Vectorcardiographic changes as predictors of cardiac complications during major vascular surgery

Vectorcardiographic Changes as Predictors of Cardiac Complications During Major Vascular Surgery Per Gannedahl, MD, PhD, Magnus Edner, MD, PhD, and Olle Ljungqvist, MD, PhD

Objective: To elucidate the relation of changes in computerized vectorcardiographic trend parameters indicating perioperative myocardial ischemia with perioperative cardiac complications. Design: Prospective clinical study. Setting: A single university hospital. Participants: Thirty-eight patients undergoing elective abdominal aortic surgery. Interventions: Computerized vectorcardiography recorded during surgery and for 48 hours postoperatively. Measurements and Main Results: Vectorcardiographic spatial alterations in the QRS complex (QRS-VD) and absolute (ST-VM) and spatial (STC-VM) ST-segment changes, previously used indicators of myocardial ischemia, were analyzed and related to the cardiac events detected clinically. In five patients with clearly ischemic (cardiac death, myocardial infarction, recurrent ischemia) and eight patients with possibly ischemic (congestive heart failure, arrhythmia) perioperative cardiac events, ST-VM and STC-VM were significantly increased intraoperatively. Postoperatively, these differences remained, but QRS-VD were also significantly increased. Intraoperative and postoperative changes

indicating ischemia were strongly related (r = 0.83). The signs of ischemia were most pronounced during the postoperative 12 to 36 hours. The presence of 60 minutes of signs of ischemia during 2 hours revealed high sensitivity (85%), specificity (80%), and positive (69%) and negative (91%) predictive values for subsequent cardiac events. Traditional vector loop analysis showed signs of non-Qwave infarctions in six patients, whereas only three of these were detected using standard clinical methods. Conclusions: Vectorcardiographic signs of myocardial ischemia were significantly increased intraoperatively, but most pronounced postoperatively in the patients subsequently suffering cardiac events. The changes could be related to the individual cardiac morbidity with acceptable precision. Thus, continuous vectorcardiographic monitoring may be beneficial for patients at risk of developing perioperative ischemia. Copyright © 1998 by W.B. Saunders Company

HE RELATION between perioperative myocardial ischemia and postoperative cardiac morbidity is still controversial? ,z as is the best method of cardiac monitoring in at-risk patients. Smith et aP showed that new wall motion abnormalities as detected by transesophageal echocardiography (TEE) were more frequent than electrocardiographic ST deviations intraoperatively. These results were confirmed by others. 4,5 However, Eisenberg et al 6 challenged these findings by showing no difference in the incidence of intraoperative signs indicating ischemia between TEE and 12-lead electrocardiogram (ECG). Clinical studies have shown that postoperative ischemia, rather than intraoperative, is predictive for postoperative cardiac morbidity. 7-9 This implies that adequate on-line ischemia monitoring in the postoperative period is a prerequisite, if interventions are possible, to decrease postoperative cardiac morbidity. For long-time postoperative monitoring, electrocardiographic methods seem to be the most feasible method today, when considering the cost and practical aspects. Computerized vectorcardiography (VCG), (MIDA, Ortivus

Medical, T~iby, Sweden) was recently introduced for perioperative ischemia monitoring. 1°,11 In brief, VCG is a personal computer system for continuous registration and spatial on-line analysis of ST and QRS changes. VCG has been evaluated by cardiologists in patients with acute myocardial infarction (AMI), 12 unstable angina, 13,14 and during coronary angioplasty. 15A6 Recently, it was reported that VCG improved the ability to predict postoperative cardiac events in major vascular surgery, as compared with the use of scalar leads. 11 However, the temporal development of the different VCG parameters indicating myocardial ischemia was not shown. Presently, detailed analysis of the VCG recordings is reported to further improve the predictive value of the method and to elucidate the temporal relation between the development of ischemic signs on the VCG and the subsequent cardiac morbidity. These studies revealed that it is possible to detect patterns of VCG changes that precede subsequent cardiac events on an individual basis with a high degree of accuracy.

