Author + information
- Received February 17, 2004
- Revision received June 17, 2004
- Accepted June 22, 2004
- Published online October 6, 2004.
- Prabhakara S. Heggunje, MD, FACC*,
- Kishore J. Harjai, MD, FACC†,
- Gregg W. Stone, MD, FACC‡,
- Rajendra H. Mehta, MD, FACC§,
- Dominic L. Marsalese, MD, FACC*,
- Judith A. Boura, MS*,
- William W. O'Neill, MD, FACC* and
- Cindy L. Grines, MD, FACC*,* ()
- ↵*Reprint requests and correspondence:
Dr. Cindy L. Grines, Director, Interventional Cardiology Fellowship, Division of Cardiovascular Medicine, William Beaumont Hospital, 3601 West 13 Mile Road, Royal Oak, Michigan 48073-6769
Objectives We evaluated whether patients' clinical status, angioplasty success, or both, should guide discharge after primary angioplasty (i.e., percutaneous coronary intervention [PCI]) for acute myocardial infarction (AMI).
Background Current guidelines do not address a discharge strategy for AMI patients undergoing successful PCI.
Methods Patients who underwent PCI in Primary Angioplasty in Myocardial Infarction (PAMI) studies (N = 3,188) were classified as “high clinical risk” if they had either age >70 years, Killip class >1, heart rate >100 beats/min, systolic blood pressure <100 mm Hg, anterior MI, or left bundle branch block, and as “low clinical risk” if none was present. Successful PCI patients were compared with those with unsuccessful PCI in both groups for 30-day major adverse cardiac events (MACE).
Results Percutaneous coronary intervention was successful in 668 (90%) of 745 low-risk clinicaland 2,104 (86%) of 2,443 high-risk clinical patients. Regardless of clinical risk status, patients with successful PCI had lower 30-day MACE than those with unsuccessful PCI (low-risk group: 4.6% vs. 22%, p < 0.0001; high-risk group: 7% vs. 21%; p < 0.0001). Moreover, successful PCI patients with either risk status had few MACE after day 4, whereas unsuccessful PCI patients had more MACE. The success of PCI was the strongest independent predictor of 30-day MACE (odds ratio [OR] 3.7, 95% confidence interval [CI] 2.8 to 5.0). A constellation of three or more high-risk clinical features also predicted higher 30-day MACE (OR 2.25, 95% CI 1.62 to 3.12).
Conclusions The success of PCI is the prime determinant of clinical outcome after PCI for AMI. The majority of AMI patients with less than three high-risk clinical features who undergo successful PCI may be discharged from the hospital by day 4. In contrast, patients with more than two high-risk clinical features or unsuccessful PCI may need longer observation.
With current emphasis on evidence-based medicine and cost effectiveness, the timing of discharge of acute myocardial infarction (AMI) patients from the hospital is crucially important. Practice guidelines for management of AMI recommend a predischarge submaximal stress electrocardiogram or some form of pharmacologic stress imaging at four to six days (1) and further invasive testing, if indicated, before discharge from the hospital. The value of routine four to six days of monitoring and subsequent noninvasive testing in patients with an uncomplicated course after successful primary angioplasty for AMI is uncertain. It has been shown that early identification of low-risk patients with MI results in safe omission of the intensive care phase and noninvasive testing, a day 3 hospital discharge strategy, and substantial cost savings (2).
In general, clinically low-risk AMI patients suffer few cardiac events early after AMI. It is also well established that patients with AMI who undergo successful percutaneous coronary intervention (PCI) fare much better than those who fail to reperfuse (3). Although angiographic success and “low-risk status” are both associated with better outcomes, a given patient may fall into one group, both groups, or neither group. Currently, patients with a low clinical risk status are often selected for early discharge, whereas high-risk patients are observed for a longer period, regardless of PCI success. However, it is unclear whether PCI success influences outcomes differently among the two risk categories or provides guidance for discharge planning.
The purpose of this analysis was to determine the importance of PCI success among the two clinical risk groups in identifying patients who could be discharged from the hospital early.
