Patterns of discharge antiplatelet therapy and late outcomes among 8,582 patients with bleeding during acute coronary syndrome: A pooled analysis from PURSUIT, PARAGON-A, PARAGON-B, and SYNERGY
Article Outline
- Abstract
- Methods
- Results
- Discussion
- Conclusions
- Disclosures
- Acknowledgements
- Appendix. Supplemental methods and data
- References
- Copyright
Background
Major bleeding during an acute coronary syndrome (ACS) is associated with increased late ischemic events. Patients with bleeding are often discharged without antiplatelet therapy (AT). The association between discharge AT use and late ischemic outcomes among ACS patients with bleeding is uncertain.
Methods
We examined discharge AT use among 8,582 ACS patients with in-hospital bleeding from a total of 26,451 patients enrolled in 4 randomized trials. After adjusting for the propensity to receive AT, we compared 6-month postdischarge outcomes between patients discharged with and those discharged without AT.
Results
Almost 1 in 10 patients with bleeding was discharged without AT (n = 826). Compared with those receiving discharge AT, those not receiving discharge AT had a higher risk of 6-month death, myocardial infarction, and stroke (14.3% vs 7.8%, propensity-adjusted hazard ratio [HR] = 1.36, 95% confidence interval = 1.01-1.85). Nonuse of AT at discharge was associated with worse outcomes among patients treated with percutaneous coronary intervention compared with those treated without it (adjusted HR = 4.22 vs 1.13, interaction P = .0003). Discharge monotherapy was associated with worse outcomes than dual AT among patients receiving stents (adjusted HR = 1.78, 95% CI = 1.04-3.03).
Conclusions
Bleeding occurred commonly among patients with ACS. AT was often not used in these patients at discharge, and lack of discharge AT was associated with an increased risk of 6-month ischemic events. These data raise the possibility that lack of AT use among patients with in-hospital bleeding may contribute to their excess risk of long-term ischemic outcomes.
Among patients hospitalized for acute coronary syndrome (ACS), approximately 30% experience bleeding events of varying severity during hospitalization.1, 2, 3 Major bleeding during an ACS is strongly associated with increased mortality and recurrent myocardial infarction (MI) up to 6 months after the initial bleeding event.4, 5, 6, 7 However, factors leading to the increased risk of future ischemic events in these patients are poorly understood. Antiplatelet therapy (AT) is often discontinued after bleeding, and many patients with bleeding are likely to be discharged without these agents.8 Data on the association between the avoidance of discharge AT and late ischemic outcomes among ACS patients with bleeding are limited.
We investigated the association between discharge AT and late clinical outcomes among ACS patients who had experienced bleeding during their index hospitalization. We hypothesized that among patients with bleeding, avoidance of discharge AT would be associated with higher risks of postdischarge ischemic outcomes compared with patients discharged on AT. We also sought to investigate whether this risk varied with the intensity of discharge AT and whether or not patients had undergone percutaneous coronary intervention (PCI) during their index hospitalization.
Methods
Data sources and study population
We merged patient-level data from 4 phase III non–ST-elevation ACS clinical trials of antithrombotic therapy: Platelet Glycoprotein IIb/IIIa in Unstable Angina, Receptor Suppression Using Integrilin Therapy (PURSUIT), Platelet IIb/IIIa Antagonism for the Reduction of Acute coronary syndrome events in a Global Organization Network (PARAGON-A), PARAGON-B, and Superior Yield of the New Strategy of Enoxaparin, Revascularization, and Glycoprotein IIb/IIIa Inhibitors (SYNERGY). The study designs and results of these trials have been previously published9, 10, 11, 12 and are described in the supplemental data in the online Appendix.
We included all patients from the merged data set who were alive at the time of discharge from their index ACS hospitalization. We excluded all patients discharged on warfarin and patients with incomplete data on bleeding, discharge AT, and 6-month outcomes (Figure 1).
Figure 1.
Patient flow. Single AT, aspirin or thienopyridine; dual AT, aspirin and thienopyridine.
Our study complies with the Declaration of Helsinki, and the Duke University Medical Center Institutional Review Board approved the study with a waiver of the requirement for written informed consent.
