Author + information
- Received August 23, 2016
- Revision received September 8, 2016
- Accepted September 8, 2016
- Published online November 23, 2016.
- Eric A. Secemsky, MD, MSca,b,c,
- Ajay Kirtane, MD, SMd,
- Sripal Bangalore, MD, MHAe,
- Ion S. Jovin, MDf,
- Rachit M. Shah, MBBSf,
- Enrico G. Ferro, BSb,
- Neil J. Wimmer, MD, MScg,
- Matthew Roe, MD, MHSh,
- Dadi Dai, PhDh,
- Laura Mauri, MD, MScb,i and
- Robert W. Yeh, MD, MScb,c,∗ ()
- aDivision of Cardiology, Department of Medicine, Massachusetts General Hospital, Boston, Massachusetts
- bHarvard Medical School, Boston, Massachusetts
- cSmith Center for Outcomes Research in Cardiology, Department of Medicine, Beth Israel Deaconess Medical Center, Boston, Massachusetts
- dDivision of Cardiology, Department of Medicine, and Center for Interventional Vascular Therapy, Columbia University, New York, New York
- eDivision of Cardiology, Department of Medicine, New York University School of Medicine, New York, New York
- fDivision of Cardiology, Department of Medicine, Virginia Commonwealth University, Richmond, Virginia
- gDivision of Cardiology, Department of Medicine, Christiana Care Health System, Newark, Delaware
- hDuke Clinical Research Institute, Duke University, Durham, North Carolina
- iDivision of Cardiovascular Medicine, Department of Medicine, Brigham and Women’s Hospital, Boston, Massachusetts
- ↵∗Reprint requests and correspondence:
Dr. Robert W. Yeh, Smith Center for Outcomes Research in Cardiology, Department of Medicine, Beth Israel Deaconess Medical Center, 375 Longwood Ave, MASCO 4, Boston, Massachusetts 02215.
Objectives The purpose of this study was to describe temporal trends and determine the comparative effectiveness of bivalirudin versus unfractionated heparin (UFH) during percutaneous coronary intervention (PCI) for ST-segment elevation myocardial infarction (STEMI).
Background Several clinical trials have compared the safety and effectiveness of bivalirudin versus UFH during PCI for STEMI, but results have been conflicting.
Methods Trends in anticoagulant use were examined among 513,775 PCIs for STEMI from July 2009 through December 2014 within the National Cardiovascular Data Registry CathPCI Registry. We conducted an instrumental variable analysis comparing bivalirudin with UFH, using operator preference for bivalirudin as the instrument. We used a test of mediation to determine the extent to which differences in outcomes between anticoagulants were due to differences in use of glycoprotein IIb/IIIa inhibitors (GPI). Primary outcomes were in-hospital bleeding and mortality.
Results Bivalirudin use increased from 2009 through 2013, followed by a new decline. GPIs were used in 74.7% of UFH PCIs versus 26.5% of bivalirudin PCIs. In unadjusted analyses, bivalirudin was associated with decreased bleeding (risk difference [RD]: -4.2%; p < 0.001) and mortality (RD: -0.84%; p < 0.001). After instrumental variable analyses, bivalirudin remained associated with less bleeding (RD: -3.75%; p < 0.001), but not mortality (RD: -0.10%; p = 0.280). The higher rate of GPI use with UFH was responsible for more than one-half of bivalrudin’s bleeding reduction (GPI-adjusted RD: -1.57%; p < 0.001). Bleeding reductions were negligible for transradial PCI (RD: -0.11%; p = 0.842).
Conclusions The use of bivalirudin during STEMI has decreased. Bivalirudin was associated with reduced bleeding and no mortality difference. The bleeding reduction with bivalirudin was largely explained by the greater use of GPIs with UFH.
