Author + information
- Received November 12, 2012
- Revision received January 29, 2013
- Accepted February 19, 2013
- Published online May 21, 2013.
- Stacie L. Daugherty, MD, MSPH⁎,⁎ (, )
- Lauren E. Thompson, MD†,
- Sunghee Kim, PhD‡,
- Sunil V. Rao, MD‡,§,
- Sumeet Subherwal, MD, MBA‡,§,
- Thomas T. Tsai, MD, MSc⁎,‖,
- John C. Messenger, MD⁎ and
- Frederick A. Masoudi, MD, MSPH⁎
- ↵⁎Reprints and correspondence:
Dr. Stacie L. Daugherty, Division of Cardiology, University of Colorado School of Medicine, 12605 East 16th Avenue, Mailstop B130, PO Box 6511, Aurora, Colorado 80045
Objectives This study sought to compared the use and effectiveness of bleeding avoidance strategies (BAS) by sex.
Background Women have higher rates of bleeding following percutaneous coronary intervention (PCI).
Methods Among 570,777 men (67.5%) and women (32.5%) who underwent PCI in the National Cardiovascular Data Registry's CathPCI Registry between July 1, 2009 and March 31, 2011, in-hospital bleeding rates and the use of BAS (vascular closure devices, bivalirudin, radial approach, and their combinations) were assessed. The relative risk of bleeding for each BAS compared with no BAS was determined in women and men using multivariable logistic regressions adjusted for clinical characteristics and the propensity for receiving BAS. Finally, the absolute risk differences in bleeding associated with BAS were compared.
Results Overall, the use of any BAS differed slightly between women and men (75.4% vs. 75.7%, p = 0.01). When BAS was not used, women had significantly higher rates of bleeding than men (12.5% vs. 6.2%, p < 0.01). Both sexes had similar adjusted risk reductions of bleeding when any BAS was used (women, odds ratio: 0.60, 95% confidence interval [CI]: 0.57 to 0.63; men, odds ratio: 0.62, 95% CI: 0.59 to 0.65). Women and men had lower absolute bleeding risks with BAS; however, these absolute risk differences were greater in women (6.3% vs. 3.2%, p < 0.01).
Conclusions Women continue to have almost twice the rate of bleeding following PCI. The use of any BAS was associated with a similarly lower risk of bleeding for men and women; however, the absolute risk differences were substantially higher in women. These data underscore the importance of applying effective strategies to limit post-PCI bleeding, especially in women.
Peri-procedural bleeding is the most common noncardiac complication following percutaneous coronary intervention (PCI) and is associated with high morbidity and mortality (1–3). Historically, women have been at higher risk for peri-procedural bleeding following PCI compared with men (4–10). Bleeding avoidance strategies (BAS), including vascular closure devices (VCDs), bivalirudin, and radial access, are increasingly used and have been associated with decreased rates of bleeding following PCI (8,9,11–13). In practice, however, those at the highest predicted risk for bleeding are often the least likely to receive BAS at the time of PCI, suggesting a “risk-treatment paradox” (11). Whether women, who are known to be at high risk for bleeding, receive BAS during PCI as frequently as men in contemporary practice has not been determined. Furthermore, whether BAS are associated with similar reductions in peri-procedural bleeding in women compared with men is not known.
To address these gaps in knowledge, we compared the use of BAS (VCDs, bivalirudin, radial access, or their combinations) by sex and conducted an observational comparative effectiveness study of BAS to determine whether the lower risk of bleeding associated with BAS use was similar between women and men. This study was designed to provide a contemporary assessment of the use of BAS and the extent to which BAS may reduce the risk of this common adverse consequence in women undergoing PCI.
Data were obtained from the National Cardiovascular Data Registry's (NCDR) CathPCI Registry, which is an initiative of the American College of Cardiology (ACC) Foundation and the Society for Cardiovascular Angiography and Interventions. A detailed description of the registry has been published previously (14). Demographic, clinical, procedural, and institutional data elements for PCI procedures are collected at more than 1,400 participating centers throughout the United States (∼80% of hospitals with invasive catheterization laboratories). Data are entered via a secure Web-based platform or via software provided by ACC certified vendors. Data quality assurance measures include automatic system validation and reporting of data completeness, random on-site auditing of participating centers, and education and training of site data managers (15).