Study definitions
Bleeding was defined according to the Global Utilization of Streptokinase and Tissue Plasminogen Activator for Occluded Coronary Arteries (GUSTO) bleeding scale in the PURSUIT and SYNERGY trials12, 13: severe or life-threatening bleeding was defined as either intracranial hemorrhage or bleeding that caused hemodynamic compromise and required intervention, moderate bleeding was defined as bleeding that required blood transfusion but did not result in hemodynamic compromise, and mild bleeding was any other bleeding. The PARAGON-A and PARAGON-B trials adopted a modified GUSTO definition9, 10 as follows: major/life-threatening bleeding was defined as intracranial hemorrhage or bleeding leading to hemodynamic compromise requiring intervention and intermediate bleeding was defined as bleeding requiring transfusion, a decrease in hemoglobin ≥5 g/dL, or a decrease in hematocrit ≥15%. For this analysis, the major/life-threatening and intermediate bleeding episodes in PARAGON-A and PARAGON-B were considered equivalent to GUSTO severe and moderate bleeding, respectively.
Patients were considered to be on AT at discharge if the following medications were reported on their case report form at the time of discharge: aspirin and/or a thienopyridine, either ticlopidine or clopidogrel. Among those discharged with AT, we further subdivided them into those receiving single (aspirin or a thienopyridine) and dual (aspirin and a thienopyridine) AT (Figure 1).
MI was defined according to the protocol of each trial.9, 10, 12, 13 All endpoints, including strokes, were adjudicated by an independent, blinded events committee.
The main outcome of interest for this analysis was the composite of death, MI, or stroke between discharge and 6 months postrandomization.
Statistical analysis
All statistical analyses were performed using SAS version 9.2 (SAS, Cary, NC). All P values were 2-tailed, with α set at .05 and all confidence intervals (CIs) calculated to the 95th percentile. We summarized baseline characteristics using percentages for categorical variables and medians and interquartile ranges for continuous variables. For comparisons, we used the Pearson χ2 test for categorical variables, the 2-sample t-test for normally distributed continuous variables, and the Wilcoxon rank-sum test for non-normally distributed data. We summarized in-hospital procedural use and nonfatal events using the cumulative incidence function and observed mortality using Kaplan-Meier estimates.
We estimated the unadjusted relationship between treatment and outcome using a Cox proportional hazards model in which treatment group was the only variable. We then estimated the adjusted relationship between treatment and outcome by entering covariates, selected based on their relevance to post-ACS and postbleeding outcomes, in the proportional hazards model. Because outcomes were measured from discharge, we included the performance of postrandomization procedures, including catheterization, PCI, and coronary artery bypass grafting (CABG), and the occurrence of postrandomization nonfatal adverse events during the index hospitalization as time-updated covariates in the outcome models, as previously described.14 We used the methodology proposed by Lin et al15 to assess the proportional hazards assumption, whereas the assumption of linearity for continuous variables was tested using restricted cubic splines. When the relationship was found to be nonlinear, the data were transformed with piecewise linear splines as previously described.16
Because patients were not randomized to receive discharge AT, we used propensity scores to account for medication selection bias. We used logistic regression to estimate the propensity to receive any AT at discharge (vs no AT) as a function of 26 independent variables based on admission clinical characteristics, in-hospital interventions, and in-hospital nonfatal events. To further account for the heterogeneity of prescription patterns in each of the 4 trials, we factored clinical trial association into the propensity model as 2 bivariate independent variables, using the most contemporary trial (SYNERGY) as the reference group.
Propensity scores were then calculated for each patient and included as a covariate in the outcome models. We prespecified this propensity-score–adjusted approach to strike a balance between study power and the distribution of treatment selection variables between treatment comparison groups. Because we expected a relatively small number of patients to have been discharged without AT compared with patients discharged with AT, the more commonly used propensity-matched analysis may result in substantial information loss with the limited number of matched pairs available for analysis. Further analytical methods are described in Methods and the online Appendix.
This analysis was internally funded by the Duke Clinical Research Institute (Durham, NC). The PURSUIT trial was funded by COR Therapeutics and Schering-Plough Research Institute, the PARAGON-A and PARAGON-B trials were funded by Hoffman La-Roche (Basel, Switzerland), and the SYNERGY trial was funded by Sanofi-Aventis (Paris, France). We, the authors, are solely responsible for the design and conduct of this study, all study analyses, the drafting and editing of the manuscript, and its final contents.
Results
In-hospital bleeding population
In total, 26,451 subjects were discharged alive and without warfarin after their index hospitalization. Of these, 8,582 (32.4%) experienced an in-hospital bleeding event (Figure 1), with 5,369 (62.6%) experiencing GUSTO mild bleeding and 3,213 (37.4%) having GUSTO moderate/severe bleeding. Compared with patients without bleeding, patients with an in-hospital bleeding event had more comorbid conditions and experienced more adverse events (online Appendix).