- percutaneous coronary intervention
- ST-segment elevation myocardial infarction
Strategies to reduce bleeding have become an integral component of current percutaneous coronary intervention (PCI) practice to decrease adverse outcomes (1,2). Bivalirudin, a direct thrombin inhibitor, has been demonstrated in multiple large-scale randomized trials to reduce major bleeding events after PCI among patients presenting with ST-segment elevation myocardial infarction (STEMI) compared with unfractionated heparin (UFH) (3–8). However, bleeding reductions associated with bivalirudin therapy have occurred at the expense of increased rates of acute stent thrombosis (3,4,6,8). These trials have also varied in the proportion of use of glycoprotein IIb/IIIa inhibitors (GPI) in the UFH arm, ranging from infrequent (6,7) to moderate (4,8) to obligatory (3). Given the association of GPI therapy with increased bleeding and reduced ischemia (9), it is not clear whether observed differences between bivalirudin and UFH may be, at least in part, secondary to the greater use of GPI therapy with UFH (10).
Currently, U.S. PCI guidelines do not endorse a primary antithrombotic strategy during PCI for STEMI (11,12), leaving the choice of procedural anticoagulant to the discretion of the physician operator. In addition, because contemporary trials have conflicted in results and varied in anticoagulant protocols, there remains a lack of consensus among U.S. clinicians regarding the safety and effectiveness of bivalirudin therapy during PCI (13–15).
The present study had 3 purposes. First, we examined temporal and physician operator variation in anticoagulant use during PCI in U.S. patients with STEMI from 2009 through 2014 as an assessment of whether new clinical trial evidence has influenced clinical practice (6). Second, we assessed the comparative effectiveness of bivalirudin versus UFH as used in clinical practice, using an instrumental variable approach to account for differences in treatment selection. Last, we assessed the extent to which bleeding reductions with bivalirudin were the result of the more frequent use of GPI therapy with UFH.
Study cohort and data sources
The study cohort was derived from the National Cardiovascular Data Registry (NCDR) CathPCI Registry, a national quality improvement program that collects in-hospital data on patients undergoing cardiac catheterizations and PCIs (16,17). Data on PCIs performed from July 1, 2009, through December 31, 2014, were analyzed. All PCIs in which patients presented with STEMI were included. PCIs were excluded if neither bivalirudin nor UFH were used, or when alternative anticoagulants including low molecular weight heparin, fondaparinux, and nonbivalirudin direct thrombin inhibitors were used (Online Figure 1). For the comparative effectiveness analysis, PCIs by operators who had an annual PCI volume ≤25th percentile of all operators (≤6 procedures per year) were also excluded before examination of outcomes data to study a population of operators regularly engaged in the invasive management of STEMI patients and for whom a clear preference for bivalirudin versus UFH could be determined.
Patient, procedural, and operator variables
Main outcome measures
The main outcome measures were in-hospital bleeding and in-hospital mortality. In-hospital bleeding was based on the NCDR Version 4 definition (19) and included any of the following occurring before hospital discharge: arterial access site bleeding; retroperitoneal, gastrointestinal, or genitourinary bleeding; intracranial hemorrhage; cardiac tamponade; post-procedure hemoglobin decrease of 3 g/dl in patients with a pre-procedure hemoglobin level of ≤16 g/dl; or post-procedure non-bypass surgery–related blood transfusion for patients with a pre-procedure hemoglobin level of ≥8 g/dl. Secondary outcome measures included access site bleeding, non-access site bleeding, use of red blood cell (RBC) transfusion, and repeat PCI for stent thrombosis, which was defined as a subsequent PCI performed during the same index hospitalization as the STEMI-related PCI for the indication of stent thrombosis (Online Methods).
Quarterly time trends of proportion of STEMI patients treated with bivalirudin, bivalirudin with GPI, UFH, and UFH with GPI were examined during the study period. Linear regression was used to test for differences in the proportion of use of each anticoagulant strategy over calendar quarters. The proportions of bivalirudin and GPI use among PCI operators were also examined, which ranged in use from 0% to 100% of all PCIs. For analyses stratified by use of bivalirudin versus UFH, the bivalirudin arm allowed for concomitant treatment with UFH, whereas the UFH arm excluded any treatment with bivalirudin.
