YouTube meets the criteria for a public online database as it is free, publicly accessible without requiring registration, and has a large membership size [18,29]. Consistent with past research [29,30], our institutional ethics board deemed ethics approval unnecessary as content relevant to this study was in the public domain. YouTube was accessed without a registered account to ensure accessed videos were publicly available and had no age restrictions.

The coding rubric for the video analysis was initially developed through a deductive approach. Specifically, we reviewed studies that used content analysis to examine health-related YouTube topics [23,25-28]. Videos were coded as either personal, where the primary focus is on sharing experiences, or informational-educational, where the focus is to provide information. Other variables included (1) tone (positive, negative, neutral, or mixed), (2) purpose(s) of video (sharing information, emotional aspects of infertility, treatment outcomes, and advice to others), (3) whether the video has live people present (character video) or absent (noncharacter video), and (4) source of video, which indicates the type of YouTube channel that posted the video (personal account, organization’s channel, news channel, or other type of channel). Before viewing videos, the coding team generated categories hypothesized to be relevant to infertility-related videos, such as the stage of fertility treatment of the poster(s) and parity of the poster(s). Two coders tested the rubric using 7 sets of 3 to 5 videos with low view counts. Over the course of this testing period, the codebook was modified until variables were refined and explicitly defined. Through an inductive process, new variables such as the code “pregnancy announcement” were added if agreed upon by both coders during the training process. Training videos were not included in the final dataset. The training period ended when the 2 coders reached agreement for all codes and code definitions (see Table 1 for codebook). Certain variables (purpose of video, type of informational video, and subtones in videos) were not mutually exclusive, and therefore codes could sum to more than 100%. Video statistics were collected for all videos in the sample. These included number of views, likes, dislikes, comments, date video was posted, transcript (generated by YouTube), subscriptions driven (number of people who subscribed to the channel following viewing of video), number of shares, length of video, number of subscribers to channel, account creation date, and total channel views (sum of views for all videos posted by user). YouTube channels are profiles for users to post their videos and share information about themselves and their videos.

In January 2017, the most-viewed infertility YouTube videos were identified using the search term “infertility” on YouTube on Google Chrome browser with the search history and cookies cleared [31]. We chose the term “infertility” to capture the most common, basic query a patient would be likely to search, as people generally use keywords and condition names when searching for health information online [32]. This search produced approximately 314,000 results. Videos were sorted by view count and the 80 most-viewed videos were selected for analysis. Previous research suggests over 90% of users click on results in the first 3 pages of a search [33]; therefore, we elected to include the first 4 pages of results to ensure selection of popular infertility videos most likely to be encountered (20 results per page). Data for each video were collected and stored in Microsoft Excel. When variable information was not provided for a video (eg, number of shares), it was coded as missing. The URL and title of video were saved as a reference during data analysis.

Videos were excluded from the sample if they were non-English (n= 16), unrelated to infertility (n= 3), or had age-restricted access (n=2). This produced a final dataset of 59 videos. Refer to Figure 1 for an inclusion flowchart.

Approximately 20% of the final dataset (14/59, 24%) was independently coded by 2 coders (MKH and FB) to examine inter-rater reliability. Cohen kappa was used to evaluate inter-rater reliability (see Table 1 for kappa coefficients). Kappa coefficients ranged from .638 (substantial agreement) to 1.00 (near perfect agreement), reflecting satisfactory or excellent inter-rater reliability for health-related research [34]. All analysis was performed using SPSS Statistics 23 software (IMB Analytics). Median value and inter-quartile range were used for most video statistics to provide a measure of central tendency, which tends to be less influenced by skewed data. A similar approach has been used in previous research of health-related topics on YouTube [35-37].

The total sample of 59 videos ranged from 50 seconds to 40 min in length (mean 6.6 min, SD 7.10 min). Videos were collectively viewed almost 23 million times (median=156,103 views). Videos received more likes (median=525 likes) than dislikes (median=25.5 dislikes), with a median of 55 comments per video. Only 35 videos indicated number of shares, with a median of 105 shares per video (see Table 2 for video popularity statistics).

Videos were classified into 4 channel categories—personal accounts (23/59, 39%), organizations’ channels (20/59, 34%), news channels (2/59, 3%), and accounts that could not be classified (eg, due to lack of information on channel; 14/59, 24%). The channels had a median of almost 6000 subscribers.

The majority of videos had people in them (character videos, 48/59, 81%), whereas 19% had animation or photos (noncharacter videos, 11/59). More than half of the videos discussed treatment outcomes (33/59, 56%) and half shared information (30/59, 51%). About one-third of videos discussed emotional aspects of infertility (20/59, 34%) and 10% offered advice to others (6/59). The sample contained more personal (35/59, 59%) than informational-educational (24/59, 41%) videos. The characteristics of these 2 categories of videos are discussed below and are summarized in Table 3 (personal videos) and Table 4 (informational-educational videos).