AT use among patients with bleeding
Among patients who experienced an in-hospital bleeding event, 9.6% were discharged without any AT; this occurred more frequently among patients with GUSTO moderate to severe bleeding than among patients with GUSTO mild bleeding (16.5% vs 5.5%, P < .001). Although patients discharged without AT were more likely to have baseline comorbidities and more adverse presenting characteristics, their Global Registry of Acute Coronary Event scores were similar to those of patients discharged with AT (Table I). Patients discharged with AT were more likely to have undergone PCI, whereas patients discharged without AT were more likely to have undergone CABG and experienced more in-hospital nonfatal events (Table II). Individual trial data are presented in the online Appendix.
Table I. Baseline characteristics
| Patients with in-hospital bleeding (N = 8573)⁎ | ||||
|---|---|---|---|---|
| No discharge AT (n = 826) | AT received on discharge | P value | ||
| Single: ASA or thieno (n = 4904) | Dual: ASA and thieno (n = 2943) | |||
| Demographics | ||||
 Age (y) | 71 (64, 77) | 67 (59, 73) | 67 (60, 74) | <.0001 |
| Women (%) | 40.1 | 37.5 | 37.1 | .292 |
| Body weight (kg) | 76 (65, 87) | 78 (68, 89) | 79 (70, 90) | <.0001 |
| White (%) | 86.1 | 87.6 | 87.7 | .412 |
| North American (%) | 60.4 | 57.1 | 73.4 | <.0001 |
| Medical history | ||||
| Hypertension (%) | 69.1 | 61.9 | 64.9 | <.0001 |
| Diabetes mellitus (%) | 30.3 | 26.2 | 27.1 | .054 |
| Current smoker (%) | 20.4 | 23.4 | 23.8 | .117 |
| Prior MI (%) | 34.4 | 30.3 | 27.0 | <.0001 |
| Prior PCI (%) | 13.2 | 14.1 | 21.4 | <.0001 |
| Prior CABG (%) | 14.3 | 10.9 | 16.0 | <.0001 |
| Prior heart failure (%) | 17.8 | 10.6 | 7.7 | <.0001 |
| Renal insufficiency (%)† | 25.3 | 16.5 | 31.0 | <.0001 |
| Presenting features | ||||
| Heart rate (min−1) | 74 (65, 85.5) | 72 (64, 83) | 71 (62, 80) | <.0001 |
| Systolic blood pressure (mm Hg)‡ | 130 (117, 146) | 132 (120, 150) | 132 (118, 150) | .018 |
| Baseline hematocrit (%)‡ | 40 (36.2, 43.3) | 41 (37.7, 44) | 41 (37.9, 44) | <.0001 |
| Baseline creatinine clearance (mL/min)‡ | 61.8 (46.3, 84.1) | 73.2 (55.6, 94.2) | 75.2 (56.7, 98.4) | <.001 |
| Killip class III/IV (%) | 3.3 | 2.0 | 2.1 | .047 |
| Global Registry of Acute Coronary Event score | 95 (90-97) | 96 (94-98) | 97 (95-98) | <.0001 |
⁎One or more episodes of GUSTO mild, moderate, or severe bleeding during hospitalization for non–ST-elevation ACS. |
†Defined as documented history of renal failure with baseline creatinine level >1.5 mg/dL. |
‡Expressed as median (25th, 75th percentile). |
Table II. In-hospital procedures and nonfatal events
| Patients with in-hospital bleeding (N = 8573)⁎ | ||||
|---|---|---|---|---|
| No discharge AT (n = 826) | AT received on discharge | P value | ||
| Single: ASA or thieno (n = 4904) | Dual: ASA and thieno (n = 2943) | |||
| In-hospital procedures† | ||||
| Coronary angiography (%) | 73.4 | 74.6 | 93.5 | <.0001 |
| PCI (%) | 13.7 | 14.2 | 75.6 | <.0001 |
| CABG (%) | 41.8 | 45.4 | 9.1 | <.0001 |
| In-hospital nonfatal events‡ | ||||
| MI (%) | 29.7 | 16.3 | 12.5 | <.0001 |
| Recurrent ischemia (%) | 24.3 | 16.3 | 9.3 | <.0001 |
| Stroke (%) | 3.9 | 1.0 | 0.6 | <.0001 |
⁎One or more episodes of GUSTO mild, moderate, or severe bleeding during hospitalization for non–ST-elevation ACS. |
†Procedures performed prior to discharge from the index hospitalization. |
‡Nonfatal events occurring prior to discharge. |
Factors independently associated with lack of antiplatelet use at discharge are shown in Table III. We included 8,355 patients in the logistic regression model predicting discharge without AT (number of events = 806; model c index = 0.79). Of the 26 predictors tested, 14 were independently predictive.
