s

BLr1

N

0d

Effect of Chest Compressions on Ventricular Activation

Jose Osorio, MDa,*, Derek J. Dosdall, PhDb, Paul B. Tabereaux, MD, MPHa,Robert P. Robichaux, Jr., MD, MPHa, Shannon Stephens, EMTPc, Jeffrey D. Kerby, MD, PhDd,

Ronald E. Stickneye, Stephen Pogwizd, MDa, and Raymond E. Ideker, MD, PhDa,f,g

External mechanical forces can cause ventricular capture and fibrillation (i.e., commotiocordis). In animals, we showed that chest compressions (CCs) can also cause the phenom-enon. The aim of the present study was to determine whether ventricular capture by CCsoccurs in humans. Electronic rhythm strips were analyzed in 31 cases of out-of-hospitalcardiac arrest. The timing of the CCs was identified from the changes in thoracic imped-ance between the defibrillator pads. Ventricular capture was defined as QRS complexes ofsimilar morphology occurring intermittently but synchronized with the CC artifact andimpedance waveform. Only intermittent ventricular capture was identified to avoid mis-classifying constant motion artifacts or intrinsic rhythm as ventricular capture. Of the 29patients who received CCs for >1 minute, minimal or stable motion artifact was present in24. Intermittent ventricular capture was found in 7 of the 24 patients. In the patients withventricular capture, the number of ventricular activations (from ventricular capture andnative beats) was greater during the CCs than when the CCs was not being performed(18 � 8.9 vs 9.7 � 4.0 activations in 15 seconds, p � 0.01). However, in patients withoutventricular capture, they were similar (6.8 � 8.2 vs 7.2 � 8.8 activations in 15 seconds,p � 0.47). Refibrillation occurred in 22 patients; it began during the CCs in 16 and closelyfollowing their initiation in 3. In conclusion, CCs during cardiopulmonary resuscitationcan electrically stimulate the heart. Additional studies evaluating the effect of ventricularcapture on cardiopulmonary resuscitation outcomes, its relation to refibrillation, andmethods to prevent or time ventricular capture by CCs are warranted. © 2012 Elsevier

Inc. All rights reserved. (Am J Cardiol 2012;109:670–674)

Externally applied mechanical forces, such as precordialthumps or chest wall impacts with baseballs (i.e., commotiocordis) can cause ventricular capture and, in certain cases,ventricular fibrillation (VF). This phenomenon has beenextensively studied1–5 and is thought to be modulated by thetretch-activated ion channels.6 We have recently reported

that the mechanical forces generated by chest compressions(CCs) can also cause ventricular capture in both swine andcanines.7,8 Such ventricular capture during CCs occurredafter defibrillation of VF of long duration, which, undercertain conditions, created long–short cycles of activationthat led to refibrillation. Although mechanistically plausible,ventricular capture has not been documented during cardio-pulmonary resuscitation (CPR) in humans. Therefore, theprimary objective of the present study was to search forevidence of ventricular capture by CCs in cases of out-of-hospital sudden cardiac arrest. The secondary objective wasto determine whether refibrillation could be linked to theventricular capture caused by CCs.

Departments of aMedicine, cEmergency Medicine, dSurgery, fBiomedi-cal Engineering, and gPhysiology, University of Alabama at Birmingham,

irmingham, Alabama; bDepartment of Medicine, University of Utah, Saltake City, Utah; and ePhysio-Control, Inc, Redmond, WA. Manuscript

eceived June 13, 2011; revised manuscript received and accepted October0, 2011.

This work is supported by grants HL 66256 and 77881 from theational Institutes of Health, Bethesda, Maryland.

*Corresponding author: Tel: (205) 934-7114; fax: (205) 996-7460.

E-mail address: josorio@uab.edu (J. Osorio).

002-9149/12/$ – see front matter © 2012 Elsevier Inc. All rights reserved.oi:10.1016/j.amjcard.2011.10.024

Methods

Patients who sustained out-of-hospital cardiac arrest inthe Greater Birmingham, Alabama area from January toAugust 2008 were studied. The subjects had been enrolledin the Resuscitation Outcomes Consortium-Cardiac ArrestPopulation-based Registry (ROC-Epistry). The Resuscita-tion Outcomes Consortium is a clinical research networkconsisting of 10 clinical centers in the United States andCanada organized to conduct randomized trials pertaining totrauma and out-of-hospital sudden cardiac arrest. The ROC-Epistry Registry was designed to assess how evidence-basedresuscitation is implemented in community-based practice.The present observational study, based on existing datasources, met the requirements for minimal risk and waiver ofinformed consent. Local institutional review board approvalwas obtained before having access to de-identified data files.Outcomes data were not obtained.

