A systematic search of videos uploaded on YouTube was performed in March 2022 with the following five search terms: “external cephalic version,” “ECV,” “fetal version,” “turning a breech baby,” and “turning baby.” The search was performed using the Google Chrome browser operated in Incognito mode and no personal YouTube or Google account was used. Search results were sorted by relevance, replicating a default YouTube search. The first 35 results were collected per search term, since 90% of YouTube users do not watch videos past the 30th resulting video [22]. Inclusion criteria were (1) videos related to cephalic version, (2) English language, and (3) acceptable audio-visual quality. Exclusion criteria were (1) length>1 hour, (2) irrelevant videos, and (3) duplicates. Videos were evaluated by two authors (MRvD and AFvdM), representing the demographic of the viewers. At the time of reviewing the videos, MRvD was a PhD candidate with 1 year of clinical experience in gynecology and obstetrics and AFvdM was a medical student in her final year. Discrepancies were discussed with authors LEvRF and JV, who are both obstetricians.

For each video, the following quantitative data were collected: day of upload, years since upload, duration of the video in minutes, number of views, total number of likes and dislikes, and uploader’s username. To assess audience engagement, the like ratio ([likes× 100]/[likes+dislikes]), view ratio (number of views/days on platform), and video power index (VPI) ([like ratio×view ratio]/100) were determined for each video.

All videos in this study are publicly available on YouTube and all analyses were performed on deidentified data. This study is an observational study in which no action or behavior was imposed on participants. Therefore, this study is not subject to the Dutch Medical Scientific Research Involving Human Subjects Act [25].

The primary outcome of the evaluation was the usefulness score. We performed several secondary analyses and report the scores obtained for prespecified groups. First, videos were categorized according to the source of upload into childbirth attendants, paramedics, vloggers, health care professionals, educational channels, and other. Second, videos were divided into medical videos, which included health care professionals, paramedics, and partly educational channels, and nonmedical videos, which included vloggers, childbirth attendants, and partly educational channels. Videos classified by educational channels were categorized according to the presenter’s profession. Third, all videos uploaded by vloggers were evaluated according to whether ECV was encouraged or discouraged.

The intraclass correlation was used to measure the reliability between the two reviewers with regard to the total usefulness score. The intraclass correlation is used to assess the consistency of measurements between raters. Among the multiple models available for this assessment, we selected the two-way mixed-effect model since we wanted to determine the level of agreement between the two raters. According to Koo and Li [26], based on the 95% CI, a value below 0.50 indicates poor correlation and a value above 0.90 indicates excellent correlation. Continuous variables are presented as mean (SD) and categorical variables are presented as n (%). For comparisons between groups, the χ², Kruskal-Wallis, and Mann-Whitney U tests were used as appropriate. SPSS (version 28) was used for the analysis, where P<.05 was considered statistically significant.

On March 16, 2022, 175 videos were selected from YouTube. Multimedia Appendix 1 shows a flowchart of the selection process of the included videos. After removing duplicates and applying the exclusion criteria, 70 videos remained for evaluation. Multimedia Appendix 2 provides an overview of all included videos. The mean duration of all videos was 8.9 minutes (SD 9.3) and the mean VPI was 203.8 (SD 449.8).

The mean total usefulness score was 7.6 (SD 5.6). The intraclass correlation coefficient of the usefulness score between the two reviewers was 0.942, indicating excellent agreement [26]. Videos were categorized into useful (14.3%), slightly useful (22.9%), and not useful (61.4%) (Table 2). Videos scored as not useful had the highest audience engagement and videos scored as useful had the lowest audience engagement.

Over 60% of the videos defined as not useful were uploaded by vloggers and childbirth attendants. Most videos included in this category were found with the search term “turning a breech baby.” The majority of the videos defined as slightly useful were uploaded by educational channels and health care professionals (37.5% and 18.8%, respectively). All videos defined as useful were uploaded by educational channels and health care professionals, with 90% of the videos found with the search term “external cephalic version.”

The included videos were evenly spread across the upload categories. Three videos could not be categorized in the predefined subgroups and were thus included in the “other” group. These videos included a news channel and a government organization, whereas one source was miscellaneous. Characteristics of the source of upload are shown in Table 3. Childbirth attendants appear to be the most influential sources on this topic, scoring the highest on audience engagement (view ratio, like ratio, and VPI). Excluding the “other” group, paramedics scored the lowest on audience engagement factors.

