Author + information
- Received January 5, 1999
- Revision received June 8, 1999
- Accepted June 30, 1999
- Published online November 1, 1999.
- ↵*Reprint requests and correspondence: Tomas Jernberg, Department of Cardiology, Cardiothoracic Center, University Hospital, 751 85 Uppsala, Sweden
The purpose of this study was to evaluate the prognostic importance of ischemic episodes detected by ST-segment monitoring with continuous 12-lead electrocardiography (ECG) in a nonselected coronary care unit (CCU) population with chest pain and ECG nondiagnostic of acute myocardial infarction (AMI).
Patients with chest pain and ECG nondiagnostic of AMI constitute a heterogeneous group concerning both diagnosis and prognosis. Continuous 12-lead ECG is a rather new method not thoroughly studied in this population.
The ST-segment monitoring with continuous 12-lead ECG was performed for 12 h in 630 consecutive patients admitted to CCU due to chest pain and a nondiagnostic ECG, i.e., no ST-segment elevations. An ST-episode was defined as a transient ST-segment depression or elevation of at least 0.10 mV. The median follow-up time was six months.
A total of 176 ST-episodes occurred in 100 (15.9%) patients. The median duration and maximal ST-segment deviation in patients with ST-episodes were 80 min and 0.20 mV, respectively. Presence of ST-episodes predicted worse outcome concerning cardiac death and cardiac death or myocardial infarction (MI) (log-rank p < 0.001). At 30 day follow-up procedure, 10% versus 1.5% died from cardiac causes or had an MI in the group with and without ST-episodes, respectively. In a multivariate analysis, only troponin T ≥0.10 μg/l and the presence of ST-episodes came out as independent predictors of cardiac death or MI.
Continuous 12-lead ECG monitoring provides prognostic information on-line and considerably improves early risk stratification in patients with ECG nondiagnostic of AMI and symptoms suggestive of acute coronary syndrome.
Patients with chest pain suggestive of acute coronary syndrome and electrocardiogram (ECG) nondiagnostic of acute myocardial infarction (AMI) constitute a heterogeneous group concerning both diagnosis and future risk of cardiac events (1,2). Early risk stratification is essential for further treatment and management and is usually based on clinical characteristics, 12-lead ECG and biochemical markers (3).
In patients with unstable coronary artery disease (UCAD), i.e., unstable angina and non–Q-wave myocardial infarction (MI), presence of ischemic episodes during ST-segment recording with the Holter technique has in several studies been related to worse outcome (4–13). However, the Holter-technique has been limited to 2 to 3 leads and is an off line method, which means that the results have not been provided until several hours or days after the ischemic episodes. This has probably caused the limited use of Holter monitoring in clinical practice.
Recently, new technologies have been introduced, providing on line ST-segment monitoring with continuous 12-lead ECG or vectorcardiography (VCG) (14,15). Whereas several studies have presented results with VCG (16,17), very limited data are available concerning continuous 12-lead ECG in the population with UCAD (18,19).
The aims of this prospective study were to evaluate, in a nonselected coronary care unit (CCU) population admitted because of chest pain and with ECG nondiagnostic of AMI, the prognostic importance of ischemic episodes detected by ST-segment monitoring with continuous 12-lead ECG and compare it with well known prognostic factors such as clinical variables, ECG on admission and biochemical markers.
In the FAST study (Fast Assessment of Thoracic pain), patients admitted to the CCU in one university hospital were eligible for participation between March 1997 and February 1998. The inclusion criterion was a history of chest pain or other symptoms suggestive of an acute coronary syndrome. Exclusion criteria were prehospital thrombolysis, presence of ST-segment elevation (≥0.20 mV in V1-V4, ≥0.10 mV in other leads) on admission ECG, ECG not interpretable for ischemia (left bundle branch block [LBBB], pacemaker treatment, atrioventricular block or ventricular rhythm) or previous inclusion in this study. After a run in period of two months, consecutive patients were included.
Of the 1,194 eligible patients, 131 were excluded due to prehospital thrombolysis or ST-segment elevation on admission ECG, 170 because of previous enrollment in the study and 90 because of an ECG not interpretable for ischemia. Another 173 patients were excluded because of less than 9 h of monitoring, due to technical reasons or a need for fast turnover of CCU beds. Thus, 630 patients with at least 9 h of ST-segment monitoring during the first 12 h from admission were included in this analysis.
