Clin Rheumatol (2014) 33:297–304 DOI 10.1007/s10067-014-2492-7
REVIEW ARTICLE
Risk of venous thromboembolism in patients with rheumatoid arthritis: a systematic review and meta-analysis Patompong Ungprasert & Narat Srivali & Ittikorn Spanuchart & Charat Thongprayoon & Eric L. Knight
Received: 31 December 2013 / Accepted: 5 January 2014 / Published online: 15 January 2014 # Clinical Rheumatology 2014
Abstract We performed this meta-analysis to assess venous thromboembolism risk in patients with rheumatoid arthritis. A comprehensive search was performed in MEDLINE, EMBASE, and the Cochrane databases. Nine observational studies met our inclusion criteria and were included in the data analysis. The pooled risk ratios of deep venous thrombosis, pulmonary embolism, and venous thromboembolism in patients with rheumatoid arthritis (RA) compared with non-RA participants were 2.08 (95 % CI 1.75–2.47), 2.17 (95 % CI 2.05–2.31), and 1.96 (95 % CI 1.81–2.11), respectively. Subgroup analysis demonstrated a consistent increased risk in every study design (cohort, case–control, and crosssectional). Our results indicate a significant increased risk of venous thromboembolism among patients with rheumatoid arthritis. Keywords Deep venous thrombosis . Pulmonary embolism . Rheumatoid arthritis . Venous thromboembolism
Introduction Venous thromboembolism (VTE), which includes deep venous thrombosis (DVT) and pulmonary embolism (PE), is a P. Ungprasert (*) : N. Srivali : E. L. Knight Department of Internal Medicine, Bassett Medical Center and Columbia University College of Physicians and Surgeons, Cooperstown, NY, USA e-mail: [email protected] I. Spanuchart Department of Internal Medicine, University of Hawaii, Honolulu, HI, USA C. Thongprayoon Division of Nephrology and Hypertension, Department of Medicine, Mayo Clinic, Rochester, MN, USA
common medical problem with an annual incidence of 1–2 new cases per 1,000 population [1, 2]. It confers a significant morbidity and mortality with a reported 30-day case fatality rate of 11–30 % [1–3]. Several conditions, such as immobilization, surgery, major trauma, and malignancy, are well recognized as risk factors for VTE. Chronic inflammatory disorders, though not generally viewed as traditional risk factors for VTE, might increase the risk of developing VTE as inflammatory cytokines have been demonstrated to modulate the coagulation pathway by upregulating procoagulants and downregulating anticoagulant and fibrinolytic systems. Furthermore, chronic inflammation has been shown to be associated with endothelial dysfunction [4]. While there are a considerable number of epidemiologic studies regarding the morbidity and mortality from cardiovascular diseases in patients with rheumatoid arthritis (RA), one the most common chronic inflammatory disorders, the data on VTE risk in this population are relatively limited. Several recent studies demonstrated an increased VTE risk among these patients though the results were fairly heterogeneous. Thus, to obtain a more accurate and precise estimated effect, we conducted a systematic review and meta-analysis of observational studies that compared the risk of VTE in patients with RA versus non-RA participants.
Method Search strategy Two investigators (P.U. and N.S.) independently searched published studies indexed in MEDLINE, EMBASE, and the Cochrane databases from inception to November 2013 using the terms “pulmonary embolism,” “deep venous thrombosis,” and “venous thromboembolism” combined with the term
298
Clin Rheumatol (2014) 33:297–304
“rheumatoid arthritis.” A manual search of references of selected retrieved articles was also performed. Abstract and unpublished studies were not included.
Inclusion criteria The inclusion criteria were as follows: (1) observational studies (case–control, cross-sectional, or cohort studies) published as original studies to evaluate the association between RA and risk of VTE, DVT, or PE; (2) odds ratios (OR), relative risk (RR), or hazard ratio (HR) with 95 % confidence intervals (CIs) or survival curves were provided; and (3) random nonRA participants were the reference group. Study eligibility was independently determined by each investigator noted above. Differing decisions were resolved by consensus. The quality of each study was independently evaluated by each investigator using Newcastle-Ottawa quality assessment scale [5]. Fig. 1 Outline of our search methodology and study selection
Statistical analysis Data analysis was performed using the Review Manager 5.2 software from the Cochrane Collaboration. We reported the pooled effect estimate of DVT, PE, and VTE (DVT and/or PE). We combined data from case–control, cross-sectional study, and cohort analyses to increase the precision of our estimates. Nonetheless, we also performed a separate analysis according to study design. As the outcome under study was relatively uncommon, we used the OR of case–control studies as the estimate of the RR to pool this data with the RR or HR from cross-sectional studies and cohort studies. Point estimates and standard error were extracted from individual studies and were combined by the generic inverse variance methods. We ran a random-effect model rather than a fixedeffects model because of the high likelihood of between study variance. Statistical heterogeneity was assessed using the Cochran’s Q test. This statistic was complemented with I2 statistics,
Potentially relevant articles identified from search of MEDLINE, EMBASE and the Cochrane database and screened for retrieval (n=335)
