Does Comorbidity Account for the Excess Mortality in Patients With Major Bleeding in Acute Myocardial Infarction?
Background— Analyses from randomized controlled trials suggest that bleeding in patients with acute myocardial infarction is associated with poor outcomes. Because these data are not generalizable to all patients with acute myocardial infarction, we sought to better understand the scope of this problem in a “real-world” setting.
Methods and Results— We examined the frequency of major bleeding in 40 087 patients with acute myocardial infarction enrolled in the Global Registry of Acute Coronary Events. Regression analyses were used to examine the association between patient and treatment characteristics, bleeding, and hospital and postdischarge outcomes. Major bleeding occurred in 2.8% of patients. These patients were older, more severely ill, and more likely to undergo invasive procedures. Patients with bleeding were more likely to die during hospitalization (hazard ratio, 1.9; 95% confidence interval, 1.6 to 2.2) but not after discharge (hazard ratio, 0.8; 95% confidence interval, 0.6 to 1.0) than patients who did not bleed. Continuation of antithrombotic therapies after day 1 was lower in patients who experienced early bleeding. Moreover, in patients who bled, hospital mortality was increased in those who discontinued aspirin, thienopyridines, or low–molecular-weight heparins.
Conclusions— Major bleeding occurred in 1 in 35 patients with acute myocardial infarction; these patients accounted for ≈10% of all hospital deaths. Nevertheless, risk of hospital mortality associated with bleeding was much lower than reported in randomized controlled trials. These data suggest that although bleeding may be causally related to adverse outcomes in some patients in the real-world setting, it is often merely a marker for patients at higher risk for adverse outcomes.
Received February 5, 2007; accepted October 2, 2007.
Increased understanding of the essential, overlapping, and synergistic roles of platelet activation and thrombin generation/activity in patients with acute myocardial infarction (AMI) has led to the development of increasingly effective antiplatelet and thrombin-inhibition therapies. Percutaneous intervention, often performed in combination with 3, 4, or even 5 of these antithrombotic therapies, is being increasingly used in the management of patients with AMI. These advances have led to impressive improvements in our ability to achieve early and sustained coronary patency. However, the impact of these more aggressive treatment approaches on bleeding complications and hospital and long-term outcomes is not clear. Recent analyses from randomized controlled trials suggest that bleeding after treatment for AMI is associated with significantly worse overall outcomes.1,2 Because randomized controlled trial data are not necessarily generalizable to all patients with AMI,3 the objective of our large observational study was to better understand the magnitude and scope of this problem as it exists in the community setting.
Editorial p 2776
Clinical Perspective p 2801
Using data from a large multinational registry of patients with acute coronary syndromes (ACS), we report the frequency of major bleeding in patients with AMI, examine the relationship between major bleeding and antithrombotic therapy (with or without invasive procedures), and explore the association between major bleeding and adverse patient outcomes. Finally, because patients who suffer major bleeding are likely to have antithrombotic therapies discontinued, we hypothesized that the cessation of these therapies may partially explain any observed association between major bleeding and excess mortality. We begin to explore this hypothesis using data from a subset of patients who suffered a major bleeding episode within the first 24 hours of hospital admission.
Full details of the Global Registry of Acute Coronary Events (GRACE) rationale and methodology have been published and are outlined below.4–6
GRACE is designed to reflect an unbiased and generalizable sample of patients with ACS. A total of 114 hospitals located in 14 countries (Argentina, Australia, Austria, Belgium, Brazil, Canada, France, Germany, Italy, New Zealand, Poland, Spain, United Kingdom, and United States) across 4 continents have contributed data to this observational study.
Patient enrollment at each study hospital is intended to reflect an unbiased sample of admissions for ACS that is independent of the annual volume of ACS patients seen at each of the participating hospitals. Individual hospital enrollment targets have been uniformly established as the first 10 qualifying cases of ACS discharged each month. Regular audits are performed at all participating hospitals.