All clinical data were prospectively collected and entered into the local database of the National Registry for Cardiac Intensive Care. The treatment of individual patients was left to the discretion of the individual physician. Because this was an observational study and all investigations were done according to the local clinical routines, including the collection of clinical data in the local database of the National Registry for Cardiac Intensive Care, informed consent was considered unnecessary according to the local ethics committee who approved the study. However, all patients received written information and could refuse participation in this study and entering of any data in the database.
ECG on admission
A 12-lead ECG was obtained from the continuous 12-lead ECG system on admission to CCU and was interpreted centrally and without knowledge of the outcome. The ST-segment depression was defined as a depression of the ST-segment of ≥0.05 mV below the isoelectric line in any lead. T-wave inversion was defined as a negative or isoelectric T-wave in leads V2–V6, I and II, in aVL if R > 0.5 mV and in aVF if R > Q and R > S.
Blood samples for immediate analyses of creatine-kinase isoenzyme (CK-MB) (mass) and Troponin T at the central laboratory were obtained on admission, after 6 and 12 h, and, if needed, at later time points. Creatine-kinase isoenzyme MB (mass) was analyzed by the IMx method (Abbot Laboratories, Abbot Park, Illinois) with a decision limit for myocardial damage of 6 μg/l and for AMI of 10 μg/l. Troponin T was determined by the Enzymun-Test system (Boehringer-Mannheim, Mannheim, Germany) with a decision limit for myocardial damage of 0.10 μg/l.
The ST-segment monitoring was started with the admission ECG as the first acquired ECG and then performed for 12 h, using a continuous ST-segment monitoring software package, ST-Guard (GE Marquette Medical Systems, Milwaukee, Wisconsin). This software calculates each min from the last 10 s of monitoring median QRS-T complexes for each of the 12 leads. From these QRS-T complexes an ST-trend for each lead is constructed, stored and displayed online. The online ST-trends were displayed during the stay in the CCU and, hence, were known to the emergency physician.
Central analysis of ST-monitoring
The ST-trends were stored on floppy disks and later analyzed centrally by a well-trained physician (TJ) without knowledge of clinical outcome. Artifacts were removed before analysis. The ST-segment changes due to changes in body position were defined as sudden ST-segment changes with simultaneous changes of QRS- and T-wave amplitudes.
If the first acquired ECG had no signs of ischemia, this was chosen as the reference ECG. Otherwise, an ECG as close as possible to normal, at a time point as soon as possible after admission, was chosen. An ST-episode was defined as a transient ST-segment depression or elevation in any lead of at least 0.1 mV compared with the reference ECG, lasting for at least 1 min. Hence, patients with persistent ST-segment depressions were not classified as having ST-episodes. Episodes had to be separated by at least 1 min to be regarded as two separate episodes. In order to reduce the effect of heart rate, ST-segment changes were measured at J-point + (1/16 × R-R interval), corresponding to J-point + 60 ms at a heart rate of 62.5 beats per minute. The time from start of monitoring to start of the first ST-episode, the number of ST-episodes, the total duration of ST-episodes and the maximal degree of ST-depression or elevation compared with the reference ECG during an episode were recorded. The recurrence of chest pain was not monitored.
The median follow up time was six months (range 1 to 13 months). The end points were cardiac death and cardiac death or MI at 30 days and at the end of follow-up. All in hospital events were registered in the local database of the National Registry for Cardiac Intensive Care. Only new MIs occurring more than 24 h after admission were considered as new events. Out of hospital information about death and cause of death was obtained from the National Registry on Mortality. Information on MI was obtained from the hospital diagnosis registry and patients’ records. This information will contain the complete information on all admissions to the only hospital providing hospital care for acute diseases for the present patient population. For the diagnosis of acute MI, as well at the index as at recurrent events, one of the following should be fulfilled:
1. pathologic Q waves (duration >0.03 s and amplitude ≥25% of R) developing in at least two leads,
2. symptoms suggestive of acute MI or nondiagnostic ECG changes and typical elevated levels of biochemical markers with CK-MB (mass) ≥10 μg/l,
3. symptoms suggestive of AMI and ST-segment elevation or LBBB and further diagnostics not possible, or
4. signs of AMI at autopsy.
