Circadian Variation in Ventricular Electrical Instability Associated with Coronary Artery Disease James McClelland, MD, Blair Halperin, MD, Joel Cutler, MD, Peter Kudenchuk, MD, Jack Kron, MD, and John McAnulty, MD
AHhough sudden cardiac deaths and ischsmk cardiac events clearly occur in a circadian pattem, suchapattemhasnotbeenshownforprimaryarrhythmic events. Recause p&nary a-ythmk events are thougM to play an important rote in sudden cardiac death, a large series of ventrkular stimuiatkn studies was analyxed to determine whether circadian variation in ventricular ekctficai i~biiii exists. if such a circadian variation could be shown, it could have implkationr for the conduct and hlterprataion of ekctmphysiokgk testing and the etiology of circadian variation in sudden cardiac death. Results of 2 dmg-free vent-r stimulationstudies performed4to28heursapartineach Ofl62patbltSWith coronary artery disease were adyxed. Rate and duration of in&tced arrhythmia, number of extrastbnuli required to induce arrhythmiaandchangesinthesefactorsbetweenthe2 tests in each patient were analyxed. Comparisons were mads by ha&day, by hour and in a temporally conti-8 manner to eliminate Man associated with any single method. No signifkant cimdian variation was found in any electrophysiologic measure of venMadar electrical itibiiity despite adequate statistkai power. Theseffndlngsshowthatthethneofdaydving which ventrkuiar stimuiatkn tests are performed doesnotaffecttestresults,andtheMoredoesnot needtobecoMmikd during ekctmphysiokgic stdes. if thew findings are parallel to those in ambulatory patients with coronary artery disease, then circadian changes in ve&icular electrical instabiiii may not play as important a role in the circadian pattern of sudden cardiac death as had been previornly kgM. (Am J Cardid 1990;65:1351-1357)
From the Division of Cardiology, Department of Medicine, Oregon Health Sciences University, Portland, Oregon. This study was supported by grant HL 07596 from the National Institutes of Health, Betheada, Maryland. Manuscript received October 6, 1989; revised manuscript received and accepted January 30,199O. Address for reprints: James McClelland, MD, Division of Cardiology, L-462, Oregon Health Sciences University, 3 18 1 S.W. Sam Jackson Park Road, Portland, Oregon 97201.
he majority of arrhythmic sudden cardiac deaths are thought to begin as primary arrhythmic events or as events secondary to ischemia. While sudden deathly2 and primary ischemic events3 have been shown to occur in a circadian pattern, no such pattern has been shown for primary arrhythmic events. Because primary arrhythmic events account for an important portion of all sudden cardiac deaths,4 but have not yet been shown to occur in a circadian pattern, we investigated whether temporal variation in ventricular electrical instability exists that could serve as the basis for a circadian pattern of arrhythmic sudden death. If such a variation exists, it could be responsible for part of the circadian variation in sudden cardiac death. If such a variation is absent, then other factors would be required to explain circadian variation in sudden cardiac death. In addition, circadian variation in the results of ventricular stimulation studies could have implications for the conduct and interpretation of these tests. Investigation of circadian variation in ventricular electrical instability has previously used review of death certificates or continuous ambulatory electrocardiographic recording. Each of these methods has limitations. We used another method by investigating ventricular electrical instability through analysis of a large series of ventricular stimulation studies. These studies are uniquely suited to this analysis, as changes in the stability of the electrophysiologic substrate alone are evaluated while variations in the frequency of initiators of arrhythmic events, such as ischemia and ventricular ectopy, are controlled. In addition, temporal inaccuracies and reporting biases, problems that may hamper analysis of circadian patterns under less well-controlled circumstances, are eliminated.
T
METHODS Patients: Between 1985 and 1988,222 patients with coronary artery disease underwent electrophysiologic evaluation of serious ventricular arrhythmia at Oregon Health Sciences University; these electrophysiologic tests were analyzed retrospectively in this study. Patients were excluded from analysis if the times of their tests were not clearly recorded (n = 31) or if they required ongoing antiarrhythmic therapy or a P-blocking medication (n = 29). Patients requiring @blockers were excluded because these agents are thought to reduce circadian variation in cardiac events. *J The remaining 162 patients each underwent 2 drug-free ventricular stimulation studies 4 to 28 hours apart. Table I lists the cliniTHE AMERICAN
JOURNAL
OF CARDIOLOGY
JUNE
1. 1990
1351
TABLE
I Patient
Characteristics Group
No. of patients Age Ws) Sex (M/F) Presenting arrhythmia Nonsustained ventricular Ventricular tachycardia Ventricular fibrillation
tachycardia
1
Group 2
Group 3
Total
p Value
55 56f14 4w7
47 6Oi13 36/11
60 57i14 52/8
162 58kl2 136/26
NS NS
3 21 31
4 21 22
2 26 32
9 68 85
0.7
Group 1 = test 1 before noon and test 2 after noon; Group 2 = test 1 after noon and test 2 before noon; Group 3 = both tests before noon or both tests after noon. NS = not significant.