This study compared the effect of bivalirudin versus UFH therapy on in-hospital bleeding and mortality after PCI in patients presenting with STEMI. The use of an instrumental variable approach was prespecified as the primary analytic strategy, because unmeasured confounders unaccounted for in traditional observational methods may affect both the operator selection of anticoagulant therapy during PCI and post-procedural outcomes. For instance, bivalirudin may be used more frequently in patients assessed as being at greater risk for bleeding. Patient frailty, which is associated with increased bleeding and adverse outcomes (20–22), might influence an operator’s determination of bleeding risk, yet is infrequently quantified in observational and registry databases. Instrumental variable analyses attempt to balance both these unmeasured and measured factors between patient groups, similar to randomized controlled trials, by exploiting situations where some degree of randomness affects how a patient is selected for treatment (23,24) (Online Methods).
For the instrumental variable analysis, a “preference-based” instrument (25,26), namely, the proportion of bivalirudin use among PCI-performing operators, was used. To develop this instrument, “high bivalirudin” operators (those in the highest quartile of bivalirudin use during PCI) and “low bivalirudin” operators (those in the lowest quartile of bivalirudin use during PCI) were identified. PCIs performed by operators outside the lowest and highest quartiles of bivalirudin use were excluded because these operators displayed weaker preferences, making them less suitable for an instrumental variable analysis, similar to prior studies (25,26).
An instrumental variable analysis was performed comparing bivalirudin versus UFH using the 2-stage least squares methodology (27,28) with adjustment for 56 potential confounders including patient, procedural, and operator variables (Online Methods). The modeling strategy generates an estimate of the effect of 100% use of bivalirudin versus 0% use of bivalirudin, derived from the effect in the top versus bottom quartile of bivalirudin users. Bivalirudin was first compared with UFH without adjusting for the different rates of GPI use in the 2 groups as a pragmatic comparison of how these 2 anticoagulants are used in clinical practice. This analytical approach is analogous to allowing operators to use GPIs at their discretion in a randomized clinical trial analyzed by the intention-to-treat principle. Subsequently, to assess the extent to which any observed differences in outcomes between treatment groups were explained by the unbalanced use of adjunctive GPI therapy, we additionally adjusted for GPI use in the 2-stage least squares approach, as a test of mediation (Online Methods).
Analyses were repeated in the prespecified subgroups of patients who received PCI by either transradial or transfemoral arterial access. As a sensitivity analysis, a propensity score analysis using 1:1 matching was performed. Covariates used for matching included those variables adjusted for in the instrumental variable models, including treatment with GPI.
A p value threshold of <0.05 was used for statistical significance. In addition, given the large sample size, a standardized difference (SD) was reported with a cutoff of >10% to define significance (29). Statistical analyses were performed using SAS software version 9.3 (SAS Institute, Cary, North Carolina).
Patterns of anticoagulant use
Between July 1, 2009, and December 31, 2014, 587,998 PCIs by 9,058 operators at 1,535 sites were performed in patients presenting with STEMI. Of these PCIs, 513,775 procedures by 6,878 operators at 1,507 sites met study criteria (Online Figure 1). PCI was performed within 12 hours of presentation for 89.1% of STEMI patients. Overall, bivalirudin was used in 47.1% of cases and UFH without bivalirudin was used in 52.9% of cases. Of PCIs using bivalirudin, UFH was administered concomitantly in 49.9% and adjunctive GPI therapy was used in 26.5%. For PCIs using UFH, adjunctive GPI therapy was administered in 74.7% (Online Figure 2).
