| JOURNAL HOME | CME HOME | THIS MONTH | PAST ISSUES | ETOC | COLLECTIONS |
| AUTHORS | REVIEWERS | EDITORIAL BOARD | FEEDBACK | RSS | HELP |
| ||||||||||||||
| ||||||||||||||
|
|
|
|
||||||||||||
|
||||||||||||||
*Division of Cardiovascular/Thoracic Anesthesiology, Mayo Clinic College of Medicine, Rochester, Minnesota; and Neuro-Autonomic Laboratory, NASA Johnson Space Center, Houston, Texas
Address correspondence and reprint requests to Thomas N. Spackman, MD, Department of Anesthesiology, Mayo Clinic, 200 First St. S.W., Rochester, MN 55905. Address e-mail to spackman.thomas{at}mayo.edu.
 |
Abstract |
|---|
 Top Abstract IntroductionMethodsResultsDiscussionAppendix 1References |
|---|
|
Introduction |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAppendix 1References |
|---|
HF QRS analysis involves using a higher sampling rate, signal averaging, and filters to monitor frequencies from 150 to 250 Hz. The usual ECG is derived from frequencies <100 Hz. A representative limb-lead tracing from a conventional surface ECG in a healthy subject is shown in Figure 1. An example of an HF QRS signal from a healthy subject is shown in Figure 2. Note that the amplitude of the HF wave form is in microvolts (Fig. 2) rather than in millivolts (Fig. 1).
|
|
The two most commonly derived variables from HF QRS analysis are the root mean squared (RMS) voltage and a wave pattern called a "reduced-amplitude zone" (RAZ) (see Appendix 1). Diminution of the RMS voltage, which is an estimate of the total energy of the signal, has been shown to occur earlier and with greater sensitivity than the ST segment during balloon angioplasty (3–5) and exercise stress testing (6). An increase in RMS voltage has also shown greater specificity for prediction of successful thrombolytic therapy than ST segment resolution (7). Figure 3 shows a HF QRS signal in a patient with myocardial ischemia and demonstrates the morphology of a RAZ. RAZ patterns are present with greater frequency in CAD (8) and may be indicative of dead or ischemic cardiac conduction tissue (3,9), although they occur in healthy individuals as well.
|
This study was performed to measure in real-time the effect of the induction of general anesthesia on multiple variables of the HF QRS, assuming that there would be no effect, and to assess the practicality of using such a sensitive monitor in the electronically noisy environment of the operating room (OR).
|
Methods |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAppendix 1References |
|---|
This study was approved by our IRB. Thirty patients between the ages of 18 and 60 yr who were scheduled for general anesthesia and surgery for a variety of procedures (ear-nose-throat, dental, orthopedic, vein stripping, and so on) consented to participate. The anesthetic was not controlled but typically consisted of small doses of midazolam and fentanyl, induction with propofol or thiopental, and maintenance with isoflurane or sevoflurane via endotracheal tube. Exclusionary factors were the presence of any of the following: atrial fibrillation, flutter, other sustained supraventricular or ventricular arrhythmia, paced rhythms, QRS interval duration >120 ms, pericarditis patterns, abnormal ECG due to electrolyte abnormalities, history of CAD, diabetes, hypertension, smoking >20 pack-years, or known hyperlipidemia.
Patients were connected to a 12-lead electrode ECG placement in the preoperative holding area. Two separate collections of at least 300 accepted beats, as described in the next paragraph, were made. After the patient was taken to the OR, a third recording was performed just before the induction of general anesthesia. Starting at least 10 min after anesthetic induction, three additional separate recordings were made over the following half-hour.
A 50-beat data average was used for each recording. The first 50-beat recording of each 300-beat collection was discarded to avoid possible start-up artifact. The software program analyzes each incoming beat and compares it with the stored template. Beats were accepted when single-channel cross-correlation was more than 97%.
