“Herzensretter” is a teaching concept developed by the German Federal Association for First Aid (BAGEH). In this concept, the students are supposed to participate in 90-minute resuscitation courses once a year over three consecutive school years (see Figure 1). In the first year, the course focuses on hands-only resuscitation, whereas the second one also includes ventilation, and the third one adds in using an automated external defibrillator (AED). Please see the supplemental material for detail on the Herzensretter concept and curriculum.

This study aimed to evaluate the first-year course, therefore the intervention we examined only includes hands-only resuscitation. This first-year course was led by schoolteachers who had been trained in resuscitation and corresponding teaching methods by professional first aid instructors in a six-hour course. This course was based on standardized first aid instructor courses, but more was more concise. The curriculum only included CPR; other first aid topics (e.g., bandages) were not included (see supplemental material). To reduce their workload, the teachers were encouraged to seek assistance from students with first aid training experience. One of the participating schools already had a student group for first aid training that was involved in the courses for the seventh grade. In the other schools, the participating teachers trained a few selected students from the seventh or higher grades to assist. Regardless of the age or experience of assisting students, the class was led and supervised by the schoolteacher. Part of the course consisted of practical training where 2–4 participants gathered around one resuscitation mannequin and were trained by a student aid under supervision by the responsible teacher (see supplement material) (Sick et al., 2018).

At the beginning of 2020, we trained one or two teachers from each of the 33 participating schools in the rural German state of Brandenburg. Originally it was planned to randomly select five of these schools, one from each school district. Due to the COVID-19 pandemic, the project was implemented and evaluated in only five of the participating schools until the end of the field phase (November 2021).

We assessed the students’ knowledge immediately before the course started (baseline) and one month afterward using self-report questionnaires. Immediately after the written assessment, we also evaluated the practical skills of 42 randomly selected students. During the COVID-19 pandemic, the one-month-assessment was only delayed in one school (by five months) whereas in the other four schools, it took place after one month. In this school, no practical assessments took place (see Figure 2).

Ethical approval for this study was obtained from the local ethics committee of Charité Universitätsmedizin Berlin, Campus Virchow-Klinikum (application number: EA2/041/20).

From the five participating schools we included all present seventh grade students whose parents provided written informed consent for the evaluation. Follow-up questionnaire data from students who were not present for the baseline assessment, or the resuscitation course were excluded from the evaluation.

Resuscitation knowledge was assessed using standardized questionnaires including demographic data (age, sex, language spoken at home, number of siblings) and previous first aid experience. No personal information to identify individuals was collected. The baseline and follow-up questionnaires were anonymized and not linked on an individual level. The open questions assessed depth and rate of chest compression and duration to check for breathing. The students were asked to order the four steps of what to do for an unconscious person. The steps were (1) self-protection, (2) check breathing, (3) call the emergency number, (4) perform cardiopulmonary resuscitation.

In the assessment one month after the course, the students were also asked whether they would help (and would perform resuscitation) only for someone familiar to them, or if they would also help a stranger in an emergency. Further, they were asked whether they felt confident in taking appropriate actions in an emergency.

The questionnaire (same for pre- and post-course assessment for knowledge but with extra questions on confidence and willingness post-course) was developed through the collaborative efforts of the responsible department of the Ministry of Education of the state of Brandenburg, health scientists from the medical faculty of the Charité Berlin, and medical professionals including doctors and trainers from emergency services, and organizations that provide first-aid training to adults, youth and children. The survey, just like the course content, was based on guidelines for first aid education and used age-appropriate language adapted to the children’s experience (De Buck et al., 2015; International Federation of Red Cross Red Crescent Societies, 2016). Soon after the start of our study, the Global First Aid Reference Center published a new guideline including an Outcome Measurement Toolkit. Fortunately, we recognized many concepts and elements we used in the new guidelines (Boase et al., 2021; Global First Aid Reference Center, 2020).

For the practical skills examination, the students were randomly selected and assessed individually without the presence of other children. The assessor asked each child to show in a role play with a resuscitation manikin how they would act if they found an unconscious person. The assessors of the practical assessment had a medical background and strictly followed a standard operating procedure developed for this assessment.