TABLE
of Rate and Duration
II Categorization
of Arrhythmia
Arrhythmia
Category
None 1 to 3 repetitive ventricular responses 4 to 6 beats of nonsustained ventricular tachycardia >6 beats of nonsustained ventricular tachycardia Sustained ventricular tachycardia, rate 100 to 160 beats/min Sustained ventricular tachycardia, rate 161 to 200 beats/min Sustained ventricular tachycardia. rate 201 to 300 beats/min Sustained ventricular tachycardia. rate >300 beats/min or ventricular
1 2 3 4 5 6 7 8
cal characteristics of these patients, grouped by time of day at which their studies were performed. Informed consent was obtained from each subject. Defin&ms: Three patient groups were defined to facilitate pairwise analysis of morning and afternoon tests. Group 1 is made up of patients who underwent test 1 (the first test) in the morning and test 2 (the second test) in the afternoon. Group 2 is made up of patients who underwent test 1 in the afternoon and test 2 the following morning. Group 3 is made up of the remaining patients in whom both studies were performed in the morning or both in the afternoon. Repetitive ventricular response was defined as an induced ventricular depolarization after the last captured ventricular pacing stimulus; nonsustained ventricular tachycardia as >4 repetitive ventricular responses and 30 seconds or requiring intervention due to hemodynamic instability; and ventricular fibrillation as completely disorganized ventricular elcctrical activity associated with hemodynamic instability. Ventricular stimulation protocok Our ventricular stimulation protocol has been previously described in detail.6 Two drug-free studies are performed in each patient; the first (test 1) uses 2 right ventricular pacing sites and 2 paced cycle lengths, with up to 4 programmed extrastimuli. Pacing proceeds from the right ventricular apex using 1, 2 and 3 extrastimuli before pacing using another catheter at the right ventricular outflow tract. Lidocaine (‘%%) is used for local anesthesia during the first study. A second study (test 2) is performed 4 to 28 hours later in the cardiac care unit using a right subclavian pacing catheter that is retained from the first study. Patients are fully ambulatory between the studies. Aside from pacing at only the apical site and the lack of local anesthesia, the second study follows a pacing protocol identical to that of the first. 1352
THE AMERICAN
JOURNAL
OF CARDIOLOGY
VOLUME
65
fibrillation
The time at which each study is performed is based solely on scheduling concerns and bears no relation to patient characteristics. Data edlection: The hour and minute at which pacing began was recorded during each ventricular stimulation study. An electrocardiogram is recorded continuously during each study and induced arrhythmia is recorded on a standardized form. Each arrhythmic response is assigned an arrhythmia category based on the number of repetitive ventricular responses and the rate of tachycardia (if sustained) using a previously reported categorization system6 (Table II). The highest category attained during single extrastimulus technique with either cycle length is entered into a computerized database as the induced arrhythmia. The same is done for double, triple and quadruple extrastimuli technique. A complete ventricular stimulation study therefore results in assigning an arrhythmia category for the results of pacing with each of 1, 2, 3 and 4 extrastimuli. The highest of these categories is taken as the arrhythmia category for the entire test. This system allows analysis of traditional clinical endpoints, such as nonsustained or sustained ventricular tachycardia, as well as rates of sustained tachycardia and duration of nonsustained tachycardia. It also allows for pairwise analysis of arrhythmia induction and electrophysiologic studies can be compared as though 300 orVF Total Mean ridit No. extrastimuli required induce sustained VT
4 44
89 162
0.51 5 23 40
6 81 27 38 172 324 >0.3*
3 19 41 20 83
8 42 81 41 172
21 89 0.49
0.51
12 37 24 16
7 37 22 17
19 74 46 33
89 0.49
83
172
0.51
>0.5*
to
1 2 3 4 Total Mean ridit
>0.5*
* Mann-Whitney-Wilcoxon test. NSVl = nonsustained ventrwmlar tachycardla; RVR = repetitive ventricular sponse; VF = ventricular fibrillation: VT = ventricular tachycardia.
re-
and 6 P.M. Because the unpaired analyses depend on tests 1 and 2 being substantially equal in capacity to induce arrhythmia, results from these tests were compared to determine if they differ in this regard. There are no significant differences between tests 1 and 2 in duration or rate of induced arrhythmia or number of extrastimuli required for induction of sustained ventricular tachycardia (Table III). Temporal variation in irwkction of arrhythmia: To avoid errors associated with any single method, temporal variation in arrhythmia induction was analyzed in the following 3 ways: continuously, by hour and by halfday. A.M.
TEMPORALLY CONTINUOUS ANALYSIS SUSTAINED VENTRICULAR TACHYCARDIA:
OFINDUCTION
OF
Harmonic regression analysis was used to derive an equation relating the probability of induction of sustained ventricular tachycardia to the time at which the test was performed. The time of day (to the nearest minute) was
Frequency of Induction of Sustained Arrhythmia
70
9
10
11
12
13
14
15
16
17
16
19 al
Time of Day THE AMERICANJOURNAL OF CARDIOLOGY JUNE 1. 1990
1353
TABLE IV Comparison
of lnducibility
Before
and After
Frequency of induction of sustained ventricular tachycardia Not induced Induced Duration of induced arrhythmia 0 RVRs 1 to 3 RVRs 4 to 6 RVRs NSVT Sustained VT or VF Mean ridit Rate of sustained VT (beats/min) 100 to 160 161 to 200 201 to 300 >3OOorVF Mean ridit No. extrastimuli required for induction of sustained VT 1 2 3 4 Mean ridit
Noon
Before Noon
After Noon
IL&!
Ll!!a!E
54 64
98 108
152 172
0.8
1 32 5 16 64 0.51
5 49 22 22 108 0.50
6 81 27 38 172
2 17 26 19 0.52
6 25 55 22 0.49
8 42 81 41
3 25 24 12 0.55
16 49 22 21 0.47
19 74 46 33
>0.7”
>0.4*
0.061
* Mann-Whitney-Wilcoxon test. Abbreviatms as in Table III.
used as the independent variable and induction of sustained ventricular tachycardia was used as the dependent variable in the following equation: PII] = 0.53 + (O.OOl)cos(?rt/12) + (O.OOl)sin(ti/12). PII] represents the probability of induction of sustained ventricular tachycardia and t represents the time of the test in hours. The coefficients of the sine and cosine terms are not statistically different from zero (F test); that is, there is no time of day when the induction of ventricular tachycardia occurs significantly more frequently than the mean rate of induction. HOURLY
VARIATION
IN
INDUCTION
OF SUSTAINED
VEN-
The percentage of tests in which ventricular tachycardia was induced during each hour of testing is shown in Figure 1. No hourly induction rate is significantly different from the mean daily frequency of induction. The smallest p value for this difference is noted for the hour between 11 A.M. and noon (chi-square = 1.0, p = 0.32), when the incidence of induction was reduced to 42%; all other p values exceed 0.46. TRICULAR
TACHYCARDIA:
INDUCTION
OF ARRHYTHMIA
BEFORE
NOON
VERSUS
AF-
NOON: When tests performed in the morning are compared to tests performed in the afternoon, there is no significant difference between frequency of induction of sustained ventricular tachycardia, duration-related arrhythmia categories, rate of induced tachycardia or number of extrastimuli required for the induction of sustained ventricular tachycardia (Table IV). Ridit analysis shows statistically insignificant trends toward longer induced arrhythmias, faster rates and greater number of extrastimuli required for arrhythmia induction in the morning compared to the afternoon.
TER
PAIRWISE BETWEEN
1354
COMPARISON
TESTS
1AND
THE AMERICAN
2:
OF
DIFFERENCES
IN
INDUCTION
If it were true that electrical in-
JOURNAL
OF CARDIOLOGY
VOLUME
65
stability is greater in the morning than the afternoon, then one would expect arrhythmia categories to reflect a decrease in the severity of arrhythmia between tests 1 and 2 in group 1 patients (who were first studied in the morning) and an increase between the 2 tests in group 2 patients (who were first studied in the afternoon). The rates of induced arrhythmia and numbers of extrastimuli required for arrhythmia induction would be expected to reflect this difference in electrical instability in a similar fashion. This pairwise analysis is listed in Table V; changes in induced arrhythmia were calculated by subtracting the arrhythmia category (listed in Table II) attained during test 2 from the arrhythmia category attained during test 1 in the same patient. There is no statistically significant difference between any of the 3 groups with respect to changes in any of these parameters. In addition, the mean ridit values do not reflect any consistent trend. DISCUSSION
In these 162 patients with coronary artery disease presenting for evaluation of serious ventricular arrhythmia, there is no detectable circadian variation in ventricular electrical instability between the hours of 7 A.M. and 11 P.M. as assessedby ventricular stimulation studies. No significant variations are present in the duration of induced arrhythmia, frequency of induction of sustained ventricular tachycardia, rate of induced tachycardia, “ease of induction” as measured by number of extrastimuli required for induction or changes in these parameters between consecutive ventricular stimulation tests in the same patient. Trends in one direction or the other are small. No circadian variations are found when patients are grouped by presenting arrhythmia or gender. Although not presented here, analysis after exclu-
TABLE V Paitwise
Analysis
of Differences
in induction Category
Change
in duration
of induced
arrhythmia
of Arrhythmia Change
-3 -2 -1 0 fl +2 +3
Mean ridit Change in rate of induced
1
0 0
Tests
1 and 2
Group 2 3 0 5 28 7
Group
3
p Value
1
3 0.49
1 5 43 6 0 4 0.47
>0.9*
4 17 3 0.52
4 14 0 0.44
5 18 4 0.52
>0.5*
1
3 2
1 9 13 3 1 0.4-8
>0.9*
7 34 9 4 1 0.52
1
tachycardia -1 0 +1
Mean ridit Change in no. of extrastimuli to induce VT
Between
Group
required -2 -1 0 +1 +2
Mean ridit
7 11 4 1 0.52
10 2
1 0.50
* Kruskal-Wallis test.