Patients entered in the registry had to be ≥18 years of age, be admitted for ACS as a presumptive diagnosis, and have at least one of the following: ECG changes consistent with ACS, serial increases in biochemical markers of cardiac necrosis, and/or documentation of coronary artery disease.4–6 The qualifying ACS must not have been precipitated by trauma or surgery. For purposes of the present study, patients with unstable angina but no increase in cardiac markers or those who were subsequently found to have non-ACS primary diagnoses and patients transferred into or out of GRACE registry hospitals were excluded from further consideration. When required, study investigators received approval from their hospital ethics or institutional review board, and a signed consent form for follow-up contact was obtained.
Data were collected at each site by a trained coordinator using a standardized case report form. Demographic and clinical characteristics, treatment practices, and hospital outcome data, including information on occurrence and timing of major bleeding, were collected. Standardized definitions of all patient-related variables, clinical diagnoses, selected hospital complications, and outcomes were used. Medications and procedures received in hospital were categorized into those received during the “first 24 hours of admission” or those received “anytime during hospitalization.” Major in-hospital bleeding was defined as life-threatening bleeding requiring a transfusion of ≥2 U of packed red blood cells, resulting in a decrease in hematocrit of ≥10%, occurring intracerebrally, or resulting in stroke or death. Date and site of each episode of major bleeding were recorded. Bleeding that occurred after coronary artery bypass graft surgery was not included in our analyses.
Differences in the demographic and clinical characteristics and treatment practices of study patients in relation to occurrence of major bleeding were analyzed with χ2 tests for discrete variables. The Wilcoxon rank-sum test was used to analyze differences between respective comparison groups for continuous variables. Differences in clinical presentation (ST-segment elevation versus non–ST-segment elevation) and subsequent diagnosis (ST-segment elevation myocardial infarction [STEMI] versus non-STEMI) in relation to major bleeding were analyzed with χ2 tests. Fisher’s exact test was used to compare mortality between patients with bleeding who stopped versus those who continued anticoagulant use. The Kaplan–Meier method was used to create survival curves for patients with and without in-hospital bleeding. Major bleeding was treated as a time-varying covariate (see below).
Cox proportional-hazards regression analyses were used to examine the association between patient demographic, clinical, and treatment characteristics and the occurrence of major bleeding. Because our initial case report form (1999) did not include information on the timing of glycoprotein IIb/IIIa inhibitors, we excluded these patients (n=5595) from our analyses. Of the remaining 34 492 patients, we had complete covariate data on 28 327 (81%), who comprised the final cohort for our bleeding models.
Candidate variables for inclusion in regression models included the demographic, clinical, and treatment characteristics listed in Table 1⇓. Only procedures (eg, cardiac catheterization, percutaneous coronary intervention) occurring before major bleeding were considered for these analyses. These procedures and fibrinolytic therapy, for which timing was available, were modeled as time-varying covariates. Similarly, only medical therapies administered within the first 24 hours of hospitalization were included in these analyses. Candidate variables possibly associated with in-hospital bleeding (P≤0.25 after univariate analysis) were included in the multivariable models. Variables with values of P>0.05 were eliminated in a backward fashion so that only variables with a statistically significant association with the outcome of interest were included in the final regression models. The assumption of proportional hazards was tested for each covariate by a covariate-by-time interaction. Recognizing that different demographic, clinical, and treatment factors may be associated with “early” bleeding (defined as occurring in days 0 to 1 or days 0 to 2) versus “late” bleeding (defined as occurring in days 2 to 30 or days 3 to 30), we reported early and late differential hazards if statistically significant (P≤0.05) using SAS programming statements (SAS Institute, Cary, NC).7
Using similar methods, we also performed multiple Cox regression analyses to test whether hospital bleeding was related to hospital and postdischarge mortality. These analyses were performed in the cohort of patients with complete covariate data (n=32 240). In these analyses, bleeding was treated as a time-varying covariate, as was hospital stay, as follows: Let X(t) represent bleeding by day t (day t extends from admission day to 180 days after admission), where X(t)=0 if the patient never bled or bled after day t and X(t)=1 if the patient bled on or before day t. Let L(t) represent whether day t occurs in hospital or after discharge, where L(t)=0 if day t is after discharge (t>length of stay in hospital) and L(t)=1 if day t is before discharge. The bleeding hazard ratio for in-hospital mortality is then B1X(t)+B2X(t)×L(t) and B1X(t) for postdischarge mortality, where B1 is the estimate for bleeding by day t and B2 is the estimate for bleeding-by-hospital-stay interaction.
