Risk assessment is central to the management of acute coronary syndromes. Often, however, assessment is not complete until the troponin concentration is available. Using 2 multicenter prospective observational studies (Evaluation of Methods and Management of Acute Coronary Events [EMMACE] 2, test cohort, 1,843 patients; and EMMACE-1, validation cohort, 550 patients) of unselected patients with acute coronary syndromes, a point-of-admission risk stratification tool using frontal QRS-T angle derived from automated measurements and age for the prediction of 30-day and 2-year mortality was evaluated. Two-year mortality was lowest in patients with frontal QRS-T angles <38° and highest in patients with frontal QRS-T angles >104° (44.7% vs 14.8%, p <0.001). Increasing frontal QRS-T angle–age risk (FAAR) scores were associated with increasing 30-day and 2-year mortality (for 2-year mortality, score 0 = 3.7%, score 4 = 57%; p <0.001). The FAAR score was a good discriminator of mortality (C statistics 0.74 [95% confidence interval 0.71 to 0.78] at 30 days and 0.77 [95% confidence interval 0.75 to 0.79] at 2 years), maintained its performance in the EMMACE-1 cohort at 30 days (C statistics 0.76 (95% confidence interval 0.71 to 0.8] at 30 days and 0.79 (95% confidence interval 0.75 to 0.83] at 2 years), in men and women, in ST-segment elevation myocardial infarction and non–ST-segment elevation myocardial infarction, and compared favorably with the Global Registry of Acute Coronary Events (GRACE) score. The integrated discrimination improvement (age to FAAR score at 30 days and at 2 years in EMMACE-1 and EMMACE-2) was p <0.001. In conclusion, the FAAR score is a point-of-admission risk tool that predicts 30-day and 2-year mortality from 2 variables across a spectrum of patients with acute coronary syndromes. It does not require the results of biomarker assays or rely on the subjective interpretation of electrocardiograms.
Age, heart rate, and systolic blood pressure are readily available on admission to the hospital and permit the discrimination of short-term mortality. Data from rest electrocardiography may also be used to predict mortality, and because international guidelines recommend that all patients with suspected acute coronary syndromes (ACS) undergo rest 12-lead electrocardiography as soon as possible, these data should be immediately available for risk estimation. Indeed, relations exist between heart rate, PR and QT intervals, QRS duration, and P, QRS, and T-wave axes and cardiovascular outcomes. Notably, the frontal T-axis and spatial QRS-T angle also predict mortality. The aims of our study were to investigate the survival of 1,843 patients admitted with the full spectrum of ACS diagnoses stratified by the frontal QRS-T angle and to evaluate the performance of a risk score based on frontal QRS-T angle and age.
Methods
We used the Evaluation of Methods and Management of Acute Coronary Events (EMMACE) 2 and EMMACE-1 prospective studies to test and validate models, respectively. The 2 studies examined outcomes in consecutively admitted, unselected patients with confirmed ACS in multiple adjacent hospitals (within the catchment area of 1 tertiary center) in Yorkshire, United Kingdom. Patients were recruited over a 3-month period during 1995 in EMMACE-1 (3,684 patients) and during a 6-month period during 2003 in EMMACE-2 (2,499 patients). Baseline patient characteristics, inpatient and discharge treatment variables, and all-cause mortality from hospital admission were recorded. All EMMACE-2 patients provided written informed consent to participate, which was approved by the Multi Research Ethics Committee and local research ethics committees, in accordance with the Declaration of Helsinki. The EMMACE-1 study was conducted as an audit and did not require consent.
We used the joint consensus document of the European Society of Cardiology and the American College of Cardiology as the diagnostic standard for acute myocardial infarction, which required ischemic symptoms and increases in cardiac biomarkers. The presence of new ST-segment elevation in ≥2 contiguous leads (measuring 0.2 mV in leads V 1 to V 3 or 0.1 mV in all other leads) provided the basis for categorization into ST-segment elevation myocardial infarction (STEMI). The 12- to 24-hour postadmission AccuTnI troponin I assay (Beckman Coulter, Ltd., High Wycombe, United Kingdom; interassay coefficient of variation 10% at 0.06 μg/L 20 ) was used to define patients as troponin I negative (≤0.06 μg/L) or troponin I positive (>0.06 μg/L).