T

KEY WORDS: anesthesia, electrocardiography, intraoperative monitoring, myocardial ischemia, vascular surgery, vectorcardiography

MATERIALS AND METHODS From the Departments of Anaesthesiology and Intensive Care, Medicine, and Surgery, Karolinska Institute at Karolinska and Danderyd Hospitals, Stockholm, Sweden. This study was supported by the Karolinska Institute, the Swedish Society of Medicine, the Swedish Medical Research Council, Grant No. 09101, Claes Groschinsky's Memorial Fund, the SALUS Fund for Medical Research, the "FOrenade Liv" Mutual Group Life Insurance Company, and the Swedish Heart and Lung Foundation. Presented in part at the American Society of Anesthesiologists meeting in Atlanta, GA, October 21-25, 1995. Address reprint requests to Per Gannedahl, MD, PhD, Department of Anaesthesiology and Intensive Care, Karolinska Hospital, S-171 76 Stockholm, Sweden. Copyright © 1998 by W.B. Saunders Company 1053-0770/98/1201-0007503.00/0

38

The study was approved by the local Ethics Committee and, after informed consent, 38 patients, including 6 women, scheduled for abdominal aortic surgery, were included. Patients with ECGs showing a bundle-branch block were excluded. The demographic and perieperative data are shown in Table 1. The preoperative assessment included the measurement of left ventricular ejection fraction at rest by the use of a scintigraphic method applying a nuclear stethoscope. The anesthetic regimen was standardized to intravenous (IV) fentanyl and isoflurane in O2/N20. The standard perioperative monitoring was uniform for all patients. The details of the perioperative care were described previously. ~1 There was no specific protocol for the detection of cardiac events, but all cardiac complications found by the clinicians in charge of the patients were registered. Cardiac death was defined as death because of circulatory insufficiency caused by a primary cardiac cause. The diagnosis of an AMI was

Journal ofCardiothoracic and VascularAnesthesia, Vo112,No 1 (February),1998:pp 38-44

PERIOPERATIVE VECTORCARDIOGRAPHY

39

Table 1. Demographic and Perioperative Data for 38 Patients Undergoing Elective Abdominal Aortic Surgery

Age (yr) Length (cm) Weight (kg) Hypertension Angina pectoris Previous myocardial infarction Coronary artery disease Cardiac disease and/or hypertension Cardiovascular medication Diabetes mellitus Left ventricular ejection fraction (LVEF) LVEF <0,45 Aortic occlusive disease Abdominal aortic aneurysm Hospitalization (days)

68,7 _+ 8.0 173.9 +_ 5.5 69.9 _+ 12.3 9/3,8 10/38 8/38 15/,28 23/38 19/,28 2/38 0.53 -+ 0.10 9/38 10/38 28/38 16 + 11

NOTE. Mean _+SD or number of patients. Reprinted by permission of the Journal of the American College of Surgeons. 11

determined when there was a rise in creatine kinase-MB isoenzyme of >0.2 pkatallL (catalyzing activity per second) (equivalent to 12 U/L), being > 3 % of the total creatine kinase level, combined with the development of a new Q wave or new persistent T inversions. Recurrent myocardial ischemia was considered as transitory ischemic ECG changes with or without anginal chest pain, but without cardiac enzyme release. An arrhythmia was judged to be a cardiac event if the arrhythmia caused hemodynamic deviations and if the clinician in charge decided on a new treatment with antiarrhythmic therapy, either by electroconversion or pharmacologic IV agents. Finally, congestive heart failure (CHF) was defined as a clinical situation when treatment with diuretics and/or inotropic support was determined to be required by the clinician in charge on chest radiograph examination and/or pulmonary artery catheter measurements in combination with clinical findings (tales, stasis of the jugular veins, $3 rhythm and/or dyspnea determined to be of cardiac origin) or clinical pulmonary edema. The VCG system used has been described previously in detail. 1°-16 Briefly, VCG recordings of myocardial spatial electrical activity, applying the Frank leads, 17 are continuously sampled and stored in a personal computer. The sampling rate is 500 Hz, the sensitivity is 1 pV, and the bandwidth is 0.05 to 200 Hz. A software program examines the VCG signals and displays an on-line analysis. By collecting and averaging the signals from the first heart beats, a baseline complex is created. Forthcoming heart beats are averaged during a chosen period of 10 to 240 seconds, and compared with the baseline complex. Extrasystoles are ignored. Differences in both the QRS and the ST segments are displayed in real-time as trend parameters (Fig 1). Of these, the QRS vector difference (QRS-VD) displays the spatial changes in the QRS complex. The ST-vector magnitude (ST-VM) shows the ST deviation in the three orthogonal leads and the ST-change vector magnitude (STC-VM) reflects the spatial ST changes, by calculating the difference between the present and the initial ST vector. In the present study, VCG was recorded during surgery and for 48 hours postoperatively. The baseline complex was collected in the operating room before the induction of anesthesia. The average sampling time was 20 seconds intraoperatively, but changed to 2 minutes postoperatively because of computer memory limitations. The VCG recordings were not available to the clinicians in charge of the patients. Thus, all clinical decisions were made without knowledge of the results of the VCG. The VCG recordings were analyzed for changes indicative of myocardial ischemia in QRS-VD, ST-VM, or STC-VM. Based on the findings in studies on patients with ischemic heart disease, 12,15As