Detailed review of potentially relevant articles
323 articles were excluded since they were not observational studies or were not done in patients with RA 12 potentially relevant articles included for fulllength article review
2 articles were excluded since participants received anticoagulants before entering the studies 1 article was excluded since the reference group included RA participants
9 articles were included in the meta-analysis
Taiwan Case–control 2012 Hospitalized or non-hospitalized patients who were diagnosed with DVT, identified from national insurance database Duplex ultrasound scan or venography Sex and age-matched subject randomly selected from same database 2001–2009 18 to >69 58.7 5,193 20,772 Diagnostic code from the registry. The diagnosis needed to be made by rheumatologist and patient needed to take DMARD Hospitalization, pregnancy, fracture, surgery, cancer, hypertension, diabetes, coronary artery disease, heart failure, renal disease, hyperlipidemia, inflammatory bowel disease Selection: 4 stars Comparability: 1 star Exposure: 2 stars
NA Sex and age-matched subject randomly recruited from same geographical area 1999–2004 18–70 54.0 4,311 5,768 Self report using questionnaire
Hospitalization, pregnancy, fracture, surgery, cancer, diabetes, coronary artery disease, renal disease, COPD, osteoporosis, medications, liver disease, obesity, recent infection, medications Selection: 4 stars Comparability: 1 star Exposure: 1 star Hospitalization, pregnancy, fracture, surgery, cancer, diabetes, coronary artery disease, renal disease, COPD, osteoporosis, medications, liver disease, obesity Selection: 3 stars Comparability: 1 star Exposure: 1 star
The Netherlands Case–control 2012 All ambulatory patients with the diagnosis with DVT and/or PE who attended anticoagulation clinics
Ocak et al. [9]
Denmark Case–control 2012 Hospitalized or non-hospitalized patients who were diagnosed with DVT and/or PE identified from national registry database NA Sex and age-matched subject randomly selected from same database 1999–2009 18–99 52.9 14,721 147,210 Diagnostic code from the registry
Johannesdottir et al. [8]
DVT deep venous thrombosis, PE pulmonary embolism, NA not available, DMARD disease modifying anti-rheumatic drug, COPD chronic obstructive pulmonary disease
Quality assessment (Newcastle-Ottawa scale)
Confounder assessed
Period of inclusion Age range, years Woman, % Number of cases Number of control Diagnosis of RA
Case verification Controls
Country Study design Year Cases
Kang et al. [7]
Table 1 Main characteristics of case–control studies included in the meta-analysis
Clin Rheumatol (2014) 33:297–304 299
57.0 69.0 7,904 37,350 5.8 Hospitalization
Sex and age-matched subject randomly selected from same database
Duplex ultrasound scan or venography or CT angiography or autopsy
Until death, migration from the system or December 31, 2008
55.6 69.0 464 464 9.6
Hospitalization
Controls
Event verification
Follow up
Mean age, Y Woman, % Number of cases Number of control Average range of follow up, Y Confounder assessed
Sex, age, and residential area-matched subject randomly selected from Swedish population registry Patients needed to fill prescriptions for vitamin K antagonist or heparin Until death, migration from the system or December 31, 2010
Diagnostic code from the registry
Fulfilled at least 4 of the 7 ACR 1987 criteria
Diagnosis of RA
Sweden Prospective cohort 2012 Ambulatory patients who were diagnosed with RA between 1997 and 2009. Cases were identified by using Swedish national registry.
USA Retrospective cohort 2012 Residences of Olmstead county who were diagnosed with RA between 1980 and 2007. Cases were identified by using Mayo Clinic and its affiliations database.
Holmqvist et al. [12]
Country Study design Year Cases
Bacani et al. [11]
Table 2 Main characteristics of cohort studies included in the meta-analysis
Until death, migration from the system or June 30, 2008 Until death, experience the outcome, migration from the system or December 31, 2010 52.0 77.0 29,238 116,952 6.6
Until death, experience the outcome or May 31, 2010
Hospitalization, varicose vein, fracture, surgery, cancer, smoking, COPD, diabetes, coronary artery disease, heart failure, renal disease, BMI, inflammatory bowel disease
Atrial fibrillation, hypertension, diabetes, hyperlipidemia, stroke, heart failure, fracture, surgery, cancer
Patients needed to receive anticoagulation therapy
Diagnostic code, confirmed by peer review
Patients needed to receive anticoagulation therapy
Hospitalization, varicose vein, fracture, surgery, COPD, diabetes, heart failure, cancer, renal disease, obesity, stroke
52.2 75.0 22,143 88,572 2.0
Sex and age-matched subject randomly selected from same database
Sex and age-matched subject randomly selected from same database
58.3 69.4 9,589 95,776 5.5
Diagnostic code from the registry + at least one use of DMARDs
Fulfilled at least 4 of the 7 ACR 1987 criteria
Kim et al. [15] USA Retrospective cohort 2013 Hospitalized or non-hospitalized patients who were diagnosed RA between 2001 and 2008. Cases were identified by using the Medicare and a private health insurance company database.
Chung et al. [14] Taiwan Prospective cohort 2013 Hospitalized or non-hospitalized patients who were diagnosed RA between 1998 and 2008. Cases were identified by using the National health insurance database (which covers 99 % of entire population).
