Quality and accuracy of cardiopulmonary resuscitation teaching in short videos: an analysis across three major short video platforms

Abstract

Objective

Cardiopulmonary resuscitation (CPR) is vital for saving patients experiencing cardiac arrest. Teaching CPR skills through short videos offers numerous advantages. However, potential inaccuracies or misinformation could mislead the public and impact the effectiveness of CPR education. This study aims to evaluate the quality and accuracy of CPR instructional videos shared on three major short video platforms in China (TikTok, Bilibili, and REDnote), analyze common irregular or erroneous practices, and provide valuable suggestions for content optimization.

Methods

The collected videos were evaluated using a five-point scoring criterion based on 2020 American Heart Association (AHA) Guidelines for Cardiopulmonary Resuscitation (CPR) and Emergency Cardiovascular Care (ECC). The videos were categorized into three levels: excellent, moderate, and poor, based on their video quality scores for CPR procedures and non-procedural content. Further categorization was made by video duration (≤5 min as short; >5 min as long). Additionally, the most critical and prevalent irregularities or errors were documented, and a detailed analysis of the most popular video from each platform was carried out. The relationship between video quality and popularity, and video duration and popularity were examined separately.

Results

A total of 100 CPR instructional videos were analyzed. While 86% of the videos were produced by healthcare professionals, substantial errors were identified in critical areas such as the extra time spent removing foreign body airway obstruction (67%), and incorrect hand position during compression (62%). Other issues with non-procedural content were identified, including video acceleration (13%), lack of step-by-step explanations (61%), etc. Statistical analysis revealed no significant differences in popularity across videos of different quality or duration (p = 0.876 among video quality groups for CPR procedures, p = 0.988 among video quality groups for non-procedural content, p = 0.260 between video duration groups).

Conclusions

This study identified the necessity for improvements in CPR procedures and non-procedural content of CPR instructional videos. To enhance video quality, measures such as rigorous review mechanisms, public feedback and promotion of certified high-quality videos are recommended.

Introduction

Out-of-hospital cardiac arrest (OHCA) is one of the major public health challenges that seriously threatens people’s lives and health. In China, there are more than 230 million people with cardiovascular disease, and 550 000 people experience cardiac arrest every year, with a survival rate of less than 1% [1]. Cardiopulmonary resuscitation (CPR) is a critical life-saving intervention for saving lives in cases of sudden OHCA [2]. Proficiency in CPR is essential for improving survival rates, reducing injuries, and strengthening community emergency response capabilities [3,4,5,6]. In the digital age, the dissemination of medical and health knowledge via the Internet has become widely accepted [7, 8]. Short videos, in particular, have become a popular method for CPR skill training due to their efficiency, accessibility, and diversity, visually appealing formats, which improve public access to basic CPR techniques [9,10,11].

However, due to the diverse identities of content creators, varying standards and target audiences, the quality and accuracy of these videos can be inconsistent. Potentially inaccurate or erroneous information contained therein is highly misleading and could lead to the widespread dissemination of misinformation, with negative consequences for public health [12, 13]. Previous studies have identified misleading content within CPR instructional videos [14,15,16,17]. However, comprehensive data and systematic analysis of specific problems and their prevalence in video instruction, as well as a systematic evaluation of the quality of CPR instructional videos on short-video platforms in China, are still lacking.

This study aims to examine the teaching methods used in CPR training short videos on mainstream platforms in China, assess their quality and instructional accuracy, identify common deficiencies, and ultimately provide valuable insights for future improvements in CPR training.

Methods

Study design and setting

The research was an observational, analytical, and descriptive cross-sectional study at Emergency Center, Zhongnan Hospital of Wuhan University, Wuhan, Hubei, China, from July to November of the 2024 academic year. The data were acquired by collecting videos on open short video platforms in China and did not contain any data which could identify the individual. Therefore, ethical approval was exempted.

Video sources

Two evaluators searched for relevant videos using the keyword “CPR 视频” on three of the most popular short video platforms in China: TikTok (https://www.douyin.com/), Bilibili (https://www.bilibili.com/), and REDnote (https://www.xiaohongshu.com/explore) from 25 to 28 July, 2024.

Video selection criteria

The inclusion criterion for the videos was adult CPR instructional videos. Non-instructional videos (e.g., news reports, footage of real-life events, comedic videos, advertisements), videos on pediatric CPR, non-Chinese videos, and duplicate videos were excluded during the screening process. The selection criteria were developed by two evaluators and two instructors.

