Out-of-hospital cardiac arrests (OHCAs) are associated with poor survival and neurological outcomes [1]. Prognoses are improved when bystanders promptly recognize cardiac arrest and initiate cardiopulmonary resuscitation (CPR) [2]. Of note, many laypeople lack the ability to identify OHCA or act appropriately. The American Heart Association and European Resuscitation Council advocate compulsory annual CPR training for individuals aged ≥12 years [3]. Generally, students receiving formal education (ie, middle school and high school students) are aged between 12 and 18 years. Introducing CPR training in schools can equip a large proportion of the country’s population with fundamental lifesaving knowledge and skills. The early introduction of these lifesaving skills during one’s developmental years not only promotes altruism but also increases one’s willingness to help people experiencing OHCA [4]. Such school-based education sessions can prepare both students and teachers in responding to cardiac arrest incidents within schools and the community at large. Moreover, regular refresher training can be arranged easily in schools because most children attend formal education [5]. However, the lack of stringent guidelines gives schools full autonomy to conduct diverse training modalities, some of which are yet to be supported by empirical evidence [6].

Standard CPR training involves didactic face-to-face lessons, skill demonstrations by qualified instructors, and hands-on practice on manikins in small groups [7]. Although this modality has been regarded as the gold standard, there are often not enough qualified instructors for large-scale implementation in schools. Such training requires numerous manikins and equipment, which are costly [8]. Furthermore, standard training consumes substantial class time and impedes adoption by schools [9].

International resuscitation guidelines suggest the incorporation of technology into CPR training as alternatives to standard training [10,11]. Particularly in the age of the COVID-19 pandemic, technology-based CPR training has become increasingly prominent. These interventions are facilitated by digital technology, including instructional videos, web-based learning, computer programs, mobile apps, or advanced manikin software [12]. Many of them use self-directed learning to decrease reliance on the availability of qualified instructors [13]. Technology-based CPR training may also be cost-effective because fewer resources are required [14]. Finally, training duration is kept minimal, which complements hectic school curricula [15]. Hence, there is a tremendous potential for the proliferation of technology-based CPR training among adolescents.

Two systematic reviews were conducted on CPR training modalities for adolescents. Plant and Taylor [16] concluded that all modalities, including technology-based training, improved knowledge and skills. Reveruzzi et al [17] added that qualified instructors, videos, and hands-on practice improved training outcomes. Both reviews had broad eligibility criteria and no restrictions on study design. This contributed to heterogenous results that prevented pooling of training effects using meta-analyses. Other systematic reviews involving technology-based training focused on health care students, health care professionals, and adult laypeople [18-20]. The conclusions from these reviews cannot be generalized to adolescents because different teaching approaches might be necessary to cater to learners of different age groups [10]. This review used meta-analysis to synthesize evidence on the effectiveness of technology-based CPR training compared with standard training in improving the skills and knowledge of adolescents aged 12 to 18 years.

Searches were conducted in PubMed, Embase, Cochrane Library, Ovid MEDLINE, CINAHL, PsycINFO, Education Resources Information Center, ProQuest Dissertations and Theses Global, and Scopus from inception to June 25, 2021. The search terms included adolescent*, schoolchild*, student*, cardiopulmonary resuscitation, basic life support, train*, and teach*. Synonyms were combined with the Boolean operator OR. Population and intervention concepts were then combined, such as adolescents AND CPR AND training. Refer to Multimedia Appendix 1 for the database search strategies. ClinicalTrials.gov and the World Health Organization International Clinical Trials Registry Platform were searched for ongoing and unpublished trials. Hand searching of the Resuscitation journal was performed for articles published between January 2000 and June 2021. The reference lists of relevant trials and systematic reviews were screened to identify additional studies.

Randomized controlled trials (RCTs) were included if they met the following criteria: (1) participants were adolescents aged 12 to 18 years; (2) participants received CPR training that included technology-based components such as videos, web-based learning, computer programs, mobile apps, or manikin software with real-time feedback; (3) technology-based CPR training was compared with standard CPR training (without the technology-based intervention component); and (4) the RCTs reported CPR skills or knowledge. CPR skills are defined as the ability to perform CPR techniques objectively measured via manikin software or as evaluated by instructors. Theoretical knowledge scores are measured by self-reported instruments, including multi-item questionnaires or multiple-choice–question tests (Table 1).

