The effect of intravenous adrenaline on electrocardiographic changes during resuscitation in patients with initial pulseless electrical activity in out of hospital cardiac arrest

Clinical paper

The effect of intravenous adrenaline on electrocardiographic changes during resuscitation in patients with initial

pulseless electrical activity in out of hospital cardiac arrest

Gunnar Waage Skjeflo

*, Eirik Skogvoll

,

Jan Pa˚l Loennechen

, Theresa Mariero Olasveengen

, Lars Wik

, Trond Nordseth

aDepartmentofCirculationandMedicalImaging,FacultyofMedicine,NTNU,NorwegianUniversityofScienceand Technology,Trondheim,Norway

bDepartmentofAnesthesiology,NordlandHospital,Bodø,Norway

cSt.OlavsHospital,TrondheimUniversityHospital,DepartmentofAnesthesiaandIntensiveCareMedicine,Trondheim, Norway

dSt.OlavsHospital,TrondheimUniversityHospital,DepartmentofCardiology,Trondheim,Norway

eDepartmentofAnesthesiology,OsloUniversityHospital,Oslo,Norway

fNorwegianNationalAdvisoryUnitonPrehospital EmergencyMedicine(NAKOS),OsloUniversityHospital, Oslo, Norway

gStOlavsHospital,TrondheimUniversityHospital,DepartmentofEmergencyMedicineandPre-HospitalServices, Trondheim,Norway

Abstract

Introduction:Presence of electrocardiographicrhythmintheabsenceof palpable pulsesdefines pulselesselectricalactivity(PEA) andthe electrocardiogram(ECG)mayprovideasourceofinformationduringresuscitation.TheaimofthisstudywastoexaminethedevelopmentofECG characteristicsduringadvancedlifesupport(ALS)fromOut-of-hospitalcardiacarrest(OHCA)withinitialPEA,andtoexplorethepotentialeffectsof adrenalineonthesecharacteristics.

Methods:PatientswithOHCAandinitialPEA,partofrandomizedcontrolledtrialofALSwithorwithoutintravenousaccessandmedications,were included.Atotalof4840combinedobservationsofQRScomplexrate(heartrate)andwidthweremadebyexaminingdefibrillatorrecordingsfrom 170episodesofcardiacarrest.

Results:WefoundIncreasedheartrate(47beatsperminute)andreducedQRScomplexwidth(62ms)duringALSinpatientswhoobtainedreturnof spontaneouscirculation(ROSC);whilepatientswhoreceivedadrenalinebutdiedincreasedtheirheartrate(22beatsperminute)withoutany concomitantdecreaseinQRScomplexwidth.

Conclusion:ECGchangesduringALSincardiacarrestwereassociatedwithprognosis,andthe administration ofadrenalineimpacted onthesechanges.

Keywords:Pulselesselectricalactivity,PEA,Electrocardiographic,ECG,Adrenaline,Epinephrine,Cardiacarrest,Advancedlifesupport

* Correspondingauthorat:DepartmentofCirculationandMedicalImaging,FacultyofMedicine,NorwegianUniversityofScienceandTechnology,NO- 7491Trondheim,Norway.

E-mailaddress:[email protected](G.W.Skjeflo).

https://doi.org/10.1016/j.resuscitation.2019.01.021

Received16July2018;Receivedinrevisedform21December2018;Accepted20January2019

0300-9572/©2019TheAuthor(s).PublishedbyElsevierB.V.ThisisanopenaccessarticleundertheCCBY-NC-NDlicense(http://creativecommons.org/

licenses/by-nc-nd/4.0/).ThisisanopenaccessarticleundertheCCBY-NC-NDlicense(http://creativecommons.org/licenses/by-nc-nd/4.0/).

Availableonlineatwww.sciencedirect.com

Resuscitation

j our na lho me pa g e :ww w. e l s e v i e r. c om/ l o ca t e / re s usc i ta t i on

Introduction

Pulselesselectricalactivity(PEA),definedasanorganizedelectro- cardiographic rhythm in the absence of palpable pulses, is the presentingrhythminaboutonequarterofpatientswithoutofhospital cardiacarrest(OHCA).^1–4Theelectrocardiogram(ECG)reflectsthe electric function of the myocardium.⁵ Changes in the ECG may representasourceofinformationduringtheprovisionofadvancedlife support(ALS)tosuchpatients.