Without the presence of other children, the assessor recorded whether the student (i) paid attention to their safety and (ii) checked the manikin for consciousness and breathing. The breathing was assessed as correct if the student tilted the manikin’s head backward, observed the torso, and approached an ear to the manikin’s mouth to hear and feel for breathing for 10 seconds. The assessor also recorded whether the student (iii) called the emergency number, (iv) started resuscitation on their own using the correct chest compression rate and depth, and (v) performed continuous resuscitation until the arrival of medical professionals.

For the statistical analysis, we used IBM SPSS Statistics 28 (Armonk, New York, USA). We followed an explorative approach to examine various research questions and did not test specific a-priori hypotheses and did not adjust for multiple testing. Due to this approach and their limited ability to measure the actual importance of a result, we refrained from using p-values or the term ‘statistically significant’ (Wasserstein & Lazar, 2016).

We did not use a statistical test to conclude that the samples of the theoretical and the practical assessments are similar. In this context, it does not seem reasonable to specify equivalence bounds for an appropriate equivalence test. Therefore, we interpreted these results with caution.

The results of continuous variables were presented using median, minimum, and maximum values of categorical variables with absolute and relative frequencies. To estimate statistical uncertainties, we calculated 95% confidence intervals using bootstrapping with 1,000 samples.

Out of 356 students who completed the baseline questionnaire immediately before the course, 340 participated in the follow-up assessment by questionnaire and 42 students (a randomly selected subgroup from 4 schools) participated in the practical examination one month after the course. Most participants were 12–13 years old (age range 10 to 15 years). Between one-third and half of the students had previously participated in a first-aid course (Table 1).

All participating teachers taught the course on their own with 20–30 students per class. They previously instructed selected students to help supervise the practical training (see supplemental material). In one school, students with previous first-aid training from extracurricular activities supported the teacher. The results of all schools were similar.

To assess knowledge on resuscitation we asked the students about the numerical values of resuscitation rate, depth, and the necessary duration to check for breathing. They were also asked to describe the correct procedure when finding an unconscious person. Although a good proportion of students improved their resuscitation knowledge, more than half of the students did not improve (Table 2).

All participating students spoke German fluently and passed their regular exams in German. However, students who spoke only or partly German at home seemed to perform better than those who spoke no German at home, although the latter subgroup was too small to draw conclusions. Overall, there were no considerable differences between girls and boys (see Table 2).

Most of the students (89%, 95%-CI 79–98%) felt confident to call the emergency number and report relevant information. Regarding performing resuscitation, slightly more than half of the students (59%, 95%-CI 49–69%) felt confident regarding their specific knowledge (see Figure 2).

Most of the students reported that they were willing to help others in emergencies, especially people whom they knew (see Table 3). This attitude did not differ considerably between different sociodemographic groups.

The demographic factors of the 42 randomly selected students were similar to those of the whole study population, except that there were slightly more male students and students who did not speak German at home in the subgroup (see Table 1). The practical results of the subgroup were considerably better than the results of the theoretical assessment of the whole sample (see Tables 2 and 4), especially regarding chest compression rate (69% versus 27%) and depth (71% versus 37%). In the practical assessment, 81% checked for breathing, but only 26% did this completely correctly. The most frequent error was not tilting the manikin’s head backward.

One month after the 90-minute resuscitation course conducted by schoolteachers, depending on the item, up to 25% more students were able to correctly answer questions on relevant resuscitation parameters than before the course. However, about 50% of the participants did not show any improvement in their resuscitation knowledge. This is lower than the 59% of students who indicated confidence in their knowledge. This could be because they were instructed that if a patient requires resuscitation and they are the only bystanders, they should attempt resuscitation even if they have doubts on how to do so. Some students may feel confident to initiate chest compressions while still having uncertainties about the exact parameters. In the practical assessment with a randomly selected subgroup, the students seemed to perform much better compared to the theoretical assessment. It is not clear why this occurred. It may be due to the emphasis on practical training of the course, or because students in this age group might remember movement patterns more easily than numerical values. More than two-thirds of them followed the correct procedure for resuscitation and performed sufficient resuscitation on a manikin. The proportion of children with previous first aid training in the whole sample (37.9%) seemed similar to the proportion of those with such experience in the randomly selected subsample for practical assessment (38.1%). Still, due to the small sample size of students doing the practical assessment, these results should be interpreted with caution.

