The Reliability of Intermittent ECG Sampling in Arrhythmia Detection By LAFRS RYDAN, M.D., ANDERS WALDENSTROM, M.D.,
AND
STIG HOLMBERG, M.D.
SUM MARY
Although intermittent ECG sampling is a commonly used method for arrhythmia detection there are no controlled studies of its efficiency. In the present study a continuous ECG was recorded for three hours in 52 patients with ventricular tachyarrhythmias. All ECGs were interpreted minute to minute to get the true arrhythmia content. Intermittent ECG samples were simulated by analyzing the first, first two and first five minutes of every 15 minutes in the ECG material. Two minute long ECG samples were also used every 30 minutes. The 15 and 30 minute long periods were classified as containing arrhythmias or not according to the arrhythmia content of the samples. These findings were subsequently compared with the true arrhythmia content. The arrhythmia detection rate depends on the sampling time and true arrhythmia occurrence. The distribution of arrhythmias within the periods was also of importance. With a five minute long ECG sample about 80% of the intervals containing any type of ventricular tachyarrhythmia will be detected. This is reduced to about 50% when one minute long ECG strips are used. Infrequent types of VPCs such as R on T or ventricular tachycardia are very poorly detected also when sampling as much as one-third of the total time. The study also included a comparison between the results obtained from the clinical material and the results obtained by the use of computerized arrhythmia models. It is concluded that intermittent ECG sampling, besides a low detection rate for infrequent arrhythmias and short ECG samples, also brings a risk of underestimating or overemphasizing the arrhythmia occurrence depending on a periodic type of arrhythmia distribution among the patients. These factors make intermittent ECG sampling an unsuitable method for evaluating antiarrhythmic drugs.
W ITH THE INTRODUCTION of coronary care units (CCU) there has been an increasing interest in ventricular arrhythmias complicating acute myocardial infarction (AMI). Early detection and treatment of potentially dangerous ventricular ectopic activity can markedly decrease hospital mortality in AMI.` The increased possibilities of detecting ventricular ectopic activity with improved monitoring facilities have also increased the interest in arrhythmias complicating other types of cardiac diseases. Intermittent ECG sampling is one of several methods used for arrhythmia detection when evaluating the efficiency of antiarrhythmic drugs.4 6 There does not, however, seem to be any critical analysis of the accuracy of this method. This study was designed in order to compare intermittent ECGsampling with a continuous ECG record in detecting different types of ventricular arrhythmias. Material and Methods The material consisted of 52 patients admitted to the From the Department of Medicine I, Division of Cardiology, Sahlgren's Hospital, S-413 45 G8teborg, Sweden. Supported by a grant from the Swedish National Association against Heart and Chest Diseases. Address for reprints: Lars Ryden, M.D., Department of Medicine 1, Division of Cardiology, Sahlgren's Hospital, S-413 45 Goteborg, Sweden. Received August 26, 1974; revision accepted for publication May
15, 1975.
CCU with a proven or suspected AMI. Pertinent clinical data are presented in table 1. As part of another study7 all patients received lidocaine either as an intravenous bolus injection of 75 mg followed by a constant infusion of 2 mg/min or as an intramuscular injection of 300 mg into the vastus lateralis muscle. Except for this antiarrhythmic therapy the patients were managed according to routine procedure in the CCU.8 A multichannel ink writing ECG recorder with a paper speed of 10 mm/sec (Mingograf 81, Elema-Schbnander AB, Sweden) was started as soon as the patient received lidocaine and was continued for three hours thereafter. This electrocardiographic paper record was used for the subsequent arrhythmia analysis and all data concerning the occurrence of ventricular ectopic activity referred to this continuous ECG. This method of arrhythmia detection has been previously described by Mogensen.9 The continuous ECG was analyzed on a minute to minute basis and each 1-min interval was analyzed separately to determine the presence or absence of ventricular arrhythmias. When ventricular premature contractions (VPC) were present they were classified into the following groups: 1-5 VPCs/min; > 5 VPCs/min; R on T VPC; multifocal VPCs; paired VPCs; and ventricular tachycardia (VT) defined as three or more VPCs in sequence at a rate of > 100/min. To illustrate the effect of intermittent ECG sampling, the continuous ECG record from each patient was divided into consecutive 15-min and 30-min periods. To be included in the analysis the ECG of a 15-min period had to be interpretable for at least ten of the fifteen 1-min intervals including the first five 1-min intervals of the period. If these criteria were not fulfilled that particular 15-min period was discarded. The same criteria were applied in analyzing the 30-min periods except that the number of interpretable 1-min intervals had to be more than twenty. Intermittent ECG samples were simulated by selectively 540 Circulation, Volume 52. October 1975
Downloaded from http://circ.ahajournals.org/ by guest on April 30, 2015
RELIABILITY OF INTERMITTENT ECG SAMPLING Table 1
Clinical Data No.
Age (yr)
< 40 41-50 51-60
61-70 > 70 Male Female AMI Angina pectoris VPC of unknown cause Cardiac insufficiency i.v. lidocaine i.m. lidocaine
Sex Diagnosis
Treatment
1 7 20 21 3 43 9 43 5 3 1 23 29
Abbreviations: AMI = acute myocardial infarction; VPC = ventricular premature contractions; i.v. = intravenous; i.m. = intramuscular.
