Global Registry of Acute Coronary Events (GRACE) hospital discharge risk score accurately predicts long-term mortality post acute coronary syndrome

Global Registry of Acute Coronary Events (GRACE) hospital discharge risk score accurately predicts long-term mortality post acute coronary syndrome Eng Wei Tang, MMed,a Cheuk-Kit Wong, MD,a and Peter Herbison, MScb Dunedin, New Zealand

Background The Global Registry of Acute Coronary Events (GRACE) hospital discharge risk score (GRACE score) developed from a multinational registry involving all subsets of acute coronary syndrome (ACS) predicted 6-month survival. There is currently no validated risk model to predict mortality beyond 6 months. Methods and Results Of the 1143 consecutive patients with ACS admitted to coronary care unit in 2000 to 2002 (mean age, 64.9 F 12.6 years), 39% had ST-elevation myocardial infarction, 39% had non–ST-elevation infarction, and 22% had unstable angina. The mortality was 7.5% during index admission, 12.1% at 6 months, 14.8% at 1 year, 18.7% at 2 years, 25.0% at 3 years, and 39.2% at 4 years. The GRACE hospital discharge risk score calculated for 1057 hospital survivors discriminated survival from death at 6 months (C index, 0.81), 1 year (C index, 0.82), 2 years (C index, 0.81), 3 years (C index, 0.81), and 4 years (C index, 0.80). The risk score worked for all 3 subsets of ACS at all time points, with C index N0.75 in all analyses. A separate multivariable mortality model for these 1057 patients over the 4 - years follow-up period identified 10 independent predictors of mortality. Seven were in the GRACE risk model (age, history of ischemic heart disease, heart failure, increased heart rate on admission, serum creatinine level, evidence of myonecrosis, not receiving in-hospital percutaneous coronary intervention). Conclusions The GRACE postdischarge risk score contains relevant prognostic factors and accurately discriminate survivors from nonsurvivors over the longer term (up to 4 years) in all subsets of ACS patients. (Am Heart J 2007;153:29235.)

The need to risk stratify acute coronary syndrome (ACS) is widely accepted. Risk stratification should be performed early, at the time of hospital admission for short-term (V30 days or in-hospital) adverse outcome1 - 3 and later on at discharge for longer-term prognosis. Many risk score algorithms were derived from clinical trials, such as the Thrombosis In Myocardial Infarction (TIMI)-111 and Platelet glycoprotein IIb/IIIa in Unstable angina: Receptor Suppression Using Integrilin Therapy (PURSUIT)2 trials, which had restricted entry criteria. Patients with comorbidities such as renal impairment were often excluded, and in the TIMI-111 trial, patients

From the aCardiology Division, Department of Medicine, Dunedin School of Medicine, University of Otago, Dunedin, New Zealand, and bStatistics Division, Department of Preventive and Social Medicine, Dunedin School of Medicine, University of Otago, Dunedin, New Zealand. Doctor E.W. Tang received support from The Cardiac Society of Australia and New Zealand/MSD Fellowship. Submitted May 16, 2006; accepted October 9, 2006. Reprint requests: Cheuk-Kit Wong, MD, FRCP, FRACP, FACC, Dunedin School of Medicine, University of Otago, Dunedin Public Hospital, 9001 Dunedin, New Zealand. E-mail: [email protected] 0002-8703/$ - see front matter n 2007, Mosby, Inc. All rights reserved. doi:10.1016/j.ahj.2006.10.004

with heart failure were also underrepresented. In contrast, the Global Registry of Acute Coronary Events (GRACE)4 hospital discharge prediction model for 6-month mortality (GRACE score) was developed from a multinational registry (94 hospitals, 14 countries, and 22,645 patients) involving all subsets of ACS including ST-segment elevation myocardial infarction (STEMI), non–ST-segment elevation myocardial infarction (NSTEMI), and unstable angina from April 1999 to March 2002. The GRACE score derived from clinical parameters at the time of hospital discharge was found to accurately predict mortality at 6 months4 and may have broader applicability in the longer term. Parameters in this GRACE risk model are age, history of congestive heart failure, history of myocardial infarction, elevated resting heart rate, low systolic blood pressure on arrival, ST-segment depression, elevated initial serum creatinine, elevated cardiac enzymes, and not having in-hospital percutaneous coronary intervention (Appendix A). Each parameter is scored, and the summed GRACE risk score corresponds to an estimated probability of all-cause mortality from hospital discharge to 6 months.4 Although the GRACE model may reflect dreal-lifeT practice enabling clinicians to risk stratify patients more accurately than the TIMI or PURSUIT

American Heart Journal January 2007

30 Tang, Wong, and Herbison

model,5 its validity beyond 6 months has not been established. In this study, we examined whether this GRACE risk score4 calculated at hospital discharge would predict longer-term (up to 4 years) mortality in a separate registry cohort and compared the GRACE model with the mortality model derived directly from the cohort.