All patients studied were resuscitated in the field usingLifePak 12 and 500 defibrillators (Physio-Control, Red-mond, Washington). These defibrillators can continuouslyrecord electrocardiographic and thoracic impedance wave-forms. The rhythm strips recorded electronically were ob-tained from the defibrillation patches used. Proprietary soft-ware (CODE-STAT, version 7.0, Physio-Control) was usedto analyze the electronic files, retrieved from the defibrilla-tors. The software allows for analysis of both rhythm stripsand thoracic impedance waveforms.

Large or chaotic artifacts produced by CCs can render

rhythm strips obtained during CPR uninterpretable. Fre-

www.ajconline.org

m

671Arrhythmias and Conduction Disturbances/Chest Compressions and Ventricular Activation

Figure 1. Artifact produced by CCs, as initiated (asterisk) and interrupted (A,C) from patient in asystole. (B,D) Same tracings, with superimposition of thoracic

impedance waveform from defibrillator pads (green), which can be used to analyze CC events. Red arrows generated by algorithm mark each CC event.

Figure 2. Rhythm strip showing evidence of ventricular capture (QRS complexes marked with asterisk). (A) Note, similarities in marked complexes. (B) Note,similar timing between impedance (green) and initiation of marked QRS complexes. Note, also, QRS seen after each of first 3 CCs but intermittentlythereafter. Arrows generated by computer algorithm mark each CC event. Dashed circles point to stable relation between QRS complexes, CC artifact, andimpedance waveform. (C) Nine QRS complexes considered ventricular capture from CCs are superimposed. Note, how QRS complexes and impedance

waveform line up.

Figure 3. Example of evidence of ventricular capture. (A) Ventricular capture (QRS complexes marked with asterisk) seen after some CC artifacts. Note,ventricular capture is intermittent but QRS complexes kept a stable relation with impedance waveform (B, green tracing). Arrows generated by algorithm

ark each CC event.

672 The American Journal of Cardiology (www.ajconline.org)

quently, however, a stable, oscillating artifact is seen(Figure 1). In these cases, we searched for evidence ofventricular capture caused by CCs. Periods of pulselesselectrical activity or asystole, either as the initial rhythmor after successful defibrillation, were used for analysis.

The external defibrillators measured and recorded im-pedance continuously by passing a safe, high-frequency (14to 20 kHz), alternating current between the defibrillationpads and measuring the voltage generated. A computeralgorithm identified CC events (Figure 1) using a methodthat has been clinically validated.9,10 The software markseach CC event with a red arrow (Figures 1 to 5).

Intermittent ventricular capture was defined as the pres-ence of QRS complexes of similar morphology, that wereintermittently associated with the CC cycles determinedfrom thoracic impedance, and that exhibited a constant timerelation with the CC artifact seen on the rhythm strip andwith the impedance signal. Only cases of intermittent ven-tricular capture were considered, because of the difficulty in

Figure 4. Example of evidence of ventricular capture (QRS complexes maevery 3 CCs thereafter. (A) Arrows generated by algorithm mark each CC ewaveforms (green). (B) Gain of electrocardiographic signal 5 mm/mV.

Figure 5. (A) Example of VF arising shortly after CCs begun (asterisk midioventricular rhythm allowing third CC to be delivered during repolariza(B) Example from another patient of VF arising during CCs.

differentiating constant ventricular capture in a 1:1 ratio

with the CC from a constant motion artifact or an intrinsicrhythm of coinciding rate. Two groups were defined accord-ing to the presence or absence of ventricular capture byCCs. Ventricular activations were defined as QRS com-plexes that were either intrinsic beats or considered fromventricular capture. Successful defibrillation was defined asthe absence of VF 5 seconds after defibrillation.

In the 24 cases in which CCs produced only minor orstable motion artifacts, the number of ventricular activationsduring a 15-second period while CCs were being performedwas compared to a 15-second period in which CCs were not(immediately before the initiation of CCs when possible). Inthe group of patients in whom ventricular capture wasfound, the first segment in which it was seen was used. In 2cases, however, CCs were started within a few seconds ofsuccessful defibrillation. In these cases, the 15 secondsbefore and after the cessation of CCs were used. In thegroup of patients in whom ventricular capture was notfound, the first segment in which CCs produced motion

th asterisk), seen once for every 2 CCs until seventh capture and once for) Note, stable relation among QRS complexes, CC artifact, and impedance

itiation of VF). Note, patient was in pulseless electrical activity and hadase, after which VF ensued (possible “compression-on-T” phenomenon).

rked wivent. (B

arks intion ph

artifact and the 15 seconds preceding it were used.

ogp

R

3wuali

lmvEt

omoIecccpwtiwi

reipvvaoc

p

Vsrdt

iedl

673Arrhythmias and Conduction Disturbances/Chest Compressions and Ventricular Activation

The number of episodes of refibrillation (in cases inwhich the initial rhythm was VF) was analyzed. In the casesin which the initial rhythm was not VF, episodes of de novoVF were quantified, and for analysis purposes were catego-rized as refibrillation. A standard electrocardiographic gainand sweep speed (10 mm/mV and 25 mm/s, respectively)was used for all the figures shown, unless noted otherwise.