All patients within the CathPCI Registry discharged after PCI between July 1, 2009 and March 31, 2011 were candidates for inclusion (n = 606,382 patients at 1,232 sites). Patients whose PCI did not use either a radial or femoral approach (n = 1,997) and those who underwent more than 1 PCI procedure during their hospital stay (n = 12,488) were excluded. Patients were also excluded if they had cardiogenic shock (n = 12,746), died the same day as the procedure (n = 568), or if they were missing data to determine a bleeding event (n = 165). Patients were also excluded if they received manual compression and a closure device (n = 7,382) because it was felt these cases might reflect failed deployment of the closure device. In addition, patients who had radial access and received a closure device (n = 106) or bivalirudin, radial access, and a closure device (n = 153) were excluded because this combination of treatments was felt to reflect procedures with combined femoral and radial access that might inherently pose a higher risk for peri-procedural bleeding irrespective of the BAS strategy used. After applying exclusions (total of 5.9% excluded), 570,777 patients at 1,230 sites remained. Among the final study cohort, 385,103 (67.5%) were men and 185,674 (32.5%) were women (Fig. 1).
In-hospital bleeding complications following PCI were ascertained and reported by participating centers. Peri-procedural bleeding was defined according to the CathPCI V4 data definitions and included: 1) any documented bleeding event that occurred within 72 h after PCI regardless of site (including access site bleeding, access site hematoma, retroperitoneal bleeding, gastrointestinal bleeding, genital-urinary bleeding, intracerebral hemorrhage); 2) pericardial tamponade; 3) any transfusion following PCI (except among patients with pre-procedure hemoglobin ≤8 g/dl or those who underwent coronary artery bypass grafting during their hospital stay); or 4) any absolute decline of ≥3 g/dl in hemoglobin level (except for patients with pre-procedure hemoglobin >16 g/dl) (16).
Bleeding avoidance strategies
BAS studied included: 1) VCDs alone (see Online Table 1 for list of specific devices included); 2) bivalirudin alone (Angiomax, The Medicines Company, Parsippany, New Jersey); 3) bivalirudin with VCD; 4) radial access alone; and 5) radial access with bivalirudin. Patients receiving manual compression who did not receive VCD, bivalirudin, or radial access served as the referent group for effectiveness comparisons.
Pre-procedural bleeding risk estimation
Estimated bleeding risk scores based upon pre-procedural patient characteristics were derived using the CathPCI bleeding risk model, version 4 (17). Risk scores were generated for each patient based on the inverse logarithmic sum of the beta coefficients for each of the following pre-PCI variables: sex, age, body mass index, previous cerebrovascular disease, chronic lung disease, previous PCI, peripheral vascular disease, diabetes mellitus, left ventricular ejection fraction, chronic kidney disease, PCI status (defined as elective, urgent, emergent, or salvage), ST-segment elevation myocardial infarction (STEMI), non–ST-segment elevation myocardial infarction (NSTEMI), cardiac arrest within 24 h, pre-procedure New York Heart Association (NYHA) class IV heart failure, estimated glomerular filtration rate, pre-procedure hemoglobin, pre-procedure Thrombolysis In Myocardial Infarction flow, number of diseased vessels, use of fibrinolytics before PCI, subacute stent thrombosis, Society for Cardiovascular Angiography and Intervention lesion class, and lesion location (proximal left anterior descending or left main vs. other).
Baseline demographic, clinical, procedural, and hospital characteristics were compared between women and men using Pearson chi-square tests for categorical variables and Kruskal-Wallis tests for continuous variables. In the primary outcomes analysis, rates of in-hospital bleeding were compared between women and men. Next, the use of each BAS was compared by sex. Given the previous studies that demonstrated a risk treatment paradox for BAS use (i.e., lower rates of use among those at highest predicted risk for bleeding), rates of use of any BAS were compared between women and men among tertiles of bleeding risk (defined as low [<2.3%], intermediate [2.3% to 5.0%], and high [>5%]) according to the CathPCI bleeding risk model, version 4 (11). Because female sex was a determinant of higher bleeding risk, there were a larger proportion of women in the higher tertiles of bleeding risk.
