Access to informative and reliable clinical content is crucial for managing complex clinical conditions, especially for individuals in rural areas or with limited access to specialist care [1]. Epilepsy, a common chronic condition, highlights the challenges faced by patients needing access to clinical specialists. Children and youth with epilepsy and their families need accurate and trustworthy information to manage their condition effectively [2].

Social media platforms, particularly YouTube, have become significant sources of information for many people. With billions of users monthly, YouTube has established its presence in over 100 countries worldwide and offers localized versions in 80 languages to serve a diverse global population [3]. However, poor-quality health videos on YouTube can be misleading, but they are still promoted via YouTube’s search algorithm that prioritizes popularity, relevance, and view history over content quality and accuracy [4]. This presents a significant challenge for individuals who seek health information without proper guidance or supervision, as they are increasingly exposed to unverified and partially misleading content. The consequences of such exposure can include the promotion of unhealthy habits and activities [5]. A 2023 study about video-sharing platforms revealed a strong negative correlation between high-quality content and popularity regarding videos providing accurate information on cancer and nutrition [6]. Out of 1200 videos, the most reliable videos on YouTube were from universities, professional organizations, and nonprofit physician groups, while unreliable videos included content related to treatments that scored low on the Journal of American Medical Association scale for quality medical content [7]. Analysis of video content using the DISCERN instrument yielded similar results for diabetes content [8]. High-quality health care educational resources developed by experts on widely accessible platforms such as YouTube offer an approach to mitigating these concerns. Through analytics available through YouTube, we can understand the factors influencing viewership and engagement with clinical videos.

Through the REACT (Reaching Out for Epilepsy in Adolescents and Children Through Telemedicine) project, Children’s Mercy Kansas City has developed online educational resources for families and children and youth with epilepsy [9]. One of these resources includes online videos covering different aspects of managing epilepsy and seizures, such as general content about seizures, types of seizures, seizure precautions, alternative treatments, and frequently asked questions. The goal was to provide access to patients and their families, primarily those who spoke English or Spanish. This design provided an opportunity to compare the use of videos with the same clinical content. A preliminary analysis of YouTube Analytics traffic patterns for one of these videos (“What is an absence seizure?”) showed high use for the Spanish version [7]. This paper presents a comprehensive traffic analysis of the REACT YouTube channel, comparing the paired English and Spanish versions of otherwise identical content.

Videos for epilepsy-related topics prioritized by the clinical team were developed in English and Spanish. A multistep approach was taken when translating the videos from English to Spanish. First, the bilingual research team translated the scripts into Spanish. The scripts were reviewed by bilingual and culturally competent staff before recording the videos. The resulting videos were uploaded to the REACT YouTube channel. The videos for each topic are considered as episodes with two versions.

YouTube Analytics for the REACT channel were accessed to evaluate online traffic patterns to the video content [10]. No personal identifiers were available or sought from YouTube. We did not attempt to correlate patient-prescribed views with the traffic patterns evaluated in this work. GraphPad Prism 9.5.1 (GraphPad Software, Inc) was used to quantify video characteristics, with continuous variables presented as means with SDs and ranges.

Tags are descriptive keywords used on YouTube to improve the discoverability of a video. In August 2022, the study team added tags to each of the videos. To analyze the impact of these tags, total monthly views were separated into a pretagging period (September 2021 to August 2022) and a posttagging period (September 2022 to August 2023).

Impressions are the number of times a video’s thumbnail is presented to a user on YouTube. The click-through rate was provided by YouTube Advanced Analytics and represents the percentage of REACT channel thumbnails that were clicked by viewers.

All study data used were anonymous.

Clinical experts developed video content for 17 topics related to treating epilepsy and seizures. The Children’s Mercy Kansas City neurology clinic adopted these videos as part of the patient departure educational process. Children and youth with epilepsy and their families were “prescribed” to view at least one video after each visit. The information and prescribed video were printed in the educational department and given to patients so they could scan a QR code that led them to a REDCapÒ survey. This survey included demographic information, a link to the prescribed YouTube video, and a pre- and postsurvey. Children and youth with epilepsy and families also had the option to view more than one video if desired. The first videos were uploaded to the REACT YouTube channel on August 7, 2020, and all videos were made publicly available by September 2, 2021 [11].