The Statistical Package for Social Sciences was used forstatistical analysis. The Wilk-Shapiro test was used to assessfor normality of the distribution of the data. The number ofventricular activations during a 15-second period was nor-mally distributed. Continuous variables are expressed as themean � SD. The paired t test with equal variances notassumed was used to compare the number of ventricularactivations before and during CCs in each group (with andwithout ventricular capture). An independent samples Stu-dent’s t test was used to compare the change in the numberf ventricular activations before and during CCs in bothroups. Statistical significance was defined as a 2-tailedvalue �0.05.

esults

The rhythm strips and thoracic impedance measurement of1 patients sustaining out-of-hospital sudden cardiac arrestere studied. The initial rhythm was VF or pulseless ventric-lar tachycardia in 21 patients, pulseless electrical activity in 7,nd asystole in 3. In 2 patients, a return of spontaneous circu-ation was achieved without the need for CCs. All the remain-ng patients received CCs for �1 minute.

Of the 29 patients who received CCs, 24 had �1 segmentasting �30 seconds in which the artifact of the CCs wasinimal or had a stable oscillating pattern that allowed

isualization of the underlying rhythm (Figures 1 to 4).vidence of ventricular capture was found in 7 (29%) of

hese 24 patients (Figures 2 to 4).In the patient whose data are shown in Figure 2, evidence

f ventricular capture was seen for the QRS complexesarked with an asterisk. The marked QRS complexes were

f the same morphology and timing in relation to the CCs.n Figure 2, the impedance waveform varied with the CCvents and the downslope in the impedance trace repeatedlyorresponded in time with the beginning of the markedomplexes (dashed circles). Also, in Figure 2, all 9 QRSomplexes identified as ventricular capture are superim-osed to show the common relation with the impedanceaveform and similarities of the QRS morphology. Ven-

ricular capture is seen in a 1:1 ratio in the first 3 CCs andntermittently thereafter, intermixed with 2 QRS complexesith a different morphology and timing in relation to the

mpedance (unmarked beats).In another patient whose data are shown in Figure 3, the

hythm strip displayed a very stable oscillating artifact gen-rated by the CCs, with an interval similar to that of thempedance waveform. The complexes are of similar mor-hology (asterisks mark the ventricular captures). Althoughentricular capture occurred intermittently, a similar inter-al was seen between the nadir of the oscillating CC artifactnd the beginning of the QRS and between the downslopef the impedance tracing and the QRS (Figure 3, dashed

ircles). The ventricular activation was not regular, making n

an intrinsic rhythm less likely. Similarly, in Figure 4, anexample is shown in which each ventricular capture occursin a 2:1 ratio (CC/ventricular capture) for the first 7 activa-tions and a 3:1 ratio thereafter.

In the cases in which ventricular capture was found,ventricular activations (intrinsic or ventricular capture fromCCs) were seen more often during the delivery of the CCs(18 � 9.5 vs 9.7 � 4.0 activations in 15 seconds, p � 0.01).In the cases in which ventricular capture was not found, thenumber of ventricular activations was similar during theCCs and immediately before CC initiation (6.8 � 8.2 vs7.2 � 8.8 activations in 15 seconds, p � 0.47). The meanchange in the number of ventricular activations was signif-icantly greater in the patients with ventricular capture thanin those without it (8.6 � 6.2 vs �0.3 � 2.0, p �0.01).

Refibrillation was seen in 22 patients (76% of those whorequired CCs). In 16 patients, VF started during the CCs,and in 13 we were able to identify the initiation (Figure 5).In 3 of these patients, VF closely followed the initiation ofCCs (within the first 5 compressions).

Discussion

In most patients, the rhythm strips obtained during CPRcontain segments with relatively minor motion artifact or astable, oscillating motion artifact that allows visualization ofthe cardiac rhythm. The major finding of the present study isthat ventricular capture occurred synchronously with CCs in29% of these patients. Although it was not possible to obtainthe definitive evidence of direct cardiac recordings, our resultsprovide strong evidence that the mechanical force generated byCCs in humans can cause ventricular capture.