Table III. Independent predictors of discharge without AT among patients with in-hospital bleeding
| Predictor | Wald χ2 | Adjusted odds ratio (95% CI) | P value |
|---|---|---|---|
| In-hospital heart failure | 92.6 | 2.81 (2.27-3.46) | <.0001 |
| In-hospital MI | 52.2 | 2.02 (1.67-2.44) | <.0001 |
| SYNERGY versus PARAGON-A and PARAGON-B⁎ | 51.5 | 4.00 (2.70-5.88) | <.0001 |
| PURSUIT versus SYNERGY† | 35.3 | 2.07 (1.63-2.64) | <.0001 |
| In-hospital stroke | 30.3 | 3.92 (2.41-6.37) | <.0001 |
| Age per 10-y increase above 55 y‡ | 29.9 | 1.35 (1.21-1.50) | <.0001 |
| No in-hospital PCI | 20.2 | 2.17 (1.54-2.94) | <.0001 |
| No in-hospital stent implantation | 18.9 | 2.44 (1.64-3.70) | <.0001 |
| Hemoglobin per 1-g/dL decrease below 15 g/dL‡ | 7.3 | 1.08 (1.02-1.14) | .007 |
| Weight per 10-kg decrease below 75 kg‡ | 6.3 | 1.15 (1.03-1.29) | .01 |
| Chronic renal insufficiency | 4.3 | 1.29 (1.01-1.64) | .04 |
| Heart rate per 10-U increase above 70 beats/min‡ | 4.2 | 1.07 (1.00-1.15) | .04 |
| Prior heart failure | 4.2 | 1.28 (1.01-1.61) | .04 |
| No prior stroke | 4.1 | 1.35 (1.01-1.82) | .04 |
⁎Adjusted probability of discharge without AT in SYNERGY compared with the PARAGON trials. |
†Adjusted probability of discharge without AT in PURSUIT compared with SYNERGY. |
‡Because the assumption of linearity was not met for these continuous variables, the data were transformed using piecewise linear splines separated by a threshold value. |
Outcomes
Among patients with bleeding, postdischarge adverse outcomes occurred more frequently in patients discharged without AT than in those discharged on AT (14.3% vs 7.8% and P < .0001 for 6-month death/MI/stroke, 7.8% vs 3.0% and P < .0001 for 6-month death). The unadjusted hazard ratios (HRs) comparing discharge without AT and discharge with AT were 2.12 (95% CI = 1.61-2.80) for death/MI/stroke and 2.16 (95% CI = 1.69-2.75) for death. These differences persisted after adjusting for covariates (HR = 1.64 [95% CI = 1.22-2.20] for death/MI/stroke and HR = 1.79 [95% CI = 1.22-2.62] for death) and the propensity to receive AT on discharge (HR = 1.36 [95% CI = 1.01-1.85] for death/MI/stroke and HR = 1.72 [95% CI = 1.36-2.17] for death) (Figure 2).
Figure 2.
Cumulative 6-month outcomes among patients with bleeding who were discharged with and without AT. For both outcomes, log-rank P < .001. HR compares the risk of 6-month adverse outcomes among patients discharged without versus patients discharged with AT.
In the treatment effect-by-treatment intensity analysis, we noted a trend toward increasing event rates corresponding to lower intensity of discharge AT (Figure 3). The adjusted HR for aspirin monotherapy vs thienopyridine monotherapy did not achieve statistical significance (HR = 1.09 [95% CI = 0.059-2.03] for death, MI, or stroke and HR = 1.18 [95% CI = 0.53-2.60] for death).
Figure 3.
Cumulative 6-month outcomes as a function of the intensity of discharge AT among patients with bleeding. For both outcomes, log-rank P < .001. ASA indicates aspirin; thieno, thienopyridine.
Resumption of AT after discharge
To further understand how discharge without AT might influence 6-month outcomes in patients with bleeding, we analyzed 30-day postdischarge use of AT among patients enrolled in SYNERGY, the most contemporary of the 4 clinical trials. Among patients with bleeding who were discharged without AT, AT was prescribed less frequently at 30 days compared with patients with bleeding who were discharged on AT (Figure 4).
Figure 4.
