Diagnostic Accuracy of Global Registry of Acute Coronary Events (GRACE) Risk Score in ST-Elevation Myocardial Infarction for In-Hospital and 360-Day Mortality in Japanese Patients

Abstract

Background: The purpose of the present study was to confirm the diagnostic accuracy of Global Registry of Acute Coronary Events (GRACE) risk score 1.0 (GRACE 1.0) and updated GRACE 1.0 (GRACE 2.0) for in-hospital and 360-day mortality in ST-elevation myocardial infarction (STEMI) in Japanese patients. GRACE 1.0 and GRACE 2.0 are the established predictive models in acute coronary syndrome, but their application to Japanese patients has not been fully verified.

Methods and Results: The present study retrospectively analyzed 412 consecutive STEMI patients who had undergone primary percutaneous coronary intervention from January 2006 to September 2011. All causes of death during hospitalization were examined to confirm the diagnostic accuracy of GRACE 1.0 on receiver operating characteristic (ROC) analysis. Similarly, all causes of death during the 360 days after hospitalization were analyzed to confirm the diagnostic accuracy of GRACE 2.0. The average GRACE 1.0 score was 175.8±50.9. In-hospital and 360-day mortality were 13.1% and 15.5%, respectively. Area under the ROC curve, which describes the diagnostic accuracy of the GRACE 1.0 predicted in-hospital mortality and the GRACE 2.0 predicted 360-day mortality, was as high as 0.95 and 0.92, respectively.

Conclusions: Both GRACE 1.0 and GRACE 2.0 had a high diagnostic accuracy for prediction of in-hospital and 360-day mortality in Japanese STEMI patients. (Circ J 2014; 78: 2950–2954)

ST-elevation myocardial infarction (STEMI) had a high mortality rate, but emergency coronary revascularization has dramatically improved this, down to 5–7%.^1–4

The Global Registry of Acute Coronary Events (GRACE) risk score (GRACE 1.0) and updated GRACE 1.0 (GRACE 2.0) have been reported to be useful for prediction of short- and long-term mortality in acute coronary syndrome (ACS).^5–7 Although these 2 scoring models consist of 8 simple parameters including the vital signs and examination findings on arrival at hospital, prediction ability is superior. These models are useful not only for predicting patient mortality, but also for identifying more critical patients who require intensive treatment.

The GRACE registry collected data from a worldwide population with a central focus on North America, South America, Europe, Australia, and New Zealand, therefore it has not been verified whether these scores can apply to Japanese patients.

The purpose of the present study was to confirm the diagnostic accuracy of GRACE 1.0 and GRACE 2.0 for in-hospital and 360-day mortality for STEMI in Japanese patients.

Methods

Study Design and Subjects

To study the diagnostic accuracy of GRACE 1.0 and GRACE 2.0 for in-hospital and 360-day mortality in STEMI patients, we retrospectively studied the medical records of STEMI patients who were admitted to Tokai University School of Medicine from January 2006 to September 2011 with acute symptom onset, and who had the diagnosis of STEMI on emergency coronary angiography and following primary percutaneous coronary intervention (PCI). Among 436 STEMI patients identified, the present study reviewed records from 412 consecutive STEMI patients whose clinical course could be tracked after 360 days to evaluate mortality.

The exclusion criteria were as follows: patients who did not undergo emergency coronary angiography to confirm the culprit lesion, and those whose mortality could not be tracked during 360 days after STEMI onset were excluded.

All causes of death during hospitalization were taken into consideration to confirm the diagnostic accuracy of GRACE 1.0 on receiver operating characteristic (ROC) analysis. In the same manner, all causes of death during 360 days after PCI were taken into consideration to confirm the diagnostic accuracy of GRACE 2.0.