episodes of ischemia were defined as an increase in QRS-VD > 15 pVs, or an elevation of the ST-VM or STC-VM of >100 ~tV, for more than 1 minute intraoperatively or at least 2 minutes postoperatively. All VCG analysis was performed retrospectively without knowledge of the patient's clinical course. The number and duration of all episodes of ischemia were registered. The ischemic intensity, defined as the area under the curve for the ischemic episodes, was also calculated. The mean area of ischernic episodes per recorded hour and the mean area per minute of ischemic episode were also determined. The correlation between the intraoperative and postoperative VCG duration of myocardial ischemia expressed as the percentage of the recorded time for each of the two periods was also calculated. The ischemic duration per hour was calculated for each patient, and the individual accumulated ischemic duration per hour was plotted. The plots were examined regarding typical patterns indicating forthcoming cardiac events. The ischemic duration per hour was also analyzed for differences between those stfffering from cardiac events and those who had an uneventful postoperative course. Adjustment to the correct hour of the day was additionally done to elucidate any diurnal variations in the occurrence of ischemia.

QRS-Vector Difference

~Ax

Y

~Ay

Z ~_Az

X

QRS-VD=V Ax2+Ay2+Az2 ST-Vector Maqnitude

y

z/ sT

ST-VM = V

STx2 + STye + STz2

STC-Vector Maqnitude Reference ST-vector

D i f f e r e n c e vector, D

Present ST-vector

S T C -VM = V

Dx2 + D y 2 + Dz2

Fig 1. Vectorcardiographic trend parameters: The QRS-VD is the area of difference (A) between the initial reference complex and the present complex. The ST-VM is the deflection of the ST segment 20 to 80 ms after the J-point (60 ms in the present study). The STC-VM is the spatial difference vector between the ST vector in the reference complex and the ST vector of the measured complex,

40

GANNEDAHL, EDNER,AND LJUNGQVIST

The first and last recorded vector loops were analyzed in a blinded fashion for signs of an acute perioperative myocardial infarction by the criteria: (1) >180 ° angle between the maximum T and QRS vectors in one lead, (2) >0.3 mV ST vector in the same direction in more than 2 leads, and (3) a localized defect in the QRS loop rotating the opposite way in more than 2 leads with duration >10 ms and amplitude >0.15 mV.~9 The mean _+ SD are given for the demographic data. For the comparison of VCG ischemia parameters between unpaired groups, the Mann-Whimey test was used because the VCG data did not appear as normally distributed. The Wilcoxon rank test was used for comparison of paired groups. However, to illustrate the difference between the groups with or without events, the mean and not the median is shown in Fig 2. If ties were present in the nonparametric analysis, tied p values were used. Pearson's linear regression analysis was used for testing correlation; p <0.05 was considered significant.

Thirteen patients had one or more postoperative cardiac events (Table 2). The cardiac complications were clinically evident during the first 48 postoperative hours in eleven patients, and on the following postoperative day for the two remaining patients. Apart from some minor recording problems because of sweating and loss of skin contact, the 48-hour postoperative recording time was achieved in all but six patients. One patient died of cardiac failure < 2 hours postoperatively. The other five recordings were interrupted after 31 and 38 hours (discomfort from the electrodes), 22 hours (cable breakage), and 36 and 28 hours, respectively (reoperation because of bleeding and compartment syndrome in the calf). Thirty patients had significant ischemic changes in at least one of the VCG trend parameters, ranging between 10.2% to 99.9% of the recorded perioperative time. In addition, four of the remaining eight patients had changes postoperatively in QRS-VD during 4, 4, 8, and 34 minutes, respectively, corresponding to 0.2%, 0.1%, 0.3%, and 1.1% of the perioperative recording time. The comparison of the vectorcardiographic trend parameters for the groups with and without cardiac events is summarized in Table 3. During surgery, all the variables related to frequency and duration of ST-VM and STC-VM episodes, as well as