Resumption of AT at 30 days. Data were only available for patients with bleeding in SYNERGY.
Bleeding, transfusion, and discharge AT
Because blood transfusion has been shown to be an independent predictor of future ischemic events,17 we further investigated the influence of transfusion on the association between discharge AT and the primary endpoint. Of the 8,582 patients who bled, 2,944 (34.3%) received one or more blood transfusions. Patients who bled and received a transfusion were less likely to receive discharge AT (32.1% vs 67.9%, P < .001) and were more likely to experience death (4.0% vs 2.9%, P = .015, unadjusted HR = 1.27 [95% CI = 1.06-1.52]) and death/MI/stroke (9.6% vs 7.6%, P = .011, unadjusted HR = 1.39 [95% CI = 1.08-1.78]) at 6 months, compared with patients who bled but did not receive a transfusion. In a proportional hazards model that adjusted for whether or not patients received a transfusion, discharge without AT remained an independent predictor of both death/MI/stroke (P = .005) and death (P = .006), but transfusion was an independent predictor of neither endpoint (P = .61 and P = .66, respectively).
Subgroup analyses
A total of 2,957 patients with bleeding underwent PCI, of whom 2,228 patients received stents and 729 patients did not receive stents. Discharge without AT was associated with a 4-fold greater risk of postdischarge adverse outcomes among patients who underwent in-hospital PCI than among patients who did not undergo PCI (interaction P = .0003) (Table IV). Among all patients who underwent PCI, there was no significant interaction effect between stent implantation and discharge AT on the composite endpoint (interaction P = .589). Among patients who received stents during PCI, those discharged on single AT experienced the composite primary endpoint more often than those discharged on dual AT (adjusted HR = 1.78 [95% CI = 1.04-3.03] for death/MI/stroke and adjusted HR = 2.38 [95% CI = 1.17-4.82] for death) (Figure 5).
Table IV. Six-month adjusted outcomes of patients discharged without versus those discharged with AT in prespecified subgroups
 |
Figure 5.
Monotherapy versus dual therapy among stented and non-stented patients.
Among patients undergoing in-hospital CABG, discharge without AT was associated with an adjusted HR of 2.04 (95% CI = 3.57-1.17) for death/MI/stroke and that of 3.10 (95% CI = 6.58-1.47) for death, compared with discharge with AT. Unlike PCI, the association was not significantly different from patients who did not undergo in-hospital CABG (interaction P = .111 for death/MI stroke and interaction P = .106 for death).
Discussion
Our study shows that bleeding was common among patients hospitalized with non–ST-elevation ACS; approximately 1 in 10 patients with in-hospital bleeding was discharged without AT. Even after accounting for differences in baseline risk and the propensity to receive AT, the lack of AT use at discharge was associated with significantly increased risks of death and the composite endpoint of death, MI, or stroke at 6 months, compared with patients who received single or dual AT at discharge. The association of nonuse of discharge AT with adverse risk was greatest among patients undergoing PCI.
Although a bleeding event puts patients at acute as well as long-term risk for adverse ischemic outcomes,7 the reasons for the increased long-term risk are not fully clarified. A failure to prescribe AT among those with bleeding has been postulated as a factor contributing to this increased risk.18 Our data lend credence to this hypothesis and merit further research to confirm these findings.
In the multivariable analysis, discharge AT was retained as an independent predictor of 6-month outcomes, but blood transfusion was not. Rao et al17 found a strong association between transfusion during an ACS and 30-day ischemic outcomes; however, a more recent study showed that the adverse effects associated with transfusion were markedly attenuated when patients were followed up for 90 days.19 More research is needed to investigate the possibility that although transfusion may contribute toward the early hazard associated with bleeding, factors such as avoidance of AT may have a greater influence on late ischemic outcomes.
Although the temporary interruption of AT is a rational response to active bleeding, it is unclear when AT should be resumed after resolution of bleeding. Despite practice guidelines recommending either continuing AT in patients with mild bleeding or restarting therapy as soon as possible in patients with more severe degrees of bleeding (class I recommendation),20 these recommendations are supported only by expert consensus (level of evidence C). Given the importance of AT in the secondary prevention of ACS,20, 21 patients with bleeding during ACS may benefit from earlier postdischarge visits to enable more timely reinitiation of AT or to monitor for recurrent bleeding after reinitiation of AT.