UK Retrospective cohort 2013 Ambulatory patients who were diagnosed RA between 1986 and 2010. Cases were identified by using The Health Improve Network (UK computerized medical record database comprised of 7.3 million patients). Diagnostic code from the registry + at least one use of DMARDs Sex and age-matched subject randomly selected from same database
Choi et al. [13]
300 Clin Rheumatol (2014) 33:297–304
DVT deep venous thrombosis, PE pulmonary embolism, DMARD disease modifying anti-rheumatic drug, COPD chronic obstructive pulmonary disease, CT computerized tomography, ACR American College of Rheumatology
Selection: 4 stars Comparability: 1 star Outcome: 2 stars Selection: 4 stars Comparability: 1 star Outcome: 3 stars Selection: 4 stars Comparability: 1 star Outcome: 3 stars Selection: 4 stars Comparability: 1 star Outcome: 3 stars Selection: 4 stars Comparability: 2 stars Outcome: 3 stars
Chung et al. [14] Choi et al. [13]
Quality assessment (Newcastle-Ottawa scale)
Table 2 (continued)
Bacani et al. [11]
Holmqvist et al. [12]
Kim et al. [15]
Clin Rheumatol (2014) 33:297–304
301
which quantify the proportion of total variation across studies that is due to heterogeneity rather than chance. A value of I2 of 0 to 25 % represents insignificant heterogeneity, 25 to 50 % low heterogeneity, 50 to 75 % moderate heterogeneity, and 75 to 100 % high heterogeneity [6].
Result Our search strategy yielded 335 potentially relevant articles. Three hundred twenty-three articles were excluded as they were not observational studies or were not done in patients with RA. Twelve articles underwent full-length article review; two of them were excluded since participants received anticoagulants before entering the studies, and one of them was excluded since the reference group included RA participant. Nine studies (five cohort studies, three case–control studies, and a cross-sectional study) with 606,427 subjects met our inclusion criteria and were included in the analysis [7–15]. Eight studies reported the RR of VTE in patients with RA compared with non-RA participants [7–13, 15], six studies reported the RR of DVT [7, 10, 12–15], while other five studies reported the RR of PE [10, 12–15]. Figure 1 outlines our search methodology and selection process. The detailed characteristics and quality assessment of the included studies are described in Tables 1 and 2. The pooled risk ratios of DVT, PE, and VTE of subjects with RA versus control subjects were 2.08 (95 % CI 1.75– 2.47), 2.17 (95 % CI 2.05–2.31), and 1.96 (95 % CI 1.81– 2.11), respectively. The statistical heterogeneity was high in DVT studies (I2 of 92 %), moderate in VTE studies (I2 of 66 %), and negligible in PE studies (I2 of 22 %).
Subgroup analysis As shown in Figs. 2, 3, and 4, the risk ratios DVT, PE, and VTE in subjects with RA were significantly higher in every study design subgroup. For DVT, the risk ratios were 2.08 (95 % CI 1.49–2.92), 2.23 (95 % CI 1.72–2.87), and 1.90 (95 % CI 1.88–1.91) for cohort, case–control, and crosssectional studies, respectively. The comparison between studies revealed no significant subgroup difference (I2 of 0 %). For PE, the risk ratios were 2.02 (95 % CI 1.84–2.23) for cohort studies and 2.25 (95 % CI 2.23–2.27) for cross-sectional study. The subgroup heterogeneity was high with I2 of 78.5 %. Lastly, the risk ratio of VTE were 1.90 (95 % CI 1.55–2.33), 1.95 (95 % CI 1.56–2.45), and 1.99 (95 % CI 1.98–2.01) for cohort, case–control, and cross-sectional studies without a significant subgroup difference (I2 of 0 %).
302
Clin Rheumatol (2014) 33:297–304
Fig. 2 Forest plot of DVT studies
Publication bias Funnel plots to evaluate publication bias are summarized in Fig. 5. The graph of the eight VTE studies, although fairly symmetric, provides a suggestion that publication bias in favor of positive studies may be present.