Inanobservationalstudyofin-hospitalcardiacarrest(IHCA),wefound anincrease inPEA heart rate and agradual narrowing of theQRS widththe last6–12minutes(min)beforereturnofspontaneouscirculation(ROSC).⁶ These findings may have prognostic importance,as well as raising questionsabouttheelectromechanicalpropertiesoftheheartduringPEA.

Administrationof adrenaline (epinephrine) has been shown to increase the proportion of patients who achieve ROSC without increasinglongtermsurvival.^4,7,8Inonerecentrandomizedcontrolled trialofadrenalineversusplaceboinOHCAalargerproportionwith ROSCwasfoundintheadrenalinegroup,butlongtermsurvivalwas alsohigherintheadrenalinegroup.⁹Someauthorshavearguedthat adrenalinetopatientswithnon-shockablerhythms(asystoleandPEA) is as essential as defibrillation to shockable rhythms (ventricular fibrillation—VFandpulselessventriculartachycardia—VT).^10,11In cardiacarrestthepresumedmaineffectofadrenalineistoimprove coronaryperfusionpressurebyincreasingaorticdiastolicpressure.¹²A generaleffectofadrenalineisanincreaseinheartrate.¹³Despitebeing aroutinedrugincardiacarrest,theimmediateeffectofadrenalineon ECGcharacteristicsduringALShasnotbeenstudied.Wehypothe- sizedthatpossiblebeneficialeffectsofadrenalinecouldbeassociated withchangeinheartrate.Theaimofthisstudywastoinvestigatethe developmentofECGcharacteristicsduringALSinpatientswithOHCA andinitialPEA,andtheeffectofadrenalineonthesecharacteristics.

Material and methods

ArandomizedcontrolledtrialwasconductedattheOsloEmergency MedicineServicesbetween2003and2008inadultOHCApatients.

PatientswererandomizedtoALSwithandwithoutintravenousaccess anddrugadministration.MainoutcomeswereROSCandsurvivalto hospitaldischarge.^4,14PatientsfromthisstudywithinitialPEAwere includedinthecurrentsub-groupanalysis.Electronicsignaldatafrom Lifepak12defibrillators(LP12,PhysioControl,Medtronic,Redmond, WA,USA),clinicaldatafromUtsteinstylecardiacarrestformsanddata fromhospitalrecordsacquiredfortheoriginalstudywereanalysed.⁴

Electrocardiographiccharacteristics

ECG and transthoracic impedance (TTI) data from LP 12 were analysedusingMatlab(R2017b,MathWorksInc.,Natick,MA),and annotatedintermsofclinicalstates(Asystole,PEA,VF/VT,ROSC)as describedinapreviouspublication.¹⁵

TheECGandTTIsignalswereplottedagainsttime,QRS-rates (heart rate) and -widths were measured during pausesin chest compressionsforanyreason(includingat endofeffortsorwhen sustained ROSC was obtained), as long as measurable QRS complexes existed. The QRS width was defined as the interval

betweentheinitialdeflectionfromthebaselinetowardstheQ-orR- waveandthebeginningoftheST-intervalontheECG.Incaseswhen noobvioustransitionfromtheQRStotheST-intervalcouldbeseen, thepointwheretheECGtracingcrossedthebaselinetowardstheT- wavewasconsideredtheQRSend-point,asdescribedpreviously.⁶ EachQRSwidthmeasurementwascoupledtoaninstantaneousheart rate, calculated from the distancebetween theQRS complex in questionandthesucceedingQRScomplex.

Statisticalanalysis

PatientsweregroupedbyROSCstatus,theROSCgroupcomprising all patients with ROSC at hospital admission; and by whether intravenous adrenaline was actually given. We disregarded the intention-to-treatstatusintheoriginalstudy,asthebiologicaleffectof adrenalinewasofmaininterestinthissubgroupanalysis.

