Resuscitation 34 (1997) 221–225

Improving cardiopulmonary resuscitation skills retention: effect of twochecklists designed to prompt correct performance

Phillip Ward a,*, Lisa A. Johnson b, Neil W. Mulligan c, Marie C. Ward d, Diana L. Jones b

a Department of Health and Human Performance, 247 Mabel Lee Hall, Uni6ersity of Nebraska-Lincoln, Lincoln, NE 68588-0229, USAb Department of Physical Education, Health and Recreation, 101 McCormic Hall, Illinois State Uni6ersity, Normal, IL 61790-5121, USA

c Department of Psychology, Campus Box 4620, Illinois State Uni6ersity, Normal, IL 61790-4620, USAd Ser6ices for Students with Disabilities, 132 Canfield Administration Building, Uni6ersity of Nebraska-Lincoln, Lincoln, NE 685588-0401, USA

Received 29 May 1996; received in revised form 29 October 1996; accepted 29 October 1996

Abstract

Previous research has shown that regardless of an individual’s experience, life support skills such as cardiopulmonaryresuscitation (CPR) are poorly performed as soon as 1 month following training. The purpose of this study was to compare theeffects of two checklists designed to prompt correct CPR performance. We compared the performance of 169 undergraduatestudents, at the time of course assessment, with retention testing that occurred 2 months following the course assessment. Studentswere randomly assigned to a control group, a short version of a CPR checklist and a longer more detailed version. Two groupsof variables were created: procedural and compression–ventilation variables. In addition, an overall-performance variable wascreated, summarizing performance on the procedural variables. Binary variables were assessed with x2-tests of independence.One-way ANOVAs, using ‘group’ as the between-subjects factor, were used to assess each continuous variable. Comparisonsbetween groups yielded significant differences of PB0.05. The long checklist generally led to superior performance on theprocedural variables. The results support the hypothesis that remembering the steps of CPR is too complex for some. Thoughpreliminary, the findings of this study indicate that the detailed checklist was an effective strategy to improve the post-courseperformance of CPR. © 1997 Elsevier Science Ireland Ltd.

Keywords: Cardiopulmonary resuscitation; Checklist; Retention; Training

1. Introduction

In 1973, the American Heart Association and laterthe American Red Cross (ARC), recognizing that oftenthe first people on the scene of emergencies were laypersons and not medical personnel, proposed and pur-sued a policy of training the lay public in basic lifesupport skills. The goal of this policy has been toincrease the number of persons trained to perform basiclife support skills and increasing the number of livessaved by emergency intervention [1]. Since 1975, 35studies have been published documenting the effective-ness of the training and retention of cardiopulmonary

resuscitation (CPR) skills [2,3]. A review of these stud-ies reveals two recurring problems, poor training inCPR at the basic life support skills course and poorretention of CPR during the months following thecourse [2,3]. These problems have typically been ad-dressed by improving the course design using alterna-tives such as computers and videodisc [4], modifiedcourses [5], and self paced instruction [6]. To date theseefforts have not improved retention.

One conclusion from the retention studies is thatperhaps remembering the steps of CPR is too difficult atask. Some have argued in favor of simplifying theprocedures so that lay public might remember them [2].An alternative strategy to simplifying the procedures ofbasic life support skills is to use a checklist to take the* Corresponding author. E-mail: pward@unlinfo.edu

0300-9572/97/$17.00 © 1997 Elsevier Science Ireland Ltd. All rights reserved.PII S 0300 -9572 (96 )01069 -6

P. Ward et al. / Resuscitation 34 (1997) 221–225222

guess work out of the procedures. Checklists have beenused to more effectively train course participants in thecorrect performance of CPR [3]. Instead of training andhoping for retention, perhaps teaching the participantsto use and follow the steps in a checklist might offersome solution to the problem of retention. In this studywe assessed the effects of two checklists which wereused to guide the behaviors of undergraduates as theyperformed CPR.

2. Methods

2.1. Organization and equipment

The subjects in this study were 169 undergraduates ata Midwest University who were enrolled in an electiveclass to be certified in basic life support skills. The basicstrategy was to assess all students at the time of CPRcertification testing, randomly assign them to a controland two experimental groups, and then re-test forretention 2 months later. At the time of re-testing, oneexperimental group used a wallet size short checklist(SC condition) published by the ARC that includeddiagrams as well as text identifying check, call and careprocedures [7] (e.g. if no pulse or breathing—give CPR:repeat sets of 15 compressions and two breaths). Theother experimental group used a longer and more de-tailed checklist (no diagrams) which outlines the specificprocedure for CPR described in ARC ‘First Aid: Re-sponding To Emergencies’ course manual [8] (the longchecklist, or LC condition). The control group used nochecklist.