To compare the effectiveness of BAS strategies by sex, the crude relative risk ratios in bleeding for any BAS and individual BAS types versus no BAS were compared between women and men. Next, among strata of women and men, multivariable logistic regression was used to assess the relationship between each BAS strategy (compared with none) and peri-procedural bleeding, adjusting for baseline patient, procedural, and site characteristics (all Tables 1 and 2⇓ variables, except estimated bleeding risk), as well as clustering by site and the propensity to receive each BAS.
Propensity scores for each BAS were calculated in women and men to minimize the effect of potential selection bias for BAS choice. The propensity score for receiving each BAS was derived using multiple logistic regression models. Variables used to derive these propensity scores included demographics (age, sex, race/ethnicity), clinical characteristics (body mass index, NYHA heart failure classification), coronary artery disease risk factors (diabetes, hypertension, dyslipidemia, smoking, family history of coronary artery disease), coronary artery disease history (previous PCI, coronary artery bypass graft surgery, myocardial infarction), other cardiovascular disease history (congestive heart failure, cerebrovascular disease, peripheral vascular disease), other disease history (chronic obstructive pulmonary disease, renal failure), and presenting syndrome (no symptoms, atypical chest pain, stable angina, unstable angina, STEMI, and NSTEMI) (11). Given the smaller numbers of patients within certain BAS groups (e.g., bivalirudin plus radial) and potential loss of a significant portion of the population with propensity matching, inverse probability-weighted estimators were used (18). Compared with matching and stratification, semiparametric inverse probability-weighted estimators require few distributional assumptions about underlying data, and they avoid the potential residual confounding that arises from stratification on a fixed number of strata (19). Finally, the absolute differences in bleeding risk for each BAS compared with none were determined for women and men using chi-square tests.
All statistical analyses were performed using Statistical Analytic Systems version 9.2 (SAS Institute, Cary, North Carolina). The study was approved by the Institutional Review Board of Duke University Medical Center and was determined to meet the definition of research not requiring informed consent.
Study population characteristics
Baseline demographic and clinical characteristics for women and men are shown in Table 1. Compared with men, women were older and more likely to have higher NYHA class, have a lower estimated glomerular filtration rate, and have a slightly lower body surface area. Admissions symptoms, procedural details, and hospital characteristics for women and men are shown in Table 2. Women more often presented with an acute coronary syndrome (NSTEMI or unstable angina) compared with men and less often underwent PCI for an emergency procedure or for an emergent indication.
According to the CathPCI Registry bleeding risk model V4, the mean ± SD estimated bleeding risk was significantly higher for women than for men (8.0 ± 7.1% vs. 4.0 ± 3.9%; p < 0.01). Women were significantly more likely to be categorized as either at high (54.1% vs. 23.3%) or intermediate (38.9% vs. 30.7%) risk for bleeding compared with men; only 7% of women compared with 46% of men were categorized as low risk for predicted bleeding (p < 0.01 for all) (Table 1).
Bleeding outcomes by sex
Following PCI, 7.8% of women and 3.7% of men experienced a bleeding event (p < 0.01). The rates of all of the components of the composite bleeding outcome were significantly higher among women compared with men, with the exception of lower rates of genitourinary bleeding (0.05% vs. 0.07%, p < 0.01). Both access and nonaccess site bleeding complications were more frequent in women. The largest contributors to bleeding events for both sexes were related to post-PCI transfusions and hemoglobin decreases ≥3 g/dl (Online Table 2). In the bleeding outcomes models adjusted for patient, procedural characteristics, site characteristics, and the inverse probability-weighted estimators to receive BAS, women were more than twice as likely to bleed compared with men (odds ratio [OR]: 2.23, 95% confidence interval [CI]: 2.17 to 2.30).
