This study was a post-test only control group design to evaluate the effectiveness of flipped learning compared to conventional learning for CPR.¹⁷ The study was conducted at the medical faculty of the University of Fukui, Japan.

Study participants were recruited from the 110 registered first-year medical students at the medical faculty of the University of Fukui, Japan. All students have agreed to participate in the study. The students knew this assessment of CPR performance would not be considered in their grades.

The 110 students in the first-year medical school class were then randomly assigned to either the conventional learning group or the flipped learning group. Both groups had one participant who did not attend. Thus, 108 participants were included in this study. Thirteen participants (12.0%) did not attend the 6-month follow-up evaluation. Overall, 50 students in the conventional learning group (92.6%) and 45 students in the flipped learning group (83.3%) have completed the study.

There were no significant differences in the demographic characteristics including age, gender, any experience of recent CPR training (defined as CPR training in the previous 2 years) and actual CPR, their own or family's history of heart disease and family history of sudden cardiac death between the two groups (Table 1). Two participants in the flipped learning group (3.7%) reported that they did not watch the video at all. Nevertheless, they were included in this study. Thirteen participants who did not complete the study were not significantly different in the demographic characteristics from the participants who completed it in both groups.

The risk of harmful events for the participants was low since our study was a simulation-based study. Consent was obtained from all of the research participants, and it was explained that their participation in this study was completely voluntary. The study protocol was approved by the Ethics Committee of the University of Fukui Hospital.

The sample size was calculated based on the number of chest compressions. It is known that a chest compression rate of 100 to 120 per minute is optimal.^18,19 In this study, the 2-minute test period included the initial assessment in order to simulate a real setting. We estimated that the actual time for chest compressions would be 1 minute, eliminating the time for the initial assessment and mouth-to-mouth ventilation from previous studies.^20,21 Therefore, we assumed that the clinically important difference in the number of total chest compressions was 20. Under the condition of an alpha error of 5% and a power of 80%, 36 participants per group were needed. Expecting a 20% dropout rate, the necessary sample size was estimated to be 90 participants in total.

The participants' skills were assessed immediately after the conventional and flipped learning and 6 months later. Each participant was evaluated individually for 2 minutes according to a standardized testing protocol including an initial assessment, chest compressions and mouth-to-mouth ventilation by examiners who were blinded to the group assignment. All of the examiners were certified, and trained instructors of the ICLS course, and the authors were not involved in the evaluation.

After the 2 minutes of CPR evaluation, the AED was brought to the participant, and their AED operation skills were evaluated. All CPR assessments were completed on the Resusci Anne mannequins connected to a SimPad SkillReporter. Before the first evaluation of CPR performance, each of the participants was asked to complete a questionnaire that included their age, sex, recent CPR training, any experience of actual CPR, their own or family's history of heart disease and their family's history of sudden cardiac death. The participants in the flipped learning group were also asked how many times they had watched the video.

Our primary outcome measures were the time to first chest compression and the number of total chest compressions performed during the 2-minute test period 6 months after training. The time to the first chest compression was defined as time from the start of CPR evaluation till the first chest compression performed in the evaluation. These were recorded by the Resusci Anne mannequins and the SimPad SkillReporter.

The secondary outcome measures were the time to the first chest compression and the number of total chest compressions performed during the 2-minute test period immediately after the training. Both immediately and 6 months after the training session, we evaluated the other outcome measures, including the hands-off time, percentage of correct hand positions, chest compression rate, average chest compression depth, percentage of appropriate chest recoil, percentage of appropriate chest compression depth, the number of total ventilation, average ventilation volume, percentage of appropriate ventilation volume and ventilation rate. These outcome measures were all recorded by the Resusci Anne mannequins and the SimPad SkillReporter. We also evaluated the time from AED arrival to shock and the key elements of CPR, which included the safe approach, check responsiveness, calling 119 (the emergency phone number in Japan), asking for an AED, breathing check, appropriate ratio of compression and ventilation, appropriate ventilation, correct positioning of defibrillator pads and safe delivery of shock. These were also measured both immediately and 6 months after the training as other outcome measures. The time from AED arrival to shock was measured by the examiners, who gave either an 'adequate' or 'inadequate' grade for the key elements. The appropriate chest compression depth was defined as more than 50 mm, and the appropriate chest compression rate was defined as 100-120 per minute based on the 2010 American Heart Association (AHA) Guidelines.²² Hands-off time was defined as periods without compressions during resuscitation, and the appropriate ventilation volume was defined as 400-700 ml.