WeexpectedthatheartrateandQRSwidthwouldbecorrelated;

thiswastestedusingPearson’sproductmomentcorrelation.

First,weinvestigatedthecombinedchangeinheartrateandQRS widthusingbivariateanalysisofvariance(MANOVA)¹⁶accordingto ROSC andadrenaline. Herewe employed thefirst and last15s averagedheartrateandQRSwidthineachpatient.Tosatisfythe requirementofbivariatenormalityandhomoscedasticity,weapplieda square roottransformation after addingaconstant term to avoid negativevalues.Q–Qplotsweresatisfactory.

Second,wemodelledthe15-sfinalaveragedheartrateandQRSwidth separatelyusingalinearmodel¹⁷usingtheinitial15-sheartrateorQRS widthascontinuouscovariates,andROSCandadrenalineasfactors.

Third, to visualize the average development of the ECG characteristics forthelast12min precedingROSCortermination ofALSefforts,wefittedadditivemixedeffectsmodels¹⁸ofheartrate andQRSwidthineachgroup;withtimeasthefixedeffectcovariate, andpatientidentityasrandomeffect.Wespecifiedautocorrelated residuals(whichimprovedmodelfit)asweexpectedthatmeasure- ments closer to each other in time would be more similar than measures further apart.The additive mixed models fit penalized regressionsplinestothedata,byaprocessofcrossvalidation.¹⁸

AllmeasurementsofheartrateandQRSthatweremadeatorlater than12minbeforeROSCorendofALSwereincludedinthismodel, irrespectiveoftiming.Noattemptsweremadetobalancedatawith respecttonumberofortimingofmeasurementsbetweenpatients.

The software R version 3.4.3,¹⁹ running in RStudio version 1.1.419,withthepackagesmcgv,nlmeandellipse,andthesoftware Stata,²⁰wereutilizedforthestatisticalanalyses.Ap-valuelessthan 0.05wasconsideredtoindicatestatisticalsignificance.

Ethics

TheRegionalCommitteesforMedicalandHealthResearchEthics approvedthestudy.Theoriginalstudywasregisteredatclinical-trials.

govwithidentifierNCT00121524.

Results

Two-hundredandthirty-threepatientshadinitialPEAintheoriginal study, of whom170 patients (73%)were includedin the current analysis. Patients were excluded due to missing defibrillator file (n=59) or that defibrillator files were illegible (n=4). ROSC at admissiontohospitalwaspresentin41patients(24%)andadrenaline

wasadministeredto101(59%)oftheincludedpatients.Atotalof 4840combinedobservationsofQRSrateandwidthweremadefrom thedefibrillatorfiles fromtheincludedpatients.Demographic and clinicaldataarepresentedinTable1.Thegroupsvariedinsize:the‘no ROSC’groupswerelarger.Otherwise,therewasanotabledifference indurationofALSbetweenthe‘noadrenalineROSC’groupandall othergroups, withshorter durationof ALSinthisgroup. The‘no adrenalineROSC’groupalsoconsistedofallbutonemale.

ChangesinheartrateandQRScomplexwidthfromstartto endofALS

Heartrate andQRSwidthwerefoundtobe negativelycorrelated (Pearson’sr: 0.35,p<0.0001).Bivariateanalysisofthecombined changeofheartrateandQRSwidthfromthebeginningtotheendof ALSshowedthatthesevariablesweresignificantlyassociatedwith both whether ROSC was obtained and whether adrenaline was administered(p<0.001forboth).Therewasnoevidenceofinteraction betweenROSCandadrenalinestatus(p=0.86).Univariateanalysisof meanfinalheartrateorQRSwidthseparatelyshowedthatthefinal mean heart rate was dependent on the mean initial heart rate (coefficient0.28,p=0.01), ROSC(46.6bpmincreasewithROSC, p<0.0001), and adrenaline (21.7bpm increase with adrenaline, p<0.0001).FinalmeanQRSwidthdependedonmeaninitialQRS width(coefficient0.45,p<0.0001)andROSC(62mslesswithROSC, p<0.0001),butnotadrenaline(p=0.4).Therewasnoevidenceof interactionbetweenROSC andadrenalinestatus in theunivariate analyses(p=0.8and0.72forheartrateandQRSwidthrespectively).