Discussion Our meta-analysis demonstrated a significant association between RA and VTE with an overall 1.94-fold (95 % CI 1.79– 2.10) increased risk compared with non-RA participants. The risk ratios were fairly consistent across the studies, ranging from 1.40 to 3.60, regardless of study design. Our study also
Fig. 3 Forest plot of PE studies
demonstrated an increased risk for DVT and PE for patients with RA with risk ratios of 2.08 and 2.17, respectively. Heterogeneity between studies was present in this metaanalysis particularly for DVT studies. We suspect that the difference in case selection was the main source of heterogeneity, as approximately half of the studies were done in both hospitalized and non-hospitalized subjects while another half was done exclusively in ambulatory setting. We did perform a sensitivity analysis for each setting and found no significant difference for the pooled RR and 95 % CI (data is not shown). Why patients with RA have a higher risk of VTE compared with non-RA subjects is unclear. The occurrence of VTE is linked to three determinative factors, known as Virchow triad, including hypercoagulability, endothelial injury/dysfunction, and venous stasis. RA appears to inter-
Clin Rheumatol (2014) 33:297–304
303
Fig. 4 Forest plot of VTE studies
fere with all of these provocative factors as chronic inflammation related to this autoimmune disease has been demonstrated to promote procoagulation, impair the anticoagulation pathway, and inhibit the fibrinolytic process [4, 16, 17]. Endothelial dysfunction in patient with RA has also been extensively documented [4, 18]. Moreover, immobilization from joint pain and compromised regional blood flow from the compression of femoral vein from the swollen, inflamed iliopsoas bursa can further pose patients at an increased risk of flow stasis and DVT [19]. Even though the nine studies included in this metaanalysis were of high quality, there are some limitations and, thus, the result should be interpreted with caution. First, we cannot exclude the possibility of publication bias as the funnel plot is relatively asymmetric. Second, most of
the included studies were conducted using medical registrybased database, thus maybe coding inaccuracy. Third, this is a meta-analysis of observational studies, and this type of study can demonstrate an association, not establish cause and effect, so we cannot be certain that RA itself versus other potential confounders cause the increase risk of VTE. Furthermore, the higher detection rate of VTE in patients with RA might be partly due to the fact that they are sicker and, thus, exposed more to medical community. In conclusion, the results of our meta-analysis demonstrate a statistically significant increased VTE risk among patients with RA. As VTE confers a high morbidity and mortality, our study suggests that physicians should carefully monitor patients with RA for VTE, although the role of VTE prophylaxis remains unclear and merits further investigations.
Disclosures None. Funding None.
References
Fig. 5 Funnel plot of VTE studies
1. Naess IA, Christiansen SC, Romundstad P et al (2007) Incidence and mortality of venous thrombosis: a population-based study. J Thromb Haemost 5:692–699 2. Cushman M, Tsai AW, White RH et al (2004) Deep venous thrombosis and pulmonary embolism in two cohorts: the longitudinal investigation of thromboembolism etiology. Am J Med 117:19–25 3. Heit JA, Silverstein MD, Mohr DN et al (1999) Predictor of survival after deep venous thrombosis and pulmonary embolism: a population-based, cohort study. Arch Intern Med 159:445–453
304 4. Xu J, Lupu F, Esmon CT (2010) Inflammation, innate immunity and blood coagulation. Hamostaseologie 30(5–6):8–9 5. Stang A (2010) Critical evaluation of the Newcastle-Ottawa scale for the assessment of the quality of nonrandomized studies in metaanalyses. Eur J Epidemiol 25:603–605 6. Higgins JP, Thompson SG, Deeks JJ, Altman DG (2003) Measuring inconsistency in meta-analyses. BMJ 327:557–560 7. Kang JH, Keller JJ, Lin YK, Lin HC (2012) A population-based case–control study on association between rheumatoid arthritis and deep vein thrombosis. J Vasc Surg 56:1642–1648 8. Johannesdottir SA, Schmidt M, Horvath-Puho E, Sorensen HT (2012) Autoimmune skin and connective tissue diseases and risk of venous thromboembolism: a population-based case–control study. J Thromb Haemost 10:815–821 9. Ocak G, Vossen CY, Verduijn M et al (2013) Risk of venous thrombosis in patients with major illnesses: result from the MEGA study. J Thromb Haemost 11:116–123 10. Matta F, Singala R, Yaekoub AY, Najjar R, Stein PD (2009) Risk of venous thromboembolism with rheumatoid arthritis. Thromb Haemost 101:134–138 11. Bacani AK, Gabriel SE, Crowson CS, Heit JA, Matteson EL (2012) Noncardiac vascular disease in rheumatoid arthritis. Arthritis Rheum 64: 53–61 12. Holmqvist ME, Neovius M, Eriksson J et al (2012) Risk of venous thromboembolism in patients with rheumatoid arthritis and
Clin Rheumatol (2014) 33:297–304
13.
14.
15.
16.
17.
18.
19.