All retrieved records were imported into EndNote X9 (Clarivate) for deduplication. Titles and abstracts of records were screened by 2 independent reviewers (AL and WX) for relevance. After removing irrelevant records, full texts of potential studies were independently assessed for eligibility. Discrepancies were resolved through discussion with a third reviewer (BS).

AL and WX collected data independently using a standardized data extraction form. Extracted data included publication year, country, study design, setting, participants, sample size, interventions, comparators, outcome measures, and instruments. Posttraining and retention data were extracted, with retention defined as at least 4 weeks after training. Indicators of trial quality were also extracted; for example, attrition rate, intention to treat, and trial registration. Results of studies reported in >1 publication were extracted as 1 study. Authors were contacted when data were incomplete or unclear. Discrepancies in extracted data were resolved through discussion with BS.

AL and WX performed quality appraisal independently for all included studies. Discrepancies were resolved through discussion with BS. The Cochrane risk-of-bias tool was used to appraise studies for risks of bias [22]. The Grading of Recommendations Assessment, Development, and Evaluation approach was used to appraise certainty of evidence for the main outcomes [23]. Ratings were categorized as high, moderate, low, or very low.

Meta-analyses were performed with Review Manager (The Cochrane Collaboration) and presented as forest plots where appropriate. The Mantel-Haenszel approach and risk ratio (RR) were selected for dichotomous outcomes, whereas the inverse-variance approach pooled mean differences (MDs) for continuous outcomes. Continuous outcomes measured using different scales were presented as standardized MDs or Cohen d. When mean and SD were not reported, values were estimated using median and IQR. Overall intervention effects were interpreted using the Z statistic, with level of significance set at P≤.05.

Heterogeneity was evaluated using the Cochran Q test and I² statistic, with level of significance set at P≤.10. A random-effects model was used because of variation in training characteristics [24]; for instance, the studies used different modes of technology-based instruction as well as different types of CPR instructors such as health care professionals, schoolteachers, or medical students, which may affect effect sizes across the studies. For meta-analyses with significant heterogeneity and with at least 6 comparisons, sensitivity or subgroup analyses were performed [25]. The subgroup analyses explored potential effect modifiers, including instructor guidance, hands-on practice, and training modalities. Sensitivity analyses were conducted when meta-analyses yielded significant heterogeneity that was attributable to an outlier study. Funnel plots were not performed because of the limited number of trials included in each meta-analysis. Where quantitative analysis could not be determined from the meta-analysis, findings were presented narratively.

The preparation of this paper did not involve primary research or data collection involving human participants; therefore, no institutional review board examination or approval was required.

The search process is illustrated in Figure 1. Seventeen RCTs were included in this review. Table 2 summarizes the characteristics of the included studies. Of the 28 intervention arms included in this review, 4 (14%) [26-29] were excluded on account of irrelevance. Studies were conducted across 11 countries: Belgium, Canada, Iran, Italy, South Korea, the Netherlands, Spain, Sweden, Turkey, the United Kingdom, and the United States. A total of 5578 secondary school or high school students were recruited (sample sizes ranged from 62 to 1426). Of the 17 studies, 6 (35%) excluded students with prior or recent CPR training [26,27,30-33].

All studies adhered to national or international resuscitation guidelines, except for the study by Rezaei et al [33], which did not mention them. Only the study by Iserbyt et al [31] used a multimedia learning theory to guide the intervention. Training was either self-directed [27-31,33,37,39] or instructor guided [26,29,30,33-36,38,40-42]. Trained schoolteachers or medical students served as instructors or facilitators in 29% (5/17) of the studies [29,31,33,41,42]. The interventions comprised video instruction [26-28,31,33,40,42], computer programs or mobile apps [29,30,36-39,41,42], or real-time feedback [32,34,35]. Of the 24 intervention arms, 4 (17%) omitted hands-on practice on manikins [28-30,33]. All studies reported up to 2 training sessions over a span of 3 weeks, with each session lasting from 1.5 minutes to 4 hours. The length of follow-up ranged from 2 months [28,29] to 1 year [34].