TheobservedmeanchangesinheartratesandQRSwidthsare illustratedasarrowsinFig.1.TheindividualmeasurementsofQRS widthsandheartratesatthebeginning,during,andattheendofALS are presented as bivariate scatterplots in Fig. 1; there was considerablevariationinthesemeasurements.

TimecourseofheartrateandQRScomplexwidthduringthe last12minofALS

TheexpectedheartrateandQRSwidthduringthelast12minofALS beforeROSCorendofALSeffortsarepresentedinFig.2,basedon predictionsfromtheadditivemixedmodels.Inboththe‘adrenaline’and the‘noadrenalineROSC’groups,amarkedriseinheartrateoccurred between3–6minbeforeROSC.Heartrateincreasedslightlyinalinear fashion inthe‘adrenalinenoROSC’group,butwasunchangedtowards the endofALSeffortsin the‘noadrenalinenoROSC’group.We observed a sharp decrease in QRS width the last 6min in the

‘adrenalineROSC’group.Inthe‘noadrenalineROSC’groupamore gradualnarrowingofQRSwidthsoccurredduringthelast12minof ALS.Inbothno-ROSCgroupsQRSwidthincreasedslightly.

Discussion

Toourknowledge,thisisthefirststudytoexaminechangesinECG characteristicsduringALSinpatientswithOHCAandinitialPEAfor the duration ofALS. It isasecondary analysis of a randomized controlledtrialofintravenousaccessversusnointravenousaccessin

Table1–Demographicandclinicaldata,stratifiedonROSCandadrenalinestatus.(n=number,yrs:years,min:

minute,IQrange:Interquartilerange,mg:milligram).

Returnofspontaneouscirculation (n=41)

Noreturnofspontaneouscirculation (n=129)

Adrenaline (n=29)

Noadrenaline (n=12)

Adrenaline (n=72)

Noadrenaline (n=57)

Age(yrs),median(IQrange) 61(56–75) 65(62–75) 78(56–83) 77(65–85)

Males,n(%) 15(52) 11(92) 47(36) 34(60)

Location

Home,n(%) 20(69) 5(42) 41(57) 41(72)

Public,n(%) 5(17) 5(42) 20(28) 11(21)

Work,n(%) 0(0) 1(8) 1(1) 0(0)

Other,n(%) 4(14) 1(8) 10(14) 5(9)

Witnessedbylayperson,n(%) 19(66) 8(67) 31(43) 38(67)

Witnessedbyparamedic,n(%) 6(21) 1(8) 28(38) 6(11)

BystanderCPR,n(%) 12(41) 3(25) 26(36) 27(47)

Responsetime(min),median(IQrange) 7.9(5.3–9.0) 10.6(6.3–11.4) 6.1(0–8.7) 9.6(6.6–11.3) DurationofALS(min),median(IQrange) 20(12.5–28.1) 6.7(5.2–11.5) 26.5(19.1–31.9) 21.4(14.4–28.9) Compressionrate(/min),median(IQrange) 117(112–120) 115(112–126) 119(112–126) 112(107–120) Handsoffratio,median(IQrange) 0.15(0.1–0.26) 0.18(0.12–0.28) 0.17(0.12–0.25) 0.20(0.13–0.29)

Defibrillationatleastonce,n(%) 4(14) 0(0) 24(33) 7(12)

Intubated,n(%) 27(93) 9(75) 62(86) 45(79)

Intravenousaccess,n(%) 29(100) 8(67) 71(99) 7(12)

Adrenalinedose(mg),median(IQrange) 2(1–3) 0(0) 3(2–5) 0(0)

Atropine,n(%) 8(28) 0(0) 35(49) 2(4)

Amiodarone,n(%) 0(0) 1(8) 9(13) 0(0)

Admittedtohospital,n(%) 29(100) 12(100) 22(30) 11(19)

Dischargedfromhospitalalive,n(%) 1(3) 3(25) 0(0) 1(2)^a

Missingdata:witnessed(n=1),age(n=1),adrenalinedose(n=19),durationofALS(n=1),IVaccess(n=1).

aOnepatientwithoutpre-hospitalROSCbutadmittedtohospitalunderALSsurvivedtohospitaldischarge.