ischemic intensity, were significantly greater for the group that subsequently developed cardiac events. In the postoperative period, the ischemic intensity of QRS-VD and ST-VM was significantly greater, as were all the STC-VM variables in the patients with cardiac events. These differences became even more apparent for the ST-VM area parameters and all the STC-VM variables when the entire perioperative recording period was used for comparison between the two groups. Five subjects among the 13 patients with cardiac events had complications that were truly ischemic (acute myocardial infarction, cardiac death, and recurrent ischemia). The remaining eight patients had cardiac events that were not necessarily related to ischemia (arrhythmia and CHF). However, there were no statistically significant differences in the VCG trend parameters between these two subgroups. Regression analysis between the percentage of ischemic time intraoperatively and postoperatively revealed intraoperative ischemia, as detected by at least one of the three trend parameters, to be strongly related to postoperative ischemia (r = 0.83, p <0.0001). This relation was also present when the cardiac event group was analyzed separately (r = 0.84, p <0.005). The duration of ischemia per hour in QRS-VD (Fig 2A) and STC-VM (Fig 2B) was significantly greater in the group with cardiac events during the end of the first 24 hours and ongoing for more than 24 hours. However, in ST-VM, only 2 hours showed significant differences between the groups. The adjustment to the exact starting hour of the recordings did not reveal any diurnal variations. The patients with cardiac events showed a longer duration of ischemia than the group with an uneventful course early in the morning of the first day after surgery and ongoing for more than 24 hours until the morning of the second day after surgery. Plots of accumulated ischemia duration per hour for four patients are shown in Fig 3. This analysis indicated that a rapid increase in accumulated ischemic time could discriminate those at high risk. By further examination of the individual curves of ischemic time, ie, the accumulated minutes of ischemia/hour, it was found that the presence of 2 consecutive hours -->60 minutes of signs of ischemia for the combination of QRS-VD

A

B

RESULTS

I'ntra°Pl

PoD,

I

Poo2

I

40 ~

I

t-i'r°°P I

~oo,

I

Poo2

I

I

min/hr

min/hr 20-

30-

152010105-

0

i

J

i

10

20

30

b~<~.

i

i

40

50

hours from start of registration

6'0

00

10

20

30

40

50

6'0

hours from start of registration

Fig 2. The mean duration of ischemia per hour for QRS-VD (A) and STC-VM (B) in the groups with or without cardiac events, Abbreviations: Min/hr, minutes of ischemia per hour; Intraop, intraoperative period (the mean for all patients); POD, postoperative day; QRS-VD, QRS vector difference; STC-VM, ST change vector magnitude. * indicates p <0.05. [] = cardiac event, ~ = no cardiac event.

41

PERIOPERATIVE VECTORCARDIOGRAPHY

Table 2. Postoperative Cardiac Events and Time of Fulfillment of Developed Criteria for Critical Ischemia Measured From the Start of the VCG Recordings Fulfillment of Criteria for Critical Ischemia Cardiac Event

Cardiac pump failure, cardiac death Acute Q-wave infarction + pulmonary edema Acute non-Q-wave infarction Acute non-Q-wave infarction + ventricular tachycardia Recurrent myocardial ischemia Congestive heart failure + atrial fibrillation Congestive heart failure + atrial fibrillation Congestive heart failure + atrial fibrillation Atrial fibrillation/atrial flutter 4- pulmonary edema Atrial fibrillation Atrial fibrillation Congestive heart failure Congestive heart failure

Major VCG Trend Parameter

QRS-VD ST-VM QRS-VD ST-VM ST-VM QRS-VD STC-VM QRS-VD QRS-VD QRS-VD ST-VM QRS-VD QRS-VD

3qme of Event From Start of Recording ( h r )

QRS-VD + ST-VM/STC-VM( h r )

STWM/STC-VM(hr)

4 46 POD 3 48 POD 1 29 51 33 POD 1 20 24 POD 3 POD 2

5 29 41 27 3 26 9 26 3 32 37 ---

5 29 41 27 2 2 9 5 2 3 20 ---

NOTE. In five patients only the POD was possible to determine for the event. - - indicates the two patients with cardiac events who never fulfilled the predictive criteria for critical VCG changes. Abbreviations: VCG, vectorcardiography; QRS-VD, QRS vector difference; ST-VM, ST vector magnitude; STC-VM, ST change vector magnitude; POD, postoperative day.