Definitions

STEMI was defined as acute when the patient presented within 24 h of symptom onset, and electrocardiography (ECG) on arrival showed persistent ST-segment elevation >1 mm in 2 contiguous leads with reciprocal ST depression, new or presumed new left bundle branch block. The final diagnosis was made on emergency coronary angiography.⁸

The estimated glomerular filtration rate was calculated using the Modification of Diet in Renal Disease (modified for Japanese) equation.⁹

GRACE 1.0 and GRACE 2.0 were calculated from 8 clinical parameters on arrival in hospital: age; heart rate; systolic blood pressure (SBP); serum creatinine; Killip classification; cardiac arrest; ST-segment deviation on ECG; and elevated cardiac enzyme.^10,11 These parameters followed the GRACE definitions. Cardiogenic shock was defined as SBP <80 mmHg. Cardiac arrest was defined as rapid ventricular tachycardia with hemodynamic instability, ventricular fibrillation, electrical mechanical dissociation or asystole and requiring cardiopulmonary resuscitation from onset to admission. ST-segment deviation was defined as ≥1-mm elevation or depression of ST-segment level from the baseline on ECG. Increase of cardiac enzyme was defined as positive troponin I.

If the predicted 360-day mortality as calculated by GRACE 2.0 was given as a range, the lower result was used as the predicted 360-day mortality.

Statistical Analysis

Normally distributed data are given as mean±SD. Skewed data are given as median (interquartile range). Student’s t-test was used to determine statistically significant differences in clinical parameters between 2 different groups. Wilcoxon rank-sum test was used to determine statistically significant differences in clinical values with skewed distribution between 2 different groups. Pearson’s chi-squared test was used to determine statistically significant differences between categorical variables. ROC analysis and the area under the ROC curve (AUC) were used to quantify the ability of GRACE 1.0 and GRACE 2.0 to estimate STEMI mortality. P<0.05 was considered statistically significant. All statistical calculations were performed using JMP version 9 (SAS Institute, Cary, NC, USA).

Results

Baseline characteristics are listed in Table 1. Mean age was 65.9±12.5 years and 79.9% of the patients were male. In initial evaluation on arrival at hospital, the mean clinical parameters were as follows: SBP, 125.5±41.7 mmHg; heart rate, 74.9±26.0 beats/min; Killip IV, 22.6%; and prevalence of cardiogenic shock, 21.6%. ST-segment deviation was observed in all patients because of STEMI. Seven percent of the patients experienced cardiac arrest from onset to arrival at hospital. Average GRACE 1.0 score was 175.8±50.9 points. Comparing the 360-day survivors (n=348) and non-survivors (n=64), 6 components of GRACE 1.0 were significantly different; heart rate and serum creatinine were not. Notably, the presence of cardiac arrest was a strong predictor of high 360-day mortality rate, which was as high as 23.4%. Among 29 patients who experienced cardiac arrest, more than half of them (15 patients, 51.7%) did not survive to discharge. There was a significant difference in GRACE 1.0 score between 360-day survivors and those who did not survive 360 days (162.4±40.4 points vs. 248.5±39.4 points, P<0.0001). Table 2 lists clinical outcome for in-hospital and 360-day mortality. In-hospital and 360-day mortality rates were 13.1% and 15.5%, respectively. For 360-day all-cause death, cardiac death accounted for 71.9%, and cause in 15 non-cardiac deaths was as follows: pneumonia, n=7; sepsis or other infection, n=4; subarachnoid hemorrhage, n=1; abdominal aneurysm rupture, n=1; intrathoracic hemorrhage, n=1; and carcinoma, n=1.

