This study used a prospective observational design, and participants were recruited from the AED Donation Project, also called the Love GOGO program, implemented by Chang Gung Memorial Hospital, Taiwan. The Love GOGO program aims to establish an educational training system for CPR and build a comprehensive teaching database encompassing participants’ attributes, learning models, and CPR parameters. Individuals from government agencies, nonprofit organizations, schools, and organizations required by current Taiwanese regulations to have AED facilities participate in this education and training program. These include transportation hubs, large long-distance vehicles, tourist spots, schools or large assembly places, large leisure places, large shopping malls, hotels, large public bathhouses, hot springs, and public service sectors such as police stations. These organizations voluntarily participated in the Love GOGO program and proactively contacted the research assistant (YTK) of this study. For this study, participants were enrolled in the Love GOGO program from January to December 2017. Based on our previous study, both traditional and blended teaching models showed a noticeable decline in skill retention after approximately 6 months [12,15]. In this study, mandatory retraining was administered every 6 months or 1 year (Figure 1), spanning a comprehensive training regimen conducted over 2 years. In the initial training phase, the participants were assigned to either traditional teaching or blended teaching modes. Learning effectiveness assessments were conducted every 12 months, with a retraining frequency of 6 or 12 months. Before refresher courses but following initial training, each group underwent evaluation at 12 and 24 months. The results of the 12-month learning effectiveness assessment were disclosed only at 24 months. The research assistants independently allocated training methodologies and the frequencies of subsequent follow-up assessments, using unit convenience and considering the practicalities of the study context. Those responsible for the execution of course training and assessments were not involved in the allocation process.

The inclusion criteria are described as follows: (1) aged at least 18 years and (2) not having undergone any CPR training within the preceding 2 years. Individuals who had physical limitations preventing them from kneeling to perform CPR, who were pregnant, or who were unwilling to sign the informed consent form were excluded from this study. Before initial training, the research assistant divided the participants into groups, and their basic characteristics—namely age, sex, educational level, exercise habits, whether they were receiving CPR training for the first time, their most recent CPR training, and their basic life support (BLS) knowledge scores—were collected through a web-based survey. The assessment of CPR learning should encompass the status of both knowledge and skills. After initial training but before refresher training, we collected data regarding BLS knowledge, skill tests, and CPR quality at the scene at 12 and 24 months. The BLS knowledge and skill tests received approval from the Chairman of the Taiwan Society of Emergency Medicine and have also been published in previous studies [12,15] (Multimedia Appendices 1 and 2).

This study was approved by the institutional review board of the Chang Gung Memorial Foundation (approvals: 201600149B0, 201900399B0, 202200559B0, CMRPG1M0081, and CMRPG1N0081), and this study was performed in accordance with relevant guidelines and regulatory requirements. The IGOGO database is anonymized or deidentified, and no type of compensation is provided to participants. Written informed consent was obtained from all the participants (Figure 2).

An appropriate sample size for this study was estimated based on a pilot study, in which the expected percentage of correct compression depth was 65.4 (SD 29.5) cm for traditional training. To achieve a statistical power of 90% by using a 2-tailed t test with a significance level of P<.05, each group was required to have 225 participants. We planned to enroll at least 900 participants in total.