These analyses were conducted in the overall study sample and in patients who did or did not receive fibrinolytic therapy. Multiple Cox regression also was used to assess the adjusted relation between in-hospital bleeding and subsequent complications of recurrent myocardial infarction and stroke during hospitalization.
The authors had full access to and take full responsibility for the integrity of the data. All authors have read and agree to the manuscript as written.
The study population consisted of 40 087 men and women with AMI (53% STEMI, 47% NON-STEMI) enrolled in GRACE from April 1999 to March 2007. The median length of stay was 6 days (interquartile range, 3 to 10 days). Of these, 1140 patients (2.8%) experienced a major bleeding episode during hospitalization for AMI. The location of major bleeding was at the site of vascular access in 29% of patients, intracerebral in 6%, at another site in 53%, and not reported in 12%. Approximately 57% of patients with major bleeding received blood transfusions of ≥2 U, 4% underwent surgery to explore or repair the bleeding site, and 3% received transfusions and underwent surgery. Approximately half (49%) of the major bleeding events occurred the day of or the day after hospital admission. Patients with major bleeding in hospital were more likely to die in the next 6 months than patients without major bleeding (25.7% versus 9.3%; P<0.001). The 6-month survival curves of patients with and without major bleeding are shown in Figure 1.
Demographic, Clinical, and Treatment Characteristics of Patients With or Without Major Bleeding
Patients experiencing major bleeding during hospitalization were older, more likely to be female, and more likely to have a history of a number of comorbidities than patients without major bleeding (Table 1⇑). Patients with major bleeding were of lower body mass index; had worse renal function, lower blood pressure, and higher pulse rates on admission; and were less likely to present in Killip class I than patients without major bleeding. Patients with major bleeding were more likely to have a discharge diagnosis of STEMI.
Patients with major bleeding were more likely to be treated during hospitalization with glycoprotein IIb/IIIa inhibitors, unfractionated heparin, thienopyridines, diuretics, and intravenous vasopressor therapy but were less likely to be treated with aspirin, β-blockers, or low-molecular-weight heparins than those without major bleeding (Table 1⇑). Overall use of fibrinolytic therapy did not differ significantly between patients with or without major bleeding. Patients with major bleeding also were more likely to have undergone percutaneous coronary intervention, intra-aortic balloon placement, or pulmonary artery catheter placement during hospitalization than patients without a major bleeding episode.
After Cox proportional-hazards regression analysis, increasing age; female sex; history of prior bleeding, smoking, or peripheral arterial disease; presentation with glomerular filtration rate <30 mL/min or ST-segment deviation on initial ECG; treatment with intravenous vasopressor agent(s) or glycoprotein IIb/IIIa inhibitors in the first 24 hours; treatment with fibrinolytics; and use of cardiac catheterization, percutaneous coronary intervention, intra-aortic balloon pump, or pulmonary artery catheter were associated with bleeding in the first 30 days (Table 2). Treatment with aspirin or unfractionated heparin in the first 24 hours was inversely associated with major bleeding in the first 30 days. History of atrial fibrillation or presentation with increased heart rate was associated with late but not early bleeding. History of hypertension was inversely associated with early but not late bleeding, whereas presentation with ST elevation on initial ECG was inversely associated with late but not early bleeding.