sion of test results obtained using 4 extrastimuli also failed to detect a circadian variation in ventricular electrical instability. These findings are of interest for several reasons. First, they suggest that the time of the study does not need to be controlled during ventricular stimulation tests. Second, ventricular electrical instability, as assessed by the technique that we use clinically for this purpose, has no discernible circadian variation. This is true even though factors other than the pacing protocol itself were left uncontrolled during this study. This fmding is concordant with investigations of ventricular electrical instability using the other primary means for assessing ventricular electrical instability-continuous ambulatory electrocardiographic monitoring (see discussion). Most interestingly, if these data apply to nonhospitalized persons with coronary artery disease, they suggest that changes in ventricular electrical instability may not play an important role in the known circadian variation in the onset of sudden death.
Others have shown that the incidence of symp tomatic ventricular arrhythmias peaks near 11 A.M.~* The factors underlying these inconsistent findings are not entirely clear, although the well-described spontaneous variability of ventricular ectopic activity is likely to be a factor.22,23 Reviews of death certificates have clearly shown circadian variation in the incidence of all forms of sudden death taken together. 1,2Hinkle and Thaler have shown that death certificates discriminate poorly between primary arrhythmic deaths and arrhythmic deaths secondary to myocardial ischemia. In their review of 142 sudden deaths, 41% of those deaths classified as arrhythmic in whom an autopsy was performed had evidence of acute occlusion of a major coronary artery.4 More than 75% of these were clinically “silent,” indicating that clinical findings available to those completing death certificates may not be helpful in distinguishing between primary ischemic and primary arrhythmic events. A major difference between our study and others is Analytic methods for the detection of circadian the use of a third method, ventricular stimulation studvariation in vehiadar electrical instability: Previous ies, for the evaluation of circadian changes in ventricuinvestigations of circadian variation in arrhythmic lar electrical instability. Errors associated with accurate events have used review of death certificates or continu- categorization and timing of the events are avoided usous ambulatory electrocardiographic recording. The ing this method. These errors may introduce a systematearliest reports using electrocardiographic monitoring ic bias that results in showing circadian variations that compared sleeping and waking hours and showed in- do not truly exist or in the obscuring of true circadian creased ectopy during certain phases of .~leep.~~-~~ Sub- relation. In particular, there has been concern that the sequent studies of sleep cycles have shown no difference well-documented morning increase in the frequency of in the frequency of arrhythmia between sleeping and cardiac events may be a manifestation of delayed diswaking hours13 or a marked decrease during ~leep.‘~J~ covery and reporting of events that occur between midStudies of 24-hour cycles of ventricular ectopic activity night and 6 A.M., when patients and observers are likely showed that peaks were either nonexistent,i6 absent for to be asleep. This delay would falsely decrease the apsimple ventricular ectopy but present for nonsustained parent frequency of very early morning events and ventricular tachycardia,” unimodal at 7 A.M.,** unifalsely increase the apparent frequency of events after 6 modal with a peak between 8 P.M. and midnight5 or bi- A.M. These potentially serious limitations to circadian modal with peaks at 8 A.M. and 4 P.M.*~ or 11 A.M. and 9 analysis have been discussed in full by others.24 By usP.M.~O
THE AMERICAN
JOURNAL
OF CARDIOLOGY
JUNE
1, 1990
1355
ing electrophysiologic tests, we have been able to investigate changes in ventricular electrical instability in isolation from changes in the frequency of triggers of arrhythmia and acute ischemic events. IMPLICATIONS FOR THE DETERMINANTS VARIATION IN THE ONSET OF SUDDEN CARDIAC
OF CIRCADIAN DEATH Stud-
ies based on continuous ambulatory electrocardiographic monitoring have shown that ectopic activity does not follow a consistent circadian pattern, and does not parallel the pattern of sudden death as determined by Muller and Willich and their co-workers.*,2 Our study also supports this lack of circadian variation in ventricular electrical instability. Sudden death does appear to follow a circadian pattern similar to that of acute myo&dial infarction; if electrical instability of these patients parallels that of ambulatory patients, our study supports the concept that circadian variation in the incidence of sudden cardiac death may be primarily a manifestation of the well-documented circadian variation in primary ischemic events alone. Limitations: The lack of circadian variation in induction of arrhythmia during ventricular stimulation studies does not preclude all circadian variation in ventricular electrical instability in these or other patients for the 4 reasons that follow. AND
METHODOLOGIC DIFFERENCES EXIST BETWEEN THE FIRST SECOND ~~~~~:‘Ideally, for statistical purposes, the
conduct of the first and second tests should be identical. In our study, the fiist test occurred in the catheterization laboratory, shortly after the insertion of electrode catheters, whereas the second test occurred at the bedside using an indwelling catheter and the stimulation protocols were identical only through 3 extrastimuli. Lidocaine (ti%) was used during the fust test, but not the second. Reanalysis of the data using only results from the first 3 extrastimuli, which eliminates diiferences in stimulation techniques between the tests, did not substantially change the findings. The effects of methodologic differences on test results are listed in Table III; they are not statistically significant. That pacing from a se&d site should play a minor role in arrhythmia induction is supported by the findings of Kudenchuk et al who showed that stimulation from multiple right ventricular sites results in only a small increment in arrhythmia induction.25 CIRCADIAN BE ABOLISHED
VARIATION IN ELECTRICAL IN HOSPITALIZED PATIENTS
INSTABILITY
MAY
A number’ of studies support the concept that hospitalization alone does not abolish the circadian variation of human physiologic parameters.26 It has been shown that circadian variation in some physiologic variables, such as renal excretion of water and electrolytes, may be lowered in amplitude by severe restriction of activity, but the circadian pattern ,persists.27Patients in this study were hospitalized but they were not acutely ilk for that reason we suspect that their circadian rhythms remained largely intact. If circadian rhythms had been disturbed by the electrophysiologic testing itself, however, their detection would be difficult or impossible using thii method. CIRCADIAN TLE TO DETECT:
13S6
VARIATION
MAY
BE PRESENT,
BUT
TOO SUB-
In the largest study of sudden death, a
THE AMERICAN
JOURNAL
OF CARDIOLOGY
VOLUME
65
33.8% increase in incidence was reported during the 6 hours before noon as compared to the 6 hours after noon1 Based on our mean induction rate of 53%, induction would have to decrease from 63% in the morning to 42% in the afternoon to attain a change of that magnitude. To attain an 80% probability of detecting such a change with a l-tailed test and a p value of 0.05, 220 observations are required2*; 324 observations are available in our study. THE HOURS BETWEEN MIDNIGHT AND 7 A.M. WERE NOT OBSERVED Because no tests were performed between midnight and 7 A.M., cyclic variation in ventricular elec-
trical instability-during those hours would not be detected by our study. Investigators of sudden cardiac death have found peak frequencies to occur during the late morning, however, when many of our tests were performed. Condusiensr Our study shows that the time of day at which ventricular stimulation studies are performed does not affect induction of ventricular arrhythmia. Therefore, from a practical standpoint, it is not a variable that needs to be controlled during serial drug testing. In addition, if these results reflect the state of ventricular electrical instability in nonhospitalized patients with coronary artery disease, then primary arrhythmic events may be a less important component of the circadian variation in sudden death than previously thought. REFERENCES
1; Muller JE, Ludmer PL, Willich SN, Tofler GH, Aylmer G, Klangos I, Stone PH. Circadian variation in the frequency of sudden cardiac death. Circulation
1987;75:131-138.
