A Brief Resilience-Enhancing Intervention and Loneliness in At-Risk Young Adults: A Secondary Analysis of a Randomized Clinical Trial

Loneliness has become a global public health concern, particularly among adolescents and young adults, with high prevalence (eg, 40% of those aged 16-24 reporting frequent loneliness). Loneliness in young people is linked to poorer general health, physical inactivity, low sleep quality, increased substance use, impaired psychosocial functioning, and is a risk factor for depression, anxiety, and suicidal ideation. Existing behavioral interventions have largely targeted older adults and show only moderate efficacy. Given associations between loneliness, negative self-evaluation, and sensitivity to social rejection, the study examined whether a brief, group-based Resilience Training (RT) program focused on improving emotion regulation, self-perception, and social interactions could reduce loneliness in at-risk young adults.

The paper situates the problem within evidence that loneliness is a significant health issue (Lancet, 2023) and highly prevalent in young adults (BBC survey, 2018). Prior epidemiological work links loneliness in young adults to a range of adverse mental and physical health outcomes (Psychol Med, 2019). Interventions to reduce loneliness have primarily focused on older adults and yielded moderate effects (JAMA Netw Open, 2022). Previous work by the authors showed RT increased resilience-related capacities and reduced psychopathology symptoms in at-risk college students (Psychol Med, 2023), motivating this secondary analysis on loneliness outcomes.

Design: Secondary ad hoc analysis of a randomized clinical trial (NCT06038786), approved by the Mass General Brigham IRB; written informed consent obtained; CONSORT guidelines followed. Participants: Boston-area college students (aged 18-25) with mild depressive symptoms and/or subclinical psychotic symptoms (elevated risk for psychiatric illness). Randomization: RT intervention (n=54; 7-12 per group) vs waitlist control (n=53). Timeframe: July 2018 to February 2020. Intervention: Resilience Training (RT), a 4-session, group-based behavioral program teaching evidence-based skills (mindfulness, self-compassion, mentalization) via didactic material, experiential exercises, and group discussions. Measures: UCLA Loneliness Scale (primary loneliness outcome) and other scales (Connor-Davidson Resilience Scale, Five Facet Mindfulness Questionnaire, Self-Compassion Scale) administered pre- and post-intervention. Demographics: Self-reported race/ethnicity collected but not used as variables. Analysis: Mixed-model ANOVA to test group × time interaction on loneliness; intent-to-treat; α=0.05 (2-tailed); SPSS v29 (IBM); additional analyses examined correlations between changes in loneliness and changes in resilience-related skills and conditional models controlling for these changes.

Sample characteristics: 100 participants (70% female), mean (SD) age 18.8 (0.9) years; racial/ethnic composition: 30% Asian, 6% Black, 12% Latinx, 55% White, 9% multiracial/other. Primary outcome: Significant group × time interaction for loneliness (η²=0.11; P=.002), reflecting a greater decrease in loneliness in RT vs waitlist. Pre/post means (mean [SD] change): - Loneliness (UCLA Loneliness Scale, 20-80): RT 40.98 (12.11) → 36.83 (11.69); change −4.27 (9.15). Waitlist 39.37 (11.61) → 40.62 (12.57); change 1.46 (7.16). Secondary measures (pre → post; change): - Resilience (0-100): RT 64.94 (12.46) → 72.70 (14.03); change 2.45 (3.62). Waitlist 55.89 (17.91) → 60.48 (16.64); change 0.44 (9.28). - Mindfulness (0-195): RT 124.44 (19.05) → 132.38 (26.82); change 6.73 (8.74). Waitlist 110.53 (15.64) → 115.53 (16.66); change 1.33 (7.60). - Self-compassion: RT 2.91 (0.86) → 3.45 (0.74); change 0.32 (0.47). Waitlist 2.65 (0.67) → 2.82 (0.73); change 0.03 (0.35). Correlational findings within RT group: Decrease in loneliness correlated with increases in resilience (r=−0.42; P=.048), mindfulness (r=0.06; P=.03), and self-compassion (r=−0.43; P=.051). No similar changes in waitlist. Conditional analyses: The group × time effect on loneliness remained significant when controlling for changes in resilience (η²=0.19; P=.02) and mindfulness (η²=0.16; P=.04) but not when controlling for increases in self-compassion (η²=0.10; P=.10), suggesting self-compassion gains contributed to loneliness reduction.

Findings indicate that a brief, 4-session resilience-focused group intervention can significantly reduce loneliness among at-risk young adults relative to a waitlist control. The observed associations between loneliness reduction and increases in resilience, mindfulness, and particularly self-compassion suggest that enhancing these capacities may be mechanisms through which RT impacts loneliness. The persistence of the group × time effect after controlling for resilience and mindfulness, but attenuation when controlling for self-compassion, supports self-compassion as a key contributor. These results align with the theoretical rationale that improving emotion regulation, self-perception, and social interactions reduces negative self-evaluation and sensitivity to rejection, thereby lowering loneliness. Given its brevity and group format, RT may be scalable for public health efforts targeting loneliness in young adult populations at risk for mental health problems.

This secondary analysis of a randomized clinical trial shows that a brief, group-based Resilience Training program was associated with significant reductions in loneliness among at-risk young adults and concurrent improvements in resilience-related skills. The findings suggest self-compassion may be a central mechanism. Future research should include active control conditions, follow-up assessments to evaluate durability and downstream effects on health and well-being, and testing in broader and more diverse populations to assess generalizability and mechanisms.

Key limitations include the absence of an active control group and the lack of follow-up assessments to determine durability of effects and longer-term impacts. The analysis is secondary and ad hoc, and the trial sample comprised Boston-area college students, which may limit generalizability to other populations.