2.2. CPR training and testing

Students were tested using two Laerdal SkillmeterResusci Anne recording manikins. Initial testing oc-curred at the time of the CPR course assessment.Retention testing occurred 2 months following the ini-tial assessment. Students performed the retention testwith little or no forewarning. Retention testing typicallyoccurred as students exited or entered their regularlyscheduled classes. At this time they were asked toaccompany a investigator to a room where they foundthe manikin. Retention testing took place during atwo-week period. No effort was made to prevent newsof the testing from becoming public, though no evi-dence of this was observed.

2.3. Definitions

Two groups of variables were created: proceduraland compression–ventilation variables. Proceduralvariables represented critical elements of the CPR pro-cedure and included: calling 911, performing the wrong

procedure (a procedure other than CPR was performedsuch as abdominal thrusts), head tilt, compression rate,and first pulse check. These critical elements repre-sented the key steps or errors in performing CPR forthe investigators. In addition, an overall-performancevariable was created, summarizing performance on theprocedural variables. Specifically, a subject’s perfor-mance was considered successful if the subject suc-ceeded on all of the procedural variables. Eachprocedural variable was treated as a yes/no binaryvariable.

The compression–ventilation variables were obtainedfrom the recording manikin and measured the accuracyof the ventilations and compressions. The recordingmanikin provided a fine-grained analysis of ventilationsand compressions, categorizing each as too little, justright, or too much. Because the depth of compressionsand ventilations had to be within a small window forthe manikin to register them as ‘just right’, we createda larger category of correct performance, countingthose compressions and ventilations that were eithertoo much or just right as correct. This yielded twomeasures for each subject (proportion correct for venti-lations and proportion correct for compressions). Sub-sequent analyses based on the stringent (i.e. just right)and the more inclusive (i.e. just right and too much)definitions of correct performance led to identical con-clusions. Only the analysis of the latter set of variablesare reported below. The compression–ventilation ratio(C–V ratio) was assessed as a yes/no binary variable.The distinction between the first and second set ofvariables was made for two reasons. First, most CPRcourses assess without the aid of recording manikinsusing checklists instead to determine competence, andas such the procedural variable represents a measurewhich is typically used in the majority of settings whereCPR training occurs. Second, data obtained from therecording manikin provided quantitative measures ofthe depth of the compression and ventilations andallowed us to assess with some sophistication theseaspects of the CPR performances which would notnormally be possible on a non-recording manikin.

2.4. Statistical methods

The effects of group on the yes/no binary variables(i.e. the procedural variables and C–V ratio) wereassessed with x2-square tests of independence (or Fish-er’s exact test where expected frequencies were low).The proportions of ventilations and compressions cate-gorized as correct were analyzed with the Kruskal–Wallis test, using group as the between-subjects factor.This non-parametric test was used because the depen-dent measures displayed significant non-normality, asassessed by the Lilliefors test.

P. Ward et al. / Resuscitation 34 (1997) 221–225 223

Table 1Initial testing: proportions of subjects sucessfully completing differentaspects of CPR

Successful completionVariable

%naGroup

Overall performance on procedural variables13NC 22

SC 11 20169LC

Procedural call 9117242NC

SC 40 7137 67LC

Correct procedureNC 54 93SC 9654

55 100LCHead tilt

NC 33 57SC 38 68LC 30 54

Compression rate4526NC

SC 28 5028 51LC

First pulse check56 97NC

SC 55 98LC 52 94

C–V ratio7644NC

47 84SC48 87LC

NC, no checklist; SC, short checklist; LC, long checklist.a Total n for each group: NC=58; SC=56; LC=55.

Table 3Retention testing: proportions of subjects sucessfully completing dif-ferent aspects of CPR as a function of group

