- Email: [email protected]
Utilization of ST-segment deviation sum and change scores to identify acute myocardial infarction
American Journal of Emergency Medicine (2010) 28, 790–797
www.elsevier.com/locate/ajem
Original Contribution
Utilization of ST-segment deviation sum and change scores to identify acute myocardial infarction☆ Francis M. Fesmire MD ⁎ Emergency Heart Center, Erlanger Medical Center and Division of Emergency Medicine, University of Tennessee Chattanooga College of Medicine, Chattanooga, TN 37405, USA Received 4 February 2009; revised 31 March 2009; accepted 1 April 2009
Abstract Objective: No information is currently available regarding the optimal cutoff values of the baseline STsegment deviation sum (STDsumbaseline) and 60-minute ST-segment deviation change (STDchange60min) for predicting acute myocardial infarction (AMI). Methods: A retrospective study was performed in 783 admitted patients with chest pain who had suspected acute coronary syndrome and absence of left ventricular hypertrophy or bundle branch block on the initial electrocardiogram (ECG). The STDsumbaseline was defined as the sum in millimeters (1 mm = 0.1 mV) of the absolute value of ST-segment deviations in all 12 leads at the initiation of continuous 12-lead ECG monitoring session. The STDchange60min was defined as the absolute value of the difference between the baseline and 60-minute STDsum. Three cutoff values are reported and represent the smallest values in which the positive likelihood ratio (+LR) for AMI was greater than or equal to 5, 10, and 20, respectively. Results: Acute myocardial infarction occurred in 162 (20.7%) patients. The smallest cutoff value of the STDsumbaseline for AMI with a +LR equal to or greater than 5, 10, and 20 was 9.6, 12.4, and 14.1 mm, respectively. In the subset of 699 patients without ST-segment elevation AMI on initial ECG, the smallest cutoff value of the STDchange60min for AMI with a +LR equal to or greater than 5, 10, and 20 was 2.4, 3.5, and 7.9 mm, respectively. Conclusions: Clinical studies need to be performed to determine if STDsum and STDchange, in conjunction with physician pretest probability of AMI, can be used to select patients who may benefit from emergent reperfusion therapy and other aggressive medical management strategies. © 2010 Elsevier Inc. All rights reserved.
1. Introduction
☆ Presented at the Scientific Assembly of the American College of Emergency Physicians (Research Forum) at Washington, DC, on September 26, 2005. ⁎ UT College of Medicine, PO Box 4045, Chattanooga, TN 37405, USA. E-mail address: [email protected].
0735-6757/$ – see front matter © 2010 Elsevier Inc. All rights reserved. doi:10.1016/j.ajem.2009.04.004
Identification of ST-segment changes diagnostic of injury or ischemia on the 12-lead electrocardiogram (ECG) is of paramount importance in the identification of patients with acute myocardial infarction (AMI) and non-AMI acute coronary syndrome (ACS) who may benefit from emergent reperfusion therapy and other aggressive anti-ischemic therapies [1-6]. Current practice guidelines from the American College of Cardiology and American Heart
Utilization of ST deviation sum score Association (AHA) recommend emergent reperfusion therapy in patients with new or presumably new left bundle branch block (BBB) on the ECG, in patients with STsegment elevations of 1 mm in 2 or more contiguous leads, or in patients with ST-segment depressions consistent with true posterior AMI [1]. Interpretation of the magnitude and morphology of STsegment elevations and depressions is often subjective and varies depending on the experience and expertise of the health care provider. Furthermore, the multitude of confounding factors that result in ST-segment elevations and depressions can make interpretation of electrocardiographic injury and ischemia difficult [6-9]. Studies have demonstrated that ST-segment elevation is frequently seen in emergency department (ED) patients with chest pain and that only approximately 15% to 30% of these patients are experiencing ACS [8,9]. Left ventricular hypertrophy (LVH) is the most common cause of non-ACS STsegment elevation on the ECG and occurs in approximately 30% of this subset of patients [8,9]. Other nonACS causes of ST-segment elevation are benign early repolarization, pericarditis, left ventricular aneurysm, paced rhythm, and BBB [8,9]. A potential method to assist health care providers in the identification of ST-segment changes suggestive of electrocardiographic injury and ischemia is to use the absolute value of the sum of ST-segment deviations in all 12 leads on the ECG (STDsum) to better predict the total ischemic burden (Fig. 1) [10]. In patients with ST-segment elevation AMI (STEMI), the STDsum has been demonstrated to be predictive of infarct size [11,12] and adverse events [13,14]. In patients with non-STEMI (NSTEMI), the STDsum has been demonstrated to be an independent predictor of 1-year death or myocardial infarction (MI) [15]. Recovery of STDsum of 70% or greater at 60 minutes has also been demonstrated to be predictive of successful reperfusion in patients treated with fibrinolytics [16]. It has been previously reported that increasing ST-vector magnitude on the ECG is associated with increasing likelihood of AMI [17,18]. ST-vector magnitude represents the sum of ST-segment deviation in 3-dimensional space and, in theory, should correlate with the STDsum [17,18]. Currently, no information is available regarding optimal STDsum values for predicting AMI in ED patients with chest pain presenting with possible ACS. Theoretically, this optimal value corresponds to electrocardiographic injury and high-risk ischemia and could potentially be a criterion for emergent reperfusion therapy and other aggressive medical management strategies, as well as a tool to assist physicians and other health care providers in the evaluation, treatment, and disposition of patients with chest pain. In addition, although decreases in STDsum over time have been shown to be predictive of successful reperfusion with fibrinolytic therapy [16], no information is available regarding whether changes in STDsum are predictive of AMI in patients without STEMI on the baseline ECG
791 (ECGbaseline) obtained on initial ED presentation. In this study, we investigate the ability of an automated computerized measurement of the baseline STDsum (STDsumbaseline) in all study patients and changes in STDsum at 60 minutes (STDchange60min) in the subset of patients without STEMI on ECGbaseline to predict AMI in ED chest pain patients with suspected ACS.
2. Methods 2.1. Study design This is a retrospective analysis of a previously reported prospective investigation on the utility of automated continuous 12-lead ECG monitoring (ECGcontinuous) for identification of injury and ischemia in patients with chest pain [19]. The study was approved by the institutional review board.
2.2. Setting The study was conducted at Erlanger Medical Center, an adult urban tertiary referral center. The hospital is affiliated with the University of Tennessee College of Medicine and has an annual ED census of 45 000 patients.
2.3. Selection of participants The study population was derived from a convenience sample of 1000 patients with chest pain who underwent at least 60 minutes of ECGcontinuous during the initial ED evaluation and were subsequently admitted to the hospital for further treatment and/or evaluation. Exclusion criteria for purposes of this current investigation included patients who were discharged from ED, patients undergoing less than 60 minutes of ECGcontinuous, and patients with electrocardiographic LVH, BBB (right, left, or atypical), or ventricular paced rhythm on the ECGbaseline obtained on initial ED presentation, as these patients already have significant ST-segment deviations on the ECG secondary to abnormalities in repolarization [8,9,17].
2.4. Measurements The ECGbaseline was obtained on presentation with the use of the PageWriter Xli Cardiograph (Philips, Andover, MA). Methodology for interpretation of the ECGbaseline has been reported previously [19]. The ELI-100 ECG machine with Lifenet STM Monitor (Mortara Instruments, Milwaukee, WI) was used to perform ECGcontinuous. ST-segment magnitudes were measured every 20 seconds 60 milliseconds past the J point. Printouts of the 12-lead ST-segment trends consist of an ST-segment magnitude for each of the
792 12 leads, a sum of ST-segment magnitude for 4 regions (anterior precordial leads, lateral precordial leads, inferior limb leads, and lateral limb leads), and an absolute sum of the ST-segment deviations in all 12 leads (STDsum). The STDsumbaseline was defined as the STDsum of the initial ECGcontinuous obtained at the beginning of the monitoring session. The STDchange60min was defined as the absolute value of the difference between the baseline STDsum and 60-minute STDsum.
