JMIR Publications

ABSTRACT

Background:

The COVID-19 pandemic has dramatically changed the physical activity (PA) landscape through the closures of gyms, recreational facilities, schools, and many outdoor spaces. Physical distancing and stay-at-home public health guidelines have also reduced opportunity for PA. In turn, the popularity of online home workout and fitness videos, particularly those hosted on free video streaming platforms like YouTube and Instagram, has spiked during the pandemic. These online exercise videos offer viewers a convenient, accessible, and cost-effective means of engaging in PA, typically through regularly posted videos or discrete programs spread over a number of days. Notably, traditional PA and exercise programs often suffer from poor adherence and high dropout rates, despite many advantages over online workout programs (e.g., specialized equipment, real-time feedback, in-person social engagement). Thus, while there are clear advantages to these online fitness videos, their ability to maintain long-term engagement and adherence is unknown.

Objective:

This study explored patterns of engagement (i.e., for online fitness videos posted by channels since the declaration of COVID-19 over four months. A secondary objective was to examine potential moderators of engagement metrics.

Methods:

An environmental scan was used to identify eligible channels. Eligible channels were (i) freely available on YouTube or Instagram and (ii) posted daily or weekday series workouts OR offered quarantine-specific workout programs. Searches were conducted on June 1 and 4, 2020. Engagement metrics of views, likes, and comments were collected from videos posted between March 11 to June 26 or 30, 2020, inclusive. A series of multilevel modeling analyses were conducted to examine longitudinal changes in each of the three outcome variables (i.e., views, likes, and comments).

Results:

Ten channels were deemed eligible and included in analyses; six channels posted regularly, while the other four posted discrete workout programs. Multilevel models revealed that both views and likes significantly decreased across days. Visually, channels display the sharpest drop in engagement within the first week. Linear change estimate indicates that the number of views were initially declining by 24,700 per day, on average across all the channels. Channels that had more subscribers declined in their views, likes, and comments at a significantly faster rate compared to channels with fewer subscribers. The day of the week a video is posted, ‘virality’, and content of a video all appear to influence engagement.

Conclusions:

Regardless of raw engagement metrics, each channel demonstrated peak engagement with the initial video followed by a decline in engagement with subsequent videos. As many countries continue to maintain restrictions on traditional PA facilities due to COVID-19, determining methods to improve engagement and adherence with online fitness videos becomes increasingly important.