The public benefits significantly from the information presented in high-quality and engaging TikTok videos related to myopia. In our study, we explored the impact of search strategy and sample size on video quality and audience engagement. Interestingly, we found that these two factors influenced audience engagement but did not affect video quality. Furthermore, specific characteristics of the videos emerged as potential contributors to both video quality and audience engagement. Our research holds significance in guiding viewers to discern high-quality, evidence-based myopia content and assisting healthcare professionals, especially ophthalmologists and optometrists, in optimizing their videos for maximum impact and reach.
Video quality and audience engagement
TikTok videos related to diabetes and chronic obstructive pulmonary disease was considered acceptable (mean total score: 47.1) or satisfactory (mean total score: 59.7) [25, 29]. In our study, the total DISCERN score of 41.3 (95% CI of 39.9 to 42.8) in our study indicated that the quality of myopia-related videos was only fair. This unsatisfactory outcome emphasized the urgent need for improvement in myopia information on TikTok. The fact that 219 myopia-related videos received 2.25 million likes and 0.2 million shares, confirmed the powerful dissemination capability of TikTok. While the Chinese version of the TikTok app cannot extract view counts, the actual audience exposed to educational information on TikTok might be largely underestimated. Therefore, TikTok emerges as a promising platform for myopia education, provided that video quality is further enhanced. This conclusion aligned with findings commonly observed in other health video research [30,31,32].
Search strategy and evaluating sample size
The video retrieval process in existing literature involved both the user-preference-based MLS strategy [19, 30, 31, 33,34,35] and the mainly used DCS strategy [13, 16, 36,37,38,39,40], and four cut-off sample sizes (n = 50 [13, 33], 100 [16, 19, 34, 35, 38,39,40], 150 [30] and 200 [31, 36, 37]) were all reported. This is the first study to analyze whether these two retrieval algorithms influence the audience-available video quality and audience engagement. The MLS strategy exhibited higher audience engagement than the DCS strategy. However, for the videos retrieved by the DCS strategy, the median likes decreased from 6828 (cut-off n = 50) to 1554 (cut-off n = 200). This indicated that TikTok by default prioritized videos with higher popularity to the audience. Sun et al. [38] suggested that videos beyond the top 100 did not significantly affect the quality analysis. Our findings further extended this conclusion to the top 50. Despite differences existing in video engagement, video quality was not influenced by the MLS or DCS retrieval processes across all four included samples. This indirectly suggested the fact that varied video engagement does not correlate with video quality as reported by previous studies [18, 39]. Furthermore, these results eliminated selection bias that might be involved in the factor analysis of video quality, and further enhanced the results’ generalization.
Video production and popularity
Video production properties were theoretically considered to potentially impact the popularity of videos [12, 41]. Our study specifically found videos featuring people were correlated with more likes and larger rate of like growth, consistent with the results of Shi et al. [22]. It indicated that users tend to prefer the direct involvement of video creators in the content dissemination process. Other production properties, as observed by Rein et al. [42], such as inclusion of education labels, background music, emojis, animation/flash did not have any correlation with the audience engagement and quality discernment of short videos. It appears that emphasizing these techniques may not necessarily enhance the popularity and quality of videos.
Video publisher and video quality and popularity
In comparison to ISCs & FPOs, videos uploaded by HCPs exhibited higher quality but paradoxical lower levels of video audience engagement. Similar results were observed in videos uploaded by NPOs. This “quality-quantity disparity” phenomenon had been reported in other studies as well [12, 13]. The overall quality of myopia-related videos on TikTok was already inherently low. Despite being reliable sources of information, HCPs failed to make their high-quality content more appealing. This posed a significant challenge to TikTok’s current role as a potential science communication tool. Thus, possibly HCPs could collaborate with NPOs to increase online reach [43]. Audiences should also be more inclined to select videos created by authoritative professionals, rather than opting for more popular content, as studies indicated that video quality and popularity were not necessarily correlated [18, 39, 44].
Video duration, video quality and popularity
The median duration of TikTok videos related to myopia was 62 s, which was close to the recommended short video length of 60 s [10]. When the video duration exceeded 60 s, a correlation with higher video quality was observed. This can be attributed to the ability of video publishers to present more comprehensive content with longer durations. However, viewer numbers attrition occurred with increasing video length, and longer videos resulted in a smaller proportion of the total video length being watched [42]. This explained the observed decline in audience engagement for videos surpassing 60 s. Wang et al. [39] identified a positive correlation between the video popularity index and the duration of thyroid cancer videos. This association may be influenced by the shorter duration of thyroid cancer videos (37 s). The conclusion drawn from their study may not be directly applicable to myopia videos exceeding 60 s. Acknowledging the contradictory impact of duration on quality and popularity, it is crucial for myopia educators to meticulously prepare condensed and shorter content and minimize the dissemination of ineffective information.
Video contents, video quality and popularity
Previous studies showed comprehensiveness of video content positively influenced video popularity [40]. Videos received higher reach and engagement when discussing disease prevention, severity, and cues to action [45]. Consistent with these findings, our study revealed a positive correlation between video content addressing myopia risk factors/ -management/outcomes and video popularity, indicated by increased likes and shares. The emotional impact of risk factors and severe outcomes may prompt the audience to seek further information on managing myopia. To improve video reach and popularity, myopia educators should not underestimate the importance of myopia content that readily resonates with audiences, especially when it comes to myopia management content that positively correlated with video quality. In response, video producers should pay more attention to myopia evaluation content that educators frequently overlooked [12].
Strengths and limitations
Before analyzing the factors related to the quality and audience engagement of myopia-related TikTok videos, we were the first to compare the impact of two search strategies (DCS and MLS) and four sample cut-offs. This comparison mitigated potential selection bias in subsequent analyses. Additionally, the dependent variables included not only video likes and shares but also the rate of likes increment (likes/days), offering a more comprehensive measure of audience engagement. However, limitations also existed. Firstly, various language versions of TikTok have possible discrepancies in their ability to meet the information needs of the public [46]. However, our evaluation was exclusively focused on Chinese myopia videos, which may undermine the generalizability of our findings. Another limitation was related to the DISCERN instrument, originally designed for assessing quality of website information. Although commonly used in TikTok information evaluation [38,39,40], certain items, such as the inclusion of evidence-based references, were rarely fulfilled in short videos. Additionally, we were unable to access information of video views, posing a limitation in evaluating the popularity of the videos.
