Timing and Correlates of Very Early Major Adverse Clinical Events Following Percutaneous Coronary Intervention

The timing of, and risk factors for, very early (< 24 hours) major adverse clinical events (death, myocardial infarction [MI], urgent target vessel revascularization [uTVR], and major bleeding) following percutaneous coronary intervention (PCI) in the contemporary era of routine coronary stenting and aggressive adjunctive periprocedural antiplatelet therapy are not well characterized. Our understanding of this temporal relationship and the correlates of very early (24 hours) and early (28 days) adverse clinical outcomes is important as a basis for decisions regarding aggressiveness of in-hospital care and the duration of hospitalization, including same-day PCI in selected patient populations. Early discharge following PCI offers a number of advantages for the patient and for those responsible for provision of cardiac services.
We sought to determine the incidence and precise timing of major adverse clinical events in all patients undergoing PCI in the Clopidogrel for the Reduction of Events During Observation (CREDO) trial, to identify the clinical, angiographic and procedural correlates of very early major adverse clinical events in this cohort, and to compare and contrast these variables with correlates of 28-day outcomes.

Methods
Study population. The CREDO trial design, methods and results have been previously reported.¹ In brief, the CREDO trial was a double-blind, placebo-controlled trial that randomized 2,116 patients who were to undergo elective PCI, or who were deemed highly likely to undergo PCI, at 99 centers in North America between June 1999 and April 2001. Major exclusion criteria included contraindications to antithrombotic or antiplatelet therapy, a > 50% stenosis of the left main coronary artery, failed PCI in the previous 2 weeks, coronary anatomy not amenable to stent implantation, persistent ST-elevation within 24 hours prior to randomization, planned staged PCI procedure, and administration of a glycoprotein (GP) IIb/IIIa inhibitor within 7 days, clopidogrel within 20 days, or thrombolytics within 24 hours.
Following randomization, and 3–24 hours prior to PCI, patients received either a 300 mg loading dose of clopidogrel or placebo. All patients were treated with 325 mg of aspirin. At the time of the procedure, both groups received 75 mg of clopidogrel and 325 mg of aspirin daily for 28 days following PCI. After 28 days and until the end of the study period, the pretreatment group continued to receive 75 mg of clopidogrel or placebo daily. Both groups received standard therapy, including 81–325 mg of aspirin daily, at the discretion of the operator until the end of the 12-month treatment period. At the discretion of the investigator, 20% of all patients could be prespecified at the time of randomization to receive a GP IIb/IIIa inhibitor at the time of PCI, and bailout GP IIb/IIIa inhibitor use was allowed.
Definitions. Death was defined as mortality from any cause. Acute Q-wave MI was defined as the presence of a new significant Q-wave with a duration of at least 0.04 seconds or a depth equal to one-fourth of the corresponding R-wave amplitude in 2 or more contiguous leads. A periprocedural non-Q-wave MI was defined as the elevation of the serum levels of creatine kinase (CK) or CK-MB isoenzyme to at least 3 times the upper limit of normal in 2 samples collected at different times, with an increase of at least 50% over the previous trough level. Stroke was defined as a new focal neurological deficit of vascular origin lasting at least 24 hours. uTVR was defined as coronary artery bypass grafting (CABG) initiated within 24 hours of the index procedure due to an inadequate or unstable result of the index PCI, even if ongoing myocardial ischemia was not present; repeat PCI or CABG of the target vessel initiated within 1 week of (re)hospitalization for acute MI or unstable angina; or repeat PCI or CABG of the culprit vessel initiated within 24 hours of the last episode(s) of ischemia. Major bleeding was defined as intracranial bleeding or bleeding associated with a decrease in hemoglobin > 5 g/dl (or when hemoglobin values were not available, a hematocrit decrease of at least 15%). Bleeding that met the criteria for major bleeding events, but was associated with a surgical procedure, including bypass surgery, was considered separately from other major bleeding events. All outcome events were adjudicated by a blinded Clinical Events Committee.