The Love GOGO program offers 2 teaching models for CPR training: the traditional instructor-led, classroom-based model and the blended model. In the traditional model, participants undergo a 90-minute session, which includes a 60-minute CPR knowledge education session involving a CPR lecture and demonstration, an AED use demonstration, an introduction to relevant laws, and a 30-minute hands-on practice session focusing on compression-only CPR. The blended program, which was approved by the Chairman of the Taiwan Society of Emergency Medicine in 2016, combines an 18-minute e-learning module with a 30-minute hands-on session for compression-only CPR. The e-learning module comprises a video that covers essential knowledge related to CPR and AEDs, including knowledge related to cardiac arrest scenes, the technique of compression-only CPR, the benefits of using CPR and AEDs for OHCA treatment, CPR and AED use steps, and an introduction to relevant laws. In this study, the participants assigned to the blended program were granted access to the e-learning video 3 days before the hands-on session. After completing the e-learning module, the participants practiced their skills in a 30-minute instructor-led, hands-on session in a classroom setting. Both CPR training programs were conducted by AHA instructors who were also emergency physicians. For hands-on CPR practice, both groups used sensor-equipped manikins (Resusci Anne with QCPR, Laerdal Medical AS). The participant-to-manikin-to-instructor ratio per class was 6:3:1, involving 4 instructors and 6 examiners. The study team provided different certification learning stimuli (traditional and blended learning) at 2 frequencies: every 6 months (at 6, 12, 18, and 24 months) and every 12 months (at 12 and 24 months). To establish groups with unique frequencies, the research assistant (YTK) conducted allocation during the initial training phase. Therefore, the traditional teaching model was applied for initial training, and certification sessions for retraining occurring every 6 or 12 months were conducted using either the blended retraining model (18-minute e-learning module with self–hands-on practice for compression-only CPR) or the on-site retraining model (30-minute instructor-led, hands-on session). These groups were called Mixed6 (initial traditional training and 6-month blended refresher training), Traditional6 (initial traditional training and 6-month traditional refresher training), and Mixed12 (initial traditional training and 12-month blended refresher training). For the Blended6 group, initial training was conducted using the blended teaching model, and for certification stimuli every 6 months, the blended retraining model was applied (Figure 1).

This study systematically assessed the participants’ CPR knowledge and performance at multiple time points. Initially, the CPR knowledge and performance of the participants were assessed immediately after training. Following initial training but before refresher training, subsequent evaluations of knowledge and performance were conducted at 12 and 24 months. CPR knowledge was examined through a written test comprising 15 multiple choice questions, with a maximum total score of 100. CPR performance was assessed through 2 methods: examiner-rated assessment and manikin feedback. Individual examiners meticulously assessed the participants’ ability to execute the BLS sequence, encompassing tasks from verifying scene safety to using an AED, with a maximum total score of 40. Objective assessment data regarding CPR quality—including compression depth, compression rate, and full chest recoil—were collected from manikin feedback. The assessment adhered to the 2015 AHA guidelines update for CPR and emergency cardiovascular care; high-quality CPR was characterized by the following three criteria: (1) achieving a compression depth of 5-6 cm, (2) maintaining a compression rate of 100-120 beats per minute (bpm), and (3) facilitating complete chest wall recoil of >80%. Notably, because of the focus on compression-only CPR, ventilation was excluded because it was therefore beyond the scope of the assessment in this study. The primary outcome measure was the comparison of high-quality CPR among the 4 groups. Secondary outcome measures were differences in the percentage of full chest recoil, the percentage of compressions delivered with adequate depth (5-6 cm), the percentage of compressions delivered at an adequate rate (100-120 bpm), written test scores, and examiner-rated skill test scores.

Descriptive statistics are expressed as mean (SD) for continuous variables and as counts and percentages for categorical variables. Linear regression analysis was conducted to determine any differences in the mean values of baseline characteristics among the groups, with adjustment for control variables—namely age, sex, educational level, exercise habits, whether CPR training was being received for the first time, most recent CPR training, and pretest BLS knowledge scores, which were based on the significance test result and which were proposed in previous research [12,15,20]. After allocation, differences in characteristics among groups were observed. To mitigate potential biases introduced by this allocation method, we applied multiple linear regression analyses and generalized estimating equation (GEE) to adjust for these variations when evaluating outcomes (Multimedia Appendices 3 and 4). The chi-square test was used to assess the differences in proportions among the groups, and the general linear model, such as analysis of covariance, was used to test differences among the groups. The control variables—namely age, sex, educational level, exercise habits, whether CPR training was being received for the first time, and pretest BLS knowledge scores—may have influenced skill retention and test scores. Therefore, the model was adjusted for these variables.