Major Bleeding and Hospital Outcomes
Patients who experienced a major bleeding episode were more likely to die in hospital than those who did not bleed (20.9% versus 5.6%; P<0.001). One in 5 patients with a major bleed did not survive until hospital discharge; these patients accounted for ≈10% of all hospital deaths observed.
After the previously described demographic and clinical variables were controlled for, patients who had major bleeding during hospitalization were significantly more likely to die in hospital (hazard ratio [HR], 1.9; 95% confidence interval [CI], 1.6 to 2.2) than those who did not experience major bleeding. Major bleeding was not associated with subsequent recurrence of AMI (HR, 1.1; 95% CI, 0.9 to 1.4) or nonhemorrhagic stroke (HR, 1.8; 95% CI, 0.8 to 3.7) during hospitalization after controlling for potential confounders.
Major Bleeding, Fibrinolytics, and Hospital Outcomes
Of 5931 patients treated with fibrinolytic therapy, 3.1% suffered major bleeding. After adjustment for potential confounders, patients who received fibrinolytic therapy and bled were at significantly increased risk of in-hospital death (HR, 3.3; 95% CI, 2.3 to 4.7) compared with patients who had received fibrinolytic therapy and did not bleed. Of 26 043 patients who did not receive fibrinolytic therapy, 2.9% suffered major bleeding. After regression analysis, these patients also were at significantly increased risk of hospital death (HR, 1.7; 95% CI, 1.4 to 2.0) compared with their counterparts who did not bleed.
Major Bleeding and Postdischarge Outcomes
Patients who survived major bleeding in hospital were more likely to die during postdischarge follow-up than patients without major bleeding (7.9% versus 5.2%; P=0.002). However, after the previously described demographic and clinical variables were controlled for, hospital survivors of a major bleeding episode did not have an increased risk of 6-month mortality (HR, 0.8; 95% CI, 0.6 to 1.0). Similarly, major bleeding in hospital was not significantly associated with postdischarge mortality in patients who received fibrinolytic therapy (HR, 0.6; 95% CI, 0.2 to 1.6) or in those who did not receive fibrinolytic therapy (HR, 0.8; 95% CI, 0.6 to 1.1).
Bleeding, Treatment, and Outcomes
We identified 506 patients who suffered major bleeding on or the day after hospital admission for AMI and had received at least 1 antithrombotic therapy. Overall use of therapy after day 1 in patients who bled compared with those who never bled was as follows: aspirin, 69% versus 86%; thienopyridines, 51% versus 52%; unfractionated heparin, 29% versus 35%; low–molecular-weight heparin, 27% versus 50%; or glycoprotein IIb/IIIa inhibitors, 9% versus 14%.
Among patients who suffered major bleeding within the first hospital day, mortality rates were higher among those who discontinued aspirin, thienopyridines, or unfractionated heparin compared with those patients who bled but continued to be treated with these agents after the first day (52% versus 13%, P≤0.001; 58% versus 13%, P<0.001; 26% versus 16%, P=0.03, respectively) (Figure 2). The corresponding odds ratios for mortality associated with discontinuation of each agent were as follows: aspirin, 7.55 (95% CI, 4.43 to 12.88); thienopyridines, 8.91 (95% CI, 4.39 to 18.12); and unfractionated heparin, 1.91 (95% CI, 1.09 to 3.36).
In this large observational registry, major bleeding occurred in 2.8% of all patients hospitalized with AMI. Although major bleeding occurred in only 1 in 35 patients, ≈10% of deaths in our entire study population occurred in this high-risk patient subset. Our data provide insight into the clinical significance of major bleeding in patients with AMI in the real world, which patients are at risk, and the magnitude of associated morbidity and mortality.