The participants were randomly assigned to either the 60-minute conventional learning group or the 30-minute flipped learning group. The randomization scheme was generated by using the web site.²³

The conventional learning group participated in a 60-minute training program consisting of an initial assessment, chest compressions, mouth-to-mouth ventilation, and an automated external defibrillator (AED) operation based on the 2010 AHA Guidelines.²² After the instruction and discussion of CPR for 30 minutes, the participants received hands-on training similarly for 30 minutes. After the training session, all of the participants in the conventional learning group directly proceeded to their testing stations and their performance was assessed by the examiners.

We defined flipped learning in our study as the use of video learning followed by hands-on training. Specifically, the flipped learning group participants were directed to watch an official video provided by the Japanese Red Cross Society on the internet one week before the hands-on training. The video is approximately 15 minutes in length and explains how to perform bystander CPR based on the 2010 AHA Guidelines.²² The participants were instructed not to use other videos or resources in order to learn bystander CPR. On the day of the hands-on training, the flipped learning group participated in a 30-minute training program. The participants and instructors first discussed the video content, and all of the participants' questions were answered by the instructors before the training session started. We assumed that they had learned how to perform CPR from the video; therefore, no lecture was provided by the instructors. Thus, the training program in this group was 30 minutes shorter than the conventional group. They also proceeded to their testing stations immediately after the training.

Six months after the training sessions, all of the participants were invited to a reassessment, and their skills were evaluated by the examiners. This study was conducted as a part of their curriculum, so the participants were notified of the reassessment in advance. These training programs were conducted using Little Anne mannequins and the AED Trainer 2. Each group consisting of five participants was provided one mannequin and the AED trainer respectively. All participants were instructed by the physicians and nurses who were certified instructors of the Immediate Cardiac Life Support (ICLS) course or who had completed the ICLS course and had instruction experience equivalent to the certified instructors. The instructor/participant ratio was 1:5 in the both groups. None of the authors were involved in the instruction. The ICLS course is the most widely conducted course for CPR training in Japan and is managed by the Japanese Association for Acute Medicine.²⁴

Summary statistics were presented as the means with standard deviations or medians with interquartile ranges as appropriate. As most variables were not normally distributed, continuous variables were described using medians with interquartile ranges, and the Mann-Whitney U test was used to assess significant differences. In terms of our primary and secondary outcomes, we also performed Student's t-test if the distribution was normal. Fisher's exact test was used for binary variables. P-values < 0.05 were considered statistically significant. Statistical analyses were performed using SPSS Ver. 22 (SPSS, Inc., Chicago, IL, USA).

There were several limitations to this study. First, this study was conducted as a part of the participants' medical school curriculum, so the participants were notified of the study process in advance, although they were randomly assigned to either group. Second, we used a video already uploaded to the internet and participants did not need to log in to view it. Thus, it was unclear how much or how deeply they learned from the video content. We asked the participants how many times they watched the video and only 2 participants (3.7%) reported that they did not watch it at all. We could overcome this concern by utilizing a log-in system or embedding a quiz or a brief test in the video.^16,28 Third, our choice of participants might have influenced the outcomes of the study. They might be highly motivated because they were studying to become doctors, although they knew the results of CPR performance would not be considered in their grades. Fourth, it is important to note that this is a mannequin study. The mannequins we used are unable to represent humans, so appropriate CPR performance on a mannequin might not necessarily assure appropriate practice of CPR. Finally, our study was first intended to determine if flipped learning was superior to conventional learning. This study was neither a non-inferiority trial nor equivalence trial, so future research, preferably through a non-inferiority trial with more samples, would be needed to prove the non-inferiority of flipped learning over conventional learning in CPR education.

We thank the Japanese Red Cross Society for permission to use their video for CPR training.

The authors declare that they have no conflict of interest.