Author + information
- aAtlanta VAMC and Emory Clinical Cardiovascular Research Institute, Atlanta, Georgia
- bStatistics and Consulting Unit, Department of Mathematics and Statistics, Boston University, Boston, Massachusetts
- ↵∗Reprint requests and correspondence:
Dr. Peter W.F. Wilson, Atlanta VAMC and Emory Clinical Cardiovascular Research Institute, 1462 Clifton Road, Atlanta, Georgia 30322.
In this issue of the Journal, McNamara et al. (1) presented results from the ACTION (Acute Coronary Treatment and Intervention Outcomes Network) Registry concerning in-hospital mortality following myocardial infarction (MI). The investigators analyzed data from >240,000 patients hospitalized for MI in the United States between 2012 and 2013. Their multivariable analyses showed, with a high degree of precision and internal validation, that heart rate, systolic blood pressure, life-threatening presentations (cardiac arrest, cardiogenic shock, or heart failure [HF]), ST-segment elevation myocardial infarction (STEMI) at presentation, creatinine clearance, and the troponin ratio in the blood were highly associated with death during hospitalization. The investigators developed a risk algorithm that might guide clinicians in terms of prognostication and care of such patients in the future.
The 2016 ACTION Registry report concerning in-hospital survival following MI provided contemporary data demonstrating the progress made in the care of patients with MI over the past decade. For example, in 2000, the TIMI (Thrombolysis In Myocardial Infarction) risk score was developed for patients presenting with unstable angina or non–ST-segment elevation myocardial infarction (NSTEMI) (2). The key components of the TIMI score are age older than 65 years, 3 cardiac risk factors (e.g., family history of coronary artery disease, hypertension, hypercholesterolemia, diabetes, or current smoking), history of coronary stenosis >50%, STEMI, presence or absence of 2 recent angina events, aspirin use in the past 7 days, and positive cardiac biomarkers. The TIMI investigators had older troponin assay information and did not assess the role of presenting for care after cardiac arrest, cardiogenic shock, or HF.
The McNamara et al. (1) results demonstrated the dynamic nature of health risk appraisals. In the setting of patients hospitalized for acute coronary syndrome (ACS) or MI, the clinician has extensive information related to risk factors, recent medications, a history of coronary disease, and severity of clinical presentation. Each of these domains can be assessed and aid in developing risk models that may provide usefulness to clinicians concerning medications and interventions.
Starting with risk factors, the ACTION investigators found that sex, diabetes mellitus, hypertension, previous HF, previous percutaneous coronary interventions (PCI), and previous coronary artery bypass grafting were related to in-hospital mortality in age-adjusted analyses, but in multivariable analyses, the key risk factor associations were statistically significant for age, heart rate, systolic pressure, and creatinine clearance. This is substantially different from the risk factors considered in the TIMI database, which included family history of coronary disease, hypertension, hypercholesterolemia, diabetes, or current smoking.
Patients in the ACTION Registry who presented as severely ill (after cardiac arrest, in cardiogenic shock, or in HF) experienced high risks for in-hospital death. These modes of presentation constitute the greatest challenge to improving survival, and use of this information differs from many previous algorithms. The ACTION Registry evaluation included contemporary troponin levels, which is the assay that has become the standard to diagnose NSTEMI (3,4), and the investigators reported a highly significant association between the troponin ratio (the comparator is the upper limit of normal for troponin) and increased hospital mortality. Use of pre-hospital and in-hospital medications and procedures known to improve survival at time of MI was common in the ACTION Registry. As McNamara et al. showed, patient management commonly included aspirin, beta-blockers, thienopyridines (e.g., clopidogrel or prasugrel), and PCI or thrombolysis.