The Primary Angioplasty in Myocardial Infarction (PAMI) studies prospectively enrolled 4,023 patients with AMI in seven different trials (PAMI-1, PAMI-2, PAMI Stent Pilot, Stent PAMI, Local PAMI, Air PAMI, and PAMI-No SOS) (2,4–10), including two studies with concomitant registry enrollment (5,8). The inclusion and exclusion criteria for these trials are published elsewhere (11). In all of the PAMI trials, patients were >18 years old, had ST-segment elevation of ≥1 mm in two or more contiguous leads and/or left bundle branch block, and symptom onset within 12 h. Patients were excluded if thrombolytic agents were given for the index ST-segment elevation myocardial infarction (MI),were in cardiogenic shock, had a stroke within a month, were of child-bearing potential, had end-stage renal disease, or a life expectancy from a noncardiac condition of <1 year. Of the 4,023 patients enrolled in PAMI trials, 3,188 who underwent PCI and had a record of final Thrombolysis In Myocardial Infarction (TIMI) flow grade and final diameter stenosis were included in this analysis.
Data collection and comparisons
For each of the clinical trials, research nurses or coordinators at each site collected data prospectively and completed detailed case-report forms. Independent data monitors traveled to the participating sites to verify hospital records for all patients. Cine angiograms, which were obtained at the time of the acute coronary intervention, were analyzed by the individual operators and subsequently by core laboratories to assess coronary anatomy and estimate TIMI flow grades, percentage diameter stenosis, left ventricular ejection fraction (LVEF), and angiographic outcomes of intervention.
We pooled the clinical, demographic, angiographic, and outcomes data on the 3,188 patients included in this analysis. Patients were considered to be high risk if they had any of the following: age >70 years, heart rate >100 beats/min, systolic blood pressure <100 mm Hg, Killip class >1, anterior MI, or left bundle branch block (LBBB). These clinical features were used to define high-risk patients in the Air-PAMI study (10) and have been previously shown to be the most important predictors of early outcome after AMI (12). Success of PCI was determined by a diameter stenosis ≤30% and TIMI flow grade 3 on final angiography. Core laboratory data were used in most analyses. Core laboratory data were considered incomplete for LVEF and initial TIMI flow grade; therefore, operator-defined data were used for these variables.
For the purpose of this analysis, patients were classified into four groups based on the technical success of PCI and the clinical risk status: group 1 = clinical low-risk and successful PCI; group 2 = clinical high-risk and successful PCI; group 3 = clinical low-risk and unsuccessful PCI; and group 4 = clinical high-risk and unsuccessful PCI. We compared baseline clinical, demographic, and angiographic characteristics between these four groups. We also compared in-hospital and 30-day outcomes (death, re-infarction, or ischemia-driven target vessel revascularization [I-TVR]) between the groups.
Study end points and definitions
The primary study outcomes included in-hospital mortality and in-hospital and 30-day incidence of major adverse cardiac events (MACE), defined as death, re-infarction, or I-TVR. Re-infarction was defined as recurrent clinical symptoms or the development of new electrocardiographic changes accompanied by new elevation of creatine kinase and creatine kinase-MB enzyme levels. Ischemia-driven TVR was defined as TVR prompted by symptoms or objective evidence of ischemia. The incidences of in-hospital complications, such as bradycardia, ventricular arrhythmia, sustained hypotension, cardiopulmonary resuscitation, disabling stroke, and the need for initiation of hemodialysis, were also compared between the four groups. Sustained hypotension was defined as systolic blood pressure <80 mm Hg unresponsive to intravenous fluids, requiring pressors for >1 h or intra-aortic balloon pump.
Statistical analyses were completed on the categorical variables using the chi-square test when appropriate (expected frequency >5). Otherwise, the Fisher exact test was used. Continuous variables were analyzed using a nonparametric two-tailed Wilcoxon rank test. Step-down multivariate logistic regression was utilized to adjust for baseline differences in clinical and angiographic characteristics, as well as technical success of PCI to derive independent predictors of death, re-infarction, and MACE. Kaplan-Meier curves were analyzed using the log-rank test comparing the four categories. All analyses were completed using SAS version 8.0 (SAS Institute, Cary, North Carolina).
Percutaneous transluminal intervention was successful in 2,772 patients (87%), and 2,443 (77%) had at least one high-risk clinical feature. Among the clinical high-risk patients, 1,368 (39.4%) had one risk factor, whereas two, three, four, and five risk factors were present in 816 (23.5%), 333 (9.6%), 96 (2.8%), and 16 (0.5%) patients, respectively. Age >70 years was seen in 844 (24.7%), heart rate >100 beats/min in 665 (19.5%), systolic blood pressure <100 mm Hg in 973 (28.5%), Killip class >1 in 448 (13.1%), anterior MI in 1,433 (42%), and LBBB in 36 (1%) patients. Baseline clinical characteristics in each of the four subgroups are given in Table 1.Patients with unsuccessful PCI more often had diabetes and previous coronary artery bypass grafting than those who had successful PCI. The high-clinical risk group was older and had more females, a higher Killip class, and more often a history of congestive heart failure, as compared with the low-risk group.