Use of BAS by sex and bleeding risk
Overall, the use of any BAS was statistically different in women and in men, but absolute differences were small (overall use of BAS in women 75.4% vs. men 75.7%, p = 0.01; absolute difference 0.3%). Compared with men, women were more likely to receive bivalirudin (31.0% vs. 27.5%, p < 0.01), less likely to undergo a radial approach (3.0% vs. 3.5%, p < 0.05), or have a closure device deployed (16.4% vs. 18.6%, p < 0.01).
BAS of any type was used the least often in both women and in men at highest risk for bleeding compared with those in the lower-risk tertiles (men: 63.9% high vs. 76.4% intermediate and 81.2% low; women: 71.4% high vs. 79.7% intermediate and 82.4% low; p < 0.01 for both). Further, the interaction of sex and risk category for receiving BAS was statistically significant (p < 0.01), suggesting that the use of BAS significantly varied by sex and risk category, with the largest sex differences in BAS use seen among the highest risk women and men (71.4% vs. 63.9%).
BAS and bleeding risk by sex
Women who did not receive BAS had significantly more bleeding events following PCI compared with men who did not receive BAS (crude rates of 12.6% vs. 6.2%, p < 0.01). For women and men, the use of all BAS types was associated with a similarly lower relative risk of bleeding compared with no BAS. In the multivariable models adjusting for patient characteristics, procedural information, hospital data, clustering by site, and the propensity to receive each BAS, both women and men had significantly lower odds of bleeding with each BAS strategy compared with no BAS (Fig. 2). There was no significant interaction between sex and specific BAS type with the exception of the radial approach plus bivalirudin group, where the relative risk ratio was significantly greater in women than in men (p = 0.02 for interaction) (Fig. 2).
Given the higher bleeding risk for women among all BAS strata, the absolute differences in bleeding associated with each BAS compared with none were greater in women (range: 4.1% to 9.5%) than in men (range: 2.2% to 4.4%) (Fig. 3). Compared with the bleeding risk associated with the use of no BAS, the largest absolute differences in bleeding risk were seen among women and men who underwent the radial approach and received bivalirudin (9.5% absolute bleeding risk difference in women; 4.4% absolute bleeding risk difference in men) (Fig. 3).
In this contemporary cohort of over 500,000 patients who underwent PCI at over 1,200 hospitals throughout the United States, women had twice the risk of bleeding compared with men. Although a risk treatment paradox was seen for both women and men, with those at highest risk for bleeding being the least likely to receive BAS, women and men were overall similarly likely to receive any form of BAS following PCI. Finally, all BAS were associated with similar differences in bleeding risk regardless of sex; however, given the higher bleeding risk in women, the absolute differences in bleeding risk associated with all BAS were almost two times higher in women compared with men.
Although others have reported sex differences in bleeding following PCI, our study expands the literature in several ways (4–10). First, our study reflects a contemporary comparison of peri-procedural bleeding in women and in men from 2009 to 2011, a time in which advances in pharmacological agents (i.e., the use of bivalirudin) and procedural techniques (i.e., increased use of a radial approach) occurred (13,20,21). Second, our study includes the largest cohort of women and men who underwent PCI throughout the United States. Finally, our study employs an established and broad definition for peri-procedural bleeding. The CathPCI Registry definition of bleeding includes not only bleeding at the access site, but also bleeding at nonaccess sites, and post-procedural transfusion or an absolute decrease in hemoglobin (22). This definition is similar to the bleeding definitions used in contemporary clinical trials of PCI procedures that focus on both access- and nonaccess-related bleeding as a primary outcome (23). Bleeding of various degrees, including minor bleeding, among patients undergoing PCI is associated with increased risk of mortality and morbidity, longer hospital stay, and increased costs; therefore, a heterogeneous definition for bleeding allows a greater sensitivity to detect patients at higher risk for poor outcomes (1,24,25).