2. Willich SN, Levy D, Rocco MB, Tofler GH, Stone PH, Muller JE. Circadian variation in the incidence of sudden cardiac death in the Framingham Heart Study population. Am J Cardiol 1987.60:801-806. 3. Muller JE, Stone PH, Turi ZG, Rutherford JD, Czeisler CA, Parker C, Poole WK. Passamani E, Roberts R, Robertson T, Sobe. BE, Willerson JT, Braunwald E, the MILIS Study Group. Circadian variation in the frequency of onset of acute myocardial infaraion. N Erg/ J Med 1985;313:1315-1322. 4. Hinkle LE. Thaler HT. Clinical classification of cardiac deaths. Circulation 1982,S:457-464. 5. Mir MA. Effect of a long acting beta-adrenoceptor blocker on diurnal variation of cardiac dysrhythmias. Postgrad Med J 1986,62:175-178. 6. Kudenchuk PJ, Kron J, Walance CG, Murphy ES, Morris CD, Griffith KK, McAnulty JH. Reproducibility of arrhythmia induction with intracardiac electrophysiologic testing: patients with clinical sustained tachyarrhythmias. JACC 1986;7:819-828.
7. McCullagh
P. Regression
models for ordinal data. J R Star Sot Ser E
1980;42:109-142.
8. Fleiss JL. The Design and Analysis of Clinical Experiments. New York: John Wiley and Sons, 1986:83-84. 9. Kantor S, Winkelstein W, lbrihim MA. A note on the interpretation of ridit as a quantile rank. Am J Epidemiol 1968:87:609-615. 10. Rosenblatt G, Zwilling G, Hartmann E. Electrocardiographic changes during sheep in patients with cardiac abnormalities. Psychophjsiology 1969,6:233. 11. Lester BK. Block R, Gunn CG. Wolf S. The relation of cardiac arrhythmias t? phases of sleep. Clin Res 1969;17:456. 12. Shahaway ME. Arrhythmias and the varieties of sleep. N Engl J Med 1970:282:815.
13. Smith R, Johnson L, Rothfeld D, Zir L, Tharp B. Sleep and cardiac arrhyth&as. Arch Intern Med 1972;130:751-753. 14. Low 9, Tykocinski M, Garfein A, Brooks P. Sleep and ventricular premature beats. Circulation 1973;48:691-701. 15. Pickering TG, Goulding L, Cobern BA. Diurnal variations in ventricular ectopic beats and heart rate. Cardiouasc Med 1977;2:1013-1022. 16. Northcote RJ, MacFarlane P, Kesson CM, Ballantyne D. Continuous 24hour electrocardiography in thyrotoxicosis before and after treatment. Am Heart J 1986;112:339-344.
17. Rebuzzi AG, Lucente M, Lanza GA, Coppola E, Manzoli U. Circadian rhythm of ventricular tachycardia. Advances in Chronobiology. New York: Alan R Liss. 1987:(part
B) 153-158.
18. Leach CN, Ruskin JN, Halberg
F. Circadian
variation of frequent human
ventricular ectopy. Fed Proc 1981;40,423. 19. Saito K, Matsuyama K, Niki T, Mori H. Characteristics of non-sustained ventricular tachycardia detected by ambulatory electrocardiography. Jpn Circ J J984:48:421-426.
20. Raeder EA, Hohnloser SH, Graboys TB, Podrid PJ, Lampert S, Lawn B. Spontaneous variability and circadian distribution of ectopic activity in patients with malignant ventricular arrhythmia. JACC 1988;12:656-661. 21. Twidale N, Taylor S, Heddle WF, Ayres BF, Tonkin AM. Morning increase in the time of onset of sustained ventricular tachycardia. Am J Cardiol 1989;64:1204-1206.
22. Michelson EL, Morganroth J. Spontaneous variability of complex ventricular arrhythmias detected by long-term electrocardiographic recordings. Circulation 1980,661:690-695.
23. Anastasiou-Nana MI, Menlove RL, Nanas JN, Anderson JL. Changes in spontaneous variability of ventricular ectopic activity as a function of time in
patients with chronic arrhythmias. Circularion J988;78:286-295, 24. Myocardial infarction community registers: results of a WHO international collaborative study coordinated by the regional oftice for Europe. Public Health in Europe, No. 5. Copenhagen Regional Offlee for Europe (World Health Organiration),
1976:J-232.
25. Kudenchuk PJ, Kron J, Walance C, McAnulty JH. Limited value of programmed electrical stimulation from multiple right ventricular pacing sites in clinically sustained ventricular fibrillation or ventricular tachycardia associated with coronary artery disease. Am J Cardiol 1988,61:303-308. 26. Conroy RTLW, Mills JN. Human Circadian Rhythms. L.ondorr J.A. Churchill, J970:127-135.
27. Lubban bed-rest and 28. Kraemer in Research.
MC, Tredre BE. Renal diurnal rhythms in human subjects during limited activity. J Physiol J964;171:26P-27P. HC, Thiemann S. How Many Subjects? Statistical Power Analysis London
THE AMERICAN
Sage Publications.