taking the first, the first two and the first five 1-min intervals of each consecutive 15-min period. For the 30-min periods the samples consisted of the first two 1-min intervals. The arrhythmia content of these intermittent samples was analyzed using the same criteria as described previously and the findings were compared with the true arrhythmia content obtained from analysis of all the 1-min intervals in the respective 15- and 30-min periods. This method of comparison is illustrated in figure 1. Panel A shows sampling periods of 1-min, 2-min and 5-min duration, each sample being taken at the beginning of a 15-min period. If period 1 of panel B is used as an example that 15-min period would have been classified as nonarrhythmia containing if only the 1-min sample had been taken despite the fact that the total 15-min period contains three 1-min intervals with VPCs. In an analogous manner if a 2-min long ECG sample was used period 1 in panel B would be classified as arrhythmia containing, whereas period 2 in panel B would not be classified as arrhythmia containing even if a 5-min long ECG sample was chosen. The results of the analysis of all the 15-min and
Astamep zLiinz
541
30-min periods were summed and the number of arrhythmia containing periods as detected by the intermittent sampling techniques described above were compared with the actual number of arrhythmia containing periods obtained from the continuous ECG. To demonstrate that the data obtained from the patient material also holds true in general two statistical models were tested. The distribution of arrhythmia containing 1-min intervals may vary between two extremes, either as a random distribution exemplified in figure 1 panel B 1 or as a periodic distribution (panel B 2). Periodic distributions means that all 1-min intervals containing arrhythmias always have to be located next to each other. A computer was used to calculate the detection rate for a varying content of arrhythmias at these two extremes of distribution. Consequently the computer was programmed to distribute arrhythmia containing 1-min intervals randomly and periodically within the 15-min periods. This distribution was made for an increasing number of arrhythmia containing 1-min intervals ranging from one to fifteen within the 15-min period. Again using figure 1 as an example, panel B 1 shows a random distribution with three 1-min intervals containing arrhythmias. Panel B 2 shows an example of a periodic distribution of three 1-min intervals which may be located at any place within the 15-min period. The computer then simulated 1, 2 and 5 min long ECG samples. The results were plotted to illustrate the detection rate at an increasing number of arrhythmia containing 1-min intervals from one to fifteen. The broken line in figures 4, 5 and 6 represent this relationship for the periodic distribution and the solid line represents the same for the random distribution. The second statistical model was designed to evaluate the probability of detecting arrhythmias of a low and a high frequency respectively over a long period of time. For each consecutive 1-min interval within the 15-min period a random nurnber which was allowed to vary from 0 to 1 was drawn. If this number happened to be < 0.3 that particular 1-min interval was considered to be arrhythmia containing, thus constituting 15-min periods with a high frequency of arrhythmias. The same was performed for random numbers < 0.03 to constitute 15-min periods with a low frequency of arrhythmias. It was then noted if there were any arrhythmias in the first, first two or first five 1-min intervals. This was repeated for one million 15-min periods. All calculations were done in Fortran IV on a PDP 8 E computer with a floating point processor. Results
The absolute incidence of arrhythmias was determined by a minute to minute evaluation of the con1 1 21 tinuous paper recording. According to the protocol for i~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ the comparative lidocaine study7 five of the original 52 5 10 patients were excluded during the period of observation leaving 592 accepted 15-min periods, that is, 8880 Figure 1 1-min intervals. One hundred and sixty-three 1-min Panel A shows the ECG samples of varying length (1, 2 and 5 min). intervals were not interpretable for technical reasons Note that all samples start at the beginning of each 15-min period. (1.9%). Thirty-two of the 15-min periods were exIn panel B are shown examples of 15-min long periods with an arrhythmia distribution of 1-min intervals of the random type (1), cluded, 23 due to exclusion of patients and nine and of the periodic type (2). The arrhythmia containing 1-min interbecause of lack of full EGG information during the vals are shaded. The arrows indicate that each arrhythmia confirst five minutes of the respective period. The taining 1-min period may take any position within the 15-min prevalence of 1-min intervals containing different period (1), or that the arrhythmia containing periods occur in setypes of VPCs as determined by minute to minute quence but their position within the 15-min period may vary (2). B
Arrhythmia
examples
0
Circulation, Volume 52, October 1975
Downloaded from http://circ.ahajournals.org/ by guest on April 30, 2015
RYDEN, WALDENSTROM, HOLMBERG
542 60
L9 4u
A
n
B
C
TL
30
1
0
30 X 30
4,
0-
U~
1-
'
15t a-
A
0
&
SW
v
= ;
gJ
W
15 AM
"I
30
_A
>5
P
M VT
RIT
A >5 P
-I
60
AX
90
=
120
zf
-
150
180
Time (minutes) M VT R/T
Figure 2 The percentage of 1-min intervals (A) and 15-min periods (B) containing different types of defined vent7icular tachyarrhythmias as determined by a minute to minute evaluation of a continuous paper recording. C shows arrhythmia prevalence at different times. A, o all VPC/min; >5, * 5 VPC/min; P, A paired VPC; M, multifocal VPC; VT, ventricular tachycardia; R/T, R on T VPC. =
A =
=
X
>
=
=
=
evaluation of the three hour continuous paper recordings are presented in figure 2 A. Figure 2 B shows the prevalence of 15-min periods containing the defined ventricular arrhythmias. Since the patients were treated with lidocaine from the beginning of the study it was of interest to see whether there was a successive change in the arrhythmia prevalence as a function of time. This is illustrated in figure 2 C. The prevalence of all arrhythmias and especially > 5 VPCs/min diminished during the first three 15-min periods. A 5-min long ECG sample reveals about 80% of the 15-min periods containing any type of VPC (fig. 3). The detection rate decreases for the 2-min and 1-min long samples. The detection rate also decreases with a decreasing arrhythmia content. Consequently, low frequency arrhythmias such as paired or multifocal VPCs are poorly detected; this is even more pronounced if only a 1-min long ECG sample is used. A 1-min sample will only reveal about 10% of the 15min periods containing paired VPCs. When comparing sampling of 1/15 of the ECG time either as a 1-min ECG sample every 15 min or a 2-min long ECG sample every 30 min no significant difference could be found in arrhythmia detection (table 2). The results of the detection rates at different sampling time and arrhythmia frequencies as worked out by the computer when using the previously described statistical method are presented in table 3. In figure 4 the detection of 15-min periods with ventricular arrhythmias is plotted against a varying number of arrhythmia-containing 1-min intervals when using 1-min long ECG samples. The same type of relationship is shown in figures 5 and 6 using 2-min and 5-min long ECG samples. When this relationship
is computerized according to the statistical models above this is described by the solid curve for random
distribution and by the broken curve for periodic distribution. The interpretation of figures 4, 5 and 6 is that a certain decrease of the number of 1-min intervals containing arrhythmias is of different importance for the rate of arrhythmia detection depending on the length of the ECG sample and whether the arrhythmia content is low or high. The distribution of arrhythmia containing 1-min intervals in the patient material is neither random nor periodic but tends to approach the periodic type which is best illustrated in figure 6. Discussion
Although some studies have shown that oscilloscope SampLing time 100 r
5min
80 [
a2
2 min
*
1 min
,z
60 V .W_ al
40
[
1 1
1~~~11
20
Type of VPC
A
>5
P
M
VT
RIT
Figure 3 Detection rate of arrhythmia containing 15-min periods for defined types of arrhythmias with sampling times of 1, 2 and 5 min. Circulation, Volume 52, October 1975
Downloaded from http://circ.ahajournals.org/ by guest on April 30, 2015
543
RELIABILITY OF INTERMITTENT ECG SAMPLING Table 2 Detection Rate when Sampling 1/15 of the Observation Time either as 1 min of 15 or 2 min of 30
Type of VPC
15-min periods with arrhythmia
Any > 5/min Paired Multifocal Ventricular tachycardia R on T
355 140 69 53 28 24
1/15 15-min periods discovered No %
174 61 6 11 5 7
30-min periods with arrhythmia
Difference
NS NS NS NS NS NS
43.2 38.4 9.8 16.7 19.1 23. 5
89 33 5 7 4 4
206 86 51 42 21 17
49.0 43.6 8.7 20.8 17.9 29.2
2/30 30-min periods discovered No %
*According to x2 discrimination.