Risk Factors for Bleeding
Similar to the results of an earlier analysis from GRACE, advancing age, female sex, and altered renal function were strongly associated with major bleeding in our study.8 Although not assessed in several other studies of ACS-associated bleeding,1,2 prior history of bleeding was a potent predictor of in-hospital bleeding in our study. These data suggest that information about history of bleeding should be collected routinely from every patient presenting with AMI and must be seriously weighed by physicians formulating a management plan. Other strong predictors of major bleeding were variables reflecting more severe illness (impaired renal function on presentation, use of intravenous vasopressor agents) or use of invasive procedures for severe illness (intra-aortic balloon pump, pulmonary artery catheter).
A recent randomized trial of antithrombotic therapy in ACS suggests that improvements in patient survival may be achieved by minimizing bleeding complications.9 Therefore, we were interested in exploring the possible associations between various antithrombotic and antiplatelet agents and bleeding in a nonrandomized trial setting. Although treatment with glycoprotein IIb/IIIa inhibitors in the first 24 hours of hospitalization was linked to the development of major bleeding, there was an inverse association between treatment with aspirin, unfractionated heparin, or low-molecular-weight heparins in the first 24 hours of hospitalization and subsequent major in-hospital bleeding events. This is in contrast to a recent analysis of the combined Organization to Assess Strategies for Ischemic Syndromes (OASIS-2) study and registry and Clopidogrel in Unstable Angina to Prevent Recurrent Events (CURE) study in which unfractionated heparin, low-molecular-weight heparin, and hirudin were associated with bleeding.1 These data suggest that in the “real-world” setting, physicians do a reasonably good job of selecting patients at low risk for bleeding when deciding to administer these agents. It also suggests that at least in the community setting, type of anticoagulant therapy may have a smaller impact on bleeding than the characteristics of the patients in whom they are used (eg, comorbidities, severity of illness) and whether invasive procedures are performed.
Unfortunately, we were not able to examine the impact of dosing of antithrombotic therapies on the risk of bleeding. A previous analysis from Can Rapid Risk Stratification of Unstable Angina Patients Suppress Adverse Outcomes With Early Implementation of the ACC/AHA Guidelines (CRUSADE) suggests that patients with non-STEMI treated with excess doses of low-molecular-weight heparin or glycoprotein IIb/IIIa inhibitors were at increased risk for major bleeding.10 Similarly, in an analysis from the CURE study, bleeding risks increased with increasing aspirin dose, with or without clopidogrel, without any increase in efficacy.11
Bleeding and Hospital Outcomes
One in every 5 patients who suffered major bleeding in our study did not survive to hospital discharge. This is even higher than the mortality rate observed in the aforementioned analysis of OASIS/CURE patients in which ≈1 in every 8 patients who bled had died at 30 days after study enrollment.1 This increase in mortality likely reflects the older age and the higher prevalence of specific comorbidities (eg, diabetes mellitus, renal dysfunction, and congestive heart failure) in our population.
It should be noted that after adjustment for demographic, clinical, and treatment characteristics, the HR for in-hospital mortality associated with bleeding in our study was much lower (1.9) compared with the HR for 30-day mortality in the OASIS/CURE study of 5.4.1 The HR for 30-day death rates associated with moderate or severe bleeding in the Rao et al2 analysis of the impact of bleeding in the pooled results of 4 randomized controlled trials (Global Use of Strategies To Open Occluded Coronary Arteries [GUSTO IIb], Platelet Glycoprotein IIb/IIIa in Unstable Angina: Receptor Suppression Using Integrilin Therapy [PURSUIT], and Platelet IIb/IIIa Antagonism for the Reduction of Acute coronary syndrome events in a Global Organization Network [PARAGON] A and B) also was higher than observed in our study (2.1 and 7.5, respectively). This difference is even more striking when one considers that neither of these studies included patients who received fibrinolytic therapy. The difference in adjusted HRs between our study and those deriving data from clinical trials reflects in part the higher unadjusted death rates observed in nonbleeding patients in our observational study. Indeed, the in-hospital mortality of patients who did not bleed in our study was 5.6% compared with 30-day mortality rates of only 2.5% and 2.9% in OASIS/CURE and the Rao et al analyses, respectively. We do not believe that having different definitions of major bleeding was a likely cause for the observed differences in adjusted bleeding risk; in each of the 3 studies, bleeding prompting blood transfusion was the lowest threshold for categorization of moderate or major bleeding. In addition, absolute mortality associated with major bleeding was higher in our study than observed in either of the other 2 studies.