We conducted the assessments of the participant’s skill levels and BLS knowledge scores at multiple time points. Accordingly, we used a GEE to examine changes over time in CPR ability indicators among the groups. This allows us to comprehend the changes in CPR skills among trainees under different training methods, using a GEE model to analyze the change over time in CPR ability indicators among groups. The GEE analysis was adjusted for the control variables. To ensure fairness, statistical analysis was conducted using data obtained at time points specific to each group. That is, only data from the postinitial training (baseline), 12-month, and 24-month assessments were included in the analysis.

CPR performance is displayed by line charts, bar charts, and radar charts. In particular, we generated radar charts to illustrate the relative CPR performance in each session. The scores were converted using percent ranking, and the average score was then calculated to represent the performance of each skill for each training method. Statistical analysis was conducted using SPSS Statistics (version 26.0; IBM Corp) and STATA (MP 16.0; Stata Corp LLC).

A total of 1163 participants were recruited for this study, and they were allocated to 4 training groups. The mean age of the participants was 41.82 (SD 11.6) years, and 62.3% (n=725) of participants were female. In this study, 332 (28.5%), 270 (23.2%), 258 (22.2%), and 303 (26.1%) participants were placed in the Mixed6, Traditional6, Mixed12, and Blended6 groups, respectively. Table 1 displays the baseline characteristics of these 4 training groups. As this study was observational rather than randomized, significant differences were observed among the 4 training groups in terms of age (P<.001), sex (P=.008), educational level (P=.006), and CPR training experience (P<.001; Table 1). Notably, the Traditional6 group had the highest average age (45.30, SD 11.39 years) and consisted of 68.9% (186/270) female participants. Additionally, this group had the highest proportion of individuals receiving CPR training for the first time (92/270, 34.1%). However, no statistically significant difference was observed in the BLS pretest knowledge score (P=.11), with an overall mean score of 67.96 (SD 15.08); this finding indicated similar baseline performance across the groups before BLS training.

According to the results of the objective assessment after the first training session, significant differences were found among the 4 groups in skill tests (P=.002), average chest compression depth (P<.001), and average compression rate (P<.001; Table 2) after adjustment for the control variables in the multivariate analysis (Multimedia Appendix 5). In the multivariate analysis, higher skill test scores were associated with younger age (P=.003), higher educational level (P<.001), more previous CPR training experience (P=.04), and higher BLS pretest scores (P=.004). Furthermore, the average compression depth was significantly associated with age (P=.02) and sex (P<.001), and the average compression rate was significantly associated with educational level (P=.04) and CPR training experience (P=.02). Although the mean chest compression depths differed among the 4 groups, the proportion of participants achieving the correct chest compression depth did not differ on average (P=.11). For the overall performance assessment, the proportion of participants achieving high-quality CPR ranged from 27.4% (91/332) to 32.3% (98/303). The lowest proportion was observed in the Mixed6 group, and the highest proportion was found in the Blended6 group. In the multivariate analysis, high-quality CPR was negatively correlated with the Mixed12 training method (adjusted odds ratio 0.65, 95% CI 0.45-0.93; P=.02; Multimedia Appendix 6).

Multimedia Appendix 7 provides the descriptive statistics for the posttraining follow-up data. The results revealed that the Mixed12 group exhibited consistent BLS knowledge scores at baseline (postinitial training), with the highest average scores observed at 12 and 24 months after training. The Traditional6 group exhibited the highest average scores on the skill test at all 3 measurement time points. Figure 3 illustrates the estimated mean scores of BLS knowledge and skill tests for each group, as assessed over time using GEE models. At 12 months after initial training, the Traditional6 group had the lowest average BLS knowledge score (mean 70.10, SE 0.854), which was significantly different from that of the Mixed12 group (mean 75.14, SE 0.762; Figure 3A presents a nonoverlapping 95% CI). Subsequently, at 24 months following initial training, the Mixed12 group exhibited significantly higher scores (mean 79.32, SE 0.741) compared with the other groups.