2.5. Main outcomes Acute myocardial infarction was defined according to World Health Organization Criteria in effect at time of study [20]. Acute myocardial infarction on initial presentation was diagnosed if there was 20 minutes or more of chest pain and (1) a serial rise and fall of creatine kinase (CK) and CK-MB fraction as measured by the Stratus Fluorometric Enzyme Immunoassay (Dade Behring, Deerfield, IL), provided that the maximum CK-MB value was 12 ng/mL or greater and the CK-MB index was at least 4% within 24 hours of ED presentation; or (2) new Q-wave formation during the initial 24 hours after presentation; or (3) patient death from cardiac causes during the initial 24 hours of presentation; or (4) coronary arteriogram with documented acute occlusion of a major coronary vessel within 24 hours of presentation. Recent MI was diagnosed in patients with MI whose serial CK-MB measurements were already on the falling curve at presentation. In-hospital MI was diagnosed if the serial CKMB measurements did not meet criteria for AMI until more than 24 hours after presentation. Unstable angina was
F.M. Fesmire diagnosed in patients who were without CK-MB or ECG criteria for MI and whose in-hospital attending physician diagnosed ACS-related chest pain. Adverse outcome (AO) was defined as MI on presentation (AMI or recent MI), inhospital AMI, percutaneous coronary intervention (PCI), coronary artery bypass grafting (CABG), life-threatening complication, or death. Life-threatening complications were defined as ventricular fibrillation, sustained ventricular tachycardia, third-degree atrioventricular block, cardiogenic shock, or cardiac arrest secondary to severe bradycardia or pulseless electrical activity.
2.6. Data analysis Data were recorded on a standardized form and placed in a computer database. Positive (+) and negative (−) likelihood ratios (LRs) were calculated according to the formula +LR = sensitivity/(1 − specificity) and −LR = (1 − sensitivity)/specificity [21]. Receiver Operator Characteristic (ROC) curves were analyzed using MedCalc 9.6 (MedCalc Software, Belgium) [22]. Cutoff values for STDsumbaseline and STDchange60min were calculated at the most accurate cutoff value (value with the lowest falsenegative and false-positive rates) and at the 3 smallest cutoff values where the +LR for AMI is greater than or equal to 5, 10, and 20, respectively. All other data analyses were performed with SYSTAT 10.0 (SPSS Inc, Chicago, IL). Sample means were compared with the 2-sample unpaired t test and are reported as means ± SD. Categorical data were compared with χ2 test with the Yates correction for sample size. A P value ≤.05 was considered significant.
Fig. 1 ECG not meeting standard criteria for emergent reperfusion therapy but with an STDsumbaseline of 18 mm with a corresponding +LR for AMI of 35.
Utilization of ST deviation sum score
793
3. Results
Table 2 Mean ± SD STDsumbaseline and STDchange60min (mm; 1 mm = 0.1 mV) according to diagnosis and STEMI location in the 783 study patients
3.1. Characteristics of study subjects
STDsumbaseline STDchange60min
During the study period, a total of 1000 admitted patients with chest pain underwent a minimum of 60 minutes of ECGcontinuous during the initial ED evaluation. A total of 217 patients were excluded because of findings of LVH, BBB, or ventricular paced rhythm on the ECGbaseline, leaving a study population of 783 patients. The ECGbaseline was obtained 16.7 ± 16.5 minutes after ED arrival and ECGcontinuous was initiated 46.5 ± 46.2 minutes after ED arrival. The mean duration of ST-segment monitoring was 127.7 ± 41.2 minutes. Of the 162 patients with AMI, 84 (10.7%) patients had STEMI on ECGbaseline and 78 (10.0%) patients had NSTEMI on ECGbaseline. Of the 78 patients with NSTEMI on ECGbaseline, 26 (33.3%) patients developed new injury during the ECGcontinuous monitoring session. In the patients with non-AMI, discharge diagnosis consisted of 17 (2.2%) patients with recent MI, 235 (30.0%) patients with unstable angina, and 369 (47.1%) patients with non-ACS chest pain. A total of 256 (32.7%) patients experienced 1 or more of the following AO events: 179 (22.9%) with AMI or recent MI, 97 (12.4%) with PCI, 63 (8.0%) with CABG, 57 (7.3%) with life-threatening complications, and 9 (1.1%) with death. Of the 84 patients with STEMI on ECGbaseline, 70 (83.3%) were treated in the ED with fibrinolytics, and 12 (14.3%), with emergent PCI. Of the 26 patients with NSTEMI on ECGbaseline who developed new injury on ECGcontinuous, 14 (53.8%) patients were treated in the ED with fibrinolytics, and 3 (11.5%) patients, with emergent PCI. Table 1 summarizes demographic features, mean CKMBbaseline and ΔCK-MB2hr, mean STDsumbaseline, and STDchange60min. Patients with AO were older and more likely to be male. Patients with AO had higher mean CK-MBbaseline, ΔCK-MB2hr, STDsumbaseline, and STDchange60min. Table 1 Demographic characteristics, CK-MBbaseline (mg/dL), ΔCK-MB2hr (mg/dL), STDsumbaseline (mm; 1 mm = 0.1 mV), and STDchange60min (mm; 1 mm = 0.1 mV) in 783 study patients with and without in-hospital AO Without AO (n = 527) Age (y) Male Race White African American Previous MI Previous PCI/CABG CK-MBbaseline ΔCK-MB2hr STDsumbaseline STDchange60min
With AO (n = 256)
P
54.2 ± 13.0 58.0 ± 12.1 300 (57.0) 180 (70.3)
b.0001 b.0001
429 (81.4) 98 (18.6) 139 (26.4) 121 (23.0) 0.9 ± 1.6 0.1 ± 0.9 5.5 ± 3.6 0.8 ± 1.0
.23 (NS) .23 (NS) .36 (NS) .95 (NS) b.0001 b.0001 b.0001 b.0001
218 (85.2) 38 (14.8) 59 (23.0) 60 (23.4) 10.3 ± 30.3 15.8 ± 42.3 11.8 ± 10.4 3.3 ± 4.8
Values are presented as mean ± SD or no. (%).
Patients with non-AMI (n = 621) Recent MI (n = 17) Unstable angina (n = 235) Non-ACS (n = 369) NSTEMI on ECGbaseline (n = 78) New injury on ECGcontinuous (n = 26) No injury on ECGcontinuous (n = 52) STEMI on ECGbaseline (n = 84) Anterior/anterolateral (n = 27) Inferior/inferolateral (n = 53) Other (n = 4)
5.6 ± 3.7
0.8 ± 1.1
6.4 ± 3.9 5.5 ± 4.1 5.7 ± 3.4 8.2 ± 5.2
0.5 ± 0.5 1.0 ± 1.4 0.7 ± 0.8 2.5 ± 3.4
11.1 ± 5.2
4.0 ± 3.3
6.7 ± 4.7
1.7 ± 3.3
21.8 ± 11.8
6.7 ± 6.4
20.5 ± 11.1
5.8 ± 4.8
22.0 ± 12.4
7.2 ± 7.0
27.0 ± 6.4
6.2 ± 7.1
Table 2 summarizes mean STDsum base line and STDchange60min according to diagnosis and STEMI location. There was an incremental increase in STDsumbaseline and STDchange60min for patients with non-AMI, NSTEMI, and STEMI. No statistical differences were seen for mean STDsumbaseline and STDchange60min between patients with recent MI, unstable angina, and non-ACS. Comparing patients with non-AMI chest pain with patients with NSTEMI revealed a statistically significant increase in both STDsumbaseline (5.6 vs 8.2 mm, respectively; P b .0001) and STDchange60min (0.8 vs 2.5 mm, respectively; P b .0001). Comparing patients with NSTEMI with patients with STEMI revealed a statistically significant increase in both STDsumbaseline (8.2 vs 21.8 mm, respectively; P b .0001) and STDchange60min (2.5 vs 6.7 mm, respectively; P b .0001). There were no statistical differences in comparing STDsumbaseline (20.5 vs 22.0 mm, respectively; P = .5) and STDchange60min (5.8 vs 7.2 mm, respectively; P = .3) for anterior/anterolateral STEMI vs inferior/inferolateral STEMI. Comparing the 26 patients with NSTEMI on ECGbaseline who developed new injury on ECGcontinuous with the 52 patients who did not develop STEMI revealed a statistically significant increase in both STDsumbaseline (11.1 vs 6.7 mm, respectively; P = .001) and STDchange60min (4.0 vs 1.7 mm, respectively; P b .01).