monitoring under optimal conditions can reach a relatively high precision'0 other studies have demonstrated the unreliability of ECG monitoring particularly in the detection of ventricular
arrhythmias with a relatively low frequency of occurrence (i.e., intermittent VT).9"11 Continuous tape recordings of the ECG with subsequent replay and interpretation by means of an automated arrhythmia detection system has also been employed."-"4 Intermittent ECG sampling is another often used method for evaluation of the efficacy of antiarrhythmic drugs.4-6 Until now no study has been performed to test any of the above mentioned methods as to their accuracy when compared to the true arrhythmia occurrence as is the case for the continuous paper recording. In the present investigation the ECG records were analyzed minute to minute to get the true arrhythmia content. The 1-min intervals were chosen because the frequency of arrhythmias is commonly expressed as VPCs/min. The continuous ECG record subsequently served as a reference against which the accuracy of intermittent ECG samples of various lengths could be tested. Continuous monitoring may be considered one type of intermittent ECG sampling depending on how long and how often the oscilloscope is being watched. Studies where sampling techniques 4- or continuous monitoring7' 1-3 have been used in evaluating antiarrhythmic agents seem to obtain much the same results. As none of these methods have been evaluated
and compared to the true arrhythmia content the consistency of the results is no proof of the adequacy of the methods. Although antiarrhythmic effects have been noted by the use of intermittent ECG sampling,4 the conclusions concerning the antiarrhythmic efficacy may have been different if a higher precision of arrhythmia detection had been used. The patients in this study were all treated with lidocaine for defined ventricular arrhythmias. Thus, the general arrhythmia content was relatively low (fig. 2). It was possible, however, by the subdivision of the 15-min periods, to illustrate the effect of the varying frequency of arrhythmia content on the accuracy of the detection rate. The administration of lidocaine is of minor importance as the true arrhythmia content is
lUU (
1-min intervaLs containing arrhythmia (No) 10 1!5 5
Sampting time 1/15
C- 80 c ----
.c
60
S *
o
a a
8 E
40
periodic distr random --
>5 VPCImin alL VPC paired VPC
multifocal VPC
v
ventricular tachycardia
x
RonT VPC
20
60 80 20 40 1-min intervals containing arrhythmia (%)
100
Table 3
Figure 4
Percent Detection of Arrhythmias in Relation to Decreasing Incidence of Arrhythmias and at Different Lengths of ECGsamples. (Computerized Results)
Detected arrhythmia containing 15-min periods (in percent) for different numbers of 1-min intervals containing arrhythmias (expressed both in percent and numbers) for a sampling time of one minute. In this and the following figures the 15-min periods have been divided into three classes consisting of 1-5, 6-10 and 11-15 arrhythmia containing 1-min intervals. The mean value of arrhythmia containing 1 -min interval was calculated for each class and for all defined types of VPCs. Rare types of arrhythmias are only seen in the first class. The computed results regarding arrhythmia detection for periodic and random arrhythmia distribution are represented by a broken and a solid line, respectively.
1-min interval with arrhythmia
%
33 3.3
15-min periods with arrhythmia (N = 106) No
995199 366827 Quotient:
Detected
'
15-min
period
Sampling time
1 min
2 min
5 min
30 8 3.8
51 16 3.2
84 39 2.2
Circulation, Volume 52, October 1975
Downloaded from http://circ.ahajournals.org/ by guest on April 30, 2015
544 100
RYDEN, WALDENSTROM, HOLMBERG 1-min intervals containing arrhythmia (No) 5 10 -
Sampling 2/15
15
80
time
periodic distr
C0vo/
_random-
/
* >5 VPC/min
ci 60
/
c/
o/ * 40
/
o al VPC
paired VPC multifocat VPC vi ventricular tachycardia
in40
20
20 40 60 80 1-min intervals containing arrhythmia (%)
100
Figure 5 Detected arrhythmia containing 15-min periods for different numbers of 1 -min intervals containing arrhythmias with a sampling time of two minutes.
known from the continuous paper recordings. During the initial 15-min periods there was a decreasing content of all VPCs and especially > 5 VPCs/min (fig. 2 C). This fact would improve the accuracy of detecting arrhythmias by the intermittent sampling techniques since the samples were always taken at the beginning of the 15-min periods where the arrhythmia content was highest. As the patients were suffering from acute coronary insufficiency (table 1) and, moreover, treated with lidocaine the arrhythmia incidence varied. To illustrate the applicability of the method to the different types of arrhythmia distributions and frequencies a computer was programmed to test the detection rate when the arrhythmia prevalence varied from low to high, and when the distribution was 1-min intervals containing arrhythmia (No) 5 10 _
Sampling time
51/5 .
CL ---
_
periodic distr random --
>5 VPC | min o all VPC a paired VPC
*
oi multifocal VPC v ventricuLar tachycardia
C
x
RonT VPC
10 L)
1-min intervals containing arrhythmia (%)
Figure 6 Detected arrhythmia containing 15-min periods for different numbers of 1 -min intervals containing defined types of arrhythmias with a sampling time of five min.
