Resuscitation 64 (2005) 41–47

A study comparing the usability of fully automatic versus semi-automaticdefibrillation by untrained nursing students�

Koenraad G. Monsieursa,∗, Catherine Vogelsb, Leo L. Bossaertc,Philippe Meertd, Paul A. Callea

a Emergency Department, Ghent University Hospital, De Pintelaan 185, Ghent 9000, Belgiumb Intensive Care Department, Cliniques Universitaires St. Luc, Avenue Hippocrate 10, Brussels 1200, Belgium

c Intensive Care Department, University Hospital Antwerp, Wilrijkstraat 10, Edegem 2650, Belgiumd Emergency Department, Cliniques Universitaires St. Luc, Avenue Hippocrate 10, Brussels 1200, Belgium

Received 1 June 2004; received in revised form 2 July 2004; accepted 6 July 2004

Abstract

I externald ility withn SA). Weh , safety andsM t scenario.D ent). TheirpR the device,F de0 shock(C ED. Theu liver threes©

K

1

mb

�

K

ofrnaler-,

bery thered the

FA

0d

ntroduction:Current international guidelines prefer the use of semi-automatic external defibrillators (SAEDs) over fully automaticefibrillators (FAEDs). However, there is a lack of evidence supporting this recommendation. We conducted a study of usabursing students comparing the FAED version against the SAED version of the Lifepak CR Plus AED (Medtronic, Redmond, Uypothesized that FAED use would limit the number of operator-device interactions, thereby increasing compliance by the rescuerpeed.ethods:Sixty-two untrained first year nursing students were randomized to use the FAED or the SAED in a simulated cardiac arresuring analysis and delivery of three shocks, the AED guided the user with six voice prompts per shock (18 voice prompts per studerformance with regard to efficacy and safety was assessed using video recording.esults:All rescuers except for two were able to attach electrodes and deliver a series of three shocks. During rhythm analysis byAED users made 30/372 (8%) errors against 62/360 (17%) errors for SAED users (P < 0.001). During shock delivery, FAED users ma/186 errors against 12/180 (7%) for SAED users (P < 0.001). FAED use eliminated long time intervals between the first to the thirdrange 47–49 s for FAED versus 41–90 s for SAED).onclusion:Despite a lack of BLS skills and AED training, the majority of students demonstrated safe and effective use of the Ase of the FAED version of the CR Plus resulted in increased compliance with the protocol and reduced variability in time to dehocks. Further research is needed to confirm these findings in other groups of first responders.2004 Elsevier Ireland Ltd. All rights reserved.

eywords:Automated external defibrillator (AED); Cardiac arrest; Defibrillation; Safety; Training; Usability

. Introduction

Previous studies have indicated that the use of an auto-ated external defibrillator (AED) by untrained rescuers maye feasible, but safe operation has been one of the major con-

A Spanish and Portuguese translated version of the Abstract andeywords of this article appears at10.1016/j.resuscitation.2004.07.003.

∗ Corresponding author. Tel.: +32 9 2402755; fax: +32 9 2403948E-mail address:koen.monsieurs@ugent.be (K.G. Monsieurs).

cerns[1–6]. Current international guidelines on the useAEDs recommend the use of semi-automatic (SA) extedefibrillators (SAEDs) reserving fully automatic (FA) extnal defibrillators (FAEDs) for special situations[7]. Howeverevidence to support this recommendation is lacking.

We hypothesized that FAED use would limit the numof operator-device interactions, increasing compliance bstudent nurses, safety and speed. Therefore, we compause by single untrained nursing students of an AED inversus SA mode.

300-9572/$ – see front matter © 2004 Elsevier Ireland Ltd. All rights reserved.oi:10.1016/j.resuscitation.2004.07.003

42 K.G. Monsieurs et al. / Resuscitation 64 (2005) 41–47

Fig. 1. Flow chart of fully and semi-automatic modes of the Lifepak CR Plus (voice prompts are printed in italic).

