Author + information
- Received August 13, 1997
- Revision received March 11, 1998
- Accepted March 18, 1998
- Published online July 1, 1998.
- Peter J. Kudenchuk, MD, FACCa,∗,*,
- Charles Maynard, PhDa,∗,
- Leonard A. Cobb, MD, FACCa,∗,
- Mark Wirkusa,∗,
- Jenny S. Martin, RNa,∗,
- J.Ward Kennedy, MD, FACCa,∗ and
- W.Douglas Weaver, MD, FACCa,∗
- ↵*Address for correspondence: Dr. Peter J. Kudenchuk, Division of Cardiology, Box 356422, University of Washington, Seattle, Washington 98195-6422
- for the MITI Investigators†
Objectives. We sought to determine whether the prehospital electrocardiogram (ECG) improves the diagnosis of an acute coronary syndrome.
Background. The ECG is the most widely used screening test for evaluating patients with chest pain.
Methods. Prehospital and in-hospital ECGs were obtained in 3,027 consecutive patients with symptoms of suspected acute myocardial infarction, 362 of whom were randomized to prehospital versus hospital thrombolysis and 2,665 of whom did not participate in the randomized trial. Prehospital and hospital records were abstracted for clinical characteristics and diagnostic outcome.
Results. ST segment and T and Q wave abnormalities suggestive of myocardial ischemia or infarction were more common on both the prehospital and hospital ECGs of patients with as compared with those without acute coronary syndromes (p ≤ 0.00001). Those with prehospital thrombolysis were more likely to show resolution of ST segment elevation by the time of hospital admission (14% vs. 5% in patients treated in the hospital, p = 0.004). In patients not considered for prehospital thrombolysis, both persistent and transient ST segment and T or Q wave abnormalities discriminated those with from those without acute coronary ischemia or infarction. Compared with ST segment elevation on a single ECG, added consideration of dynamic changes in ST segment elevation between serial ECGs improved the sensitivity for an acute coronary syndrome from 34% to 46% and reduced specificity from 96% to 93% (both p < 0.00004). Overall, compared with abnormalities observed on a single ECG, consideration of serial evolution in ST segment, T or Q wave or left bundle branch block (LBBB) abnormalities between the prehospital and initial hospital ECG improved the diagnostic sensitivity for an acute coronary syndrome from 80% to 87%, with a fall in specificity from 60% to 50% (both p < 0.000006).
Conclusions. ECG abnormalities are an early manifestation of acute coronary syndromes and can be identified by the prehospital ECG. Compared with a single ECG, the additional effect of evolving ST segment, T or Q waves or LBBB between serially obtained prehospital and hospital ECGs enhanced the diagnosis of acute coronary syndromes, but with a fall in specificity.
Early diagnosis is pivotal in the contemporary management of patients with acute coronary syndromes and is an essential component of a rapid intervention strategy. Although newer enzymatic assays offer promise for earlier identification of patients with suspected acute myocardial infarction (AMI), the electrocardiogram (ECG) remains the most widely used initial screening test for evaluating patients with chest pain. Early electrocardiography may facilitate early intervention by identifying patients with true ischemic syndromes sooner. In contrast, if obtained too soon, patients with ECG abnormalities that are subtle or too nonspecific for definitive diagnosis may be misidentified and inappropriately treated. The purposes of this study were to characterize the frequency and serial evolution of very early ECG changes in patients with and without acute myocardial ischemia or infarction and to determine whether ECG screening before hospital admission improves the diagnosis of an acute coronary syndrome in patients with chest pain of suspected cardiac origin.
The Myocardial Infarction Triage and Intervention (MITI) Project was a prospective, randomized trial comparing prehospital with in-hospital thrombolysis in patients with suspected AMI; the design and results of MITI have been reported previously (1). Briefly, consecutive patients with suspected symptoms of AMI and no clinical contraindications to thrombolytic therapy received a 12-lead prehospital ECG, which was interpreted by computer and transmitted by cellular telephone to a remote emergency department for review by a physician. During the treatment phase of the trial, patients whose ECG met criteria for acute epicardial injury were randomized to prehospital or in-hospital treatment with alteplase. Before and at the end of this treatment phase of the trial, prehospital ECGs continued to be obtained in all patients suspected of having an AMI, irrespective of their clinical eligibility for thrombolytic therapy. These patients did not receive thrombolytic therapy before hospital admission. In addition to the prehospital ECG, all patients received a second ECG immediately on hospital arrival.