varied from periodic to random. The sampling time was 1, 2 or 5 min (figs. 4, 5 and 6). The arrhythmia detection rate in the patient material is consistent with that of the computer based material and is well situated within the theoretical margins provided by the detection rate for random distribution (solid line) and for periodic distribution (broken line). This suggests that the results of the described sampling technique are generic and applicable to the patient material. The results obtained in this study illustrate the importance that the length of the ECG sample, the distribution of the arrhythmias and the rate of arrhythmia occurrence have on the detection rate of arrhythmias by intermittent sampling. Considering the length of the ECG samples it was found that the collection had to approach one-third of the total time (5-min long ECG samples) to produce a representative discovery of arrhythmia containing periods (fig. 6 and table 3). This holds true only for a high arrhythmia prevalence. For patients with a low arrhythmia frequency a long ECG sample will still be insufficient to adequately detect the true arrhythmia occurrence. Subgroups such as R on T VPC and VT which are considered important predictors of ventricular fibrillation9' 1517 will be very poorly detected. The arrhythmia distribution in the patient material tended to be of the periodic type which is best illustrated in figure 6. This means that if a specific 1-min interval contains arrhythmias the probability is high for the next interval to be arrhythmia containing as well. This study demonstrates that if the arrhythmia occurrence is high from the beginning and has a periodic distribution it can be reduced considerably without being detected by the use of intermittent ECG samples (fig. 6). On the other hand, if the arrhythmia occurrence is low from the beginning the decrease in detection rate is proportionally higher than the decrease in arrhythmia containing 1-min intervals. Thus in such a case the reduction of arrhythmias will be overestimated as shown by looking at the left part of the periodic curve in figure 6. This is, however, not the case if 1-min long ECG samples are used (fig. 4). The detection rate is to a large extent dependent on the actual arrhythmia content. This explains the poor discovery of relatively infrequent arrhythmias such as paired VPCs, multifocal VPCs, R on T VPC, and VT by the means of intermittent ECG sampling. When making a final evaluation of the accuracy of intermittent ECG sampling the various factors of importance in the detection rate must be considered together. This is because the relative influence of each factor may vary with the actual condition. The Circulation, Volume 52, October 1975
Downloaded from http://circ.ahajournals.org/ by guest on April 30, 2015
545
RELIABILITY OF INTERMITTENT ECG SAMPLING
arrhythmia content tends to be more and more underestimated the shorter the ECG samples are and the lower the true arrhythmia content is. Therefore intermittent ECG sampling must be considered an unsuitable method for evaluating the effect of antiarrhythmic drugs. The detection rate would only be adequate in patients with stable arrhythmias. As soon as the arrhythmia frequency changes so does the detection rate. The present results should be considered when investigating the accuracy of other types of intermittent ECG sampling. Although very time-consuming, continuous ECG interpretation must remain the base against which a more practical automated arrhythmia detection system should be tested. Acknowledgment The authors wish to thank Lars Palmer, Ph.D., for skilled assistance with the computerization. We are also most grateful to Richard Harper, M.R.A.C.P. for invaluable linguistic assistance.
References 1. CHIISTIANSEN I, IVERSEN K, SKOUBY AP: Nytten af et Koronarafsnitt. En sammenlignende undersogelse. Ugeskr Laeger 132: 1063, 1970 2. LOWN B, KLEIN MD, HERSHBERG PhI: Coronary and precoronary care. Am J Med 46: 705, 1969 3. HOFVENDAHL S: Influence of treatment in a coronary care unit on prognosis in acute myocardial infarction. A controlled study in 271 cases. Acta Med Scand (suppl 519), 1971 4. SCHWARTZ ML, WEBB NC, COVINo BC, FINCK EM, HEIDER B: Comparative antiarrhythmic effects of intravenously ad-
ministered lidocaine and procainamide and orally administered quinidine. Am J Cardiol 26: 520, 1970 5. BELLET S, ROMAN L, KoSTIS JB, FLEISCHMANN D: Intramuscular lidocaine in the therapy of ventricular arrhythmias. Am j Cardiol 27: 291, 1971
6. BERNSTEIN V BERINSTEIN M, GRITFsFiis J, PE;RETZ DI: Lidocaine intramuscularly in acute myocardial infarction. JAMA 219: 1027, 1972 7. RYD)AN L, WALDEN.STBRM A, EHN L, HOLMBER( S, HUSAINI M: Comparison between effectiveness of intramuscular and intravenous lidocaine on ventricular arrhythmia complicating acute myocardial infarction. Br Heart J 35: 1124, 1973 8. HIE-NNING R, HOLMBERC; S: Erfarenheter fran Sahlgrenska sjukhusets Hjirtinfarktavdelning. Lakartidningen 68: 3603, 1971 9. MOGENSEN L: Ventricular tachyarrhythmias and lignocaine prophylaxis in acute myocardial infarction. A clinical and therapeutic study. Acta Med Scand (suppl 513), 1970 10. RYDLAN L, HOLMBERG S, WALDENSTRSM A: The accuracy of nurses based arrhythmia detection in a coronary care unit. In Scientific Abstracts. First World Congress on Intensive Care, edited by McA LEDINCHAM I. Glasgow, Bell and Bain Ltd, 1974, p 23 11. ROMHILT DW, BLOOMFIELD SS, CHOu TC, FOWLEB NO: Unreliability of conventional electrocardiographic monitoring for arrhythmia detection in coronary care units. Am J Cardiol 31: 457, 1973 12. BLOOMFIELD SS, ROMHILT DW, CHOu TC, FOWLER NO:
Quinidine for prophylaxis of arrhythmias in acute myocardial infarction. N Engl J Med 285: 979, 1971 13. FEHMERS MCO, DUNNING AJ: Intramuscularly and orally administered lidocaine in the treatment of ventricular arrhythmias in acute myocardial infarction. Am J Cardiol 29: 514, 1972 14. GIANELLY R, VON DER GROEBEN JO, SPIVACK AP, HARRISON DC:
Effect of lidocaine on ventricular arrhythmias in patients with coronary heart disease. N Engl J Med 277: 1215, 1967 15. LOWN B, FAKHRO AM, HOOD WB, THORN GW: The coronary care unit. New perspectives and directions. JAMA 199: 188, 1967 16. DHURANDHAR RW, MAC MILLAN RL, BROWN KWG: Primary ventricular fibrillation complicating acute myocardial infarction. Am J Cardiol 27: 347, 1971 17. BRUYNELL KJJ, OPIE LH: The value of warning arrhythmias in the prediction of ventricular fibrillation within one hour of coronary occlusion. Experimental studies in the baboon. Am Heart J 86: 373, 1973
Circulation, Volume 52, October 1975
Downloaded from http://circ.ahajournals.org/ by guest on April 30, 2015
The reliability of intermittent ECG sampling in arrhythmia detection. L Rydén, A Waldenström and S Homberg Circulation. 1975;52:540-545 doi: 10.1161/01.CIR.52.4.540 Circulation is published by the American Heart Association, 7272 Greenville Avenue, Dallas, TX 75231 Copyright © 1975 American Heart Association, Inc. All rights reserved. Print ISSN: 0009-7322. Online ISSN: 1524-4539
The online version of this article, along with updated information and services, is located on the World Wide Web at: http://circ.ahajournals.org/content/52/4/540
Permissions: Requests for permissions to reproduce figures, tables, or portions of articles originally published in Circulation can be obtained via RightsLink, a service of the Copyright Clearance Center, not the Editorial Office. Once the online version of the published article for which permission is being requested is located, click Request Permissions in the middle column of the Web page under Services. Further information about this process is available in the Permissions and Rights Question and Answer document. Reprints: Information about reprints can be found online at: http://www.lww.com/reprints Subscriptions: Information about subscribing to Circulation is online at: http://circ.ahajournals.org//subscriptions/
Downloaded from http://circ.ahajournals.org/ by guest on April 30, 2015
AND
STIG HOLMBERG, M.D.