Furthermore, at baseline, a notable difference was observed in the average scores of the skill tests between the Mixed6 and Traditional6 groups (P=.003; Figure 3B shows a nonoverlapping 95% CI). Moreover, in the follow-up assessment, the Traditional6 group exhibited significantly higher scores than the other groups. Table 3 presents the proportion in each group for the achievement of high-quality CPR. At 12 and 24 months after initial training, this proportion in the Mixed12 group exhibited the most substantial decrease compared with those at 12 and 24 months after training. At baseline, no substantial differences were observed in these proportions among the 4 groups. However, no substantial differences were observed among these proportions among the Blend6, Mixed6, and Traditional6 groups at 12 or 24 months after initial training. We concurrently used multiple linear regression and GEE models to examine the performance indicators; the corresponding results are provided in Multimedia Appendices 5, 6, 8, and 9.

We used an alternative method to rank the 4 training methods based on objectively evaluated items. The scores were converted using percent ranking, and the average score was then calculated to represent the performance of each skill in each training method. Subsequently, we visualized the results as a radar chart (Figure 4). Overall, the 4 groups exhibited comparable average performance in the tests after the first training session. However, in the follow-up assessment (ie, 12 and 24 months after training), differences emerged among the groups (Multimedia Appendix 10). The Traditional6 group exhibited outstanding performance in the skill test and correct recoil rate. The Blended6 group demonstrated superiority in correct depth rate, whereas no significant difference was observed between the Blended6 and Traditional6 groups in terms of correct compression rate or high-quality CPR achievement. The Mixed12 group exhibited a lower correct recoil rate, compression rate, depth rate, and skill test performance compared with the other 3 groups.

This study provides 3 major findings regarding the effectiveness of traditional and blended training methods for CPR education. First, no significant difference was observed in knowledge acquisition after initial training, and all the training groups exhibited proficient CPR skills that met the requirements for high-quality CPR. However, a higher proportion of participants receiving blended training initially achieved high-quality CPR; this finding served as the basis for our comparative analysis. The second major finding highlights the importance of timely retraining. When retraining was conducted 12 months after initial training, significant decreases were observed in the proficiency of CPR skills and the proportion of participants achieving high-quality CPR. Our third major finding suggests that more frequent retraining could maintain CPR skills more effectively. The participants who underwent retraining every 6 months exhibited slight decreases in their proficiency in CPR skills and their achievement of high-quality CPR. Additionally, we explored the potential of web-based self-directed learning as an alternative, and this learning method demonstrated effectiveness for skill retention regardless of the initial training method (traditional or blended), with no significant difference observed between the 2 methods.

Research has demonstrated that blended learning and traditional CPR methods [19,21,22] are practical and reasonably effective alternatives to traditional CPR training; however, large-scale comparisons of these methods or the integration of these instructional methods into CPR education have not been conducted. To the best of our knowledge, this study was the first study to demonstrate that blended learning and retraining stimuli are not inferior to traditional methods when it comes to CPR performance. Chien et al [15] found that blended learning for CPR training does not have inferior learning outcomes relative to traditional methods but that CPR skills at 6 months did not meet the AHA’s CPR guidelines. This finding was consistent with our findings. Although traditional instruction may lead to slightly more favorable performance initially, providing self-directed blended learning stimuli every 6 months is effective for maintaining CPR skills. We found that among learners who received CPR training every 12 months, the performance of high-quality CPR decreased by 35% more than that of those retrained every 6 months. Therefore, consistent with previous research recommendations, stimulating learning every 6 months appears to be favorable to doing so every 12 months. This observation aligns with the AHA’s 2020 guidelines, which suggest that for the general public, the use of convenient learning methods alongside retraining is a viable alternative to traditional face-to-face CPR training.

