(PS8-68) Enhancing engagement in an Internet intervention for depression using gamification

Abstract Text:

Background: Internet interventions can be an important tool for individuals experiencing depression. Despite the demonstrated effectiveness of Internet interventions (Buntrock et al., 2016; Taylor et al., 2020), they frequently suffer from low engagement and high attrition Eysenbach, 2005; van Ballegooijen et al., 2014). Gamification, or the addition of game-like elements, has been shown to improve engagement with mental health intervention (Bakker et al., 2016; Johnson et al., 2016). However, the effects of a single, simple, and familiar to most users method of gamification, namely awarding badges for desirable behaviors, have not been well studied.

Methods: Individuals were recruited (mainly via Google Ads) to BetterBET, a fully-automated Internet intervention for depression. Eligible participants were adults, able to read English, with regular access to email and internet. Of 830 consented individuals who logged in at least once (and therefore learned their randomization condition), 418 were in the gamified version of the intervention and 412 – in the non-gamified version. The self-paced intervention consisted of eight interactive lessons covering a variety of strategies for managing depression based on cognitive-behavioral therapy. Participants could access lessons in any order at their convenience. A single form of gamification was implemented: badges awarded for certain interactions with the intervention: after 1, 5, and 10 logins, after 1 and 4 mood ratings, after 1, 4, and all lessons completed, and after each of the four follow-ups). Badges were prominently displayed on the main page of the website. Aside from badges, both randomized conditions received an identical intervention.

Results: On average, participants were 32.1 (SD=15.2) years old, female (70.6%), with 15.2 (SD=3.5) years of education, and with moderately severe depression (mean PHQ-9 = 17.0, SD = 6.4). Compared to non-gamified condition, gamified condition resulted in greater number of logins (1.02 (SD = 1.82) vs 0.82 (SD = 1.08); p = 0.025, OR = 1.25, 95%CI: 1.03-1.53), more mood ratings (3.34 (SD = 8.58) vs 2.75 (SD = 4.45); p = 0.016, OR = 1.21, 95%CI: 1.04-1.42), and were more likely to complete at least one follow-up (21.5% vs 16.0%, Fisher’s exact test = 0.05). Those in the gamified condition were also more likely to use a resource page (.93 (SD = 3.85) vs. .47 (SD = 1.64) times; OR = 1.98, 95%CI: 1.59-2.46) even though this feature was not incentivized by badges. No differences were observed in the number or duration of lesson visits. Furthermore, 455 participants provided lesson feedback. Though all four feedback questions had higher average scores in the gamified condition, none had reached the significant threshold, and only one (helpfulness; m = 5.35, SD = 1.40 vs 5.07, SD = 1.47) approached significance (p = .051, d = 0.21).

Conclusion: Even a simple gamification system such as awarding badges for desirable behaviors can improve several metrics of engagement and retention. Future interventions should consider thoughtful integration of gamification to foster greater engagement. When implemented, gamification elements should be introduced early and offered frequently for maximum benefit.