ORIGINAL ARTICLE
Circulation Journal Official Journal of the Japanese Circulation Society http://www. j-circ.or.jp
Peripheral Vascular Disease
Hormone Therapy and Risk of Venous Thromboembolism Among Postmenopausal Women in Taiwan – A 10-Year Nationwide Population-Based Study – Cheng-Han Lee, MD, PhD; Ching-Lan Cheng, PhD; Yea-Huei Kao Yang, BSc; Li-Jen Lin, MD
Background: The incidence of venous thromboembolism (VTE) in Asians is lower than in Caucasians, but the risk of VTE associated with hormone therapy (HT) in Taiwanese postmenopausal women has not been determined. Methods and Results: From Taiwan’s National Health Insurance Research Database, we established matched cohorts (HT users and nonusers) of postmenopausal women aged ≥50 years between 1 January 1998 and 31 December 2008. We calculated the 2-year incidence of VTE in HT users and nonusers. HT users and nonusers were matched 1:1 based on propensity-score matching. Cox regression hazard model was used to identify risk factors of VTE. We initially identified 499,594 HT users and 424,963 nonusers. There were higher percentages of cancer and cardiovascular events among the HT nonusers. After matching, the VTE incidence was 4.4 vs. 2.6 per 10,000 patient-years (adjusted hazard ratio 1.796, 95% confidence interval 1.272–2.537) in HT users and nonusers, respectively. The Cox regression hazard model showed that HT use, older age, malignancy, heart failure, and recent major surgery were independent risk factors for VTE. Conclusions: Although the incidence of VTE was very low among this cohort of Taiwanese postmenopausal women, oral HT was still associated with an increased risk of VTE. Therefore, physicians should be aware of other potential VTE risk factors when prescribing oral HT to postmenopausal women. (Circ J 2015; 79: 1107 – 1114) Key Words: Deep vein thrombosis; Hormone therapy; Postmenopausal women; Pulmonary embolism; Venous thromboembolism
H
ormone therapy (HT) can improve quality of life for women with hypo-estrogenic symptoms1 and is also effective for preventing osteoporotic fractures among current users.2,3 In contrast, harmful effects of HT include breast cancer and venous thromboembolism (VTE).4 Furthermore, randomized controlled trials showed that HT might increase the risk of coronary artery disease and stroke.2,5 Many women are still prescribed estrogen therapy to ameliorate postmenopausal symptoms despite recent data showing that overall health risks may exceed the benefits of long-term HT.2 Despite evidence showing that oral estrogen activates blood coagulation in postmenopausal women,6 HT had, until 1996, long been believed to have little effect on the risk of VTE.7 Recent observational studies, however, have shown consistent associations between current use of HT and an increased risk of VTE in postmenopausal women. These findings have been confirmed in randomized controlled trials.2,5 Since the publication of the Women’s Health Initiative (WHI) results,2 the medical practice of HT has been dramatically altered.8 Although a
striking decrease in HT use has been noted, many women remain eligible for this treatment to correct postmenopausal climacteric symptoms and to prevent osteoporosis. Cardiovascular disease, including VTE, is an important determinant of the benefit-to-risk profile of HT.9 Both observational studies10–13 and clinical trials2,14 have shown a significant increase in VTE risk among postmenopausal women using HT. However, most of these studies have been conducted in Caucasian women, so the results are not essentially generalizable to other races. According to Japan’s guidelines for management and prevention of VTE, estrogen therapy is a weak risk factor of VTE, and there is no recommendation for HT in postmenopausal women.15 Nakamura et al reported that estrogen or/and HT did not contribute to a higher risk of VTE recurrence, but their sample size was too small to make a definitive conclusion.16 In our previous studies, the overall crude incidence of VTE was 14.4 events per 100,000 person-years in men and 17.4 events per 100,000 person-years in women, which was much lower (71–117 cases per 100,000 persons) than in Western
Received November 11, 2014; revised manuscript received January 6, 2015; accepted January 14, 2015; released online March 4, 2015 Time for primary review: 16 days Department of Internal Medicine, National Cheng Kung University Hospital, Tainan (C.-.H.L., L.-J.L.); Institute of Clinical Pharmacy and Pharmaceutical Sciences, College of Medicine, National Cheng Kung University, Tainan (C.-.H.L., C.-L.C., Y.-H.K.Y., L.-J.L.), Taiwan Mailing address: Li-Jen Lin, MD, Institute of Clinical Pharmacy and Pharmaceutical Sciences, College of Medicine, National Cheng Kung University, 1 University Road, Tainan, Taiwan. E-mail: [email protected] ISSN-1346-9843 doi: 10.1253/circj.CJ-14-1227 All rights are reserved to the Japanese Circulation Society. For permissions, please e-mail: [email protected] Circulation Journal Vol.79, May 2015
1108
LEE CH et al.