Two groups of patients, those randomized to prehospital versus in-hospital thrombolysis, and those who did not participate in the randomized trial were the focus of this ECG study. This trial was conducted with the approval of the Human Subjects’ Review Committee at the University of Washington.
Prehospital ECGs were obtained by a battery-powered, computer-integrated ECG system (Marquette Electronics). Prehospital and initial hospital ECGs were read independently by two experienced electrocardiographers who had no knowledge of the corresponding ECG, patient identity and outcome. Disagreements between readers regarding the presence or absence of acute epicardial injury were uncommon (<1% of tracings) and were resolved by consensus with a third electrocardiographer. ST segments were coded categorically as ≥1 mm, <1 mm or absent. ST segment elevation ≥1 mm in at least two contiguous ECG leads, in the absence of left bundle branch block (LBBB), ventricular pacing or an idioventricular rhythm, was defined as diagnostic of acute epicardial injury. In the presence of left ventricular hypertrophy, anterior ST segment elevation ≥4 mm in at least two contiguous leads was required for diagnosis of injury. Ischemic ST segment depression was defined as ≥1 mm horizontal or downsloping ST segment depression in two or more contiguous leads in the absence of LBBB or a ventricular rhythm. If accompanied by inferior ST segment elevation, ST segment depression in leads V2and V3was coded as indicative of probable posterior ST segment elevation. Pathologic Q waves were defined as an initial negative QRS deflection ≥40 ms in duration in at least two contiguous ECG leads in the absence of LBBB or a ventricular rhythm.
Prehospital and hospital records of all patients were abstracted for clinical characteristics and diagnostic outcome. Recognizing that the diagnosis of an acute coronary syndrome may be substantiated by a variety of means (characteristic symptoms, elevations in cardiac enzymes, cardiac catheterization or autopsy findings), the final hospital diagnosis was used to determine patients with acute myocardial ischemia or infarction (2).
Differences between continuous and categoric variables were evaluated by using the two-tailed Student ttest and the chi-square test (using the Yates correction as necessary), respectively. Sensitivity, specificity and predictive accuracy of a positive test were conventionally defined (3). The potential interaction between abnormalities observed on a single ECG and evolving abnormalities between serial ECGs on diagnosis was evaluated using a hierarchic log linear model. Statistical significance was defined as p ≤ 0.05 (two-tailed).
Between January 25, 1988 and February 28, 1993, 3,027 consecutive patients with symptoms of suspected AMI had prehospital electrocardiography performed by paramedics, and subsequently an in-hospital ECG. The clinical and ECG characteristics of these patients are shown in Table 1. Prehospital ECGs were obtained a mean of 91 ± 76 min after the onset of symptoms. In-hospital ECGs were obtained a mean of 24 ± 11 min after prehospital tracings. Either the prehospital or in-hospital ECG was technically uninterpretable in 10 patients (0.3%).
Of these patients in whom a prehospital and in-hospital ECG were obtained for evaluation of suspected AMI, two groups were identified for further study: those who had been randomized to receive prehospital versus in-hospital thrombolysis and those who were not participants in the randomized treatment trial. The clinical and ECG characteristics of these patients are also shown in Table 1.