Methods This is a retrospective study including consecutive patients with ACS admitted into 2 related centers in New Zealand, including the tertiary teaching hospital in Dunedin, Otago, and the regional hospital in Invercargill, Southland, from the years 2000 to 2002. Patients having ACS precipitated by significant noncardiac comorbidity, trauma, or surgery were excluded. This study protocol was in accordance with the local hospital research guidelines. All clinical data were collected by a research physician (Table I), which includes

1. Baseline characteristics: age, sex, cardiac risk factors,

2. 3. 4. 5. 6.

history of ischemic heart disease, history of stroke, peripheral vascular disease, smoking, time-to-presentation, and door-to-needle time for thrombolysis. Presenting clinical features: heart rate, blood pressure, Killip class, episodes of cardiac arrest on arrival, and cardiogenic shock. Electrocardiogram (ECG) characteristics: degree of ST deviation and T-wave changes in the initial ECG. Laboratory findings: initial and maximum troponin rise, creatinine level on arrival. Left ventricular function on echocardiography or left ventriculography during cardiac catheterization. Treatment: in-hospital medications in the first 24-hours, reperfusion and revascularization therapy, and the use of intra-aortic balloon pump.

Death was defined as all-cause mortality during hospitalization and over the 4-year follow-up period. Information on the deaths (until April 1, 2005) was obtained from medical records and the national death registry.

Definition of ACS Patients with ACS were classified into the following:

1. ST-segment elevation myocardial infarction: defined as having ST-segment elevation z1 mm in 2 contiguous leads (or z2 mm in V1 to V3 leads) or new left bundle branch block together with chest pain for N30 minutes and/or evidence of myonecrosis with elevated troponin I (Abbott AxSYM assay) z2.0 Ag/L. 2. Non–ST-segment elevation myocardial infarction: defined as no ST-segment elevation on ECG despite elevated troponin I (Abbott AxSYM assay) z2.0Ag/L and chest pain for more than 30 minutes. 3. Unstable angina: defined as ischemic chest pain lasting more than 30 minutes with no evidence of myonecrosis or ST elevation.

Table I. Baseline characteristics, investigations, in-hospital treatments, and complications of the hospital survivors with ACS Demographics Age, mean (SD) Men Medical history History of ischemic heart disease History of hypertension Diabetes mellitus Dyslipidemia Coronary artery bypass graft Family history of coronary disease History of smoking Prior medical therapy Been on h-blockers Clinical signs at presentation Heart rate, mean (SD), beat/min Systolic blood pressure, mean (SD), mm Hg Diastolic blood pressure, mean (SD), mm Hg Initial serum creatinine level, mean (SD), mmol/L Killip class I II III IV Cardiac arrest Initial troponin I level, mean (SD) Maximum troponin I level, mean (SD) ST-segment depression on arrival for NSTEACS No ST-segment change V1 mm N1 to b2 mm 2 to b3 mm z3 mm Maximum ST-segment depression for NSTEACS No ST-segment change V1 mm N1 to b2 mm 2 to b3 mm z3 mm In-hospital medical treatment during first 24 h Aspirin h-Blockers Statins Heparin (intravenous or subcutaneous) Clopidogrel Glycoprotein IIb/IIIa Lytics In-hospital coronary revascularization PCI CABG In-hospital complications Congestive cardiac failure Cardiogenic shock

64.9 (12.6) 666 (63.0) 493 515 175 678 88 278 579

(46.6) (48.7) (16.6) (64.2) (8.3) (26.3) (54.8)

336 (31.8) 77 (20) 139 (28) 78 (18) 0.104(0.042) 868 141 30 1 77 29.8 210.9

(83.5) (13.6) (2.9) (0.1) (7.4) (139.2) (480.3)

428 116 48 39 18

(66.0) (17.9) (7.4) (6.0) (2.7)

376 127 60 59 27

(57.9) (19.6) (9.2) (9.1) (4.2)

1039 845 317 910 195 136 344

(98.4) (80.0) (30.0) (86.0) (16.5) (12.9) (32.5)

225 (21.3) 123 (11.7) 285 (27.0) 35 (3.3)

Values are number (percentage) unless otherwise indicated. NSTEACS, Non–ST-segment elevation acute coronary syndrome; PCI, percutanous coronary intervention; CABG, coronary artery bypass graft.