The blended learning method used in this study offers considerable economic benefits and is time saving for both learners and instructors. By incorporating 18 minutes of web-based learning and self-training into a course, the face-to-face instruction and relearning time were collectively shortened by approximately 72 minutes initially and by 12 minutes in subsequent training. These decreases reduced the expenditure, human resources, and time requirements for learners and instructors in CPR training courses [21]. One study investigated the cost-effectiveness of blended learning for CPR training; the results revealed that blended learning decreased training costs while achieving similar maintenance of CPR skills relative to the traditional method [23]. However, some researchers have indicated that despite the costs and time reductions offered by blended learning, such learning does not ensure that participants will acquire further professional knowledge and proficiency in a demanding training environment [22]. The maintenance of CPR skills contributes to the willingness of the public to perform CPR. When EMSs are activated, guiding individuals to identify cardiac arrest and to implement CPR with dispatcher assistance is challenging as is ensuring that members of the public are able to perform high-quality CPR [24]. Accordingly, blended teaching and retraining models, which appear to be as effective as traditional learning models, can address the challenge of instructing individuals during emergency calls. The characteristics of blended teaching models, including time saving and environmental efficiency, can be beneficial for promoting CPR education among the public and for addressing challenges in maintaining CPR skills among the public.

In this study, 95.1%% (1106/1163) of the participants were high school graduates who were approximately 40 years old and who exhibited higher learning and web-based operating abilities. This demographic advantage likely contributed to the success of blended learning in this study. Moreover, this study used a participant-to-manikin ratio of 2-3:1, leading to higher costs compared with the traditional method (1 manikin to 6 students). The increased investment in training infrastructure may affect the overall cost-effectiveness of blended learning in various settings. The study did not record the frequency of learners’ usage of blended relearning stimuli; the effectiveness of self-paced web-based learning may be related to the time spent engaging with the material. Nevertheless, the primary objective of blended web-based learning is to enable individuals to learn at their convenience. In contrast to traditional face-to-face classroom learning, in blended learning, participants have the flexibility to arrange their web-based and in-class training according to their convenience and location. Accordingly, this learner-centric approach can lead to an environment that is more conducive to the maintenance of CPR skills.

In this study, favorable exercise habits and previous CPR learning experiences enhanced the effectiveness of CPR training. Even if learning had occurred more than 2 years previously, blended CPR training could effectively maintain CPR skills. Ettl et al [20] found that incorporating CPR learning into fitness exercise training increased learners’ motivation and confidence in performing CPR. Therefore, establishing exercise habits helps maintain CPR skills and for fostering rescue skills.

Finally, although blended learning with a retraining frequency of 6 months demonstrated significant economic benefits and time-saving ability in this study, its cost-effectiveness depended on factors such as participant demographics, the training environment, and the level of engagement with web-based learning opportunities. Accordingly, consideration of these factors could maximize the potential of blended learning in various CPR training scenarios.

This study had some limitations. First, in observational studies, the random allocation of samples is infeasible and could result in disparities between groups. Consequently, we used a multivariate regression model to mitigate the impact of variables; thus, we impartially assessed the differences between the groups. Moreover, this study involved tracking the training status of each group to understand the importance of the interval between retraining sessions and whether the given training method was appropriate. Second, we collected demographic data from a subset of learners, but our comprehension of these learners’ economic backgrounds and technology use was limited; consequently, whether blended learning is effective among individuals with relatively low socioeconomic status should be further explored. Third, our research cohort lacked the representation of older adults. As a result, uncertainties persist regarding the applicability of blended training for this demographic; accordingly, future studies are recommended to address this crucial gap. Finally, the absence of an analysis of the participants’ willingness to perform CPR leaves a significant gap in our understanding. Accordingly, individuals’ willingness to administer CPR after blended retraining should be investigated in future research.

Blended learning for CPR with a retraining frequency of 6 months provides higher retention of high-quality CPR skills than does retraining every 12 months. Notably, the blended method demonstrated effects similar to those of traditional relearning methods.