association with disease duration and hospitalization. JAMA 308: 1350–1356 Choi HK, Rho YH, Zhu Y, Cea-Soriano LC, Avina-Zubieta JA, Zhang Y (2013) The risk of pulmonary embolism and deep vein thrombosis in rheumatoid arthritis: a UK population-based outpatient cohort study. Ann Rheum Dis 72:1182–1187 Chung WS, Peng CL, Lin CL et al (2013) Rheumatoid arthritis increases the risk of deep vein thrombosis and pulmonary thromboembolism: a nationwide cohort study. Ann Rheum Dis August 13 Kim SC, Schneeweiss S, Liu J, Solomom DH (2013) Risk of venous thromboembolism in patients with rheumatoid arthritis. Arthritis Care Res 65:1600–1607 Montecucco F, Mach F (2009) Common inflammatory mediators orchestrate pathophysiological processes in rheumatoid arthritis and atherosclerosis. Rheumatology (Oxford) 48:11–22 Conway EM, Rosenberg RD (1988) Tumor necrosis factor suppresses transcription of the thrombomodulin gene in endothelial cells. Mol Cell Biol 8:5588–5592 Hansen S, Lassig G, Pistrosch E, Passauer J (2003) Endothelial dysfunction in young patients with long-term rheumatoid arthritis and low disease activity. Atherosclerosis 170:177–180 McLaughlin GE (2002) Sudden death in rheumatoid arthritis: pulmonary embolism—a fatal complication of iliopsoas bursitis. J Clin Rheumatol 8:208–211
REVIEW ARTICLE
Risk of venous thromboembolism in patients with rheumatoid arthritis: a systematic review and meta-analysis Patompong Ungprasert & Narat Srivali & Ittikorn Spanuchart & Charat Thongprayoon & Eric L. Knight
Received: 31 December 2013 / Accepted: 5 January 2014 / Published online: 15 January 2014 # Clinical Rheumatology 2014
Abstract We performed this meta-analysis to assess venous thromboembolism risk in patients with rheumatoid arthritis. A comprehensive search was performed in MEDLINE, EMBASE, and the Cochrane databases. Nine observational studies met our inclusion criteria and were included in the data analysis. The pooled risk ratios of deep venous thrombosis, pulmonary embolism, and venous thromboembolism in patients with rheumatoid arthritis (RA) compared with non-RA participants were 2.08 (95 % CI 1.75–2.47), 2.17 (95 % CI 2.05–2.31), and 1.96 (95 % CI 1.81–2.11), respectively. Subgroup analysis demonstrated a consistent increased risk in every study design (cohort, case–control, and crosssectional). Our results indicate a significant increased risk of venous thromboembolism among patients with rheumatoid arthritis. Keywords Deep venous thrombosis . Pulmonary embolism . Rheumatoid arthritis . Venous thromboembolism
Introduction Venous thromboembolism (VTE), which includes deep venous thrombosis (DVT) and pulmonary embolism (PE), is a P. Ungprasert (*) : N. Srivali : E. L. Knight Department of Internal Medicine, Bassett Medical Center and Columbia University College of Physicians and Surgeons, Cooperstown, NY, USA e-mail: [email protected] I. Spanuchart Department of Internal Medicine, University of Hawaii, Honolulu, HI, USA C. Thongprayoon Division of Nephrology and Hypertension, Department of Medicine, Mayo Clinic, Rochester, MN, USA
common medical problem with an annual incidence of 1–2 new cases per 1,000 population [1, 2]. It confers a significant morbidity and mortality with a reported 30-day case fatality rate of 11–30 % [1–3]. Several conditions, such as immobilization, surgery, major trauma, and malignancy, are well recognized as risk factors for VTE. Chronic inflammatory disorders, though not generally viewed as traditional risk factors for VTE, might increase the risk of developing VTE as inflammatory cytokines have been demonstrated to modulate the coagulation pathway by upregulating procoagulants and downregulating anticoagulant and fibrinolytic systems. Furthermore, chronic inflammation has been shown to be associated with endothelial dysfunction [4]. While there are a considerable number of epidemiologic studies regarding the morbidity and mortality from cardiovascular diseases in patients with rheumatoid arthritis (RA), one the most common chronic inflammatory disorders, the data on VTE risk in this population are relatively limited. Several recent studies demonstrated an increased VTE risk among these patients though the results were fairly heterogeneous. Thus, to obtain a more accurate and precise estimated effect, we conducted a systematic review and meta-analysis of observational studies that compared the risk of VTE in patients with RA versus non-RA participants.
Method Search strategy Two investigators (P.U. and N.S.) independently searched published studies indexed in MEDLINE, EMBASE, and the Cochrane databases from inception to November 2013 using the terms “pulmonary embolism,” “deep venous thrombosis,” and “venous thromboembolism” combined with the term
298
Clin Rheumatol (2014) 33:297–304
“rheumatoid arthritis.” A manual search of references of selected retrieved articles was also performed. Abstract and unpublished studies were not included.
Inclusion criteria The inclusion criteria were as follows: (1) observational studies (case–control, cross-sectional, or cohort studies) published as original studies to evaluate the association between RA and risk of VTE, DVT, or PE; (2) odds ratios (OR), relative risk (RR), or hazard ratio (HR) with 95 % confidence intervals (CIs) or survival curves were provided; and (3) random nonRA participants were the reference group. Study eligibility was independently determined by each investigator noted above. Differing decisions were resolved by consensus. The quality of each study was independently evaluated by each investigator using Newcastle-Ottawa quality assessment scale [5]. Fig. 1 Outline of our search methodology and study selection