The computer program calculated multiple variables from the HF QRS. The RMS voltage, the HF QRS energy (HFQE), and the HF QRS integral of absolute value (HFAV) (11) (see Appendix 1) were trended continuously. In addition, baseline noise values were recorded. The presence of a RAZ, as defined by NASA criteria (RAZ N; see Appendix 1), was recorded as a percentage of time present during each 50-beat recording. For continuous response variables of the HF QRS (i.e., the RMS voltage, HFQE, and HFAV), the mean, standard deviation, and coefficient of variation were estimated for each lead by using the combined data from the 15 measurements taken before and after the induction of anesthesia. Student's t-test was used to compare before-and-after measurements for each patient.
|
Results |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAppendix 1References |
|---|
|
|
The overall mean preinduction RMS voltage (Table 3) was 3.41 µV (range, 1.64–8.10 µV). It decreased slightly to 3.26 µV (range, 1.52 to 6.05 µV) after induction (P < 0.001). Of a total combination of 360 leads, RMS voltage decreased significantly (P < 0.05) after induction in 165 leads (46%) and increased significantly in 97 leads (27%).
|
The variables HFAV and HFQE followed a pattern similar to the RMS values but showed more variability. The coefficient of variation was 0.11 for HFAV (before and after induction) and was 0.22 for HFQE (both before and after induction). After induction, the baseline noise level decreased significantly (P < 0.05) in 186 leads (52%) and increased in 36 leads (10%).
Before induction, a RAZ N was present more than 67% of the time in 38 leads (10.5%), or an average of 1.27 RAZ Ns per patient. After induction, it was present in 40 leads (11.1%), or 1.33 per patient. After induction, the RAZ N decreased to less than the 67% threshold in 23 leads and was observed de novo above that level in 25 leads. A RAZ N was observed in three or more spatially contiguous leads in four patients (13%) before induction and in two patients (7%) after induction. Only one patient had this finding both before and after induction.
The beat-rejection algorithm in the software program is designed to reject beats that are not within 97% of agreement with the template established at the start of each data-collection run. When the patients were talking, obviously moving a limb, or being touched by someone else, the rejection rate often increased. The rejection rate ranged from 1.2% to 63%. Electrocautery or use of radiofrequency ablation devices in the OR always interfered with data collection. In one patient only, having the computer plugged into the back of the anesthesia machine resulted in 60-cycle interference, which disappeared when the computer was unplugged and on battery power only.
|
Discussion |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAppendix 1References |
|---|
Our result for the coefficient of variation for RMS voltage was very close to that of Aversano et al. (7). They determined that the average coefficient of variation in 32 control subjects was 10%, or 0.3 µV, and that a significant change for RMS voltage would therefore be 20%, or 0.6 µV. Although RMS voltage changed statistically in many leads with the induction of anesthesia in this study, the change never exceeded the 20% threshold, consistent with the expectation in these patients without known cardiac disease.
We also found that the measurement of HF QRS was quite robust during the induction of anesthesia and surgery. Although electrocautery interrupted the addition of incoming beats to the signal-averaging algorithm, the trend display was not affected, and trending continued as soon as the cautery ceased. Baseline noise levels tended to decrease after induction, probably because of loss of voluntary patient movement.
Why did RMS voltage decrease after the induction of anesthesia? With a threshold of cross-correlation between beats of only 97%, it is possible that voluntary movement contributed slightly to the measurement of RMS voltage through the background noise level. Thus, with the induction of general anesthesia and loss of muscle tone, the RMS voltage may have decreased slightly. An alternative possibility would be that general anesthesia causes a slight decrease in the overall voltage generated in the QRS complex.
Although the presence of a RAZ is a marker for prior myocardial infarction or damage, it was present (at least two thirds of the time) in 11% of leads before induction in this healthy population. Overall, patients had 1.3 RAZ Ns on average. These results are similar to those of Schlegel et al. (10), who found a RAZ N presence of 1.46 per patient in a group of 28 asymptomatic patients who had no risk factors and 1.92 RAZ Ns per patient in a group of 13 patients being evaluated for chest pain who had no CAD by angiography.