2. Methods

Sixty-two first year nursing students (Institut Superieurd’Enseignement Infirmier, Universite Catholique de Louvain,Belgium) were randomized to use either the FAED version orthe SAED version of the Lifepak CR Plus AED (both trainingAEDs, Medtronic, Redmond, USA) in a simulated cardiac ar-rest scenario with a fully dressed training manikin (RecordingResusci Anne, Laerdal, Norway), delivering a series of threeconsecutive shocks. We have used this particular device be-cause it is the only latest generation AED currently availablein both fully and semi-automatic mode. Apart from a smalllid covering the shock button in FA mode, both types of AEDare identical on the outside. The sounds during charging arealso identical. During analysis and shock delivery, the AEDsonly differ in a single voice prompt (seeFig. 1). The studentswere unprepared; they were expecting a scheduled lectureon first aid. A questionnaire was taken immediately beforethe test to assess demographics and previous basic life sup-port (BLS)/AED training and experience. The performanceof the students was recorded on video and their behaviourwas scored afterwards by KM and CV using the Cardiff scor-ing system[8]. All participants gave prior written informedconsent for their actions to be filmed.

2

rsta eri-

enced chest pain”. No further information was given to therescuer, except the clinical condition of the manikin, whichwas communicated after each appropriate assessment by therescuer. Ventilation and compression data were recorded onthe manikin. All BLS actions were also recorded on video.After a maximum of 3 min, the rescue attempt was inter-rupted.

2.2. Definitions of BLS performance

The Cardiff scoring criteria were used, but for some tech-niques this scoring system did not allow meaningful con-clusions because of the poor performance of our untrainedparticipants. We therefore modified the definitions for thefollowing items:

• Call for an ambulance: the candidate mentions a “call foran ambulance” at any time during the resuscitation effort.

• Free airway: the airway is opened before assessment ofbreathing and before two initial rescue breaths.

• Effective rescue breathing: a minimum of six effectivemouth-to-mouth ventilations per minute (tidal volume800–1200 ml).

• Correct hand position: hand position on the lower half ofthe sternum

• at

• on

.1. Assessment of BLS skills

To determine BLS skill level, all participants were fisked to help “a victim who collapsed after having exp

Effective chest compression: depth between 4 and 5 cma rate between 80 and 120 per min.Correct ratio: at least two cycles showing two ventilatiattempts and from 14 to 16 chest compressions.

K.G. Monsieurs et al. / Resuscitation 64 (2005) 41–47 43

2.3. Assessment of AED skills

After the assessment of the BLS skills, the manikin wasredressed and an AED was given without explaining its pur-pose, only that they should “attempt to use the device as itcould help the victim”. The AED actions were recorded onvideo and interrupted after a maximum of 4 min.

The AEDs were programmed to deliver a series of threeconsecutive shocks. During cardiac rhythm analysis andshock delivery, the AED guided the user with six voiceprompts:

1. Do not touch patient2. Evaluating heart rhythm3. Stand by4. Preparing to shock5. a. SAED: everyone clear, press flashing button

b. FAED: do not touch patient, delivering shock6. Shock delivered

Because each participant delivered three shocks, 18voice prompts were given per rescuer, i.e. a total of 558voice prompts for the FAED group (n = 31) and 540 forthe SAED group (n = 30). Based on the voice prompts,each rescue attempt was subdivided into three sectionscovering six episodes. These episodes were used to tabulateincidents occurring during the rescue efforts. After the firstv ctedt icep ayn lastt tingt ss thb , ana

TheF lish,e ichw ans“ re-c rieso

f them heA rodest

thec motec

2

iques( on,s o firsts s ont uringt livery

of three shocks. The position of the electrodes on the manikinwas assessed by comparing the video images with the correctarea as defined by the Cardiff scoring system[8].

2.5. Statistics

Continuous variables were assessed with the Mann–WhitneyU-test, discrete variables with the chi-square testand Fisher exact test where appropriate using Statistica forWindows (Statsoft).