Variable ComparisonSuccessful comple-tion

nbGroup % P-valuea

Overall performance on procedural All groups, B0.01variables

8NC 14 NC vs. SC, ns7SC 13 NC vs. LC, B0.02

LC 18 33 SC vs. LC, B0.01

Procedural call 911 All groups, B0.00129NC 50 NC vs. SC, ns

SC 29 52 NC vs. LC, B0.00149 89 SC vs. LC, B0.001LC

Correct procedure All groups, ns51 88NC NC vs. SC, ns48SC 86 NC vs. LC, ns

98LC 54 SC vs. LC, B0.03

Head tilt All groups, B0.003NC 28 48 NC vs. SC, B0.007SC NC vs. LC, B0.0037341

7642 SC vs. LC, nsLC

Compression rate All groups, ns25NC 43 NC vs. SC, ns

34SC NC vs. LC, ns19LC SC vs. LC, B0.035430

First pulse check All groups, nsNC vs. SC, nsNC 48 83NC vs. LC, ns78SC 44

93LC SC vs. LC, B0.0551

All groups, B0.04C–V ratio36 62NC NC vs. SC, ns

SC 39 70 NC vs. LC, B0.0146LC 84 SC vs. LC, ns

NC, no checklist; SC, short checklist; LC, long checklist.a P-value denoted if P-value B0.05; otherwise ns, non-significant atthe 0.05 level.b Total n for each group: NC=58; SC=56; LC=55.

3. Results

Data from the initial testing was used to verify thatthere were no group differences prior to instituting theexperimental manipulation. Analyses of the binary vari-ables and the proportions revealed no significant differ-ences between groups at the time of initial testing (seeTables 1 and 2).

The results of the retention testing for the binary

variables are presented in Table 3. For each variable,an analysis was conducted to determine if the propor-tions of successful subjects differed across all groups.Pairwise comparisons were conducted to determine ifpairs of groups differed. Table 3 lists P-values forcomparisons which yielded significant differences ofPB0.05. The long checklist led to generally superiorperformance on the summary variable. Here, the effectof group was significant, indicating that the long check-list led to superior performance compared with eitherno checklist or the short checklist. In addition, the longchecklist was superior to the short checklist for thefollowing specific procedural variables: call 911, per-forming the wrong procedure, compression rate, andthe first pulse check. The long checklist was superior tono checklist for call 911 and head tilt. It should be

Table 2Initial testing: proportions of correct ventilations and compressionsas a function of group

Group

SC LCNC

S.D.Mean Mean S.D. Mean S.D.

0.52 0.43 0.53Ventilations 0.41 0.430.520.460.370.46Compressions 0.380.450.34

NC, no checklist; SC, short checklist; LC, long checklist.

P. Ward et al. / Resuscitation 34 (1997) 221–225224

Table 4Retention testing: proportions of correct ventilations and compressions as a function of groupa

Group

NC SC LC

Mean S.D.Mean S.D. Mean S.D.

0.56 (0.51)0.41 (0.41) 0.40 (0.41)Ventilations 0.47 (0.44)0.50 (0.44) 0.38 (0.39)0.40 (0.37)0.43 (0.39)Compressions 0.34 (0.33) 0.34 (0.34) 0.34 (0.32) 0.31 (0.31)

NC, no checklist; SC, short checklist; LC, long checklist.a Means and S.D.s excluding (and in parentheses, including) subjects with zero valid ventilations or compressions.

noted that with regard to the procedural variables, theshort checklist and the no checklist conditions led tocomparable levels of performance, except for the headtilt variable, where the short checklist led to a signifi-cantly better performance. Finally, the experimentalmanipulation affected the C–V ratio as well. Specifi-cally, the long checklist led to significantly better per-formance than no checklist.

Several students failed to register any compressionsor ventilations on the manikin recording equipment.The number of subjects registering zero compressionswas 11 (6.5% of the total number): three from each ofthe no checklist and short checklist groups and fivefrom the long checklist group. The number of subjectsregistering zero ventilations was 14 (8% of the totalnumber): seven from the no checklist group, three fromthe short checklist group and four from the long check-list group. These frequencies did not significantly differacross groups for either ventilations or compressions.All analyses were carried out in two ways, once exclud-ing subjects with zero compressions/ventilations andagain including these subjects, assigning a value of zerofor the correct proportion. For all analyses, the conclu-sions were identical. The results of the ventilation andcompression data are presented in Table 4. The analy-ses indicate that neither the proportions of correctventilations nor compressions significantly varied acrossexperimental groups.

4. Discussion

A number of important findings arise from thisstudy. First, the results of the summary variable indi-cate that the group using the detailed checklist wasmore effective in performing correct CPR than eitherthe short version group or the group performing with-out a checklist. In addition, the individual variables ofthe procedure were either at least as effective with thedetailed checklist or significantly more effective thaneither of the other groups. For example, 89% of thestudents in the long checklist group remembered to call911 compared with approximately 50% of those in the

no checklist or the short version groups. There werealso fewer wrong procedures performed by members ofthe long checklist group (98% correct) than either of theother groups (88 and 86% correct).