JOURNAL
J987:76-83.
OF CARDIOLOGY
JUNE
1, 1990
1357
AHhough sudden cardiac deaths and ischsmk cardiac events clearly occur in a circadian pattem, suchapattemhasnotbeenshownforprimaryarrhythmic events. Recause p&nary a-ythmk events are thougM to play an important rote in sudden cardiac death, a large series of ventrkular stimuiatkn studies was analyxed to determine whether circadian variation in ventricular ekctficai i~biiii exists. if such a circadian variation could be shown, it could have implkationr for the conduct and hlterprataion of ekctmphysiokgk testing and the etiology of circadian variation in sudden cardiac death. Results of 2 dmg-free vent-r stimulationstudies performed4to28heursapartineach Ofl62patbltSWith coronary artery disease were adyxed. Rate and duration of in&tced arrhythmia, number of extrastbnuli required to induce arrhythmiaandchangesinthesefactorsbetweenthe2 tests in each patient were analyxed. Comparisons were mads by ha&day, by hour and in a temporally conti-8 manner to eliminate Man associated with any single method. No signifkant cimdian variation was found in any electrophysiologic measure of venMadar electrical itibiiity despite adequate statistkai power. Theseffndlngsshowthatthethneofdaydving which ventrkuiar stimuiatkn tests are performed doesnotaffecttestresults,andtheMoredoesnot needtobecoMmikd during ekctmphysiokgic stdes. if thew findings are parallel to those in ambulatory patients with coronary artery disease, then circadian changes in ve&icular electrical instabiiii may not play as important a role in the circadian pattern of sudden cardiac death as had been previornly kgM. (Am J Cardid 1990;65:1351-1357)
From the Division of Cardiology, Department of Medicine, Oregon Health Sciences University, Portland, Oregon. This study was supported by grant HL 07596 from the National Institutes of Health, Betheada, Maryland. Manuscript received October 6, 1989; revised manuscript received and accepted January 30,199O. Address for reprints: James McClelland, MD, Division of Cardiology, L-462, Oregon Health Sciences University, 3 18 1 S.W. Sam Jackson Park Road, Portland, Oregon 97201.
he majority of arrhythmic sudden cardiac deaths are thought to begin as primary arrhythmic events or as events secondary to ischemia. While sudden deathly2 and primary ischemic events3 have been shown to occur in a circadian pattern, no such pattern has been shown for primary arrhythmic events. Because primary arrhythmic events account for an important portion of all sudden cardiac deaths,4 but have not yet been shown to occur in a circadian pattern, we investigated whether temporal variation in ventricular electrical instability exists that could serve as the basis for a circadian pattern of arrhythmic sudden death. If such a variation exists, it could be responsible for part of the circadian variation in sudden cardiac death. If such a variation is absent, then other factors would be required to explain circadian variation in sudden cardiac death. In addition, circadian variation in the results of ventricular stimulation studies could have implications for the conduct and interpretation of these tests. Investigation of circadian variation in ventricular electrical instability has previously used review of death certificates or continuous ambulatory electrocardiographic recording. Each of these methods has limitations. We used another method by investigating ventricular electrical instability through analysis of a large series of ventricular stimulation studies. These studies are uniquely suited to this analysis, as changes in the stability of the electrophysiologic substrate alone are evaluated while variations in the frequency of initiators of arrhythmic events, such as ischemia and ventricular ectopy, are controlled. In addition, temporal inaccuracies and reporting biases, problems that may hamper analysis of circadian patterns under less well-controlled circumstances, are eliminated.
T
METHODS Patients: Between 1985 and 1988,222 patients with coronary artery disease underwent electrophysiologic evaluation of serious ventricular arrhythmia at Oregon Health Sciences University; these electrophysiologic tests were analyzed retrospectively in this study. Patients were excluded from analysis if the times of their tests were not clearly recorded (n = 31) or if they required ongoing antiarrhythmic therapy or a P-blocking medication (n = 29). Patients requiring @blockers were excluded because these agents are thought to reduce circadian variation in cardiac events. *J The remaining 162 patients each underwent 2 drug-free ventricular stimulation studies 4 to 28 hours apart. Table I lists the cliniTHE AMERICAN
JOURNAL
OF CARDIOLOGY
JUNE
1. 1990
1351
TABLE
I Patient
Characteristics Group
No. of patients Age Ws) Sex (M/F) Presenting arrhythmia Nonsustained ventricular Ventricular tachycardia Ventricular fibrillation
tachycardia
1
Group 2
Group 3
Total
p Value
55 56f14 4w7
47 6Oi13 36/11
60 57i14 52/8
162 58kl2 136/26
NS NS
3 21 31
4 21 22
2 26 32
9 68 85
0.7
Group 1 = test 1 before noon and test 2 after noon; Group 2 = test 1 after noon and test 2 before noon; Group 3 = both tests before noon or both tests after noon. NS = not significant.
TABLE
of Rate and Duration
II Categorization
of Arrhythmia
Arrhythmia
Category
None 1 to 3 repetitive ventricular responses 4 to 6 beats of nonsustained ventricular tachycardia >6 beats of nonsustained ventricular tachycardia Sustained ventricular tachycardia, rate 100 to 160 beats/min Sustained ventricular tachycardia, rate 161 to 200 beats/min Sustained ventricular tachycardia. rate 201 to 300 beats/min Sustained ventricular tachycardia. rate >300 beats/min or ventricular
1 2 3 4 5 6 7 8
cal characteristics of these patients, grouped by time of day at which their studies were performed. Informed consent was obtained from each subject. Defin&ms: Three patient groups were defined to facilitate pairwise analysis of morning and afternoon tests. Group 1 is made up of patients who underwent test 1 (the first test) in the morning and test 2 (the second test) in the afternoon. Group 2 is made up of patients who underwent test 1 in the afternoon and test 2 the following morning. Group 3 is made up of the remaining patients in whom both studies were performed in the morning or both in the afternoon. Repetitive ventricular response was defined as an induced ventricular depolarization after the last captured ventricular pacing stimulus; nonsustained ventricular tachycardia as >4 repetitive ventricular responses and 30 seconds or requiring intervention due to hemodynamic instability; and ventricular fibrillation as completely disorganized ventricular elcctrical activity associated with hemodynamic instability. Ventricular stimulation protocok Our ventricular stimulation protocol has been previously described in detail.6 Two drug-free studies are performed in each patient; the first (test 1) uses 2 right ventricular pacing sites and 2 paced cycle lengths, with up to 4 programmed extrastimuli. Pacing proceeds from the right ventricular apex using 1, 2 and 3 extrastimuli before pacing using another catheter at the right ventricular outflow tract. Lidocaine (‘%%) is used for local anesthesia during the first study. A second study (test 2) is performed 4 to 28 hours later in the cardiac care unit using a right subclavian pacing catheter that is retained from the first study. Patients are fully ambulatory between the studies. Aside from pacing at only the apical site and the lack of local anesthesia, the second study follows a pacing protocol identical to that of the first. 1352
THE AMERICAN
JOURNAL
OF CARDIOLOGY
VOLUME
65
fibrillation
The time at which each study is performed is based solely on scheduling concerns and bears no relation to patient characteristics. Data edlection: The hour and minute at which pacing began was recorded during each ventricular stimulation study. An electrocardiogram is recorded continuously during each study and induced arrhythmia is recorded on a standardized form. Each arrhythmic response is assigned an arrhythmia category based on the number of repetitive ventricular responses and the rate of tachycardia (if sustained) using a previously reported categorization system6 (Table II). The highest category attained during single extrastimulus technique with either cycle length is entered into a computerized database as the induced arrhythmia. The same is done for double, triple and quadruple extrastimuli technique. A complete ventricular stimulation study therefore results in assigning an arrhythmia category for the results of pacing with each of 1, 2, 3 and 4 extrastimuli. The highest of these categories is taken as the arrhythmia category for the entire test. This system allows analysis of traditional clinical endpoints, such as nonsustained or sustained ventricular tachycardia, as well as rates of sustained tachycardia and duration of nonsustained tachycardia. It also allows for pairwise analysis of arrhythmia induction and electrophysiologic studies can be compared as though 300 orVF Total Mean ridit No. extrastimuli required induce sustained VT
4 44
89 162
0.51 5 23 40
6 81 27 38 172 324 >0.3*
3 19 41 20 83
8 42 81 41 172
21 89 0.49
0.51
12 37 24 16
7 37 22 17
19 74 46 33
89 0.49
83
172
0.51
>0.5*
to
1 2 3 4 Total Mean ridit
>0.5*
* Mann-Whitney-Wilcoxon test. NSVl = nonsustained ventrwmlar tachycardla; RVR = repetitive ventricular sponse; VF = ventricular fibrillation: VT = ventricular tachycardia.