Effect of prehospital thrombolysis on the ECG in treated patients
During the treatment phase of the MITI prehospital trial (November 2, 1988 to December 3, 1991), patients with suspected AMI, with no clinical contraindications to thrombolysis, and whose ECG met criteria for epicardial injury were eligible for randomization to prehospital or in-hospital treatment with alteplase. Results have been previously reported, based on intention to treat (1); they are presented here again based on actual treatment received. In this analysis, a total of 362 patients received prehospital (n = 173) or in-hospital thrombolytic therapy (n = 189) during the treatment phase of the trial. The clinical and ECG characteristics of these patients (based on actual treatment received) are described in Table 2. These patients differed in age and initial heart rate but were otherwise comparable in terms of gender, clinical and hemodynamic characteristics on presentation and incidence of AMI. Because of the study design, the group who received prehospital thrombolysis had an ∼15 min greater time interval between the prehospital and initial hospital ECG. In patients who had prehospital thrombolysis, the frequency of ST segment elevation on the prehospital ECG was insignificantly higher, whereas on the initial hospital ECG it tended to be lower than that in patients who had in-hospital thrombolysis (p = 0.07). Among all patients randomized to thrombolysis, the prevalence of ST segment elevation, ST segment depression or T wave inversion declined, whereas the prevalence of Q waves increased between the prehospital and initial hospital ECG (p < 0.005) (Table 1). These differences in the prevalence of ST segment and T or Q wave abnormalities between the prehospital and initial hospital ECG were more pronounced in those with prehospital versus in-hospital thrombolysis (Table 2), possibly owing to the effect of prehospital thrombolysis itself or to the different time intervals between ECGs in the two treatment groups.
Table 3describes the serialECG changes between the prehospital and initial hospital ECG in those with prehospital thrombolysis and those in whom thrombolysis was delayed until after hospital admission. No overall differences were noted in the evolution of ECG abnormalities in the groups, with two notable exceptions. First, ST segment elevation was more likely to resolve completely by hospital admission in those with prehospital thrombolysis: 14% versus 5% of those treated before versus after hospital admission, respectively (p = 0.004). Second, patients who had thrombolysis in the hospital and who had ST segment elevations <1 mm on the prehospital ECG were more likely to progress in severity to ≥1 mm by hospital admission. Nondiagnostic ST segment elevation progressed to diagnostic proportions between the prehospital and initial hospital ECG in no patient who received prehospital thrombolysis compared with 3% of those who received thrombolysis after hospital admission (p = 0.02).
ECG changes in patients not evaluated for prehospital thrombolysis
To avoid the confounding effect of thrombolysis on the ECG during the treatment phase of the MITI prehospital trial, serial changes in the prehospital and initial hospital ECG were evaluated strictly in the 2,665 patients with suspected AMI who were not participants in the randomized trial. The clinical characteristics and prevalence of ECG abnormalities in these patients are shown in Table 4.
Compared with patients whose chest pain was not determined to be related to ischemia or infarction, those with acute myocardial ischemia or infarction were older, more commonly male, more likely to have a previous cardiac history of infarction, angina or bypass surgery and had a higher incidence of ECG abnormalities suggestive of acute ischemia or infarction, but demonstrated a slower heart rate and lower blood pressure on presentation. The intervals between symptom onset and the prehospital and hospital ECG were similar between the two groups.
ST segment elevation or depression, T wave inversion and Q waves on both prehospital and hospital ECGs were more common in patients with acute coronary syndromes than in those without myocardial infarction or ischemia (Table 4). In patients with acute ischemia or infarction, Q waves were more prevalent on the initial hospital ECG than on the prehospital ECG (31% vs. 21%, p < 0.00001). Otherwise, the overall prevalence of ST segment elevation, ST segment depression, T wave inversion and LBBB did not differ on the prehospital versus initial hospital ECG in patients with and without documented acute myocardial ischemia or infarction (Table 4). In comparing prehospital with initial hospital ECGs, the identification of any abnormality in ST segment, T or Q wave or LBBB had a virtually identical sensitivity, specificity and positive predictive value for detecting acute myocardial ischemia or infarction—∼80%, 60% and 64%, respectively (Table 4).
Table 5describes the serialECG changes between the prehospital and initial hospital ECG in patients whose chest pain presentation was determined to be caused by acute myocardial ischemia or infarction as compared with those in whom it was not. The serial (dynamic) evolution of these changes between the prehospital and initial hospital ECG was an important discriminator between patients with and without acute coronary ischemia or infarction. In a total of 15% of patients with acute ischemia or infarction (as compared with 4% of those without an acute coronary syndrome), ST segment elevation evolved from or to diagnostic proportions between the prehospital and initial hospital ECG (p < 0.0000001). Representative changes are illustrated in Figures 1 to 3. ⇓⇓⇓In an additional 5% of patients with ischemia or infarction (as compared with 2% of those without), less pronounced but suggestive changes of injury were observed in ST segments between the two tracings (p < 0.0000001). That is, in addition to 26% of patients who had persistent ST segment elevation on both ECGs, an additional 20% of patients with subsequently documented myocardial ischemia or infarction could have been identified by evolutionary changes in ST segment elevation that transpired in the ∼30 min between their prehospital and initial hospital ECG evaluation. Compared with ST segment elevation on a single ECG (whether prehospital or initial hospital), persistent ordynamic changes in ST segment elevation between two serial ECGs improved the diagnostic sensitivity for acute myocardial ischemia or infarction from ∼34% to 46%, with only a modest reduction in specificity from 96% to 93% and in positive predictive value from 88% to 84% (all p < 0.00004) (Fig. 4).