There were notable changes in the use of procedural anticoagulants by PCI operators during the study period. Among all PCIs performed for STEMI at the start of the study in 2009, bivalirudin was used in 29.0% (18.6% as monotherapy; 10.4% with GPI) and UFH was used in 71.0% (13.6% as monotherapy; 57.4% with GPI) (Figure 1). Between 2009 and 2013, bivalirudin use increased linearly, concomitant with a reduction in UFH with GPI use, with 44.7% of PCIs performed using bivalirudin monotherapy and 27.2% of PCIs performed using UFH with a GPI by the end of 2013 (p < 0.001 for both trends). Starting in 2014, there was a new decline in bivalirudin use, paralleled by an increase in use of UFH monotherapy (by end of 2014: 37.7% bivalirudin monotherapy, 12.0% bivalirudin with GPI, 20.8% UFH monotherapy, and 29.4% UFH with GPI).
The preference for bivalirudin varied significantly among individual PCI operators (Figure 2). The highest quartile operators used bivalirudin in ≥84.2% of all PCIs, whereas the lowest quartile operators used bivalirudin in ≤10.7% of all PCIs. High and low bivalirudin operators had similar annual PCI volumes and rates of affiliation with teaching hospitals. However, low bivalirudin operators more commonly practiced in the Midwest region and in urban locations (Online Table 1).
Patient and lesion characteristics of PCIs by bivalirudin use
Patients treated with bivalirudin versus UFH were of similar age, sex and race. There were also nonsignificant differences in rates of cardiovascular risk factors, including hypertension, dyslipidemia, peripheral arterial disease, and diabetes between treatment groups (Table 1). Bivalirudin-treated patients presented with less severe New York Heart Association functional class heart failure symptoms, had higher rates of drug-eluting stent placement (61.8% vs. 56.8% for UFH; SD: -10.1%), and more often received prasugrel (23.3% vs. 16.1% for UFH; SD: -18.0%) or ticagrelor (16.2% vs. 11.2% for UFH; SD: -14.5%) (Table 2). There were otherwise no differences in high-risk lesions, number of diseased vessels, or number of stents placed between treatment groups.
In the study population, in-hospital bleeding occurred in 12.9% (4.0% with access site bleeding, 1.2% with non–access site bleeding, and 4.8% with RBC transfusion), in-hospital mortality in 5.6%, and need for repeat PCI for stent thrombosis in 0.71% (Online Table 2). In unadjusted analyses, treatment with bivalirudin was associated with lower rates of in-hospital bleeding (10.7% with bivalirudin vs. 14.8% with UFH; risk difference [RD]: -4.2%; 95% confidence interval [CI]: -4.3% to -4.0%; p < 0.001) and in-hospital mortality (5.1% with bivalirudin vs. 6.0% with UFH; RD: -0.84%; 95% CI: -0.97% to -0.71%; p < 0.001) compared with UFH. In addition, there were lower rates of access site bleeding (3.1% with bivalirudin vs. 4.7% with UFH; RD: -1.7%; 95% CI: -1.8% to -1.6%; p < 0.001), non–access site bleeding (0.93% with bivalirudin vs. 1.35% with UFH; RD: -0.42%; 95% CI: -0.48% to -0.36%; p < 0.001), and need for RBC transfusion (4.0% with bivalirudin vs. 5.5% with UFH; RD: -1.5%; 95% CI: -1.7% to -1.4%; p < 0.001) among bivalirudin recipients. The rate of repeat PCI for stent thrombosis during the index hospitalization occurred more frequently in patients treated with bivalirudin compared with UFH (0.92% with bivalirudin vs. 0.51% with UFH; RD: 0.41%; 95% CI: 0.36% to 0.46%; p < 0.001). Rates of outcomes by different anticoagulant regimens are displayed in Online Table 3.
