Author + information
- Received June 26, 2001
- Revision received November 21, 2001
- Accepted December 11, 2001
- Published online March 6, 2002.
- Yaron Elad, MD*,
- William J French, MD, FACC*,* (, )
- David M Shavelle, MD†,
- Lori S Parsons, BS‡,
- Mark J Sada, MD, FACC*,
- Nathan R Every, MD, MPH, FACC†,
- Participants in the National Registry of Myocardial Infarction 2
- ↵*Reprint requests and correspondence:
Dr. William J. French, Division of Cardiology, Box 405, Harbor-UCLA Medical Center, 1000 W. Carson Street, Torrance, California 90509 USA.
Objectives This study was designed to compare the in-hospital outcome of patients presenting with >12 h from onset of chest pain and acute ST elevation myocardial infarction (AMI) who received either immediate invasive or conservative therapy.
Background The benefits of fibrinolytic therapy diminish in patients presenting with AMI and onset of chest pain >12 h. Primary angioplasty has been suggested as a possible treatment for such patients, but they have been excluded from most trials of primary angioplasty. It remains unclear if an invasive treatment strategy is beneficial to these patients.
Methods Patients presenting with >12 h of chest pain and AMI were identified from the National Registry of Myocardial Infarction 2 database. Patients receiving invasive therapy <6 h after hospital admission were compared with those receiving conservative therapy. Short-term outcomes were compared on the basis of the initial therapy received. To help control for baseline differences in the groups, patients were matched with controls by propensity score methodology.
Results On preliminary analysis, in-hospital outcome was improved in terms of recurrent ischemia, angina, myocardial infarction and mortality in patients receiving initial invasive therapy (odds ratio [OR] = 0.67; 95% confidence interval [CI] 0.49 to 0.92 for mortality). After matching by propensity score, the mortality benefit persisted on bivariate analysis (3.5% vs. 5.0%, p = 0.036), though on multivariate analysis, only a non-significant but strong trend toward decreased mortality remained (OR = 0.73; 95% CI 0.53 to 1.01).
Conclusions Patients receiving early invasive therapy had lower risk features on presentation. Selection bias may play an important role in choosing these patients’ course of treatment and their subsequent outcomes. Certain patients presenting with AMI and duration of chest pain >12 h may benefit from early invasive therapy. These patients could be characterized in a randomized trial.
Patients presenting with acute ST elevation myocardial infarction (AMI) and ST-segment elevation on their initial electrocardiogram (ECG) should be considered for reperfusion therapy. This can be achieved by the use of fibrinolytic agents or invasively through the use of coronary angiography and subsequent percutaneous transluminal coronary angioplasty (PTCA) or coronary artery bypass graft (CABG) surgery. A substantial number of patients with AMI have relative or absolute contraindications to fibrinolysis, and many of these patients receive primary PTCA. The most common reasons given for fibrinolytic ineligibility in patients presenting with ST elevation AMI are advanced age, stroke or bleeding risk and presentation with >6 h of chest pain (1). The time to administration of fibrinolytic therapy is a major factor in reducing mortality, with the degree of benefit decreasing sharply when >12 h of chest pain is present (2). Indeed, the American College of Cardiology/American Heart Association Practice Guidelines for the Management of Patients with Acute Myocardial Infarction considers the administration of fibrinolytic therapy to patients presenting >12 h after onset of chest pain a Class IIb (or less well established) indication (3). Although there is a Class I indication for primary PTCA or CABG surgery in this setting, there are few objective data to support this recommendation.
Because no randomized trials have been published, the only way to compare outcomes of patients who are fibrinolytic-ineligible because of chest pain for >12 h and receive either primary PTCA or no immediate reperfusion therapy would be an observational study. The National Registry of Myocardial Infarction provides an ideal database with which to study this issue.