Data analysis Each patient’s individual GRACE hospital discharge risk score for all-cause 6-month mortality from hospital discharge was calculated using the published risk score calculator from the GRACE registry (Appendix A). This model uses a patient’s (i) medical history (advanced age, history of congestive heart

American Heart Journal Volume 153, Number 1

Figure 1

Tang, Wong, and Herbison 31

Figure 2

Use of GRACE hospital discharge risk score to predict mortality up to 4 years in the different subsets of ACS. The original data in the GRACE registry4 were included for reference.

failure, and history of myocardial infarction), (ii) findings at hospital presentation (resting heart rate, systolic blood pressure, and ST depression in ECG), and (iii) findings during hospitalization (initial serum creatinine, elevated cardiac enzymes, and in-hospital percutaneous coronary intervention) to work out a total risk score, which would correspond to the likelihood of dying within 6 months of hospital discharge (http://www.outcomes-umassmed.org/grace/). For this study, we used in-hospital congestive cardiac failure instead of history of congestive heart failure and history of ischemic heart disease instead of history of myocardial infarction because these data were likely to be more accurate in the clinical records. In addition, elevated troponin levels were read as a positive score for cardiac enzyme in the GRACE score calculation. We determined survival status for each value of the GRACE score and constructed the receiver operating characteristic curve. This process was repeated for the different follow-up time points (from 6 months to 4 years). The C index (a measure of model discrimination, equivalent to the area under a receiver operating characteristic curve),6 was used to determine the performance of the GRACE hospital discharge risk score in discriminating survival status at all follow-up time points up to 4 years. If a pair of patients, one survivor and one nonsurvivor, were randomly selected, the C index is the probability that the nonsurvivor had a higher GRACE score than the survivor. Thus, a C index of 0.75 refers to a 75% (3 of 4) chance of a correct discrimination of survival status (survivors versus nonsurvivors) in such pairs of patients. Multivariable modeling using backward stepwise Cox proportional hazards regression was also done to create directly a model (the Otago-Southland model) on the association between different variables and mortality over the whole follow-up period. Variables used in developing the GRACE model were included in the first step of the modeling. Variables significant at P V .05 were retained in the final model. The C index from this analysis was determined. We compared this C index with the C index obtained using only the GRACE risk score at all follow-up time points.6

The relationship between GRACE risk score (x-axis) and actual observed mortality (y-axis, as proportion of initial hospital survivors) at different time intervals up to 4 years postdischarge. The dark bold line is the reference line of the 6-month mortality curve predicted by the GRACE model.4 The observed 6-month mortality curve can be seen to be approximating this reference line. To examine the mortality for different ranges of the GRACE score, patients were divided into 3 subgroups. This was tabulated with the vital status at each time point. The confidence interval for the proportion dead was calculated using the binomial distribution.

Results Patients A total of 1143 consecutive patients with ACS (mean age, 64.9 F 12.6 years) including 446 (39.0%) with STEMI, 450 (39.4%) with NSTEMI, and 247 (21.6%) with unstable angina were studied. Among them, 1057 (92.5%) survived hospital admission. The mortality of the 1143 patients was 7.5% in-hospital during index admission, 12.1% at 6 months, 14.8% at 1 year, 18.7% at 2 years, 25.0% at 3 years, and 39.2% at 4 years. This study focused on the 1057 hospital survivors. Their demographic characteristics, risk factors, clinical signs at presentation and initial investigational results, inhospital treatments, procedures, and complications are shown in Table I. Use of GRACE risk score model in predicting mortality The median GRACE risk score is 116, and the interquartile range is 91 to 145. The GRACE risk model4 predicted mortality at 6 months (C index, 0.81), 1 year (C index, 0.82), 2 years (C index, 0.81), 3 years (C index, 0.81), and 4 years (C index, 0.80). Furthermore, the GRACE risk model is robust in