Instrumental variable outcomes
After application of the instrumental variable, patient, procedural, and operator characteristics among those treated by high versus low bivalirudin operators were well-balanced, with 52 of 56 variables having an SD of <10% (Online Tables 1, 4, and 5). In addition, the instrument was highly predictive of actual bivalirudin receipt, with 93.6% of patients treated by high bivalirudin operators receiving bivalirudin and 3.5% of patients treated by low bivalirudin operators receiving bivalirudin.
After adjustment for clinical and operator covariates, bivalirudin use was associated with a 3.75% absolute risk reduction in in-hospital bleeding (95% CI: -4.04% to -3.47%; p < 0.001), but no difference in in-hospital mortality (RD: -0.10%; 95% CI: -0.28% to 0.08%; p = 0.280) (Figure 3A, Online Table 6). Bivalirudin use also corresponded with absolute risk reductions in access site bleeding (RD: -1.56%; 95% CI: -1.73% to -1.39%; p < 0.001), non–access site bleeding (RD: -0.29%; 95% CI: -0.38% to -0.19%; p < 0.001), and the need for RBC transfusion (RD: -1.13%; 95% CI: -1.31% to -0.95%; p < 0.001). There was a 0.43% absolute increase in the rate of repeat PCI for stent thrombosis associated with bivalirudin therapy (95% CI: 0.36% to 0.51%; p < 0.001).
The bleeding reduction with bivalirudin was, in part, mediated by the higher rate of GPI usage with UFH. When differences in GPI use were accounted for, bivalirudin was associated with an absolute risk reduction of 1.57% (95% CI: -1.90% to -1.25%; p < 0.001) (Figure 3A, Online Table 6). Absolute risk reductions in access-site bleeding (RD: -0.59%; 95% CI: -0.79% to -0.39%; p < 0.001) and need for RBC transfusion (RD: -0.90%; 95% CI: -1.11% to -0.69%; p < 0.001) associated with bivalirudin use were diminished whereas the reduction in non–access site bleeding was no longer observed (RD: 0.05%; 95% CI: -0.06% to 0.16%; p = 0.400). The observed increase in repeat PCI for stent thrombosis was also attenuated but remained significant after accounting for GPI usage (RD: 0.20%; 95% CI: 0.12% to 0.29%; p < 0.001).
Among patients receiving PCI by transradial access, bivalirudin was not associated with differences in any outcome, irrespective of adjustment for GPI use (p > 0.05 for all) (Figure 3B, Online Table 6). However, in patients treated by transfemoral access, there were significant reductions in in-hospital bleeding (RD: -4.04%; 95% CI: -4.34% to -3.74%; p < 0.001), access site bleeding (RD: -1.69%; 95% CI: -1.87% to -1.51%; p < 0.001), and need for RBC transfusion (RD: -1.23%; 95% CI: -1.42% to -1.04%; p < 0.001), as well as a 0.48% increase in repeat PCI for stent thrombosis (95% CI: 0.40% to 0.56%; p < 0.001). After adjustment for GPI use, these RDs were attenuated, yet remained significant (in-hospital bleeding: RD: -1.72%; 95% CI: -2.07% to -1.37%; p < 0.001; access site bleeding: RD: -0.65%, 95% CI: -0.86% to -0.44%; p < 0.001; RBC transfusion: RD: -0.96%, 95% CI: -1.18% to -0.74%; p < 0.001; repeat PCI for stent thrombosis: RD: 0.24%; 95% CI: 0.15% to 0.33%; p < 0.001) (Figure 3C, Online Table 6). In addition, accounting for differential GPI use resulted in a significant reduction in in-hospital mortality associated with bivalirudin therapy among patients undergoing transfemoral access (RD: -0.78%; 95% CI: -1.00% to -0.44%; p < 0.001).
In the sensitivity analysis, RDs were overall similar between the propensity score analysis and the instrumental variable analysis that included adjustment for GPI use (Online Table 7). There were also comparable RDs when stratified by transradial versus transfemoral access.