re-
and 6 P.M. Because the unpaired analyses depend on tests 1 and 2 being substantially equal in capacity to induce arrhythmia, results from these tests were compared to determine if they differ in this regard. There are no significant differences between tests 1 and 2 in duration or rate of induced arrhythmia or number of extrastimuli required for induction of sustained ventricular tachycardia (Table III). Temporal variation in irwkction of arrhythmia: To avoid errors associated with any single method, temporal variation in arrhythmia induction was analyzed in the following 3 ways: continuously, by hour and by halfday. A.M.
TEMPORALLY CONTINUOUS ANALYSIS SUSTAINED VENTRICULAR TACHYCARDIA:
OFINDUCTION
OF
Harmonic regression analysis was used to derive an equation relating the probability of induction of sustained ventricular tachycardia to the time at which the test was performed. The time of day (to the nearest minute) was
Frequency of Induction of Sustained Arrhythmia
70
9
10
11
12
13
14
15
16
17
16
19 al
Time of Day THE AMERICANJOURNAL OF CARDIOLOGY JUNE 1. 1990
1353
TABLE IV Comparison
of lnducibility
Before
and After
Frequency of induction of sustained ventricular tachycardia Not induced Induced Duration of induced arrhythmia 0 RVRs 1 to 3 RVRs 4 to 6 RVRs NSVT Sustained VT or VF Mean ridit Rate of sustained VT (beats/min) 100 to 160 161 to 200 201 to 300 >3OOorVF Mean ridit No. extrastimuli required for induction of sustained VT 1 2 3 4 Mean ridit
Noon
Before Noon
After Noon
IL&!
Ll!!a!E
54 64
98 108
152 172
0.8
1 32 5 16 64 0.51
5 49 22 22 108 0.50
6 81 27 38 172
2 17 26 19 0.52
6 25 55 22 0.49
8 42 81 41
3 25 24 12 0.55
16 49 22 21 0.47
19 74 46 33
>0.7”
>0.4*
0.061
* Mann-Whitney-Wilcoxon test. Abbreviatms as in Table III.
used as the independent variable and induction of sustained ventricular tachycardia was used as the dependent variable in the following equation: PII] = 0.53 + (O.OOl)cos(?rt/12) + (O.OOl)sin(ti/12). PII] represents the probability of induction of sustained ventricular tachycardia and t represents the time of the test in hours. The coefficients of the sine and cosine terms are not statistically different from zero (F test); that is, there is no time of day when the induction of ventricular tachycardia occurs significantly more frequently than the mean rate of induction. HOURLY
VARIATION
IN
INDUCTION
OF SUSTAINED
VEN-
The percentage of tests in which ventricular tachycardia was induced during each hour of testing is shown in Figure 1. No hourly induction rate is significantly different from the mean daily frequency of induction. The smallest p value for this difference is noted for the hour between 11 A.M. and noon (chi-square = 1.0, p = 0.32), when the incidence of induction was reduced to 42%; all other p values exceed 0.46. TRICULAR
TACHYCARDIA:
INDUCTION
OF ARRHYTHMIA
BEFORE
NOON
VERSUS
AF-
NOON: When tests performed in the morning are compared to tests performed in the afternoon, there is no significant difference between frequency of induction of sustained ventricular tachycardia, duration-related arrhythmia categories, rate of induced tachycardia or number of extrastimuli required for the induction of sustained ventricular tachycardia (Table IV). Ridit analysis shows statistically insignificant trends toward longer induced arrhythmias, faster rates and greater number of extrastimuli required for arrhythmia induction in the morning compared to the afternoon.
TER
PAIRWISE BETWEEN
1354
COMPARISON
TESTS
1AND
THE AMERICAN
2:
OF
DIFFERENCES
IN
INDUCTION
If it were true that electrical in-
JOURNAL
OF CARDIOLOGY
VOLUME
65
stability is greater in the morning than the afternoon, then one would expect arrhythmia categories to reflect a decrease in the severity of arrhythmia between tests 1 and 2 in group 1 patients (who were first studied in the morning) and an increase between the 2 tests in group 2 patients (who were first studied in the afternoon). The rates of induced arrhythmia and numbers of extrastimuli required for arrhythmia induction would be expected to reflect this difference in electrical instability in a similar fashion. This pairwise analysis is listed in Table V; changes in induced arrhythmia were calculated by subtracting the arrhythmia category (listed in Table II) attained during test 2 from the arrhythmia category attained during test 1 in the same patient. There is no statistically significant difference between any of the 3 groups with respect to changes in any of these parameters. In addition, the mean ridit values do not reflect any consistent trend. DISCUSSION
In these 162 patients with coronary artery disease presenting for evaluation of serious ventricular arrhythmia, there is no detectable circadian variation in ventricular electrical instability between the hours of 7 A.M. and 11 P.M. as assessedby ventricular stimulation studies. No significant variations are present in the duration of induced arrhythmia, frequency of induction of sustained ventricular tachycardia, rate of induced tachycardia, “ease of induction” as measured by number of extrastimuli required for induction or changes in these parameters between consecutive ventricular stimulation tests in the same patient. Trends in one direction or the other are small. No circadian variations are found when patients are grouped by presenting arrhythmia or gender. Although not presented here, analysis after exclu-
TABLE V Paitwise
Analysis
of Differences
in induction Category
Change
in duration
of induced
arrhythmia
of Arrhythmia Change
-3 -2 -1 0 fl +2 +3
Mean ridit Change in rate of induced
1
0 0
Tests
1 and 2
Group 2 3 0 5 28 7
Group
3
p Value
1
3 0.49
1 5 43 6 0 4 0.47
>0.9*
4 17 3 0.52
4 14 0 0.44
5 18 4 0.52
>0.5*
1
3 2
1 9 13 3 1 0.4-8
>0.9*
7 34 9 4 1 0.52
1
tachycardia -1 0 +1
Mean ridit Change in no. of extrastimuli to induce VT
Between
Group
required -2 -1 0 +1 +2
Mean ridit
7 11 4 1 0.52
10 2
1 0.50
* Kruskal-Wallis test.