For each patient, we selected the earliest available electrocardiographic (ECG) recording in chronologic order with computerized values for ≥1 of the following variables: heart rate, PR interval, QRS duration, QT interval, corrected QT interval, P-wave axis, QRS axis, and T-wave axis. The absolute difference between the frontal QRS and frontal T-wave axes was calculated as (T-wave axis − QRS axis) and if >180° was subtracted from 360° to give a continuous variable ranging from 0° to 180°.
Continuous data are expressed as mean ± SD or as medians and interquartile ranges for skewed distributions. Discrete or categorical data are summarized using frequencies and percentages. Associations between mortality and potential predictor factors are quantified by odds ratios with 95% confidence intervals (CIs).
Using the EMMACE-2 cohort, stepwise logistic regression was used to predict 30-day and 2-year mortality. To explore the extent to which specific ECG data contributed to the discriminative performance of the models, the models were adjusted for covariates selected from patient characteristics and the results of investigations. Parsimonious models were built. A frontal QRS-T angle–age risk (FAAR) score was constructed, with 1 point for age >69 years, 2 points for age >82 years, 1 point for a frontal QRS-T angle of 38° to 104°, and 2 points for a frontal QRS-T angle >104°. The aforementioned ranges were based on tertiles of the respective variables. The risk score was validated in the EMMACE-1 cohort. The discriminative performance of the models (assessed using C statistics) was compared to that of the Global Registry of Acute Coronary Events (GRACE) risk score, derived using coefficients generated from the EMMACE-2 data set. The calibration of the models was evaluated using the Hosmer-Lemeshow test, and the net reclassification over all possible cutoffs for the probability of mortality was measured using the integrated discrimination improvement index. Additional analyses were conducted using time to death as the response variable. Kaplan-Meier survival curves were used to compare the survival of patients in different FAAR groups and tertiles (strata) of frontal QRS-T angle.
Of the 2,499 patients recruited in EMMACE-2, there were 1,843 first recorded admissions with ≥1 available electrocardiogram comprising computerized measurements taken during the hospital admission. Twelve-month mortality was equal for patients included and excluded from this study (22.8%). The proportions of patients with a range of other descriptive characteristics were similar for those included and excluded from this study (see Online Appendix ). Hospital admission ECG data were available for 550 patients in EMMACE-1. One-year mortality (33.6%) was comparable for patients included and excluded, and the proportions of patients with key clinical characteristics were also similar for those with and without electrocardiograms (see Online Appendix ). The median times between admission and recording of the electrocardiogram used for subsequent analysis were 11 and 13 minutes in the EMMACE-2 and EMMACE-1 cohorts, respectively. For all tests, p values <0.05 were considered statistically significant.
Results
Of the 1,843 patients in the EMMACE-2 cohort, the mean age was 70.1 ± 13.1 years, and 1,547 (61.9%) were men. Of the cohort, 942 (37.7%) had non-STEMI, 755 (30.2%) had STEMI, and 802 (32.1%) had troponin-negative ACS. The management of patients with ACS in 2003 differed from current practice: 12.7% of patients were revascularized during their admissions, 73.5% received heparin, and 74.4%, 38.9%, 74.6%, 58.7%, and 59.8% of patients were discharged on aspirin, clopidogrel, statins, angiotensin-converting enzyme inhibitors, and β blockers, respectively. Equally, the management of patients in the EMMACE-2 cohort differed from than of the EMMACE-1 cohort.
