If you don't remember your password, you can reset it by entering your email address and clicking the Reset Password button. You will then receive an email that contains a secure link for resetting your password
If the address matches a valid account an email will be sent to __email__ with instructions for resetting your password
Corresponding author at: Department of Anaesthesiology and Pain Medicine, Bern University Hospital, University of Bern, Freiburgstrasse, 3010 Bern, Switzerland.
Department of Anaesthesiology and Pain Medicine, Bern University Hospital, University of Bern, Bern, SwitzerlandDepartment of Anesthesia and Pain Management, Sinai Health System, University of Toronto, Toronto, CanadaERC ResearchNET
Department of Anaesthesiology and Pain Medicine, Bern University Hospital, University of Bern, Bern, SwitzerlandStatistical Unit, Department of Anaesthesiology and Pain Medicine, Bern University Hospital, University of Bern, Bern, Switzerland
The ideal group size for effective teaching of cardiopulmonary resuscitation is currently under debate. The upper limit is reached when instructors are unable to correct participants’ errors during skills practice. This simulation study aimed to define this limit during cardiopulmonary resuscitation teaching.
Medical students acting as simulated Basic Life Support course participants were instructed to make three different pre-defined Basic Life Support quality errors (e.g., chest compression too fast) in 7 min. Basic Life Support instructors were randomized to groups of 3–10 participants. Instructors were asked to observe the Basic Life Support skills and to correct performance errors. Primary outcome was the maximum group size at which the percentage of correctly identified participants’ errors drops below 80%.
Sixty-four instructors participated, eight for each group size. Their average age was 41 ± 9 years and 33% were female, with a median [25th percentile; 75th percentile] teaching experience of 6 [2;11] years. Instructors had taught 3 [1;5] cardiopulmonary resuscitation courses in the year before the study. A logistic binominal regression model showed that the predicted mean percentage of correctly identified participants’ errors dropped below 80% for group sizes larger than six.
This randomized controlled simulation trial reveals decreased ability of instructors to detect Basic Life Support performance errors with increased group size. The maximum group size enabling Basic Life Support instructors to correct more than 80% of errors is six. We therefore recommend a maximum instructor-to-participant ratio of 1:6 for cardiopulmonary resuscitation courses.
successful CPR involves communication and collaboration with other rescuers. To achieve this, CPR is trained in small groups, where an instructor corrects learners’ mistakes and provides feedback on their performance.
Jensen et al. found significantly higher AED use (odds ratio 21.78) and increased chest compression depth and rate (ORs of 7.47 and 15.47, respectively) while evaluating BLS courses if there were six or fewer participants per instructor.
However, the optimal group size for group CPR teaching, which would enable instructors to oversee all participants and support all learners optimally during the course, is unknown while it may have strong implications on quality and resources needed.
In its current course rules, the European Resuscitation Council (ERC) recommends six to eight participants per instructor for BLS teaching.
However, due to tight budgets and economic restrictions, training centres might be pushed to allow group sizes of up to ten participants per instructor.
This single-centre randomized controlled simulation trial assessed instructors’ ability to correct errors during CPR training in groups of various sizes. The goal of the study was to identify the threshold at which certified life support instructors are still able to detect and correct 80% of participants’ errors during performance of CPR. Knowing this threshold might be helpful to establish an evidence-based suggestion for the upper limit of group size during CPR training with one instructor.
After Cantonal Ethics Committee of Bern, Switzerland review, (Req-2017-00577 and amendment 2019-00951), this trial was performed at the Bern Simulation and CPR Centre (BeSiC) at the Bern University Hospital, between March and May 2019. The study was registered at clinicaltrials.gov (NCT03773861) in December 2018.
Medical students and nurses were recruited as simulated “BLS course participants”. All were certified BLS providers with valid status. Each was instructed to make three different errors randomly assigned from a list of 12 predefined errors during the seven-minute study period. They were instructed to perform standard BLS, to behave pleasantly, but to make the errors pointed and obvious, and to continue making them until the instructor had corrected them. After correction, they continued BLS without this error. These “BLS course participants” trained the predefined errors with the study team before the start of the study.
The predefined errors were recognized CPR quality parameters:
wrong chest compression hand position (clearly over the abdomen);
chest compression frequency markedly over 100 beats per minute
chest compression frequency markedly below 100 beats per minute
chest compressions too deep (markedly greater than 6 cm)
chest compressions too shallow (markedly less than 5 cm)
chest compression not relieved properly
head not reclined for bag-mask ventilation
only 1 ventilation after 30 chest compressions
3 ventilations after 30 chest compressions
wrong positioning of AED electrodes (over the abdomen)
AED electrodes not plugged into the AED
markedly long pause after giving shock (>30 s)
The simulated “BLS course participants” provided baseline characteristics (age, gender, primary language spoken, year of medical studies, number of life-support courses attended, registration as a Bern First Responder).