The purpose of all 24 informational-educational videos was coded as sharing information and each was neutral in tone. Emotional aspects of infertility treatment and advice to viewers were each present in 2 videos. Over half of the videos contained information relating to infertility causes or detection (16/24, 67%) and/or medical procedures (14/24, 58%), whereas less than one-third discussed lifestyle factors (7/24, 29%) and/or psychological factors related to infertility (6/24, 25%).

Personal videos (mean 2507.66, SD 2663.87) had significantly more likes than informational-educational videos (mean 241.22, SD 338.38; t₅₆=4.04; P<.001). Personal and informational-educational videos did not differ on number of views, dislikes, subscriptions driven, or shares. Similarly, personal videos (mean 347.32, SD 405.63) had significantly more comments than informational-educational videos (mean 33.48, SD 36.48; t₅₅=3.69; P<.001).

Positive (n=26) and neutral (n=25) videos were further analyzed for features relating to video popularity. Personal videos with mixed tone (n=8) were not analyzed due to small sample size. Positive and neutral videos did not differ on number of views, dislikes, subscriptions driven, or shares. Positive videos had significantly more likes (mean 2059.56, SD 2355.67) than neutral videos (mean 254.29, SD 337.08; t₄₇=3.79; P<.001). Positive videos also had significantly more comments (mean 267.32, SD 321.34) than neutral videos (mean 34.38, SD 35.95; t₄₆=3.63; P<.001). As all positive videos (n= 26) were personal videos, it was not possible to determine whether greater engagement was due to videos being positive or personal.

This study was the first to examine infertility-related videos on YouTube using a content analysis method. These videos had high numbers of comments, likes, shares, and subscriptions, indicating that many individuals seek out and engage with infertility-related information on YouTube. The videos were categorized into informational-educational videos and personal videos. Personal and informational-educational videos did not differ in number of views, supporting claims that people use YouTube for both health-related information [38] and social interaction and discussion [38,39]. However, personal videos did elicit more engagement as evidenced by more likes and comments. This suggests that, compared with informational-educational videos, personal videos may resonate more strongly with viewers. One reason may be that personal videos allow people to observe others undergoing similar experiences to seek reassurance, normalization, and niche support [14,15,40]. In addition, these videos may provide a platform for the posters to share and possibly validate their own infertility experiences.

Viewers also showed greater engagement with positive videos, which garnered more likes and comments than neutral videos. This may indicate viewer preference for videos that convey a positive outlook related to fertility. Video messages showing positive fertility outcomes may increase hope and optimism in the viewer, but they may also give a false impression of the success rates of fertility treatment. In actuality, less than half of embryo transfers for in vitro fertilization result in a live birth in the United States and Canada, with rates dropping with increasing maternal age [41,42]. Reliance on YouTube for infertility information may foster unrealistic expectations regarding the success rates of treatment, which may influence treatment decisions [38,43]. As fertility patients frequently access the internet for information [13,14], research is needed to establish whether YouTube content can affect the perceptions of infertility treatment among fertility patients.

In line with research suggesting that men are less likely to share YouTube videos [39], there was an absence of videos containing men only. When men were present, they were always accompanied by their female partner or other people (eg, friends or family). This finding is consistent with research showing that online activity and communication about infertility is more common among women [44-46]. Although we found a lack of male characters in videos, it remains possible that men seek out infertility content through other online avenues, such as websites and discussion forums. Research is needed to determine whether men use YouTube to search for information related to infertility or if they prefer other online platforms.

There are several limitations to the study and future directions to consider. As we only examined the content of high-rated videos, the results may not be generalizable to all infertility videos on YouTube. It is possible that less viewed and favored videos may also contain stories of negative experiences and treatment failures, which were mostly absent from highly viewed videos. YouTube videos with negative tones, such as sadness and anger, may be less viewed and preferred by viewers. Our study was a cross-sectional examination of videos and did not explore how video popularity changed over time. Due to the low number of videos where advice was offered to viewers, we were not able to assess quality of the information posted. Further research is needed to undertake such an assessment, as some studies have found health information on YouTube to be false or misleading [28,47,48]. Finally, as we analyzed publicly available content, we were only able to assess metrics available from YouTube and did not ask viewers their motivations for watching videos or the impact of these videos on their attitudes and beliefs. Future research should explore how infertile men and women search for videos, their reasons for doing so, and their reactions to the content of the videos.

This analysis establishes highly viewed YouTube videos as a source of both technical and experience-based information related to infertility. Viewers appeared to prefer and engage more with videos containing personal experiences and positive tones. Although these videos may provide hope to viewers by frequently sharing positive treatment outcomes such as pregnancy announcements, they may also create and perpetuate unrealistic expectations about the success rates of fertility treatments.