Timing of postprocedural MI. The timing of any postprocedural MI was determined by the time of occurrence of a new episode of prolonged angina (> 20 minutes) preceding the aforementioned electrocardiographic (ECG) or enzyme changes. If a preceding prolonged episode of angina was not documented, then the time of measurement of a trough enzyme level preceding the new enzyme elevation was used as the time of onset of new MI, provided the trough enzyme level occurred after the index PCI. If the trough enzyme level was prior to the index PCI, the timing of MI was based on the assessment of the Clinical Events Committee. For patients without prolonged angina or enzyme changes, the time of the first diagnostic changes on the ECG were used as the time of onset of a new MI.
Statistical methods. The Wilcoxon two-sample test was used for group comparisons of continuous variables. Chi-square or Fisher’s Exact tests were used for comparison of categorical variables. Results were considered statistically significant if the p-value was £ 0.05. Univariable logistic regression was used for assessment of individual variables. Tables provided include odds ratios with their confidence intervals and p-values. The p-values were generated by the Wald chi-square test for the significance of the individual effects. The outcomes of interest were death, MI and uTVR at 24 hours and 28 days, and only patients undergoing PCI were included. A p-value of 0.1 was used for entry, and 0.05 was considered statistically significant. A stepwise selection technique was then used for multivariable logistic regression. SAS statistical software (SAS, Inc., Version 8.0, Cary, North Carolina, 2000) was used for all the reported analyses.

Results

A total of 2,116 patients were enrolled in the CREDO trial. The majority (1,815 patients, 85.8%) underwent a PCI procedure. The baseline clinical, angiographic and procedural characteristics for the patients who underwent PCI and comprise the study cohort are described in Tables 1 and 2. Patients who suffered a major adverse clinical event were more likely to be older, weigh less, have a past medical history of peripheral vascular disease, and be taking aspirin. With respect to angiographic and procedural patient characteristics, patients who suffered a major adverse clinical event had more extensive and severe disease (≥ 2 vessels with > 50% stenosis, maximum preprocedural stenosis, preprocedural thrombolysis in myocardial infarction [TIMI] flow grade, American Heart Association [AHA] lesion grade) and more complex procedures (number of lesions treated, total stent length, use of non-stent PCI devices [Rotoblator^®, transluminal extraction coronary or directional coronary atherectomy, laser], postprocedural TIMI flow grade) that were more commonly associated with procedural complications (distal embolization, minor dissection, transient occlusion, perforation and side branch occlusion).

One-hundred and thirty-nine patients (7.7%) experienced a major adverse clinical event (death, MI, uTVR) within the first 28 days (Table 3, Figures 1 and 2). The majority of these events occurred within the first 24 hours (111 patients, 80%). The risk of an event was greatest within the first 6 hours.
The multivariable predictors of major complications within 24 hours and 28 days, respectively, are described in Table 4. Age, AHA lesion grade and history of peripheral vascular disease are consistent predictors of very early (24 hours) and early (28 days) outcomes. Similarly, preprocedural TIMI flow grade and the number of vessels with obstructive coronary artery disease (> 50% stenosis) predict very early and early outcomes, respectively. Diabetes mellitus (28-day outcomes) or insulin therapy (24- hour outcomes) was protective and predicted a lower incidence of major adverse clinical events. The overall C-statistics for the final multivariable models were 0.69 (24 hours) and 0.66 (28 days), respectively.
The incidence and timing of all non-CABG and major non-CABG bleeding events (Table 3, Figure 3) are reported. Forty-one patients suffered a non-CABG bleeding event (2.3%), and 18 patients suffered a major non-CABG bleeding event (1.0%). The risk of an event was greatest within the first 6 hours.