3. Results

3.1. Characteristics of rescuers (Table 1)

A limited number of participants reported that they hadfollowed a BLS course previously. In addition, 13 subjectsof each group had participated in a trial studying BLS froma CD-ROM, 7 FAED users and five SAED users had beenexposed to interactive CD-ROM AED content showing a car-diac arrest situation where a Forerunner SAED (Philips) wasused. None of the rescuers had received previous hands-ontraining in AED use.

3.2. Baseline BLS skills (Table 1)

s ando res-s

3

dedd roup.I nots ipantf As ar s nota hedo l andt s.

•t of(i.e.andthe

enthis

g theodesnceoupse of

oice prompt “do not touch patient”, rescuers are expeo stand clear until shock delivery. The first four vorompts guide the rhythm analysis; touching the victim megatively influence the accuracy of this process. The

wo voice prompts relate to shock delivery; not respechem exposes the rescuer to a safety hazard. To asseehaviour of the participants after each voice promptdditional detailed scoring sheet was used.

The language of the training AEDs was French.rench voice prompts are exact translations from Engxcept for an additional voice prompt “continue care” whas translated into “maintenir surveillance” (which me

maintain surveillance”). This “continue care” prompt (peded by a “check breathing” prompt) is given after the sef three shocks.

The correct position of the electrodes on the chest oanikin could be identified from pictorial information on tED, on the package of the electrodes and on the elect

hemselves.After the participants had attached the electrodes to

hest, an instructor pressed a button on an infrared reontrol to allow the AED to proceed.

.4. Outcome measures

Outcome measures were: performance of BLS technbefore AED use), AED time intervals (receipt to switchwitch on to electrode placement, electrode placement thock, first shock to third shock), position of electrodehe chest, rescuer compliance after each voice prompt dhe series of three shocks and BLS performance after de

e

None of the rescuers delivered effective rescue breathnly five rescuers (8%) performed effective chest compion.

.3. AED skills

The moment of receipt of the AED was not recorue to technical reasons for one participant in each g

n addition, one participant from the FAED group diducceed in attaching either electrode and one particrom the SAED group attached only a single electrode.esult the time interval switch to electrode placement wavailable in four participants. The participant who attacnly one electrode was allowed to continue the protoco

herefore data on compliance were collected on 61 case

Time intervals independent of AED type (Table 1)Since the AEDs perform identically up to the momenshock delivery, the performance data before this pointincluding the intervals between receipt to switch on,switch on to electrode placement) are not related todifferences between SAED and FAED.The interval between switch on and electrode placemwas statistically different between the two groups. Tinterval includes tasks such as undressing, openinelectrode package and correct positioning of the electron the chest. A possible explanation for the differeobserved could be an imbalance between the two grcaused by the wide variability in speed of performancthese tasks.

44 K.G. Monsieurs et al. / Resuscitation 64 (2005) 41–47

Table 1Characteristics of participants and baseline BLS skills (not influenced byAED type)

FA (n = 32) SA (n = 30) P

Gender (M/F) 4/28 0/30 0.04Age (S.D.) 23 (8) 21 (5) 0.94Practical BLS course 7 8 0.66CD-Rom BLS 13 13 0.83

BLS skillsShake and shout 13 (41%) 8 (27%) 0.25Free airway 7 (22%) 5 (17%) 0.60Call ambulance 8 (25%) 11 (37%) 0.32Effective rescue breathing 0 (0%) 0 (0%)Correct hand position 14 (44%) 15 (50%) 0.62Effective chest compression 3 (9%) 2 (7%) 0.70Correct ratio 15:2 8 (25%) 11 (37%) 0.32

AED time interval (sec)Receipt to switch on (S.D.) 15 (8) 20 (21) 0.7Switch on to electrodes (S.D.) 64 (23) 56 (19) 0.04

• Position of electrodes (Table 2)Only 22/62 (35%) rescuers positioned electrodes cor-rectly. There was no difference between FAED and SAEDusers, as would be expected from the fact that the AEDsbehave identically up to the moment of shock delivery.