3.2. Likelihood ratio analysis The ROC curve area of STDsumbaseline was 0.81 (95% confidence interval [CI], 0.78-0.83) for AMI and 0.71 (95% CI, 0.68-0.74) for AO in the 783 study patients. Table 3 summarizes sensitivity and specificity data for predicting
794
F.M. Fesmire
Table 3 Sensitivities, specificities, +LRs, and −LRs of the STDsumbaseline (mm; 1 mm = 0.1 mV) for AMI at the most accurate cutoff value and at the lowest cutoff value where the +LR for AMI is greater than or equal to 5, 10, and 20 in the 783 study patients
Most accurate +LR ≥5 +LR ≥10 +LR ≥20
Baseline STDsumbaseline
Sensitivity for predicting AMI (%)
Specificity for predicting AMI (%)
+LR
−LR
N8.5 N9.6 N12.4 N14.1
65.4 59.9 46.9 42.6
84.5 (81.5-87.3) 88.1 (85.3-90.5) 95.3 (93.4-96.9) 97.9 (96.4-98.9)
4.2 5.0 10.1 20.4
0.41 0.46 0.56 0.59
(57.6-72.7) (51.9-67.5) (39.0-54.9) (34.9-50.6)
AMI at the most accurate cutoff point on the ROC curve and at the 3 cutoff points where the +LR is greater than or equal to 5, 10, and 20, respectively. The 3 STDsumbaseline cutoff values for predicting AMI with +LR greater than or equal to 5, 10, and 20 were 9.6, 12.4, and 14.1 mm, respectively. Fig. 2 demonstrates a graph of baseline STDsumbaseline vs +LR for AMI and demonstrates that increasing values are correlated with increasing likelihood of AMI. Combining initial ECG finding of injury with an STDsumbaseline value greater than 12.4 mm (ie, +LR for AMI N10) identified an additional 13 patients with a final diagnosis of AMI.
3.3. Subset analysis in NSTEMI patients In the subset of 699 patients without STEMI on ECGbaseline, the ROC curve area of STDchange60min was 0.70 (95% CI, 0.67-0.74) for AMI and 0.63 (95% CI, 0.59-0.66) for AO. Table 4 summarizes sensitivity and specificity data of the STDchange60min for predicting AMI at the most accurate cutoff point on the ROC curve and at the 3 cutoff points where the +LR is greater than or equal to 5, 10, and 20. The 3 STDchange60min cutoff values for predicting AMI with +LR greater than or equal to 5, 10, and 20, were 2.4, 3.5, and 7.9 mm, respectively. Fig. 3 demonstrates a graph of STDchange60min value vs +LR for AMI and demonstrates that increasing values are correlated with increasing likelihood of AMI. An STDchange60min greater than 3.5 mm (ie, +LR for AMI N10) identified an additional 19 patients with a final diagnosis of AMI.
Fig. 2 Graph of STDsumbaseline value vs +LR ratio for AMI for the 783 study patients.
3.4. Risk stratification Combining the 2 tests at the cutoff value with a +LR greater than or equal to 10 for AMI (ie, baseline STDsumbaseline N12.4 mm in all patients and/or STDchange60min N3.5 mm in patients with NSTEMI) has a sensitivity and specificity for identification of AMI of 54.3% (95% CI, 46.3-62.2) and 93.4% (95% CI, 91.1-95.2), respectively, and effectively risk stratifies patients for AO. Patients with either an STDsumbaseline greater than 12.4 mm or an STDchange60min greater than 3.5 mm had 16.8 times increased odds of AMI (95% CI, 10.8-26.2; P b .0001), 4.0 times increased odds of PCI/CABG (95% CI, 3.6-6.0; P b .0001), 5.1 times increased odds of life-threatening complications (95% CI, 2.9-8.9; P b .0001), and 2.6 times increased odds of death (95% CI, 0.610.4; P = .19, not significant) as compared with patients with a negative STDsumbaseline and STDchange60min.
4. Discussion It has been previously reported that measurements of the STDsum is predictive of infarct size [11,12], adverse outcome [13,14], and total ischemic burden [10]. In patients with NSTEMI, the STDsum has been demonstrated to be an independent predictor of 1-year death or MI [15]. Recovery of STDsum of 70% or greater at 60 minutes has also been demonstrated to be predictive of successful reperfusion in patients treated with fibrinolytics [16]. Of the aforementioned studies, the studies by Aldrich et al [11], Windhausen et al [15], and French et al [16] are limited by the fact that only leads with ST-segment elevations or depressions of 1 mm or greater were summed. Other investigators have measured the sum of total STsegment elevation without any measurements of the potential additive effects of including the magnitude of ST-segment depression [23-25]. Birnbaum et al [23,24] has reported that baseline ST-segment elevation sum of anterior leads in anterior AMI does not correlate with infarct size or severity of left ventricular dysfunction. Gwechenberger et al [25] found that ST-segment elevation sum magnitude in patients treated with fibrinolytics was predictive of congestive heart failure and significant arrhythmias within 24 hours. In Gwechenberger's investigation, ROC curve analysis revealed that optimal values for ST-segment elevation sum
Utilization of ST deviation sum score
795
Table 4 Sensitivities, specificities, +LRs, and −LRs of the STDchange60min for AMI (mm; 1 mm = 0.1 mV) at the most accurate cutoff value and at the lowest cutoff values where the +LR for AMI is greater than or equal to 5, 10, and 20 in the 699 patients without STEMI on ECGbaseline
Most accurate +LR ≥5 +LR ≥10 +LR ≥20
STDchange60min
Sensitivity for AMI (%)
Specificity for AMI (%)
+LR
−LR
N2.6 N2.4 N3.5 N7.9
37.2 (26.5-48.9) 37.2 (26.5-48.9) 24.4 (15.4-35.2) 3.9 (0.8-10.8)
95.6 92.6 97.6 99.8
8.6 5.0 10.1 23.9
0.66 0.68 0.78 0.96
for predicting an adverse event were 13 mm for anterior MI and 9 mm for inferior AMI. There are only a few studies comparing a true STDsum to the ST-segment elevation sum. Schreiber et al [14] found that the STDsum is a better predictor of 24-hour complications (defined as congestive heart failure or significant rhythm disturbances) in patients treated with fibrinolytic therapy as compared with the ST-segment elevation sum. ROC analysis revealed that optimum values for prediction of adverse events were 16 mm for anterior AMI and 13 mm for inferior AMI. In our study, we found no differences in magnitudes of STDsumbaseline for anterior AMI vs inferior AMI (20.5 vs 22.0 mm, respectively), although we had insufficient numbers of patients with STEMI to investigate optimal STDsum for predicting adverse events. A true STDsum is a summation of the absolute magnitude of all 12 leads on the ECG. Because patients with reciprocal ST depressions in the presence of STEMI have larger infarcts and are at higher risk of adverse outcome [16,26], and as degree of ST depression in anterior leads correlates with infarct size in patients with posterior AMI [12,27], STDsum, theoretically, is a better representative of the total ischemic burden in patients with STEMI as compared with STsegment elevation sum [10]. This study also demonstrates that STDsum is predictive of AMI in patients with NSTEMI. Because many ECGs in patients with perceived NSTEMI have borderline elevations and depressions, a true STDsum has the potential to assist physicians in the identification of electrocardiographic injury (especially in health care providers less experienced in ECG interpretation).
Fig. 3 Graph of the STDchange60min value vs +LR ratio for AMI for the 699 study patients without STEMI on ECGbaseline.
(93.7097.1) (90.2-94.5) (94.8-97.9) (99.1-100)
This article is unique in that we report cutoff values for clinically meaningful +LRs. Traditional ROC analysis uses the optimum point as the most accurate cutoff value (lowest false-positive and false-negative rates). Because Bayes' theorem states that pretest odds × LR = posttest odds of the condition under investigation [21], we also report the lowest cutoff values in which the +LR is greater than or equal to 5, 10, and 20 to better assist the physician in decision making. Traditional teaching states that a +LR greater than or equal to 10 should drive changes in clinical decision making [21]. However, as Figs. 2 and 3 demonstrate, increasing cutoff values for STDsumbaseline and STDchange60min have their own unique +LR for predicting AMI. An important finding in this article is that patients with absence of injury on ECGbaseline who subsequently developed new injury during ECGcontinuous had a mean STDsumbaseline of 11 mm and STDchange60min of 4 mm. These values correspond to a +LR for AMI of 7 and 11, respectively. This finding supports the belief that one can use +LR analysis of STDsumbaseline and STDchange60min in selecting patients who may benefit from more aggressive anti-ischemic therapy and consideration for emergent coronary arteriogram. Furthermore, information obtained from STDchange60min values potentially could be programmed into the alarms of ECGcontinuous devices.