sion of test results obtained using 4 extrastimuli also failed to detect a circadian variation in ventricular electrical instability. These findings are of interest for several reasons. First, they suggest that the time of the study does not need to be controlled during ventricular stimulation tests. Second, ventricular electrical instability, as assessed by the technique that we use clinically for this purpose, has no discernible circadian variation. This is true even though factors other than the pacing protocol itself were left uncontrolled during this study. This fmding is concordant with investigations of ventricular electrical instability using the other primary means for assessing ventricular electrical instability-continuous ambulatory electrocardiographic monitoring (see discussion). Most interestingly, if these data apply to nonhospitalized persons with coronary artery disease, they suggest that changes in ventricular electrical instability may not play an important role in the known circadian variation in the onset of sudden death.
Others have shown that the incidence of symp tomatic ventricular arrhythmias peaks near 11 A.M.~* The factors underlying these inconsistent findings are not entirely clear, although the well-described spontaneous variability of ventricular ectopic activity is likely to be a factor.22,23 Reviews of death certificates have clearly shown circadian variation in the incidence of all forms of sudden death taken together. 1,2Hinkle and Thaler have shown that death certificates discriminate poorly between primary arrhythmic deaths and arrhythmic deaths secondary to myocardial ischemia. In their review of 142 sudden deaths, 41% of those deaths classified as arrhythmic in whom an autopsy was performed had evidence of acute occlusion of a major coronary artery.4 More than 75% of these were clinically “silent,” indicating that clinical findings available to those completing death certificates may not be helpful in distinguishing between primary ischemic and primary arrhythmic events. A major difference between our study and others is Analytic methods for the detection of circadian the use of a third method, ventricular stimulation studvariation in vehiadar electrical instability: Previous ies, for the evaluation of circadian changes in ventricuinvestigations of circadian variation in arrhythmic lar electrical instability. Errors associated with accurate events have used review of death certificates or continu- categorization and timing of the events are avoided usous ambulatory electrocardiographic recording. The ing this method. These errors may introduce a systematearliest reports using electrocardiographic monitoring ic bias that results in showing circadian variations that compared sleeping and waking hours and showed in- do not truly exist or in the obscuring of true circadian creased ectopy during certain phases of .~leep.~~-~~ Sub- relation. In particular, there has been concern that the sequent studies of sleep cycles have shown no difference well-documented morning increase in the frequency of in the frequency of arrhythmia between sleeping and cardiac events may be a manifestation of delayed diswaking hours13 or a marked decrease during ~leep.‘~J~ covery and reporting of events that occur between midStudies of 24-hour cycles of ventricular ectopic activity night and 6 A.M., when patients and observers are likely showed that peaks were either nonexistent,i6 absent for to be asleep. This delay would falsely decrease the apsimple ventricular ectopy but present for nonsustained parent frequency of very early morning events and ventricular tachycardia,” unimodal at 7 A.M.,** unifalsely increase the apparent frequency of events after 6 modal with a peak between 8 P.M. and midnight5 or bi- A.M. These potentially serious limitations to circadian modal with peaks at 8 A.M. and 4 P.M.*~ or 11 A.M. and 9 analysis have been discussed in full by others.24 By usP.M.~O
THE AMERICAN
JOURNAL
OF CARDIOLOGY
JUNE
1, 1990
1355
ing electrophysiologic tests, we have been able to investigate changes in ventricular electrical instability in isolation from changes in the frequency of triggers of arrhythmia and acute ischemic events. IMPLICATIONS FOR THE DETERMINANTS VARIATION IN THE ONSET OF SUDDEN CARDIAC
OF CIRCADIAN DEATH Stud-
ies based on continuous ambulatory electrocardiographic monitoring have shown that ectopic activity does not follow a consistent circadian pattern, and does not parallel the pattern of sudden death as determined by Muller and Willich and their co-workers.*,2 Our study also supports this lack of circadian variation in ventricular electrical instability. Sudden death does appear to follow a circadian pattern similar to that of acute myo&dial infarction; if electrical instability of these patients parallels that of ambulatory patients, our study supports the concept that circadian variation in the incidence of sudden cardiac death may be primarily a manifestation of the well-documented circadian variation in primary ischemic events alone. Limitations: The lack of circadian variation in induction of arrhythmia during ventricular stimulation studies does not preclude all circadian variation in ventricular electrical instability in these or other patients for the 4 reasons that follow. AND
METHODOLOGIC DIFFERENCES EXIST BETWEEN THE FIRST SECOND ~~~~~:‘Ideally, for statistical purposes, the
conduct of the first and second tests should be identical. In our study, the fiist test occurred in the catheterization laboratory, shortly after the insertion of electrode catheters, whereas the second test occurred at the bedside using an indwelling catheter and the stimulation protocols were identical only through 3 extrastimuli. Lidocaine (ti%) was used during the fust test, but not the second. Reanalysis of the data using only results from the first 3 extrastimuli, which eliminates diiferences in stimulation techniques between the tests, did not substantially change the findings. The effects of methodologic differences on test results are listed in Table III; they are not statistically significant. That pacing from a se&d site should play a minor role in arrhythmia induction is supported by the findings of Kudenchuk et al who showed that stimulation from multiple right ventricular sites results in only a small increment in arrhythmia induction.25 CIRCADIAN BE ABOLISHED
VARIATION IN ELECTRICAL IN HOSPITALIZED PATIENTS
INSTABILITY
MAY
A number’ of studies support the concept that hospitalization alone does not abolish the circadian variation of human physiologic parameters.26 It has been shown that circadian variation in some physiologic variables, such as renal excretion of water and electrolytes, may be lowered in amplitude by severe restriction of activity, but the circadian pattern ,persists.27Patients in this study were hospitalized but they were not acutely ilk for that reason we suspect that their circadian rhythms remained largely intact. If circadian rhythms had been disturbed by the electrophysiologic testing itself, however, their detection would be difficult or impossible using thii method. CIRCADIAN TLE TO DETECT:
13S6
VARIATION
MAY
BE PRESENT,
BUT
TOO SUB-
In the largest study of sudden death, a
THE AMERICAN
JOURNAL
OF CARDIOLOGY
VOLUME
65
33.8% increase in incidence was reported during the 6 hours before noon as compared to the 6 hours after noon1 Based on our mean induction rate of 53%, induction would have to decrease from 63% in the morning to 42% in the afternoon to attain a change of that magnitude. To attain an 80% probability of detecting such a change with a l-tailed test and a p value of 0.05, 220 observations are required2*; 324 observations are available in our study. THE HOURS BETWEEN MIDNIGHT AND 7 A.M. WERE NOT OBSERVED Because no tests were performed between midnight and 7 A.M., cyclic variation in ventricular elec-
trical instability-during those hours would not be detected by our study. Investigators of sudden cardiac death have found peak frequencies to occur during the late morning, however, when many of our tests were performed. Condusiensr Our study shows that the time of day at which ventricular stimulation studies are performed does not affect induction of ventricular arrhythmia. Therefore, from a practical standpoint, it is not a variable that needs to be controlled during serial drug testing. In addition, if these results reflect the state of ventricular electrical instability in nonhospitalized patients with coronary artery disease, then primary arrhythmic events may be a less important component of the circadian variation in sudden death than previously thought. REFERENCES
1; Muller JE, Ludmer PL, Willich SN, Tofler GH, Aylmer G, Klangos I, Stone PH. Circadian variation in the frequency of sudden cardiac death. Circulation
1987;75:131-138.