Statistical analysis Data analysis was performed using the Review Manager 5.2 software from the Cochrane Collaboration. We reported the pooled effect estimate of DVT, PE, and VTE (DVT and/or PE). We combined data from case–control, cross-sectional study, and cohort analyses to increase the precision of our estimates. Nonetheless, we also performed a separate analysis according to study design. As the outcome under study was relatively uncommon, we used the OR of case–control studies as the estimate of the RR to pool this data with the RR or HR from cross-sectional studies and cohort studies. Point estimates and standard error were extracted from individual studies and were combined by the generic inverse variance methods. We ran a random-effect model rather than a fixedeffects model because of the high likelihood of between study variance. Statistical heterogeneity was assessed using the Cochran’s Q test. This statistic was complemented with I2 statistics,
Potentially relevant articles identified from search of MEDLINE, EMBASE and the Cochrane database and screened for retrieval (n=335)
Detailed review of potentially relevant articles
323 articles were excluded since they were not observational studies or were not done in patients with RA 12 potentially relevant articles included for fulllength article review
2 articles were excluded since participants received anticoagulants before entering the studies 1 article was excluded since the reference group included RA participants
9 articles were included in the meta-analysis
Taiwan Case–control 2012 Hospitalized or non-hospitalized patients who were diagnosed with DVT, identified from national insurance database Duplex ultrasound scan or venography Sex and age-matched subject randomly selected from same database 2001–2009 18 to >69 58.7 5,193 20,772 Diagnostic code from the registry. The diagnosis needed to be made by rheumatologist and patient needed to take DMARD Hospitalization, pregnancy, fracture, surgery, cancer, hypertension, diabetes, coronary artery disease, heart failure, renal disease, hyperlipidemia, inflammatory bowel disease Selection: 4 stars Comparability: 1 star Exposure: 2 stars
NA Sex and age-matched subject randomly recruited from same geographical area 1999–2004 18–70 54.0 4,311 5,768 Self report using questionnaire
Hospitalization, pregnancy, fracture, surgery, cancer, diabetes, coronary artery disease, renal disease, COPD, osteoporosis, medications, liver disease, obesity, recent infection, medications Selection: 4 stars Comparability: 1 star Exposure: 1 star Hospitalization, pregnancy, fracture, surgery, cancer, diabetes, coronary artery disease, renal disease, COPD, osteoporosis, medications, liver disease, obesity Selection: 3 stars Comparability: 1 star Exposure: 1 star
The Netherlands Case–control 2012 All ambulatory patients with the diagnosis with DVT and/or PE who attended anticoagulation clinics
Ocak et al. [9]
Denmark Case–control 2012 Hospitalized or non-hospitalized patients who were diagnosed with DVT and/or PE identified from national registry database NA Sex and age-matched subject randomly selected from same database 1999–2009 18–99 52.9 14,721 147,210 Diagnostic code from the registry
Johannesdottir et al. [8]
DVT deep venous thrombosis, PE pulmonary embolism, NA not available, DMARD disease modifying anti-rheumatic drug, COPD chronic obstructive pulmonary disease
Quality assessment (Newcastle-Ottawa scale)
Confounder assessed
Period of inclusion Age range, years Woman, % Number of cases Number of control Diagnosis of RA
Case verification Controls
Country Study design Year Cases
Kang et al. [7]
Table 1 Main characteristics of case–control studies included in the meta-analysis
Clin Rheumatol (2014) 33:297–304 299
57.0 69.0 7,904 37,350 5.8 Hospitalization
Sex and age-matched subject randomly selected from same database
Duplex ultrasound scan or venography or CT angiography or autopsy
Until death, migration from the system or December 31, 2008
55.6 69.0 464 464 9.6
Hospitalization
Controls
Event verification
Follow up
Mean age, Y Woman, % Number of cases Number of control Average range of follow up, Y Confounder assessed
Sex, age, and residential area-matched subject randomly selected from Swedish population registry Patients needed to fill prescriptions for vitamin K antagonist or heparin Until death, migration from the system or December 31, 2010
Diagnostic code from the registry
Fulfilled at least 4 of the 7 ACR 1987 criteria
Diagnosis of RA
Sweden Prospective cohort 2012 Ambulatory patients who were diagnosed with RA between 1997 and 2009. Cases were identified by using Swedish national registry.
USA Retrospective cohort 2012 Residences of Olmstead county who were diagnosed with RA between 1980 and 2007. Cases were identified by using Mayo Clinic and its affiliations database.
Holmqvist et al. [12]
Country Study design Year Cases
Bacani et al. [11]
Table 2 Main characteristics of cohort studies included in the meta-analysis
Until death, migration from the system or June 30, 2008 Until death, experience the outcome, migration from the system or December 31, 2010 52.0 77.0 29,238 116,952 6.6
Until death, experience the outcome or May 31, 2010
Hospitalization, varicose vein, fracture, surgery, cancer, smoking, COPD, diabetes, coronary artery disease, heart failure, renal disease, BMI, inflammatory bowel disease
Atrial fibrillation, hypertension, diabetes, hyperlipidemia, stroke, heart failure, fracture, surgery, cancer
Patients needed to receive anticoagulation therapy
Diagnostic code, confirmed by peer review
Patients needed to receive anticoagulation therapy
Hospitalization, varicose vein, fracture, surgery, COPD, diabetes, heart failure, cancer, renal disease, obesity, stroke
52.2 75.0 22,143 88,572 2.0
Sex and age-matched subject randomly selected from same database
Sex and age-matched subject randomly selected from same database
58.3 69.4 9,589 95,776 5.5
Diagnostic code from the registry + at least one use of DMARDs
Fulfilled at least 4 of the 7 ACR 1987 criteria
Kim et al. [15] USA Retrospective cohort 2013 Hospitalized or non-hospitalized patients who were diagnosed RA between 2001 and 2008. Cases were identified by using the Medicare and a private health insurance company database.
Chung et al. [14] Taiwan Prospective cohort 2013 Hospitalized or non-hospitalized patients who were diagnosed RA between 1998 and 2008. Cases were identified by using the National health insurance database (which covers 99 % of entire population).