Video evaluation

Two AHA BLS training instructors developed a comprehensive CPR scoring standard based on the CPR content from the 2020 AHA Guidelines for CPR and ECC [18,19,20]. Each scoring criterion used a five-point scale to evaluate the videos in two key areas: CPR procedures, and non-procedural content (including media attributes, presentation methods, contraindications and complications, resuscitation success rate, and importance of first aid), while also highlighting potentially misleading behaviors that require attention.

For each video, the source, the identity of the content creators, the time the video was published, the duration of the video, and engagement metrics, including the number of likes and bookmarks, were recorded. Videos were classified as short (≤5 min) or long (>5 min) based on their duration. Their quality and accuracy were assessed by two evaluators trained by AHA BLS Instructor (also BLS faculty). Two evaluators rated each video independently and submitted the results to two BLS instructors (also active BLS faculty) for review, resulting in a consensus on the final scores. Based on their scores for CPR procedures (totaling 90 points) and non-procedural content (totaling 25 points), the videos were categorized into three levels: excellent (80–90 points, 21–25 points), moderate (70–79 points, 17–20 points), and poor (≤69 points, ≤16 points). This study also summarized the most critical and frequent irregularities or errors in these videos.

In addition, the most popular videos from each platform, as measured by average daily likes and bookmarks, were selected for detailed analysis.

Statistical analysis

Numbers and proportions for the categorical data were calculated, and means and standard deviations were calculated for measures. Using IBM SPSS Statistics 25 software, we employed the Kruskal-Wallis test to compare differences in popularity among three video quality groups (excellent, moderate and poor) and Mann-Whitney U test between two video duration groups (long and short), with a significance level set at p < 0.05.

Results

A total of 235 videos were retrieved from TikTok, Bilibili, and REDnote. 99 non-instructional videos, 12 non-Chinese videos, 1 pediatric CPR video, and 23 duplicate videos were excluded. Ultimately, 100 videos were selected for evaluation and analysis (Fig. S1).

CPR video characteristics

A review of the basic information presented in the videos indicated that 86% of the content creators were healthcare professionals and 88% of the videos performed CPR simulations inside the hospital. While 94% chose to use a manikin as the demonstration subject, only 2% chose to use both a real person and a manikin. With return of spontaneous circulation (ROSC) serving as an indicator of resuscitation success, the demonstration results indicated successful resuscitation in 62% of the cases, while 38% did not specify the outcome. No cases of unsuccessful resuscitation were reported. In addition, 20% of the videos included automated external defibrillator (AED) instructions (Table 1).

Table 1 CPR video characteristics

Evaluation of CPR procedures

We firstly evaluated the quality and accuracy of the videos based on the CPR procedural standards. The videos excelled in the sections of calling for help, assessing breathing, preparations before compression, chest compression posture, chest wall recoil, continuous compression, and number of chest compressions with high overall scores (all means ≥4.9, Fig. 1A). However, the scores for checking for responsiveness, chest compression position, chest compression rate, opening the airway, and ventilation number were varied significantly. Some videos scored poorly due to irregular or erroneous practices, particularly in checking for responsiveness and chest compression position, where the mean score was below 4.

Fig. 1

Scores on video quality

In details, 67% of the videos showed extra time spent removing foreign body airway obstruction, 66% showed failure to check for responsiveness, and the incidence of incorrect hand position during compression was as high as 62% (primarily failing to consider the effects of age and gender by commonly compressing at the midpoint between the nipples). In addition, a small number of videos showed incorrect compression rate (too fast or too slow), incorrect compression depth (too deep or too shallow), failure to observe chest rise and fall during ventilation, and failure to release the patient’s nose after ventilation, with occurrence rates of 15%, 6%, 10%, and 8%, respectively (Fig. 2).

Fig. 2

Percentage of irregular and erroneous practices

Evaluation of non-procedural content

We also focused on the non-procedural content in these CPR instructional videos and found that the mean score was significantly low in the “resuscitation success rate” Sect. (2.63 ± 0.95) (Fig. 1B).

Further analysis of the deficiencies in the non-procedural content revealed that 86% of the videos provided viewers with subtitles of specific practices, while 4% of the videos had spelling errors in the text descriptions and 9% of the videos exhibited deficiencies in audio clarity. In addition, 61% of the videos only demonstrated a complete CPR procedure without step-by-step explanation. And 92% did not mention contraindications, indications, complications, or infectious diseases related to CPR, nor did they emphasize that the success rates of CPR are not 100%. In addition, 83% failed to adequately stress the importance of CPR as an emergency intervention. Other common problems included video acceleration and no complete CPR procedural demonstrations (Table 2).