Figure 1. Flow chart of participants’ disposition. HT, hormone therapy; VTE, venous thromboembolism.
countries.17,18 For women, the overall VTE recurrence rate was 10.1% vs. 11.4% (P=0.41) among HT users and nonusers, respectively. Meanwhile, there was no significant difference in the prevalence of HT use (including estrogen and/or progesterone) between VTE recurrent cases and VTE non-recurrent controls. Therefore, the VTE risk of HT in postmenopausal women in Taiwan needs to be determined.
Methods Study Design We retrospectively performed a nationwide population-based matched cohort study to evaluate HT and the risk of VTE during a 2-year follow-up of Taiwanese postmenopausal women. The study was approved by the Institutional Review Board of National Cheng Kung University Hospital, Tainan, Taiwan. Source of Data The National Health Insurance (NHI) databases used in this study included all inpatient and outpatient medical claims from January 1, 1997 to December 31, 2008. In these databases, medical information of disease diagnosis, prescription drugs, procedures, and surgery incurred during a hospitalization or at an outpatient visit are documented. For administration processing by the NHI in Taiwan, all healthcare service providers are requested to submit all diagnosis information using the International Classification of Disease-Clinical Modification, 9th revision (ICD-9-CM), together with the service claims.
Study Population Our 2 study cohorts consisted of randomly selected women who were between 50 and 79 years old during the study period from year 2000 through 2008 and (1) had a prescription for HT or medical service for a postmenopausal condition (ICD9: 627.xx) or (2) had neither prescription for HT nor medical service for a postmenopausal condition. We excluded all subjects with a diagnosis of VTE prior to the enrollment, who were ever prescribed HT in the previous 3 years, or who had a history of undergoing a hysterectomy. In addition, women without any visit and medical records after age 50 or who had a follow-up duration less than 90 days were also excluded. We used a propensity-score method to match the selected cohort populations to overcome initial selection bias for HT users and nonusers (Figure 1). The selected variables in the propensityscore method included age and potential VTE confounding factors. Graphs of before and after propensity-score matching are shown in Figure S1 and S2, respectively. Exposure Definition HT users were defined as women having any HT prescriptions during the study period, and the first HT prescription date was the index date. A list of all medications containing estrogens and/or progestogens recommended for HT and available in Taiwan during the study period was extracted from the database. In Taiwan, there were no pharmaceutical products for transdermal HT, tibolone, or estradiol implant during the study years. For HT nonusers, the index date was the date of the first medical visit during 2000–2008.
Circulation Journal Vol.79, May 2015
HT and Risk of VTE
1109
Figure 2. Kaplan-Meier estimates of VTE-free survival distribution before propensity-score matching. The incidence rate of VTE was 6.7 per 10,000 patient-years for HT users and 6.2 per 10,000 patient-years for nonusers [adjusted hazard ratio: 0.997 (0.88– 1.13)]. HT, hormone therapy; VTE, venous thromboembolism.