and ST-VM or STC-VM. The combination of SI'-VM or STC-VM showed equal sensitivity (84.6%) and similar values in specificity (80% to 76%), positive (68.8% to 64.7%), and negative predictive values (90.9% to 90.5%) for cardiac events. 1 If the criterion of 60 minutes or more of ischemia during 2 consecutive hours was applied to either of the two combinations of QRS-VD and ST-VM/STC-VM, or ST-VM/STC-VM, 11 of 13 patients who subsequently developed cardiac events would have been detected to be at risk. The two patients who would not have been discovered using these criteria suffered from CHF on postoperative days 2 and 3, respectively. When the exact times for the fulfillment of either of the two criteria were related to the approximate times when the cardiac events were clinically apparent, 8 and 10 cardiac complications, respectively, would have been detected by the VCG ahead of the complication (Table 2). Yet another patient developing cardiogenic shock during surgery and dying postoperatively showed signs of severe ischemia on the VCG before the event was noted clinically. However, cardiac failure developed so rapidly that he never fulfilled the criteria until the event was clinically evident. The analysis of the first and last recorded vector loops revealed that six patients fulfilled the vector loop criteria for a new non-Q-wave infarction. Five of these were in :he group with cardiac events found clinically, but only three patients were discharged with a clinical diagnosis of a new rr.yocardial infarction. The remaining two patients suffered arrhy~hmia and arrhythmia/CHF, respectively, during their in-hospital stay. The sixth patient had an uneventful course without clinical signs of a cardiac event. DISCUSSION

This is the first study to analyze VCG changes in delail for the detection of perioperative ischemia developing into cardiac events. The perioperative use of VCG may identify the individual patients likely to suffer from cardiac complications. In most cases, the VCG could identify the risk well ahead of the events becoming clinically apparent. Because the VCG was not

available to the clinician in charge, the question of whether intervention based on signs of myocardial ischemia is beneficial cannot be answered. However, the present data form the basis for such interventional studies applying the suggested VCG criteria for critical myocardial ischemia. These patients were recruited from the waiting list for elective surgery. The number of cardiac events was high, but is consistent with other reports, 2°-23 in particular for the incidence of AMI. The total cardiac morbidity in these studies ranges between 9% to 28%. However, it is a difficult task to compare the studies because of the use of different criteria, detection systems, and patient groups. 24 CHF and arrhythmias, including atrial fibrillation/flutter, were included, defined as when the clinician in charge decided that an intervention (pharmacologic treatment or electrical cardioversion) was necessary, because they can be regarded as clinically relevant. Perioperative AMI and angina pectoris are considered to be truly ischemic cardiac complications. Besides increased oxygen supply, these complications can have other etiologies such as postoperative hypercoagulability, endothelial factors, coronary spasm, or leukocyte activation. 24 The cardiac events arrhythmia and CHF may have different causes, also including etiologies related to the development of myocardial ischemia. 25 Even if the origin might have been nonischemic, the event may still have caused subsequent ischemia. Thus, it is not possible to exclude an ischemic etiology for any of these events, and they were subsequently included. In support of this, although the number of patients was small, no difference was found in VCG changes between the five patients with truly ischemic events and the other eight patients with arrhythmias or CHE Besides, all patients with subsequent cardiac events had VCG signs indicating ischemia, before or concomitantly with the event. Although no pathophysiologic relationship to the present clinical cardiac events can be revealed from these data, there seems to be a temporal relationship between signs of ischemia and the cardiac events, as shown in previous studies. 26,27

42

GANNEDAHL, EDNER, AND LJUNGQVIST

~

"

~ o ~

~

~ o o o ~ o

0 6 0 0 0 6 6 o o o o o o o

O o O o o o o o o o o o

$ 9 ~ 6 f f ~ o £ o o o £ £

~ 9 9 9 9 9 ~ 9 ~ ~ o £ o o o £ £ £

o

o o o o

o

..8 0

#.

0

._~ ®

~ A ~

~

._e ~

'

,

~

~,

,

~

.