5. Limitations Limitations of this study predominantly are due to the retrospective design and lack of inclusion of patients not admitted for further evaluation. Theoretically, because nonadmitted patients have less STDsum baseline and STDchange60min as compared with admitted patients, inclusion of these patients would have lowered the cutoff values because specificity for each value would be increased. Although fibrinolytic therapy was the primary means of reperfusion therapy during the study period, there potentially could be selection bias in that patients with larger AMIs and hemodynamic instability may have been taken preferentially for emergent PCI before the 60-minute eligibility criteria for inclusion in this study, thus affecting cutoff values for the STDsumbaseline. We have tried to minimize the effects of bias on the STDchange60min by excluding patients with STEMI on ECGbaseline because most of these patients would be
796 expected to have larger STDchange60min values, as one of the hallmarks of STEMI is ST-segment instability [28]. Another limitation is that the STDsumbaseline was not measured until a mean of 47 minutes after ED presentation as compared with 17 minutes for the ECGbaseline. Future studies need to incorporate measurement of the STDsumbaseline as soon as possible after ED presentation with a goal of less than 10 minutes as per current American College of Cardiology/ AHA guideline recommendations [1]. Finally, another major limitation is the use of outdated modified World Health Organization criteria in effect at the time of the study. The 2007 Joint European Society of Cardiology, American College of Cardiology Foundation, American Association, and World Health Federation Task Force for Redefinition of Myocardial Infarction Universal recommends that one preferably use the 99th percentile of troponin, as the cutoff value for cardiac marker diagnosis of MI though use of CKMB is allowed as an alternative [29]. Utilization of 99th percentile of troponin undoubtedly would result in an increase in the incidence of diagnosis of NSTEMI in the subgroup of patients with final diagnosis of unstable angina, thus influencing sensitivities, specificities, and LRs. Because the diagnosis of STEMI is based on ECG interpretation and not on cardiac marker results, utilization of the 2007 criteria would not have influenced the diagnosis of STEMI in our patient population.
6. Conclusions In conclusion, increasing STDsumbaseline and STDchange60min result in increasing likelihood of AMI and risk stratify patients for AO. Clinical studies need to be performed to determine if STDsum and STDchange, in conjunction with physician pretest probability of AMI, can be used to select ED patients in need of emergent reperfusion therapy and other aggressive medical management strategies.
Acknowledgment The author would like to thank Benjamin Baker for assistance in manuscript preparation and Jason Lee, Amir Kaki, and Deyaa Wadie for assistance in data collection.
References [1] Antman EM, Anbe DT, Armstrong PW, et al. ACC/AHA guidelines for the management of patients with ST-elevation myocardial infarction: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines (Committee to Revise the 1999 Guidelines for the Management of Patients With Acute Myocardial Infarction). 2004. Available at: http://www.acc. org/qualityandscience/clinical/guidelines/stemi/STEMI%20Full% 20Text.pdf.
F.M. Fesmire [2] Antman EM, Hand M, Armstrong PW, et al. 2007 focused update of the ACC/AHA 2004 Guidelines for the Management of Patients With ST-Elevation Myocardial Infarction: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines (Writing Group to Review New Evidence and Update the ACC/AHA 2004 Guidelines for the Management of Patients With ST-Elevation Myocardial Infarction). J Am Coll Cardiol 2008;51:210-47. [3] Van de Werf F, Bax J, Betriu A, et al. Management of acute myocardial infarction in patients presenting with persistent ST-segment elevation: the Task Force on the Management of ST-Segment Elevation Acute Myocardial Infarction of the European Society of Cardiology. Eur Heart J 2008;29:2909-45. [4] Anderson JL, Adams CD, Antman EM, et al. ACC/AHA 2007 guidelines for the management of patients with unstable angina/non– ST-elevation myocardial infarction: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines (Writing Committee to Revise the 2002 Guidelines for the Management of Patients With Unstable Angina/Non-ST-Elevation Myocardial Infarction). Available at: http://content.onlinejacc.org/cgi/ reprint/50/7/e1.pdf. [5] Fesmire FM, Decker WW, Diercks DB, et al. Clinical policy: critical issues in the evaluation and management of adult patients with non– ST-segment elevation acute coronary syndromes. Ann Emerg Med 2006;48:270-301. [6] Fesmire FM, Brady WJ, Hahn S, et al. Clinical policy: indications for reperfusion therapy in emergency department patients with suspected acute myocardial infarction. Ann Emerg Med 2006;48:358-83. [7] Otto LA, Aufderheide TP. Evaluation of ST-segment elevation criteria for the prehospital electrocardiographic diagnosis of acute myocardial infarction. Ann Emerg Med 1994;23:17-24. [8] Brady WJ, Syverud SA, Beagle C, et al. Electrocardiographic ST segment elevation: the diagnosis of acute myocardial infarction by morphologic analysis of the ST-segment. Acad Emerg Med 2001; 8:961-7. [9] Brady WJ, Perron AD, Martin ML, et al. Cause of ST segment abnormality in ED chest pain patients. Am J Emerg Med 2001; 19:25-8. [10] Caldwell MA, Pelter M, Drew BJ. Estimating ischemic burden: comparison of two formulas. Am J Crit Care 1997;6:463-6. [11] Aldrich HR, Wagnes NB, Boswick J, et al. Use of initial ST-segment deviation for prediction of final electrocardiographic size of acute myocardial infarcts. Am J Cardiol 1988;61:749-53. [12] Ripa RS, Holmavang L, Maynard C, et al. Consideration of the total ST-segment deviation on the initial electrocardiogram for predicting final acute posterior myocardial infarct size in patients with maximum ST-segment deviation as depression in leads V1 through V3. A FRISC II substudy. J Electrocardiol 2005;38:180-6. [13] Hathaway WR, Peterson ED, Wagner GS, et al. Prognostic significance of the initial electrocardiogram in patients with acute myocardial infarction. JAMA 1998;279:387-91. [14] Schreiber W, Kittler H, Pieper O, Woisetschlaeger C, et al. Prediction of 24 h, nonfatal complications in patients with acute myocardial infarction receiving thrombolytic therapy by calculation of the ST segment deviation value. Can J Cardiol 2003;19:151-7. [15] Windhausen F, Hirsch A, Tijssen G, et al. ST-segment deviation on the admission electrocardiogram, treatment strategy, and outcome in non– ST-elevation acute coronary syndromes. A substudy of the invasive versus Conservative Treatment in Unstable Coronary Syndromes (ICTUS) Trial. J Electrocardiol 2007;40:408-41. [16] French JK, Ramanathan K, Stewart JT, et al. A value predicts failure of reperfusion after fibrinolytic therapy for acute myocardial infarction. Am Heart J 2003;145:508-14. [17] Fesmire FM, Eriksson SV, Stout PK, et al. Utilization of baseline STvector magnitude to identify electrocardiographic injury in patients with suspected acute myocardial infarction. Am J Emerg Med 2002; 20:535-40.
Utilization of ST deviation sum score [18] Fesmire FM, Eriksson SV. Utilization of ST-vector magnitude to risk stratify chest pain patients with left ventricular hypertrophy on the initial ECG. Ann Noninvasive Electrocardiol 2004;9:149-55. [19] Fesmire FM, Percy RF, Bardonner JB, Wharton DR, Calhoun FB. Usefulness of automated serial ECG monitoring during the ED evaluation of chest pain patients. Ann Emerg Med 1998;31:3-11. [20] WHO MONICA Project. Myocardial infarction and coronary deaths in the World Health Organization MONICA Project. Registration procedures, event rates, and case-fatality rates in 38 populations from 21 countries in four continents. Circulation 1994;90:583-612. [21] Gallagher EJ. Clinical utility of likelihood ratios. Ann Emerg Med 1998;21:391-7. [22] Schoonjans F, Zalata A, Depuydt CE, et al. Medcalc: a new computer program for medical statistics. Comput Methods Programs Biomed 1995;48:257-62. [23] Birnbaum Y, Maynard C, Wolfe S, et al. Terminal QRS distortion on admission is better than ST-segment measurements in predicting final infarct size and assessing the potential effect of thrombolytic therapy in anterior wall acute myocardial infarction. Am J Cardiol 1999;84:530-4. [24] Birnbaum Y, Criger DA, Wagner GS, et al. Prediction of the extent and severity of left ventricular dysfunction in anterior acute myocardial
797
[25]
[26]
[27]
[28]
[29]
infarction by the admission electrocardiogram. Am Heart J 2001; 141:915-24. Gwechenberger M, Schreiber W, Kittler H, et al. Prediction of early complications in patients with acute myocardial infarction by calculation of the ST value. Ann Emerg Med 1997;3:563-70. Peterson ED, Hathaway WR, Zabel KM, et al. Prognostic significance of precordial ST segment depression during inferior myocardial infarction in the thrombolytic era: results in 16,521 patients. J Am Coll Cardiol 1996;28:305-12. Asfour W, Bell S, Amkieh AM, et al. The correlation between presenting ST-segment depression and the final size of acute myocardial infarcts in patients with acute coronary syndromes. J Electrocardiolol 2000;33(Suppl):61-3. Fesmire FM, Wharton DR, Calhoun FB. Instability of ST-segments in the early stages of acute myocardial infarction in patients undergoing continuous 12-lead ECG monitoring. Am J Emerg Med 1995; 13:158-63. Thygesen K, Alpert JS, White HD, Joint ESC/ACCF/AHA/WHF Task Force for the Redefinition of Myocardial Infarction. Universal definition of myocardial infarction. Circulation 2007; 116:2634-53.
