(Circulation. 1999;100:446-449.)
© 1999 American Heart Association, Inc.
Correspondence |
Individual Circadian Variation and Sudden Death
Heart Institute (InCor) University of São Paulo Medical School, São Paulo, Brazil
Professor of Cardiology, Cardiovascular Research and Teaching Institute, OLV Hospital, Aalst, Belgium
 |
Introduction |
|---|
 Top Introduction ReferencesIntroduction |
|---|
To the Editor:
Because the identification of patients with certain circadian trends may permit individualized prevention of cardiac arrest or sudden death, we read the article by Peckova et al with interest.1 We were, however, puzzled by the investigators' results emphasizing that the phenomenon of individual circadian variation for sudden cardiac death was not present. We believe there are important caveats to heed when one interprets these results.
In 1995, when analyzing the distribution of sudden death due to ventricular fibrillation (VF) aborted by implanted defibrillators (ICDs), we reported a clear tendency for shocks to occur in the morning hours in a group of 22 patients.2 3 Interestingly, 16 patients who received multiple appropriate shocks showed a trend for the repeated shocks to occur around the same period of the day that the initial shock occurred.2 Two patients each experienced 6 shocks that occurred within the same 3-hour period of the day. This trend was noted even when subsequent shocks occurred as long as 1 year after the initial shock. Such findings strongly suggested that an individual circadian variance in sudden death due to VF could exist in patient subgroups.
Our results, however, were limited by our small study group. Peckova et al have offered us a large cohort of patients with their analysis of temporal variation in 6603 out-of-hospital cardiac arrests and suggested that women formed the sole group to demonstrate any significant similarity in the time of day the first and second arrests occurred. However, some characteristics in Peckova's study group might have concealed individual circadian variance.
Namely, Peckova's data consider all arrhythmias, regardless of the type. Therefore, it is not clear whether patients with recurrent cardiac arrest had the same type of arrhythmic event during the first and subsequent episodes. Different arrhythmic events have different circadian variations.1 We believe this fact necessitates that analysis should contemplate only patients in whom the cause is the same for the first and subsequent arrests.
Additionally, the authors included data acquired by rescue services. Generally, patients without an ICD experience a limited number of cardiac arrests. Data obtained from the ICD are incomparably more precise than data acquired by rescue services. Therefore, data obtained from ICDs are the best way to identify individual circadian rhythms for arrhythmic sudden death.
Because of these limitations, we believe the issue of individual circadian variance for arrhythmic death still begs for further analysis and continued discussion.
|
References |
|---|
|
TopIntroductionReferencesIntroduction |
|---|
1. Peckova M, Fahrenbruch CE, Cobb LA, Hallstrom AP. Circadian variations in the occurrence of cardiac arrests: initial and repeat episodes. Circulation. 1998;98:31–39.
2.
d'Avila A, Wellens F, Andries E, Brugada P. At what
time are implantable defibrillator shocks delivered? Evidence for
individual circadian variance in sudden cardiac death. Eur
Heart J. 1995;16:1231–1233.
3.
Boyle NG, Josepheson ME. Sudden cardiac death,
circadian rhythms and defibrillators. Eur Heart J. 1995;16:1162–1164. Editorial comment.
Response
Department of Biostatistics
Division of Cardiology, University of Washington, Seattle, Wash
|
Introduction |
|---|
|
TopIntroductionReferencesIntroduction |
|---|
The letter from Drs Avila and Brugada is interesting, and we certainly agree with their conclusion that the issue of individual circadian variation for arrhythmic death begs for further analysis and continued discussion. For the patients we studied with recurrent cardiac arrest, the rhythm when the emergency medical services team dispatched for the first cardiac arrest arrived was ventricular fibrillation. It is therefore reasonable to assume that the initial rhythm was ventricular fibrillation or ventricular tachycardia degenerating to ventricular fibrillation. However, we cannot know with certainty what the initial rhythm was for the second arrests.
It is relevant to note that there is a difference between shocks by an
implantable defibrillator and sudden death. Most postimplant studies
report at least 50% of patients have received 
1 shock, whereas
sudden death rates at 1 year in survivors of cardiac arrest in the
predefibrillator era are reported in the 15% range. Another potential
point of difference between our patient population and the 16 patients
Avila and Brugada describe may be related to the type of
arrhythmia. It appears from their article that perhaps half
were treated with an ICD because of an episode of
ventricular tachycardia that resulted in
syncope or compromising symptoms.
It is only fair to point out that Drs Avila and Brugada found clustering of arrhythmias in only 5 of 16 patients with multiple shocks. There are insufficient data in their article to try to calculate the probabilities of this occurring by chance, but the probability is not zero. Nevertheless, we don't dispute that some of those 5 surely do have clustering over and above that inherent in the global circadian variation. We suspect that there are individual patients in our sample who also might have clustering, but to determine clustering in a single individual would require many more than 2 episodes, and, sadly, few patients survive >2 events.
We should clarify that our analysis of a clustering pattern of arrests within individuals was not based on 6603 patients but on fewer than 500 patients who had an initial and subsequent cardiac arrest. However, that number is sufficient that we had a power of 0.8 of rejecting the null if 5% of our patients had clustering (against a traditional 2-sided 0.05 level test of the null) and a power exceeding 0.99 of rejecting the null if 30% of our patients had clustering.
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||