Mechanoelectrical coupling occurs in different settingswith a wide range of externally applied mechanical forces,from precordial thumps11–14 to chest wall impact with arojectile such as a baseball (commotio cordis).11–13 We

have recently reported ventricular capture caused by CCsduring CPR in a swine model in which ventricular capturewas verified by direct intracardiac recordings.7 In that study,CCs were performed with a pneumatic device, and ventric-ular capture was seen in all pigs, most often in a 1:1 ratio(CC/ventricular capture). In another study of canines, weobserved ventricular capture when CCs were performedwith either a pneumatic device or manually.8

The clinical use of precordial thumps, which is no longeradvised by the current American Heart Association Guide-lines for cardiopulmonary resuscitation,15 illustrates how

F could arise from externally applied mechanical forces,uch as CCs, when delivered during the vulnerable period ofepolarization. This proarrhythmic phenomenon has beenocumented both during attempts to cardiovert ventricularachycardia12 and during complete heart block when the

thump is delivered during repolarization of an escape beat.11

The exact amount of mechanical force needed to cause VF(“lower limits of vulnerability”) in humans is not known.

In our animal model,16 refibrillation closely followed thenitiation of CCs in most episodes. An analysis of thepisodes of refibrillation revealed that ventricular captureuring CCs interacted with the intrinsic rhythm, creatingong–short cycles of activation, a well-recognized mecha-

ism by which VF may be initiated.17,18

mrqntt

1

1

1

2

2

2

2

674 The American Journal of Cardiology (www.ajconline.org)

Only a few studies have investigated the association be-tween CCs and refibrillation. Most recently, Berdowski et al19

studied cases of out-of-hospital CPR and found that VF com-monly recurs within the first few seconds after CPR is initiatedafter successful defibrillation. Their study is the first to dem-onstrate a temporal relation between the onset of CCs and VFrecurrence. The mechanism underlying VF recurrence, how-ever, was not studied. Ventricular capture by CCs is 1 of thepotential explanations for such a finding. It is plausible thatcompression during the vulnerable period of repolarization(“compression-on-T”) or that creates a long–short activationsequence would initiate refibrillation. Conclusive evidence forrefibrillation initiated by CCs was not found in our study.

It is not known whether ventricular capture by CCs duringCPR has an effect on the outcomes. It also remains to bedetermined whether ventricular capture caused by CCs influ-ences cardiac output produced by CPR. The current AmericanHeart Association Guidelines for cardiopulmonary resuscita-tion15 recommend a 1-shock protocol with immediate resump-tion of CCs, minimizing the “no-flow” time, which is known toadversely affect the outcomes.20,21 The current protocol might

ake it possible for CCs to be delivered during an organizedhythm, producing a “compression-on-T” or long–short se-uences that, in theory, could lead to refibrillation. However,o indication was found that this adversely affects resuscita-ion, and increased survival rates have been reported sincehese changes in the guidelines.22

During CPR, VF frequently recurs and is thought to be amarker of a poor outcome.19,23 If the mechanical forcesgenerated by CCs during the vulnerable period are capableof causing VF, timing compressions with respect to intrinsiccardiac electrical activity or with respect to an externalpacing stimuli to prevent long–short sequences or “com-pression-on-T” could prevent refibrillation.

Our study had several limitations. Continuous intracar-diac electrical recordings would need to be acquired whileCCs are performed to obtain definitive evidence of ventric-ular capture by CCs. It was not possible to obtain suchrecordings during out-of-hospital CPR. It is possible that thegreater number of ventricular activations seen during CCs isa result of myocardial perfusion during CPR. The incidenceof ventricular capture caused by CCs might have beenunderestimated if CC and ventricular capture occurred in a1:1 ratio, because such cases were not considered ventric-ular captures in the present study. Patient demographics andoutcomes were not obtained; hence, a correlation betweenventricular capture and survival could not be made.

Acknowledgment: We thank Kate Sreenan for assistance inpreparing the report.

1. Alsheikh-Ali AA, Akelman C, Madias C, Link MS. Endocardial map-ping of ventricular fibrillation in commotio cordis. Heart Rhythm2008;5:1355–1356.

2. Link MS, Wang PJ, Pandian NG, Bharati S, Udelson JE, Lee MY,Vecchiotti MA, VanderBrink BA, Mirra G, Maron BJ, Estes NA III.

An experimental model of sudden death due to low-energy chest-wallimpact (commotio cordis). N Engl J Med 1998;338:1805–1811.