www.elsevier.com/locate/ajem
Original Contribution
Utilization of ST-segment deviation sum and change scores to identify acute myocardial infarction☆ Francis M. Fesmire MD ⁎ Emergency Heart Center, Erlanger Medical Center and Division of Emergency Medicine, University of Tennessee Chattanooga College of Medicine, Chattanooga, TN 37405, USA Received 4 February 2009; revised 31 March 2009; accepted 1 April 2009
Abstract Objective: No information is currently available regarding the optimal cutoff values of the baseline STsegment deviation sum (STDsumbaseline) and 60-minute ST-segment deviation change (STDchange60min) for predicting acute myocardial infarction (AMI). Methods: A retrospective study was performed in 783 admitted patients with chest pain who had suspected acute coronary syndrome and absence of left ventricular hypertrophy or bundle branch block on the initial electrocardiogram (ECG). The STDsumbaseline was defined as the sum in millimeters (1 mm = 0.1 mV) of the absolute value of ST-segment deviations in all 12 leads at the initiation of continuous 12-lead ECG monitoring session. The STDchange60min was defined as the absolute value of the difference between the baseline and 60-minute STDsum. Three cutoff values are reported and represent the smallest values in which the positive likelihood ratio (+LR) for AMI was greater than or equal to 5, 10, and 20, respectively. Results: Acute myocardial infarction occurred in 162 (20.7%) patients. The smallest cutoff value of the STDsumbaseline for AMI with a +LR equal to or greater than 5, 10, and 20 was 9.6, 12.4, and 14.1 mm, respectively. In the subset of 699 patients without ST-segment elevation AMI on initial ECG, the smallest cutoff value of the STDchange60min for AMI with a +LR equal to or greater than 5, 10, and 20 was 2.4, 3.5, and 7.9 mm, respectively. Conclusions: Clinical studies need to be performed to determine if STDsum and STDchange, in conjunction with physician pretest probability of AMI, can be used to select patients who may benefit from emergent reperfusion therapy and other aggressive medical management strategies. © 2010 Elsevier Inc. All rights reserved.
1. Introduction
☆ Presented at the Scientific Assembly of the American College of Emergency Physicians (Research Forum) at Washington, DC, on September 26, 2005. ⁎ UT College of Medicine, PO Box 4045, Chattanooga, TN 37405, USA. E-mail address: [email protected].
0735-6757/$ – see front matter © 2010 Elsevier Inc. All rights reserved. doi:10.1016/j.ajem.2009.04.004
Identification of ST-segment changes diagnostic of injury or ischemia on the 12-lead electrocardiogram (ECG) is of paramount importance in the identification of patients with acute myocardial infarction (AMI) and non-AMI acute coronary syndrome (ACS) who may benefit from emergent reperfusion therapy and other aggressive anti-ischemic therapies [1-6]. Current practice guidelines from the American College of Cardiology and American Heart
Utilization of ST deviation sum score Association (AHA) recommend emergent reperfusion therapy in patients with new or presumably new left bundle branch block (BBB) on the ECG, in patients with STsegment elevations of 1 mm in 2 or more contiguous leads, or in patients with ST-segment depressions consistent with true posterior AMI [1]. Interpretation of the magnitude and morphology of STsegment elevations and depressions is often subjective and varies depending on the experience and expertise of the health care provider. Furthermore, the multitude of confounding factors that result in ST-segment elevations and depressions can make interpretation of electrocardiographic injury and ischemia difficult [6-9]. Studies have demonstrated that ST-segment elevation is frequently seen in emergency department (ED) patients with chest pain and that only approximately 15% to 30% of these patients are experiencing ACS [8,9]. Left ventricular hypertrophy (LVH) is the most common cause of non-ACS STsegment elevation on the ECG and occurs in approximately 30% of this subset of patients [8,9]. Other nonACS causes of ST-segment elevation are benign early repolarization, pericarditis, left ventricular aneurysm, paced rhythm, and BBB [8,9]. A potential method to assist health care providers in the identification of ST-segment changes suggestive of electrocardiographic injury and ischemia is to use the absolute value of the sum of ST-segment deviations in all 12 leads on the ECG (STDsum) to better predict the total ischemic burden (Fig. 1) [10]. In patients with ST-segment elevation AMI (STEMI), the STDsum has been demonstrated to be predictive of infarct size [11,12] and adverse events [13,14]. In patients with non-STEMI (NSTEMI), the STDsum has been demonstrated to be an independent predictor of 1-year death or myocardial infarction (MI) [15]. Recovery of STDsum of 70% or greater at 60 minutes has also been demonstrated to be predictive of successful reperfusion in patients treated with fibrinolytics [16]. It has been previously reported that increasing ST-vector magnitude on the ECG is associated with increasing likelihood of AMI [17,18]. ST-vector magnitude represents the sum of ST-segment deviation in 3-dimensional space and, in theory, should correlate with the STDsum [17,18]. Currently, no information is available regarding optimal STDsum values for predicting AMI in ED patients with chest pain presenting with possible ACS. Theoretically, this optimal value corresponds to electrocardiographic injury and high-risk ischemia and could potentially be a criterion for emergent reperfusion therapy and other aggressive medical management strategies, as well as a tool to assist physicians and other health care providers in the evaluation, treatment, and disposition of patients with chest pain. In addition, although decreases in STDsum over time have been shown to be predictive of successful reperfusion with fibrinolytic therapy [16], no information is available regarding whether changes in STDsum are predictive of AMI in patients without STEMI on the baseline ECG
791 (ECGbaseline) obtained on initial ED presentation. In this study, we investigate the ability of an automated computerized measurement of the baseline STDsum (STDsumbaseline) in all study patients and changes in STDsum at 60 minutes (STDchange60min) in the subset of patients without STEMI on ECGbaseline to predict AMI in ED chest pain patients with suspected ACS.
2. Methods 2.1. Study design This is a retrospective analysis of a previously reported prospective investigation on the utility of automated continuous 12-lead ECG monitoring (ECGcontinuous) for identification of injury and ischemia in patients with chest pain [19]. The study was approved by the institutional review board.
2.2. Setting The study was conducted at Erlanger Medical Center, an adult urban tertiary referral center. The hospital is affiliated with the University of Tennessee College of Medicine and has an annual ED census of 45 000 patients.
2.3. Selection of participants The study population was derived from a convenience sample of 1000 patients with chest pain who underwent at least 60 minutes of ECGcontinuous during the initial ED evaluation and were subsequently admitted to the hospital for further treatment and/or evaluation. Exclusion criteria for purposes of this current investigation included patients who were discharged from ED, patients undergoing less than 60 minutes of ECGcontinuous, and patients with electrocardiographic LVH, BBB (right, left, or atypical), or ventricular paced rhythm on the ECGbaseline obtained on initial ED presentation, as these patients already have significant ST-segment deviations on the ECG secondary to abnormalities in repolarization [8,9,17].
2.4. Measurements The ECGbaseline was obtained on presentation with the use of the PageWriter Xli Cardiograph (Philips, Andover, MA). Methodology for interpretation of the ECGbaseline has been reported previously [19]. The ELI-100 ECG machine with Lifenet STM Monitor (Mortara Instruments, Milwaukee, WI) was used to perform ECGcontinuous. ST-segment magnitudes were measured every 20 seconds 60 milliseconds past the J point. Printouts of the 12-lead ST-segment trends consist of an ST-segment magnitude for each of the
792 12 leads, a sum of ST-segment magnitude for 4 regions (anterior precordial leads, lateral precordial leads, inferior limb leads, and lateral limb leads), and an absolute sum of the ST-segment deviations in all 12 leads (STDsum). The STDsumbaseline was defined as the STDsum of the initial ECGcontinuous obtained at the beginning of the monitoring session. The STDchange60min was defined as the absolute value of the difference between the baseline STDsum and 60-minute STDsum.