Discussion

This retrospective analysis of the CREDO database confirms that the incidence of major adverse clinical events following contemporary PCI is low, that the majority of these events are postprocedural MIs, and that the majority of these events occur early (within 24 hours), with the greatest risk of an event within the first 6 hours. The risk of death is particularly low (hazard functions < 0.001) and constant. Similarly, the risks of uTVR and major bleeding events are very low (hazard function < 0.001).
Hazard functions and clinical decision making. The hazard function is a measure of risk of event per unit time. In this analysis, the issue of “safe discharge” is, of course, a relative judgement — the risk of death, MI and urgent TVR is never zero in patients with coronary artery disease. Therefore, events will continue to accrue over time, including between 24 hours and 28 days, as evident from this analysis. However, the critical issues are: 1) when is the risk/hazard “low enough?” 2) when is the risk (hazard function) flat, signifying no further benefit to ongoing observation since it is not possible to observe indefinitely in anticipation of rare events? and 3) is there likely to be clinical benefit to observation — will the treatment be different if an event is observed in-hospital?
With respect to the clinical endpoints studied, and bearing these three issues in mind, we suggest the following interpretation of this analysis: the incidence of death is very low at all time points (0.06% within 6 hours, 0.06% between 6 hours and 24 hours, and 0.22% between 24 hours and 28 days) (Table 3, Figure 2); the hazard function is flat (Figure 2); and it is unclear whether in-hospital observation might prevent death following PCI. However, it must be conceded that we cannot exclude the possibility that mortality may have been reduced by in-hospital observation, as practiced in this study, for at least 24 hours.
The incidence of urgent TVR is also very low (0.44% within 6 hours, 0.22% between 6 hours and 24 hours, and 1.16% between 24 hours and 28 days) (Table 3, Figure 2); the hazard function is flat after 6 hours with minor late fluctuations due to rare events; and, similar to death, it is unclear whether in-hospital observation might improve patient care, but intuitively, this is assumed to be so. We can also only speculate as to when the need for uTVR became apparent, since we can only report the timing of the TVR procedure.
With regard to MI, the risk extends beyond the 6-hour period (Table 3, Figure 2). However, this is largely an artifact of the measurement protocol — it can be assumed periprocedural infarctions are due to events at the time of balloon inflation/stent deployment, although we have recorded the timing of the event based on the timing of the enzyme assay. It is unclear whether prolonged hospital observation would prevent further events in patients with asymptomatic and electrocardiographically “silent” postprocedural elevation of cardiac enzymes,² but the timing of the final enzyme assay and the clinical significance of such enzyme elevation would need to be considered when devising any early (same-day) discharge strategy post-PCI.
Timing of and risk factors for clinical events after balloon angioplasty. The Integrilin to Minimize Platelet Aggregation and Coronary Thrombosis-II (IMPACT-II) trial investigators have published the results of a retrospective analysis that evaluated the time to the primary trial endpoint (death, myocardial infarction, repeat nonelective PCI, nonelective bypass surgery, or stenting for abrupt closure) in this trial of 4,010 patients randomized to placebo or one of two eptifibatide regimens during PCI (plain-old balloon angioplasty — coronary stenting was permitted only as a “bailout” treatment for threatened or manifest abrupt closure).³ The risk of major adverse cardiac events declined rapidly between 6 and 9 hours (66% occurred within 6 hours and the hazard function plot was flat after 9 hours). Independent predictors in the IMPACT-II study population of “early” events (< 6 hours) included dissection, pre- and postprocedural coronary blood flow, side branch occlusion, procedural thrombolytic use, previous bypass surgery, presentation with unstable angina, absence of diabetes and hyperlipidemia.
These results support the safety of early discharge following plain-old balloon angioplasty with “bailout” stenting. However, the relevance of these results to contemporary PCI practices is limited due to the adoption of routine coronary stenting as a preferred PCI treatment strategy. Routine stenting is certainly associated with better angiographic outcomes, but there is no consistent evidence that this strategy reduces the incidence of death or periprocedural MI. Furthermore, subacute closure following balloon angioplasty is a well-documented complication that typically occurs within 6 hours of the PCI procedure, and most commonly while the patient is still in the cardiac catheterization laboratory. But although the incidence of this complication is dramatically reduced by routine stenting, stent thrombosis is less predictable and patients may remain at risk of this highly morbid complication.
Incidence of major adverse clinical events following contemporary PCI. Despite the fact that indices of patient comorbidity have increased, and that operators are attempting to treat more complex lesions, procedural and clinical outcomes continue to improve and procedural safety has been further enhanced.⁴ For example, a review of 18,593 PCI procedures performed between 1992 and 2000 reported an overall need for emergency CABG of 0.61%.⁵ The need for emergency CABG declined during the course of this study period from 1.5% of PCI procedures in 1992, to 0.14% in 2000 (p < 0.001). Among patients undergoing elective PCI not requiring emergency CABG, there was also a significant decline in the incidence of death (2.0% to 0.9%; p < 0.001) and Q-wave MI (0.7% to 0.2%; p < 0.001).
Analysis of the published results of the Sirolimus-Eluting Stent in De Novo Coronary Lesions (SIRIUS) trial support the contention that contemporary PCI techniques are associated with very low rates of morbidity and mortality.⁶ The SIRIUS trial randomized a low-risk population (238 patients; stable or unstable angina or documented silent ischemia; single, de novo coronary lesions; 15–30 mm long; 2.5–3.5mm reference vessel diameter) to either bare-metal stent or sirolimus-eluting stent implantation, and documented exceptionally low rates of in-hospital Q-wave MI (0.25%), TVR (0%) and death (0%).
Major bleeding post-PCI. Bleeding events post-PCI procedures are typically due to either vascular trauma at the access site, or the consequences of intensive periprocedural anticoagulation and intravenous and oral antiplatelet therapies. A number of recent developments in vascular access techniques have impacted access site complications, including development of, and experience with, transradial angioplasty, percutaneous suture of the femoral artery and other vascular closure devices.^7,8 However, a meta-analysis of closure devices has raised serious questions about their safety and efficacy and further studies are required, particularly in the context of early discharge following PCI.⁹
Similarly, there are ongoing efforts to define an optimal anticoagulation and antiplatelet regimen for the prevention of periprocedural thrombotic complications and stent thrombosis. Patients in the CREDO trial all received aspirin (325 mg of aspirin daily; a 300 mg loading dose of clopidogrel 3–24 hours before the procedure or matching placebo; and 75 mg of clopidogrel immediately after the PCI procedure and daily thereafter for the first 28 days). Procedural heparin was administered, achieving a mean activated clotting time of 282 seconds. Approximately 45% of patients received a GP IIb/IIIa inhibitor, either electively or as a “bailout” intervention.
Predictors of early major adverse clinical events. The predictors of early (< 6 hours) major adverse cardiac events in the balloon-only era of the IMPACT-II study correlate quite closely with the predictors of major adverse clinical events within 24 hours of the PCI procedure in our analysis of the stent-dominated CREDO study. These predictor variables are mostly consistent with prior studies of procedural outcomes, with the exception of baseline clinical variables that are indicative of the presence of diabetes mellitus (insulin therapy and history of diabetes mellitus, respectively) and which predicted a favorable outcome following PCI in both the IMPACT-II and CREDO analyses. This is certainly counterintuitive, but the relationship of diabetes to procedural outcomes has always been unclear and remains open to debate and speculation regarding the possible underlying mechanisms. It also remains possible that this is a statistical aberration and reflects the play of chance. This possibility is of particular concern when attempts to define correlates of a rare outcome are confounded by “overfitting” artifacts.¹³
Similarly, the predictors of 28-day outcomes in this analysis of the CREDO trial are consistent with prior analyses and do not differ significantly from the aforementioned predictors of outcomes within 24 hours of the PCI procedure.³
Recent randomized trials. Two recent randomized trials evaluated the safety and efficacy of same-day discharge following PCI in selected populations. Choussat et al, in a trial examining access techniques and anticoagulation strategies, randomized 1,005 patients after a bolus of abciximab and uncomplicated transradial percutaneous coronary stent implantation to either same-day discharge and no infusion of abciximab (Group 1, n = 504) or overnight hospitalization and standard 12-hour infusion of abciximab (Group 2, n = 501). The primary endpoint of the 30- day incidence of death, MI, urgent revascularization, major bleeding, repeat hospitalization, access site complications and severe thrombocytopenia was evaluated. Abciximab bolus followed by same-day discharge was found to be noninferior to continued standard infusion and overnight hospitalization, with similar incidences of the primary endpoint in both groups, and with an overall low rate of bleeding in either group (Group 1: 0.8% vs. Group 2: 0.2%).¹⁰
Another trial randomized 800 consecutive patients to sameday discharge or overnight hospital stay using a femoral approach and heparin only during the intervention with clopidogrel loading following the procedure. Four hours after PCI, patients were evaluated and discharged or kept overnight. The primary endpoints were death, MI, CABG, repeat PCI, or puncture-related complications occurring within 24 hours after PCI. Of the patients that met criteria for same-day discharge, the overall event rate was low, with the composite endpoint occurring in 1 of 326 patients randomized to same-day discharge and 2 of 312 overnight stay patients. All 3 events were related to puncture site complications.¹¹
The two randomized trials demonstrate differing approaches to minimizing very early complications related to coronary interventions. Both approaches attempt to minimize bleeding complications. The first trial used a transradial vascular access site, which appears to be associated with fewer bleeding complications.¹² The second study used a femoral approach, which is still the most common access approach used, but did not allow for the use of GP IIb/IIIa inhibitors, thus selecting for a lowerrisk patient population than is seen in the CREDO patients.
Study limitations. This is a retrospective analysis and is susceptible to the limitations of this methodology.
However, this is a largely descriptive analysis, and therefore the results are not susceptible to the confounding and bias that is present in retrospective comparative analyses. The detailed information available to the study investigators for the purposes of the randomized, controlled trial facilitate such descriptive analyses and is a strength of this research methodology. However, the timing of cardiac enzyme measurement (typically according to a predetermined study protocol) will impact the results of a retrospective analysis. In all likelihood, the clinical event resulting in the periprocedural release of cardiac enzyme can be presumed to be the procedural balloon inflation. This is not reflected in our analysis (or prior analyses), since we have recorded the time of the first measurement of an elevated cardiac biomarker as the timing of the event.