In a large, contemporary cohort of PCI patients presenting with STEMI in the United States, we have the following observations. First, bivalirudin use increased from 2009 through 2013, followed by a new decrease in 2014. This decrease was paralleled by an increase in use of UFH monotherapy. Next, throughout the study period, there was a wide range in the individual operator preference for bivalirudin versus UFH therapy. By exploiting this heterogeneity in practice, we demonstrated through an instrumental variable approach that bivalirudin use was associated with decreased in-hospital bleeding. Third, more than one-half of the associated risk reduction in bleeding with bivalirudin therapy was due to the higher rate of concomitant GPI use with UFH. Finally, the bleeding reduction associated with bivalirudin treatment was greatest among patients undergoing transfemoral access and negligible when transradial access was used.
The observed change in temporal trends and heterogeneity in individual practice patterns may reflect the lack of consensus among U.S. operators regarding which anticoagulant strategy provides optimal outcomes during PCI for STEMI. This is evident by the timing of major shifts in patterns of anticoagulant use, which seem to be largely and immediately influenced by the release of clinical trial data. For instance, the increased use of bivalirudin therapy and corresponding decreased use of UFH plus GPI seen in 2009 closely followed the publication of the HORIZONS-AMI (Harmonizing Outcomes with Revascularization and Stents in Acute Myocardial Infarction) trial (3), a landmark study that demonstrated bleeding reductions and similar ischemic outcomes with bivalirudin use compared with UFH plus obligatory GPI for patients with STEMI. This shift in practice was supported by a subsequent large positive trial (4), reinforcing the safety and effectiveness of bivalirudin therapy. The observed rise in bivalirudin use was notably reversed starting in 2014, corresponding with the presentation of the results of the HEAT-PPCI (Unfractionated Heparin Versus Bivalirudin in Primary Percutaneous Coronary Intervention) trial (6). HEAT-PPCI demonstrated no difference in major bleeding between bivalirudin and UFH monotherapy and an increase in the composite ischemic endpoint with bivalirudin use. As such, with data suggesting equivalence in bleeding events between bivalirudin and UFH monotherapy, and in the setting of overall declining use of GPIs nationwide (30), more operators seemed willing to return to a strategy of UFH.
In addition, a large number of physicians in our study used bivalirudin and UFH with similar frequency, suggesting that individual operators frequently alternate between anticoagulation strategies. Although our data do not allow for a complete understanding of which variables drive anticoagulant choice, some factors, such as radial access, need for mechanical support, and left main coronary artery disease, strongly favored the use of UFH over bivalirudin in our analysis.
The results from our comparative effectiveness analysis are consistent with the results of randomized clinical trials (4,5,8), excluding the findings from HEAT-PPCI (6). When comparing anticoagulant strategies as used in contemporary clinical practice, which included a 75% rate of GPI use with UFH, bivalirudin use was associated with a 3.75% absolute reduction in in-hospital bleeding. In addition, the rate of repeat PCI for stent thrombosis was greater with bivalirudin, likely due to the frequent use of GPI therapy with UFH.
By additionally adjusting for GPI use, we were able to measure the extent to which the differences in outcomes observed between bivalirudin and UFH were associated with the much greater use of GPIs with UFH. When accounting for GPI use, the bleeding reduction associated with bivalirudin remained significant, but was more than halved. Similarly, the increase in repeat PCI for stent thrombosis associated with bivalirudin was considerably attenuated. Our results were supported by the sensitivity analysis, in which propensity score-based methods, which included use of GPIs as a covariate, demonstrated similar treatment effects between bivalirudin and UFH. These findings are also consistent with recent meta-analyses of randomized trials, which have demonstrated that after adjustment for use of GPIs, the bleeding reduction associated with bivalirudin is negated (10,31). However, our analysis importantly adds to this current body of literature by reproducing these findings within a real-world population of patients and operators, and by quantifying the absolute magnitude of bleeding reduction associated with bivalirudin that is due to the greater use of GPIs with UFH.