2. Willich SN, Levy D, Rocco MB, Tofler GH, Stone PH, Muller JE. Circadian variation in the incidence of sudden cardiac death in the Framingham Heart Study population. Am J Cardiol 1987.60:801-806. 3. Muller JE, Stone PH, Turi ZG, Rutherford JD, Czeisler CA, Parker C, Poole WK. Passamani E, Roberts R, Robertson T, Sobe. BE, Willerson JT, Braunwald E, the MILIS Study Group. Circadian variation in the frequency of onset of acute myocardial infaraion. N Erg/ J Med 1985;313:1315-1322. 4. Hinkle LE. Thaler HT. Clinical classification of cardiac deaths. Circulation 1982,S:457-464. 5. Mir MA. Effect of a long acting beta-adrenoceptor blocker on diurnal variation of cardiac dysrhythmias. Postgrad Med J 1986,62:175-178. 6. Kudenchuk PJ, Kron J, Walance CG, Murphy ES, Morris CD, Griffith KK, McAnulty JH. Reproducibility of arrhythmia induction with intracardiac electrophysiologic testing: patients with clinical sustained tachyarrhythmias. JACC 1986;7:819-828.
7. McCullagh
P. Regression
models for ordinal data. J R Star Sot Ser E
1980;42:109-142.
8. Fleiss JL. The Design and Analysis of Clinical Experiments. New York: John Wiley and Sons, 1986:83-84. 9. Kantor S, Winkelstein W, lbrihim MA. A note on the interpretation of ridit as a quantile rank. Am J Epidemiol 1968:87:609-615. 10. Rosenblatt G, Zwilling G, Hartmann E. Electrocardiographic changes during sheep in patients with cardiac abnormalities. Psychophjsiology 1969,6:233. 11. Lester BK. Block R, Gunn CG. Wolf S. The relation of cardiac arrhythmias t? phases of sleep. Clin Res 1969;17:456. 12. Shahaway ME. Arrhythmias and the varieties of sleep. N Engl J Med 1970:282:815.
13. Smith R, Johnson L, Rothfeld D, Zir L, Tharp B. Sleep and cardiac arrhyth&as. Arch Intern Med 1972;130:751-753. 14. Low 9, Tykocinski M, Garfein A, Brooks P. Sleep and ventricular premature beats. Circulation 1973;48:691-701. 15. Pickering TG, Goulding L, Cobern BA. Diurnal variations in ventricular ectopic beats and heart rate. Cardiouasc Med 1977;2:1013-1022. 16. Northcote RJ, MacFarlane P, Kesson CM, Ballantyne D. Continuous 24hour electrocardiography in thyrotoxicosis before and after treatment. Am Heart J 1986;112:339-344.
17. Rebuzzi AG, Lucente M, Lanza GA, Coppola E, Manzoli U. Circadian rhythm of ventricular tachycardia. Advances in Chronobiology. New York: Alan R Liss. 1987:(part
B) 153-158.
18. Leach CN, Ruskin JN, Halberg
F. Circadian
variation of frequent human
ventricular ectopy. Fed Proc 1981;40,423. 19. Saito K, Matsuyama K, Niki T, Mori H. Characteristics of non-sustained ventricular tachycardia detected by ambulatory electrocardiography. Jpn Circ J J984:48:421-426.
20. Raeder EA, Hohnloser SH, Graboys TB, Podrid PJ, Lampert S, Lawn B. Spontaneous variability and circadian distribution of ectopic activity in patients with malignant ventricular arrhythmia. JACC 1988;12:656-661. 21. Twidale N, Taylor S, Heddle WF, Ayres BF, Tonkin AM. Morning increase in the time of onset of sustained ventricular tachycardia. Am J Cardiol 1989;64:1204-1206.
22. Michelson EL, Morganroth J. Spontaneous variability of complex ventricular arrhythmias detected by long-term electrocardiographic recordings. Circulation 1980,661:690-695.
23. Anastasiou-Nana MI, Menlove RL, Nanas JN, Anderson JL. Changes in spontaneous variability of ventricular ectopic activity as a function of time in
patients with chronic arrhythmias. Circularion J988;78:286-295, 24. Myocardial infarction community registers: results of a WHO international collaborative study coordinated by the regional oftice for Europe. Public Health in Europe, No. 5. Copenhagen Regional Offlee for Europe (World Health Organiration),
1976:J-232.
25. Kudenchuk PJ, Kron J, Walance C, McAnulty JH. Limited value of programmed electrical stimulation from multiple right ventricular pacing sites in clinically sustained ventricular fibrillation or ventricular tachycardia associated with coronary artery disease. Am J Cardiol 1988,61:303-308. 26. Conroy RTLW, Mills JN. Human Circadian Rhythms. L.ondorr J.A. Churchill, J970:127-135.
27. Lubban bed-rest and 28. Kraemer in Research.
MC, Tredre BE. Renal diurnal rhythms in human subjects during limited activity. J Physiol J964;171:26P-27P. HC, Thiemann S. How Many Subjects? Statistical Power Analysis London
THE AMERICAN
Sage Publications.
JOURNAL
J987:76-83.
OF CARDIOLOGY
JUNE
1, 1990
1357