UK Retrospective cohort 2013 Ambulatory patients who were diagnosed RA between 1986 and 2010. Cases were identified by using The Health Improve Network (UK computerized medical record database comprised of 7.3 million patients). Diagnostic code from the registry + at least one use of DMARDs Sex and age-matched subject randomly selected from same database
Choi et al. [13]
300 Clin Rheumatol (2014) 33:297–304
DVT deep venous thrombosis, PE pulmonary embolism, DMARD disease modifying anti-rheumatic drug, COPD chronic obstructive pulmonary disease, CT computerized tomography, ACR American College of Rheumatology
Selection: 4 stars Comparability: 1 star Outcome: 2 stars Selection: 4 stars Comparability: 1 star Outcome: 3 stars Selection: 4 stars Comparability: 1 star Outcome: 3 stars Selection: 4 stars Comparability: 1 star Outcome: 3 stars Selection: 4 stars Comparability: 2 stars Outcome: 3 stars
Chung et al. [14] Choi et al. [13]
Quality assessment (Newcastle-Ottawa scale)
Table 2 (continued)
Bacani et al. [11]
Holmqvist et al. [12]
Kim et al. [15]
Clin Rheumatol (2014) 33:297–304
301
which quantify the proportion of total variation across studies that is due to heterogeneity rather than chance. A value of I2 of 0 to 25 % represents insignificant heterogeneity, 25 to 50 % low heterogeneity, 50 to 75 % moderate heterogeneity, and 75 to 100 % high heterogeneity [6].
Result Our search strategy yielded 335 potentially relevant articles. Three hundred twenty-three articles were excluded as they were not observational studies or were not done in patients with RA. Twelve articles underwent full-length article review; two of them were excluded since participants received anticoagulants before entering the studies, and one of them was excluded since the reference group included RA participant. Nine studies (five cohort studies, three case–control studies, and a cross-sectional study) with 606,427 subjects met our inclusion criteria and were included in the analysis [7–15]. Eight studies reported the RR of VTE in patients with RA compared with non-RA participants [7–13, 15], six studies reported the RR of DVT [7, 10, 12–15], while other five studies reported the RR of PE [10, 12–15]. Figure 1 outlines our search methodology and selection process. The detailed characteristics and quality assessment of the included studies are described in Tables 1 and 2. The pooled risk ratios of DVT, PE, and VTE of subjects with RA versus control subjects were 2.08 (95 % CI 1.75– 2.47), 2.17 (95 % CI 2.05–2.31), and 1.96 (95 % CI 1.81– 2.11), respectively. The statistical heterogeneity was high in DVT studies (I2 of 92 %), moderate in VTE studies (I2 of 66 %), and negligible in PE studies (I2 of 22 %).
Subgroup analysis As shown in Figs. 2, 3, and 4, the risk ratios DVT, PE, and VTE in subjects with RA were significantly higher in every study design subgroup. For DVT, the risk ratios were 2.08 (95 % CI 1.49–2.92), 2.23 (95 % CI 1.72–2.87), and 1.90 (95 % CI 1.88–1.91) for cohort, case–control, and crosssectional studies, respectively. The comparison between studies revealed no significant subgroup difference (I2 of 0 %). For PE, the risk ratios were 2.02 (95 % CI 1.84–2.23) for cohort studies and 2.25 (95 % CI 2.23–2.27) for cross-sectional study. The subgroup heterogeneity was high with I2 of 78.5 %. Lastly, the risk ratio of VTE were 1.90 (95 % CI 1.55–2.33), 1.95 (95 % CI 1.56–2.45), and 1.99 (95 % CI 1.98–2.01) for cohort, case–control, and cross-sectional studies without a significant subgroup difference (I2 of 0 %).
302
Clin Rheumatol (2014) 33:297–304
Fig. 2 Forest plot of DVT studies
Publication bias Funnel plots to evaluate publication bias are summarized in Fig. 5. The graph of the eight VTE studies, although fairly symmetric, provides a suggestion that publication bias in favor of positive studies may be present.
Discussion Our meta-analysis demonstrated a significant association between RA and VTE with an overall 1.94-fold (95 % CI 1.79– 2.10) increased risk compared with non-RA participants. The risk ratios were fairly consistent across the studies, ranging from 1.40 to 3.60, regardless of study design. Our study also
Fig. 3 Forest plot of PE studies
demonstrated an increased risk for DVT and PE for patients with RA with risk ratios of 2.08 and 2.17, respectively. Heterogeneity between studies was present in this metaanalysis particularly for DVT studies. We suspect that the difference in case selection was the main source of heterogeneity, as approximately half of the studies were done in both hospitalized and non-hospitalized subjects while another half was done exclusively in ambulatory setting. We did perform a sensitivity analysis for each setting and found no significant difference for the pooled RR and 95 % CI (data is not shown). Why patients with RA have a higher risk of VTE compared with non-RA subjects is unclear. The occurrence of VTE is linked to three determinative factors, known as Virchow triad, including hypercoagulability, endothelial injury/dysfunction, and venous stasis. RA appears to inter-
Clin Rheumatol (2014) 33:297–304
303
Fig. 4 Forest plot of VTE studies
fere with all of these provocative factors as chronic inflammation related to this autoimmune disease has been demonstrated to promote procoagulation, impair the anticoagulation pathway, and inhibit the fibrinolytic process [4, 16, 17]. Endothelial dysfunction in patient with RA has also been extensively documented [4, 18]. Moreover, immobilization from joint pain and compromised regional blood flow from the compression of femoral vein from the swollen, inflamed iliopsoas bursa can further pose patients at an increased risk of flow stasis and DVT [19]. Even though the nine studies included in this metaanalysis were of high quality, there are some limitations and, thus, the result should be interpreted with caution. First, we cannot exclude the possibility of publication bias as the funnel plot is relatively asymmetric. Second, most of
the included studies were conducted using medical registrybased database, thus maybe coding inaccuracy. Third, this is a meta-analysis of observational studies, and this type of study can demonstrate an association, not establish cause and effect, so we cannot be certain that RA itself versus other potential confounders cause the increase risk of VTE. Furthermore, the higher detection rate of VTE in patients with RA might be partly due to the fact that they are sicker and, thus, exposed more to medical community. In conclusion, the results of our meta-analysis demonstrate a statistically significant increased VTE risk among patients with RA. As VTE confers a high morbidity and mortality, our study suggests that physicians should carefully monitor patients with RA for VTE, although the role of VTE prophylaxis remains unclear and merits further investigations.