ALS,whereadrenaline was onlyadministered to theintravenous accessgroup.

WediscoveredthatpatientswhoobtainedROSChadincreased heartratesanddecreasedQRSwidthsbeforeROSC,incontrastto

patientswhoweredeclareddeadonscene.Patientsgivenadrenaline hadalargermeanchangeinheartrate,butasimilarchangeinQRS widthcomparedtopatientswhodidnotgetadrenaline,bothinthe ROSCandnoROSCgroups.

Fig.1–BivariatePlotofHeartrateandQRSwidth,ms:milliseconds,min:minutes. Theheartrateaxis waslog transformedforvisualizationpurposes.

Topfigure:adrenalinegroup,lowerfigureno-adrenalinegroup.

Thecircles,dotsandplussignsrepresentindividualmeasurementsofheartrateandQRSwidthatthebeginning, duringandattheendofALS.ThesearenotgroupedbyeitherpatientorROSCstatus,butillustratethevariabilityinthe individualmeasurements.

ThearrowsrepresentmeanchangefromstartofALS(baseofarrow)toendofALS(tipofarrow).Greenarrows representtheROSCgroups,greyarrowsrepresentthenoROSCgroups.

Thegreenellipsesrepresentthe50,75and90%coverageareasfortheROSC-groups’endpoints.

Overall,thepatternofincreaseinheartrateandnarrowingofQRS widthintheROSCgroupsareconsistentwiththefindingsofourearlier studyofin-hospitalcardiacarrestwithinitialPEA.⁶

TheincreaseinheartrateintheROSCgroupsoccurredmainly duringthelast3–6minbeforeROSCwasobtained.Inpatientswho received adrenaline, the QRS narrowed simultaneously with the increaseinheartrate,whileQRSwidthdecreasedmoregraduallyin the‘noadrenalineROSC’group.Interestingly,inpatientswhodidnot obtainROSC;thosewhoreceivedadrenalinedifferedmarkedlyfrom thosewhodidnot,inthatheartrateincreasedintheadrenalinegroup whileremainingunchangedintheno-adrenalinegroup.

ThechangeofheartrateandQRSwidthintheROSCgroupsinthis studywerechangestowardsmorenormalvalues.Thisisinlinewith otherstudiesofOHCAthatessentiallyhavefoundahigherprevalence of normalinitial ECG characteristics in survivors presenting with PEA.^21–23

Themainlybroadcomplexeswithoutdetectableatrialactivityseen inthisstudyweremostlikelyofventricularorigin.Subendocardial Purkinje cells have been implicated in arrhythmogenesis after myocardial infarction, and has been observed to function as pacemakersin thedamagedheart in dogs.²⁴ Ithas beenshown that in ventricular Purkinje cells that surviveacute ischemia, the

resting membrane potential is less negative, with increased automaticityandprolongedactionpotentials.²⁵Further,adrenaline hasbeenshowntoincreaseactionpotentialgenerationinthesecells, aneffectthatwasabolishedusingabeta-blockingdrug.²⁶

ThoughheartrateandQRSwidthwereinverselycorrelatedinthis study, the development over time differed. Both these variables dependon theelectrical functionoftheheart,andare somewhat correlatedinhealthyhumansaswell,thoughtheexactmechanisms areunclear.²⁷BasedonourresultswespeculatethattheQRSwidth more closely reflects the underlying physiological state of the myocardium,butthattheheartmaybeabletorespondwithincreased heartratetoadrenalineeveniftheunderlyingmetabolicstateofthe myocardium is notimproved. This couldtheoretically explainthe increaseinheartrateandQRSwidthseeninthe‘adrenalinenoROSC’ group.ThedifferentpatternofQRSwidthnarrowingintheROSC groupsmaybeduetoamoresuddenimprovementinmyocardialstate inthe‘adrenalineROSC’group,perhaps,againtheoretically,bya rapidadrenalinemediatedincreaseincoronaryperfusionpressure.