Patients presenting >12 h after onset of chest pain with ST-segment elevation on their initial ECG and no other contraindications to fibrinolytic therapy were identified from the National Registry of Myocardial Infarction 2 (NRMI 2) database. The NRMI is a multicenter, ongoing observational study designed to collect demographic, clinical, treatment and outcome data on patients with acute myocardial infarction (MI). The NRMI 2 collected data from 1,482 hospitals from 1994 through 1998 and included 772,586 patients. Hospitals were encouraged to enroll consecutive patients, regardless of treatment strategy. Because the NRMI is an anonymous observational study, no informed consent is required of participating patients by the institutional review boards at any of the participating institutions. All patients were diagnosed with MI as defined by either: 1) a history and presentation suggestive of MI accompanied by (a) total creatine kinase (CK) or CK-MB greater than or equal to twice the upper limit of the hospital laboratory normal or (b) electrocardiographic evidence of MI or (c) in the absence of definitive cardiac enzyme or electrocardiographic data, alternative enzymatic, scintigraphic, echocardiographic, angiographic or autopsy evidence of MI; or 2) an International Classification of Diseases, 9th revision, clinical modification discharge diagnosis code for AMI, 410.01 through 410.91. Patients were excluded from the study if they either: 1) were transferred from the NRMI 2 hospital; 2) received fibrinolytic therapy; 3) were in cardiogenic shock (Killip class IV); or 4) were treated in hospitals that could not perform coronary angiography, angioplasty and/or CABG surgery.
Study variables included in this analysis are listed in Table 1. Patients who received initial invasive therapy were defined as those who underwent cardiac catheterization, PTCA or CABG within 6 h of presentation to the hospital.
All analyses were performed using SAS Version 8.0 (SAS Institute, Cary, North Carolina). Differences between the patients who received initial invasive therapy and those who received conservative therapy were evaluated with the Student ttest for continuous variables and the chi-square test for categorical variables. A stepwise, multivariate logistic regression analysis was performed to predict hospital mortality. The following variables were chosen as independent predictors of hospital mortality and allowed to enter the model: age, gender, white race, history of MI, history of congestive heart failure (CHF), previous PTCA, previous CABG surgery, diabetes mellitus, smoking history, Killip class, pulse >100 beats/min, systolic blood pressure ≤100 mm Hg, time from symptom onset to first ECG, the use of aspirin, intravenous unfractionated heparin, intravenous or oral beta-blockers, calcium channel blockers (all medications within the first 24 h) and anterior location of MI. An indicator variable for initial invasive therapy versus conservative therapy was forced into the model. Odds ratios (OR) and 95% confidence intervals (CI) were calculated.
Because of the substantial differences in baseline characteristics between the treatment groups, propensity score methodology was used to identify comparable patients treated with each strategy. The propensity score was initially proposed by Rosenbaum and Rubin (4)and has been used in prior observational studies to help adjust for treatment selection bias (5,6).
In this study, the treatment selection is for initial invasive therapy, and factors that predicted which patients were significantly (p < 0.05) more likely to receive this treatment were first identified. These included white race, previous PTCA, chest pain at presentation and admission diagnosis of MI. Factors that predicted which patients were significantly (p < 0.05) less likely to receive invasive therapy were also identified and included older age, history of prior MI, history of CHF, being transferred in from an outside hospital and hospital discharge in 1994, 1995 or 1996 (when compared with 1997 as the reference year). A stepwise, multivariate logistic regression analysis was then performed to predict initial invasive therapy. The predicted probability of receiving initial invasive therapy (the propensity score) was calculated for each patient. Patients with the same propensity score in either treatment group therefore had the same probability of receiving initial invasive therapy based on the factors identified above. Patients receiving initial invasive therapy (cases) were then matched to patients receiving conservative therapy (controls) on propensity score using the nearest available pair matching method. If more then one unmatched control matched to a case, the control was selected at random. Cases were first matched on five digits of the propensity score. For those that were left unmatched, matching continued on four digits, then three, then two and then one digit of the propensity score. Ninety percent of the cases matched to a control. Bivariate differences between the matched pairs were evaluated using the signed tank test for continuous data and the McNemar’s test for binary data. After the matched-pair analysis, the original multivariate logistic regression model to predict hospital death was re-run with the propensity score forced in. Odds ratios and 95% CIs were calculated.
A total of 7,358 patients met inclusion criteria for this study (Table 1). From this group 1,631 patients (22%) were treated with initial invasive therapy and 5,727 (78%) received conservative therapy. Of the patients in the invasive group, 1,290 (79%) had primary PTCA.
Patients in the invasive group were younger, more likely to be men, more likely white and more likely to have had a history of smoking and prior PTCA (Table 1). They were less likely to have had a prior history of CHF, diabetes mellitus or CABG. Although presenting systolic blood pressure was similar between the two groups, patients who received invasive therapy were less likely to be tachycardic, more likely to be in Killip class I and had lower risk MI features, as defined by the Thrombolysis In Myocardial Infarction (TIMI) (7)and Primary Angioplasty in Myocardial Infarction (PAMI) (8)trials (Table 1).