SUM MARY
Although intermittent ECG sampling is a commonly used method for arrhythmia detection there are no controlled studies of its efficiency. In the present study a continuous ECG was recorded for three hours in 52 patients with ventricular tachyarrhythmias. All ECGs were interpreted minute to minute to get the true arrhythmia content. Intermittent ECG samples were simulated by analyzing the first, first two and first five minutes of every 15 minutes in the ECG material. Two minute long ECG samples were also used every 30 minutes. The 15 and 30 minute long periods were classified as containing arrhythmias or not according to the arrhythmia content of the samples. These findings were subsequently compared with the true arrhythmia content. The arrhythmia detection rate depends on the sampling time and true arrhythmia occurrence. The distribution of arrhythmias within the periods was also of importance. With a five minute long ECG sample about 80% of the intervals containing any type of ventricular tachyarrhythmia will be detected. This is reduced to about 50% when one minute long ECG strips are used. Infrequent types of VPCs such as R on T or ventricular tachycardia are very poorly detected also when sampling as much as one-third of the total time. The study also included a comparison between the results obtained from the clinical material and the results obtained by the use of computerized arrhythmia models. It is concluded that intermittent ECG sampling, besides a low detection rate for infrequent arrhythmias and short ECG samples, also brings a risk of underestimating or overemphasizing the arrhythmia occurrence depending on a periodic type of arrhythmia distribution among the patients. These factors make intermittent ECG sampling an unsuitable method for evaluating antiarrhythmic drugs.
W ITH THE INTRODUCTION of coronary care units (CCU) there has been an increasing interest in ventricular arrhythmias complicating acute myocardial infarction (AMI). Early detection and treatment of potentially dangerous ventricular ectopic activity can markedly decrease hospital mortality in AMI.` The increased possibilities of detecting ventricular ectopic activity with improved monitoring facilities have also increased the interest in arrhythmias complicating other types of cardiac diseases. Intermittent ECG sampling is one of several methods used for arrhythmia detection when evaluating the efficiency of antiarrhythmic drugs.4 6 There does not, however, seem to be any critical analysis of the accuracy of this method. This study was designed in order to compare intermittent ECGsampling with a continuous ECG record in detecting different types of ventricular arrhythmias. Material and Methods The material consisted of 52 patients admitted to the From the Department of Medicine I, Division of Cardiology, Sahlgren's Hospital, S-413 45 G8teborg, Sweden. Supported by a grant from the Swedish National Association against Heart and Chest Diseases. Address for reprints: Lars Ryden, M.D., Department of Medicine 1, Division of Cardiology, Sahlgren's Hospital, S-413 45 Goteborg, Sweden. Received August 26, 1974; revision accepted for publication May
15, 1975.
CCU with a proven or suspected AMI. Pertinent clinical data are presented in table 1. As part of another study7 all patients received lidocaine either as an intravenous bolus injection of 75 mg followed by a constant infusion of 2 mg/min or as an intramuscular injection of 300 mg into the vastus lateralis muscle. Except for this antiarrhythmic therapy the patients were managed according to routine procedure in the CCU.8 A multichannel ink writing ECG recorder with a paper speed of 10 mm/sec (Mingograf 81, Elema-Schbnander AB, Sweden) was started as soon as the patient received lidocaine and was continued for three hours thereafter. This electrocardiographic paper record was used for the subsequent arrhythmia analysis and all data concerning the occurrence of ventricular ectopic activity referred to this continuous ECG. This method of arrhythmia detection has been previously described by Mogensen.9 The continuous ECG was analyzed on a minute to minute basis and each 1-min interval was analyzed separately to determine the presence or absence of ventricular arrhythmias. When ventricular premature contractions (VPC) were present they were classified into the following groups: 1-5 VPCs/min; > 5 VPCs/min; R on T VPC; multifocal VPCs; paired VPCs; and ventricular tachycardia (VT) defined as three or more VPCs in sequence at a rate of > 100/min. To illustrate the effect of intermittent ECG sampling, the continuous ECG record from each patient was divided into consecutive 15-min and 30-min periods. To be included in the analysis the ECG of a 15-min period had to be interpretable for at least ten of the fifteen 1-min intervals including the first five 1-min intervals of the period. If these criteria were not fulfilled that particular 15-min period was discarded. The same criteria were applied in analyzing the 30-min periods except that the number of interpretable 1-min intervals had to be more than twenty. Intermittent ECG samples were simulated by selectively 540 Circulation, Volume 52. October 1975
Downloaded from http://circ.ahajournals.org/ by guest on April 30, 2015
RELIABILITY OF INTERMITTENT ECG SAMPLING Table 1
Clinical Data No.
Age (yr)
< 40 41-50 51-60
61-70 > 70 Male Female AMI Angina pectoris VPC of unknown cause Cardiac insufficiency i.v. lidocaine i.m. lidocaine
Sex Diagnosis
Treatment
1 7 20 21 3 43 9 43 5 3 1 23 29
Abbreviations: AMI = acute myocardial infarction; VPC = ventricular premature contractions; i.v. = intravenous; i.m. = intramuscular.