Table 1. Baseline Characteristics and Clinical Status on Arrival

	All (n=412)	Survivors at 360 days (n=348)	Non-survivors in hospital (n=54)	Non-survivors at 360 days (n=64)	P-value¶
Age (years)	65.9±12.5	64.5±12.0	73.4±12.6	73.5±12.2	<0.0001
Male	79.9	79.9	77.8	79.7	0.9711
Height (cm)	162.3±8.4	162.6±8.4	160.6±9.1	161.0±8.6	0.1886
Weight (kg)	63.2±12.5	63.6±12.3	63.6±14.5	63.6±14.0	0.9614
Current smoking	65.3	65.2	61.1	65.6	0.9513
Hypertension	77.9	76.7	87.0	84.4	0.0045
Dyslipidemia	68.7	70.7	55.6	57.8	0.0412
Diabetes mellitus	35.4	33.3	48.1	46.9	0.0374
Insulin	4.6	3.1	11.1	12.5	0.0011
Old MI	10.9	8.6	24.1	23.4	0.0005
Prior PCI	8.3	8.3	7.4	7.8	0.8893
Prior CABG	0.2	0.3	0	0	0.6677
Prior stroke	10.7	10.0	13.0	14.1	0.3404
Hemodialysis	2.4	2.6	1.9	1.6	0.6248
Hemoglobin (mg/dl)	14.1±2.3	14.3±2.1	13.1±2.9	13.0±2.8	0.1409
LDL-C (mg/dl)	124.2±40.6	127.3±40.2	105.5±40.9	106.7±38.9	0.0002
HDL-C (mg/dl)	50.0±15.6	51.0±15.7	43.8±14.1	44.6±13.5	0.0034
Triglyceride (mg/dl)	118.6±116.7	121.8±107.9	101.8±168.2	100.9±156.8	0.1946
Serum creatinine (mg/dl)	1.1±1.3	1.10±1.3	1.3±0.9	1.9±0.9	0.1105
eGFR (ml·min−1·1.73 m−2)	68.9±27.6	72.1±26.2	53.3±30.6	51.8±29.1	<0.0001
BNP (pg/ml)	73 (25.3–234.9)	62.6 (23.6–208.6)	142.1 (48.8–748.5)	178 (53.6–737.2)	<0.0001
Onset to door (min)†	204.0±224.1	209.5±219.8	163.7±260.0	173.2±246.6	0.2444
Onset to reperfusion (min)	309.7±232.1	308.0±222.2	297.7±283.3	319.0±280.3	0.7431
SBP (mmHg)	125.5±41.7	133.3±34.8	74.3±47.3	83.3±50.5	<0.0001
DBP (mmHg)	69.8±32.3	75.6±27.0	29.8±36.9	37.8±39.6	<0.0001
Cardiogenic shock	21.6	12.6	77.8	70.3	<0.0001
Heart rate (beats/min)	74.9±26.0	74.9±22.4	70.2±42.1	74.5±40.5	0.9121
Killip I/II/III/IV (n)	181/89/49/93	179/84/38/47	0/2/9/43	2/5/11/46	<0.0001
Cardiac arrest‡	7.0	4.0	27.8	23.4	<0.0001
Peak CK (IU/L)	4,155.3±4,971.9	3,442.8±3,554.9	9,185.8±9,427.2	8,090.5±9,049.4	<0.0001
LVEF (%)	50.6±13.3	52.8±11.6	34.9±14.3	37.0±14.8	<0.0001
GRACE 1.0 score	175.8±50.9	162.4±40.4	257.0±33.0	248.5±39.4	<0.0001

Data given as mean±SD, %, or median (IQR). ^†Excluded 5 patients with in-hospital onset. ^‡Patients who experienced fatal arrhythmia (pulseless ventricular tachycardia, ventricular fibrillation, electromechanical dissociation) or asystole and requiring cardiopulmonary resuscitation from onset to admission. ^¶360-day survivors vs. non-survivors. BNP, brain natriuretic peptide; CABG, coronary artery bypass graft; CK, creatinine kinase; DBP, diastolic blood pressure; eGFR, estimated glomerular filtration rate; GRACE, Global Registry of Acute Coronary Events; HDL-C, high-density lipoprotein cholesterol; LDL-C, low-density lipoprotein cholesterol; LVEF, left ventricular ejection fraction; MI, myocardial infarction; PCI, percutaneous coronary intervention; SBP, systolic blood pressure.