Schlegel et al. (10) also performed a simple retrospective analysis of their data by using the presence of three or more RAZ Ns in spatially contiguous ECG leads and found it to be 92% sensitive and 77% specific for identifying uncomplicated CAD. Thus, the presence of this finding in 7%–13% of healthy patients in this study is consistent with their results.
Prior studies mentioned in the introduction point toward the potential use of these variables in the OR. Studies in patients having balloon angioplasty are particularly interesting with regard to ischemic changes that may occur during the stress of a surgical procedure. Pettersson et al. (4) found that in a small series of 19 patients receiving prolonged angioplasty, 4 patients who had no changes in the ST segment showed significant reductions in RMS voltage. In their larger series of 52 patients (5), 88% had significant (>20%) reductions in RMS voltage, as compared with 79% of patients who had ST segment elevation or depression during angioplasty. The American College of Cardiology guidelines (12) for monitoring patients with cardiac disease during anesthesia recommend computerized ST segment analysis, noting an average sensitivity of 74% and an average specificity of 73% for this real-time monitor.
We have shown that although the induction of general anesthesia causes many changes in HF QRS variables—notably, a small decrease in RMS voltage—these changes are within the normal limits of variability, as described in prior studies. In addition, the electrically noisy environment of the OR does not prohibit collection of HF QRS data. Further studies are warranted to assess the sensitivity and specificity of HF QRS variables in the perioperative period as an additional monitor for cardiac ischemia.
|
Appendix 1 |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAppendix 1References |
|---|
Several numerical measures of HF QRS complexes have been devised (11). One of the most commonly used measures is the root mean square (RMS) voltage of the filtered QRS signal, which is equivalent to the area under the curve of the power spectrum, defined as
|
|
where Xi is the filtered voltage at a given sampling point, fqon and fqoff are the onset and offset, respectively, of the HF QRS signal, and FQRSD is the filtered QRS interval duration as defined by fqon and fqoff. The onset and offset of the filtered QRS occur when the voltage exceeds some multiple of the average noise level in an isoelectric portion of the filtered ECG (i.e., in the ST or PR segment).
A better, alternative measure defines RMS voltage in the same fashion except that it uses the unfiltered QRS duration (UQRSD = uqoff – uqon) rather than the FQRSD shown in Equation 1. This alternative measure is likely the most physiologically accurate and is therefore the primary measure of RMS voltage in the software used for this study.
|
|
The values for RMS voltage can nonetheless be overly influenced by noise within the HF QRS complex. Therefore, a third measure, the HF QRS energy (HFQE), has been defined, where
|
|
In this measure, the interval over which the energy is calculated is extended to include that period from 10 ms before the QRS onset until 10 ms after the QRS offset. This extension may reduce the variation of noise that might otherwise corrupt the RMS voltage. Because by definition the RMS voltage and HFQE both use the square of the HF QRS voltage, these measures are very sensitive to small changes in the voltage amplitude. Such sensitivity is not always desirable, because the HF QRS signal will likely exhibit small physiologic variations over the period of monitoring. Therefore, a less sensitive measure, without using the square of the signal amplitude, is the HF integral of absolute values (HFAV). HFAV is also measured from 10 ms before the QRS onset until 10 ms after the QRS offset as
|
|
Reduced-Amplitude Zone
A reduced-amplitude zone (RAZ), as defined by Abboud (9), occurs when at least two local maxima (or corresponding minima) are found within the envelope of the HF QRS complex. Local maxima (or corresponding minima) are in turn defined at a given envelope sample point if, and only if, the amplitude of the HF QRS ECG at that point is higher than (or, for minima, lower than) the amplitude of the three envelope sample points immediately before and after it.
RAZ as Defined by NASA Criteria
A RAZ as defined by NASA criteria (RAZ N) (10) uses stricter criteria than the original definition. It has both a secondary local maximum and a secondary local minimum, both with an absolute voltage of at least X% of their respective primary local maximum and minimum. The variable X% is user adjustable in software but presently defaults to 30%.
|
Footnotes |
|---|
|
References |
|---|
|
TopAbstractIntroductionMethodsResultsDiscussionAppendix 1References |
|---|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
|