Changing ST segment depression, T wave inversion and diagnostic Q waves between the prehospital and initial hospital ECG were more common than ST segment elevation in patients with acute coronary ischemia or infarction. Between the ∼30 min that transpired between the prehospital and initial hospital ECG, transient ST segment depression, transient T wave inversion and newly developed Q waves were observed in 21%, 23% and 14%, respectively, of patients with an acute coronary syndrome, as compared with 7%, 15% and 6%, respectively, of those without ischemia or infarction (p < 0.0000001). In contrast, transient or persistent changes in LBBB were infrequent and similar in patients with and without acute coronary ischemia and infarction.
In all “evolving” changes in ST segment elevation or depression, T wave inversion, developing Q waves or LBBB between the prehospital and initial hospital ECG were observed in 57% of patients with acute myocardial ischemia or infarction as compared with 32% of patients without an acute coronary syndrome (Table 5)(p = 0.0000001). Taken by themselves, these serial changes were less sensitive but more specific for detecting acute myocardial ischemia or infarction than ST segment, T or Q wave or LBBB abnormalities observed on a single ECG (sensitivity 57% vs. 80%, specificity 68% vs. 60%, respectively, p = 0.0000001). However, compared with a single abnormal ECG, the presence orserial evolution of ST segment, T or Q waves or LBBB between the prehospital and initial hospital ECG improved the diagnostic sensitivity of the ECG for an acute coronary syndrome from 80% to 87%, with a decline in specificity from 60% to 50% and a decline in positive predictive value from 64% to 60% (all p < 0.000006) (Fig. 4). In a hierarchic log linear model, inclusion of serial abnormalities between the prehospital and initial hospital ECG enhanced the ability to predict the presence of an acute coronary syndrome over isolated findings on either ECG alone (p < 0.05).
Electrocardiography has an established role in the evaluation of myocardial ischemia and is the most immediately available diagnostic modality for patients with suspected acute coronary syndromes. The relative ease of performing electrocardiography, together with development of portable, battery-operated electrocardiograph machines with land-based or cellular telephone transmission capabilities, and computer-facilitated interpretation of tracings have broadened the potential availability of this important tool to use outside the hospital or physician’s office (4,5).
Limited acceptance of prehospital electrocardiography
Prehospital electrocardiography has been demonstrated to be a feasible and reliable method of screening patients with suspected AMI, particularly when thrombolytic therapy is contemplated (2,5). At present, however, it is infrequently used in the evaluation of patients with suspected AMI (6). In part, this may be related to the uncertain need for prehospital electrocardiography when prehospital thrombolytic therapy is not under consideration. In contrast, because of the relatively small number of patients found to be eligible for thrombolytic therapy in the prehospital setting, the wide deployment of prehospital electrocardiography may not be seen as justifiable (7).
A second problem relates to whether the prehospital ECG provides useful diagnostic information beyond an ECG obtained on hospital arrival. The interval over which such diagnostic changes make their ECG appearance is not precisely known. Most information used to make the ECG diagnosis of acute myocardial ischemia and infarction has stemmed from hospital evaluations of patients who have oftentimes presented many hours after symptom onset. It could be argued that if obtained too early, before ECG changes of myocardial ischemia become manifest, interpretation of the prehospital ECG could be misleading and result in the inappropriate triage of patients away from emergency coronary interventions.