Differences in the patient profile of our patients compared with those enrolled in clinical trials and use of a different mortality time point (in-hospital versus 30-day) may partially explain the differences in the impact of bleeding on death rates in our study compared with prior investigations. Nevertheless, our data suggest that in the community setting, bleeding, although still associated with worse outcomes, also is a marker for patients with more severe illness or comorbidities.
Bleeding and Postdischarge Outcomes
It has been suggested that the occurrence of major bleeding in hospital continues to affect mortality even after hospital discharge. In the Canadian Acute Coronary Syndrome Registry, major bleeding was a predictor of 1-year mortality.12 In the aforementioned report by Rao et al,2 mild, moderate, and severe bleeding remained significantly associated with 1-year mortality (HR, 1.4, 2.1, and 7.5, respectively). However, both of these studies were cumulative analyses including early in-hospital deaths. In the OASIS/CURE analysis, major bleeding remained associated with an increased hazard of death after 30 days after the exclusion of patients who died before 30 days.1 In contrast, although patients in our study who suffered major bleeding experienced higher postdischarge mortality (7.9% versus 5.2%), bleeding was no longer a predictor of late mortality after controlling for differences in clinical and treatment characteristics. These findings are concordant with that of a meta-analysis of 25 studies of ACS in which the impact of bleeding on mortality was confined to in-hospital or 30-day mortality.13
Bleeding and Mortality: Potential Mechanisms
Although the GRACE registry was not specifically designed to explore the issue of bleeding and increased mortality in patients with AMI, available data allow us to make some observations and to generate hypotheses about possible mechanisms involved. First, as noted previously, patients who bled were older, sicker, and more likely to undergo invasive procedures. Although we attempted to control for these and other variables, it is still likely that in some patients bleeding was merely a marker for severe illness and did not contribute directly to their death.
That said, bleeding remained associated with in-hospital mortality after controlling for important confounders. Mechanisms by which bleeding may directly lead to death in patients with a myocardial infarction warrant further study. One potential mechanism includes the premature discontinuation of antiplatelet and antithrombotic therapies in patients who bleed. Clearly, early cessation of these therapies in patients with ACS, particularly those who may have undergone percutaneous coronary intervention with stent placement, is problematic. Although we were able to explore this mechanism in only a limited fashion, our data suggest that this may be an important issue. Institution or continuation of all antiplatelet and/or antithrombotic therapies assessed after day 1 was lower in patients who experienced early bleeding (within 24 hours). Moreover, in patients who bled, hospital mortality was significantly increased in those who discontinued aspirin, thienopyridines, or low-molecular-weight heparins compared with those who continued these agents after the first day of hospitalization.
Given the proven efficacy of these agents for the prevention of recurrent ischemic events (particularly in patients undergoing stent implantation), the inability to use them would be expected to have adverse sequelae. Unfortunately, our sample size for this subanalysis was not large enough to allow a more detailed analysis. It must also be acknowledged that antithrombotic therapy may have been discontinued more often in sicker patients with more serious bleeds. We are unable to directly explore the relationship between bleeding, avoidance of antiplatelet and antithrombotic medications, and adverse outcomes in this observational study; future observational studies of bleeding in ACS should be designed to examine these relationships.
Other postulated mechanisms by which major bleeding may lead to recurrent coronary ischemia and/or increased mortality might include the adverse effects of resulting hypotension on end organs, platelet and coagulation activation associated with anemia, or even adverse effects of the resulting blood transfusions themselves. Finally, interventions performed to manage bleeding complications (eg, exploratory surgery) likely result in increased risk. Although we could not fully explore these issues, we believe that a better understanding of the pathophysiological relationship between bleeding and subsequent mortality is critical and deserves further study.