Second, although there were no significant differ-ences among groups on the proportions of correctcompressions or ventilations, the groups did differ onthe C–V ratio variable. However, the successful com-pletion by members of the long checklist group wasonly 84%. It appears that while procedural steps can beprompted, the depth of compression and ventilationsmay be more difficult to train for and retain. Thisfinding is compounded by the differences between train-ing and assessment conditions. During the CPR courseperformances were conducted on non-computerizedmanikins. These manikins were older and discrimina-tions between the depth of compressions and ventila-tions were approximate. In contrast, students receivedtwo trials using the recording manikins—one at thetime of the initial testing and the other at the time ofretention assessment. It may be that not having re-ceived the feedback from the computerized manikinduring training, regarding the depth of compressionsand ventilations, that the results are not as representa-tive of how well trained individuals would perform. Onthe other hand the majority of CPR training in the USdoes not occur using recording manikins in which casethese data have high ecological validity.

Third, using the detailed version of the checklist didnot appear difficult for the students in this study. Wegave the short and long checklists to the students at thetime retention testing, this was the first time that theyhad seen them. What would have been the effects ofgroup performances if they had been taught to use thechecklist? Perhaps the results of the retention test mayhave been more significant on the overall summaryvariable.

Arguing in favor of the use of a checklist to promptCPR performance assumes that a checklist is present atthe time of need. We think it is unlikely that the laypublic would carry around such a checklist of their ownvolition. However, many people do carry car insuranceand health insurance cards in their wallets and purses.

P. Ward et al. / Resuscitation 34 (1997) 221–225 225

Typically these cards are blank on one side. Thoughour findings represent a single study, if further studiesreplicated our effects then a case might be made toinsurance companies to print CPR procedures on theback of insurance cards. Furthermore, basic life sup-port course instructors might train the lay public totake the cards from the wallet/purse and use the card toprompt CPR performance during the course.

Though the detailed checklist improved performanceon the overall performance variable, an important out-come of this study is the difference between the groupson the variable calling 911. There is abundant evidencein the form of anecdotal reports from citizens who havesaved the life of another, from television reports in thepopular press (e.g. Rescue 911) and from the on-scenepolice reports, that armed with nothing more than atelephone, individuals regardless of their experience canbe prompted to perform basic life support skills andcan do so well enough to save a life. The one step thatthe lay public needs to remember to do in an emergencyis to call 911 immediately. For lay persons at least,perhaps calling 911 is the critical skill to maintain, notmerely to ensure that an ambulance is dispatched, butalso to ensure that the rescuer is in contact with trainedpersonnel who can prompt and direct the behavior ofthe rescuer. In this study using the checklist consider-ably improved the likelihood of 911 being called imme-diately.

5. Conclusions

The results of this study support our original hypoth-esis that remembering the steps of CPR is too complexfor some. Using the detailed checklist was an effective

alternative strategy to prompt the performance of CPRskills following course conclusion. This is one of thefew studies that has reported an effective alternative totraining and hoping that CPR will be performed cor-rectly.

Acknowledgements

This research was supported by grants from theAsmund S. Laerdal Foundation and the Laerdal Medi-cal Corporation.

References

[1] National conference on cardiopulmonary resuscitation andemergency cardiac care. Guidelines for cardiopulmonary resusci-tation and emergency cardiac care, 1992.

[2] Kaye W, Rallis SF, Mancini ME, Linhares KC, Angell ML,Donovan DS, Zajano NC, and Finger JA. The problem of poorretention of cardio-pulmonary resuscitation skills may lie withthe instructor, not the learner or the curriculum. Resuscitation1991; 21: 61–87.

[3] Ward P, and Ward MC. The effects of classwide peer tutoringon correct cardiopulmonary resuscitation performance by physi-cal education majors. J Behav Educ 1996; 6: 331–342. (In Press)

[4] Kaye W, Montgonery W, Hon D et al., Interactive computer-videodisc CPR training and testing. Circulation 1983; 68 (suppl.III): 111–114.

[5] Plank CH, and Steinke KR. Effect of two teaching methods onCPR retention. J Nurs Staff Dev 1989; 5: 145–147.

[6] Friesen L, and Stotts NA. Retention of basic cardiac life supportcontent: The effect of two teaching methods. J Nurs Educ 1984;23: 184–191.

[7] American Red Cross. Adult Lifesaving Steps: Check Call Care.Mosby, St. Louis, 1993.

[8] American Red Cross. First Aid: Responding to Emergencies.Mosby, St. Louis, 1993.

. .