Even if abnormalities are detected on the prehospital ECG, a third problem relates to the specificity of such findings. ST segment of acute epicardial injury can be mimicked by pericarditis or by chronic changes associated with left ventricular hypertrophy or aneurysm, intraventricular conduction abnormalities and early repolarization. Similarly, ST segment depression, T wave inversion and even Q waves may accompany conditions that are not necessarily associated with acute myocardial ischemia or infarction. In the presence of such abnormalities, it could be argued that prehospital electrocardiography offers no advantage over an ECG obtained on hospital arrival, or might even result in the overdiagnosis of acute coronary syndromes.
The findings of this study suggest that ECG abnormalities are an early manifestation of acute coronary ischemia and can be readily identified by prehospital electrocardiography. Such changes were commonly observed early (mean ∼60 to 90 min) after symptom onset in patients with acute myocardial ischemia or infarction.
During the treatment phase of the MITI prehospital trial (1), early electrocardiography did not hamper diagnosis in patients eligible for thrombolysis. In those with prehospital thrombolysis, the prehospital ECG was, as expected, more likely to be abnormal than their initial hospital ECG. Even in patients who were not allocated to receive thrombolysis until after hospital admission, the prevalence of ST segment elevation on the prehospital ECG was higher than that on the hospital ECG obtained ∼20 to 30 min later (p = 0.03) (Table 2). On the basis of serial changes, the vast majority of patients who were eligible for thrombolysis before hospital admission by clinical and ECG criteria remained so on hospital admission (Table 3). An added benefit was observed in 14% of those with prehospital thrombolytic therapy in whom ST segment elevation resolved by hospital arrival, suggesting that coronary reperfusion was successful (8,9)and that additional invasive interventions to restore infarct-related artery patency were not likely to be necessary.
Prehospital electrocardiography was also of value in patients with suspected pain of myocardial ischemia or infarction who might have become candidates for thrombolysis. On average, prehospital ECGs were obtained 20 min later after symptom onset in patients who were not considered for randomization compared with patients who were randomized to receive prehospital or in-hospital thrombolysis during the treatment phase of the trial (p = 0.0001). Persistent, dynamic or suggestive ST segment elevation of epicardial injury was observed in a total of 46% of these patients with subsequently documented acute myocardial ischemia or infarction, as compared with 7% of those without documented ischemia or infarction (p < 0.0000001) (Table 5). Compared with ST segment elevation on a single ECG, serial data improved the diagnostic sensitivity for an acute coronary syndrome potentially treatable with thrombolytic agents from ∼34% to 46%, with only a modest decline in specificity from 96% to 93% and in positive predictive value from 88% to 84% (all p < 0.00004) (Fig. 4).
An attribute of this study was the opportunity to compare and contrast prehospital ECG findings with those of the initial hospital ECG. Unlike patients without evidence of myocardial ischemia or infarction, ECG abnormalities (ST segment elevation, ST segment depression, T wave inversion and abnormal Q waves) were not only more prevalent in patients with confirmed unstable angina or infarction (Table 4), but were also more frequently dynamic (Table 5). When considered together, either an abnormal ECG or serial changes between the prehospital and initial hospital ECG significantly enhanced the diagnostic sensitivity for an acute coronary event, as compared with abnormalities observed on a single ECG.
Serial ECGs obtained in the hospital might have demonstrated the same diagnostic utility as those obtained in this study. However, obtaining a prehospital ECG utilized transport time efficiently, allowing for serial ECG changes to occur during otherwise “down time” while en route to the emergency department. Time to treatment could have been advanced by ∼30 min or more with prehospital electrocardiography, as compared with awaiting serial ECG evaluation in the hospital.
Previous studies have determined that diagnostic-quality prehospital 12-lead ECGs can be obtained with great success by paramedics (10), can enhance the sensitivity and specificity of the clinical diagnosis of acute myocardial ischemia and infarction (11), can shorten time to treatment after hospital admission (12)and can even potentially lower in-hospital mortality (6). To our knowledge, this is the first study to systematically evaluate the diagnostic utility of serial changes between the prehospital and hospital ECGs in patients with and without acute coronary syndromes, regardless of their candidacy for early thrombolysis. Although only a minority of patients with acute coronary syndromes are currently eligible for thrombolytic therapy (7), many more may be eligible for other effective early interventions (such as heparin) (13). Accordingly, prehospital electrocardiography appears to have potential utility for the screening and treatment of all patients with suspected cardiac chest pain.