Baseline angiographic data is given in Table 2.Patients with unsuccessful PCI had a longer chest pain onset to balloon time. They also had a higher prevalence of multivessel disease, TIMI flow grade 0 or 1 in the infarct-related artery on the initial angiogram, and a higher prevalence of thrombus at the culprit site. The incidence of dissection was more frequent in patients with unsuccessful PCI than in those with successful PCI. The high-clinical risk group had a lower LVEF and, by definition, had a left anterior descending coronary artery as the predominant infarct-related artery.
Data on postdischarge medical treatment were not available, and the data on in-hospital medication use were incomplete. Analysis of the available information showed no significant differences in the in-hospital use of aspirin, ticlid, beta-blockers, or angiotensin-converting enzyme inhibitors between the groups of successful and failed PCI within the clinical low-risk group. Within the clinical high-risk group, however, patients with failed PCI received beta-blockers less often than those who had successful PCI did (57% vs. 77%, p < 0.0001). Information on beta-blocker use was missing in 334 of the 2,443 high-risk patients. There was no difference in the in-hospital use of aspirin and angiotensin-converting enzyme inhibitors. No data were available on the use of statins.
In-hospital clinical outcomes
Table 3shows the in-hospital complications and outcomes. In-hospital mortality was highest in the high-risk group with unsuccessful PCI (10.9%) and lowest in the low-risk group with successful PCI (0.5%). The group with unsuccessful PCI had higher in-hospital MACE than the group with successful PCI, irrespective of their clinical risk status (16.9% vs. 3.7%, p < 0.0001 in the low-risk group; 17.4% vs. 5.6%, p < 0.0001 in the high-risk group). This was primarily due to an increased rate of recurrent MI and I-TVR. Sustained hypotension and intra-aortic balloon pump use was more frequent in patients who had unsuccessful PCI, irrespective of their risk status.
Step-down multivariate logistic regression analysis showed that unsuccessful PCI, advancing age, and multi-vessel disease were the only independent predictors of in-hospital death (Table 4).In-hospital MI was predicted by unsuccessful PCI and not using a stent. Other variables, including diabetes and abciximab use, did not predict in-hospital outcome. In-hospital beta-blocker use was not an independent predictor of MACE (p = 0.48). Clinical high-risk status was not an independent predictor of either in-hospital death or re-infarction and predicted a lower incidence of in-hospital TVR.
30-day clinical outcomes
Table 5shows one-month clinical outcomes in different subgroups. Irrespective of their clinical risk status, patients with unsuccessful PCI had a higher incidence of death (9.8%), re-infarction (5.2%), and I-TVR (10.3%) at one month than those who had successful PCI. Patients with unsuccessful PCI had more MACE at one month (22% vs. 6%, p < 0.0001), mostly attributable to a substantially higher number of TVR in the low-risk group with unsuccessful PCI (18% vs. 4%, p < 0.0001) during the first month.
Multivariate logistic regression analysis once again showed that unsuccessful PCI, age, and multi-vessel disease were strong predictors of both 30-day MACE and mortality (Table 6).In addition, female gender emerged as an independent predictor of 30-day mortality, whereas female gender as well as not using a stent independently predicted 30-day MACE. In contrast, clinical high-risk status was not an independent predictor of higher 30-day MACE and showed only a trend toward higher 30-day mortality. Other baseline variables, including a history of diabetes, hypertension, previous MI, previous congestive heart failure, peripheral vascular disease, and previous coronary artery bypass graft surgery were not independent predictors of either 30-day MACE or mortality.
Figure 1shows Kaplan-Meier curves showing the cumulative incidence of MACE (death, MI, and I-TVR) at 30 days. Patients with successful PCI had very few MACE after hospital day 4, whereas patients with unsuccessful PCI continued to experience MACE beyond the first four hospital days at a much higher rate (2.5% vs. 6.7%, p < 0.001). Figures 2and 3show Kaplan-Meier survival curves for death and MI as well as death at 30 days, respectively. Once again, the group with successful PCI had a low event rate after the initial two to three days, whereas the group with unsuccessful PCI continued to experience a higher rate of death and MI up to 10 days.
A subgroup analysis of the clinical high-risk group showing different gradations of risk and 30-day outcome is given in Table 7.Patients with three or more high-risk clinical features had a significantly higher 30-day MACE than those with one or two high-risk clinical features, which was primarily due to higher mortality in the very high-risk group. Having three or more high-risk clinical features was also an independent predictor of 30-day MACE (odds ratio [OR] 2.25, 95% confidence interval [CI] 1.62 to 3.12, p < 0.0001) and death (OR 3.9, 95% CI 2.53 to 6.02, p < 0.0001).