Patients in the invasive group were more likely to receive intravenous heparin, aspirin and intravenous beta-blockers. They were less likely to receive oral beta-blockers. There was no statistically significant difference in the use of calcium channel blockers.
The number of procedures, complications and clinical outcomes of both patient groups is shown in Table 2. Patients in the invasive therapy group were less likely to have recurrent ischemia or angina and recurrent MI than the patients receiving conservative therapy. There was no significant difference in the incidence of stroke between the two groups. Major bleeding (causing hemodynamic compromise) was more likely to occur in the patients who underwent initial invasive therapy (Table 2). It should be noted that 72% of the patients in the conservative therapy group eventually underwent cardiac catheterization during their initial hospitalization, but most of these procedures occurred >24 h after the onset of chest pain.
Comparison of groups matched by propensity analysis
As expected, baseline differences between the treatment groups were less pronounced when the patients were matched using the propensity score. As shown in Table 3, patients receiving conservative therapy were more likely to have had a slightly longer duration of chest pain and had higher rates of non–Q-wave MI. They were less likely to receive intravenous heparin and more likely to receive oral beta-blockers. Excluding these factors, there were no other statistically significant (p < 0.05) differences between the groups.
The number of procedures, complications and clinical outcomes of both patient groups matched by propensity score is shown in Table 4. Similar proportions of patients underwent coronary angiography, PTCA and CABG as in the non-propensity score matched groups. In addition, the rates of complications and the clinical outcomes were also similar. The lower rate of recurrent ischemia in the invasive therapy group remained statistically significant; however, the previously seen improvement in the rate of recurrent MI was no longer statistically significant.
In the non-propensity analysis, unadjusted hospital mortality was significantly lower in patients treated with invasive therapy (3.4% vs. 6.6%, p < 0.001). To help adjust for baseline differences between patient groups, we performed multivariate logistic regression models (Fig. 1). Factors associated with higher hospital mortality included older age (OR = 1.70 per 10-year increment; 95% CI 1.54 to 1.89), prior history of CHF (OR = 1.373; 95% CI 1.01 to 1.87), presentation in Killip class II (OR = 2.01; 95% CI 1.56 to 2.58) and presentation in Killip class III (OR = 1.84; 95% CI 1.26 to 2.7). Patients presenting with systolic blood pressure ≤100 mm Hg (OR = 3.15; 95% CI 2.33 to 4.27), or pulse >100 beats/min (OR = 1.69; 95% CI 1.33 to 2.15) also had higher hospital mortality. Lower hospital mortality was seen in patients who smoked (OR = 0.70; 95% CI 0.52 to 0.94) received aspirin (OR = 0.57; 95% CI 0.44 to 0.74) or received oral beta-blockers (OR = 0.61; 95% CI 0.48 to 0.77) within 24 h of the diagnosis of MI. Patients with non–Q-wave MI also had lower mortality (OR = 0.64; 95% CI 0.49 to 0.83). After adjustment for these factors, there remained an association between receiving initial invasive therapy and lower hospital mortality (OR = 0.67; 95% CI 0.49 to 0.92).
Hospital mortality in the groups matched by propensity analysis
In the groups matched by propensity analysis, the unadjusted hospital mortality remained significantly lower in patients treated with invasive therapy (3.5% vs. 5.0%, p = 0.036). After matching by propensity score, the odds ratios for older age, Killip classes, pulse >100 beats/min, and systolic blood pressure ≤100 mm Hg continued to predict higher mortality for patients with these factors (OR all >1, Fig. 2). The favorable odds ratios for smoking status, aspirin, oral beta-blockers and non–Q-wave MI also persisted (OR < 1.0, Fig. 2). A history of CHF was no longer associated with increased mortality and dropped out of the multivariate analysis with a neutral effect on hospital mortality. There was a trend toward decreased hospital mortality in patients receiving initial invasive therapy (OR 0.731), and although its CI of 95%—which just crossed 1.0 (range 0.529 to 1.009)—was not statistically significant, it suggests a strong trend favoring an early invasive strategy.