Study Strengths and Limitations
The strengths of the present investigation include its large sample size, the amount of data collected for each subject, and the use of a uniform predetermined definition of major bleeding throughout the study. Another strength is that our study population is more reflective of patients seen in physicians’ actual practice than patients enrolled in randomized controlled trials.
However, several limitations should be borne in mind when our study findings are interpreted. Most important, we acknowledge that in this large-scale observational study, we cannot exclude “underreporting” of bleeding events. That said, the observed rate of major bleeding in our study was slightly more than that observed in the aforementioned OASIS/CURE analysis. Obviously, no firm conclusions about a causal relationship between hospital bleeding and hospital outcomes can be made on the basis of this analysis. Because we did not stratify bleeding events by severity, we were unable to examine the relationship between increasing severity and subsequent outcomes. A number of prior analyses have demonstrated a consistent increase in the likelihood for dying with increasing bleed severity; the consistency of these findings lends strength to the validity of an association.1,2 Finally, we were unable to fully explore possible relationships between antithrombotic therapy (dosing, timing, and cessation), bleeding, and observed outcomes. Future observational studies specifically designed to address these associations are required.
This observational study of >40 000 patients with AMI suggests that major bleeding events occur in ≈1 in every 35 patients. One in 5 of these patients will not survive until hospital discharge; these patients accounted for ≈10% of all hospital deaths observed. Our data suggest that in the community setting, bleeding tends to occur in patients who are older and have more comorbidities or more severe illness and is a marker for poor hospital outcome. A better understanding of the specific causes of AMI-associated bleeding and the relationship between major bleeding events and death is needed if we are to decrease morbidity and mortality in this high-risk patient subset.
We thank the physicians and nurses who are participating in GRACE and David Hosmer, Jr, PhD, for his statistical expertise. We are grateful to Sophie Rushton-Smith, PhD, for editorial assistance in the preparation of this article.
Sources of Funding
Funding and sponsorship for GRACE are provided by sanofi-aventis (Paris, France). Sanofi-aventis had no involvement in the collection, analysis, and interpretation of data; in the writing of the report; and in the decision to submit the paper for publication. The corresponding author had complete access to the data for the preparation of this manuscript and had final responsibility for the decision to submit for publication. The design, conduct, and interpretation of the GRACE data are undertaken by an independent steering committee.
Dr Spencer has served on an advisory board and received a research grant from sanofi-aventis. Dr Moscucci has received grant support from Blue Cross Blue Shield of Michigan and has served as a consultant and on the speakers’ bureaus for Pfizer and the Medicine Company. Dr Granger has received research grants from Novartis, Proctor and Gamble, sanofi-aventis, The Medicines Company, Alexion, AstraZeneca, Boehringer Ingelheim, BMS, deCode Genetics, Genentech, and GSK. He has served as a consultant for Novartis, Proctor and Gamble, sanofi-aventis, The Medicines Company, Alexion, AstraZeneca, Genentech, GSK, INO Therapeutics, and Medicure. Dr Gore has received a research grant from sanofi-aventis. Dr Steg has received a research grant from sanofi-aventis; has served on the speakers’ bureaus for Boehringer Ingelheim, BMS, GSK, MSD, Novartis, Nycomed, sanofi-aventis, Sankyo, Servier, and ZLB-Behring; and has served on consulting or advisory boards for AstraZeneca, BMS, GSK, MSD, Pfizer, sanofi-aventis, Servier, and Takeda. Dr Goodman has received research grants or other research support from AstraZeneca, Bayer, Biovail, Boehringer Ingelheim, Bristol Myers Squibb, Eli Lilly, GlaxoSmithKline, Guidant, Hoffman La-Roche, Johnson & Johnson, Key Schering/Schering Plough, Merck Frosst, Pfizer, sanofi-aventis, and The Medicines Company; has received honoraria from AstraZeneca, Boehringer Ingelheim, Bristol Myers Squibb, Hoffman La-Roche, Key Schering/Schering Plough, Merck Frosst, Pfizer, sanofi-aventis, and The Medicines Company; and has served as a consultant or on the advisory boards for Bristol Myers Squibb, GlaxoSmithKline, Hoffman La-Roche, and sanofi-aventis. Dr Budaj has received research grants and honoraria from sanofi-aventis, GSK, AstraZeneca, Boehringer Ingelheim and has served as a consultant or on the advisory boards for sanofi-aventis and GSK. Dr Fox has received research grants and honoraria and served as a consultant or on the advisory boards for sanofi-aventis, BMS, and GSK. The other authors report no conflicts.