Based on theoccurrence ofROSC in theadrenaline and no adrenaline groups, a larger number of survivors to hospital dischargein theadrenaline groupwould be expected. However, anincreasedproportionofpatientswithROSCbutalesserorno Fig.2–Developmentofheartrate(top)andQRSwidth(bottom)forthelast12minofALS,i.e.thelast12minbefore sustainedROSC(green),orbeforeALSeffortswerestopped(grey/Noresponse),accordingtotheadditivemixed effectsmodel.Theadrenalinegroupisattheleft,theno-Adrenalinegroupisattheright.Dashedlines:95%confidence intervals.(ms:milliseconds.min:minutes.ROSC:returnofspontaneouscirculation”).

increase in the proportion of survivors in the long term has repeatedlybeenshowninbothobservationalstudiesandrandom- izedclinicaltrialsofadrenaline inALSwithundifferentiatedinitial rhythms.^7–9Thiswasalsothemainresultofthestudyforwhichthe data analysed in the currentstudy was gathered.⁴ A numberof factorsmaycontributetothis,oneofwhichisapossibledetrimental effectof adrenalineon longterm myocardialfunction. In animal models,beta-adrenergicstimulationoftheheartduringcardiopul- monaryresuscitation (CPR)has beenshowntoincreaseoxygen consumptionwithoutimprovingoxygensupply.^28,29

Adrenalineseemstoincreasethetimewindowwhereitispossible toobtainROSC.¹⁵ThedifferencebetweentheshortdurationofALSin the‘noadrenalineROSC’groupandthelongerdurationofALSinall othergroupsinthecurrentstudyreflectsthis.Itispossiblethatthis increased timewith low flow, orCPR dependent flow,results in irreparabledamagetothebrainandotherorgansinsomepatients.

The objective of this study was to explore the general developmentofheartrateandQRS-widthduringALSinpatients withPEAandtheimpactofadrenalineonthesedevelopments.But such overall trends cannot immediately be applied in decision makingwhenprovidingcaretoindividualpatients.Caveatinmind, the clinical application of these findings may be to pay more attentiontothedevelopmentofheartrateandQRSwidthduring ALS. Narrowing of QRS width and increase in heart rate is a possible marker of an overall positive clinical development that couldmotivatecontinuingALSefforts.Itispossiblethatabsenceof thesechangescouldbeutilizedasfeedbackduringALS,prompting improvementsintheALSeffortsorapplyingconcurrenttreatment modalities(e.g. fibrinolysis)to improvethe myocardialstateand thusincreasingtheprobabilityofROSC.AnincreaseinQRSwidth seemstobeanegativeprognosticsign,evenifheartrateincreases after adrenaline administration. The time dependent effects of adrenalinecouldnotbeassessedinthisstudy;thus,wedonotknow ifQRSwidthorheartrateincreasedfirstintheadrenalinenoROSC group,orifthedevelopmentofQRSwidthchangedinanywayafter adrenaline administration. Any such pattern could have clinical impactandthetemporalrelationbetweenadrenalineadministration, heartrateincreasesandQRSwidthdevelopmentisofgreatinterest, andshouldbestudiedfurther.

Limitations

Thenumberofpatientsincludedisrelativelylowandcomprised only one pre-hospital emergency response system, potentially limitinggeneralizability. Whetheror notagivenpatient present withPEAorasystolemaydependonseveralcircumstancesand patients with initial asystole was not included in this analysis.

Unfortunately, the exacttime during ALSwhen adrenaline was administered is not known, as records could not be kept with enough detail in the EMS system. Thus, the immediate time dependenteffectsofadrenalinecouldnotbeexamined.Demon- strationofatimedependencebetweenadrenalineadministration andECGchangeswouldmaketheargumentthatthedifferences betweenthe adrenalineand noadrenaline groupsweredue to adrenalinealonestronger.

