In the 6 years since the publication of the previous American College of Physicians guidelines on risk stratification after myocardial infarction [1], the management of patients who have myocardial infarction has changed dramatically. Three major factors have prompted this change: the increased availability of empirical data from randomized clinical trials, the proliferation of diagnostic and interventional technologies, and the growth of managed care organizations.

The recent completion of several large clinical trials has not only shown the efficacy of many new therapeutic agents but has supplied rich clinical databases for use in outcomes research. The dramatic spread of diagnostic and therapeutic technologies has also influenced acute coronary care in the United States. For example, the use of cardiac catheterization and revascularization in Medicare patients presenting with acute infarction increased by 45% and 70%, respectively, between 1987 and 1992 [2, 3]. Among 350 775 patients in the United States who were enrolled in the National Registry of Myocardial Infarction between 1990 and 1994, overall rates of catheterization, angioplasty, and bypass surgery were 56%, 20%, and 12%, respectively [4]. The counterbalance to this increase in technology-based medicine has been a growing, nationwide concern about the escalating cost of medical care. In the managed care settings of the 1990s, providers have been encouraged to shorten hospital stays and limit resource utilization.

To continue to provide high-quality care while controlling medical costs in today's health care environment, providers must use an evidence-based strategy to quantify a patient's overall short- and long-term risk. In this position paper, we highlight the four major phases of the hospitalization of a patient with acute myocardial infarction-the acute evaluation phase, the cardiac care unit phase, the hospital phase, and the predischarge phase-and we outline the important risk-stratification processes that should occur during each phase. We also provide a quantitative measure of the strength of published evidence supporting each of our positions. We have defined strength-of-evidence grades (given in parentheses after each position) as follows:

A: Supported by data from large, randomized clinical trials

B: Supported by well-designed clinical studies

C: Supported by a synthesis of small observational reports

D: Supported by consensus and practice norms without empirical documentation

We also provide brief rationales for our positions and refer the reader to the appropriate sections in the accompanying background paper for a more detailed discussion of the evidence. Finally, we have summarized our conclusions into an algorithm for risk stratification after myocardial infarction (Figure 2 and 3 in the background paper).

Acute Evaluation Phase

1. The initial clinical history, physical examination, and electrocardiogram provide critical information for risk stratification of a patient who may be having an acute infarction. Important predictors of outcome include age, hemodynamics (systolic blood pressure and heart rate) at time of presentation, evidence of congestive heart failure on physical examination, location of the infarction, and history of previous infarction (A).

2. Although the acute use of such specialized technologies as echocardiography and perfusion imaging may provide additional information in specific clinical situations, their incremental value in the acute evaluation of patients who may have an infarction has not been proven. Thus, we cannot currently recommend their routine use (C).

3. Rapid triage of reperfusion candidates and patients in cardiogenic shock is necessary to optimize myocardial salvage and, ultimately, the number of lives saved (A).

Rationale

Patients who are suspected of having acute myocardial infarction at presentation must rapidly receive a diagnosis, be accurately stratified for risk, and be appropriately treated for optimal outcome. Many published studies (single-institution studies; community-based registries; and large, randomized clinical trials) have shown that a few key clinical variables can accurately identify patients with infarction who are at increased risk for death [3.1]. The baseline electrocardiogram also provides both diagnostic and prognostic information [3.2], as do serum markers of myocardial damage [3.3]. Studies of specialized acute testing (such as echocardiography and cardiac perfusion imaging) have not shown that these tests add prognostic information to that obtained from the history, physical examination, and electrocardiogram [3.4]. Finally, two groups of patients must be rapidly identified: those who are in cardiogenic shock and those who are candidates for acute reperfusion therapy. The benefits of acute, aggressive intervention in these two groups of patients have been shown in several large studies [3.5, 3.6].

Cardiac Care Unit Phase

1. Patients with clear evidence of myocardial damage should be admitted to an intensive care setting in which staff are trained to interpret hemodynamic and electrocardiographic data and provide rapid therapy, such as defibrillation, if needed (B).

2. Patients receiving thrombolytic therapy should be monitored for clinical signs of reperfusion. Those who do not have reperfusion have a higher risk for death or complications after infarction (B) and may benefit from "rescue" reperfusion therapy (B).

3. Patients who do not have complications within 24 hours after infarction can be moved to a less intensive, but monitored, medical setting (B).