._e

.o

g

A ~ A A ~ 6 6 6 6 6 9

e-

£~

6 6 6 6 6 6 6

6 6 6 6 6 6 6 6 6

~ 0

0

0

0

0

0

0

0

0

0

0

0

0

0

0

0

E 0

0

"o ,+-

._~

,_o

o

~)

.E

A .--

o o o o o ~ o o o ~ ~ o ~

.

.

.

o ~ o o o ~ o ~ ~ .

~ o ~ o o o o o o

.

,

l

l

l

l

l

l

l

l

o o o

.,-- ~.

,

*4 A

.E •"o

"E ~

~

~

..

•~_ ~.~ a=~o.9.

-~-~

~

-~

~~-

'~.~

~-~

~ ~ ~

~

oo

o

:

._

~

£

"~

:o

o~,.~.

~

~ a ,-,- ~

.~.

e)

== I,

_

>co

~oc0

~:._o aoa

F-

~.~

~•

z,

o d

o d

o d

II ~

+-~<~

PERIOPERATIVE VECTORCARDIOGRAPHY

43

Minutes of ischemia 17509 ¢. 6

1500-

1250-

e

1000-

.0

750 -

e,o 6

500-

O--lllllll,l,llllllllil~llll,lll ,...... 0 10 20

.0 ¢

I..,..,.,.,]

~i 30

I

I

I

40

50

60

Hours from start of recording Fig 3. Plots of accumulated duration of VCG signs of ischemia/ hour in four patients. ~, []. © indicate patients experiencing postoperative cardiac events. The events were detected clinically, approximately 46, 24, and 51 hours from the start of the recordings, respectively.

Four patients showed minor changes in QRS-VD lasting 0.1% to 1.1% of the recording time. The minor clzanges in QRS-VD for these four patients were considered to be most likely of nonischemic origin. The QRS-VD cannot dis(;riminate among changes in the loop configuration because of ~n altered spatial level of the vector loop per se, ie, caused by variations in body position,1°,z8 or actual changes in loop configuration produced by pathologic processes. Changes in QRS-VD may also be induced by alterations in left ventricular blood volumes29 and hematocrits. 3° This is also indicated in Table 3, where QRS-VD is the least specific of the three trenzl parameters in detecting cardiac events. Nevertheless, QRS-VD was significantly increased during several hours in the patients with subsequent cardiac morbidity (Fig 2A). Studies in coronary care units have also shown that QRS-VD changes may be present without concomitant ST changes in patients with unstable angina31 or during coronary angioplasty. ]5 Thus, the QRS-VD may be less specific, but still seems valuable for the overall interpretation during ischemia monitoring. The relation between perioperative ischemia and postoperative cardiac morbidity has been previously studied using Holter ECG. 7-9,27,32Some authors have found that only postoperative ischemia is associated with an increased risk ot cardiac complications.7-9 However, the present results (Table 3) chal-

lenge this view because signs of ischemia were already significantly increased intraoperatively in the group with subsequent cardiac events. One shortcoming of the largest of the previous studies 9 was that postoperative cardiac events occurring up to 49 days after surgery were included, which raises the question of the causal relationship between the ischemia and the cardiac events included)3 The greatest differences were noted in the STC-VM variables, indicating that this parameter may be the most important to monitor. This is in line with findings in patients with unstable angina pectoris. 31 Table 3 also indicates that the ischemic intensity, measured as the area under the trend parameter curve, may become an important variable as a predictor of postoperative cardiac morbidity. The regression analysis showed that the presence of intraeperative ischemia was related to postoperative ischemia. Thus, signs of VCG ischemia intraoperatively should be considered as serious warnings of potential postoperative morbidity, and patients with such signs should subsequently be monitored extensively postoperatively for ischemia. The incidence of VCG signs indicating ischemia was high, in particular for the QRS-VD. A comparison of the present findings with other reports is difficult, but other studies have found up to a 74% incidence intraoperative ischemia. 34 However, the gold standard for monitoring myocardial ischemia perioperatively is highly controversial. Regardless of which method is used for monitoring, the aim is always to detect changes that could signal an upcoming cardiac complication and initiate various types of interventions. The calculated duration of ischemia per hour (Fig 2) may be criticized as arbitrary, but it still shows the temporal development of ischemic differences between groups. Ischemia was predominantly present in the cardiac event group toward the end of the first postoperative day, and lasted more than half of the second postoperative 24-hour period. When adjusted to real-time, the differences were greatest during the first day after surgery. This finding is important because it is common to move these patients from the intensive care unit on the first day after surgery. The present findings suggest that extended monitoring may be warranted for many patients. However, previous reports of a nocturnal pattern of ischemia could not be confirmed. 35 By analyzing the individual plots of accumulated duration of ischemia per hour, criteria for the highest possible figures for sensitivity, specificity, positive predictive value, and negative predictive value were found.2 However, these criteria were based on a retrospective analysis. To evaluate the potential benefits of intervention regarding cardiac morbidity, prospective studies are needed. The present data could help form the basis for such studies. Interestingly, the time criterion of -----60minutes of ischemia during 2 hours is within the time range of two previous studies, applying ST analysis on Holter ECG registrations. 7,zv The vector loop analysis revealed signs of a new non-Qwave infarction in six patients, in only three of whom was it evident clinically. This supports the findings that postoperative AMIs may often be silent. 24,36However, the diagnosis of AMI should also include typical increases in plasma enzymes, and these vector loop findings may be considered with some caution. This detailed analysis of the perioperative use of VCG in