Various prognostic scores have been developed for use in patients evaluated for ischemic symptoms in the hospital (examples in Table 1). For example, investigators from the GRACE (Global Registry of Acute Coronary Events) study developed a risk score based on >24,000 patients admitted with a presumptive ACS diagnosis across numerous geographic regions worldwide (5). The participants in GRACE were followed up to 6 months with a variety of surveillance methods. The investigators reported that ST-segment elevation on the electrocardiogram, positive troponin levels, serum creatinine, heart rate, and HF were related to survival at 6 months following hospitalization. A subsequent GRACE analysis that used a simplified algorithm included 8 factors (age, Killip class, systolic blood pressure, ST-segment deviation, presentation with cardiac arrest, serum creatinine, positive cardiac enzymes, and heart rate) and demonstrated significant associations with worse survival over 6 months of follow-up (6).
Coronary disease risk algorithms have also been developed for patients evaluated in the emergency department (ED), and the HEART (History, ECG, Age, Risk factor, Troponin) Score has been proposed to help guide the care of patients being evaluated for ischemic symptoms in the ED to determine whether a patient should be admitted or sent home. The 5 domains of the HEART score include history, electrocardiographic findings, age, risk factors, and troponin levels. This scoring system has recently been validated, and may be used in the future to guide decisions for early discharge from ED observation units for patients with minimal findings of ischemia (7).
Along the same line of thinking, GRACE investigators have developed an algorithm to help identify individuals who have not developed clinical events, and this approach has been recommended to predict the in-hospital course of NSTEMI or unstable angina (8). Fifteen factors were found to be predictive of no adverse events, including age, Killip class, unstable angina, no hypotension, no ST-segment deviation, no cardiac arrest at presentation, normal creatinine, decreased pulse, no hospital transfer, no history of diabetes, no history of HF, no peripheral arterial disease, no atrial fibrillation, pre-hospital statin use, and no long-term warfarin treatment. Many of the predictive variables in the GRACE algorithm for “freedom from an event” are the same as those used in the ACTION Registry.
Data analyses for the risk models developed by the ACTION Registry generally showed good accuracy and precision. The calibration information demonstrated that patients with a cardiac arrest experienced much greater risk for mortality as the other major groups (STEMI, NSTEMI, or no cardiac arrest). Until now, clinicians and researchers have generally used either the TIMI or GRACE score to guide therapeutic decisions. With the advent of the ACTION score, which appears to be most helpful for patients with moderate-to-severe disease, and the HEART score, which targets care for patients with minimal-to-mild disease, there are other options. Recently, the DAPT (Dual Antiplatelet Therapy) investigators published a prediction algorithm that provides yet another prognostic score to assess risk of ischemic events and risk of bleeding in patients who have undergone PCI (9). The key variables in the DAPT score are age, cigarette smoking, diabetes mellitus, MI at presentation, previous PCI or previous MI, use of a paclitaxel-eluting stent, stent diameter <3 mm, HF or reduced ejection fraction, and use of a vein graft stent.
A comprehensive cross validation and comparison across at least some of the algorithms—TIMI, GRACE, HEART, DAPT, and ACTION—would help at this point. Interventions and decision points have evolved over the past 15 years, and evaluation of relatively contemporary data would be especially helpful. For example, the HEART score is likely to be used in situations in which the negative predictive capabilities are most important. The ACTION score is likely to be most useful in severely ill patients and to provide guidance for newer interventions. If detailed information concerning stents is available, then the DAPT score should prove helpful.
It is likely that 1 score does not fit all. Each algorithm provides a useful summary of risk to help guide decision-making for patients with ischemic symptoms, depending on the severity of the signs and symptoms at presentation and the duration of the follow-up interval. Consensus building would help to move this field forward for hospital-based management of patients evaluated for cardiac ischemia.
↵∗ Editorials published in the Journal of the American College of Cardiology reflect the views of the authors and do not necessarily represent the views of JACC or the American College of Cardiology.
Both authors have reported that they have no relationships relevant to the contents of this paper to disclose.
- American College of Cardiology Foundation
- McNamara R.L.,
- Kennedy K.F.,
- Cohen D.J.,
- et al.
- Roe M.T.,
- Messenger J.C.,
- Weintraub W.S.,
- et al.
- Mahler S.A.,
- Riley R.F.,
- Hiestand B.C.,
- et al.
- Brieger D.,
- Fox K.A.,
- Fitzgerald G.,
- et al.