In current practice, however, AT is withheld long after discharge from the hospitalization associated with the initial bleeding event, as observed in the contemporary Prospective Registry Evaluating Myocardial Infarction: Events and Recovery (PREMIER) study (2003-2004), in which 17% of patients with bleeding were discharged without aspirin and 41% were discharged without a thienopyridine.8 This discrepancy in antiplatelet use persisted up to 6 months post-MI, a trend that was not observed with other non–platelet-directed medications such as β-blockers and statins. This finding may be attributed to the lack of evidence supporting the efficacy and safety of early AT reinitiation in patients who suffer from in-hospital bleeding. Because the sample size of the bleeding population in PREMIER was too small to investigate the association between discharge AT and outcomes, the present study extends this important finding by showing, in a much larger bleeding cohort, that withholding AT at discharge is associated with an increased risk of downstream adverse clinical outcomes.
Another concern is the intensity at which AT should be restarted after resolution of bleeding. Our findings are consistent with those seen in the Clopidogrel in Unstable Angina to Prevent Recurrent Ischemic Events randomized trial, in which prolonged dual AT achieved a much greater relative risk reduction in death or MI of 31% in the PCI population,22 compared with only 18% in the overall study population.23 As in our study, the majority of patients undergoing PCI in CURE received stents (2,172 of 2,658).22 This raises the hypothesis that although monotherapy may be an acceptable compromise for the majority of patients with ACS who experience bleeding, every possible effort should be made, if feasible, to initiate dual AT at or soon after discharge in those who have had stents implanted. The use of newer, more potent P2Y12 receptor antagonists, with their greater propensity to cause bleeding, will likely introduce further challenges to reinitiating dual AT among patients receiving stents. A reasonable strategy in patients who experience bleeding with the more potent P2Y12 receptor antagonists, such as prasugrel or ticagrelor, may be to reinitiate therapy with a lower dose of these agents or switch to clopidogrel, although there are presently no clinical studies to support this recommendation.
Our study has several limitions: First, our analysis was retrospective and observational in nature. Thus, our results may represent statistical chance rather than true biological interactions and need cautious interpretation and further investigation. Second, recurrent bleeding outcomes were not captured, and thus the safety of early AT resumption after bleeding cannot be examined. However, regulatory agencies support the concept that although many bleeds are reversible and nonfatal, what matters are the less common bleeding events that result in severe consequences such as mortality, MI, and stroke.24 Third, the heterogeneity of study populations and study periods in this pooled analysis may have influenced our findings; we adjusted as much as possible for these differences in our main effects and propensity models. Fourth, we studied highly selected patients from randomized trial populations. Therefore, the outcomes in our analysis likely underestimate real-world risks due to major differences in baseline comorbidities and risks.
Conclusions
Bleeding occurred commonly among patients with ACS. One in 10 of these patients did not receive AT at discharge. This lack of AT use among patients with bleeding was associated with increased risk of late ischemic events, particularly among those undergoing PCI. Reinitiation of AT in a monitored setting before or soon after hospital discharge in patients who experience bleeding, and with dual agents when feasible among those receiving stents, may have the potential to promote long-term anti-ischemic benefit while ensuring the safety of longitudinal continuation. More research is needed to determine the risk of recurrent bleeding versus the benefits for preventing ischemic events among patients who are restarted on AT after an initial bleeding event.
Disclosures
Disclosures for S.V.R., K.W.M., L.K.N., R.A.H., and E.D.P. can be found at http://www.dcri.duke.edu/research/coi.jsp. M.Y.C., R.D.L., M.E.J., and K.S.P. have nothing to declare.
Acknowledgements
We thank Dr Rajendran H. Mehta (Duke Clinical Research Institute, Durham, NC) for critical review of our manuscript and Charmian S. Choong (National University Healthcare System, Singapore, Singapore) for editorial assistance.
Appendix. Supplemental methods and data
Supplemental methods
To determine if resumption of AT after discharge was delayed among patients discharged without AT, we analyzed longitudinal postdischarge use of AT among patients enrolled in SYNERGY, the only trial for which these data were available. Patients were asked if they were still taking aspirin and/or a thienopyridine at their scheduled 30-day visit or by telephone contact. We then compared 30-day aspirin and thienopyridine use among patients who were initially discharged with versus those discharged without AT.
Because blood transfusion has been shown to be an independent predictor of future ischemic events, we further investigated the influence of transfusion on the association between discharge AT and the primary endpoint. We analyzed the independent effect of transfusion on the primary endpoint and then included transfusion in the multivariable model to determine if discharge without AT remained an independent predictor of the primary endpoint after adjusting for transfusion as a covariate.