The standard training used included face-to-face qualified-instructor–led demonstration, with hands-on practice on manikins. Other comparators included static pictures [31], classroom-based video instruction [31,34,38,41], or didactic teaching without hands-on practice [42]. Refer to Multimedia Appendix 2 for details.

Most (12/17, 71%) of the studies had overall moderate risk of bias (Figure 2). There was low risk of selection bias in 47% (8/17) of the RCTs because of adequate random sequence allocation and in 12% (2/17) because of allocation concealment. Because of the nature of CPR training, blinding of participants and personnel was not possible in all of the trials. Of the 17 RCTs, 7 (41%) had low risk of detection bias because of the blinding of outcome assessors and another 4 (24%) trials had all outcomes objectively measured through manikins, minimizing bias attributable to the lack of blinding. Of the 17 studies, 13 (76%) had low risk for attrition bias because of similar reasons for attrition across the groups or no missing data. Although only 12% (2/17) of the studies published protocols, 59% (10/17) reported all outcomes completely and were thus rated low risk for reporting bias. Certainty of evidence was appraised as very low for skills and knowledge using the Grading of Recommendations Assessment, Development, and Evaluation approach (Multimedia Appendix 3). Domains were downgraded because of high risks of bias, statistical and methodological heterogeneity, small sample sizes, and wide CIs.

Meta-analyses performed for CPR skill components are summarized in Table 3.

The strengths of this review include an extensive search in multiple bibliographic databases and comprehensive synthesis of results. However, the review was limited by the low quality of the included studies. Most (16/17, 94%) of the studies inadequately reported or took measures to reduce selection and performance biases. In addition, variations in intervention designs across the studies increased heterogeneity; for instance, videos and computer programs or mobile apps may emphasize theoretical knowledge, whereas interventions involving real-time feedback focused on CPR skills. Furthermore, several (11/17, 65%) of the technology-based interventions involved active interaction and engagement with students, whereas others (7/17, 41%) involved passive learning through videos. These variations made it challenging to draw conclusions on training elements required for optimal effectiveness. Finally, this review only included trials published in English.

More high-quality RCTs with clear descriptions of study procedures—for example, allocation concealment and blinding of participants and personnel—are needed. These efforts will improve the credibility of evidence and contribute to stronger conclusions on the effectiveness of technology-based training for adolescents. Future studies should consider incorporating learning theories to guide their interventions [20]. Technology-based training can be considered a routine refresher training modality in schools for future research.

Overall, technology-based training demonstrated equivalence or improvements in skills and knowledge after training and at retention when compared with standard training among adolescents. From an educational perspective, the noninferiority of technology-based training offers a desirable alternative to standard training. Schools can consider using videos, computer programs, or mobile apps for self-directed theoretical instruction. However, instructor guidance and hands-on practice are still necessary components of training. Real-time feedback devices may also be used to increase students’ compliance to resuscitation guidelines. Such a blended learning approach, comprising technology-based resources and face-to-face teaching, may reduce class time and reliance on instructor availability and increase schools’ capacity for wider training dissemination. Furthermore, refresher training should focus on challenging skills; for example, compression depth and ventilation.

This review explored the use of technology-based training as an alternative to standard CPR training among secondary school and high school students. Our findings supported the use of technology-based components such as videos, computer programs, or mobile apps for self-directed theoretical instruction; these components potentially improve skills and knowledge retention. However, instructor guidance, hands-on practice, and real-time feedback are still necessary components of training to achieve better learning outcomes for adolescents. Such a blended learning approach may reduce class time and reliance on instructor availability. Regular refresher training is necessary for challenging skills such as compression depth and ventilation. Caution must be exercised when interpreting the results of this review because of the high heterogeneity of intervention characteristics. The overall low quality of evidence indicated the need for high-quality RCTs with large sample sizes and follow-up data.