44

GANNEDAHL, EDNER, AND LJUNGQVIST

elective abdominal aortic surgery revealed that the three V C G trend parameters QRS-VD, ST-VM, and STC-VM were significantly increased in the group of patients suffering from postoperative cardiac complications. The increase was present already during surgery, but was most pronounced toward the second postoperative day. The data indicate that continued monitoring

should be a minimum precaution in patients showing signs of ischemia during -----60 minutes during 2 hours. Hence, these findings indicate possible criteria that may be used in future prospective studies applying VCG for the evaluation of active intervention in cases of ischemia on perioperative cardiac morbidity.

REFERENCES 1. Thomson IR: Con: Intraoperative myocardial ischemia is not benign. J Cardiothor Vasc Anesth 8:593-595, 1994 2. Nathan H: Pro: Perioperative myocardial ischemia is benign. J Cardiothor Vasc Anesth 8:589-592, 1994 3. Smith J, Cahalan M, Benefiel D, et al: Intraoperative detection of myocardial ischemia in high-risk patients: Electrocardiography versus two-dimensional transesophageal echocardiography. Circulation 72: 1015-1021, 1985 4. Koolen J, Visser C, Reichert S, et al: Improved monitoring of myocardial ischaemia during major vascular surgery using transoesophageal echocardiography. Eur Heart J 13:1028-1033, 1992 5. Ellis JE, Shah MN, Btiller JE, et al: A comparison of methods for the detection of myocardial ischemia during noncardiac surgery: Automated ST-segment analysis systems, electrocardiography, and transesophageal echocardiography. Anesth Analg 75:764-772, 1992 6. Eisenberg MJ, London MJ, Leung JM, et al: Monitoring for myocardial ischemia during noncardiac surgery. A technology assessment of transesophageal echocardiography and 12-lead electrocardiography. JAMA 268:210-216, 1992 7. Landesberg G, Lutia MH, Cotev S, et al: Importance of longduration postoperative ST-segment depression in cardiac morbidity after vascular surgery. Lancet 341:715-719, 1993 8. Ouyang P, Gerstenblith G, Furman WR, et al: Frequency and significance of early postoperative silent myocardial ischemia in patients having peripheral vascular surgery. Am J Cardiol 64:11131116, 1989 9. Mangano DT, Browner WS, Hollenberg M, et al: Association of perioperative myocardial ischemia with cardiac morbidity and mortality in men tmdergoing noncardiac surgery. N Engl J Med 323:1781-1788, 1990 10. Gannedahl R Edner M, Lindahl SGE, Ljungqvist O: Minimal influence of anaesthesia and abdominal surgery on computerized vectorcardiography recordings. Acta Anaesth Scand 39:71-78, 1995 11. Gannedahl R Edner M, Ljungqvist O: Computerized vectorcardiography for improved perioperative cardiac monitoring in vascular surgery. JAm Coll Surg 182:530-536, 1996 12. Dellborg M, Riha M, Swedberg K: Dynamic QRS-complex and ST-segment monitoring in patients with acute myocardial infarction during thrombolytic therapy. A double-blind study with rt-PA vs placebo. Am J Cardio167:343-349, 1991 13. Lundin R Eriksson SV, Fredtikson M, Rehnqvist N: Prognostic information from on-line vectorcardiography in unstable angina pectotis. Cardiology 86:60-66, 1995 14. Dellborg M, Malmberg K, Ryddn L, et al: Dynamic on-line vectorcardiography improves and simplifies in-hospital ischemia monitoring of patients with unstable angina. J Am Coll Cardiol 26:15011507, 1995 15. Dellborg M, Emanuelsson H, Riha M, Swedberg K: Dynamic QRS-complex and ST-segment monitoring by continuous vectorcardiography during coronary angioplasty. Coron Artery Dis 2:43-52, 1991 16. Jensen SM, Johansson G, Osterman G, Reiz S: On-line computerized vectorcardiography of myocardial isehemia during coronary angioplasty: Comparison with 12-lead electrocardiography. Coron Artery Dis 5:507-514, 1994 17. Frank E: An accurate, clinically practical system for spatial vectorcardiography. Circulation 13:737-749, 1956 18. von Arnim T, Reuschel-Janetschek E: Continuous bedside moni-