Table 2 Deficiencies in non-procedural content of CPR instructional videos

Evaluation of the most popular video on each platform

We selected the most popular video from each platform for detailed analysis. In terms of CPR procedures, all three top videos exhibited irregular maneuvers in the airway opening section (all 4 points). In additional, the most popular video on TikTok scored 1 point in the section of ventilation number with the rate of ventilation too fast. Meanwhile, the top video on Bilibili had notable problems in the responsiveness check (4 points). The most popular video on REDnote not only scored low in the responsiveness check for the same reason as Bilibili video (3 points), but also omitted the resuscitation assessment section entirely (0 points). However, in other sections, these videos all performed well (Fig. 3).

Fig. 3

CPR procedure scores for the most popular videos across platforms

In terms of non-procedural systematic reviews, the top three videos scored high on “media attributes” Sect. (5 points) but low on “contraindications and complications” and “resuscitation success rate” sections, as all failed to mention contraindications, indications, complications, or infectious diseases, or emphasize that CPR success rates are not 100% (both sections <4). Furthermore, the most popular videos on TikTok and REDnote exhibited notable issues in presentation methods with no step-by-step explanations in TikTok video and no full demonstrations of CPR in REDnote video (both 4 points) (Fig. 4).

Fig. 4

Non-procedural content scores for the most popular videos across platforms

Relationship between video quality, duration and popularity respectively

Two independent evaluations of the relationship between video quality, video duration and their popularity indicated no significant differences in video popularity amomg different groups (Table 3).

Table 3 Relationship between video quality, duration and popularity

Discussion

As a primary source of health and medical information for an increasing number of individuals [21], short videos have become an important medium for disseminating medical knowledge. This study systematically evaluated the quality of 100 CPR instructional videos on three major Chinese short video platforms and found that these videos performed well overall in terms of CPR procedures. However, there were still shortcomings in content quality and instructional accuracy, which may mislead the public. In addition, the popularity of the videos was not associated with their quality or duration.

Evaluation on CPR video characteristics

In our study, the majority of the content creators (86%) were healthcare professionals, validating the reliability of the professional content source. Most (94%) presenters chose to use a manikin for CPR demonstrations, which may not adequately represent the realities of performing CPR on a real person due to the more variable and complex nature of real-life resuscitation. We recommend that the optimal demonstration method should involve real-person demonstrations supplemented by manikin utilization for specific procedures that are not suitable for a real-person demonstration [22]. Furthermore, most of the videos depicted CPR simulations in safe environments, even often conducted directly within medical facilities (88%). These videos typically showcased successful and rapid rescue cases (62%), which starkly contrasts with the complexities and challenges of real-life rescues, potentially leading to misconceptions about the authenticity and difficulty of resuscitation efforts. It is therefore crucial to clarify the inherent limitations of rescue techniques while emphasizing the importance of CPR. During the evaluation, we also found that although most videos did not address the topic, a portion (20%) provided comprehensive instructions on the use of AEDs, which is commendable. Particularly following a cardiac arrest, immediate CPR combined with the use of an AED can significantly improve survival rates [18, 19, 23]. Therefore, the dissemination of knowledge regarding the application of AEDs to the general public is of paramount importance.

Evaluation on video quality for CPR procedures and non- procedural content

While short videos play a valuable role in sharing medical knowledge, the quality issues need to be addressed. Yilmaz et al. assessed the reliability, utility, and quality of CPR videos shared on YouTube and reported low information quality and limited accuracy and reliability, without providing detailed analysis of the problems within the videos [17]. Our study revealed overall high scores for CPR procedures but still deficiencies. When checking patient’s responsiveness, 66% of the videos employed the method of shouting in both ears, which may result in not being able to observe the patient’s response at the same time, leading to an incorrect assessment of the patient’s condition. Optimal practice should involve calling out loudly while gently tapping the patient’s shoulders and closely observing their responses [18, 19, 23]. Additionally, 62% of the videos demonstrated incorrect hand placement for compression. The correct practice should be to press on the lower half of the sternum, rather than the midpoint between the nipples, as age and gender could affect the latter’s position [20]. Notable issues were also identified during ventilation. According to the results, 67% of the videos cleared oral obstructions during airway opening, which may not only fail to reduce coughing during ventilation but could also unnecessarily cost valuable time during the CPR process, as evidenced by real-life experience. It is recommended that oral obstruction be removed only in patients with a clear history of foreign body airway obstruction or repeated ventilation failures, given the critical importance of resuscitation time. These discrepancies may be related to the different protocols followed by various demonstrators, as not all demonstrators are healthcare professionals and some may not be professionally trained. Consequently, there is no guarantee that every procedure will meet the standard requirements. Furthermore, 13% of the videos accelerated repetitive actions such as chest compressions, rescue breaths, and resuscitation cycles, directly influencing the comprehensiveness and accuracy of CPR skill dissemination and potentially misleading the public. These findings underscore the widespread misleading issues present in current CPR instructional videos, and highlight the urgent need to improve video quality and standardize demonstration procedures to enhance public understanding and proficiency in CPR techniques.