Transradial arterial access was used by a minority of operators in the study population, which is similar to prior analyses that have demonstrated its slow adoption during PCI for STEMI (32,33). However, treatment effects associated with bivalirudin were modified by site of arterial access. Patients receiving transradial access did not experience a reduction in in-hospital bleeding associated with bivalirudin, irrespective of GPI use. This is accordant with a previous analysis that observed negligible reductions in post-PCI bleeding when combing radial access with bivalirudin treatment and accounting for differential GPI use (34). This finding may also account for the lack of bleeding reduction with bivalirudin treatment observed in the HEAT-PPCI trial (6), which had >80% use of transradial access and <15% use of GPIs. In contrast, patients receiving transfemoral access had reductions in bleeding with bivalirudin treatment both with and without adjustment for GPI use, largely due to decreased access site bleeding and need for transfusion. As the use of transradial access during PCI for STEMI continues to increase, these findings suggest that there may be less of a role for bivalirudin as a means of reducing periprocedural bleeding.
The results of this analysis must be considered in the context of the study design. The definition of in-hospital bleeding was developed for a quality improvement registry and all outcomes were not adjudicated, which could have resulted in underestimation or overestimation of actual rates of in-hospital events. In particular, stent thrombosis is not a validated event in the CathPCI registry. The outcome “repeat PCI for stent thrombosis” was constructed for this analysis and only included those events occurring during the same hospitalization as the index PCI. Data were only available for in-hospital outcomes, whereas differences in treatment effects may occur with longer follow-up (35). Our analysis also lacked data on the dosage, timing, and duration of use of periprocedural antithrombotic agents and whether GPIs were used as routine or bailout therapy. For instance, differences in anticoagulant dosing, which is standardized by weight and renal function for bivalirudin and GPI therapy but not for UFH, may have affected the observed rate of periprocedural bleeding. In addition, the timing of the loading dose of the oral P2Y12 inhibitor and the duration and post-PCI dose of bivalirudin infusion may have influenced the observed rate of repeat PCI for stent thrombosis.
The validity of the study results also depends on the strength of the methodology used in the analysis (Online Methods). Our study prespecified the use of an instrumental variable analysis as the primary analytic approach, versus traditional methods of adjustment, as unmeasured patient characteristics may be important determinants of treatment and outcomes (20–22). The instrumental variable we used may have its own limitations, particularly if operators who used more versus less bivalirudin also differed in other unmeasured ways. In our case, however, both the results of the instrumental variable and the propensity score analysis yielded similar results, strengthening our confidence in the results.
Significant temporal and individual variations exist in anticoagulant preference during PCI for STEMI among U.S. operators. By the end of the study period, bivalirudin was the primary anticoagulant used for the treatment of STEMI; however, there was a decline in bivalirudin use in favor of UFH monotherapy starting in 2014. Through an instrumental variable approach, we found that bivalirudin was associated with reduced in-hospital bleeding, a difference that was in part explained by the greater use of GPIs with UFH.
WHAT IS KNOWN? Bivalirudin is commonly used during PCI for STEMI and has been associated with less bleeding in clinical trials, primarily when GPIs are co-administered with UFH.
WHAT IS NEW? In the present study, we show that since 2014, near the time of the presentation of the HEAT-PPCI trial, bivalirudin use during PCI for STEMI has decreased in exchange for UFH monotherapy. We also demonstrate that the greater use of GPIs with UFH partly, but not completely, account for the bleeding reduction observed with bivalirudin therapy.
WHAT IS NEXT? Future studies are needed to examine how changes in the use of adjunctive antithrombotics and arterial access site strategies will reshape the optimal role of bivalirudin in the management of patients with STEMI.