Students that did not speak German at home seemed to have more difficulty with the material, however we cannot draw definitive conclusions. The sample size of these students was very small; therefore the 95%-confidence intervals were overlapping with the whole study population (see Table 2). As these students were able to speak German fluently, different reasons apart from possible language barriers, for example, social status and integration into the class should be explored in future evaluations. Regarding attitude to help in emergencies, over 80% of the participating students were willing to perform resuscitation on strangers and more than half of them were confident that they would do this correctly.

The strengths of our study include a relatively large sample size with 340 grade seven students. Further, we were able to complete the study during the recent COVID-19 pandemic with several long periods of school closure in Germany. We noted an overall high willingness to perform resuscitation despite the increased awareness of contagious diseases in the context of the pandemic. Almost all invited students agreed to participate in theoretical as well as practical assessments. Thus, our results can be considered representative of the participating schools and their catchment area.

Our study had some limitations. First, the resuscitation courses were planned to take place in 33 different schools, and for their evaluation, we planned to randomly select 5–6 schools from the 4 different districts of the state. However, due to the pandemic, this was not possible. Nonetheless, five schools implemented the resuscitation course. All of them agreed to participate in the evaluation. The schools and teachers implementing the project under such difficult conditions were likely to be the most enthusiastic institutions (and teachers) among all 33 schools that had sent teachers to the special training. Thus, we cannot rule out a selection bias and may have overestimated the benefits of the course. On the other hand, we might have underestimated the benefits of the course, considering that 38% of the participating students had previous first-aid training experience.

Secondly, to respect possible data protection concerns of the parents and schools, we did not collect personal data like names or addresses from the students and collected only anonymous data. Therefore, we were not able to link the baseline with the follow-up questionnaires on an individual level. As the time between both assessments was relatively short and the number of students did not differ much, we assume that this should not have influenced our results considerably. On the other hand, the anonymous approach helped us to include almost all eligible students and achieve a high participation rate. Any differences in individual learning ability that we were unable to detect in this anonymous approach are not relevant to scaling across the breadth of the school landscape anyway, as students’ learning readiness and ability are routinely heterogeneous across classes. Such a teaching concept must be robust in this regard.

Thirdly, as the baseline and follow-up questionnaires on knowledge were identical in design, some students may have memorized the correct answers without any real understanding. The questionnaires were collected a day before the course and not discussed in class. The students did not receive any feedback on the baseline questionnaires. Further, the questionnaires were not part of the material used during the course. Therefore, we do not expect a considerable impact on the follow-up questionnaires.

Fourthly, while the high willingness to help in emergencies that we observed is a pleasing result, in its appraisal social desirability bias must be considered. Some children may be inclined to choose the more socially acceptable answer to feel competent and brave instead of indicating their actual feelings toward the situation. Our study staff informed the students that this questionnaire was anonymous, and that no teacher would have access to the questionnaires. Furthermore, we ensured privacy while the questionnaires were being filled out. We also expressed the importance of accurate statements for a realistic evaluation of the course.

Finally, the very positive results of the practical examination were only based on a small subgroup. While the theoretical assessment was delayed for some students, the subgroup with practical examinations was assessed one month after the course without exception. Due to the smaller sample size, the results of this part of the evaluation may be regarded as less representative. As we conducted an anonymous assessment to achieve high participation rates, we were not able to compare the individual results of the subgroup’s self-report questionnaire results with the whole sample to examine if the subgroup may have also performed better in the theoretical assessment. However, we selected all subgroup students randomly, therefore the results of the practical assessment may suggest that students of this age group learn resuscitation at least as well or better by watching the instructors and imitating their performances rather than by theoretical teaching methods. As the state-wide role out of the Herzensretter concept is planned, further theoretical and practical evaluations of larger samples in different settings are necessary and were recommended to the state ministry of education.