In terms of non-procedural content, 86% of the videos provided viewers with subtitles or textual explanations of specific steps, which helped the public to better understand the details and significance of each procedure more clearly. However, 4% of the videos had spelling errors in the text descriptions, which could mislead viewers and negatively impact their video viewing experience. And 9% of the videos exhibited deficiencies in audio clarity, frequently occurring concurrently with the absence of subtitles, which severely hindered viewers’ accurate comprehension of the video content. Furthermore, 78% of the videos employed either complete demonstrations or step-by-step explanations, each with its own respective advantages and disadvantages. Complete demonstrations allow viewers to intuitively grasp the processes in rescue situations. In contrast, conducting step-by-step explanation and analysis can help viewers accurately master the details of each technique, thus conveying CPR implementation methods more clearly. The ideal demonstration approach should start with theoretical explanations, followed by a complete demonstration, and conclude with a detailed step-by-step explanation to ensure that viewers does not miss a single detail of the resuscitation while grasping the full theory of the procedure [22]. Moreover, 92% of the videos did not address contraindications or complications related to CPR techniques, nor did they emphasize that CPR does not guarantee a 100% success rate, which could potentially mislead the public.

Evaluation on the most popular videos in details

Considering that video popularity may influence data accuracy and stability, as more popular videos have more impact, we further analyzed the specific scoring situations of the most popular CPR instructional videos on various platforms and also found common quality issues such as too fast ventilation rate in TikTok, failure to check for responsiveness and lack of resuscitation assessment in REDnote, and lack of emphasis on resuscitation success rate in all three videos. These issues could lead to substantial misinformation, given the extensive dissemination of these videos, underscoring the necessity to enhance the quality of such instructional videos. Furthermore, our study revealed that there was no correlation between video popularity and their quality or duration. Viewers may lack the ability to accurately assess video quality and select information correctly, which could potentially lead to the spread of misinformation. So ensuring video quality is vital for their effective educational role. It would be beneficial to encourage viewers to provide feedback on the video instructional content, while also actively promoting certified high-quality instructional videos. Furthermore, while no significant distinction was identified between the short and long video groups, the three most popular videos were all shorter than 5 min, suggesting that controlling for a range of video duration also makes some sense, as viewers may be more inclined to view shorter videos.

Limitations

Although this study simulated patients’ autonomous search behavior for relevant data, several potential limitations exist. Firstly, we only used “CPR 视频” as a keyword for video searches on specific platform and limited our analysis to a small number of 100 CPR instructional videos on specific Chinese online platforms. Consequently, this number may not effectively represent the overall quality of CPR instructional videos on online platforms as we did not include all videos in the screening, and further analysis using larger and more diverse samples is needed. Although the scoring scale we used is based on CPR guidelines, the scoring process relies on human evaluation, which may introduce subjectivity and uncertainty. Therefore, we considered the comprehensive results from two independent evaluators and two other instructors’ review to reduce this subjectivity and uncertainty. This is why we encourage official platforms to establish strict review mechanisms and promote standardized videos. Additionally, considering the long duration of the study, the number of likes and bookmarks on the videos may have changed, thus affecting the statistics of the results. Nevertheless, given the considerable advantage in likes and bookmarks for the most popular videos, the evaluation results still hold meaningful reference value. It should also be noted that due to the marked differences in the algorithms behind platforms like TikTok in China versus the USA or Europe, it is difficult to draw parallels between such videos on platforms outside of China.

Conclusion

This study systematically evaluated CPR instructional videos published on specific platforms, and found that improvements are needed in both the CPR procedures and non-procedural content. This study also provides valuable insights into how to improve the quality of these CPR instructional videos, such as establishing a rigorous video review mechanism, and encouraging public feedback and evaluation. Meanwhile, certified high-quality instructional videos should be actively promoted to guide the public in learning and mastering correct CPR skills.