All data analysis was performed using the Statistical Package for Social Sciences (SPSS 8.0.0) software (SPSS Inc., Chicago, Illinois). Differences in proportions were judged by chi-square analysis using Yates’ correction and by Fisher exact test as appropriate. A significant difference was considered to exist with p < 0.05. The cumulative survival curves were constructed using the Kaplan-Meier method. The end points were cardiac death alone and cardiac death or MI. Statistical assessment was performed using the log-rank test with p < 0.05 considered as significant. To identify predictors of the double end point—cardiac death or MI—the univariate Cox regression analyses were used after all variables had been checked for proportional hazards. The following dichotomized variables were introduced in the analyses: presence or not of age ≥70, male gender, diabetes mellitus, hypertension, current smoking, history of previous MI, previous revascularization, peak-value of troponin T ≥0.10 μg/l, peak-value of CK-MB (mass) ≥6 μg/l, ST-segment depression ≥0.05 mV on admission ECG, ST-segment depression ≥0.10 mV on admission ECG, isolated T-wave inversion on admission ECG and ST-episodes during monitoring. The results were expressed as relative risk with a 95% confidence interval together with the p value from the log-rank test. All variables with a p < 0.05 were then tested in a multivariate Cox regression analysis using forward stepwise selection. Variables were entered if p < 0.05 and removed if p < 0.10.
The clinical characteristics and findings in the ECG on admission of the 630 patients are shown in Table 1. During the first 24 h from admission, 163 patients (26%) were classified as having an acute MI, 192 (30%) had unstable angina or angina, whereas 64 (10%) were considered to have other cardiac causes to their symptoms and 211 (33%) had noncardiac or unknown causes. Creatine-kinase isoenzyme MB (mass) ≥6 μg/l was present in at least one sample in 216 (34%) patients, whereas 193 (31%) had at least one sample of Troponin T ≥0.10 μg/l. During the follow-up period there were 32 (5.1%) deaths, of which 24 (3.8%) were cardiac deaths. Forty-four (7.0%) had an MI.
A total of 176 ischemic episodes occurred in 100 (15.9%) patients during ST-segment monitoring. Of these, 87 patients had only episodes of ST-segment depression, 5 patients had only episodes of ST-segment elevation and 8 patients had both depressions and elevations of the ST-segment. More than one episode was present in 36 patients. The episodes were located in the anterior or anterolateral leads in 62 (62%), inferior or inferolateral leads in nine (9%) and in both anterior and inferior leads in 29 (29%). The median duration of ischemia in patients with episodes was 80 min (25th to 75th percentile: 30 to 167), whereas the median value of the maximal ST-segment deviation of each patient was 0.20 mV (25th to 75th percentile: 0.15–0.29 mV). Sixty-one (61%) of the patients with ST-episodes had their first episodes at start of monitoring, whereas, of the remaining 39, 36 patients (92%), had their first episodes during the first 6 h from admission.
The number of patients with ST-episodes in relation to admission ECG and diagnosis are shown in Tables 2 and 3, respectively. ⇓⇓There was a significantly higher proportion of patients with ST-episodes in the group with ST-segment depressions ≥0.10 mV and the group with AMI on admission compared with the other groups (p < 0.001).
Predictors of outcome
In the univariate analysis (Table 4), the only clinical variables significantly predicting cardiac death or MI were age ≥70 years, diabetes mellitus and previous MI. Both Troponin T ≥0.10 μg/l and CK-MB (mass) ≥6 μg/l predicted worse outcome with a respective 6.9-fold and 5.9-fold increased risk of subsequent cardiac death or MI compared with those without this finding. Whereas ST-segment depression on admission ECG was a significant predictor, isolated T-wave inversions was not.
For the end points of cardiac death alone and cardiac death or MI, the presence of ST-episodes predicted a worse outcome (Fig. 1, log-rank p < 0.001). All patients were followed for 30 days. During this period there were 10 (10%) and 8 (1.5%) patients who died from cardiac causes or who had an MI in the group with and without ST-episodes, respectively. During the entire follow-up period, there were 24 (24%) versus 24 (4.5%) cardiac deaths or MIs.
In the multivariate analysis, including all characteristics showed to be significant predictors in the univariate analysis (Table 4), only Troponin T ≥0.10 μg/l (p < 0.001) and presence of ST-episodes (p = 0.004) were independent predictors of cardiac death or MI. Presence of ST-episodes was a better predictor of worse outcome than ST-segment depression ≥0.10 mV on admission-ECG. For patients with and without ST-segment depression ≥0.10 mV on admission, the occurrence of ST-episodes at monitoring added significantly to prognostic information (Fig. 2).