Table 2. Clinical Outcomes

	All (n=412)	Survivors at 360 days (n=348)	Non-survivors in hospital (n=54)	Non-survivors at 360 days (n=64)	P-value†
CCU/ICU stay (days)	5.8±9.0	4.4±4.0	14.5±20.9	13.3±19.5	<0.0001
Length of hospitalization (days)	16±19.9	15.2±18.3	19.3±27.9	20.5±27.0	0.0050
In-hospital death	54 (13.1)	0	54 (100)	54 (84.4)	<0.0001
30-day death	46 (11.1)	0	45 (83.3)	46 (71.9)	<0.0001
360-day death	64 (15.5)	0	54 (100)	64 (100)	<0.0001
Cardiac			45 (83.3)	46 (71.9)
Non-cardiac			9 (16.7)	15 (23.4)
Unknown			0	3 (4.7)

Data given as mean±SD or n (%). ^†360-day survivors vs. non-survivors. CCU, coronary care unit; ICU, intensive care unit.

Diagnostic Accuracy

To confirm the diagnostic accuracy of GRACE 1.0 and GRACE 2.0, ROC curves were constructed for GRACE 1.0 predicted in-hospital mortality and GRACE 2.0 predicted 360-day mortality in STEMI patients (Figure). The AUC for in-hospital and 360-day mortality were as high as 0.95 and 0.92, respectively.

Figure.

Receiver operating characteristic curves for (A) Global Registry of Acute Coronary Events (GRACE) 1.0 predicted in-hospital mortality and (B) GRACE 2.0 predicted 360-day mortality for all-cause death. The diagnostic accuracy was very high. The area under the curve (AUC) for in-hospital and 360-day mortality was 0.95 and 0.92, respectively.

Discussion

The present study has demonstrated the diagnostic accuracy of GRACE 1.0 and GRACE 2.0 for in-hospital and 360-day mortality for STEMI in Japanese patients. Both GRACE 1.0 and GRACE 2.0 had a very high ability to predict in-hospital and 360-day death.

Myocardial infarction is one of the leading causes of death, resulting in more than 40,000 deaths per year in Japan, even though primary PCI has improved the mortality associated with STEMI.^12–15 In order to differentiate the critical patients at high risk of mortality, it is important to improve the prediction of mortality rate of such patients with STEMI in emergency care. GRACE 1.0 was established as a useful scoring model to predict the short-term mortality in acute myocardial infarction.^5–7,16,17 This score is so simple to calculate that early evaluation can be done at onset. GRACE 2.0 was an update on GRACE 1.0, issued in 2014 in order to evaluate long-term mortality of these patients. The GRACE registry consists of 10,000 records of ACS patients from 18 cluster sites in 14 countries with a central focus on North America, South America, Europe, Australia, and New Zealand.¹⁸ Asian patients, especially Japanese, were not included, however, therefore it is important to validate whether this scoring system can be used in Japanese patients. To our knowledge, this study is the first to validate the diagnostic ability of these scoring models for Japanese patients.

The clinical value of the GRACE predictive model is not inferior to Thrombolysis in Myocardial Infarction risk score or other predictive models.^19–21 GRACE 1.0 is a simplified method of producing an initial evaluation in the emergency department, and was reported to have a high diagnostic performance for ACS in predicting short-term mortality, in the range of AUC 0.80–0.91.^19,20,22,23 In a prior study in USA to validate the accuracy of GRACE 1.0 for in-hospital mortality in 698 STEMI patients, GRACE 1.0 had a high accuracy on ROC analysis, with an AUC as high as 0.84.¹⁷ Although minor differences were observed in some studies,²² validation studies have confirmed the accurate predictive ability of GRACE 1.0 for short-term mortality.