For the EMMACE-2 cohort, the mean age was 70.7 ± 12 years, and 1,331 (60.6%) were men. The clinical and ECG characteristics of the cohorts are listed in Table 1 . Clinical characteristics, ECG data, and mortality rates of the 3 frontal QRS-T angle groups are listed in Table 2 . Table 3 lists the predictors of all-cause 30-day and 2-year mortality.
| Variable | EMMACE-1 | EMMACE-2 |
|---|---|---|
| (n = 550) | (n = 1,843) | |
| Clinical characteristics | ||
| Age (years) | 70.7 ± 12 | 70.1 ± 13.1 |
| Men | 60.6% | 61.9% |
| Hypertension | 28.5% | 42.9% |
| Stroke | 9.7% | 11.8% |
| Diabetes mellitus | 12.7% | 19.1% |
| Hypercholesterolemia | 6.8% | 33.7% |
| Peripheral vascular disease | 7.5% | 8.4% |
| Previous acute myocardial infarction | 25.9% | 25.4% |
| Angina pectoris | 37.2% | 45.2% |
| Coronary artery bypass grafting | 3.0% | 6.9% |
| ECG data | ||
| Heart rate (beats/min) | 84 ± 26 | 82 ± 25 |
| PR interval (ms) | 168 ± 39 | 171 ± 38 |
| QRS duration (ms) | 104 ± 26 | 102 ± 24 |
| QT interval (ms) | 386 ± 62 | 384 ± 56 |
| Corrected QT interval (ms) | 437 ± 45 | 436 ± 37 |
| P axis (°) | 54 ± 32 | 53 ± 35 |
| QRS axis (°) | 23 ± 53 | 16 ± 50 |
| T axis (°) | 79 ± 63 | 73 ± 69 |
| QRS-T angle (°) | 68 (103) | 67 (102) |
| Mortality rate (months) | ||
| 1 | 24.3% | 11.6% |
| 24 | 38.0% | 29.3% |
| Variable | QRS-T Angle | p Value | ||
|---|---|---|---|---|
| Tertile 1 (<37°) | Tertile 2 (38°–104°) | Tertile 3 (>105°) | ||
| Clinical characteristics | ||||
| Age (years) | 68 ± 13 | 73 ± 13.0 | 79 ± 11 | <0.001 |
| Men | 408 (66.3%) | 376 (62%) | 365 (59.2%) | 0.009 |
| Heart rate on admission (beats/min) | 75 ± 21.8 | 80 ± 22 | 91 ± 26 | <0.001 |
| Systolic blood pressure (mm Hg) | 142 ± 28 | 143 ± 29 | 142 ± 33 | 0.945 |
| Blood glucose on admission (mmol/L) | 7.7 ± 4.2 | 7.8 ± 3.2 | 8.4 ± 4.2 | 0.008 |
| Peak troponin I (μg/L) | 12.1 ± 25.6 | 12.3 ± 21.2 | 8.7 ± 17.4 | 0.230 |
| ST-segment depression on admission electrocardiography | 98 (15.9%) | 117 (19.3%) | 148 (24.0%) | 0.002 |
| Heart failure | 20 (3.3%) | 36 (5.9%) | 74 (12%) | <0.001 |
| Previous acute myocardial infarction | 122 (19.8%) | 150 (24.8%) | 212 (34.4%) | <0.001 |
| Hypertension | 262 (42.6%) | 264 (43.6%) | 277 (44.9%) | 0.416 |
| Angina pectoris | 224 (36.4%) | 268 (44.2%) | 334 (54.1%) | <0.001 |
| Peripheral vascular disease | 45 (7.3%) | 46 (7.6%) | 60 (9.7%) | 0.748 |
| Hyperlipidemia | 213 (34.6%) | 222 (36.6%) | 203 (32.9%) | 0.792 |
| Cerebrovascular disease | 42 (6.8%) | 58 (9.6%) | 105 (17%) | <0.001 |
| Chronic renal failure | 16 (2.6%) | 24 (4.0%) | 53 (8.6%) | <0.001 |
| Diabetes mellitus | 88 (14.4%) | 86 (14.2%) | 117 (18.9%) | 0.089 |
| Current smoker | 223 (36.3%) | 167 (27.6%) | 121 (19.6%) | <0.001 |