We evaluated traffic patterns between September 2, 2021, when all videos were available, and August 31, 2023. In the aggregate, views per day for the Spanish videos surpassed the English versions (Figure 1A). Compared to the English videos, traffic to the Spanish-language videos was significantly higher (P<.001) based on click-through rate (Figure 1B). Some episodes, such as 4 and 14, exhibited different click-through rates between the two languages.

The addition of tags correlates with higher traffic (Figure 2A). The highest total monthly views for the English-language group of videos was 320 in January 2023, while the Spanish-language group of videos’ highest monthly views were 939 in March 2023. To further assess the impact of the tags, the total monthly views were divided into two periods: pretagging period (September 2021 to August 2022) and posttagging period (September 2022 to August 2023). In the pretagging period, the views per month for the English-language group increased after tagging (pretagging mean 61, SD 28; posttagging mean 220, SD 53). A similar trend was observed in the Spanish-language group as well (pretagging mean 201, SD 71; posttagging mean 743, SD 114). Both the English- and Spanish-language videos showed significant increases in traffic after the addition of tags (P<.001; Figure 2B).

Traffic to individual episodes demonstrated differing patterns. The Spanish versions of videos 3 and 4 showed a strong association between impressions and views (Figure 3A-D). For video 14, impressions and views were similar for the Spanish version, but for the English version, impressions increased earlier (Figure 3E-F). In the English category, video 14 had the highest impressions (n=106,936) and views (n=1087; Table 1). In the Spanish category, video 15 had the highest impressions (n=49,039), followed closely by video 3 (n=41,892). However, video 3 had more views (n=3839) than video 14.

While the location of the viewer was only available for 2096 (14.3%) of the 14,705 views, we observed that countries with predominantly non–English-speaking populations tended to have higher view counts reported by YouTube than primarily English-speaking countries. Users in the United States generated the most views (n=884) for the Spanish language, followed by Mexico (n=859) and Argentina (n=110; Multimedia Appendix 1).

The results from the analysis of YouTube search terms used to find English and Spanish videos are presented in Table 2. The top searches that resulted in views for English videos were “tonic,” “clonic,” or “tonic-clonic.” The term “absence seizure” also generated strong interest, resulting in 118 views. Spanish videos had more views for the comparable search term “crisis de ausencia” or “absence seizure,” with 527 views. In comparison, the terms “convulsiones tonico clonico” or “tonic clonic seizures” only garnered 118 views. The search term “tipos de convulsiones” or “types of seizures” resulted in the most views for the Spanish-language group, with 798 views. These findings suggest differences in viewers’ search behavior when seeking content in different languages on YouTube.

Table 3 presents a comprehensive list of external traffic sources used to find video content on the REACT YouTube channel. English viewers predominantly accessed the content through the Children’s Mercy website, with 443 views, compared to the Spanish count of 151. Spanish-language content was embedded on an epilepsy education website maintained by the government of Chile and generated 121 views. Some of the Spanish-language video views originated from a Spanish-language occupational health education consultancy group’s website. Spanish-language viewers relied more heavily on WhatsApp as their preferred means of finding content, with 94 views compared to only 1 among English-language viewers.

Much of the traffic to the REACT channel originated from sources outside the YouTube platform. These include internet searches, direct URL entries, bookmarks, or embedded links. External traffic sources accounted for 38,172 (12.8%) of the 297,632 impressions. However, despite the relatively lower number of impressions, these external sources, especially WhatsApp, contributed significantly to the overall viewership. In contrast, videos directly delivered by YouTube accounted for 259,460 (87.2%) impressions. Further analysis examined views by internal versus external sources, grouped by language. The Spanish-language groups contributed 8697 views internally and 2642 views externally to the YouTube platform, while the English videos generated 2508 views from within YouTube and 858 from external sources.

We investigated the impact of YouTube algorithmic suggestions on driving views to REACT content. These suggestions were presented as pop-ups after a video was completed or while users were browsing the website. Within the English group, we observed that videos discussing atonic seizures, tonic-clonic seizures, and the differences between seizures and epilepsy received the highest number of views, with 61, 31, and 14 views, respectively (Multimedia Appendix 2). The Spanish videos addressing convulsions and absence seizures garnered the most attention from these algorithmic recommendations, with 146 and 96 views, respectively.