On preliminary analysis, these findings suggest that an initial invasive approach may be more beneficial than medical therapy alone in patients presenting with chest pain for >12 h and ST elevation AMI. A statistically significant mortality benefit was apparent in patients who received invasive therapy within 6 h of hospital presentation with a nearly twofold increase in survival to hospital discharge. This effect persisted on multivariate logistic regression analysis. After subject matching based on their propensity to receive invasive therapy, patients in the invasive therapy group still had lower rates of recurrent ischemia. Although the rates of recurrent MI and in-hospital mortality remained lower in the invasive therapy group, these were no longer statistically significant when compared with rates for propensity score matched patients who received only medical therapy, though a strong trend was present.
Patients in the invasive therapy group were less likely to have had prior histories of diabetes mellitus, MI, CHF and CABG; they were more likely to have had prior PTCA and more likely to have been smokers. Aside from a history of CABG, these differences were no longer statistically significant after matching by propensity score. These findings indicate that selection bias may play a significant role in determining the choice of therapy and outcome in this group of patients.
Of interest, patients in the medical therapy group appeared to be more acutely ill upon presentation, with more in Killip Class II and in either TIMI or PAMI high-risk subsets. Physician bias may favor initial invasive therapy in lower risk patients. The patients in the invasive therapy group also received more complete medical therapy than patients in the conservative therapy group. A significantly larger proportion of patients in the invasive therapy group received aspirin, heparin and (intravenous but not oral) beta-blockers. Both aspirin (9)and beta-blockade (10)have been shown to significantly decrease AMI mortality, which might explain at least a portion of the mortality benefit seen in our data. Thus, although the early use of invasive therapy may lead to substantial benefits in the care of patients with AMI, it may also be a marker for the use of other beneficial therapies. Indeed, the use of invasive therapy would certainly require the involvement of cardiologists, who have been shown to be more likely than non-cardiologists to utilize beneficial medical therapies (11).
Timing and fibrinolyis
On the basis of findings from large randomized trials, most clinicians have limited the use of fibrinolytic therapy to patients who present within 12 h of the onset of chest pain. The Fibrinolytic Therapy Trialists’ Collaborative Group reviewed the data from nine trials that randomized more than 1,000 patients with suspected MI between fibrinolytic therapy and control (2). An excess of deaths was noted in patients presenting with >12 h of chest pain since symptom onset, and statistically uncertain overall mortality benefit was seen for those presenting at 13 to 18 h. The Global Utilization of Streptokinase and Tissue Plasminogen Activator for Occluded Coronary Arteries (GUSTO-I) investigators similarly noted increasing mortality, as well as an increase in the frequency of shock, heart failure and stroke with increasing delay in the administration of fibrinolytic therapy (12). Thus, a perceived risk to benefit quotient has evolved that favors fibrinolysis until 12 h of chest pain, with the perceived risks outweighing the benefit for patients presenting with more prolonged chest pain.
Timing and angioplasty
Primary PTCA has compared favorably with fibrinolytic therapy in the treatment of AMI in a number of randomized controlled trials (8,13,14). These trials largely excluded patients who were fibrinolytic ineligible for various reasons, with the presence of chest pain for >12 h among them. Although it is widely assumed that these patients benefit from late reperfusion therapy with PTCA, this issue has not been well studied. A study of 139 patients presenting with 6 to 48 h of chest pain with a median time to angioplasty of 15 h found a 5.5% in-hospital mortality rate in patients who had successful primary PTCA (15). Mortality was substantially higher when angioplasty was unsuccessful, or when performed in the presence of cardiogenic shock, advanced age or an ejection fraction ≤30%. However, there was no control group in this trial.
The MATE trial (Medicine vs. Angiography in Thrombolytic Exclusion) randomized fibrinolytic ineligible patients with acute coronary syndromes to early triage angiography versus conventional medical therapy (16). The majority of patients were ineligible for fibrinolysis because of ECG ineligibility (lack of ST elevation). However, approximately half of the patients were ineligible because they presented with chest pain for >6 h. There was no statistically significant in-hospital mortality benefit, though the composite in-hospital end point of recurrent ischemia, reinfarction or death was statistically significant, with a benefit seen for the triage angiography group. There was no statistically significant benefit in the composite end point of recurrent infarction or death on long-term follow-up.