2.5. Main outcomes Acute myocardial infarction was defined according to World Health Organization Criteria in effect at time of study [20]. Acute myocardial infarction on initial presentation was diagnosed if there was 20 minutes or more of chest pain and (1) a serial rise and fall of creatine kinase (CK) and CK-MB fraction as measured by the Stratus Fluorometric Enzyme Immunoassay (Dade Behring, Deerfield, IL), provided that the maximum CK-MB value was 12 ng/mL or greater and the CK-MB index was at least 4% within 24 hours of ED presentation; or (2) new Q-wave formation during the initial 24 hours after presentation; or (3) patient death from cardiac causes during the initial 24 hours of presentation; or (4) coronary arteriogram with documented acute occlusion of a major coronary vessel within 24 hours of presentation. Recent MI was diagnosed in patients with MI whose serial CK-MB measurements were already on the falling curve at presentation. In-hospital MI was diagnosed if the serial CKMB measurements did not meet criteria for AMI until more than 24 hours after presentation. Unstable angina was
F.M. Fesmire diagnosed in patients who were without CK-MB or ECG criteria for MI and whose in-hospital attending physician diagnosed ACS-related chest pain. Adverse outcome (AO) was defined as MI on presentation (AMI or recent MI), inhospital AMI, percutaneous coronary intervention (PCI), coronary artery bypass grafting (CABG), life-threatening complication, or death. Life-threatening complications were defined as ventricular fibrillation, sustained ventricular tachycardia, third-degree atrioventricular block, cardiogenic shock, or cardiac arrest secondary to severe bradycardia or pulseless electrical activity.
2.6. Data analysis Data were recorded on a standardized form and placed in a computer database. Positive (+) and negative (−) likelihood ratios (LRs) were calculated according to the formula +LR = sensitivity/(1 − specificity) and −LR = (1 − sensitivity)/specificity [21]. Receiver Operator Characteristic (ROC) curves were analyzed using MedCalc 9.6 (MedCalc Software, Belgium) [22]. Cutoff values for STDsumbaseline and STDchange60min were calculated at the most accurate cutoff value (value with the lowest falsenegative and false-positive rates) and at the 3 smallest cutoff values where the +LR for AMI is greater than or equal to 5, 10, and 20, respectively. All other data analyses were performed with SYSTAT 10.0 (SPSS Inc, Chicago, IL). Sample means were compared with the 2-sample unpaired t test and are reported as means ± SD. Categorical data were compared with χ2 test with the Yates correction for sample size. A P value ≤.05 was considered significant.
Fig. 1 ECG not meeting standard criteria for emergent reperfusion therapy but with an STDsumbaseline of 18 mm with a corresponding +LR for AMI of 35.
Utilization of ST deviation sum score
793
3. Results
Table 2 Mean ± SD STDsumbaseline and STDchange60min (mm; 1 mm = 0.1 mV) according to diagnosis and STEMI location in the 783 study patients
3.1. Characteristics of study subjects
STDsumbaseline STDchange60min
During the study period, a total of 1000 admitted patients with chest pain underwent a minimum of 60 minutes of ECGcontinuous during the initial ED evaluation. A total of 217 patients were excluded because of findings of LVH, BBB, or ventricular paced rhythm on the ECGbaseline, leaving a study population of 783 patients. The ECGbaseline was obtained 16.7 ± 16.5 minutes after ED arrival and ECGcontinuous was initiated 46.5 ± 46.2 minutes after ED arrival. The mean duration of ST-segment monitoring was 127.7 ± 41.2 minutes. Of the 162 patients with AMI, 84 (10.7%) patients had STEMI on ECGbaseline and 78 (10.0%) patients had NSTEMI on ECGbaseline. Of the 78 patients with NSTEMI on ECGbaseline, 26 (33.3%) patients developed new injury during the ECGcontinuous monitoring session. In the patients with non-AMI, discharge diagnosis consisted of 17 (2.2%) patients with recent MI, 235 (30.0%) patients with unstable angina, and 369 (47.1%) patients with non-ACS chest pain. A total of 256 (32.7%) patients experienced 1 or more of the following AO events: 179 (22.9%) with AMI or recent MI, 97 (12.4%) with PCI, 63 (8.0%) with CABG, 57 (7.3%) with life-threatening complications, and 9 (1.1%) with death. Of the 84 patients with STEMI on ECGbaseline, 70 (83.3%) were treated in the ED with fibrinolytics, and 12 (14.3%), with emergent PCI. Of the 26 patients with NSTEMI on ECGbaseline who developed new injury on ECGcontinuous, 14 (53.8%) patients were treated in the ED with fibrinolytics, and 3 (11.5%) patients, with emergent PCI. Table 1 summarizes demographic features, mean CKMBbaseline and ΔCK-MB2hr, mean STDsumbaseline, and STDchange60min. Patients with AO were older and more likely to be male. Patients with AO had higher mean CK-MBbaseline, ΔCK-MB2hr, STDsumbaseline, and STDchange60min. Table 1 Demographic characteristics, CK-MBbaseline (mg/dL), ΔCK-MB2hr (mg/dL), STDsumbaseline (mm; 1 mm = 0.1 mV), and STDchange60min (mm; 1 mm = 0.1 mV) in 783 study patients with and without in-hospital AO Without AO (n = 527) Age (y) Male Race White African American Previous MI Previous PCI/CABG CK-MBbaseline ΔCK-MB2hr STDsumbaseline STDchange60min
With AO (n = 256)
P
54.2 ± 13.0 58.0 ± 12.1 300 (57.0) 180 (70.3)
b.0001 b.0001
429 (81.4) 98 (18.6) 139 (26.4) 121 (23.0) 0.9 ± 1.6 0.1 ± 0.9 5.5 ± 3.6 0.8 ± 1.0
.23 (NS) .23 (NS) .36 (NS) .95 (NS) b.0001 b.0001 b.0001 b.0001
218 (85.2) 38 (14.8) 59 (23.0) 60 (23.4) 10.3 ± 30.3 15.8 ± 42.3 11.8 ± 10.4 3.3 ± 4.8
Values are presented as mean ± SD or no. (%).
Patients with non-AMI (n = 621) Recent MI (n = 17) Unstable angina (n = 235) Non-ACS (n = 369) NSTEMI on ECGbaseline (n = 78) New injury on ECGcontinuous (n = 26) No injury on ECGcontinuous (n = 52) STEMI on ECGbaseline (n = 84) Anterior/anterolateral (n = 27) Inferior/inferolateral (n = 53) Other (n = 4)
5.6 ± 3.7
0.8 ± 1.1
6.4 ± 3.9 5.5 ± 4.1 5.7 ± 3.4 8.2 ± 5.2
0.5 ± 0.5 1.0 ± 1.4 0.7 ± 0.8 2.5 ± 3.4
11.1 ± 5.2
4.0 ± 3.3
6.7 ± 4.7
1.7 ± 3.3
21.8 ± 11.8
6.7 ± 6.4
20.5 ± 11.1
5.8 ± 4.8
22.0 ± 12.4
7.2 ± 7.0
27.0 ± 6.4
6.2 ± 7.1
Table 2 summarizes mean STDsum base line and STDchange60min according to diagnosis and STEMI location. There was an incremental increase in STDsumbaseline and STDchange60min for patients with non-AMI, NSTEMI, and STEMI. No statistical differences were seen for mean STDsumbaseline and STDchange60min between patients with recent MI, unstable angina, and non-ACS. Comparing patients with non-AMI chest pain with patients with NSTEMI revealed a statistically significant increase in both STDsumbaseline (5.6 vs 8.2 mm, respectively; P b .0001) and STDchange60min (0.8 vs 2.5 mm, respectively; P b .0001). Comparing patients with NSTEMI with patients with STEMI revealed a statistically significant increase in both STDsumbaseline (8.2 vs 21.8 mm, respectively; P b .0001) and STDchange60min (2.5 vs 6.7 mm, respectively; P b .0001). There were no statistical differences in comparing STDsumbaseline (20.5 vs 22.0 mm, respectively; P = .5) and STDchange60min (5.8 vs 7.2 mm, respectively; P = .3) for anterior/anterolateral STEMI vs inferior/inferolateral STEMI. Comparing the 26 patients with NSTEMI on ECGbaseline who developed new injury on ECGcontinuous with the 52 patients who did not develop STEMI revealed a statistically significant increase in both STDsumbaseline (11.1 vs 6.7 mm, respectively; P = .001) and STDchange60min (4.0 vs 1.7 mm, respectively; P b .01).
3.2. Likelihood ratio analysis The ROC curve area of STDsumbaseline was 0.81 (95% confidence interval [CI], 0.78-0.83) for AMI and 0.71 (95% CI, 0.68-0.74) for AO in the 783 study patients. Table 3 summarizes sensitivity and specificity data for predicting
794
F.M. Fesmire
Table 3 Sensitivities, specificities, +LRs, and −LRs of the STDsumbaseline (mm; 1 mm = 0.1 mV) for AMI at the most accurate cutoff value and at the lowest cutoff value where the +LR for AMI is greater than or equal to 5, 10, and 20 in the 783 study patients
Most accurate +LR ≥5 +LR ≥10 +LR ≥20
Baseline STDsumbaseline
Sensitivity for predicting AMI (%)
Specificity for predicting AMI (%)
+LR
−LR
N8.5 N9.6 N12.4 N14.1
65.4 59.9 46.9 42.6
84.5 (81.5-87.3) 88.1 (85.3-90.5) 95.3 (93.4-96.9) 97.9 (96.4-98.9)
4.2 5.0 10.1 20.4
0.41 0.46 0.56 0.59
(57.6-72.7) (51.9-67.5) (39.0-54.9) (34.9-50.6)
AMI at the most accurate cutoff point on the ROC curve and at the 3 cutoff points where the +LR is greater than or equal to 5, 10, and 20, respectively. The 3 STDsumbaseline cutoff values for predicting AMI with +LR greater than or equal to 5, 10, and 20 were 9.6, 12.4, and 14.1 mm, respectively. Fig. 2 demonstrates a graph of baseline STDsumbaseline vs +LR for AMI and demonstrates that increasing values are correlated with increasing likelihood of AMI. Combining initial ECG finding of injury with an STDsumbaseline value greater than 12.4 mm (ie, +LR for AMI N10) identified an additional 13 patients with a final diagnosis of AMI.