Rationale

The risk for death and complications after acute infarction decreases nonlinearly over time. Therefore, the first 24 to 48 hours after infarction is the period of highest risk; during this time, the patient must be monitored closely for signs of arrhythmia, hemodynamic instability, and other complications [4]. Large, randomized trials show that current thrombolytic strategies result in patency rates in the infarction-related artery of 65% to 85% at 90 minutes. Direct coronary angioplasty, although it has been studied in fewer patients, is successful in 90% to 95% of cases. Patients who do not have reperfusion, however, have a significantly increased risk [4.1]. To date, nonrandomized comparisons have shown that patients who do not have reperfusion may benefit from either repeated administration of thrombolytic therapy or "rescue" angioplasty. In addition, a single small, randomized study showed that patients who had anterior infarction and underwent rescue angioplasty had better outcomes than patients who had anterior infarction and received conservative care. If these results are confirmed in larger randomized studies, determination of the patency of the infarction-related artery will be indicated for optimal patient management. Although it is not highly sensitive, patency of the infarction-related artery can often be assessed on clinical grounds alone, such as relief of chest pain and a recovery of 50% or more in ST-segment elevation.

Hospital Phase

1. The optimal duration of continuous rhythm monitoring is unknown, but this monitoring should continue for at least 24 hours after transfer from an intensive care unit (C).

2. Patients who have recurrent ischemia, high-grade arrhythmia, or congestive heart failure after infarction are at high risk for death, complications, or both (B) and may benefit from cardiac catheterization (D).

3. Patients who have no complications should be considered for early discharge from the hospital. In many clinical settings, these patients may be discharged safely after 4 to 5 days of hospitalization (B).

Rationale

Continuous electrocardiographic monitoring allows faster detection of life-threatening arrhythmias, but it is labor-intensive and expensive and no definitive empirical data support its current duration of use. Clearly, however, the risk for sudden cardiac death is highest during the first 2 days after infarction and declines exponentially thereafter [5] (Figure 1 in the background paper). An aggressive diagnostic and interventional approach seems to be indicated in patients who have congestive heart failure, recurrent ischemia, or high-grade arrhythmia and are considered potential candidates for revascularization. On the other end of the spectrum, patients with no significant complications during the first 4 days after infarction have been shown in some series to have a 30-day mortality rate approaching 1% [5]. Thus, they probably would not benefit from prolonged hospitalization.

Predischarge Phase

1. Left ventricular function should be assessed in all patients who have an infarction (A). This assessment can be done accurately on the basis of clinical factors alone in at least 40% of patients (B).

2. Specialized methods, such as Holter monitoring, signal-averaged electrocardiography, heart rate variability assessment, and programmed electrical stimulation studies, can detect patients at increased risk for sudden cardiac death (B). However, because abnormal results on these tests have not been shown to alter patient management, they are not currently recommended for routine care after an infarction (B).

3. Patients who have markers for residual ischemia on noninvasive stress testing are at increased risk for later infarction or death (B).

4. Adding radionuclide ventriculography, perfusion imaging, or echocardiography to electrocardiographic stress testing increases the sensitivity of electrocardiographic stress testing for the prediction of future cardiac events but decreases the overall specificity of this testing. The incremental value of additional imaging has not been shown to be worth the increased cost (C).

5. Among patients who have uncomplicated infarctions, those with significant left ventricular dysfunction are more likely to have multivessel disease and to incur a long-term survival advantage from coronary revascularization (A). Thus, they may benefit from routine cardiac catheterization.

6. Patients who have uncomplicated infarctions and preserved left ventricular function are at low risk (mortality rate, <3% over the first year after infarction); thus, noninvasive tests can be used safely for risk stratification (B).

7. Screening and intervention for the purposes of secondary prevention, including education about smoking cessation; therapy for hyperlipidemia, hypertension, or diabetes; and cardiac rehabilitation, can significantly improve long-term prognosis and should be part of routine care after an infarction (A). Additional life-prolonging medical therapies (with aspirin, ß-blockers, and angiotensin-converting enzyme inhibitors for patients with significant ventricular dysfunction) should also be instituted in patients who have no clinical contraindications to these therapies (A).

Rationale

Left ventricular function is a powerful predictor of outcomes after infarction, and both medical and surgical interventions can reduce the rate of subsequent cardiac events in patients with left ventricular dysfunction. Thus, left ventricular function should be assessed in all patients who have an infarction, either by using surrogate clinical markers or by using specialized tests (echocardiography, radionuclide ventriculography, or cardiac catheterization) when the clinical evaluation yields unclear results [6.2].

In contrast, interventions proposed for the prevention of sudden cardiac death have been shown to be either ineffective or harmful-for example, anti-arrhythmic agents in the Cardiac Arrhythmia Suppression Trial (CAST) study [6]-or beneficial in almost all subsets of the population with infarction. ß-Blockers, for example, reduce the relative risk for death and other events in most patients who have had myocardial infarction [7, 8]. Because the results of specialized electrocardiographic testing (Holter monitoring, signal-averaged electrocardiography, heart rate variability testing, and invasive electrophysiology) have not been shown to significantly affect clinical management, we cannot recommend their routine use as part of risk stratification after infarction [6.3].