Further analysis to examine the timing of deaths in patients who underwent successful PCI revealed a total of 58 deaths in the first 30 days of which 24 occurred by day 4. Of the 34 patients who died between day 5 and day 30, 16 had three or more high-risk clinical features, and interestingly, 13 of these 16 patients had sustained hypotension or congestive heart failure. Of the remaining 18 patients, the vast majority had cardiac or noncardiac complications that prolonged the hospital stay, and these deaths occurred in the hospital. Only three patients died of presumed cardiac causes after discharge from the hospital on days 13, 20, and 19, and these patients had an estimated LVEF of 30%, 35%, and 50%, respectively. The last patient was recorded to have a re-infarction on day 12 requiring I-TVR.
Although the advent of thrombolytic therapy in late 1980s improved the overall outcome of AMI patients, recurrent ischemic events continued to occur unpredictably in the post-MI period, necessitating prolonged observation and additional testing. Thus, the latest national guidelines still recommend four to six days of in-hospital observation of AMI patients, followed by some form of noninvasive testing for adequate risk stratification (1). Primary PCI, on the other hand, is associated with a low incidence of recurrent ischemic events (4,8,13) and allows adequate risk stratification of AMI patients at the time of initial angiography. The PAMI-II investigators have shown that the low-risk subsets of AMI patients undergoing successful primary PCI can be safely discharged from the hospital on day 3 (2).
What defines “low risk,” however, is less clear. Is it the clinical risk status or the angiographic success? It is not uncommon to encounter patients with angiographic success and “high-risk” clinical feature(s) or those with unsatisfactory angiographic results in a clinically “low-risk” patient. Our study shows that the angiographic result is the prime determinant of clinical outcome. In fact, patients undergoing successful PCI had very few events after day 4, irrespective of their clinical risk status. It was only when three or more high-risk features were present concurrently that the outcome was less satisfactory. Based on these findings, we think, the majority of AMI patients with <3 high-risk clinical features who undergo successful PCI may be discharged from the hospital early, possibly by day 4. The economic impact of using PCI success as a major criterion for discharge planning is huge, as five of every six patients undergoing primary PCI fulfilled the criteria for successful PCI.
As expected, in this study, the patients with high-risk clinical features and unsuccessful PCI had a significant number of MACE during the first post-MI month. This observation supports the current practice of keeping such patients in the hospital for a longer observation period. However, the immediate post-MI course of the clinically low-risk group with unsuccessful PCI is less well known, however. The present study, for the first time, shows that this subset of AMI patients is not at “low risk,” contrary to being classified clinically as being low risk. Hard end points such as death and re-infarction continue to occur well beyond the initial 4 days, mandating an additional hospital stay for these “low-risk” patients.
High-risk clinical features chosen in this study have been shown to be the best independent predictors of 30-day outcome after AMI (12). However, this study shows that the presence of one or two high-risk clinical features does not negatively affect 30-day outcome in AMI patients undergoing successful PCI. In contrast, the presence of three or more high-risk clinical features dramatically increases the risk of an adverse outcome during the first post-MI month. The difference in 30-day outcome between patients with two and three clinical high-risk features is statistically highly significant. In this study, three or more high-risk clinical features were present in 13% of the study population. This small group of highest risk patients is at a substantial risk of adverse cardiovascular outcome, regardless of PCI success, and may merit longer observation and in-patient treatment.
The criteria used in the classification of high- and low-risk patients in this study are crucial. We used the same clinical high-risk features used in the Air-PAMI study (10), which have been previously shown to be the most important predictors of early outcome after AMI (12). In an analysis of 41,021 AMI patients enrolled in the GUSTO-1 study, advanced age, anterior infarction, lower systolic pressure, higher Killip class, and elevated heart rate were found to be the strongest predictors of 30-day mortality, and together, these five characteristics accounted for 90% of the prognostic information in the baseline clinical data (12).
As the clinical high-risk group had older patients by definition, it is not surprising to find significant differences in the distribution of other baseline characteristics between the clinical high- and low-risk groups (Table 1). However, based on the GUSTO-1 observations already mentioned, the differences in the distribution of “other” baseline variables between high and low clinical risk groups would explain only a 10% difference in 30-day outcome between the groups. Although some of the baseline variables may have predicted procedural success, on multivariate analysis, only female gender independently predicted 30-day outcome.