Outcome Definition and Follow-up The primary endpoints were a principal or secondary admission diagnosis of VTE, which included DVT or PE. All incident cases of VTE occurring during the study period were identified from the claims database by the ICD9-CM codes of 451.1x, 451.2, 451.83, 453.1, 453.2, 453.4, 453.8, 453.9, and 415.1x. To avoid misdiagnoses, we only selected inpatients who met the following criteria: (1) a discharge diagnosis of DVT or PE; (2) received a course of subcutaneous or intravenous anticoagulation therapy with unfractionated heparin or surgical thrombectomy during hospitalization and continued oral warfarin therapy after discharge; and (3) hospitalization of at least 3 days, unless the patient died. We also identified outpatients with VTE who met the following criteria: (1) principle diagnosis of DVT or thrombophlebitis; and (2) received a course of subcutaneous anticoagulation therapy with lowmolecular-weight heparin and continued oral warfarin therapy. We followed the cohort population until the date of the first study endpoint (VTE), death, or 2 years, whichever occurred first. Potential Confounding Variables We considered other medications (systemic steroids, NSAIDs, statins, antiplatelet agents, anticoagulant drugs, selective serotonin reuptake inhibitors) and variables believed to affect the risk of VTE (age, sex, hypertension, heart failure, cancer, varicose veins, diabetes mellitus, extremity fractures, myocardial infarction, ischemic and hemorrhagic cerebrovascular events, chronic lung diseases, liver diseases, renal diseases) as
potential confounders. The comorbidities predisposing patients to VTE were recorded using ICD-9-CM codes (Table S2). The medications before or within 90 days of the index date were recorded, and other variables were measured for 1 year prior to the index date. Statistical Analysis Demographic data are expressed as mean (± SD) or percentage. In general, differences in proportions were tested with the chi-square test or Fisher’s exact test, and differences in location parameters of continuous variables were tested with Student’s t-test. We calculated the incidence rates of VTE per 10,000 person-years for the HT users and nonusers. Intent-to-treat analysis was used to evaluate the association between HT and risk of VTE; once a patient met the definition for exposure to HT, she was considered exposed from that point forward, even if she discontinued therapy. Time-toevent curves for each cohort were calculated by the KaplanMeier method and compared by means of log-rank test. The estimates of relative risk, with 95% confidence intervals (CIs), were derived from Cox proportional-hazard models, adjusted for potential confounders including medications and comorbidities. We conducted several sensitivity analyses to test the robustness of our findings. First, we performed an “as-treated” analysis to test the HT-user effect. We followed HT nonusers for 2 years and followed HT users who only used HT for less than 90 days. Second, we conducted a nested case-control study within unmatched cohorts. Cases were defined as admis-
Circulation Journal Vol.79, May 2015
1110
LEE CH et al.
Table 1. Clinical Characteristics of the Taiwanese Postmenopausal Women in Matched Cohorts From 2000 to 2008 HT users (n=179,351) Mean age, years ± SD
Non-HT users (n=179,351)
P value
60.7±8.1
59.5±7.6
Circulation Journal Official Journal of the Japanese Circulation Society http://www. j-circ.or.jp
Peripheral Vascular Disease
Hormone Therapy and Risk of Venous Thromboembolism Among Postmenopausal Women in Taiwan – A 10-Year Nationwide Population-Based Study – Cheng-Han Lee, MD, PhD; Ching-Lan Cheng, PhD; Yea-Huei Kao Yang, BSc; Li-Jen Lin, MD
Background: The incidence of venous thromboembolism (VTE) in Asians is lower than in Caucasians, but the risk of VTE associated with hormone therapy (HT) in Taiwanese postmenopausal women has not been determined. Methods and Results: From Taiwan’s National Health Insurance Research Database, we established matched cohorts (HT users and nonusers) of postmenopausal women aged ≥50 years between 1 January 1998 and 31 December 2008. We calculated the 2-year incidence of VTE in HT users and nonusers. HT users and nonusers were matched 1:1 based on propensity-score matching. Cox regression hazard model was used to identify risk factors of VTE. We initially identified 499,594 HT users and 424,963 nonusers. There were higher percentages of cancer and cardiovascular events among the HT nonusers. After matching, the VTE incidence was 4.4 vs. 2.6 per 10,000 patient-years (adjusted hazard ratio 1.796, 95% confidence interval 1.272–2.537) in HT users and nonusers, respectively. The Cox regression hazard model showed that HT use, older age, malignancy, heart failure, and recent major surgery were independent risk factors for VTE. Conclusions: Although the incidence of VTE was very low among this cohort of Taiwanese postmenopausal women, oral HT was still associated with an increased risk of VTE. Therefore, physicians should be aware of other potential VTE risk factors when prescribing oral HT to postmenopausal women. (Circ J 2015; 79: 1107 – 1114) Key Words: Deep vein thrombosis; Hormone therapy; Postmenopausal women; Pulmonary embolism; Venous thromboembolism