3. Link MS, Maron BJ, Wang PJ, VanderBrink BA, Zhu W, Estes NA III.Upper and lower limits of vulnerability to sudden arrhythmic deathwith chest-wall impact (commotio cordis). J Am Coll Cardiol 2003;41:99–104.

4. Link MS. Mechanically induced sudden death in chest wall impact(commotio cordis). Prog Biophys Mol Biol 2003;82:175–186.

5. Link MS, Estes NA III. Mechanically induced ventricular fibrillation(commotio cordis). Heart Rhythm 2007;4:529–532.

6. Bode F, Franz MR, Wilke I, Bonnemeier H, Schunkert H, WiegandUK. Ventricular fibrillation induced by stretch pulse: implications forsudden death due to commotio cordis. J Cardiovasc Electrophysiol2006;17:1011–1017.

7. Osorio J, Dosdall DJ, Robichaux RP, Tabereaux PB, Ideker RE. In aswine model, chest compressions cause ventricular capture, and bymeans of a long-short sequence, ventricular fibrillation. Circ Arrhyth-mia Electrophysiol 2008;1:282–289.

8. Dosdall DJ, Osorio J, Robichaux RP, Huang J, Ideker RE. Chestcompressions after defibrillation cause ventricular electrical activationin dogs. Presented at the Annual Fall Meeting of the BiomedicalEngineering Society, October 2–4, 2008, St. Louis, MO.

9. Stickney R, Marx R, Olsufka M, Doherty A, McMahon M, Walker C,O’Grady S. Accurate measurement of chest compressions from tho-racic impedance signals. Circulation 2005;112:326.

0. Stecher FS, Olsen JA, Stickney RE, Wik L. Transthoracic impedanceused to evaluate performance of cardiopulmonary resuscitation duringout of hospital cardiac arrest. Resuscitation 2008;79:432–437.

11. Cayla G, Macia JC, Pasquie JL. Images in cardiovascular medicine:precordial thump in the catheterization laboratory experimental evi-dence for commotio cordis. Circulation 2007;115:e332.

12. Kohl P. In: Zipes DP, Jalife J, eds. Cardiac Electrophysiology: FromCell to Bedside. 5th ed. Philadelphia: WB Saunders; 2004:115–125.

13. Maron BJ. Sudden death in young athletes. N Engl J Med 2003;349:1064–1075.

14. Allred JD, Epstein AE, Ideker RE. The sign of four and the SDNN ofone of three. Heart Rhythm 2006;3:287–289.

15. 2005 American Heart Association Guidelines for cardiopulmonaryresuscitation and emergency cardiovascular care. Circulation 2005;112(24 Suppl):IV1–IV203.

16. Hess EP, White RD. Ventricular fibrillation is not provoked by chestcompression during post-shock organized rhythms in out-of-hospitalcardiac arrest. Resuscitation 2005;66:7–11.

17. Viskin S, Fish R, Zeltser D, Belhassen B, Heller K, Brosh D, LaniadoS, Barron HV. Arrhythmias in the congenital long QT syndrome: howoften is torsade de pointes pause dependent? Heart 2000;83:661–666.

8. Kay GN, Plumb VJ, Arciniegas JG, Henthorn RW, Waldo AL. Tor-sade de pointes: the long-short initiating sequence and other clinicalfeatures: observations in 32 patients. J Am Coll Cardiol 1983;2:806–817.

9. Berdowski J, Tijssen JG, Koster RW. Chest compressions cause re-currence of ventricular fibrillation after the first successful conversionby defibrillation in out-of-hospital cardiac arrest. Circ Arrhythm Elec-trophysiol 2010;3:72–78.

0. Berg RA, Sanders AB, Kern KB, Hilwig RW, Heidenreich JW, PorterME, Ewy GA. Adverse hemodynamic effects of interrupting chestcompressions for rescue breathing during cardiopulmonary resuscita-tion for ventricular fibrillation cardiac arrest. Circulation 2001;104:2465–2470.

1. Berg RA, Hilwig RW, Kern KB, Sanders AB, Xavier LC, Ewy GA.Automated external defibrillation versus manual defibrillation for pro-longed ventricular fibrillation: lethal delays of chest compressionsbefore and after countershocks. Ann Emerg Med 2003;42:458–467.

2. Rea TD, Helbock M, Perry S, Garcia M, Cloyd D, Becker L, EisenbergM. Increasing use of cardiopulmonary resuscitation during out-of-hospital ventricular fibrillation arrest: survival implications of guide-line changes. Circulation 2006;114:2760–2765.

3. van Alem AP, Post J, Koster RW. VF recurrence: characteristics and

patient outcome in out-of-hospital cardiac arrest. Resuscitation 2003;59:181–188.