SomeuncertaintyappliestothemeasurementsofQRSwidths, becauseofthesometimesaberrantmorphologyobservedintheQRS complexes.AconsistentapproachtothemeasurementofQRSwidth as described in the methods section was utilized to reduce this uncertainty.

Conclusion

InpatientswithOHCAandinitialPEAwhoobtainedROSC,heartrates increasedandQRSwidthsdecreasedduringALS.Inpatientswhodid nothaveROSC,heartratedecreasedandQRSwidthincreased, exceptinpatientswhodidgetadrenaline.InthisgroupQRSwidthalso increased,butheartrateincreased.

AbsenceofdecreaseinQRSwidthduringALSmaybeapoor prognosticfactorinOHCAwithinitialPEA.

Funding

TheLiaisonCommitteebetweentheCentralNorwayRegionalHealth Authority (RHA) and the Norwegian University of Science and Technology(NTNU).

Conflict of interest

GunnarWaageSkjeflo,JanPålLoennechen,EirikSkogvoll,Theresea MarieroOlasveengenandTrondNordsethdeclarenoconflictofinterest.

LarsWik:NAKOSrepinMABPhysioControl,Prinicpalinvestigator forCIRC,LUCAS2ADstudy.Patentholderofpatentslicensedto ZOLLandPhysioControl.

Acknowledgements

WewishtothankthemembersoftheALSsysteminOslo,Norway,for thecollectionofdatafortheoriginalstudy.Also,wewishtothankthe EasternNorwayRegionalHealthAuthority,OsloUniversityHospital, NorwegianAirAmbulanceFoundation,LaerdalFoundationforAcute Medicine,andtheAndersJahresFundforfundingtheoriginalstudy.

REFERENCES

1.KuismaM,RepoJ,AlaspääA.Theincidenceofout-of-hospital ventricularfibrillationinHelsinki,Finland,from1994to1999.Lancet 2001;358:473–4.

2.CobbLA,FahrenbruchCE,OlsufkaM,CopassMK.Changing incidenceofout-of-hospitalventricularfibrillation,1980-2000.JAMA 2002;288:3008–13.

3.HerlitzJ,AnderssonE,BångA,etal.Experiencesfromtreatmentof out-of-hospitalcardiacarrestduring17yearsinGöteborg.EurHeartJ 2000;21:1251–8.

4.OlasveengenTM,SundeK,BrunborgC,ThowsenJ,SteenPA,WikL.

Intravenousdrugadministrationduringout-of-hospitalcardiacarrest:

arandomizedtrial.JAMAJAmMedAssoc2009;302:2222–9.

5.MeekS,MorrisF.Introduction.I—leads,rate,rhythm,andcardiac axis.BMJ2002;324:415–8.

6.SkjefloGW,NordsethT,LoennechenJP,BergumD,SkogvollE.ECG changesduringresuscitationofpatientswithinitialpulselesselectrical activityareassociatedwithreturnofspontaneouscirculation.

Resuscitation2019;127:31–6Availablefrom:https://www.

sciencedirect.com/science/article/pii/S030095721830159X.[Cited4 April2018].

7.JacobsIG,FinnJC,JelinekGA,OxerHF,ThompsonPL.Effectof adrenalineonsurvivalinout-of-hospitalcardiacarrest:arandomised double-blindplacebo-controlledtrial.Resuscitation2011;82:1138–43.

8.HagiharaA,HasegawaM,AbeT,NagataT,WakataY,MiyazakiS.

Prehospitalepinephrineuseandsurvivalamongpatientswithout-of- hospitalcardiacarrest.JAMA2012;307:1161–8.

9.PerkinsGD,JiC,DeakinCD,etal.Arandomizedtrialofepinephrinein out-of-hospitalcardiacarrest.NEnglJMed2018;379:711–21.