For supplemental materials, figures, and 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. Dr. Kirtane receives grants to institution from Boston Scientific, Abbott Vascular, Medtronic, Abiomed, St. Jude Medical, Eli Lilly, and GlaxoSmithKline. Dr. Bangalore is on the advisory board of The Medicines Company. Dr. Roe receives research funding from Eli Lilly, Sanofi, Daiichi-Sankyo, Janssen Pharmaceuticals, Ferring Pharmaceuticals, AstraZeneca, American College of Cardiology, American Heart Association, and the Familial Hypercholesterolemia Foundation. He provides consulting or has received honoraria from PriMed, AstraZeneca, Boehringer Ingelheim, Merck, Amgen, Myokardia, Eli Lilly, Daiichi-Sanyko, and Elsevier Publishers. Dr. Mauri receives grants to institution from Abbott, Boston Scientific, Medtronic Vascular, Eli Lilly, Daiichi Sankyo, Biotronik, Boehringher-Ingelheim, Sanofi Aventis, and Bristol-Myers Squibb. She provides consulting to Medtronic Vascular, AstraZeneca, St. Jude Medical, Biotronik, Boehringher-Ingelheim, and Eli Lilly. Dr. Yeh receives research funding from Abiomed and Boston Scientific; and is a consultant and serves on advisory boards for Abbott Vascular and Boston Scientific. All other authors reported that they have no relationships relevant to the contents of this article to disclose.
- Abbreviations and Acronyms
- confidence interval
- glycoprotein IIb/IIIa inhibitor
- National Cardiovascular Data Registry
- percutaneous coronary intervention
- red blood cells
- risk difference
- ST-segment elevation myocardial infarction
- unfractionated heparin
- Received August 23, 2016.
- Revision received September 8, 2016.
- Accepted September 8, 2016.
- American College of Cardiology Foundation
- Baklanov D.V.,
- Kim S.,
- Marso S.P.,
- Subherwal S.,
- Rao S.V.
- Schulz S.,
- Richardt G.,
- Laugwitz K.L.,
- et al.
- Levine G.N.,
- Bates E.R.,
- Blankenship J.C.,
- et al.
- O'Gara P.T.,
- Kushner F.G.,
- Ascheim D.D.,
- et al.
- Stone G.W.,
- Mehran R.,
- Steg P.G.
- Brindis R.G.,
- Fitzgerald S.,
- Anderson H.V.,
- Shaw R.E.,
- Weintraub W.S.,
- Williams J.F.
- ↵National Cardiovascular Data Registry. Data elements & definitions, technology downloads and risk adjustment. Available at: http://cvquality.acc.org/∼/media/QII/NCDR/Data%20Collection%20Forms/CathPCI%20Registry_DataCollectionForm.ashx. Accessed December 10, 2014.
- Rao S.V.,
- McCoy L.A.,
- Spertus J.A.,
- et al.
- Ekerstad N.,
- Swahn E.,
- Janzon M.,
- et al.
- Alonso Salinas G.L.,
- Sanmartin Fernandez M.,
- Pascual Izco M.,
- et al.
- Perera V.,
- Bajorek B.V.,
- Matthews S.,
- Hilmer S.N.
- Yeh R.W.,
- Vasaiwala S.,
- Forman D.E.,
- et al.
- Rassen J.A.,
- Brookhart M.A.,
- Glynn R.J.,
- Mittleman M.A.,
- Schneeweiss S.
- Subherwal S.,
- Peterson E.D.,
- Dai D.,
- et al.
- Bittl J.A.,
- He Y.,
- Lang C.D.,
- Dangas G.D.
- Howe M.J.,
- Seth M.,
- Riba A.,
- Hanzel G.,
- Zainea M.,
- Gurm H.S.
- Baklanov D.V.,
- Kaltenbach L.A.,
- Marso S.P.,
- et al.
- Perdoncin E.,
- Seth M.,
- Dixon S.,
- et al.
- Stone G.W.,
- Witzenbichler B.,
- Guagliumi G.,
- et al.