We further assessed the heterogeneity of risk associated with the absence of discharge AT among 4 prespecified subgroups in a bivariate fashion: mild versus moderate to severe bleeding, in-hospital versus no in-hospital PCI, age <75 years versus age ≥75 years, and men versus women. Among patients undergoing PCI, we further investigated the differential effect of discharge AT among patients receiving stents versus patients not receiving stents. To determine the heterogeneity of risk, we introduced treatment-by-subgroup interaction terms into separate propensity score-adjusted Cox proportional hazards models.
Supplemental data
Baseline characteristics: no bleeding vs. bleeding
| All patients (N = 26 451) | |||
|---|---|---|---|
| No in-hospital bleeding (n = 17 869) | In-hospital bleeding (n = 8582)⁎ | P | |
| Demographics | |||
| Age (y) | 64 (55, 72) | 67.9 (60, 74) | <.001 |
| Female sex (%) | 33.1 | 37.6 | <.001 |
| Body weight (kg) | 79 (69, 89.9) | 78 (68, 89.1) | <.001 |
| White | 88.1 | 87.5 | .188 |
| North American | 43.8 | 62.8 | <.001 |
| Medical history | |||
| Hypertension | <.001 | ||
| Diabetes mellitus | 57.3 | 63.6 | <.001 |
| Current smoker | 29.0 | 23.2 | <.001 |
| Prior MI | 30.3 | 29.6 | .274 |
| Prior PCI | 15.2 | 16.4 | .015 |
| Prior CABG | 13.7 | 12.9 | .071 |
| Prior heart failure | 9.4 | 10.3 | .023 |
| Renal insufficiency | 14.2 | 22.1 | <.001 |
| Presenting features | |||
| Heart rate | 72 (62, 82) | 72 (63, 82) | .039 |
| Systolic blood pressure | 131 (120, 150) | 132 (118, 150) | .822 |
| Baseline hematocrit (%) | 41.5 (38.4, 44.3) | 40.9 (37.6, 43.9) | <.001 |
| Baseline creatinine clearance (mL/min) | 77.8 (59.4, 100) | 72.9 (55, 94.4) | <.001 |
| Killip class III/IV | 1.6 | 2.1 | .003 |
| In-hospital procedures | |||
| Coronary angiography | 62.9 | 80.7 | <.001 |
| PCI | 30.3 | 34.5 | <.001 |
| CABG | 3.2 | 33.0 | <.001 |
| In-hospital events | |||
| MI | 6.8 | 16.4 | <.001 |
| Recurrent ischemia | 11.3 | 14.7 | <.001 |
| Stroke | 0.2 | 1.1 | <.001 |
⁎Patients who experienced one or more episodes of GUSTO mild, moderate, or severe bleeding during the index hospitalization for acute coronary syndrome. |
Cumulative 6-month outcomes: bleeding versus no bleeding
Rates of death/MI/stroke at 6 months were higher among patients with in-hospital bleeding compared with patients without bleeding (6.9% vs 8.4%, P < .001, adjusted HR = 1.22, 95% CI = 1.08-1.37); a similar pattern was seen for 6-month mortality alone (3.3% vs 2.7%, P < .001, adjusted HR = 1.15, 95% CI = 0.97-1.37). As shown in Figure 1, event curves continued to diverge over time, implying a pattern of sustained event accrual over time beyond the index hospitalization. Adjusted HRs for 6-month death/MI/stroke were 1.17 (95% CI = 1.02-1.33) for GUSTO mild versus no bleeding and 1.39 (95% CI = 1.14-1.70) for GUSTO moderate/severe versus no bleeding.