Patients admitted to CCU due to chest pain or other symptoms suggestive of an acute coronary syndrome and ECG nondiagnostic of AMI encompass a heterogeneous group that varies widely as well regarding diagnosis as future risk for cardiac events (20). There are several potential benefits with early risk stratification in this group. Patients identified as having a high risk of subsequent event might need a more intense pharmacologic treatment and should be considered for earlier interventions than patients with a low risk, in whom a more conservative approach might be appropriate. Considerable economic gains may be achieved by early identification of patients who are at sufficiently low risk for early transfer to lower level of care and early discharge.
Both 12-lead ECG and ST-recording with Holter technology have in previous studies been shown to contain important prognostic information in patients with a suspicion of acute coronary syndromes (4–13,21,22). However, both these methods have definite limitations. A 12-lead ECG at rest is only a “snapshot,” unable to reflect the dynamic nature of coronary thrombosis and myocardial ischemia. Because 70% to 90% of all ischemic episodes are silent in these situations (4–9)even repeated 12-lead ECGs in conjunction with chest pain will often be insufficient. Long-term 24 to 48 h Holter-recordings might be used for ST-analysis, but commonly only records 2–3 leads. In a previous study, 12-lead recordings had a better sensitivity for detection of ST-episodes compared with 3-lead recordings (19). However, the major disadvantage with Holter-recording is the off line analysis, which means that information obtained might concern events and risks occurring several hours or days earlier. Therefore, on line methods using either continuous 12-lead ECG or VCG monitoring have been developed (14,15). Whereas the continuous 12-lead ECG has been evaluated during thrombolysis in several studies (23,24), very little has been published concerning the population analyzed in this study.
This is thus far the largest study concerning ST-monitoring in patients admitted to CCU due to suspicion of an acute coronary syndrome with ECG nondiagnostic of AMI. The aim of this study was to evaluate the usefulness of continuous 12-lead ECG for early risk-stratification in a nonselected CCU population. Accordingly, consecutive patients admitted to CCU were included, excluding only patients with a definite diagnosis based on ST-segment elevation or the occurrence of LBBB.
In this study, 16% of the patients had ST-episodes during the first 12 h from admission. The incidence of ST-episodes in previous studies using Holter or VCG-monitoring in patients with unstable angina has varied widely (11% to 66%), probably mostly due to different inclusion criteria (4–13,16). Most of the patients with ST-episodes in this study had only episodes of ST-segment depression (87%), and only five patients had only episodes of ST-segment elevation. This is in accordance with some (6,25), but not all, previous studies (8,9).
Our results suggest that a 6 to 9 h period of monitoring in this kind of group might be sufficient to answer the question whether ST-episodes are present or not, because most (97%) patients with ST-episodes had their first episode started during the first 6 h from start of monitoring. This is in contrast with a previous study in which only about 50% had their first ST-episode during the first 12 h of monitoring (19). However, that study included a different population with an incidence of ST-episodes of 77%. Regarding ST-segment monitoring for other purposes, e.g., for assessment of the effect on ischemic episodes of a particular treatment, a longer time of monitoring might be necessary.
Predictors of outcome
In this study ST-segment monitoring with continuous 12-lead ECG was a significant predictor of both cardiac death alone and the double end point cardiac death or MI. Patients with ST-episodes had a 5.7-fold risk of death or MI compared with patients without ST-episodes. These results are in accordance with a recently published Holter-study by Patel et al. (13)that included a fairly unselected group with unstable angina, although it excluded patients with acute MI. In that study, 15% had ST-episodes during the 48 h of ST-monitoring, with a subsequent 4.1-fold increased risk of death or MI. However, none of the previous Holter-studies has shown the presence of ST-episodes to be a significant predictor of death or cardiac death alone.
In another study by Andersen et al. (16), including a population similar to the one in this study, 29% had ischemic episodes detected as ST-vector magnitude (ST-VM) episodes during 24 h of VCG monitoring. In that study 23% and 10% died or had an MI during one-year follow-up in the group with and without ST-VM episodes, respectively. The larger proportion of patients with ischemic episodes in that study was probably due to differences in the patient population, e.g., a smaller proportion of noncardiac or unknown causes (20%). However, part of the difference might also be related to discrepancies between the two monitoring techniques, e.g., an ST-episode of 0.10 mV during 12-lead monitoring does not necessarily correspond to an ST-VM episode of 0.05 mV during VCG monitoring.