The study took place in the BeSiC. BLS training manikins with CPR feedback devices (Resusci Anne QCPR, Laerdal Medical GmbH, Puchheim, Germany), the corresponding number of bags and masks for ventilation, and AEDs (AED Trainer 3, Laerdal Medical GmbH, Puchheim, Germany) were provided based on the randomized group size and the 2015 ERC Guidelines.
BeSiC BLS instructors were invited to participate in the study. After explaining the study purpose, study participants provided written informed consent, and participants’ characteristics were recorded: age, gender, primary language spoken, occupation, experience as instructors in years, number of life support courses taught in the 12 months prior to the study, registration as Bern First Responder.
Using computer-generated codes (www.randomizer.org), eight different participating instructors were randomized to one group size, ranging between three and ten simulated “BLS course participants”, which summed up to 64 participating instructors. Each instructor participated only once to exclude a learning effect.
Participating instructors were informed that BLS theory had been taught beforehand. Now they were asked to oversee and monitor a BLS practice session (including AED use and bag-mask ventilation) for seven minutes, and to correct all errors that “BLS course participants” made. BLS followed the 2015 ERC Guidelines.
Participating instructors were informed that the group would comprise typical BLS course participants, that errors would occur and needed to be corrected.
Directly before and after the study we collected saliva samples. Participating instructors moistened a cotton roll in their mouth for one to two minutes and then place it in a Salivette® (Sarstedt AG & Co., Nümbrecht, Germany) to measure cortisol levels (in nmol/L). The Salivettes® were stored at 2−8 °C and protected from heat or direct sunlight. Electro-chemi-luminescence immunoassay (ECLIA) tests (Elecsys Cortisol II, cobas e 801, F. Hoffmann-La Roche Ltd., Basel, Switzerland) were used to analyse cortisol concentration.
Before and after the study, participating instructors were asked to rate their stress level on a Numeric Rating Scale (NRS; 1 = no stress to 10 = maximal imaginable stress).
Immediately after they finished each participating instructor’s trial session, simulated “BLS course participants” were asked if their errors had been correctly identified and corrected. Additionally, to ensure data quality all study sessions were video recorded to allow for later assessment of error identification and correction.
The primary outcome was the maximal group size at which instructors were still able to correct 80% of BLS course participants’ errors. We based the analysis of the primary outcome on the percentages of correctly identified errors for each instructor as a function of group size. Secondary outcomes were influences of the instructors’ age and experience on instructor performance, and differences in stress levels as represented by subjective stress scores and cortisol levels.
A formal sample size calculation was not performed, because none of the published studies addressing that question provided sufficient data for calculations. We therefore included as many instructors as possible and enrolled a convenient number of eight instructors per group (size ranging from 3 to 10 simulated “BLS course participants”). This resulted in a total of 64 instructors participating in the study.
Statistical analysis was performed using Stata version 16.0 (StataCorp LT, Texas, USA) and R Software (R Core Team (2020), Vienna, Austria). Descriptive statistics were used for participants’ characteristics. A Logistic-binominal regression modelled the relationship between success rate and group size. The maximum group size (primary outcome) was defined as the point where the mean prediction of the regression model dropped below the predefined threshold of 80% of correctly identified errors. The variable “group size” was considered a continuous predictor in the logistic-binominal regression model. Goodness-of-fit was examined by computing the model’s Brier-Score and the ratio of residual deviance to the residual degrees of freedom.
A linear mixed-effect model with a random intercept for each instructor was used to examine (i) the relationship between pre- and post-measurements of cortisol levels (nmol/L) using estimated marginal means (EMM) and (ii) the two-way interaction between group size and cortisol levels. A likelihood ratio test was used to assess if the pre-post change in cortisol levels depends on the group size. We used the log-transformed values of cortisol levels after examination of QQ-Plots of the distribution of cortisol levels. Contrasts and associated 95% confidence intervals are shown on the original scale. Median and interquartile ranges of cortisol levels are presented as summary measures.
In terms of stress scores, we considered the numerical rating score (NRS) as an ordinal variable and median and interquartile ranges of stress scores are presented as summary measures. A proportional odds regression model was used to examine (i) the relationship between pre- and post-measurements of stress scores using estimated marginal means (EMM) and (ii) the two-way interaction between group size and stress scores. A likelihood ratio test was used to assess if the pre-post change in stress levels depends on the group size.