H
ormone therapy (HT) can improve quality of life for women with hypo-estrogenic symptoms1 and is also effective for preventing osteoporotic fractures among current users.2,3 In contrast, harmful effects of HT include breast cancer and venous thromboembolism (VTE).4 Furthermore, randomized controlled trials showed that HT might increase the risk of coronary artery disease and stroke.2,5 Many women are still prescribed estrogen therapy to ameliorate postmenopausal symptoms despite recent data showing that overall health risks may exceed the benefits of long-term HT.2 Despite evidence showing that oral estrogen activates blood coagulation in postmenopausal women,6 HT had, until 1996, long been believed to have little effect on the risk of VTE.7 Recent observational studies, however, have shown consistent associations between current use of HT and an increased risk of VTE in postmenopausal women. These findings have been confirmed in randomized controlled trials.2,5 Since the publication of the Women’s Health Initiative (WHI) results,2 the medical practice of HT has been dramatically altered.8 Although a
striking decrease in HT use has been noted, many women remain eligible for this treatment to correct postmenopausal climacteric symptoms and to prevent osteoporosis. Cardiovascular disease, including VTE, is an important determinant of the benefit-to-risk profile of HT.9 Both observational studies10–13 and clinical trials2,14 have shown a significant increase in VTE risk among postmenopausal women using HT. However, most of these studies have been conducted in Caucasian women, so the results are not essentially generalizable to other races. According to Japan’s guidelines for management and prevention of VTE, estrogen therapy is a weak risk factor of VTE, and there is no recommendation for HT in postmenopausal women.15 Nakamura et al reported that estrogen or/and HT did not contribute to a higher risk of VTE recurrence, but their sample size was too small to make a definitive conclusion.16 In our previous studies, the overall crude incidence of VTE was 14.4 events per 100,000 person-years in men and 17.4 events per 100,000 person-years in women, which was much lower (71–117 cases per 100,000 persons) than in Western
Received November 11, 2014; revised manuscript received January 6, 2015; accepted January 14, 2015; released online March 4, 2015 Time for primary review: 16 days Department of Internal Medicine, National Cheng Kung University Hospital, Tainan (C.-.H.L., L.-J.L.); Institute of Clinical Pharmacy and Pharmaceutical Sciences, College of Medicine, National Cheng Kung University, Tainan (C.-.H.L., C.-L.C., Y.-H.K.Y., L.-J.L.), Taiwan Mailing address: Li-Jen Lin, MD, Institute of Clinical Pharmacy and Pharmaceutical Sciences, College of Medicine, National Cheng Kung University, 1 University Road, Tainan, Taiwan. E-mail: [email protected] ISSN-1346-9843 doi: 10.1253/circj.CJ-14-1227 All rights are reserved to the Japanese Circulation Society. For permissions, please e-mail: [email protected] Circulation Journal Vol.79, May 2015
1108
LEE CH et al.
Figure 1. Flow chart of participants’ disposition. HT, hormone therapy; VTE, venous thromboembolism.
countries.17,18 For women, the overall VTE recurrence rate was 10.1% vs. 11.4% (P=0.41) among HT users and nonusers, respectively. Meanwhile, there was no significant difference in the prevalence of HT use (including estrogen and/or progesterone) between VTE recurrent cases and VTE non-recurrent controls. Therefore, the VTE risk of HT in postmenopausal women in Taiwan needs to be determined.
Methods Study Design We retrospectively performed a nationwide population-based matched cohort study to evaluate HT and the risk of VTE during a 2-year follow-up of Taiwanese postmenopausal women. The study was approved by the Institutional Review Board of National Cheng Kung University Hospital, Tainan, Taiwan. Source of Data The National Health Insurance (NHI) databases used in this study included all inpatient and outpatient medical claims from January 1, 1997 to December 31, 2008. In these databases, medical information of disease diagnosis, prescription drugs, procedures, and surgery incurred during a hospitalization or at an outpatient visit are documented. For administration processing by the NHI in Taiwan, all healthcare service providers are requested to submit all diagnosis information using the International Classification of Disease-Clinical Modification, 9th revision (ICD-9-CM), together with the service claims.
