The study team, led by Li Lv of Hubei University together with co-first author Yating Yang of China University of Geosciences, recruited 124 college students. Using the Zung Self-Rating Depression Scale, a validated self-report instrument first introduced in 1965, they identified 93 students with depressive tendency and 31 mentally healthy students. The students with depressive tendency were then randomly assigned to one of three groups of 31 participants each: a serious game intervention group that played Celeste, an active control group that played an ordinary entertainment game, and a no-treatment control group that received no game at all. The healthy students served as a fourth reference group, allowing the researchers to ask whether the intervention could bring struggling students back to the psychological baseline of their peers.

Celeste is an unusual candidate for a clinical trial. The game follows a young climber named Madeline as she ascends a mountain while confronting anxiety and self-doubt, and its narrative of struggle, failure, and eventual self-compassion has resonated with millions of players. Unlike purpose-built therapeutic games such as SPARX, which was designed as a self-help intervention for adolescents with depression, Celeste was created as an entertainment product. That distinction matters for the theory behind the trial: the researchers reasoned that the game’s themes of accepting imperfection could operate as a vehicle for self-acceptance, a psychological construct closely linked in prior literature to lower depression, reduced perfectionism, and better mental health outcomes.

All 124 participants completed pre-tests and post-tests on two measures: the Self-Rating Depression Scale and the Self-Acceptance Questionnaire, a Chinese instrument compiled in 1999 by Cong and Gao that assesses both general self-acceptance and acceptance of one’s specific experiences. The statistical analysis used a 2 (Time) by 4 (Group) mixed-design analysis of variance, a technique that examines how scores change over time within each group and whether those changes differ between groups. The results were striking. For depression scores, the Time by Group interaction was highly significant, F(3, 120) = 268.55, p < .001, with a partial eta squared of 0.87, an effect size far beyond the 0.14 threshold conventionally considered large. For self-acceptance scores, the interaction was also significant, F(3, 120) = 33.09, p < .001, with a partial eta squared of 0.45.

Simple effect analyses confirmed where the improvement actually occurred. Only the serious game intervention group showed significant gains from pre-test to post-test. Their depression scores dropped dramatically, F(3, 120) = 1066.53, p < .001, with a partial eta squared of 0.90, while their self-acceptance scores rose sharply, F(3, 120) = 149.92, p < .001, with a partial eta squared of 0.81. By contrast, the active control group that played an ordinary entertainment game and the no-treatment control group showed no comparable improvement. The most remarkable result, however, came from the comparison with healthy students: after the intervention, the post-test depression and self-acceptance scores of the Celeste group no longer differed significantly from those of the mentally healthy control group, with both comparisons yielding p values greater than .05. In practical terms, students who began the study with depressive tendency ended it statistically indistinguishable from peers who had never shown such symptoms.

The design of the trial addresses a persistent weakness in serious game research. Many prior studies compare a game intervention only with a passive control group, making it impossible to tell whether the benefit comes from the game’s therapeutic content or simply from the experience of playing something engaging. By including an active control group that played an entertainment game, the researchers could isolate the specific contribution of Celeste’s narrative and mechanics. The fact that the entertainment game produced no significant change suggests that the benefit was not mere distraction or enjoyment, but something tied to the game’s content. This distinction echoes earlier work by Granic, Lobel, and Engels, who argued in American Psychologist that video games can offer emotional and social benefits, and by Klimmt and colleagues, who described how players undergo enjoyable alterations of self-perception through identification with game characters.

The theoretical framing of the study rests on the construct of self-acceptance, which has deep roots in cognitive behavior therapy and acceptance and commitment therapy. Unconditional self-acceptance, as described by Chamberlain and Haaga and by Fergusson and colleagues in work on perfectionism and depression, refers to the capacity to accept oneself without harsh judgment even in the face of failure or negative feedback. Prior Chinese studies have repeatedly linked low self-acceptance to depressive symptoms among college and medical students, and group counseling interventions targeting self-acceptance have been shown to reduce depression in female college students. The researchers propose that Celeste’s story of a climber who learns to accept her own imperfections may function as an implicit intervention on this construct, allowing players to rehearse self-compassion through gameplay rather than through explicit therapeutic instruction.

