cells,^[23,24] causing patients to be in a hypercoagulable and hypofibrinolytic state for a long time, ultimately leading to venous thrombosis, which is consistent with the conclusions drawn from our analysis. In an observational study, plasma vWF and factor VIII returned to normal values in patients with hyperthyroidism treated with thiamazole,^[25] but no studies have further explored changes in the prevalence of VTE after treatment of hyperthyroidism.

In addition, thyroid dysfunction is often strongly associated with autoimmune diseases. Studies have shown that 22% of patients with primary antiphospholipid syndrome have hypothyroidism.^[26] In contrast, multiple cohort studies suggest that the prevalence of positive serum antiphospholipid antibodies is also increased in patients with hyperthyroidism,^[27,28] and there may be a specific association between thyroid dysfunction and primary antiphospholipid syndrome. Some studies have also discovered that (subclinical) hyperthyroidism and (subclinical) hypothyroidism are both found to be closely related to rheumatoid arthritis.^[29] Autoimmune diseases often lead to endothelial cell damage and increase the incidence of VTE. Therefore, this may also factor in the increased risk of VTE in patients with thyroid dysfunction. Gender may also act as an important risk factor for increased VTE risk in patients with thyroid dysfunction. Firstly, female patients are at higher risk of autoimmune diseases,^[30] and there is also a higher risk of having thyroid dysfunction along with other autoimmune diseases.^[31] In addition, young and middle-aged women often have an increased risk of VTE due to oral contraceptives or bed rest during pregnancy and lower limb blood stasis.^[32] Therefore, we found that women with thyroid dysfunction had a higher risk of VTE than men in the included studies,^[19] and further large-scale cohort studies are still needed to evaluate further the association between gender and the risk of VTE after thyroid dysfunction in the future.

There are large contradictions in studies related to hypothyroidism and coagulation changes. In early studies, Simone et al^[33] found decreased coagulation factor VIII, factor IX, and factor XI levels and corresponding decreased coagulation function in hypothyroid patients. Gullu et al^[34] observed coagulation-related factors before and after levothyroxine treatment in 15 patients with hypothyroidism and found that factor VIII and vWF activities in hypothyroid patients were lower than those in the normal population before treatment, and prothrombin time and activated partial thromboplastin time were increased; these coagulation abnormalities could be corrected after 3 months of levothyroxine replacement therapy. In a study of coagulation status in patients with subclinical hypothyroidism, different phenomena were found. A recent meta-analysis of 12 studies with a total of 1325 patients^[35] showed that although activated partial thromboplastin time and D-dimer were not significantly different in patients with subclinical hypothyroidism compared with the normal group, tissue plasminogen activator and PAI-1 were higher than those in the normal control group and appeared to have a bidirectional effect on coagulation. The increase of tissue plasminogen activator may be a compensatory reaction of human body after the increase of PAI-1. In conclusion, most studies on coagulation function in hypothyroidism suggest that patients with overt hypothyroidism have a hypocoagulable state and bleeding tendency; while subclinical hypothyroidism may induce a prothrombotic state, PAI-1 may play an important role in it.

In our clinical observational studies based on meta-analysis, both hypothyroidism and subclinical hypothyroidism suggested an increased risk of VTE. Studies^[19] have further explored the relative changes in the risk of VTE after thyroid hormone replacement therapy: ①The risk of VTE in hypothyroid people using thyroid hormone replacement therapy remains high compared with people without hypothyroidism (HR: 1.54 [1.17–2.04]), but the risk is lower compared with hypothyroid patients without thyroid hormone replacement therapy (HR: 0.73 [0.52–1.01]). This supports the notion that hypothyroid patients are at increased risk of VTE, and hormone replacement therapy effectively reduces this risk.

In addition to the above factors causing damage to vascular endothelial cells in patients with thyroid dysfunction often associated with autoimmune-related diseases, blood stasis may also be one of the risk factors for increased risk of VTE in hypothyroid patients, and hypothyroid patients often have myxedema, which leads to venous blood stasis. It has been shown that myxedema is considered to be one of the risk factors for VTE.^[11] Therefore, although patients with overt hypothyroidism have decreased coagulation and increased bleeding risk, the risk of VTE remains increased under the influence of other combined factors. Prophylactic anticoagulant therapy for high-risk VTE patients with hypothyroidism requires additional vigilance for the risk of bleeding during clinical diagnosis and treatment.

Studies have shown that thyroid dysfunction is often closely related to the prognosis of VTE. Analysis of Pohl et al^[36] on the short-term and long-term outcomes in 831 patients with pulmonary embolism showed that hyperthyroidism increased in-hospital mortality in patients with pulmonary embolism. In contrast, in long-term follow-up, both hyperthyroidism and hypothyroidism were associated with an increased risk of long-term mortality. In a chronic thromboembolic pulmonary hypertension (CTEPH) study, thyroid hormone replacement therapy in hypothyroid patients appeared to lead to an increased proportion of CTEPH,^[37] and hypothyroid patients with CTEPH have more severe clinical symptoms, higher NT-proBNP and lower cardiac index than those without CTEPH,^[38] of which the mechanism is still unclear, and further studies are needed to elucidate their association. Nevertheless, current observational evidence suggests that early diagnosis and treatment of thyroid dysfunction are important in clinical VTE prevention and prognosis.

There are 2 points worth noting in this study, the funnel plot of VTE risk in (subclinical) hyperthyroid patients was initially not symmetrical, and the Egger test suggested publication bias. There was no substantial effect on the results after supplementing 5 theoretical studies with the trim-and-fill method. Therefore, our finding is still significant even in the presence of publication bias. In the VTE risk analysis of (subclinical) hypothyroidism, Danesce’s studies^[20] had a major impact on the pooled results. Not only was the study of low quality (NOS score = 5), but it also accounted for a large weight in the pooled analysis, resulting in a significant heterogeneity of the results. After excluding the 2 studies of this article, the results were robust and only suggested mild heterogeneity. The conclusion was consistent with or without Danesce’s studies, indicating an increased risk of VTE in (subclinical) hypothyroid patients.

This study has several limitations. Firstly, patients with thyroid dysfunction are often associated with other symptoms or diseases due to hormonal disturbances, and the included studies did not systematically count data on other potential risk factors for VTE (e.g., obesity, estrogen therapy/contraception, immobilization, etc.), which may limit the accuracy of confounding factor adjustment in multivariate analysis. Secondly, the definition of thyroid dysfunction differed among the included studies, and there was insufficient data to distinguish subclinical thyroid dysfunction from overt thyroid dysfunction, which may have led to a slight bias in the results. More clinical studies are needed to confirm our conclusions in the future.

5. Conclusion

In conclusion, this literature review-based meta-analysis suggests that patients with 2 opposite types of thyroid dysfunction ([subclinical] hyperthyroidism [subclinical] hypothyroidism) both have an increased risk of VTE. Therefore, routine thyroid function screening in patients at high risk for VTE is warranted in clinical practice.

Author contributions

Conceptualization: Yunmeng Wang, Jingru Wang.

Data curation: Yunmeng Wang, Chaowei Ding.

Formal analysis: Yunmeng Wang.

Investigation: Yunmeng Wang, Chang Guo.

Methodology: Yunmeng Wang, Chaowei Ding.

Visualization: Yunmeng Wang.

Writing – original draft: Yunmeng Wang.

Writing – review & editing: Suyun Liu.