American Heart Journal January 2007

32 Tang, Wong, and Herbison

Table II. Predictors for mortality over the follow-up period Hazard ratio (95% confidence interval)

Table III. Multivariable predictors in the GRACE model and the Otago-Southland model4 P

Otago-Southland Model4

GRACE model Univariable predictors Risk factor Demographics Age, per 10-y increase Male sex Diabetes mellitus Dyslipidemia History of IHD History of hypertension History of CABG Smoking history Regional (vs tertiary) hospital Diagnosis STEMI NSTEMI Unstable angina Presenting characteristics Pulse, per 10 beat/min Systolic blood pressure, per 10-mm Hg increase Diastolic blood pressure, per 10-mm Hg increase Initial creatinine level, per 0.10 mmol/L increase Congestive heart failure Killip class II (vs class I) Killip class III (versus class I) Clinical shock Cardiac arrest on arrival ST-deviation on ECG Maximum TNI4 In-hospital medical treatment during first 24 h Aspirin h-Blockers Lytics In-hospital revascularization Coronary angiography PCI CABG Multivariable predictors Age, per 10-y increase History of IHD Dyslipidemia Smoking history Pulse, per 10 beat/minute Congestive cardiac failure Initial creatinine level, per 0.10 mmol/L increase Maximum TNI; per 100 Ag/L increase In-hospital PCI In-hospital CABG

1.99 0.67 2.24 0.46 2.79 1.15 1.21 1.06 1.50

(1.70-2.33) (0.48-0.94) (1.55-3.22) (0.33-0.65) (1.95-3.99) (0.82-1.60) (0.70-2.10) (0.76-1.47) (1.05-2.14)

b.0005 .019 b.0005 b.0005 b.0005 .415 .501 .747 .025

0.79 (0.58-1.13) 1.51 (1.08-2.10) 0.75 (0.49-1.14)

.196 .015 .176

1.23 (1.15-1.31) 0.95 (0.89-1.01)

b.0005 .089

0.90 (0.82-0.99)

.028

1.27 (1.19-1.35)

b.0005

5.90 4.74 3.97 2.14 1.13 1.12 1.01

b.0005 b.0005 b.0005 .036 .680 .506 .379

(4.17-8.33) (3.32-6.76) (1.99-7.92) (1.05-4.37) (0.62-2.11) (0.80-1.58) (0.98-1.04)

0.34 (0.15-0.76) 0.44 (0.31-0.62) 0.78 (0.54-1.14)

.009 b.005 .198

0.65 (0.38-1.11) 0.29 (0.18-0.48) 0.26 (0.11-0.64)

.115 b.0005 .003

1.43 2.31 0.49 1.84 1.11 3.77 1.13

b.0005 b.0005 b.0005 .002 .005 b.0005 .014

(1.19-1.73) (1.52-3.52) (0.33-0.71) (1.26-2.67) (1.03-1.19) (2.56-5.55) (1.03-1.23)

Age, per 10-y increase History of myocardial infarction History of congestive heart failure Pulse per 30/min increase Initial serum creatinine level per 1-mg/dL increase Initial cardiac enzyme elevation No in-hospital PCI – – – Systolic blood pressure per 20 mm Hg decease ST-segment depression

Age, per 10-y increase History of ischemic heart disease Clinical congestive heart failure Pulse per 10/min increase Initial serum creatinine level per 0.1 mmol/L increase Maximum troponin I elevation No in-hospital PCI No in-hospital CABG History of smoking No history of dyslipidemia – –

4Variables used in developing the GRACE model4 were included in the first step of the backward-stepwise Cox proportional hazards regression in building the OtagoSouthland model. Proportion of patients reaching the follow-up time points was 100% at 2 years, 82% at 3 years and 54% at 4 years. A small proportion (approximately 10%) of patients was excluded because of some missing data field.

Table IV. Comparing the C index of the GRACE model versus the Otago-Southland model at the different time points

6 1 2 3 4

mo y y y y

GRACE model

Otago-Southland model

n4

C index (95% confidence interval)

C index (95% confidence interval)

963 962 962 781 493

0.80 0.81 0.78 0.78 0.76

0.89 0.88 0.85 0.85 0.82

(0.72-0.87) (0.75-0.86) (0.74-0.83) (0.74-0.82) (0.73-0.82)

(0.84-0.93) (0.84-0.93) (0.81-0.89) (0.81-0.89) (0.78-0.87)

P b.0005 b.0005 b.0005 b.0005 b.001

4Because approximately 10% of patients had missing data fields in the construction of the Otago-Southland model, we presented the patients number on which the performance of each model was compared.