3.3. Subset analysis in NSTEMI patients In the subset of 699 patients without STEMI on ECGbaseline, the ROC curve area of STDchange60min was 0.70 (95% CI, 0.67-0.74) for AMI and 0.63 (95% CI, 0.59-0.66) for AO. Table 4 summarizes sensitivity and specificity data of the STDchange60min for predicting AMI at the most accurate cutoff point on the ROC curve and at the 3 cutoff points where the +LR is greater than or equal to 5, 10, and 20. The 3 STDchange60min cutoff values for predicting AMI with +LR greater than or equal to 5, 10, and 20, were 2.4, 3.5, and 7.9 mm, respectively. Fig. 3 demonstrates a graph of STDchange60min value vs +LR for AMI and demonstrates that increasing values are correlated with increasing likelihood of AMI. An STDchange60min greater than 3.5 mm (ie, +LR for AMI N10) identified an additional 19 patients with a final diagnosis of AMI.
Fig. 2 Graph of STDsumbaseline value vs +LR ratio for AMI for the 783 study patients.
3.4. Risk stratification Combining the 2 tests at the cutoff value with a +LR greater than or equal to 10 for AMI (ie, baseline STDsumbaseline N12.4 mm in all patients and/or STDchange60min N3.5 mm in patients with NSTEMI) has a sensitivity and specificity for identification of AMI of 54.3% (95% CI, 46.3-62.2) and 93.4% (95% CI, 91.1-95.2), respectively, and effectively risk stratifies patients for AO. Patients with either an STDsumbaseline greater than 12.4 mm or an STDchange60min greater than 3.5 mm had 16.8 times increased odds of AMI (95% CI, 10.8-26.2; P b .0001), 4.0 times increased odds of PCI/CABG (95% CI, 3.6-6.0; P b .0001), 5.1 times increased odds of life-threatening complications (95% CI, 2.9-8.9; P b .0001), and 2.6 times increased odds of death (95% CI, 0.610.4; P = .19, not significant) as compared with patients with a negative STDsumbaseline and STDchange60min.
4. Discussion It has been previously reported that measurements of the STDsum is predictive of infarct size [11,12], adverse outcome [13,14], and total ischemic burden [10]. In patients with NSTEMI, the STDsum has been demonstrated to be an independent predictor of 1-year death or MI [15]. Recovery of STDsum of 70% or greater at 60 minutes has also been demonstrated to be predictive of successful reperfusion in patients treated with fibrinolytics [16]. Of the aforementioned studies, the studies by Aldrich et al [11], Windhausen et al [15], and French et al [16] are limited by the fact that only leads with ST-segment elevations or depressions of 1 mm or greater were summed. Other investigators have measured the sum of total STsegment elevation without any measurements of the potential additive effects of including the magnitude of ST-segment depression [23-25]. Birnbaum et al [23,24] has reported that baseline ST-segment elevation sum of anterior leads in anterior AMI does not correlate with infarct size or severity of left ventricular dysfunction. Gwechenberger et al [25] found that ST-segment elevation sum magnitude in patients treated with fibrinolytics was predictive of congestive heart failure and significant arrhythmias within 24 hours. In Gwechenberger's investigation, ROC curve analysis revealed that optimal values for ST-segment elevation sum
Utilization of ST deviation sum score
795
Table 4 Sensitivities, specificities, +LRs, and −LRs of the STDchange60min for AMI (mm; 1 mm = 0.1 mV) at the most accurate cutoff value and at the lowest cutoff values where the +LR for AMI is greater than or equal to 5, 10, and 20 in the 699 patients without STEMI on ECGbaseline
Most accurate +LR ≥5 +LR ≥10 +LR ≥20
STDchange60min
Sensitivity for AMI (%)
Specificity for AMI (%)
+LR
−LR
N2.6 N2.4 N3.5 N7.9
37.2 (26.5-48.9) 37.2 (26.5-48.9) 24.4 (15.4-35.2) 3.9 (0.8-10.8)
95.6 92.6 97.6 99.8
8.6 5.0 10.1 23.9
0.66 0.68 0.78 0.96
for predicting an adverse event were 13 mm for anterior MI and 9 mm for inferior AMI. There are only a few studies comparing a true STDsum to the ST-segment elevation sum. Schreiber et al [14] found that the STDsum is a better predictor of 24-hour complications (defined as congestive heart failure or significant rhythm disturbances) in patients treated with fibrinolytic therapy as compared with the ST-segment elevation sum. ROC analysis revealed that optimum values for prediction of adverse events were 16 mm for anterior AMI and 13 mm for inferior AMI. In our study, we found no differences in magnitudes of STDsumbaseline for anterior AMI vs inferior AMI (20.5 vs 22.0 mm, respectively), although we had insufficient numbers of patients with STEMI to investigate optimal STDsum for predicting adverse events. A true STDsum is a summation of the absolute magnitude of all 12 leads on the ECG. Because patients with reciprocal ST depressions in the presence of STEMI have larger infarcts and are at higher risk of adverse outcome [16,26], and as degree of ST depression in anterior leads correlates with infarct size in patients with posterior AMI [12,27], STDsum, theoretically, is a better representative of the total ischemic burden in patients with STEMI as compared with STsegment elevation sum [10]. This study also demonstrates that STDsum is predictive of AMI in patients with NSTEMI. Because many ECGs in patients with perceived NSTEMI have borderline elevations and depressions, a true STDsum has the potential to assist physicians in the identification of electrocardiographic injury (especially in health care providers less experienced in ECG interpretation).
Fig. 3 Graph of the STDchange60min value vs +LR ratio for AMI for the 699 study patients without STEMI on ECGbaseline.
(93.7097.1) (90.2-94.5) (94.8-97.9) (99.1-100)
This article is unique in that we report cutoff values for clinically meaningful +LRs. Traditional ROC analysis uses the optimum point as the most accurate cutoff value (lowest false-positive and false-negative rates). Because Bayes' theorem states that pretest odds × LR = posttest odds of the condition under investigation [21], we also report the lowest cutoff values in which the +LR is greater than or equal to 5, 10, and 20 to better assist the physician in decision making. Traditional teaching states that a +LR greater than or equal to 10 should drive changes in clinical decision making [21]. However, as Figs. 2 and 3 demonstrate, increasing cutoff values for STDsumbaseline and STDchange60min have their own unique +LR for predicting AMI. An important finding in this article is that patients with absence of injury on ECGbaseline who subsequently developed new injury during ECGcontinuous had a mean STDsumbaseline of 11 mm and STDchange60min of 4 mm. These values correspond to a +LR for AMI of 7 and 11, respectively. This finding supports the belief that one can use +LR analysis of STDsumbaseline and STDchange60min in selecting patients who may benefit from more aggressive anti-ischemic therapy and consideration for emergent coronary arteriogram. Furthermore, information obtained from STDchange60min values potentially could be programmed into the alarms of ECGcontinuous devices.