Conclusions
We propose that these data support the safety of same-day PCI in a select population. The incidence of death and urgent TVR is very low, and the hazard functions suggest that risk is constant, certainly after a 6-hour postprocedural observation period. Similarly, major bleeding is very rare and risk is constant after 6 hours. Postprocedural MI is more common and risk appears to be more prolonged, but the clinical significance of this biochemical abnormality remains uncertain in the absence of symptoms or ECG changes consistent with myocardial ischemia, and conclusions from this study regarding the duration of elevated risk is confounded by the limitations of this study methodology.
It would seem reasonable to propose same-day discharge following successful, elective PCI in patients with a similar clinical, angiographic and procedural profile as the population in the CREDO trial. Further risk stratification on the basis of baseline characteristics is of questionable clinical utility, but age, history of peripheral vascular disease, diabetes or prior MI, and the presence of multivessel disease might be considered when making such a decision regarding early discharge. Additional factors include assessment of the patient’s home circumstances and ability to access medical care immediately in the event of symptoms or signs of complications.
We believe that this retrospective analysis provides important information regarding incidence, correlates and timing of major adverse events following contemporary PCI and may be used to further inform the discussion regarding appropriateness of same-day discharge following PCI. However, the final decision will always remain a clinical judgement, including assessment of when the risk is low enough, when the risk is constant, and what is the likelihood that in-hospital observation will offer the opportunity to prevent morbidity and mortality following PCI.