The present study demonstrates that in contemporary practice women have almost twice the rate of bleeding compared with men following PCI (7.8% vs. 3.7%, adjusted OR: 2.23, 95% CI: 2.17 to 2.30), despite the relatively frequent use of BAS. Over the last 2 decades, numerous studies have similarly demonstrated significantly higher rates of post-procedural bleeding in women (4–10,26). Another important contribution of this study is that we found that approximately 3 of 4 women and men undergoing PCI receive any type of BAS. Although overall BAS use was similar in women and in men, our study confirms the presence of a risk treatment paradox for both women and men (11). Future studies are needed to investigate potential reasons for lower rates of BAS use among the highest-risk patients (11).
One of the most important findings of our study is a comparison of the relative and absolute risk differences in bleeding associated with different types of BAS among women and men. Overall, each BAS strategy studied was associated with similarly lower relative risks of bleeding in women and men compared with the use of no BAS. Although several studies have demonstrated reductions in bleeding risk with BAS compared with none, few have evaluated these relationships by sex (4,9). Our study suggests a slightly greater benefit of the combined use of the transradial approach and bivalirudin in women compared with men (reduction in bleeding: 69% in women vs. 54% in men, p < 0.02 for interaction). However, these findings should be interpreted with caution because the numbers of women and men undergoing a radial approach and receiving bivalirudin was relatively small. Additional studies are needed to confirm whether 1 type of BAS will prove to be more effective in women than in men.
Although we found similarly lower relative risk differences associated with all BAS types, the absolute differences in bleeding risk for women were approximately 2-fold higher than in men because of the higher baseline predicted risk of bleeding in women. For example, women who received bivalirudin and a VCD had an 8.3% absolute difference in bleeding risk compared with a 4.2% absolute difference among men. Absolute differences of this magnitude would result in an estimated number needed to treat to prevent 1 bleed of 12 for women and 24 for men. Given their significantly higher risk of bleeding, use of BAS in women might prevent a greater proportion of bleeding events; therefore, a low clinical threshold for BAS use among women should be considered.
First, controversy exists as to whether the BAS types examined in this study are efficacious in reducing bleeding. Specifically, studies on the safety of VCDs have shown increased, decreased, or neutral bleeding complications (13). Further, the extent to which bivalirudin reduces bleeding compared with unfractionated heparin alone is not well known (13). The present study suggests that the BAS types examined are associated with similar relative differences in bleeding in women and in men following PCI; however, randomized trials are needed to define the most efficacious strategy for bleeding avoidance. Second, bleeding outcomes are based on site documentation and may be underestimated. However, we used a definition for bleeding that included documentation of a bleeding event and objective drops in hemoglobin, which were the largest portion of bleeding events for women and men. Further, the definition was applied equally to women and men, and documentation would not be expected to vary by sex. Third, the CathPCI Registry only captures in-hospital bleeding events and does not evaluate longitudinal outcomes following discharge. However, in multiple studies, in-hospital bleeding events have been associated with longer-term morbidity and mortality (1–3). Fourth, the present analysis does not account for sheath size, which has been shown to be associated with differences in post-procedural bleeding among women compared with men (4,9). However, sheath size has been associated with adverse bleeding at the access site, and many of our bleeding outcomes were not access specific. Fifth, we were unable to account for possible sex differences in the dosing of either antithrombotic or antiplatelet agents. Previous studies have demonstrated that women were more likely to receive inappropriately high doses of both antithrombotic and antiplatelet agents, which may partially explain higher rates of bleeding in women (27). It is not clear, however, that rates of inappropriate dosing would necessarily differ among strata of BAS use. Finally, given the observational nature of our data, the possibility of residual unmeasured confounding may explain the differences in risk associated with BAS in our study. We attempted to minimize confounding by calculating a non-parsimonious propensity score for receiving each type of BAS among women and men and using inverse probability-weighted estimators.