| Previous percutaneous coronary intervention | 46 (7.7%) | 39 (6.5%) | 35 (5.9%) | 0.004 |
| LBBB | 7 (1.1%) | 14 (2.3%) | 113 (18.3%) | <0.001 |
| STEMI | 199 (32.4%) | 162 (26.7%) | 87 (14.1%) | <0.001 |
| Non-STEMI | 162 (26.3%) | 176 (29.0%) | 229 (37.1%) | <0.001 |
| Troponin I–negative ACS | 29 (4.7%) | 33 (5.4%) | 24 (3.9%) | 0.492 |
| Corrected QT interval (ms) | 427 ± 33 | 432 ± 35 | 449 ± 40 | <0.001 |
| P axis (°) | 53 ± 29 | 52 ± 31 | 57 ± 45 | 0.072 |
| QRS axis (°) | 36 ± 37 | 16 ± 42 | −1 ± 61 | <0.001 |
| T axis (°) | 41 ± 39 | 57 ± 56 | 134.5 ± 76.1 | <0.001 |
| Mortality rate (months) | ||||
| 1 | 4.1% | 8.4% | 17.7% | <0.001 |
| 24 | 14.8% | 21.9% | 44.7% | <0.001 |
| Predictor | 30-Day Mortality | 2-Year Mortality | ||
|---|---|---|---|---|
| Odds Ratio (95% CI) | p Value | Odds Ratio (95% CI) | p Value | |
| Univariate predictors | ||||
| Age (per year) | 1.07 (1.057–1.084) | <0.001 | 1.088 (1.078–1.099) | <0.001 |
| Female gender | 0.718 (0.560–0.922) | 0.009 | 0.698 (0.585–0.833) | <0.001 |
| PR interval (per ms) | 0.999 (0.915–1.004) | 0.751 | 1.002 (0.999–1.005) | 0.113 |
| QRS duration (per ms) | 1.013 (1.007–1.018) | <0.001 | 1.017 (1.012–1.021) | <0.001 |
| QT interval (per ms) | 0.996 (0.993–0.999) | 0.006 | 0.995 (0.993–0.997) | <0.001 |
| Corrected QT interval (per ms) | 1.007 (1.003–1.01) | <0.001 | 1.009 (1.006–1.012) | <0.001 |
| P axis (per degree) | 1.002 (0.998–1.007) | 0.305 | 1.004 (1.001–1.007) | 0.014 |
| QRS axis (per degree) | 1.001 (0.998–1.004) | 0.468 | 0.999 (0.996–1.001) | 0.176 |
| T axis (per degree) | 1.005 (1.003–1.007) | <0.001 | 1.006 (1.005–1.008) | <0.001 |
| QRS-T angle (per degree) | 1.011 (1.009–1.014) | <0.001 | 1.012 (1.100–1.140) | <0.001 |
| QRS-T angle Q1 (<38°) | 1 | 1 | ||
| QRS-T angle Q2 (38°–104°) | 2.169 (1.325–3.548) | 0.002 | 1.619 (1.206–2.173) | 0.001 |
| QRS-T angle Q3 (>104°) | 5.064 (3.127–7.946) | <0.001 | 2.291 (1.650–3.180) | <0.001 |
| Heart failure | 1.854 (1.242–2.768) | 0.003 | 4.12 (3.004–6.649) | <0.001 |
| Previous acute myocardial infarction | 0.992 (0.747–1.318) | 0.956 | 1.548 (1.276–1.878) | <0.001 |
| Admission heart rate (per beat/min) | 1.013 (1.008–1.017) | <0.001 | 1.018 (1.014–1.021) | <0.001 |
| Admission systolic blood pressure (per mm Hg) | 0.984 (0.980–0.989) | <0.001 | 0.993 (0.990–0.995) | <0.001 |
| ST-segment depression on admission electrocardiography | 1.555 (1.175–2.057) | 0.002 | 1.484 (1.207–1.825) | <0.001 |
| Creatinine on admission (per μmol/L) | 1.003 (1.001–1.004) | <0.001 | 1.009 (1.007–1.011) | <0.001 |
| Elevated cardiac biomarkers | 2.301 (1.367–3.873) | 0.020 | 1.398 (1.044–1.874) | 0.025 |
| Inpatient percutaneous coronary intervention | 0.110 (0.041 to 0298) | <0.001 | 0.135 (0.079–0.229) | <0.001 |