5. Limitations Limitations of this study predominantly are due to the retrospective design and lack of inclusion of patients not admitted for further evaluation. Theoretically, because nonadmitted patients have less STDsum baseline and STDchange60min as compared with admitted patients, inclusion of these patients would have lowered the cutoff values because specificity for each value would be increased. Although fibrinolytic therapy was the primary means of reperfusion therapy during the study period, there potentially could be selection bias in that patients with larger AMIs and hemodynamic instability may have been taken preferentially for emergent PCI before the 60-minute eligibility criteria for inclusion in this study, thus affecting cutoff values for the STDsumbaseline. We have tried to minimize the effects of bias on the STDchange60min by excluding patients with STEMI on ECGbaseline because most of these patients would be
796 expected to have larger STDchange60min values, as one of the hallmarks of STEMI is ST-segment instability [28]. Another limitation is that the STDsumbaseline was not measured until a mean of 47 minutes after ED presentation as compared with 17 minutes for the ECGbaseline. Future studies need to incorporate measurement of the STDsumbaseline as soon as possible after ED presentation with a goal of less than 10 minutes as per current American College of Cardiology/ AHA guideline recommendations [1]. Finally, another major limitation is the use of outdated modified World Health Organization criteria in effect at the time of the study. The 2007 Joint European Society of Cardiology, American College of Cardiology Foundation, American Association, and World Health Federation Task Force for Redefinition of Myocardial Infarction Universal recommends that one preferably use the 99th percentile of troponin, as the cutoff value for cardiac marker diagnosis of MI though use of CKMB is allowed as an alternative [29]. Utilization of 99th percentile of troponin undoubtedly would result in an increase in the incidence of diagnosis of NSTEMI in the subgroup of patients with final diagnosis of unstable angina, thus influencing sensitivities, specificities, and LRs. Because the diagnosis of STEMI is based on ECG interpretation and not on cardiac marker results, utilization of the 2007 criteria would not have influenced the diagnosis of STEMI in our patient population.
6. Conclusions In conclusion, increasing STDsumbaseline and STDchange60min result in increasing likelihood of AMI and risk stratify patients for AO. Clinical studies need to be performed to determine if STDsum and STDchange, in conjunction with physician pretest probability of AMI, can be used to select ED patients in need of emergent reperfusion therapy and other aggressive medical management strategies.
Acknowledgment The author would like to thank Benjamin Baker for assistance in manuscript preparation and Jason Lee, Amir Kaki, and Deyaa Wadie for assistance in data collection.
References [1] Antman EM, Anbe DT, Armstrong PW, et al. ACC/AHA guidelines for the management of patients with ST-elevation myocardial infarction: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines (Committee to Revise the 1999 Guidelines for the Management of Patients With Acute Myocardial Infarction). 2004. Available at: http://www.acc. org/qualityandscience/clinical/guidelines/stemi/STEMI%20Full% 20Text.pdf.
F.M. Fesmire [2] Antman EM, Hand M, Armstrong PW, et al. 2007 focused update of the ACC/AHA 2004 Guidelines for the Management of Patients With ST-Elevation Myocardial Infarction: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines (Writing Group to Review New Evidence and Update the ACC/AHA 2004 Guidelines for the Management of Patients With ST-Elevation Myocardial Infarction). J Am Coll Cardiol 2008;51:210-47. [3] Van de Werf F, Bax J, Betriu A, et al. Management of acute myocardial infarction in patients presenting with persistent ST-segment elevation: the Task Force on the Management of ST-Segment Elevation Acute Myocardial Infarction of the European Society of Cardiology. Eur Heart J 2008;29:2909-45. [4] Anderson JL, Adams CD, Antman EM, et al. ACC/AHA 2007 guidelines for the management of patients with unstable angina/non– ST-elevation myocardial infarction: a report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines (Writing Committee to Revise the 2002 Guidelines for the Management of Patients With Unstable Angina/Non-ST-Elevation Myocardial Infarction). Available at: http://content.onlinejacc.org/cgi/ reprint/50/7/e1.pdf. [5] Fesmire FM, Decker WW, Diercks DB, et al. Clinical policy: critical issues in the evaluation and management of adult patients with non– ST-segment elevation acute coronary syndromes. Ann Emerg Med 2006;48:270-301. [6] Fesmire FM, Brady WJ, Hahn S, et al. Clinical policy: indications for reperfusion therapy in emergency department patients with suspected acute myocardial infarction. Ann Emerg Med 2006;48:358-83. [7] Otto LA, Aufderheide TP. Evaluation of ST-segment elevation criteria for the prehospital electrocardiographic diagnosis of acute myocardial infarction. Ann Emerg Med 1994;23:17-24. [8] Brady WJ, Syverud SA, Beagle C, et al. Electrocardiographic ST segment elevation: the diagnosis of acute myocardial infarction by morphologic analysis of the ST-segment. Acad Emerg Med 2001; 8:961-7. [9] Brady WJ, Perron AD, Martin ML, et al. Cause of ST segment abnormality in ED chest pain patients. Am J Emerg Med 2001; 19:25-8. [10] Caldwell MA, Pelter M, Drew BJ. Estimating ischemic burden: comparison of two formulas. Am J Crit Care 1997;6:463-6. [11] Aldrich HR, Wagnes NB, Boswick J, et al. Use of initial ST-segment deviation for prediction of final electrocardiographic size of acute myocardial infarcts. Am J Cardiol 1988;61:749-53. [12] Ripa RS, Holmavang L, Maynard C, et al. Consideration of the total ST-segment deviation on the initial electrocardiogram for predicting final acute posterior myocardial infarct size in patients with maximum ST-segment deviation as depression in leads V1 through V3. A FRISC II substudy. J Electrocardiol 2005;38:180-6. [13] Hathaway WR, Peterson ED, Wagner GS, et al. Prognostic significance of the initial electrocardiogram in patients with acute myocardial infarction. JAMA 1998;279:387-91. [14] Schreiber W, Kittler H, Pieper O, Woisetschlaeger C, et al. Prediction of 24 h, nonfatal complications in patients with acute myocardial infarction receiving thrombolytic therapy by calculation of the ST segment deviation value. Can J Cardiol 2003;19:151-7. [15] Windhausen F, Hirsch A, Tijssen G, et al. ST-segment deviation on the admission electrocardiogram, treatment strategy, and outcome in non– ST-elevation acute coronary syndromes. A substudy of the invasive versus Conservative Treatment in Unstable Coronary Syndromes (ICTUS) Trial. J Electrocardiol 2007;40:408-41. [16] French JK, Ramanathan K, Stewart JT, et al. A value predicts failure of reperfusion after fibrinolytic therapy for acute myocardial infarction. Am Heart J 2003;145:508-14. [17] Fesmire FM, Eriksson SV, Stout PK, et al. Utilization of baseline STvector magnitude to identify electrocardiographic injury in patients with suspected acute myocardial infarction. Am J Emerg Med 2002; 20:535-40.
Utilization of ST deviation sum score [18] Fesmire FM, Eriksson SV. Utilization of ST-vector magnitude to risk stratify chest pain patients with left ventricular hypertrophy on the initial ECG. Ann Noninvasive Electrocardiol 2004;9:149-55. [19] Fesmire FM, Percy RF, Bardonner JB, Wharton DR, Calhoun FB. Usefulness of automated serial ECG monitoring during the ED evaluation of chest pain patients. Ann Emerg Med 1998;31:3-11. [20] WHO MONICA Project. Myocardial infarction and coronary deaths in the World Health Organization MONICA Project. Registration procedures, event rates, and case-fatality rates in 38 populations from 21 countries in four continents. Circulation 1994;90:583-612. [21] Gallagher EJ. Clinical utility of likelihood ratios. Ann Emerg Med 1998;21:391-7. [22] Schoonjans F, Zalata A, Depuydt CE, et al. Medcalc: a new computer program for medical statistics. Comput Methods Programs Biomed 1995;48:257-62. [23] Birnbaum Y, Maynard C, Wolfe S, et al. Terminal QRS distortion on admission is better than ST-segment measurements in predicting final infarct size and assessing the potential effect of thrombolytic therapy in anterior wall acute myocardial infarction. Am J Cardiol 1999;84:530-4. [24] Birnbaum Y, Criger DA, Wagner GS, et al. Prediction of the extent and severity of left ventricular dysfunction in anterior acute myocardial
797
[25]
[26]
[27]
[28]
[29]
infarction by the admission electrocardiogram. Am Heart J 2001; 141:915-24. Gwechenberger M, Schreiber W, Kittler H, et al. Prediction of early complications in patients with acute myocardial infarction by calculation of the ST value. Ann Emerg Med 1997;3:563-70. Peterson ED, Hathaway WR, Zabel KM, et al. Prognostic significance of precordial ST segment depression during inferior myocardial infarction in the thrombolytic era: results in 16,521 patients. J Am Coll Cardiol 1996;28:305-12. Asfour W, Bell S, Amkieh AM, et al. The correlation between presenting ST-segment depression and the final size of acute myocardial infarcts in patients with acute coronary syndromes. J Electrocardiolol 2000;33(Suppl):61-3. Fesmire FM, Wharton DR, Calhoun FB. Instability of ST-segments in the early stages of acute myocardial infarction in patients undergoing continuous 12-lead ECG monitoring. Am J Emerg Med 1995; 13:158-63. Thygesen K, Alpert JS, White HD, Joint ESC/ACCF/AHA/WHF Task Force for the Redefinition of Myocardial Infarction. Universal definition of myocardial infarction. Circulation 2007; 116:2634-53.