IHD, Ischemic heart disease; TNI, troponin I. 4Per 100 Ag/L increase.

Figure 2 shows the relationship between patient’s individual GRACE risk score and their actual mortality among the 1057 hospital survivors at different time points up to 4 years. The GRACE model predicted 6-month mortality curve (reference curve) is drawn using a dark bold line and this approximates well to the observed 6-month mortality curve. All mortality curves at other time points (1, 2, 3, and 4 years) were separating from the standard reference in a graded manner with higher mortality at later time points.

predicting all-cause mortality for all subsets of ACS (STEMI, NSTEMI, and unstable angina) at all analyzed time points postdischarge, with C index N0.75 for all analyses (Figure 1).

The Otago-Southland risk score model The univariable predictors for mortality are shown in Table II. The multivariable Cox proportional hazards regression model identified 10 independent predictors of mortality (Table II). Seven were in the GRACE risk

1.032 (1.002-1.061) 0.39 (0.23-0.67) 0.15 (0.06-0.42)

.033 .001 b.0005

American Heart Journal Volume 153, Number 1

model (age, history of ischemic heart disease, heart failure, increased heart rate on admission, elevated initial creatinine level, evidence of myonecrosis, not receiving in-hospital percutanous coronary intervention) (Table III). The C index for predicting mortality using the OtagoSouthland model was higher than the C index obtained using only the GRACE risk score at all time points from 6 months to 4 years (Table IV).

Discussion This is the first time the GRACE hospital discharge risk score has been independently shown to accurately discriminate survivors from nonsurvivors at different time points up to 4 years in a separate cohort of consecutive patients with ACS. Of note, the mortality discrimination is observed in all 3 subsets of ACS (STEMI, NSTEMI, and unstable angina) at multiple time points from 6 months to 4 years with a C index of N0.75. To ascertain the accuracy of a prediction model, the classical method is to analyze the expected versus the observed event rate over a fixed period. In its original derivation, the GRACE score predicted event rate in the first 6 months post-ACS,4 but there was no information after this time point (ie, the expected event rate was actually unknown). We investigated the use of the GRACE risk score by using C-statistics methods to see if the GRACE score could discriminate between the survivors and the nonsurvivors at later time points and found high C indexes (all N0.75) at all time points up to 4 years. This means a N3 of 4 chances of correctly identifying the survival status using the GRACE score in a pair of randomly selected patients including one survivor and one nonsurvivor. In the separate Otago-Southland model, a higher C index at all time points of between 0.8 and 0.9 was observed, this is an expected finding as the model was constructed directly from the cohort. An important finding is that this bdirect modelQ contains 7 common mortality predictors used in the GRACE model (Table III), again highlighting the prognostic importance of both nonmodifiable baseline characteristics (age, history of myocardial infarction, heart rate on arrival, heart failure, initial serum creatinine, and cardiac enzymes) and therapeutic in-hospital intervention by revascularization. The prognostic importance of an elevated initial serum creatinine on admission with an ACS is noteworthy. In the GRACE registry7 and randomized studies,8-10 renal impairment was more common in older, female patients, and more likely to occur with other comorbidities including hypertension, diabetes mellitus, and cardiac failure. More importantly, renal impairment has been shown to independently predict higher in-hospital7 and short-term (90 days8 and 180 days9) mortality after an ACS, regardless of the ACS