The findings of the mortality analysis in the successful PCI group are consistent with an early discharge strategy for AMI patients undergoing successful PCI. Only a very small number of deaths will be missed by a day 4 discharge strategy for AMI patients with less than three high-risk clinical features who undergo successful PCI and do not have any overt cardiac or noncardiac complications. Some of these deaths are likely to be arrhythmic and may need specialized electrophysiologic evaluation and/or treatment rather than a prolonged hospital stay to prevent them. Temporary use of external defibrillators in selected patients with a low LVEF is one such consideration and needs further evaluation in this setting.
The finding of PCI success as the most important determinant of adverse cardiac events during the first post-MI month is not surprising. Both experimental and clinical evidence indicates that the benefits of a patent infarct-related artery include a favorable effect on ventricular remodeling (improved healing of infarcted tissue and prevention of infarct expansion) (14,15), improvement in left ventricular diastolic and systolic function, better electrical stability, and reduced long-term recurrent ischemic events and mortality (16,17). We used a more stringent definition of PCI success by including only TIMI flow grade III and also a final diameter stenosis ≤30% instead of 50%. This definition of PCI success probably contributed to better risk stratification of AMI patients (18,19).
This is a post-hoc analysis of pooled data from multiple clinical trials and is subject to the inherent limitations of such analysis. Although TIMI flow grade 3 in the infarct-related artery is associated with a good outcome, a few patients may do poorly due to microvascular pathology. We did not include myocardial blush score or TIMI frame count to define PCI success. However, this is unlikely to influence the findings of this study, as once-month mortality in low-risk patients with successful PCI was only 0.5%. We only evaluated MACE, but other problems such as heart failure and bradyarrhythmias, which may influence discharge planning significantly, are not evaluated. Information on medical treatment was incomplete and hence could not be analyzed. By design, PAMI studies excluded the highest risk patients, such as those presenting with cardiogenic shock, recent cerebrovascular accident or end-stage renal disease, and our data are not applicable to this population. Finally, our data are applicable to patients undergoing primary PCI and may not apply to patients undergoing rescue or delayed PCI.
In patients undergoing primary PCI, procedural success provides significant additional prognostic value to the “clinical risk stratification” algorithm that could be useful in the discharge planning of such patients. Barring other complications such as heart failure, arrhythmia, or noncardiac complications, most patients with angiographically successful PCI can safely be discharged from the hospital early, possibly by day 4. In contrast, patients with an angiographically suboptimal outcome might need longer periods of hospitalization and remain at relatively high risk for MACE up to 7 to 10 days. Very high-risk patients with three or more high-risk clinical features may also need longer periods of hospitalization, regardless of PCI outcome.
The authors thank Ms. Sue Tomaszycki for her technical assistance in generating Kaplan-Meier survival graphs for the study.
- Abbreviations and acronyms
- acute myocardial infarction
- ischemia-driven target vessel revascularization
- left bundle branch block
- left ventricular ejection fraction
- major adverse cardiac events
- Primary Angioplasty in Myocardial Infarction
- percutaneous coronary intervention
- Thrombolysis In Myocardial Infarction
- Received February 17, 2004.
- Revision received June 17, 2004.
- Accepted June 22, 2004.
- American College of Cardiology Foundation
- Ryan T.J.,
- Antman E.M.,
- Brooks N.H,
- et al.
- Grines C.L.,
- Marsalese D.,
- Brodie B,
- et al.
- GUSTO-1 Angiographic Investigators,
- Ross A.M.,
- Lundergan C.F.,
- Rohrbeck S.C., et al
- Second Primary Angioplasty in Myocardial Infarction (PAMI-II) Trial Investigators,
- Stone G.W.,
- Marsalese D.,
- Brodie B.R., et al
- Stone G.W.,
- Brodie B.R.,
- Griffin J.J,
- et al.
- Stone G.W.,
- Brodie B.R.,
- Griffin J.J,
- et al.
- Air PAMI Study Group,
- Grines C.L.,
- Westerhausen D.R.,
- Grines L.L., et al
- Lee K.L.,
- Woodlief L.H.,
- Topol E.J,
- et al.
- Grines C.L.,
- Ellis S.G.,
- Jones M,
- et al.
- Pfeffer M.A.,
- Braunwald E.
- Hochman J.S.,
- Choo H.
- Lamas G.V.,
- Flaker G.C.,
- Mitchell G,
- et al.
- White H.D.,
- Cross D.B.,
- Elliott J.M,
- et al.
- Lenderink T.,
- Simoons M.L.,
- Van Es G.-A,
- et al.