The study also situates itself within a broader movement toward digital mental health interventions. Internet-based cognitive behavior therapy has accumulated substantial meta-analytic support for depression and anxiety, and the European Psychologist published a taskforce report in 2018 documenting the efficacy of internet- and mobile-based psychological interventions. Serious games represent a related but distinct approach: rather than delivering therapy through structured modules, they embed therapeutic principles in engaging gameplay. The SPARX trial in New Zealand, published in the BMJ in 2012, demonstrated that a purpose-built fantasy game could serve as a non-inferior self-help intervention for adolescents seeking help for depression. The Celeste trial extends this evidence by showing that a commercial game, never designed as therapy, may nonetheless produce measurable psychological change when its themes align with therapeutic goals.

Subthreshold depression, the condition targeted by the trial, refers to depressive symptoms that fall below the diagnostic threshold for major depression but still cause functional impairment. Research has shown that subthreshold depression in adolescence predicts poorer mental health outcomes in adulthood, and recent Chinese surveys have documented substantial detection rates of subthreshold depression among college students. The National Health Commission of China held a press conference on mental health work in December 2024, reflecting growing official attention to youth mental health. Against this backdrop, a low-cost, scalable intervention that requires no clinician and no diagnosis could have significant public health implications, particularly in university settings where counseling resources are often stretched thin.

The authors caution that the study has limitations typical of early intervention research. The sample consisted of Chinese college students, and the generalizability of the findings to other populations, age groups, and cultural contexts remains to be established. The trial measured outcomes immediately after the intervention, so the durability of the improvement over weeks or months is unknown. The mechanism by which Celeste produced its effects, whether through narrative identification, emotional engagement, or some other process, was not directly tested. Nevertheless, the magnitude of the observed effects, with partial eta squared values of 0.90 for depression reduction and 0.81 for self-acceptance increase, is unusually large for a psychological intervention, and the finding that post-test scores matched those of healthy peers suggests that a commercial game may serve as a genuine vehicle for psychological change. As serious games continue to attract attention as scalable mental health tools, the Celeste trial offers a vivid demonstration that the boundary between entertainment and therapy may be thinner than commonly assumed.

Subject of Research: Serious game intervention for self-acceptance and depressive tendency in college students

Article Title: An intervention study of serious game Celeste on college students’ self – acceptance and depressive tendency

Article References: Lv, L., Yang, Y., Cao, Y., Huang, Y., Feng, Y., Yuan, G., Zhang, B., Chen, Y., & Zhai, Y. (2026). An intervention study of serious game Celeste on college students’ self – acceptance and depressive tendency. Current Psychology, 45(20), Article 1624. https://doi.org/10.1007/s12144-026-10187-5

Image Credits: AI Generated

DOI: 10.1007/s12144-026-10187-5

Keywords: serious games, Celeste, depressive tendency, self-acceptance, college students, randomized controlled trial, mental health intervention, Current Psychology, mixed-design ANOVA, Self-Rating Depression Scale, video game therapy, subthreshold depression

The data came from the face-to-face household component of the 2023 Chinese General Social Survey, one of the country’s most comprehensive ongoing social surveys. The researchers focused on 2,307 adults aged 60 and older who had complete information on all twelve leisure-activity indicators, which ranged from watching television to socializing, exercising, and participating in community events. Rather than treating each activity as an isolated behavior, the team asked whether activities cluster together into recognizable lifestyles. This question matters because gerontologists increasingly suspect that health in later life reflects configurations of behavior, the whole pattern of how people fill their days, rather than the isolated effect of any one pastime.

Latent class analysis works by looking for hidden subgroups within a population. The method assumes that observed responses, in this case how frequently each person engages in each of twelve activities, are generated by an unobserved categorical variable: membership in a particular lifestyle class. The researchers fit models with increasing numbers of classes and used statistical criteria including the Akaike information criterion, the Bayesian information criterion, and the Lo–Mendell–Rubin and Vuong–Lo–Mendell–Rubin likelihood-ratio tests to decide how many groups the data genuinely supported. A four-class solution emerged as the best fit, and the resulting portrait of late-life leisure in China is both detailed and sobering.

The largest group, accounting for 34.09 percent of the sample, was labeled the moderate-frequency home-based screen-oriented class. These older adults engaged in leisure activities at moderate levels, but their participation leaned heavily toward home-based screen activities such as television viewing. Close behind, at 31.79 percent, came the low-frequency television-oriented class, whose members participated in leisure overall at low levels and were similarly anchored to the television set. Together, these two screen-centered groups made up nearly two-thirds of Chinese seniors in the survey. The remaining two classes were smaller but notably healthier-looking: 21.47 percent belonged to a high-frequency socially and physically active class, and 12.64 percent belonged to a high-frequency diversified class that combined social, physical, and other leisure pursuits at high levels.

