Hormone Replacement Therapy and Adverse Outcomes in Women With Atrial Fibrillation
An Analysis From the Atrial Fibrillation Follow-Up Investigation of Rhythm Management Trial
Background and Purpose—Hormone replacement therapy (HRT) use has been related to thromboembolism, but whether HRT increases adverse outcomes in females with atrial fibrillation is uncertain.
Methods—We used the Atrial Fibrillation Follow-Up Investigation of Rhythm Management (AFFIRM) trial data set that included 1594 women (39.3% of the population, mean age 71±8), of whom 376 (23.6%) were taking HRT at baseline. The primary end point, a composite of all-cause death, stroke, systemic/pulmonary embolism, and myocardial infarction, and secondary outcomes (ie, each individual end point) and major bleeding, were considered.
Results—HRT was not independently associated with the primary end point (hazard ratio=0.894; 95% confidence interval, 0.658–1.214; P=0.473) or any secondary outcome. Age (P<0.001), diabetes mellitus (P<0.001), previous stroke (P=0.011), and heart failure (P<0.001) predicted the primary end point. Lack of association between HRT and the primary end point was confirmed in a propensity score–matched control group (hazard ratio=0.966; 95% confidence interval, 0.663–1.409; P=0.858).
Conclusions—HRT does not independently predict mortality, thromboembolism, or bleeding in a large cohort of women with atrial fibrillation.
Compared with men, women with atrial fibrillation (AF) over age 65 are at greater risk of stroke and thromboembolism.1 We hypothesized that women with AF taking hormone replacement therapy (HRT) were at greater risk of all-cause death, ischemic stroke, systemic/pulmonary embolism, myocardial infarction (MI), and major bleeding and explored this hypothesis using the large Atrial Fibrillation Follow-Up Investigation of Rhythm Management (AFFIRM) trial data set.2
For this analysis, the primary outcome was a combined end point of all-cause death, stroke, systemic/pulmonary embolism, and MI. Secondary outcomes were each individual end point and major bleeding. The association of HRT and quality of oral anticoagulation or thromboembolic risk profile was investigated. We considered a composite end point of all vascular events based on a net clinical benefit outcome definition,3 which included stroke, MI, cardiovascular death, pulmonary/systemic embolism, or major bleeding. For detailed methods, see online-only Data Supplement.
Adjusted Cox proportional hazards models were used to assess the impact of HRT on outcomes. To reduce bias introduced by lack of randomization, we assembled a more homogeneous subcohort using a matching algorithm based on the estimated propensity score. Survival curves were calculated using the mean of covariates method, in which average values of covariates are entered into a proportional hazards regression equation, adjusted for prognostic variables similar to the total cohort.
AFFIRM recruited 4060 patients, of whom 1594 (39.3%) were women (mean age, 71±8 years) and 376 (23.6%) were prescribed HRT at baseline (Table I in the online-only Data Supplement). Compared with non-HRT-treated females, HRT users were younger and had a lower thromboembolic risk as reflected by CHADS2 (cardiac failure, hypertension, age >75, diabetes, stroke [doubled]) and CHA2DS2VASc (cardiac failure or dysfunction, hypertension, age ≥75 [doubled], diabetes, stroke [doubled]–vascular disease, age 65–74 and sex category [female]) scores (Figures I and II in the online-only Data Supplement). Women prescribed HRT were more likely to be active professionally and have a higher education level (Table II in the online-only Data Supplement, full data not shown). There was no significant difference in warfarin anticoagulation between HRT and non-HRT users, but HRT users had a higher percent time in the therapeutic range (0.61±0.24 versus 0.57±0.27, P=0.02).
Overall, 326 primary end point events occurred over a 3.5-year median follow-up corresponding to an annualized incidence of 5.8 events/100 patient-years. During follow-up, 116 major bleeding events occurred (2.1 events/100 patient-years).
The primary composite end point was more common among non-HRT-treated women versus HRT-treated women (6.3 events/100 patient-years versus 4.2 events/100 patient-years; univariate analysis; P=0.002). All-cause death and the composite net clinical benefit occurred more often in non-HRT women (P=0.007 and P=0.001, respectively); no statistically significant differences were observed in the other individual outcomes (Tables 1 and 2). In a univariate analysis of the propensity-matched cohort, event rate comparisons did not reach the cut-off point of statistical significance (Tables 1 and 2).