In the univariate analysis, age, diabetes mellitus and previous MI were, as expected, associated with increased risk of cardiac death or MI. Previous Holter studies have shown that the presence of ST-episodes independently predicts future risk of cardiac events in multivariate analyses (5,8,11,13). However, these studies were performed without information from the new and sensitive biochemical markers [CK-MB (mass) and troponin T] which are important prognostic markers in this population (2,20). In this study, the occurrence of ST-episodes persisted as an independent predictor of cardiac death or MI even when these new biochemical markers were included in the multivariate analysis. These findings are in accordance with the VCG study (17), where the occurrence of ST-VM episodes during 24 h of monitoring was an independent predictor of death and death or MI in a model including clinical variables and CK-MB (mass).
A limitation with continuous ST-segment monitoring is the relatively low sensitivity to identify all patients at risk of cardiac death or MI (56% at 30 days in this study), which seems to be regardless of the method used (13,16). This stresses that ST-segment monitoring should be used in addition to the clinical evaluation and biochemical markers in the early risk assessment.
There are some limitations to this study. First, recurrence of chest pain was not registered. However, several previous studies have convincingly demonstrated ST-episodes to be a better prognostic marker than recurrence of chest pain (5,8,11,17). Second, although the monitoring-system enables on-line analysis, all the ST-trends were analyzed centrally off line. Therefore, a new study is needed for evaluation of ST-monitoring in clinical practice, where the individual physicians base their decision on on line analyses. Third, the individual physician had free access to the ST-trends online. Thus, the observation of ST-episodes online might to some extent have influenced decisions. However, this would be likely to result in an under- rather than overestimation of the prognostic value of ST-episodes detected by continuous 12-lead ECG.
Continuous 12-lead ECG monitoring during the first 6 to 9 h after admission to CCU provides prognostic information on line and, thereby, the early risk stratification in patients with ECG nondiagnostic of AMI and symptoms suggestive of acute coronary syndromes might be improved. Therefore continuous online multilead ST-monitoring is a useful part of the clinical routine, for the early identification of higher risk patients who need more intensive treatment than those with lower risk, who may benefit from a more conservative approach with rapid transfer to intermediate care and for early discharge.
We wish to thank the clinical staff at the CCU in Uppsala and especially the research nurses Matts Högberg, Anna Isaksson, Gerd Ålsjö and Catrin Henriksson for their skillful and invaluable assistance.
☆ This study was supported by grants from the Swedish Heart and Lung Foundation and the Uppsala County Association Against Heart and Lung Diseases, Sweden and GE Marquette Medical Systems, Milwaukee, Wisconsin.
- acute myocardial infarction
- coronary care unit
- creatine kinase isoenzyme
- left bundle branch block
- myocardial infarction
- unstable coronary artery disease
- Received January 5, 1999.
- Revision received June 8, 1999.
- Accepted June 30, 1999.
- American College of Cardiology
- Braunwald E,
- Jones R.H,
- Mark D.B,
- Brown J,
- et al.
- Johnson S.M,
- Mauritson D.R,
- Winniford M.D,
- et al.
- Nademanee K,
- Intarachot V,
- Josephson M.A,
- Rieders D,
- Vaghaiwalla Mody F,
- Singh B.N
- Arnim T.V,
- Krawietz G.W,
- Höfling B
- Langer A,
- Freeman M.R,
- Armstrong P.W
- Larsson H,
- Jonasson T,
- Ringqvist I,
- Fellenius C,
- Wallentin L
- Patel D.J,
- Holdright D.R,
- Knight C.J,
- et al.
- Patel D.J,
- Knight C.J,
- Holdright D.R,
- et al.
- Krucoff M.W,
- Wagner N.B,
- Pope J.E,
- et al.
- Andersen K,
- Eriksson P,
- Dellborg M
- ROMIO Study Group,
- Gomez M.A,
- Anderson J.L,
- Karagounis L.A,
- Muhlestein J.B,
- Mooers F.B
- Klootwijk P,
- Meij S,
- Es G.A,
- et al.
- Ravkilde J,
- Nissen H,
- Hörder M,
- Thygesen K
- GUSTO-I ECG-ischemia monitoring substudy,
- Klootwijk P,
- Langer A,
- Meij S,
- et al.
- TAMI 7 study group,
- Krucoff M.W,
- Croll M.A,
- Pope J.E,
- et al.
- Patel D.J,
- Knight C.J,
- Holdright D.R,
- et al.