Data is presented as either number (percentage), mean (standard deviation) or median (25th, 75th percentiles), where the latter values are based on R’s generic quantile-function which uses by default definition 7 of Hyndman, R. J. and Fan, Y. (1996).
A p value of < 0.05 was considered statistically significant. The p values of the secondary outcomes are not adjusted for multiple comparison as these results are considered exploratory in nature.
Twenty-eight simulated “BLS course participants” with a mean age of 25 (3) years were recruited for the study. Sixteen (57%) were female. First language was German for 26 (93%), French for one (3.5%) and other for one (3.5%). Twenty-four (86%) were medical students and four (14%) were nurses. All had previously attended BLS courses. Four (14%) were Bern First Responders.
Characteristics of the 64 instructors are displayed in Table 1. Twenty-one (33%) were women, and only 22 (34%) were Bern First Responders.
Table 1Participating BLS-certified instructors’ characteristics. Data are value (percentage), mean (standard deviation), or median [25th percentile; 75th percentile].
n = 64
- Emergency Medical Services paramedic
- Medical student
Experience as instructor, years
6 [2; 11]
Total number of courses taught as instructor before start of study
- Basic Life Support
2 [0; 40]
- Immediate Life Support
0 [0; 2]
- Advanced Life Support
Number of CPR courses instructed (last 12 months before the study)
The percentage of correctly identified and corrected errors for each group size is displayed in Fig. 1A. Fig. 1B displays the relationship between success rate and group size derived from a logistic-binomial regression model. The regression coefficient for group size was statistically significant (p = 0.001) and the goodness-of-fit estimates are a Brier-Score of is 0.17 and a ratio of residual deviance to the residual degrees of freedom of 1.13. Fig. 1B illustrates the average decline in the ability to achieve the pre-defined 80% threshold if the group size increases (above a group size of 6 the predefined threshold cannot be achieved). Neither age (p = 0.81) nor experience of the participating instructors (p = 0.60) showed a significant influence on success rate.
With regard to secondary analyses, subjectively perceived stress levels are summarized in Table 2. A proportional odds regression model showed that stress scores after the study were on average 1.0 (0.3–1.7, 95% confidence interval) units larger than before the study (p = 0.003). No interaction between group size and stress scores was found (p = 0.67).
Table 2Self-reported stress level between group sizes (rows) as well as before and after the study within one group size (columns). NRS = Numeric Rating Scale. Median and interquartile ranges are shown as well as the number of available observations in each group.
Cortisol levels in saliva are summarized in Table 3 and a linear mixed-effect model analysis indicated that cortisol levels after the study were on average 0.52 nmol/L (0.16−0.88 nmol/L) larger than before the study (p = 0.006). No interaction between group size and cortisol levels was found (p = 0.17).
Table 3Cortisol level (in nmol/L) in saliva between-group sizes (rows) as well as before and after the study within one group size (columns). Median and interquartile ranges are shown as well as the number of available observations in each group.
There were no missing values for our primary outcome or the demographic data, however there were some missing values for the secondary outcomes (NRS and cortisol levels, Table 2, Table 3).
The percentage of correctly identified and corrected errors varied between 89% for a group of three participants and 72% for a group of ten participants. A logistic-binominal regression model showed that the threshold of group size at which a life support instructor is still able to detect and correct 80% of participants’ errors during BLS training is six participants. Subjectively perceived stress levels and objectively measured saliva cortisol levels were comparable between groups.
Limited data exists about effects of group size on CPR courses. Mahling et al. compared resuscitation quality in groups of three, five and eight participants and found improvements in all groups; however, scores in groups of eight were lower. Participants in groups of eight were less likely to ask questions and more likely to engage in unrelated conversations during the course.
Cho et al. evaluated the effect of group size on feedback delivery after BLS training. Groups of three to five were compared to groups of seven to ten. Smaller groups spent more time on feedback and discussed more topics during feedback. This study recommended a group size between 4.4 and 5.8 participants as ideal for BLS training.
To the best of our knowledge there is no study systematically evaluating effectiveness of CPR teaching with varying group sizes. Our results show that an instructor is unable to effectively detect 80% of errors made by BLS course participants if the instructor-to-participant ratio is greater than one to six. The success rate declines with increasing group size. Even if uncertainty is accounted for by using 95% confidence intervals, all detected errors are below the threshold of 80% for groups of seven or more participants.
One might argue that our results do not reflect real-life CPR courses due to the experimental setting. However, all our simulated BLS course participants were trained to make their errors very obvious. Therefore, our study setup very conservatively measures the real threshold at which it is no longer possible to detect errors that happen during a BLS course. Our study setting being obvious is counteracted by less obvious errors made by real BLS-course participants.