taking the first, the first two and the first five 1-min intervals of each consecutive 15-min period. For the 30-min periods the samples consisted of the first two 1-min intervals. The arrhythmia content of these intermittent samples was analyzed using the same criteria as described previously and the findings were compared with the true arrhythmia content obtained from analysis of all the 1-min intervals in the respective 15- and 30-min periods. This method of comparison is illustrated in figure 1. Panel A shows sampling periods of 1-min, 2-min and 5-min duration, each sample being taken at the beginning of a 15-min period. If period 1 of panel B is used as an example that 15-min period would have been classified as nonarrhythmia containing if only the 1-min sample had been taken despite the fact that the total 15-min period contains three 1-min intervals with VPCs. In an analogous manner if a 2-min long ECG sample was used period 1 in panel B would be classified as arrhythmia containing, whereas period 2 in panel B would not be classified as arrhythmia containing even if a 5-min long ECG sample was chosen. The results of the analysis of all the 15-min and
Astamep zLiinz
541
30-min periods were summed and the number of arrhythmia containing periods as detected by the intermittent sampling techniques described above were compared with the actual number of arrhythmia containing periods obtained from the continuous ECG. To demonstrate that the data obtained from the patient material also holds true in general two statistical models were tested. The distribution of arrhythmia containing 1-min intervals may vary between two extremes, either as a random distribution exemplified in figure 1 panel B 1 or as a periodic distribution (panel B 2). Periodic distributions means that all 1-min intervals containing arrhythmias always have to be located next to each other. A computer was used to calculate the detection rate for a varying content of arrhythmias at these two extremes of distribution. Consequently the computer was programmed to distribute arrhythmia containing 1-min intervals randomly and periodically within the 15-min periods. This distribution was made for an increasing number of arrhythmia containing 1-min intervals ranging from one to fifteen within the 15-min period. Again using figure 1 as an example, panel B 1 shows a random distribution with three 1-min intervals containing arrhythmias. Panel B 2 shows an example of a periodic distribution of three 1-min intervals which may be located at any place within the 15-min period. The computer then simulated 1, 2 and 5 min long ECG samples. The results were plotted to illustrate the detection rate at an increasing number of arrhythmia containing 1-min intervals from one to fifteen. The broken line in figures 4, 5 and 6 represent this relationship for the periodic distribution and the solid line represents the same for the random distribution. The second statistical model was designed to evaluate the probability of detecting arrhythmias of a low and a high frequency respectively over a long period of time. For each consecutive 1-min interval within the 15-min period a random nurnber which was allowed to vary from 0 to 1 was drawn. If this number happened to be < 0.3 that particular 1-min interval was considered to be arrhythmia containing, thus constituting 15-min periods with a high frequency of arrhythmias. The same was performed for random numbers < 0.03 to constitute 15-min periods with a low frequency of arrhythmias. It was then noted if there were any arrhythmias in the first, first two or first five 1-min intervals. This was repeated for one million 15-min periods. All calculations were done in Fortran IV on a PDP 8 E computer with a floating point processor. Results
The absolute incidence of arrhythmias was determined by a minute to minute evaluation of the con1 1 21 tinuous paper recording. According to the protocol for i~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~~ the comparative lidocaine study7 five of the original 52 5 10 patients were excluded during the period of observation leaving 592 accepted 15-min periods, that is, 8880 Figure 1 1-min intervals. One hundred and sixty-three 1-min Panel A shows the ECG samples of varying length (1, 2 and 5 min). intervals were not interpretable for technical reasons Note that all samples start at the beginning of each 15-min period. (1.9%). Thirty-two of the 15-min periods were exIn panel B are shown examples of 15-min long periods with an arrhythmia distribution of 1-min intervals of the random type (1), cluded, 23 due to exclusion of patients and nine and of the periodic type (2). The arrhythmia containing 1-min interbecause of lack of full EGG information during the vals are shaded. The arrows indicate that each arrhythmia confirst five minutes of the respective period. The taining 1-min period may take any position within the 15-min prevalence of 1-min intervals containing different period (1), or that the arrhythmia containing periods occur in setypes of VPCs as determined by minute to minute quence but their position within the 15-min period may vary (2). B
Arrhythmia
examples
0
Circulation, Volume 52, October 1975
Downloaded from http://circ.ahajournals.org/ by guest on April 30, 2015
RYDEN, WALDENSTROM, HOLMBERG
542 60
L9 4u
A
n
B
C
TL
30
1
0
30 X 30
4,
0-
U~
1-
'
15t a-
A
0
&
SW
v
= ;
gJ
W
15 AM
"I
30
_A
>5
P
M VT
RIT
A >5 P
-I
60
AX
90
=
120
zf
-
150
180
Time (minutes) M VT R/T
Figure 2 The percentage of 1-min intervals (A) and 15-min periods (B) containing different types of defined vent7icular tachyarrhythmias as determined by a minute to minute evaluation of a continuous paper recording. C shows arrhythmia prevalence at different times. A, o all VPC/min; >5, * 5 VPC/min; P, A paired VPC; M, multifocal VPC; VT, ventricular tachycardia; R/T, R on T VPC. =
A =
=
X
>
=
=
=
evaluation of the three hour continuous paper recordings are presented in figure 2 A. Figure 2 B shows the prevalence of 15-min periods containing the defined ventricular arrhythmias. Since the patients were treated with lidocaine from the beginning of the study it was of interest to see whether there was a successive change in the arrhythmia prevalence as a function of time. This is illustrated in figure 2 C. The prevalence of all arrhythmias and especially > 5 VPCs/min diminished during the first three 15-min periods. A 5-min long ECG sample reveals about 80% of the 15-min periods containing any type of VPC (fig. 3). The detection rate decreases for the 2-min and 1-min long samples. The detection rate also decreases with a decreasing arrhythmia content. Consequently, low frequency arrhythmias such as paired or multifocal VPCs are poorly detected; this is even more pronounced if only a 1-min long ECG sample is used. A 1-min sample will only reveal about 10% of the 15min periods containing paired VPCs. When comparing sampling of 1/15 of the ECG time either as a 1-min ECG sample every 15 min or a 2-min long ECG sample every 30 min no significant difference could be found in arrhythmia detection (table 2). The results of the detection rates at different sampling time and arrhythmia frequencies as worked out by the computer when using the previously described statistical method are presented in table 3. In figure 4 the detection of 15-min periods with ventricular arrhythmias is plotted against a varying number of arrhythmia-containing 1-min intervals when using 1-min long ECG samples. The same type of relationship is shown in figures 5 and 6 using 2-min and 5-min long ECG samples. When this relationship
is computerized according to the statistical models above this is described by the solid curve for random
distribution and by the broken curve for periodic distribution. The interpretation of figures 4, 5 and 6 is that a certain decrease of the number of 1-min intervals containing arrhythmias is of different importance for the rate of arrhythmia detection depending on the length of the ECG sample and whether the arrhythmia content is low or high. The distribution of arrhythmia containing 1-min intervals in the patient material is neither random nor periodic but tends to approach the periodic type which is best illustrated in figure 6. Discussion
Although some studies have shown that oscilloscope SampLing time 100 r
5min
80 [
a2
2 min
*
1 min
,z
60 V .W_ al
40
[
1 1
1~~~11
20
Type of VPC
A
>5
P
M
VT
RIT
Figure 3 Detection rate of arrhythmia containing 15-min periods for defined types of arrhythmias with sampling times of 1, 2 and 5 min. Circulation, Volume 52, October 1975
Downloaded from http://circ.ahajournals.org/ by guest on April 30, 2015
543
RELIABILITY OF INTERMITTENT ECG SAMPLING Table 2 Detection Rate when Sampling 1/15 of the Observation Time either as 1 min of 15 or 2 min of 30
Type of VPC
15-min periods with arrhythmia
Any > 5/min Paired Multifocal Ventricular tachycardia R on T
355 140 69 53 28 24
1/15 15-min periods discovered No %
174 61 6 11 5 7
30-min periods with arrhythmia
Difference
NS NS NS NS NS NS
43.2 38.4 9.8 16.7 19.1 23. 5
89 33 5 7 4 4
206 86 51 42 21 17
49.0 43.6 8.7 20.8 17.9 29.2
2/30 30-min periods discovered No %
*According to x2 discrimination.