In terms of long-term mortality, some reported an acceptable performance for GRACE 1.0.²¹ The original target of GRACE 1.0, however, was in-hospital death, and for that reason the GRACE 2.0 updated version of GRACE 1.0 was reported for long-term mortality in 2014. To our knowledge, the present study is the first to validate the predictive performance of GRACE 2.0 for Japanese patients. GRACE 1.0 is calculated based on the severity of 8 stratified clinical parameters, and the predicted mortality has a linear relationship to the score. In contrast, GRACE 2.0 provides the predicted mortality directly, bypassing scores, because the predicted mortality has a non-linear relationship with score. GRACE 1.0 was updated to GRACE 2.0 in order to evaluate the long-term mortality without changing the scoring parameters. Although the GRACE 2.0 scoring algorithm is complicated compared to GRACE 1.0, the population histogram, which is a feature of the GRACE registry, provides an easy-to-understand visual representation of individual risk stratification.

There are several limitations in this study. First, the present study was a retrospective study. Second, the present study might be underpowered to draw a conclusion on diagnostic ability because of the small sample size, although all-cause death was as high as 15.1%. Third, the subject group was selected with a focus on STEMI. Both GRACE risk scores are the established models for not only STEMI but also unstable angina or non-STEMI, which could produce a different result from STEMI. Fourth, this study had a high mortality because it included a high number of critical patients (cardiopulmonary arrest, 7%; Killip IV, 23%). Therefore, the present subject group might differ slightly from the average Japanese STEMI patients.

Conclusions

GRACE risk score can be applied to Japanese STEMI patients for predicting in-hospital and 360-day mortality with a high diagnostic accuracy.