• Time intervals related to AED typeThe interval between applications of electrodes to firstshock (S.D.) was 22 (2) sec for FA (n = 32) and 21 (2)sec for SA (n = 30). This difference was statisticallysignificant (P = 0.014), though without clinical relevance.The interval between the first shock to third shock (S.D.)was 48 (0.5) sec for FA and 47 (7) for SA (P< 0.001) witha range of 47–49 s for FA and 41–80 s for SA. This meansthat SAED users were sometimes faster than FAED usersbecause they pressed the shock button before the end ofthe voice prompt. On the other hand, FAED use preventedlonger intervals.

• Rescuer compliance during analysis and shock deliveryWe observed three main types of non-compliance(Table 3):o Circulation assessment (Fig. 2a)

During analysis and shock delivery, rescuers assessedcirculation in various ways: by palpating the wrist, theneck or even the precordial area. This occurred much

TP

P

BO )BO 2O 2BN

T

T

Table 3Types of non-compliance during the different episodes delineated by thevoice prompts during the series of three shocks

FA SA Total P

Circulationassessment

15/558 51/540 66/1098 < 0.0001

Touchingelectrodes

12/558 10/540 22/1098 0.72

Other touchingerrors

3/558 13/540 16/1098 0.01

Total 30/558(5%)

74/540(14%)

104/1098(9%)

< 0.0001

It should be noted that not more than one error was found during each episode.

more frequently in the SA group. Most errors occurredat the beginning of the rhythm analysis sequence.

o Touching electrodes (Fig. 2b)Except for one case, electrodes were touched onlyduring the first shock and most errors occurred duringthe first voice prompts. This indicates that rescuerswanted to adjust the position of the electrodes or ensureproper attachment, even if the voice message promptedthem not to touch the patient.

o Other touching errors (Fig. 2c)These include accidental or intentional touching of thevictim, such as touching clothing.

Overall, rescuers made non-compliance errors in 9%(104/1098), with SAED users making 14% errors (74/540)and FA users only 5% (30/558) (Table 3). Table 4lists thecompliance for each voice prompt. Most errors occurred dur-ing the second voice prompt “evaluating heart rhythm”. Com-bining the voice prompts into an “analysis” group (prompts1–4) reveals that 30/372 (8%) errors were made by FA usersand 62/360 (17%) errors by SA users (P < 0.001). Errorsduring this analysis episode might interfere with the rhythmanalysis but are not a safety risk. Combining the results forvoice prompts five and six reveals that 0/186 errors were madeby the FA users against 12/180 (7%) by the SA group (P <0.001). These errors occurred immediately before or durings e FAg f 18v madeb

uents ED.

TC

123456T

able 2osition of electrodes

osition of electrodes FA SA TOTAL

oth within area 11 11 22 (35%)ne within area, one crossing border 13 11 24 (39%oth crossing border 3 4 7 (11%)ne within area, one outside area 1 1ne crossing border, one outside area 1 1oth outside area 2 1 3ot attached 1 1 2

otal 32 30 62

here were no statistical differences between the FA and SA groups.

hock delivery and are therefore a safety hazard. In throup, rescuers made a maximum of four errors out ooice prompts, whereas in the SA group seven rescuersetween 5 to 11 errors.

Fig. 3 shows that compliance increased with subseqhocks for FAED, whereas compliance decreased for SA

able 4ompliance errors for each voice prompt

FA SA Total P

6/93 5/90 11/183 0.815/93 25/90 40/183 0.065/93 17/90 22/183 0.0054/93 15/90 19/183 0.0060/93 9/90 9/183 0.0020/93 3/90 3/183 0.08

otal 30/558 74/540 104/1098 <0.0001

K.G. Monsieurs et al. / Resuscitation 64 (2005) 41–47 45

Fig. 2. (a) Circulation assessment (Erroneous circulation assessment for each voice prompt). (b) Touching electrodes (Erroneously touching electrodes for eachvoice prompt). (c) Other touching errors (Other touching errors for each voice prompt). Shock 1: grey, shock 2: white and shock 3: barred

46 K.G. Monsieurs et al. / Resuscitation 64 (2005) 41–47

Fig. 3. Errors during shocks (touching electrodes: white, circulation assessment: grey and other touching errors: barred).