10.KheraR,ChanPS,DonninoM,GirotraS.Hospitalvariationintimeto epinephrinefornonshockablein-hospitalcardiacarrestclinical perspective.Circulation2016;134:2105–14.

11.DonninoMW,SalciccioliJD,HowellMD,etal.Timetoadministrationof epinephrineandoutcomeafterin-hospitalcardiacarrestwithnon- shockablerhythms:retrospectiveanalysisoflargein-hospitaldata registry.BMJ2014;348:g3028.

12.ParadisNA,WenzelV,SouthallJ.Pressordrugsinthetreatmentof cardiacarrest.CardiolClin2002;20:61–78viii.

13.StruthersAD,ReidJL,WhitesmithR,RodgerJC.Effectofintravenous adrenalineonelectrocardiogram,bloodpressure,andserum potassium.BrHeartJ1983;49:90–3.

14.OlasveengenTM,WikL,SundeK,SteenPA.Outcomewhen adrenaline(epinephrine)wasactuallygivenvs.notgiven—posthoc analysisofarandomizedclinicaltrial.Resuscitation2012;83:327–32.

15.NordsethT,OlasveengenTM,KvaløyJT,WikL,SteenPA,SkogvollE.

Dynamiceffectsofadrenaline(epinephrine)inout-of-hospitalcardiac arrestwithinitialpulselesselectricalactivity(PEA).Resuscitation 2012;83:946–52.

16.JohnsonR,WichernD.Appliedmultivariatestatisticalanalysis.5thed.

UpperSaddleRiver:Prentice-Hall;2002.

17.VickersAJ,AltmanDG.Analysingcontrolledtrialswithbaselineand followupmeasurements.BMJ2001;323:1123–4.

18.WoodS.Generalizedadditivemodels—anintroductionwithR.1sted.

BocaRaton:Chapmann&Hall/CRCTextsinStatisticalSience;2006.

19.RCoreTeam.R:alanguageandenvironmentforstatistical computing. Availablefrom:.Vienna,Austria:RFoundationfor StatisticalComputing;2015.https://www.R-project.org/.

20.StataCorp.statastatisticalsoftware:release15.CollegeStation,TX:

StataCorpLLC;2017.

21.AufderheideTP,ThakurRK,StuevenHA,etal.Electrocardiographic characteristicsinEMD.Resuscitation1989;17:183–93.

22.StuevenHA,AufderheideT,ThakurRK,HargartenK,VanagsB.

Definingelectromechanicaldissociation:morphologicpresentation.

Resuscitation1989;17:195–203.

23.WeiserC,PoppeM,SterzF,etal.Initialelectricalfrequencypredicts survivalandneurologicaloutcomeinoutofhospitalcardiacarrest patientswithpulselesselectricalactivity.Resuscitation

2018;125:34–8.

24.FriedmanPL,StewartJR,WitAL.Spontaneousandinducedcardiac arrhythmiasinsubendocardialpurkinjefiberssurvivingextensive myocardialinfarctionindogs.CircRes1973;33:612–26.

25.FriedmanPL,StewartJR,FenoglioJJ,WitAL.Survivalof subendocardialPurkinjefibersafterextensivemyocardialinfarctionin dogs:invitroandinvivocorrelations.CircRes1973;33:597–611.

26.PappanoAJ,CarmelietEE.Epinephrineandthepacemaking mechanismatplateaupotentialsinsheepcardiacPurkinjefibers.

PflugersArch1979;382:17–26.

27.MasonJW,BadiliniF,VaglioM,etal.Afundamentalrelationship betweenintraventricularconductionandheartrate.JElectrocardiol 2016;49:362–70.

28.LindnerKH,AhnefeldFW,SchuermannW,BowdlerIM.

Epinephrineandnorepinephrineincardiopulmonaryresuscitation:

effectsonmyocardialoxygendeliveryandconsumption.Chest 1990;97:1458–62.

29.DitcheyRV,LindenfeldJ.Failureofepinephrinetoimprovethe balancebetweenmyocardialoxygensupplyanddemandduring closed-chestresuscitationindogs.Circulation1988;78:382–9.