Eikelboom JW, Mehta SR, Anand SS, Xie C, Fox KA, Yusuf S. Adverse impact of bleeding on prognosis in patients with acute coronary syndromes. Circulation. 2006; 114: 774–782.
Steg PG, Goldberg RJ, Gore JM, Fox KA, Eagle KA, Flather MD, Sadiq I, Kasper R, Rushton-Mellor SK, Anderson FA. Baseline characteristics, management practices, and in-hospital outcomes of patients hospitalized with acute coronary syndromes in the Global Registry of Acute Coronary Events (GRACE). Am J Cardiol. 2002; 90: 358–363.
Eagle KA, Lim MJ, Dabbous OH, Pieper KS, Goldberg RJ, Van de Werf F, Goodman SG, Granger CB, Steg PG, Gore JM, Budaj A, Avezum A, Flather MD, Fox KA. A validated prediction model for all forms of acute coronary syndrome: estimating the risk of 6-month postdischarge death in an international registry. JAMA. 2004; 291: 2727–2733.
Hosmer DW, Lemeshow S. Applied Survival Analysis. New York, NY: Wiley and Sons Inc; 1999.
Moscucci M, Fox KA, Cannon CP, Klein W, Lopez-Sendon J, Montalescot G, White K, Goldberg RJ. Predictors of major bleeding in acute coronary syndromes: the Global Registry of Acute Coronary Events (GRACE). Eur Heart J. 2003; 24: 1815–1823.
Peters RJG, Mehta SR, Fox KAA, Zhao F, Lewis BS, Kopecky SL, Diaz R, Commerford PJ, Valentin V, Yusuf S, for the CURE Trial Investigators. Effects of aspirin dose when used alone or in combination with clopidogrel in patients with acute coronary syndromes: observations from the Clopidogrel in Unstable Angina to Prevent Recurrent Events (CURE) Study. Circulation. 2003; 108: 1682–1687.
Analyses from randomized controlled trials suggest that bleeding in patients with acute myocardial infarction is strongly associated with poor outcomes. Because these data are not generalizable to all patients with acute myocardial infarction, we sought to understand better the scope of this problem in a “real-world” setting. Using data from >40 000 patients enrolled in the Global Registry of Acute Coronary Events, we found that 2.8% of patients with acute myocardial infarction experienced major bleeding during their hospital stay. These patients were older, more severely ill, and more likely to undergo invasive procedures. These patients accounted for ≈10% of all hospital deaths. Continuation of antithrombotic therapies after day 1 was lower in patients who experienced early bleeding. Moreover, in patients who bled, hospital mortality was increased in those who discontinued aspirin, thienopyridines, or low–molecular-weight heparins. Nevertheless, after adjustment for differences in age and comorbidities between patients with and without major bleeding, the risk of hospital mortality associated with bleeding was much lower than reported in randomized controlled trials. These data suggest that although bleeding may be causally related to adverse outcomes in some patients in the real-world setting, it is often merely a marker for patients at higher risk for adverse outcomes. The effect of antithrombotic therapy discontinuation after major bleeding in patients with acute myocardial infarction requires further study.