Usefulness scores were calculated for each category. Educational channels scored the highest, followed by health care professionals. Childbirth attendants scored the lowest on the total usefulness score, with vloggers scoring only slightly above. There was a statistically significant difference in usefulness scores according to the source of upload (P<.001).

When comparing the mean score for each ECV RCOG–specific score statement, “putting pressure on the mothers’ belly” and “aim for vaginal birth” scored the highest, whereas the statements “50% chance of succeeding” and “chance of turning back” scored the lowest. The questions “Are the aims clear and achieved?” and “Is the information presented balanced and unbiased?” had the highest mean scores of the four mDISCERN statements. Points that were deducted most often came from the misinformation categories proven risks and unproven procedure.

All evaluated videos were divided into medical (n=32) and nonmedical (n=38) videos and their characteristics are summarized in Table 4. Overall, nonmedical videos had higher audience engagement than medical videos.

Approximately one-quarter of the medical videos were considered useful, whereas only 5.3% of the nonmedical videos were considered useful. Almost all of the nonmedical videos were considered not useful, whereas 34.4% of the medical videos were considered not useful. The medical videos that were categorized as not useful scored low on the ECV RCOG–specific score and average on the mDISCERN. Moreover, over 75% of medical videos were uploaded by educational channels and health care professionals. The majority of nonmedical videos were uploaded by childbirth attendants and vloggers. Two-thirds of the medical videos were found with the search term “external cephalic version,” whereas most of the nonmedical videos were found with the search terms “turning a breech baby” and “ECV.”

Lastly, we evaluated whether videos uploaded by vloggers were encouraging or discouraging of ECV. As shown in Table 5, ECV was not mentioned in 4 videos in favor of other methods to attempt to turn the baby from a breech to cephalic position. Almost all vloggers (12/15, 80%) encouraged ECV to their followers. Videos encouraging ECV had higher audience engagement compared to videos discouraging ECV, with a particularly large difference in the VPI (313.0 vs 17.2). The like ratio did not differ between the groups (mean 92.6 vs 93.6, P=.47). Three-quarters of the encouraging videos were uploaded by vloggers who experienced a successful ECV.

In this study, according to the accuracy of medical information and content quality, only a minority of videos related to ECV on YouTube were found to be useful. Strikingly, the group with the highest audience engagement scored the lowest on usefulness and the groups with lower audience engagement had higher usefulness scores. The presence of misinformation was low. Importantly, vloggers, the group with the highest audience engagement and the lowest usefulness score, largely encouraged their audience to opt for ECV.

In general, our findings are consistent with previous research evaluating the quality of health-related subjects on YouTube [16,20,27,28]. In a study performed by Lee et al [27] in 2014 evaluating YouTube videos on cholecystolithiasis, 56.5% of the videos were classified as misleading and only 13.8% were classified as useful or very useful. Similarly, we found that very useful videos about ECV had a significantly lower mean view average than that of the misleading videos. In a study on content analysis of infertility-related videos on YouTube performed in 2018, Kelly-Hendrick et al [29] found that videos delivered from a personal point of view had significantly more likes than informational-educational videos. These findings suggest that people who are looking for videos on health-related subjects appear to rely more on personal experiences and information from their peers. This results in overexposure to scientifically inaccurate videos, although these are not necessarily misleading and may even encourage pregnant people to opt for ECV.

Among the search terms, medical videos with accurate information were largely identified when using medical terms, whereas nonmedical videos were identified with nonmedical terminology in the YouTube search.

As a result of an aging YouTube collection along with continuously developing ECV guidelines, medical videos on YouTube also contain misinformation. For example, medical videos contained misinformation on risks by stating that a previous cesarean section is a contraindication for ECV, whereas the RCOG guideline states that ECV is safe and successful in women with one previous cesarean section [5]. YouTube does not filter these outdated videos, underpinning that responsible information providers should also have a plan for updating or removing outdated videos.