Edmonds and Brown highlight the importance of the trainer’s communication and cognitive skills in their educational guide on effective small-group teaching.
Our study is the first evaluating the effectiveness of small-group CPR teaching focusing on instructors. Our experienced instructors were circa 41 years old, with 6 years of experience as instructors. Neither age nor previous experience was shown to be a significant confounder for the results of the interaction between group size and success rate in detecting performance errors.
Interestingly, our secondary outcome – the stress experienced by instructors in different group sizes measured subjectively on a Numeric Rating Scale as well as objectively in saliva cortisol levels – did not reveal differences between group sizes. It seems that simulated teaching might increase the stress level of an instructor, however the teaching sessions were very short; the slight increase therefore might be explained merely by the fact that participants were part of an educational study. To exclude a learning effect in our study, each participating instructor was confronted with only one group size. An inter-individual stress level comparison was therefore not possible, and we cannot provide data on how effective each individual instructor would be with a different group size.
In this study we chose to investigate effects of group sizes ranging from three to ten participants. Others compared only smaller groups with bigger groups
We chose the range of three to ten participants in order to be able to display the real threshold. Three participants is the smallest unit we can refer to as a group, and teaching BLS to more than ten persons at once seemed a priori ineffective. Even if the results for the smallest and largest group sizes seemed obvious, we chose a continuum between three and ten to establish a clear cut-off value for effective small-group skills teaching.
Interestingly, only 34% of CPR instructors in this study were registered as Bern First Responders, although there has been a campaign in Bern to recruit healthcare professionals for an app-based CPR-rescuer alarm system outside the organized Emergency Medical Service, with the aim being to shorten intervention times. The BeSiC CPR instructors are urged to recruit such First Responders during CPR courses. However, the low number of First Responders in our study reflects previous findings of our group that “professional” first responders may feel easily overwhelmed by their professional burden and don’t want to volunteer additionally.
High-quality CPR is only achievable by setting high standards in CPR education, and our study found that with more than 6 participants per instructor, the quality of teaching decreases. Failure to detect and correct even very obvious BLS quality errors with higher group sizes is one result of our study. We therefore recommend an instructor-to-participant ratio of 1–6.
The strengths of this study are that each instructor was only included once to minimize the bias of a learning effect and that all included CPR instructors were educated to the same high level of teaching, resulting in a homogenous group of study participants. A limitation of this study is the simulated teaching setting, which created an artificial environment for examination of our primary outcome. We don’t know if the study’s results would differ if the study were to be performed during real-life CPR courses. However, we chose not to pursue that option because it would expose real BLS course participants to possibly suboptimal training in larger groups, which we considered unethical. Because this was a single-centre study and we focused on CPR training, our results may only be extrapolated to other settings with caution, and their generalizability may be limited.
In conclusion, this randomized controlled simulation trial demonstrates that a BLS instructor is able to detect 80% of all errors made during CPR training when group size is limited to six or less participants. Logistic regression confirms a linear relationship between success rate of error detection and group size. We therefore recommend a maximum instructor-to-participant ratio of one to six for cardiopulmonary resuscitation courses.
All authors have made substantial contributions to all of the following: (1) the conception and design of the study, or acquisition of data, or analysis and interpretation of data, (2) drafting the article or revising it critically for important intellectual content, (3) final approval of the version to be submitted.
Conflicts of interest
SN is the current education representative of the European Resuscitation Council’s “young ERC”.
RG is the Board Director of Training and Education in the European Resuscitation Council and Chair of the ILCOR Task Force on Education, Implementation and Team.
SH, LT, MH, and KP report no conflicts of interest.
This project was funded by an institutional research grant of the Department of Anaesthesiology and Pain Medicine, Bern University Hospital, Inselspital, University of Bern, Bern, Switzerland .
This project was supported by an institutional research grant of the Department of Anaesthesiology and Pain Medicine, Bern University Hospital, Bern, Switzerland, awarded to Dr. Sabine Nabecker. The authors would like to thank Thora Ottenhausen and Carl Conrad (both doctoral thesis students) for their support in conduct of the study. Furthermore, the authors thank Sophie Hugger and Thilo Schweizer (both doctoral thesis students) for their assistance in data control. Additionally, the authors thank all participating Life Support instructors of the Bern Simulation and CPR-Centre at the Bern University Hospital, Bern, Switzerland, as well as all medical students and study nurses who worked as simulated BLS course participants during the conduct of the study. Last but not least we thank Jeannie Wurz for the critical review of the English in the manuscript.