To connect these lifestyle patterns with health, the researchers examined three outcomes spanning different dimensions of wellbeing: depressive tendency, health limitation, and poor self-rated health. They used survey-weighted logistic regression models that adjusted for a wide battery of covariates, including demographic characteristics, family factors, socioeconomic status, social security coverage, and body mass index. This adjustment is critical, because older adults with different incomes, education levels, family structures, and body compositions differ systematically in both how they spend leisure time and how healthy they are. The question was whether leisure lifestyle class membership predicted health outcomes above and beyond these confounding influences.

The answer, for the screen-oriented groups, was a clear yes. Compared with the high-frequency diversified class, members of the low-frequency television-oriented class had 2.242 times the adjusted odds of depressive tendency, 1.903 times the odds of health limitation, and 2.318 times the odds of poor self-rated health. The moderate-frequency home-based screen-oriented class fared little better, with corresponding odds ratios of 2.037, 1.820, and 1.812. In other words, seniors whose leisure revolved around screens, whether at high or low overall participation levels, showed less favorable indicators across all three health dimensions examined: psychological, functional, and subjective. The consistency of the pattern across outcomes strengthens the case that these lifestyle configurations are meaningfully tied to health rather than to any single measure.

The high-frequency socially and physically active class presented a more nuanced picture. Its members had higher adjusted odds of depressive tendency (odds ratio 1.659) and poor self-rated health (odds ratio 1.796) than the diversified class, but the difference in health limitation did not reach statistical significance. Intriguingly, when the researchers accounted for classification uncertainty using a 200-draw pseudo-class sensitivity analysis, the association between this active class and depressive tendency was attenuated. This suggests that some of the apparent excess risk may reflect the inherent fuzziness of assigning individuals to statistical classes rather than a genuine health disadvantage. The finding is a useful reminder that even sophisticated clustering methods carry uncertainty, and that robust conclusions require testing how sensitive results are to that uncertainty.

The team did not stop at the primary analysis. They reran the models using survey-weighted latent class analysis and available-item models, which accommodate missing responses rather than discarding participants, and found substantively similar class structures. They also tested an alternative definition of poor self-rated health and conducted exploratory subgroup analyses by sex and by urban versus rural residence. Most of the subgroup interactions were not statistically significant, hinting that the broad lifestyle-health associations hold reasonably consistently across major demographic strata, though the authors characterize these subgroup findings as exploratory rather than definitive.

What should readers make of all this? The most important caveat is that the study is cross-sectional: it captures a single moment in time and cannot establish whether screen-heavy leisure causes poorer health, whether poorer health pushes people toward sedentary screen activities, or whether unmeasured factors drive both. An older adult with chronic pain may retreat to the television precisely because other activities have become difficult. The authors are explicit that these findings do not establish temporal ordering or causality. Longitudinal studies that follow individuals over years, and ideally intervention trials that encourage broader participation, will be needed to determine whether reshaping leisure lifestyles actually improves health outcomes.

Even with that caveat, the study carries real weight for a rapidly aging society. China’s older population is enormous and growing, and nearly two-thirds of surveyed seniors fell into classes dominated by home-based screen leisure. If future longitudinal work confirms that diversified, socially and physically engaged leisure protects health, the policy implications could be substantial: community centers, accessible exercise facilities, and programs that lower the barriers to meaningful participation might pay dividends across psychological, functional, and subjective dimensions of aging. The study’s deeper contribution, however, may be methodological. By demonstrating that leisure activities cohere into identifiable lifestyle patterns with distinct health profiles, it argues persuasively that researchers and policymakers should stop evaluating hobbies one at a time and start thinking about the whole architecture of how people spend their later years.

Subject of Research: Leisure lifestyle patterns and multidimensional health outcomes among older adults in China

Article Title: Leisure lifestyle patterns and multidimensional health outcomes among older adults in China: a latent class analysis of the 2023 Chinese General Social Survey

Article References: Luo, L., Yu, M., Ma, J., Wen, Y., Zeng, H., Liu, L., & Zhang, S. (2026). Leisure lifestyle patterns and multidimensional health outcomes among older adults in China: a latent class analysis of the 2023 Chinese General Social Survey. BMC Geriatrics. https://doi.org/10.1186/s12877-026-08344-3

Image Credits: AI Generated

DOI: 10.1186/s12877-026-08344-3

Keywords: older adults, leisure participation, leisure lifestyle patterns, latent class analysis, depressive tendency, health limitation, self-rated health, healthy ageing, China, gerontology, Chinese General Social Survey, screen time