Disclosures None. Funding None.
References
Fig. 5 Funnel plot of VTE studies
1. Naess IA, Christiansen SC, Romundstad P et al (2007) Incidence and mortality of venous thrombosis: a population-based study. J Thromb Haemost 5:692–699 2. Cushman M, Tsai AW, White RH et al (2004) Deep venous thrombosis and pulmonary embolism in two cohorts: the longitudinal investigation of thromboembolism etiology. Am J Med 117:19–25 3. Heit JA, Silverstein MD, Mohr DN et al (1999) Predictor of survival after deep venous thrombosis and pulmonary embolism: a population-based, cohort study. Arch Intern Med 159:445–453
304 4. Xu J, Lupu F, Esmon CT (2010) Inflammation, innate immunity and blood coagulation. Hamostaseologie 30(5–6):8–9 5. Stang A (2010) Critical evaluation of the Newcastle-Ottawa scale for the assessment of the quality of nonrandomized studies in metaanalyses. Eur J Epidemiol 25:603–605 6. Higgins JP, Thompson SG, Deeks JJ, Altman DG (2003) Measuring inconsistency in meta-analyses. BMJ 327:557–560 7. Kang JH, Keller JJ, Lin YK, Lin HC (2012) A population-based case–control study on association between rheumatoid arthritis and deep vein thrombosis. J Vasc Surg 56:1642–1648 8. Johannesdottir SA, Schmidt M, Horvath-Puho E, Sorensen HT (2012) Autoimmune skin and connective tissue diseases and risk of venous thromboembolism: a population-based case–control study. J Thromb Haemost 10:815–821 9. Ocak G, Vossen CY, Verduijn M et al (2013) Risk of venous thrombosis in patients with major illnesses: result from the MEGA study. J Thromb Haemost 11:116–123 10. Matta F, Singala R, Yaekoub AY, Najjar R, Stein PD (2009) Risk of venous thromboembolism with rheumatoid arthritis. Thromb Haemost 101:134–138 11. Bacani AK, Gabriel SE, Crowson CS, Heit JA, Matteson EL (2012) Noncardiac vascular disease in rheumatoid arthritis. Arthritis Rheum 64: 53–61 12. Holmqvist ME, Neovius M, Eriksson J et al (2012) Risk of venous thromboembolism in patients with rheumatoid arthritis and
Clin Rheumatol (2014) 33:297–304
13.
14.
15.
16.
17.
18.
19.
association with disease duration and hospitalization. JAMA 308: 1350–1356 Choi HK, Rho YH, Zhu Y, Cea-Soriano LC, Avina-Zubieta JA, Zhang Y (2013) The risk of pulmonary embolism and deep vein thrombosis in rheumatoid arthritis: a UK population-based outpatient cohort study. Ann Rheum Dis 72:1182–1187 Chung WS, Peng CL, Lin CL et al (2013) Rheumatoid arthritis increases the risk of deep vein thrombosis and pulmonary thromboembolism: a nationwide cohort study. Ann Rheum Dis August 13 Kim SC, Schneeweiss S, Liu J, Solomom DH (2013) Risk of venous thromboembolism in patients with rheumatoid arthritis. Arthritis Care Res 65:1600–1607 Montecucco F, Mach F (2009) Common inflammatory mediators orchestrate pathophysiological processes in rheumatoid arthritis and atherosclerosis. Rheumatology (Oxford) 48:11–22 Conway EM, Rosenberg RD (1988) Tumor necrosis factor suppresses transcription of the thrombomodulin gene in endothelial cells. Mol Cell Biol 8:5588–5592 Hansen S, Lassig G, Pistrosch E, Passauer J (2003) Endothelial dysfunction in young patients with long-term rheumatoid arthritis and low disease activity. Atherosclerosis 170:177–180 McLaughlin GE (2002) Sudden death in rheumatoid arthritis: pulmonary embolism—a fatal complication of iliopsoas bursitis. J Clin Rheumatol 8:208–211
