Trial design and oversight
We conducted a Danish, investigator-initiated, stratified, randomised controlled trial comparing two new teaching methods (video- and simulation-based) with the conventional lecture-based method on adult BLS course participants’ ability to recognise different breathing patterns. Participants and investigators were not blinded, except for the trial statistician (AKE) who conducted the pre-planned analyses blinded to the intervention groups.
Participants were randomly assigned to the three groups in a 1:1:1-ratio using computer-generated random allocation sequences with permuted blocks of varying sizes (three and six) stratified by participant type (see section Participants) . Allocation concealment was ensured through the use of sequentially numbered, opaque, sealed envelopes. A person, not involved in any other aspect of the study, prepared both the randomisation sequences and the sealed envelopes.
The Committee on Health Research Ethics in the Capital Region of Denmark waived the need for ethical approval, as this was a teaching project (journal number: 17021633). All methods and experiments were performed in accordance with relevant guidelines and regulations. Participation in the trial was optional and voluntary and all participants gave written informed consent, which could be withdrawn without explanation at any time. The trial was approved by the Danish Data Protection Agency (journal number HGH-2017-131, I-Suite number: 06089). The trial was not publicly registered, as this is not mandatory for studies assessing teaching interventions in healthy volunteers. The trial was conducted in accordance with a pre-specified protocol and statistical analysis plan (available in Danish from the corresponding author upon request). There were no changes to trial outcomes after the trial commenced. The trial is reported in compliance with the Consolidated Standards of Reporting Trials (CONSORT) Statement  (completed checklist included in the Additional file 1).
Three types of adult participants (≥18 years) were included: 1) university students (medical students excluded) from the University of Copenhagen, who had applied for a course through the organisation Student2Student (a voluntary organisation of medical students who teach BLS courses to non-medical university students); 2) conscripts in the Danish Military; and 3) retired elderly people who had applied for a BLS course through the Danish foundation TrygFonden. Participants were excluded if they had ever studied to become a health care professional (paramedic, nurse, physician, or other). The only additional criterion was a satisfactory continuous assessment at an ERC BLS course (4 h) where enrolment took place through the three organisations listed above. All BLS courses were provided free of charge. Participants received a reimbursement of 200 DKK (approximately €27) for their participation.
BLS courses and interventions
Agonal breathing was not mentioned during the 4-h ERC BLS course and was only taught as part of the intervention according to the randomisation. After the course, participants were randomised to receive lecture-based, video-based or simulation-based teaching on how to recognise agonal breathing. All teaching sessions lasted approximately 2 min. The lecture group was taught using the ERC BLS 2015 course materials (a single slide in a lecture) and teaching methods according to ERC Guidelines for Resuscitation 2015 Section 2: Adult basic life support and automated external defibrillation . The video group saw a video in plenary with text and verbal explanation and examples of agonal breathing (see Additional file) using recordings of HM simulating agonal breathing (originally developed for another purpose by Laerdal Medical, Stavanger, Norway; used with permission). In order to standardise the teaching methods, the instructor was not allowed to answer questions or to give examples on agonal breathing in the lecture group or video group. The participants in the simulation group simulated agonal breathing individually with the instructor. The instructor explained and gave some examples on agonal breathing the first 30 s. The following 90 s, the participants simulated with feedback from the instructor. All three interventions are described in more detail in the Additional file.
To minimise differences between the three ERC BLS instructors (NB, AG, and TWJ) responsible for the interventions, they all followed a detailed instructions manual on what to say and how to interact with the participants in all three groups (see Additional file 1). Participants were not allowed to interact with each other between the intervention and the test.
Testing was performed immediately after the course using nine different videos of actors simulating normal breathing, no breathing and agonal breathing (three videos of each). The study group and a study investigator (AL) with extensive experience in medical teaching and BLS courses selected specific video clips for maximum clarity of various aspects of normal and agonal breathing. Videos consisted of 2 female and 7 male actors, all 40–60 years old and wearing T-shirts with high necks to avoid visualisation of the carotid artery while simulating breathing patterns. The videos lasted 30 s each and showed a side profile of a simulated patient’s head and torso. All videos were edited to give the actors a pale colour and to hide their carotid pulse. Three experienced paramedics (see acknowledgements) not otherwise involved in the study or the video production classified the videos with a 100% concordance to the intended categories. For the test, participants were informed that the persons in the videos were unconscious and had patent airways and watched all nine videos in the same order (see Additional file 1). After each video, participants had 10 s to classify the breathing pattern as either: 1) normal breathing, 2) no breathing, 3) agonal breathing, or 4) “do not know”, in order to minimise random guesses. Option 4) “do not know” was also used if participants did not answer or answered after the 10 s between each video, which was considered an incorrect response in the analyses. Participants were under observation by one of the instructors during the entire test. Each video could only be seen once, and participants were not allowed to communicate during the test. The testing procedures are fully accounted for in the Additional file 1. Data were registered on paper-based forms, and all data were electronically entered twice and compared to ensure data quality and avoid transcription errors.
The primary outcome was the participants’ overall ability to recognise all three breathing patterns in terms of the total number of correct breathing assessments out of all nine videos. The secondary outcomes were the participants’ ability to recognise each breathing pattern (normal breathing, no breathing, and agonal breathing) in three videos of each pattern.
Statistical analyses and sample size
The required sample size was calculated using a 5% significance level and 80% power. Assuming that participants in the lecture group would have a mean number of correct answers of 4–5 out of 9 videos, and that the participants in the video group and the simulation group would have a mean of 2–3 correct answers more than the participants in the lecture group, 42–48 participants were needed in each group. This assumption was made according to our best estimate as instructors teaching laypersons using the lecture method. To ensure an adequate sample size and allow for possible withdrawals of consent, we planned to include 50 participants in each intervention group. We offered inclusion to all participants present at each course (maximum of six participants per course) until we reached a minimum of 150 included participants in total. No interim analyses were planned or performed. There were no missing data and thus no need for imputation of missing data.
Baseline characteristics, including gender, age, education and previous BLS experience with an estimate of years since last BLS course, are presented separately for each group and for the total trial population. Categorical variables are presented as frequencies (counts and percentages) and numerical variables as medians with interquartile ranges (IQR) due to non-normality (inspected using histograms and Q-Q-plots), and post-hoc also as categorical variables due to non-normality.
Primary and secondary outcomes are presented as counts and percentages of correct answers and analysed using logistic regression models adjusted for participant type (as described above). The effect of teaching method on recognition of breathing patterns was tested using likelihood ratio tests, and the relative effects of teaching methods were compared using odds ratios (ORs) with 95% confidence intervals (CIs) calculated using the Wald method. In addition to the pre-planned analyses, we conducted a post-hoc analysis of the primary outcome. This analysis was conducted using a logistic regression model including the main effect of teaching method and participant type and the interaction between teaching method and participant type. A likelihood ratio test of the interaction was conducted. The relative effects of teaching methods within participant types were compared using odds ratios (ORs) with 95% confidence intervals (CIs) calculated using the Wald method. We considered P-values < 0.05 and 95% CIs for ORs not including 1.00 as statistically significant. All analyses were performed using SAS version 9.4 (SAS Institute, Cary, NC, USA), with logistic regression analyses performed using the GENMOD procedure.