Tang, Wong, and Herbison 33

subset. Of note, in patients with documented left ventricular impairment post–myocardial infarction, even mild renal dysfunction (creatinine clearance [CrCL] b 75 mL/min per 1.73 m2) can be a strong, independent predictor of mortality and cardiovascular complications. This risk increases proportionally with the decline in renal function.10 The GRACE algorithm does not only include renal impairment but also takes it as a continuous variable like age, heart rate, or blood pressure, allowing more refined prognostic prediction. Renal dysfunction predicted long-term mortality as well,10 perhaps a contributing factor to the longer-term prognostic use of the GRACE score. Elevated cardiac troponin is the currently used marker of myonecrosis11 and predicts adverse outcome (death or reinfarction) in patients presenting with ACS.12,13 Both GRACE and Otago-Southland models found that elevated cardiac biomarkers independently predicted poorer survival. However, elevated cardiac enzymes (a dichotomous variable) in the GRACE model4 make up only 15 of a possible 263 points, a relatively low value compared with the weighting of other major factors: age (100/263), resting heart rate (43/263), systolic blood pressure (24/263), history of heart failure (24/263), initial serum creatinine (20/263), and history of myocardial infarction (12/263). Baseline risk considerations are most relevant in interpreting results from clinical trials on ACS using mortality end points. From the GRACE model, a young patient (b70 years old) with no history of myocardial infarction, no clinical heart failure, stable hemodynamics, and normal renal function would have a GRACE score of around 90 (even if cardiac biomarkers are elevated and in-hospital revascularization is not performed) and a low-predicted 6-month mortality of b3%. In these patients, it will be difficult to demonstrate shortterm mortality benefit from newer therapies or interventions/revascularization. In the recent randomized interventional trials for non–ST-elevation ACS (often using elevated troponin or ST-segment changes as major inclusion criteria),14 -17 1-year mortality was consistently less than 5% regardless of treatment arm, with no mortality benefit demonstrated for early invasive approach compared with a more conservative or selectively invasive approach. The low mortality would be consistent with the fact that high-risk patients (age N80, hemodynamically unstable, overt congestive cardiac failure, renal impairment which was sometimes deemed a contraindication to percutanous coronary intervention because of contrast nephropathy) were underrepresented in trials. This also explained the possible drawbacks of the risk scores based on trial cohorts. On the contrary, registry data such as the GRACE and the current cohorts provide a more reallife description of the ACS spectrum including also the truly high-risks patients.

American Heart Journal January 2007

34 Tang, Wong, and Herbison

Although a survival benefit from revascularization over the following years was noted in this cohort, we included only hospital survivors in the analysis, thus excluding those who died after having complications from interventional procedures. This study is based on retrospective data collection on consecutive patients with ACS in Otago and Southland of New Zealand. However, the high C index of the Otago-Southland model in predicting outcome at multiple time points up to 4 years indicates that important prognostic data were captured. Clinical management after discharge would have possibly influenced longterm outcome. Despite not having these data, both risk models have performed well in an unselected real-life ACS cohort. We found that the C index of the GRACE score for all patients with ACS was 0.81 at 6 months, which was identical to that in the GRACE cohort (Figure 1), and importantly, the C index stayed N0.80 at all time points up to 4 years. The GRACE score will be superior to any single individual parameter in providing long-term prognostic information for the patients at the time of discharge from the initial ACS.

Conclusion The GRACE 6-month all-cause mortality risk model can accurately discriminate survivors from nonsurvivors in all subsets of ACS for up to 4 years. We thank the Southland hospital for allowing us to have full access to their patient files.

References 1. Antman EM, Cohen M, Bernink PJ, et al. The TIMI risk score for unstable angina/non-ST elevation MI: a method for prognostication and therapeutic decision making. JAMA 2000;284:835 - 42. 2. Boersma E, Pieper KS, Steyerberg EW, et al. Predictors of outcome in patients with acute coronary syndromes without persistent STsegment elevation. Results from an international trial of 9461 patients. The PURSUIT Investigators. Circulation 2000;101:2557 - 67. 3. Granger CB, Goldberg RJ, Dabbous O, et al. Global Registry of Acute Coronary Events Investigators. Predictors of hospital mortality in the global registry of acute coronary events. Arch Intern Med 2003;163:2345 - 53. 4. Eagle KA, Lim MJ, Dabbous OH, et al. A validated prediction model for all forms of acute coronary syndrome: estimating the risk of 6-month post-discharge death in an international registry. JAMA 2004;291:2727 - 33. 5. de Araujo Goncalves P, Ferreira J, et al. TIMI, PURSUIT, and GRACE risk scores: sustained prognostic value and interaction with revascularization in NSTE-ACS. Eur Heart J 2005;26:865 - 72. 6. DeLong ER, Delong EM, Clarke-Pearson DL. Comparing the areas under two or more correlated receiver operating curves: a nonparametric approach. Biometrics 1988;44:837 - 45. 7. Santopinto JJ, Fox KA, Goldberg RJ, et al. Creatinine clearance and adverse hospital outcomes in patients with acute coronary syndromes: findings from the Global Registry of Acute Coronary Events (GRACE). Heart 2003;89:1003 - 8.