This was a retrospective study whose focus was on how serial prehospital and in-hospital electrocardiography in patients with suspected cardiac chest pain impacted the hospital diagnosis of ischemia or infarction. How the ECG ultimately impacted physician behavior, patient treatment, time to treatment or the economy of treatment were not evaluated. From the standpoint of these issues, wider use of the prehospital ECG merits further study. It is notable that findings from the National Registry of Myocardial Infarction have suggested that prehospital electrocardiography has the potential to favorably impact both more appropriate and more timely treatment in patients with AMI (6). The present study suggests that one mechanism by which prehospital electrocardiography may facilitate patient management is by improved diagnosis.
Interpretation of the ECG remains limited by the expertise of the reader (14). All ECGs in this study were interpreted by experienced electrocardiographers who may not be representative of other readers. In contrast, even among less experienced readers, changes between serial ECGs may be more easily identified than an isolated abnormality on a single tracing. Thus, it is possible that prehospital electrocardiography could even improve the accuracy of ECG interpretation on hospital admission by offering the opportunity to observe changes between tracings that might be more difficult to interpret individually (Fig. 1 to 3). However, some patients without documented acute ischemia or infarction manifested both persistent and transient ECG abnormalities. Hence, it is appropriate to be cautious about the absolute specificity of a single ECG or serial ECGs to discriminate chest pain that is of ischemic versus nonischemic origin. Indeed, in this study as well as others, increasing sensitivity for detecting acute ischemia or infarction resulted in some compromise of specificity (15,16).
Data from the National Registry of Myocardial Infarction suggest that patients who underwent prehospital electrocardiography had a longer delay to hospital admission (6). Such delays were not observed in the present study, except in patients who received prehospital thrombolysis. We observed that obtaining a prehospital ECG should add only a few minutes of time to the evaluation process (2). However, it is possible that the presence of abnormalities on the ECG could trigger other treatments before hospital admission, resulting in a longer delay to hospital admission. Thus, the deployment of any diagnostic or therapeutic maneuvers before hospital admission should be weighed against the potential harm of postponing hospital admission and the more definitive interventions available there.
ECG abnormalities are an early manifestion of acute coronary syndromes and could be readily identified by prehospital electrocardiography. Taken by themselves, serial changes between early prehospital and in-hospital ECGs were less sensitive but more specific for acute ischemia or infarction than isolated abnormalities on a single ECG. Taken together, the presence or serial evolution of abnormalities between the prehospital and initial hospital ECG significantly enhanced the diagnostic sensitivity for an acute coronary event, at the expense of a decline in specificity, as compared with abnormalities observed on a single ECG.
☆ This study was supported in part by Grant RO1HL38454 from the National Heart, Lung, and Blood Institute, National Institutes of Health, Bethesda, Maryland and by unrestricted grants from Genentech Inc., South San Francisco, California; Marquette Electronics, Milwaukee, Wisconsin; and Pharmacia Deltec, St. Paul, Minnesota.
- acute myocardial infarction
- electrocardiogram, electrocardiographic
- left bundle branch block
- Myocardial Infarction Triage and Intervention trial
- Received August 13, 1997.
- Revision received March 11, 1998.
- Accepted March 18, 1998.
- American College of Cardiology
- Peterson D.R.
- Kudenchuk P.J.,
- Ho M.T.,
- Weaver W.D.,
- et al.
- Canto J.G.,
- Rogers W.J.,
- Bowlby L.J.,
- French W.J.,
- Pearce D.J.,
- Weaver W.D.
- Brugemann J.,
- vanderMeer J.,
- deGraeff P.A.,
- Takens L.H.,
- Lie KI.
- Buszman P.,
- Szafranek A.,
- Kalarus Z.,
- Gasior M.
- Aufderheide T.P.,
- Hendley G.E.,
- Thakur R.K.,
- et al.
- Kereiakes D.J.,
- Gibler W.B.,
- Martin L.H.,
- Pieper K.S.,
- Anderson L.C.
- Th’eroux P.,
- Waters P.,
- Qiu S.,
- McCans J.,
- deGuise P.,
- Juneau M.
- McCrea W.A.,
- Saltissi S.