In the evaluated videos, alternatives for ECV to encourage spontaneous fetal version were often mentioned. The childbirth attendants advised yoga postures, warm and cold compresses, and acupuncture in combination with misbutton and chiropractic treatment to support a baby in turning to be in a cephalic position. This is in contradiction with the RCOG guideline statement that there is no evidence that postural management alone promotes spontaneous version to a cephalic presentation [5]. People who refuse ECV might be interested in alternative methods for spontaneous version. Therefore, future research should focus on the association between spontaneous version aided by chiropractors and postural management in combination with moxibustion and acupuncture.

The importance of appropriate and encouraging information about ECV for breech presentation is evident, since ECV is of great significance to reduce the CD rate [30]. In 2014, Vlemmix et al [9] found that less than half of pregnant people with a breech baby at term had an ECV in the Netherlands, 20%-30% of pregnant people refused an ECV, and it was estimated that approximately 4%-33% of patients were not given the option of having an ECV by their health care providers. YouTube can be valuable as a source of information for this young generation that is adept at using digital technology for communication and information gathering.

This study has several strengths and limitations. First, we used the validated mDISCERN score and the customized ECV RCOG–specific score based on the RCOG guideline. However, it must be acknowledged that the partly self-developed usefulness score remains somewhat subjective due to the absence of validated tools for evaluating social media content. Although high agreement was achieved between the two video reviewers, this limitation underscores the pressing need for a standardized scoring instrument in modern research. The increasing prevalence of video content across various fields demands a reliable method of assessing video quality and content. Without a standardized and validated scoring instrument, the evaluation is prone to subjectiveness, inconsistency, and lack of scientific accuracy. Establishing and validating this instrument is essential for contemporary research to conduct meaningful studies, make informed decisions, and enhance the medical accuracy of video content. Such studies can help pave the way for more precise assessments and comparisons.

Second, the RCOG guideline was used as the up-to-date and high-quality standard, although this is not necessarily the standard in all contexts.

Third, the methods of the search may have been selective. We only used five search terms, only the first 35 results were retained, and the search was conducted in “Incognito” mode. Because a YouTube search is dynamic and dependent on algorithms, the findings may be variable based on new uploaded videos and personal algorithms.

Finally, we only searched YouTube and omitted other social media platforms that play a role as a source of information for patients with unregulated content. However, YouTube is most likely to be the platform that patients use in the search of health information, as it is the largest video-sharing website with the highest number of videos without time limits and has a user-friendly search function that does not require an account to watch videos.

While our study has shed light on the medical accuracy of YouTube videos about ECV, it is important to acknowledge that we did not explicitly address the potential impact of various video characteristics, including video length, characteristics of the audience, the uploader’s username, or word-of-mouth recommendations, on the usefulness scores of the videos. These factors are of great importance in understanding the dynamics of online health content consumption. For example, video length could affect viewers’ engagement and retention of information, while the uploader’s name or reputation can influence viewers’ trust in the content. YouTube adds a view when a video is watched for over 30 seconds; however, YouTube does not disclose the actual watched duration. Furthermore, individual user data are anonymized and demographic information of the viewer is only available for the uploader. Future research should explore these additional factors comprehensively, as they have the potential to significantly influence viewers’ interactions with health-related videos, ultimately impacting their overall usefulness and the quality of health information accessible on the platform.

This study highlights the compelling need for health care professionals to collaborate with social media influencers to create medically accurate videos with a broad reach and high viewership. In an era where online content holds substantial sway, leveraging the reach and engagement of influencers can significantly enhance the dissemination of credible health information and promote public awareness. Partnerships between hospitals and vloggers can bridge the gap between traditional health care communication channels and the rapidly evolving digital landscape, thereby improving health literacy and encouraging informed decision-making among the public. Balancing content popularity with medical accuracy will be a paramount challenge, requiring careful oversight to ensure that influencers and health care professionals collaboratively contribute to a well-informed and healthier society.

Despite the fact that only few YouTube videos on ECV were considered to be useful, the level of misinformation on ECV in the videos was low. The mismatch between audience engagement and information accuracy was striking, although the vlogs with the highest engagement were appropriately encouraging their audience to opt for an ECV. The findings from this study provide the insight that informational videos should contain valid information along with encouraging stories from the patient perspective using nonmedical terminology to have a maximum impact. This can likely best be achieved by collaboration between medical and community stakeholders. This would generate more awareness about ECV, support pregnant people with a baby in breech presentation, and lower the chance for a high-risk delivery because of term breech presentation.