8. Reddan DN, Szczech L, Bhapkar MV, et al. Renal function, concomitant medication use and outcomes following acute coronary syndromes. Nephrol Dial Transplant 2005;20:2105 - 12. 9. Al Suwaidi J, Reddan DN, Williams K, et al. Prognostic implications of abnormalities in renal function in patients with acute coronary syndromes. Circulation 2002;106:974 - 80. 10. Anavekar NS, McMurray JJ, Velazquez EJ, et al. Relation between renal dysfunction and cardiovascular outcomes after myocardial infarction. N Engl J Med 2004;351:1285 - 95. 11. The Joint European Society of Cardiology/American College of Cardiology CommitteeMyocardial infarction redefined—a consensus document of the Joint European Society of Cardiology/ American College of Cardiology Committee for the Redefinition of Myocardial infarction. J Am Coll Cardiol 2000;36:959 - 69. 12. Ottani F, Galvani M, Nicolini FA, et al. Elevated cardiac troponin levels predict the risk of adverse outcome in patients with acute coronary syndromes. Am Heart J 2000;140:917 - 27. 13. Antman EM, Tanasijevic MJ, Thompson B, et al. Cardiac-specific troponin I levels to predict the risk of mortality in patients with acute coronary syndromes. N Engl J Med 1996;335:1342 - 9. 14. de Winter RJ, Windhausen F, Cornel JH, et al. Invasive versus Conservative Treatment in Unstable Coronary Syndromes (ICTUS) Investigators. Early invasive versus selectively invasive management for acute coronary syndromes. N Engl J Med 2005;353:1095 - 104. 15. Wallentin L, Lagerqvist B, Husted S, et al. Outcome at 1 year after an invasive compared with a non-invasive strategy in unstable coronary-artery disease: the FRISC II invasive randomised trial. Lancet 2000;356:9 - 16. 16. Cannon CP, Weintraub WS, Demopoulos LA, et al. TACTICS (treat angina with aggrastat and determine cost of therapy with an invasive or conservative strategy)—thrombolysis in myocardial infarction 18 investigators. Comparison of early invasive and conservative strategies in patients with unstable coronary syndromes treated with the glycoprotein IIb/IIIa inhibitor tirofiban. N Engl J Med 2001;344:1879 - 87. 17. Fox KA, Poole-Wilson PA, Henderson RA, et al. Randomized intervention trial of unstable angina investigators. Interventional versus conservative treatment for patients with unstable angina or non-ST-elevation myocardial infarction: the British heart foundation RITA 3 randomised trial. Lancet 2002;360:743 - 51.

Appendix A The GRACE Risk Calculator for 6-Month Postdischarge Mortality After Hospitalisation for Acute Coronary Syndrome Medical History Age in Years b29 30-39 40-49 50-59 60-69 70-79 80-89 z90 History of Congestive Heart Failure History of Myocardial Infarction

Points 0 0 18 36 55 73 91 100 24 12

American Heart Journal Volume 153, Number 1

Tang, Wong, and Herbison 35

Findings at Initial Hospital Presentation Resting Heart Rate, beats/min b49.9 50-69.9 70-89.9 90-109.9 110-149.9 150-199.9 z200 Systolic Blood Pressure, mm Hg V79.9 80-99.9 100-119.9 120-139.9 140-159.9

0 3 9 14 23 35 43 24 22 18 14 10

160-199.9 z200 ST-Segment Depression Findings During Hospitalization Initial Serum Creatinine, mg/dL 0-0.39 0.4-0.79 0.8-1.19 1.2-1.59 1.6-1.99 2-3.99 z4 Elevated Cardiac Enzymes No In-hospital Percutanous Coronary Intervention

Receive tables of contents by e-mail To receive the tables of contents by e-mail, sign up through our Web site at http://www.ahjonline.com Choose E-mail Notification Simply type your e-mail address in the box and click on the Subscribe button Alternatively, you may send on e-mail message to [email protected] Leave the subject line blank, and type the following as the body of your message: subscribe ahj_toc You will receive an e-mail to confirm that you have been added to the mailing list. Note that TOC e-mails will be sent when a new issue is posted to the Web site.

4 0 11

1 3 5 7 9 15 20 15 14