3.4. BLS skills after shock delivery

After delivery of three shocks, 54/61 (88%) complied withthe “check breathing” voice prompt by checking breathingand/or the pulse (count irrespective of correctness). After thenext voice prompt “if no breathing, start CPR” only half ofthe rescuers attempted rescue breathing, few attempted chestcompression and one third did nothing at all. After the voiceprompt “maintenir surveillance” 6/61 (10%) rescuers evenceased the resuscitation efforts which they had started earlier.There were no differences in behaviour between the FAEDand SAED group.

4. Discussion

According to the recommendations of the InternationalLiaison Committee on Resuscitation (ILCOR) basic res-cuers, or even lay people, are required to operate AEDs toimprove outcome after cardiac arrest[7]. This strategy posesan enormous training challenge with regard to the operatorsof these devices. The current generation AEDs showimproved usability and studies are emerging that the use ofAEDs by minimally trained rescuers may be feasible[1–6].

Fromm and Varon showed that untrained lay rescuersc ng as ft ety-s ocksd dataw onA parel f theF tedc cew and

all remained clear of the patient during shock delivery. How-ever, before the test the children received verbal instructionsregarding pad handling and placement. The “pass” criteriafor pad placement were quite liberal and differ from currentguidelines. Mattei et al. studied 15 nurses and physiothera-pists [3]. A Forerunner AED was used during a simulatedcardiac arrest. Fifty-three percent showed wrong pad place-ment and 67% were unsafe, thereby indicating that pad posi-tion and safety are areas for improvement. Wik et al. showedthat 15/27 pairs of flight attendants with no previous expo-sure to an AED decided to use a Forerunner AED during asimulated cardiac arrest scenario[4]. Eighty-six percent ofpads were judged to be located correctly and no safety errorswere observed. However, all the subjects had previous BLStraining and all received a BLS/AED course manual beforethe test. Moreover 12 pairs of flight attendants were reluctantto use the AED. The study certainly suggests that a signifi-cant number of rescuers proficient in BLS would need onlywritten information to use an AED effectively. Bradley et al.tested 30 untrained individuals at an airport[6]. Only eightwere able to defibrillate within 3 min, the main problem beingfailure to apply the electrodes properly. The study supportsour findings that visual and oral instructions are not adequateenough to guide untrained users, especially for applying theelectrodes.

thata d byt y layr im-p den.H evicei

EDsic wered inte-

ould use a Heartstart 1000 (Laerdal, Norway) AED, usiimple written instruction sheet[1]. Approximately 50% ohe 25 subjects had previously received BLS training. Ninix percent of the rescuers were able to deliver three shuring a simulated cardiac arrest scenario. No safetyere published. It was concluded that brief instructionED use during BLS courses might be sufficient to pre

ay rescuers for AED use. Gundry et al. studied the use oorerunner AED by 15 untrained children during simulaardiac arrest[2]. None had prior BLS training or experienith AED. All subjects placed electrode pads correctly

From these mentioned studies, one may concludelthough current AEDs have been developed to be use

rained individuals, at least some AEDs can be used bescuers with minimal or no training at all. This is anortant finding because it could reduce the training burowever, most of the studies do not address operator-d

nteractions adequately.Literature data on the choice between FAEDs and SA

s very scarce. In 1987, Stults and Cummins[5] wrote that thehoice is largely subjective and that safety and efficiencyepending on good initial training and adequate skill ma

K.G. Monsieurs et al. / Resuscitation 64 (2005) 41–47 47

nance. The perceived increased risk of inadvertent shocks inFA mode would not exist when operated by trained rescuers.Still, current international guidelines on the use of AEDsfavour the use of SAEDs reserving FAEDs for “special sit-uations”, without detailing these[7]. Our results show thatthe Lifepak CR Plus FAED version eliminated the slower de-livery of shocks by inexperienced rescuers. In addition, weobserved that the students using the FAED were more com-pliant with the protocol than rescuers using the SAED, bothduring rhythm analysis and during shock delivery.