monitoring under optimal conditions can reach a relatively high precision'0 other studies have demonstrated the unreliability of ECG monitoring particularly in the detection of ventricular
arrhythmias with a relatively low frequency of occurrence (i.e., intermittent VT).9"11 Continuous tape recordings of the ECG with subsequent replay and interpretation by means of an automated arrhythmia detection system has also been employed."-"4 Intermittent ECG sampling is another often used method for evaluation of the efficacy of antiarrhythmic drugs.4-6 Until now no study has been performed to test any of the above mentioned methods as to their accuracy when compared to the true arrhythmia occurrence as is the case for the continuous paper recording. In the present investigation the ECG records were analyzed minute to minute to get the true arrhythmia content. The 1-min intervals were chosen because the frequency of arrhythmias is commonly expressed as VPCs/min. The continuous ECG record subsequently served as a reference against which the accuracy of intermittent ECG samples of various lengths could be tested. Continuous monitoring may be considered one type of intermittent ECG sampling depending on how long and how often the oscilloscope is being watched. Studies where sampling techniques 4- or continuous monitoring7' 1-3 have been used in evaluating antiarrhythmic agents seem to obtain much the same results. As none of these methods have been evaluated
and compared to the true arrhythmia content the consistency of the results is no proof of the adequacy of the methods. Although antiarrhythmic effects have been noted by the use of intermittent ECG sampling,4 the conclusions concerning the antiarrhythmic efficacy may have been different if a higher precision of arrhythmia detection had been used. The patients in this study were all treated with lidocaine for defined ventricular arrhythmias. Thus, the general arrhythmia content was relatively low (fig. 2). It was possible, however, by the subdivision of the 15-min periods, to illustrate the effect of the varying frequency of arrhythmia content on the accuracy of the detection rate. The administration of lidocaine is of minor importance as the true arrhythmia content is
lUU (
1-min intervaLs containing arrhythmia (No) 10 1!5 5
Sampting time 1/15
C- 80 c ----
.c
60
S *
o
a a
8 E
40
periodic distr random --
>5 VPCImin alL VPC paired VPC
multifocal VPC
v
ventricular tachycardia
x
RonT VPC
20
60 80 20 40 1-min intervals containing arrhythmia (%)
100
Table 3
Figure 4
Percent Detection of Arrhythmias in Relation to Decreasing Incidence of Arrhythmias and at Different Lengths of ECGsamples. (Computerized Results)
Detected arrhythmia containing 15-min periods (in percent) for different numbers of 1-min intervals containing arrhythmias (expressed both in percent and numbers) for a sampling time of one minute. In this and the following figures the 15-min periods have been divided into three classes consisting of 1-5, 6-10 and 11-15 arrhythmia containing 1-min intervals. The mean value of arrhythmia containing 1 -min interval was calculated for each class and for all defined types of VPCs. Rare types of arrhythmias are only seen in the first class. The computed results regarding arrhythmia detection for periodic and random arrhythmia distribution are represented by a broken and a solid line, respectively.
1-min interval with arrhythmia
%
33 3.3
15-min periods with arrhythmia (N = 106) No
995199 366827 Quotient:
Detected
'
15-min
period
Sampling time
1 min
2 min
5 min
30 8 3.8
51 16 3.2
84 39 2.2
Circulation, Volume 52, October 1975
Downloaded from http://circ.ahajournals.org/ by guest on April 30, 2015
544 100
RYDEN, WALDENSTROM, HOLMBERG 1-min intervals containing arrhythmia (No) 5 10 -
Sampling 2/15
15
80
time
periodic distr
C0vo/
_random-
/
* >5 VPC/min
ci 60
/
c/
o/ * 40
/
o al VPC
paired VPC multifocat VPC vi ventricular tachycardia
in40
20
20 40 60 80 1-min intervals containing arrhythmia (%)
100
Figure 5 Detected arrhythmia containing 15-min periods for different numbers of 1 -min intervals containing arrhythmias with a sampling time of two minutes.
known from the continuous paper recordings. During the initial 15-min periods there was a decreasing content of all VPCs and especially > 5 VPCs/min (fig. 2 C). This fact would improve the accuracy of detecting arrhythmias by the intermittent sampling techniques since the samples were always taken at the beginning of the 15-min periods where the arrhythmia content was highest. As the patients were suffering from acute coronary insufficiency (table 1) and, moreover, treated with lidocaine the arrhythmia incidence varied. To illustrate the applicability of the method to the different types of arrhythmia distributions and frequencies a computer was programmed to test the detection rate when the arrhythmia prevalence varied from low to high, and when the distribution was 1-min intervals containing arrhythmia (No) 5 10 _
Sampling time
51/5 .
CL ---
_
periodic distr random --
>5 VPC | min o all VPC a paired VPC
*
oi multifocal VPC v ventricuLar tachycardia
C
x
RonT VPC
10 L)
1-min intervals containing arrhythmia (%)
Figure 6 Detected arrhythmia containing 15-min periods for different numbers of 1 -min intervals containing defined types of arrhythmias with a sampling time of five min.