This study has several implications for clinical care and future research. First, we demonstrated that sex differences in bleeding following PCI are not largely due to differences in BAS use. Second, although the risk treatment paradox for BAS use is present for women and men, it did not differ markedly between them. Third, the relative risk difference associated with each BAS was similar between women and men, with the exception of slightly greater differences in bleeding risk with the combined use of the radial approach and bivalirudin in women. Taken together, these findings suggest that persistent sex differences in bleeding post-PCI are not due to differences in the apparent effectiveness of BAS. Finally, the absolute reductions in bleeding risk were almost 2 times higher in women compared with men for all BAS types. Therefore, our findings underscore the importance of comparing available strategies and developing new approaches to reduce bleeding risk in women following PCI. One such trial is currently ongoing (SAFE-PCI for Women; NCT01406236).
In contemporary practice, women have almost twice the risk of bleeding following PCI. The use of BAS was associated with significantly lower bleeding risks for both men and women; however, the absolute risk differences were higher in women. These data underscore the importance of applying effective strategies to limit post-PCI bleeding, especially in women.
For supplemental tables, please see the online version of this article.
This research was supported by the American College of Cardiology Foundation's National Cardiovascular Data Registry (NCDR). Dr. Daugherty is supported by the National Heart, Lung, and Blood Institute (Award Number K08HL103776). The views expressed in this paper represent those of the authors, and do not necessarily represent the official views of the NCDR, its associated professional societies identified at www.ncdr.com, or the National Heart, Lung, and Blood Institute. Dr. Rao is a consultant for and receives consulting honoraria from The Medicines Company and Terumo. Dr. Masoudi is the Senior Medical Officer for the NCDR; and has received research grant support from the American College of Cardiology Foundation, and the Oklahoma Foundation for Medical Quality. All other authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- Abbreviations and Acronyms
- American College of Cardiology
- bleeding avoidance strategy
- confidence interval
- National Cardiovascular Data Registry
- non–ST-elevation myocardial infarction
- New York Heart Association
- odds ratio
- percutaneous coronary intervention
- ST-elevation myocardial infarction
- vascular closure device
- Received November 12, 2012.
- Revision received January 29, 2013.
- Accepted February 19, 2013.
- American College of Cardiology Foundation
- Ndrepepa G.,
- Berger P.B.,
- Mehilli J.,
- et al.
- Lindsey J.B.,
- Marso S.P.,
- Pencina M.,
- et al.
- Eikelboom J.W.,
- Mehta S.R.,
- Anand S.S.,
- Xie C.C.,
- Fox K.A.A.,
- Yusuf S.
- Duvernoy C.S.,
- Smith D.E.,
- Manohar P.,
- et al.
- Akhter N.,
- Milford-Beland S.,
- Roe M.T.,
- Piana R.N.,
- Kao J.,
- Shroff A.
- Ahmed B.,
- Piper W.D.,
- Malenka D.,
- et al.
- Dauerman H.L.,
- Rao S.V.,
- Resnic F.S.,
- Applegate R.J.
- Brindis R.G.,
- Fitzgerald S.,
- Anderson H.V.,
- Shaw R.E.,
- Weintraub W.S.,
- Williams J.F.
- Messenger J.C.,
- Ho K.K.,
- Young C.H.,
- et al.
- ↵(2011) NCDR CathPCI Registry v4. 4 Coder's Data Dictionary (American College of Cardiology Foundation, Washington, DC).
- Rao S.V.,
- Kaltenbach L.A.,
- Spertus J.A.,
- Krone R.J.,
- Singh M.,
- Peterson E.D.
- Rao S.V.,
- Ou F.S.,
- Wang T.Y.,
- et al.
- Roe M.T.,
- Messenger J.C.,
- Weintraub W.S.,
- et al.
- Mehta S.K.,
- Frutkin A.D.,
- Lindsey J.B.,
- et al.
- Jolly S.S.,
- Niemelä K.,
- Xavier D.,
- et al.
- Ndrepepa G.,
- Schuster T.,
- Hadamitzky M.,
- et al.
- Lansky A.J.,
- Hochman J.S.,
- Ward P.A.,
- et al.
- Alexander K.P.,
- Chen A.Y.,
- Newby L.K.,
- et al.