| Independent predictors ⁎ | ||||
| PR interval (per ms) | 0.996 (0.99–1.001) | 0.142 | 0.998 (0.993–1.002) | 0.243 |
| QRS duration (per ms) | 1.004 (0.995–1.013) | 0.403 | 1.009 (1.002–1.016) | 0.011 |
| QT interval (per ms) | 0.999 (0.994–1.004) | 0.598 | 0.998 (0.994–1.002) | 0.262 |
| P axis (per degree) | 0.999 (0.994–1.004) | 0.644 | 0.999 (0.995–1.003) | 0.605 |
| QRS axis (per degree) | 1.006 (1.002–1.01) | 0.002 | 1.003 (1.000–1.006) | 0.041 |
| T axis (per degree) | 0.999 (0.996–1.002) | 0.369 | 1.001 (0.999–1.004) | 0.281 |
| QRS-T angle (per degree) | 1.01 (1.006–1.015) | <0.001 | 1.005 (1.002–1.009) | 0.004 |
| QRS-T angle Q1 (<37°) | 1 | 1 | ||
| QRS-T angle Q2 (38°–104°) | 1.629 (0.97–2.738) | 0.065 | 1.105 (0.795–1.537) | 0.552 |
| QRS-T angle Q3 (>105°) | 2.66 (1.627–4.347) | <0.001 | 1.926 (1.404–2.643) | <0.001 |
⁎ After adjustment for age, heart failure, previous acute myocardial infarction, heart rate, systolic blood pressure, ST-segment depression on admission electrocardiography, creatinine, elevated cardiac markers, and inpatient percutaneous coronary intervention.
After adjustment for age, heart failure, previous acute myocardial infarction, heart rate, blood pressure, ST-segment depression (measuring ≥0.1 mV in ≥2 contiguous leads), creatinine, elevated cardiac markers, and inpatient percutaneous coronary intervention, the frontal QRS-T angle carried the greatest significance of the ECG parameters.
Figure 1 shows the Kaplan-Meier curves for unadjusted survival probabilities by frontal QRS-T angle tertile for 2-year mortality. Survival was highest in patients with frontal QRS-T angles <38°, followed by patients with frontal QRS-T angles of 38° to 104°, and lowest in patients with frontal QRS-T angles >104°. The risk for death increased with frontal QRS-T angle.
Increasing 30-day and 2-year mortality rates (30-day rate: FAAR score 0 = 0.7%, score 1 = 4.4%, score 2 = 8.7%, score 3 = 18.2%, score 4 = 22.3%; 2-year rate: FAAR score 0 = 3.7%, score 1 = 14.4%, score 2 = 30.2%, score 3 = 38.2%, score 4 = 57.3%), and significantly different survival experiences ( Figure 2 ) were evident by FAAR score. The scoring system offered good discriminative performance for 30-day and 2-year mortality, maintained its performance in the EMMACE-1 validation cohort at 30 days and at 2 years, and compared favorably to the GRACE score ( Table 4 ). The integrated discrimination improvement index for each model comparison (age to age QRS-T angle regression and age to FAAR score at 30 days and 2 years in EMMACE-1 and EMMACE-2) was p <0.001. The p values for the Hosmer-Lemeshow test indicated good calibration for the FAAR score in the EMMACE-1 but not the EMMACE-2 cohort.