Data across clinical trials
| PURSUIT (N = 10 655) | PARAGON-A (N = 2188) | PARAGON-B (N = 5126) | SYNERGY (N = 9838) | Total (N = 27 807) | |
|---|---|---|---|---|---|
| In-hospital GUSTO bleeding (%) | 29.5 | 13.3 | 31.8 | 40.1 | 32.9 |
| Discharge without AT among subjects with bleeding (%) | 15.2 | 0.0 | 3.0 | 8.8 | 9.6 |
| Six-month death/MI/stroke among subjects with bleeding treated with discharge AT (%) | 6.1 | 9.6 | 9.0 | 6.3 | 7.6 |
| Six-month death/MI/stroke among subjects with bleeding not treated with discharge AT (%) | 10.6 | 0.0 | 16.7 | 14.9 | 12.4 |
Outcomes: bleeding with transfusion versus bleeding without transfusion
References
- A comparison of the clinical impact of bleeding measured by two different classifications among patients with acute coronary syndromes. J Am Coll Cardiol. 2006;47:809–816
- Impact of clopidogrel in patients with acute coronary syndromes requiring coronary artery bypass surgery: a multicenter analysis. J Am Coll Cardiol. 2008;52:1693–1701
- Baseline risk of major bleeding in non–ST-segment-elevation myocardial infarction: the CRUSADE (Can Rapid risk stratification of Unstable angina patients Suppress ADverse outcomes with Early implementation of the ACC/AHA Guidelines) Bleeding Score. Circulation. 2009;119:1873–1882
- Adverse impact of bleeding on prognosis in patients with acute coronary syndromes. Circulation. 2006;114:774–782
- Predictors of major bleeding in acute coronary syndromes: the Global Registry of Acute Coronary Events (GRACE). Eur Heart J. 2003;24:1815–1823
- Associations of major bleeding and myocardial infarction with the incidence and timing of mortality in patients presenting with non–ST-elevation acute coronary syndromes: a risk model from the ACUITY trial. Eur Heart J. 2009;30:1457–1466
- Bleeding, blood transfusion, and increased mortality after percutaneous coronary intervention: implications for contemporary practice. J Am Coll Cardiol. 2009;53:2019–2027
- Antiplatelet therapy use after discharge among acute myocardial infarction patients with in-hospital bleeding. Circulation. 2008;118:2139–2145
- . International, randomized, controlled trial of lamifiban (a platelet glycoprotein IIb/IIIa inhibitor), heparin, or both in unstable angina. Platelet IIb/IIIa Antagonism for the Reduction of Acute coronary syndrome events in a Global Organization Network. Circulation. 1998;97:2386–2395
- . Randomized, placebo-controlled trial of titrated intravenous lamifiban for acute coronary syndromes. Circulation. 2002;105:316–321
- Enoxaparin vs unfractionated heparin in high-risk patients with non–ST-segment elevation acute coronary syndromes managed with an intended early invasive strategy: primary results of the SYNERGY randomized trial. JAMA. 2004;292:45–54
- . Inhibition of platelet glycoprotein IIb/IIIa with eptifibatide in patients with acute coronary syndromes. Platelet Glycoprotein IIb/IIIa in Unstable Angina: Receptor Suppression Using Integrilin Therapy. N Engl J Med. 1998;339:436–443
- The SYNERGY trial: study design and rationale. Am Heart J. 2002;143:952–960
- Prevalence, predictors, and impact of conservative medical management for patients with non–ST-segment elevation acute coronary syndromes who have angiographically documented significant coronary disease. JACC CI. 2008;1:369–378
- . Smoothing spline-based score tests for proportional hazards models. Biometrics. 2006;62:803–812
- Long-term mortality of patients undergoing cardiac catheterization for ST-elevation and non–ST-elevation myocardial infarction. Circulation. 2009;119:3110–3117
- Relationship of blood transfusion and clinical outcomes in patients with acute coronary syndromes. JAMA. 2004;292:1555–1562
- Does comorbidity account for the excess mortality in patients with major bleeding in acute myocardial infarction?. Circulation. 2007;116:2793–2801
- Transfusion and mortality in patients with ST-segment elevation myocardial infarction treated with primary percutaneous coronary intervention. Eur Heart J. 2009;30:2575–2583
- Guidelines for the diagnosis and treatment of non–ST-segment elevation acute coronary syndromes. Eur Heart J. 2007;28:1598–1660
- ACC/AHA 2007 guidelines for the management of patients with unstable angina/non–ST-elevation myocardial infarction: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines (Writing Committee to Revise the 2002 Guidelines for the Management of Patients with Unstable Angina/Non–ST-Elevation Myocardial Infarction) developed in collaboration with the American College of Emergency Physicians, the Society for Cardiovascular Angiography and Interventions, and the Society of Thoracic Surgeons endorsed by the American Association of Cardiovascular and Pulmonary Rehabilitation and the Society for Academic Emergency Medicine. J Am Coll Cardiol. 2007;50:e1–e157
- The PCI-CURE study. Lancet. 2001;358:527–533
- Effects of clopidogrel in addition to aspirin in patients with acute coronary syndromes without ST-segment elevation. N Engl J Med. 2001;345:494–502
- . Weighing benefits and risks—the FDA's review of prasugrel. N Engl J Med. 2009;361:942–945
PII: S0002-8703(10)00789-1
doi:10.1016/j.ahj.2010.09.001
© 2010 Mosby, Inc. All rights reserved.