In a multivariable Cox regression analysis adjusted for age, prior stroke, hypertension, diabetes mellitus, heart failure, prior MI, peripheral arterial disease, warfarin treatment on recruitment, and treatment arm, HRT use was not independently associated with the primary end point (hazard ratio, 0.894; 95% confidence interval, 0.658–1.214; P=0.473; Tables 1 and 2 and Figure [A]). A Cox regression analysis in the total cohort found that age (P=0.00000002), diabetes mellitus (P=0.001), previous stroke (P=0.0003), heart failure (P=0.00000001), and time in therapeutic range (P=0.000001) were strongly correlated with the primary end point (Table 3).
HRT was not associated with any individual end point or net clinical benefit outcome (Tables 1 and 2). In the propensity score–matched cohort, HRT was also not significantly associated with the primary outcome (hazard ratio, 0.966; 95% confidence interval, 0.663–1.409; P=0.858; Table III in the online-only Data Supplement; Figure [B]).
HRT use among female patients with AF enrolled in AFFIRM use was not independently associated with adverse cardiovascular outcomes, major bleeding, or mortality. Indeed, HRT use did not increase risk of the composite primary end point, nor end points of all-cause death, ischemic stroke, major bleeding, or the composite of MI, pulmonary embolism, and systemic embolism—even in a propensity score–matched cohort. This analysis, investigating the impact of HRT on stroke risk in women with AF, is the largest of its kind.
In these subjects, independent predictors of the primary end point were age, diabetes mellitus, prior stroke, and heart failure. Age, diabetes mellitus, and prior stroke are well-validated risk factors for stroke.4 Data for heart failure are less consistent, but moderate-to-severe systolic impairment5 and recent decompensated heart failure increase stroke risk irrespective of ejection fraction.6
Notwithstanding the benefits of HRT on osteoporosis and menopausal symptoms, the influence of HRT on cardiovascular disease and stroke has been subject to debate, with some studies reporting that HRT increased stroke risk7 while others have not.8 Prior studies have been conducted in general populations not confined to AF per se,7 and the Heart and Estrogen/Progestin Replacement Study (HERS) trial was a much smaller trial and not powered for stroke as an end point although an excess of venous thromboembolism was seen.8
The present analysis represents the largest cohort investigating a link between HRT use and stroke in female patients with AF. The only available prior data assessing HRT use on the risk of stroke in women with AF were from the Stroke Prevention in Atrial Fibrillation (SPAF) I-III trials and that data included 90 women in the HRT group within a cohort of only 274 women.9 Our findings are pertinent given the interest into understanding the excess risk of thromboembolic events associated with females with AF.1 Possibly, social factors affect overall health and medication adherence, leading to improved outcomes although general studies of adherence suggest that education beyond high school and current working status may decrease adherence.10
Residual confounding due to incomplete control for imperfectly measured and unmeasured parameters cannot be excluded. Propensity score matching may reduce bias from lack of randomization but cannot control for unmeasured confounders. Specific hormonal preparations prescribed were not evaluated. Finally, the study may be underpowered to detect harm from HRT.
In conclusion, this analysis of female patients with AF in AFFIRM, HRT did not predict the primary composite end point, nor all-cause death, ischemic stroke, or major bleeding.
Drs Apostolakis, Olshansky, and Lip have received research funding and honoraria from various pharmaceutical companies in relation to atrial fibrillation for meetings and educational symposia. Dr Sullivan has no conflicts to declare.
The online-only Data Supplement is available with this article at http://stroke.ahajournals.org/lookup/suppl/doi:10.1161/STROKEAHA.114.006668/-/DC1.
- Received July 7, 2014.
- Revision received July 7, 2014.
- Accepted July 15, 2014.
- © 2014 American Heart Association, Inc.
- Wagstaff AJ,
- Overvad TF,
- Lip GY,
- Lane DA
- Wyse DG,
- Waldo AL,
- DiMarco JP,
- Domanski MJ,
- Rosenberg Y,
- Schron EB,
- et al
- Hohnloser SH,
- Oldgren J,
- Yang S,
- Wallentin L,
- Ezekowitz M,
- Reilly P,
- et al
- Rossouw JE,
- Anderson GL,
- Prentice RL,
- LaCroix AZ,
- Kooperberg C,
- Stefanick ML,
- et al
- Hulley S,
- Grady D,
- Bush T,
- Furberg C,
- Herrington D,
- Riggs B,
- et al
- Hart RG,
- Pearce LA,
- McBride R,
- Rothbart RM,
- Asinger RW