References

• 1.    Nakayama  N,  Kimura  K,  Endo  T,  Fukui  K,  Himeno  H,  Iwasawa  Y, et al. Current status of emergency care for ST-elevation myocardial infarction in an urban setting in Japan. Circ J 2009; 73: 484–489.
• 2.    Hao  K,  Yasuda  S,  Takii  T,  Ito  Y,  Takahashi  J,  Ito  K, et al. Urbanization, life style changes and the incidence/in-hospital mortality of acute myocardial infarction in Japan: Report from the MIYAGI-AMI Registry Study. Circ J 2012; 76: 1136–1144.
• 3.    Daida  H,  Miyauchi  K,  Ogawa  H,  Yokoi  H,  Matsumoto  M,  Kitakaze  M, et al. Management and two-year long-term clinical outcome of acute coronary syndrome in Japan: Prevention of atherothrombotic incidents following ischemic coronary attack (PACIFIC) registry. Circ J 2013; 77: 934–943.
• 4.    Nakamura  M,  Yamagishi  M,  Ueno  T,  Hara  K,  Ishiwata  S,  Itoh  T, et al. Current antiplatelet therapy for Japanese patients with ST elevation acute myocardial infarction: J-AMI registry. Cardiovasc Interv Ther 2013; 28: 162–169.
• 5.    Granger  CB,  Goldberg  RJ,  Dabbous  O,  Pieper  KS,  Eagle  KA,  Cannon  CP, et al. Predictors of hospital mortality in the global registry of acute coronary events. Arch Intern Med 2003; 163: 2345–2353.
• 6.    Eagle  KA,  Lim  MJ,  Dabbous  OH,  Pieper  KS,  Goldberg  RJ,  Van de Werf  F, et al. A validated prediction model for all forms of acute coronary syndrome: Estimating the risk of 6-month postdischarge death in an international registry. JAMA 2004; 291: 2727–2733.
• 7.    Fox  KA,  Fitzgerald  G,  Puymirat  E,  Huang  W,  Carruthers  K,  Simon  T, et al. Should patients with acute coronary disease be stratified for management according to their risk? Derivation, external validation and outcomes using the updated GRACE risk score. BMJ Open 2014; 4: e004425, doi:10.1136/bmjopen-2013-004425.
• 8.   Standard definitions adopted in GRACE. http://www.outcomes-umassmed.org/grace/Files/Standard_Definitions.pdf (accessed August 11, 2014).
• 9.    Matsuo  S,  Imai  E,  Horio  M,  Yasuda  Y,  Tomita  K,  Nitta  K, et al. Revised equations for estimated GFR from serum creatinine in Japan. Am J Kidney Dis 2009; 53: 982–992.
• 10.   GRACE. http://www.outcomes-umassmed.org/grace/ (accessed July 22, 2014).
• 11.   GRACE 2.0 ACS Risk Score App. http://www.gracescore.org/WebSite/default.aspx?ReturnUrl=%2f (accessed July 22, 2014).
• 12.   Number and rate of gender-based death by classification of the cause of death (per 100,000 population). http://www.mhlw.go.jp/toukei/saikin/hw/jinkou/kakutei10/dl/11_h7.pdf (accessed August 11, 2014) (in Japanese).
• 13.    Kimura  K,  Kosuge  M,  Okuda  J. Percutaneous coronary intervention in ST-segment elevation myocardial infarction. Cardiovasc Interv Ther 2010; 25: 53–59.
• 14.    Hsieh  MJ,  Chen  YC,  Chen  CC,  Wang  CL,  Wu  LS,  Wang  CC. Renal dysfunction on admission, worsening renal function, and severity of acute kidney injury predict 2-year mortality in patients with acute myocardial infarction. Circ J 2013; 77: 217–223.
• 15.    Kojima  S,  Matsui  K,  Ogawa  H; Kumamoto Acute Coronary Events (KACE) Study Group. Temporal trends in hospitalization for acute myocardial infarction between 2004 and 2011 in Kumamoto, Japan. Circ J 2013; 77: 2841–2843.
• 16.    Fox  KA,  Dabbous  OH,  Goldberg  RJ,  Pieper  KS,  Eagle  KA,  Van de Werf  F, et al. Prediction of risk of death and myocardial infarction in the six months after presentation with acute coronary syndrome: Prospective multinational observational study (GRACE). BMJ 2006; 333: 1091.
• 17.    Aragam  KG,  Tamhane  UU,  Kline-Rogers  E,  Li  J,  Fox  KA,  Goodman  SG, et al. Does simplicity compromise accuracy in ACS risk prediction? A retrospective analysis of the TIMI and GRACE risk scores. PLoS One 2009; 4: e7947, doi:10.1371/journal.pone.0007947.
• 18.    Investigators  G. Rationale and design of the GRACE (Global Registry of Acute Coronary Events) Project: A multinational registry of patients hospitalized with acute coronary syndromes. Am Heart J 2001; 141: 190–199.
• 19.    Correia  LC,  Freitas  R,  Bittencourt  AP,  Souza  AC,  Almeida  MC,  Leal  J, et al. Prognostic value of GRACE scores versus TIMI score in acute coronary syndromes. Arq Bras Cardiol 2010; 94: 613–619.
• 20.    D’Ascenzo  F,  Biondi-Zoccai  G,  Moretti  C,  Bollati  M,  Omede  P,  Sciuto  F, et al. TIMI, GRACE and alternative risk scores in acute coronary syndromes: A meta-analysis of 40 derivation studies on 216,552 patients and of 42 validation studies on 31,625 patients. Contemp Clin Trials 2012; 33: 507–514.
• 21.    Filipiak  KJ,  Koltowski  L,  Grabowski  M,  Karpinski  G,  Glowczynska  R,  Huczek  Z, et al. Comparison of the seven-year predictive value of six risk scores in acute coronary syndrome patients: GRACE, TIMI STEMI, TIMI NSTEMI, SIMPLE, ZWOLLE and BANACH. Kardiol Pol 2014; 72: 155–165.
• 22.    Tang  EW,  Wong  CK,  Herbison  P. Global Registry of Acute Coronary Events (GRACE) hospital discharge risk score accurately predicts long-term mortality post acute coronary syndrome. Am Heart J 2007; 153: 29–35.
• 23.    Chan  MY,  Shah  BR,  Gao  F,  Sim  LL,  Chua  T,  Tan  HC, et al. Recalibration of the Global Registry of Acute Coronary Events risk score in a multiethnic Asian population. Am Heart J 2011; 162: 291–299.