References

1. Steinhubl SR, Berger PB, Mann JT 3rd, et al. Early and sustained dual oral antiplatelet therapy following percutaneous coronary intervention: A randomized controlled trial. JAMA 2002;288:2411–2420.
2. Ellis SG, Chew D, Chan A, et al. Death following creatine kinase-MB elevation after coronary intervention: Identification of an early risk period: Importance of creatine kinase-MB level, completeness of revascularization, ventricular function, and probable benefit of statin therapy. Circulation 2002;106:1205–1210.
3. Thel MC, Califf RM, Tardiff BE, et al. Timing of and risk factors for myocardial ischemic events after percutaneous coronary intervention (IMPACT-II). Integrilin to Minimize Platelet Aggregation and Coronary Thrombosis. Am J Cardiol 2000;15;85:427–434.
4. Laskey WK, Williams DO, Vlachos HA, et al. Changes in the practice of percutaneous coronary intervention: A comparison of enrollment waves in the National Heart, Lung, and Blood Institute (NHLBI) Dynamic Registry. Am J Cardiol 2001;87:964–969; A3-4.
5. Seshadri N, Whitlow PL, Acharya N, et al. Emergency coronary artery bypass surgery in the contemporary percutaneous coronary intervention era. Circulation 2002;106:2346–2350.
6. Morice MC, Serruys PW, Sousa JE, et al. A randomized comparison of a sirolimus-eluting stent with a standard stent for coronary revascularization. N Engl J Med 2002;346:1773–1780.
7. Applegate RJ, Grabarczyk MA, Little WC, et al. Vascular closure devices in patients treated with anticoagulation and IIb/IIIa receptor inhibitors during percutaneous revascularization. J Am Coll Cardiol 2002;40:78–83.
8. Morice MC, Dumas P, Lefevre T, et al. Systematic use of transradial approach or suture of the femoral artery after angioplasty: Attempt at achieving zero access site complications. Catheter Cardiovasc Interv. 2000;51:417–421.
9. Koreny M, Riedmuller E, Nikfardjam M, et al. Arterial puncture closing devices compared with standard manual compression after cardiac catheterization: Systematic review and meta-analysis. JAMA 2004;291:350–357.
10. Bertrand OF, De Larochelliere R, Rodes-Cabau J, et al. A randomized study comparing same-day home discharge and abciximab bolus only to overnight hospitalization and abciximab bolus and infusion after transradial coronary stent implantation. Circulation 2006;114:2636–2643.
11. Heyde GS, Koch KT, de Winter RJ, et al. Randomized trial comparing same-day discharge with overnight hospital stay after percutaneous coronary intervention: Results of the Elective PCI in Outpatient Study (EPOS). Circulation 2007;115:2299–2306.
12. Choussat R, Black A, Bossi I, et al. Vascular complications and clinical outcome after coronary angioplasty with platelet IIb/IIIa receptor blockade. Comparison of transradial vs transfemoral arterial access. Eur Heart J 2000;21:662–667.
13. Harrell FE Jr, Lee KL, Califf RM, et al. Regression modelling strategies for improved prognostic prediction. Stat Med 1984;3:143–152.