Study Population Our 2 study cohorts consisted of randomly selected women who were between 50 and 79 years old during the study period from year 2000 through 2008 and (1) had a prescription for HT or medical service for a postmenopausal condition (ICD9: 627.xx) or (2) had neither prescription for HT nor medical service for a postmenopausal condition. We excluded all subjects with a diagnosis of VTE prior to the enrollment, who were ever prescribed HT in the previous 3 years, or who had a history of undergoing a hysterectomy. In addition, women without any visit and medical records after age 50 or who had a follow-up duration less than 90 days were also excluded. We used a propensity-score method to match the selected cohort populations to overcome initial selection bias for HT users and nonusers (Figure 1). The selected variables in the propensityscore method included age and potential VTE confounding factors. Graphs of before and after propensity-score matching are shown in Figure S1 and S2, respectively. Exposure Definition HT users were defined as women having any HT prescriptions during the study period, and the first HT prescription date was the index date. A list of all medications containing estrogens and/or progestogens recommended for HT and available in Taiwan during the study period was extracted from the database. In Taiwan, there were no pharmaceutical products for transdermal HT, tibolone, or estradiol implant during the study years. For HT nonusers, the index date was the date of the first medical visit during 2000–2008.
Circulation Journal Vol.79, May 2015
HT and Risk of VTE
1109
Figure 2. Kaplan-Meier estimates of VTE-free survival distribution before propensity-score matching. The incidence rate of VTE was 6.7 per 10,000 patient-years for HT users and 6.2 per 10,000 patient-years for nonusers [adjusted hazard ratio: 0.997 (0.88– 1.13)]. HT, hormone therapy; VTE, venous thromboembolism.
Outcome Definition and Follow-up The primary endpoints were a principal or secondary admission diagnosis of VTE, which included DVT or PE. All incident cases of VTE occurring during the study period were identified from the claims database by the ICD9-CM codes of 451.1x, 451.2, 451.83, 453.1, 453.2, 453.4, 453.8, 453.9, and 415.1x. To avoid misdiagnoses, we only selected inpatients who met the following criteria: (1) a discharge diagnosis of DVT or PE; (2) received a course of subcutaneous or intravenous anticoagulation therapy with unfractionated heparin or surgical thrombectomy during hospitalization and continued oral warfarin therapy after discharge; and (3) hospitalization of at least 3 days, unless the patient died. We also identified outpatients with VTE who met the following criteria: (1) principle diagnosis of DVT or thrombophlebitis; and (2) received a course of subcutaneous anticoagulation therapy with lowmolecular-weight heparin and continued oral warfarin therapy. We followed the cohort population until the date of the first study endpoint (VTE), death, or 2 years, whichever occurred first. Potential Confounding Variables We considered other medications (systemic steroids, NSAIDs, statins, antiplatelet agents, anticoagulant drugs, selective serotonin reuptake inhibitors) and variables believed to affect the risk of VTE (age, sex, hypertension, heart failure, cancer, varicose veins, diabetes mellitus, extremity fractures, myocardial infarction, ischemic and hemorrhagic cerebrovascular events, chronic lung diseases, liver diseases, renal diseases) as
potential confounders. The comorbidities predisposing patients to VTE were recorded using ICD-9-CM codes (Table S2). The medications before or within 90 days of the index date were recorded, and other variables were measured for 1 year prior to the index date. Statistical Analysis Demographic data are expressed as mean (± SD) or percentage. In general, differences in proportions were tested with the chi-square test or Fisher’s exact test, and differences in location parameters of continuous variables were tested with Student’s t-test. We calculated the incidence rates of VTE per 10,000 person-years for the HT users and nonusers. Intent-to-treat analysis was used to evaluate the association between HT and risk of VTE; once a patient met the definition for exposure to HT, she was considered exposed from that point forward, even if she discontinued therapy. Time-toevent curves for each cohort were calculated by the KaplanMeier method and compared by means of log-rank test. The estimates of relative risk, with 95% confidence intervals (CIs), were derived from Cox proportional-hazard models, adjusted for potential confounders including medications and comorbidities. We conducted several sensitivity analyses to test the robustness of our findings. First, we performed an “as-treated” analysis to test the HT-user effect. We followed HT nonusers for 2 years and followed HT users who only used HT for less than 90 days. Second, we conducted a nested case-control study within unmatched cohorts. Cases were defined as admis-
Circulation Journal Vol.79, May 2015
1110
LEE CH et al.
Table 1. Clinical Characteristics of the Taiwanese Postmenopausal Women in Matched Cohorts From 2000 to 2008 HT users (n=179,351) Mean age, years ± SD
Non-HT users (n=179,351)
P value
60.7±8.1
59.5±7.6