A study of the importance of time to reperfusion in 1,352 consecutive patients with AMI treated by primary angioplasty included 275 patients who underwent primary PTCA >6 h after symptom onset (17). Although the recovery of left ventricular function decreased with increasing reperfusion time, patients treated after >6 h of chest pain had recovery of left ventricular function that was better than expected. This was felt to be secondary to an increased incidence of collateral flow to the infarct region, a viewpoint that is supported by data from other trials (18,19). Primary angioplasty >6 h after the onset of chest pain appears to prevent left ventricular dilatation, facilitates favorable ventricular remodeling and possibly prevents the onset of heart failure in the long run (20,21).
The presence of persistent or stuttering chest pain also appears to influence the subsequent benefit of late primary PTCA. One study reported on a small group of patients who were felt to have persistent perfusion to the infarct zone and underwent primary PTCA within 24 h of presentation (18). Ejection fraction rose significantly in these patients after the procedure, and nearly 70% had residual flow to the infarct zone whether from the infarct-related artery or by collaterals.
A recent report from two German databases compared outcomes in patients receiving primary angioplasty versus conservative management in patients with AMI and a pre-hospital delay between 12 and 24 h (22). The study reported on only 94 patients in the primary angioplasty group, and an initial mortality benefit seen on univariate analysis was no longer statistically significant after multiple logistic regression. The authors attributed the loss of statistical significance to either inadequate study power or a selection bias. Our data report on a much larger cohort, and indeed, the multivariate analysis still showed a statistically significant mortality benefit. However, once we controlled for selection bias with the propensity analysis, our results too were no longer statistically significant.
There are several limitations to our study. The NRMI database is an observational study, collecting data from various hospitals across the U.S. Inherent to this are issues relating to retrospective collection and analysis of data, differences in physician practice and reporting as well as the subjective nature of some of the presenting and outcome measurements. Several important measures of patient outcome were either not collected or not available for inclusion in this study. These include socioeconomic status, amount and fluctuation of ST elevation and chest pain, measurement of troponin T or I and the use of newer agents such as low-molecular-weight heparins, glycoprotein IIb/IIIa inhibitors and stents. Several of these measures have been shown to influence subsequent patient outcome and prognosis. In addition, no long-term follow-up was available for this analysis, because the NRMI database includes only in-hospital data.
We observed an association between invasive therapy and lower hospital mortality in patients presenting late with >12 h of chest pain and AMI. Patients receiving invasive therapy had lower risk features on presentation, perhaps as a result of physician bias, and received better medical therapy, perhaps as a result of specialist care. The mortality benefit persisted after multivariate logistic regression analysis, and though it was not statistically significant, a strong trend remained after subject matching by propensity to receive invasive therapy. Primary angioplasty may be beneficial in certain patients presenting late after onset of chest pain, but this clinically important hypothesis should be tested in a randomized prospective trial.
☆ The National Registry of Myocardial Infarction is supported by Genentech, Inc., South San Francisco, California.
- acute ST elevation myocardial infarction
- coronary artery bypass grafting
- congestive heart failure
- confidence interval
- creatine kinase
- Global Utilization of Streptokinase and Tissue Plasminogen Activator for Occluded Coronary Arteries
- myocardial infarction
- National Registry of Myocardial Infarction
- odds ratio
- Primary Angioplasty in Myocardial Infarction
- percutaneous transluminal coronary angioplasty
- Thrombolysis In Myocardial Infarction
- Received June 26, 2001.
- Revision received November 21, 2001.
- Accepted December 11, 2001.
- American College of Cardiology Foundation
- Fibrinolytic Therapy Trialists’ (FTT) Collaborative Group
- Ryan T.J,
- Antman E.M,
- Brooks N.H,
- et al.
- Rosenbaum P.R,
- Rubin D.R
- Hillis L.D,
- Forman S,
- Braunwald E
- Casale P.N,
- Jones W.L,
- Wolf F.E,
- Pei Y,
- Eby L.M
- Newby L.K,
- Rutsch W.R,
- Califf R.M,
- et al.
- Michels K.B,
- Yusuf S
- The Global Use of Strategies to Open Occluded Coronary Arteries in Acute Coronary Syndromes (GUSTO IIb) Angioplasty Substudy Investigators
- McCullough P.A,
- O’Neill W.W,
- Graham M,
- et al.
- Brodie B.R,
- Stuckey T.D,
- Wall T.C,
- et al.
- Hirayama A,
- Adachi T,
- Asada S,
- et al.
- Horie H,
- Takahashi M,
- Minai K,
- et al.
- Zahn R,
- Schiele R,
- Schneider S,
- et al.