varied from periodic to random. The sampling time was 1, 2 or 5 min (figs. 4, 5 and 6). The arrhythmia detection rate in the patient material is consistent with that of the computer based material and is well situated within the theoretical margins provided by the detection rate for random distribution (solid line) and for periodic distribution (broken line). This suggests that the results of the described sampling technique are generic and applicable to the patient material. The results obtained in this study illustrate the importance that the length of the ECG sample, the distribution of the arrhythmias and the rate of arrhythmia occurrence have on the detection rate of arrhythmias by intermittent sampling. Considering the length of the ECG samples it was found that the collection had to approach one-third of the total time (5-min long ECG samples) to produce a representative discovery of arrhythmia containing periods (fig. 6 and table 3). This holds true only for a high arrhythmia prevalence. For patients with a low arrhythmia frequency a long ECG sample will still be insufficient to adequately detect the true arrhythmia occurrence. Subgroups such as R on T VPC and VT which are considered important predictors of ventricular fibrillation9' 1517 will be very poorly detected. The arrhythmia distribution in the patient material tended to be of the periodic type which is best illustrated in figure 6. This means that if a specific 1-min interval contains arrhythmias the probability is high for the next interval to be arrhythmia containing as well. This study demonstrates that if the arrhythmia occurrence is high from the beginning and has a periodic distribution it can be reduced considerably without being detected by the use of intermittent ECG samples (fig. 6). On the other hand, if the arrhythmia occurrence is low from the beginning the decrease in detection rate is proportionally higher than the decrease in arrhythmia containing 1-min intervals. Thus in such a case the reduction of arrhythmias will be overestimated as shown by looking at the left part of the periodic curve in figure 6. This is, however, not the case if 1-min long ECG samples are used (fig. 4). The detection rate is to a large extent dependent on the actual arrhythmia content. This explains the poor discovery of relatively infrequent arrhythmias such as paired VPCs, multifocal VPCs, R on T VPC, and VT by the means of intermittent ECG sampling. When making a final evaluation of the accuracy of intermittent ECG sampling the various factors of importance in the detection rate must be considered together. This is because the relative influence of each factor may vary with the actual condition. The Circulation, Volume 52, October 1975
Downloaded from http://circ.ahajournals.org/ by guest on April 30, 2015
545
RELIABILITY OF INTERMITTENT ECG SAMPLING
arrhythmia content tends to be more and more underestimated the shorter the ECG samples are and the lower the true arrhythmia content is. Therefore intermittent ECG sampling must be considered an unsuitable method for evaluating the effect of antiarrhythmic drugs. The detection rate would only be adequate in patients with stable arrhythmias. As soon as the arrhythmia frequency changes so does the detection rate. The present results should be considered when investigating the accuracy of other types of intermittent ECG sampling. Although very time-consuming, continuous ECG interpretation must remain the base against which a more practical automated arrhythmia detection system should be tested. Acknowledgment The authors wish to thank Lars Palmer, Ph.D., for skilled assistance with the computerization. We are also most grateful to Richard Harper, M.R.A.C.P. for invaluable linguistic assistance.
References 1. CHIISTIANSEN I, IVERSEN K, SKOUBY AP: Nytten af et Koronarafsnitt. En sammenlignende undersogelse. Ugeskr Laeger 132: 1063, 1970 2. LOWN B, KLEIN MD, HERSHBERG PhI: Coronary and precoronary care. Am J Med 46: 705, 1969 3. HOFVENDAHL S: Influence of treatment in a coronary care unit on prognosis in acute myocardial infarction. A controlled study in 271 cases. Acta Med Scand (suppl 519), 1971 4. SCHWARTZ ML, WEBB NC, COVINo BC, FINCK EM, HEIDER B: Comparative antiarrhythmic effects of intravenously ad-
ministered lidocaine and procainamide and orally administered quinidine. Am J Cardiol 26: 520, 1970 5. BELLET S, ROMAN L, KoSTIS JB, FLEISCHMANN D: Intramuscular lidocaine in the therapy of ventricular arrhythmias. Am j Cardiol 27: 291, 1971
6. BERNSTEIN V BERINSTEIN M, GRITFsFiis J, PE;RETZ DI: Lidocaine intramuscularly in acute myocardial infarction. JAMA 219: 1027, 1972 7. RYD)AN L, WALDEN.STBRM A, EHN L, HOLMBER( S, HUSAINI M: Comparison between effectiveness of intramuscular and intravenous lidocaine on ventricular arrhythmia complicating acute myocardial infarction. Br Heart J 35: 1124, 1973 8. HIE-NNING R, HOLMBERC; S: Erfarenheter fran Sahlgrenska sjukhusets Hjirtinfarktavdelning. Lakartidningen 68: 3603, 1971 9. MOGENSEN L: Ventricular tachyarrhythmias and lignocaine prophylaxis in acute myocardial infarction. A clinical and therapeutic study. Acta Med Scand (suppl 513), 1970 10. RYDLAN L, HOLMBERG S, WALDENSTRSM A: The accuracy of nurses based arrhythmia detection in a coronary care unit. In Scientific Abstracts. First World Congress on Intensive Care, edited by McA LEDINCHAM I. Glasgow, Bell and Bain Ltd, 1974, p 23 11. ROMHILT DW, BLOOMFIELD SS, CHOu TC, FOWLEB NO: Unreliability of conventional electrocardiographic monitoring for arrhythmia detection in coronary care units. Am J Cardiol 31: 457, 1973 12. BLOOMFIELD SS, ROMHILT DW, CHOu TC, FOWLER NO:
Quinidine for prophylaxis of arrhythmias in acute myocardial infarction. N Engl J Med 285: 979, 1971 13. FEHMERS MCO, DUNNING AJ: Intramuscularly and orally administered lidocaine in the treatment of ventricular arrhythmias in acute myocardial infarction. Am J Cardiol 29: 514, 1972 14. GIANELLY R, VON DER GROEBEN JO, SPIVACK AP, HARRISON DC:
Effect of lidocaine on ventricular arrhythmias in patients with coronary heart disease. N Engl J Med 277: 1215, 1967 15. LOWN B, FAKHRO AM, HOOD WB, THORN GW: The coronary care unit. New perspectives and directions. JAMA 199: 188, 1967 16. DHURANDHAR RW, MAC MILLAN RL, BROWN KWG: Primary ventricular fibrillation complicating acute myocardial infarction. Am J Cardiol 27: 347, 1971 17. BRUYNELL KJJ, OPIE LH: The value of warning arrhythmias in the prediction of ventricular fibrillation within one hour of coronary occlusion. Experimental studies in the baboon. Am Heart J 86: 373, 1973
Circulation, Volume 52, October 1975
Downloaded from http://circ.ahajournals.org/ by guest on April 30, 2015
The reliability of intermittent ECG sampling in arrhythmia detection. L Rydén, A Waldenström and S Homberg Circulation. 1975;52:540-545 doi: 10.1161/01.CIR.52.4.540 Circulation is published by the American Heart Association, 7272 Greenville Avenue, Dallas, TX 75231 Copyright © 1975 American Heart Association, Inc. All rights reserved. Print ISSN: 0009-7322. Online ISSN: 1524-4539
The online version of this article, along with updated information and services, is located on the World Wide Web at: http://circ.ahajournals.org/content/52/4/540
Permissions: Requests for permissions to reproduce figures, tables, or portions of articles originally published in Circulation can be obtained via RightsLink, a service of the Copyright Clearance Center, not the Editorial Office. Once the online version of the published article for which permission is being requested is located, click Request Permissions in the middle column of the Web page under Services. Further information about this process is available in the Permissions and Rights Question and Answer document. Reprints: Information about reprints can be found online at: http://www.lww.com/reprints Subscriptions: Information about subscribing to Circulation is online at: http://circ.ahajournals.org//subscriptions/
Downloaded from http://circ.ahajournals.org/ by guest on April 30, 2015