The most common error was touching the patient to checkfor signs of a circulation (Table 3). This error occurred morein relation to the second voice prompt (“analyzing heartrhythm”) (Fig. 2a). We believe some rescuers were confusedby this voice prompt which was often interpreted as an incen-tive to feel for a pulse whereas its goal is merely to provideinformation. Still, although the voice prompts were identicalin both groups, the effect was much smaller in the FA group.A probable explanation is that the FA mode puts the rescuer ina more passive and secure position during analysis and shockdelivery, preventing the rescuer from inappropriate actions.

We recognize a number of limitations of our study.Nursing students may have a special interest in resuscitation,or a more positive attitude towards helping than the generalpopulation. The results cannot therefore be applied to otherg diaca viourd ictedt toa s thep rs oro lysisa

5

un-t siono thirds highn LS

after defibrillation was largely inadequate. Future design ofAEDs should focus on these issues by avoiding unwantedrescuer–victim interactions during rhythm analysis and shockdelivery, and by improved guidance for the rescuer when BLSis required. Further research is needed before the results canbe applied to other groups of first responders.

Acknowledgements

We thank the students of the Institut Superieurd’Enseignement Infirmier for their participation and CybeleDelwiche, Anne-Marie Champagne and Benoıt Dupuis forproviding logistical support. We also thank all instructors forrunning the testing stations. Medtronic provided the AEDsfor the study.Conflict of interests: The authors of this article have no

conflict of interest to report.

References

[1] Fromm Jr RE, Varon J. Automated external versus blind manual de-fibrillation by untrained lay rescuers. Resuscitation 1997;33:219–21.

[2] undry JW, Comess KA, DeRook FA, Jorgenson D, Bardy GH.Comparison of naıve sixth-grade children with trained profession-

tion

[ sio-lator?

[ tor-l per-

[ sory7;12:

[ fib-ed

[ sci-

[ ithualtyrding.

roups of first responders with certainty. A simulated carrrest scenario may not necessarily predict rescuer behauring a real resuscitation event. Our findings are restr

o the Medtronic CR Plus AED. The study was limitedsingle rescuer situation and did, therefore, not asses

erformance of rescuers in the presence of bystandether rescuers who may touch the patient during anand shock delivery.

. Conclusion

Protocol compliance was increased in this group ofrained nursing students using the fully automatic verf the CR Plus. Prolonged intervals between the first tohock were eliminated. In addition, we observed a ratherumber of errors with regard to electrode positioning. B

als in the use of an automated external defibrillator. Circula1999;100:1703–7.

3] Mattei LC, McKay U, Lepper MW, Soar J. Do nurses and phytherapists require training to use an automated external defibrilResuscitation 2002;53:277–80.

4] Wik L, Dorph E, Auestad B, Steen P. Evaluation of a defibrillabasic cardiopulmonary resuscitation programme for non medicasonnel. Resuscitation 2003;56:167–72.

5] Stults KR, Cummins RO. Fully automatic versus shock advidefibrillators: what are the issues? J Emerg Med Serv 19871–3.

6] Bradley RN, Hamilton WH, Boyle MR, et al. Public access derillation programs without training are effective. Acad Emerg M2003;10:502–3 (abstract).

7] International Guidelines 2000 for CPR and ECC-A consensus onence. Resuscitation 2000;46:1–448.

8] Lester CA, Morgan CL, Donnelly PD, Assar D. Assessing wCARE: an innovative method of testing the approach and casassessment components of basic life support, using video recoResuscitation 1997;34:4.