With regard to the assessment of ischemic risk, our review of the literature on noninvasive stress testing showed that although many markers of ischemia indicate an increased likelihood of later cardiac events, the positive predictive values of these markers are low. Studies evaluating the prognostic accuracy of these noninvasive methods have generally been small, retrospective studies that did not adjust sufficiently for baseline risk. Thus, we are unable to state that one noninvasive testing method is clearly superior to another [6.5]. Despite this limited information, the current standard of care should include an exercise electrocardiographic evaluation before discharge.

Few studies have compared the outcomes seen with noninvasive testing strategies with those seen with routine cardiac catheterization. Given that patients with depressed left ventricular function and multivessel coronary disease derive a proven survival benefit from revascularization, an aggressive catheterization strategy appears reasonable for those who have significant ventricular dysfunction if the patient is a candidate for revascularization. Alternatively, given the overall low event rate in patients who have preserved left ventricular function, we believe that a conservative strategy with noninvasive testing is sufficient in these low-risk patients [6.6].

Finally, mounting evidence shows that screening and intervention for the purposes of secondary prevention can improve the long-term outcomes of patients who have myocardial infarction. Secondary cardiac risk-factor modification should include education about smoking cessation; control of hypertension, hyperglycemia, and hypercholesterolemia; improvements in exercise ability; and dietary modification. Other medical and nonmedical interventions that prolong life, including the use of aspirin, ß-blockers, angiotensin-converting enzyme inhibitors (for patients who have left ventricular dysfunction), and cardiac rehabilitation, should also be instituted in all patients who have no contraindications to these interventions. General recommendations for risk modification through the reduction of stress and isolation, hormone replacement therapy (for postmenopausal women), and modification of coagulation factors await the results of ongoing clinical trials [6.7].

Future Work

We have summarized the available empirical data on risk stratification after myocardial infarction into a usable clinical guideline, but many areas clearly need more research. Similarly, as clinical research improves our therapeutic arsenal and our understanding of prognostic indicators, our recommendations will probably require modification. Below, we highlight areas in which research is needed; the results of this research should prompt a reconsideration of our recommendations.

First, much of the available literature that addresses prognosis after myocardial infarction has been drawn from databases from clinical trials. As noted in the accompanying background paper, the spectrum of patients enrolled in clinical studies of myocardial infarction may differ significantly from that in community-based practices. Specifically, we have only limited information about infarctions in elderly persons, women, and minority populations. Similarly, little information has been available for patients who have an infarction without ST-segment elevation; these patients may account for 50% to 75% of infarctions seen in the community. Thus, confirmation of these findings and of our recommendations in more generalizable, community-based registries should be a high priority. Studies such as those by the Cooperative Cardiovascular Project [9] and the National Registry of Myocardial Infarction [4] should contribute greatly.

Second, although considerable observational and clinical trial data have been gathered to identify markers for risk in the clinical history, physical examination, and results of noninvasive testing, less is known about the complex interrelations among these markers. One example is the degree to which noninvasive test results add knowledge to that obtained from the clinical history. Although some studies have examined the independent prognostic value of noninvasive testing beyond that of the initial clinical examination, few have tested whether real-time integration of this information would reduce the need for further diagnostic evaluation. The incremental cost of obtaining additional prognostic information also merits investigation. In many circumstances, a less expensive evaluation may provide similar information. The value of simple clinical predictive rules must not be underestimated. Alternatively, the application of specialized technology, although it is initially more costly, may in some cases limit the length of hospitalization, reduce the need for downstream interventions, or both [10, 11]. Thus, alternate "care maps" or patient management strategies that use these diagnostic tools must be rigorously compared with regard to their overall diagnostic yield and use of resources.

Advances in treatment can also affect risk stratification. The proper use of new medications or interventions whose efficacy is specific for a given subset of patients who have infarction will require that these patients be identified. For example, if a drug or device is found to reduce the risk for death specifically in patients at high risk for sudden cardiac death, then specialized electrocardiographic testing may become a necessary part of the risk stratification algorithm for patients who have infarction. Similarly, if future studies find that acute perfusion imaging or echocardiographic imaging of patients without ST-segment elevation can identify a high-risk subgroup of patients who would benefit from acute reperfusion therapy (such as primary angioplasty), the use of these methods may become necessary in the routine initial evaluation of these patients.

Conclusions

We have attempted to summarize the current state of knowledge about risk assessment after acute myocardial infarction. We have proposed a simple clinical algorithm for this process, emphasizing the need for early and continuous risk assessment throughout hospitalization. We have tried to highlight some of the limitations in our current knowledge as a way to encourage further investigation. As our understanding of these issues, which affect daily clinical decision making, is clarified, these practice guidelines will need to be modified and updated.