toting of the ECG for detection of silent myocardial ischemia. Eur Heart J 9:89-92, 1988, (suppl N) 19. Edenbrandt L, Lundh B, Pahlm O: Vectorcardiographic bites. A method for detection and quantification applied on a normal material. J Electrocardio122:325-331, 1989 20. Kalra M, Charlesworth D, Morris JA, A1-Khaffat H: Myocardial infarction after reconstruction of the abdominal aorta. Br J Surg 80:28-31, 1993 21. Baron JF, Mundler O, Bertrand M, et al: Dipyridamole-thallium scintigraphy and gated radionuclide angiography to assess cardiac risk before abdominal aortic surgery. N Engl J Med 330:663-669, 1994 22. Krupski WC, Layag EL, Reilly LM, et al: Comparison of cardiac morbidity between aortic and infrainguinal operations. J Vasc Surg 15:354-365, 1992 23. Lachapelle K, Graham AM, Symes JF: Does the clinical evaluation of the cardiac status predict outcome in patients with abdominal aortic aneurysms. J Vasc Surg 15:964-971, 1992 24. Mangano DT: Perioperative cardiac morbidity. Anesthesiology 72:153-184, 1990 25. Davis RF: Etiology and treatment of perioperative cardiac dysrythmias, in Kaplan JA (ed): Cardiac Anesthesia, vol 1. Orlando, FL, C-rune & Stratton, 1987, pp 421-422 26. Slogoff S, Keats AK: Does perioperative myocardial ischemia lead to postoperative myocardial infarction. Anesthesiology 62:107114, 1985 27. Fleisher LA, Nelson AH, Rosenbaum SH: Postoperative myocardial ischemia: Etiology of cardiac morbidity or manifestation of underlying disease? J Clin Anesth 7:97-102, 1995 28. Riekkinen H, Rautaharju P: Body position, electrode level, and respiration effects on the Frank lead electrocardiogram. Circulation 53:40-45, 1976 29. Feldman T, Borow KM, Neuman A, et al: Relation of electrocardiographic R-wave amplitude to changes in left ventticular chamber size and position in normal subjects. Am J Cardiol 55:1168-1174, 1985 30. Rosenthal A, Restieaux NJ, Feig SA: Influence of acute variations in haematocrit on the QRS complex of the Frank electrocardiogram. Circulation 44:456-465, 1971 31. Dellborg M, Gustafsson G, Riha M, Swedberg K: Dynamic changes of the QRS complex in unstable angina pectoris. Int J Cardiol 36:151-162, 1992 32. McCann RL, Clements FM: Silent myocardial ischemia in patients undergoing peripheral vascular surgery: Incidence and association with perioperative cardiac morbidity and mortality. J Vasc Surg 9:583-587, 1989 33. Langer A: Perioperative myocardial isehemia during noncardiae surgery. N Engl I Med 324:1594, 1991 34. H~iggmark S, Hohner P, Ostman M, et al: Comparison of hemodynamic, electrocardiographic, mechanical, and metabolic indicators of intraoperative myocardial ischemia in vascular surgical patients with coronary artery disease. Anesthesiology 70:19-25, 1989 35. Reeder MK, Muir AD, Forx P, et al: Postoperative myocardial ischaemia: Temporal association with nocturnal hypoxaemia. Br J Anaesth 67:626-631, 1991 36. Charlson ME, Mackenzie CR, Ales K, et al: Surveillance for postoperative myocardial infarction after noncardiac operations. Surg Gynecol Obstet 167:407-414, 1988