The KIND Challenge community intervention to reduce loneliness and social isolation, improve mental health, and neighbourhood relationships: an international randomized controlled trial

The study addresses the pressing public health problem of loneliness and social isolation, which are linked to poorer mental and physical health and increased mortality. Existing interventions often show small, heterogeneous effects and are resource-intensive, limiting scalability. Community-based and digitally supported approaches may offer scalable solutions, but rigorous evaluations are sparse. Acts of kindness—aligned with provision of social support (emotional, informational, tangible, belonging, companionship)—may strengthen social ties and reduce loneliness, yet their stand-alone impact on loneliness has not been rigorously tested at scale. The research question was whether a low-intensity, community-delivered kindness intervention (the KIND Challenge) could reduce loneliness and social isolation, improve mental health, and enhance neighbourhood relationships across diverse national contexts (USA, UK, Australia).

Prior work establishes loneliness and social isolation as risk factors for morbidity and mortality. While some clinical and community interventions can reduce loneliness, effect sizes are often small and interventions require substantial resources and trained personnel. Digital tools designed to facilitate offline connections may help, especially during conditions like the COVID-19 pandemic. Experimental and small-scale studies show that performing acts of kindness can improve general well-being, relationship functioning, feelings of social support, and happiness, and providing social support confers psychological and health benefits to providers. However, the stand-alone effect of kindness on loneliness reduction in the general community had not been robustly evaluated in large RCTs. This study builds on evidence that social support provision can enhance social capital and reduce self-focus, potentially decreasing loneliness.

Design: Three parallel, two-arm randomized controlled trials (USA, UK, Australia) conducted July–September 2020, following CONSORT guidelines and preregistered (NCT04398472). Arms: KIND Challenge intervention vs waitlist control. Baseline assessment prior to randomization; follow-up at 4 weeks post-intervention. Recruitment and participants: Community adults (18–90 years) recruited via the Nextdoor social networking platform. Inclusion required English proficiency; exclusion was lack of proficient English reading comprehension. Target N was 1452 per country; total N randomized across countries was 4284. Country baseline counts (with loneliness measured at baseline and follow-up): USA baseline N=1413 (Challenge n≈705, Waitlist n≈708); Australia baseline N=1436 (Challenge n≈702, Waitlist n≈734); UK baseline N=1352 (Challenge n≈687, Waitlist n≈665). Follow-up counts: USA (Challenge n=292, Waitlist n=351), Australia (Challenge n=343, Waitlist n=452), UK (Challenge n=330, Waitlist n=390). Demographic characteristics showed broad adult age range with majority female in the USA sample; baseline balance was confirmed. Randomization and blinding: Simple 1:1 randomization implemented via Qualtrics. Due to intervention nature, participant blinding was not possible; research personnel were blind until after data collection. Intervention (KIND Challenge): Participants were encouraged to perform at least one self-selected act of kindness per week for four weeks, adhering to local COVID-19 safety guidelines. Suggested acts mapped to support types: emotional (e.g., check on a neighbor), informational (e.g., provide local tips), tangible (e.g., errands, meals), belonging (e.g., contribute to neighborhood initiatives), and companionship (e.g., regular contact). Weekly email reminders were sent. Measures: Administered online via Qualtrics. Primary outcome: Loneliness (UCLA Loneliness Scale Version 3). Secondary outcomes: Social isolation risk (Lubben Social Network Scale), depression (PHQ-8), social anxiety (Mini-SPIN), quality of life (EUROHIS-QOL-8), positive affect (PANAS-SF), stress (PSS-4). Neighborhood indicators: social cohesion and trust (Social Capital Scale), neighbourhood perception of change and importance, social relationship index—neighbourhood modified. Bespoke items assessed neighbourhood conflict, number of neighbourhood contacts, and COVID-19 social restrictions. Internal consistencies and detailed measures are provided in supplementary materials. Statistical analysis: Conducted separately by country. Intention-to-treat analyses compared intervention vs waitlist, adjusting for covariates (e.g., age, gender, baseline outcome). For continuous outcomes, linear regression models were used; for categorical outcomes, binomial/multinomial logistic regression models were used. Standardized betas and Cohen’s d quantified effects; odds ratios reported for categorical outcomes. Normality checks indicated acceptable distributions; robust MLR estimators in Mplus 8.4 were used. Missing data exceeded 5% but were missing at random; FIML-based procedures enabled inclusion of partially/fully observed cases. Sensitivity analyses added additional explanatory variables (e.g., baseline social/mental health measures, Nextdoor membership length, baseline neighborhood indices, COVID-19 restrictions) to test robustness. Feasibility and acceptability: Retention was below preset feasibility criteria (>40% attrition). No adverse events reported. Average number of kind acts over four weeks: USA 3.01, UK 2.78, Australia 2.58. Acceptability ratings exceeded 6/10 across connectedness, meaningfulness, safety, positivity, and comfort.

Primary outcome (loneliness): Significant reductions post-intervention in the USA (B=−0.06, SE=0.02, p=0.003; Cohen’s d=−0.13) and UK (B=−0.05, SE=0.02, p=0.026; d=−0.21); no significant effect in Australia (B=−0.02, SE=0.02, p=0.331; d=−0.12). Effects in USA and UK were maintained in sensitivity analyses. Secondary outcomes: - Social isolation (LSNS): USA showed a significant reduction (higher LSNS indicating less isolation) in the intervention group (B=0.05, SE=0.02, p=0.033; Cohen’s d=0.46; moderate effect). - Social anxiety (Mini-SPIN): Lower social anxiety in intervention vs control in USA (B=−0.06, SE=0.03, p=0.027; d=−0.12) and Australia (B=−0.05, SE=0.02, p=0.031; d=−0.01). - Stress (PSS-4): Reduced stress in Australia (B=−0.05, SE=0.02, p=0.031; d=0.17). No intervention effects for depression (PHQ-8), quality of life (EUROHIS-QOL-8), or positive affect (PANAS-SF). Neighbourhood outcomes: - USA: More likely to report no neighbourhood conflict post-intervention (B=0.09, SE=0.05, p=0.039; OR=1.62). More likely to know 6+ neighbours (B=0.09, SE=0.05, p=0.017; OR=1.66). - UK: Higher feelings of neighbourhood importance (B=0.13, SE=0.05, p=0.039; OR=1.62). Greater likelihood of perceiving the neighbourhood as stable vs declining (B=0.25, SE=0.11, p=0.022; OR=1.97). - Australia: Less likely to report living in an aversive neighbourhood (B=−0.20, SE=0.08, p=0.011; OR=0.46). More likely to know 6+ neighbours (B=0.08, SE=0.04, p=0.045; OR=1.47). Some effects attenuated in sensitivity analyses as expected, but core loneliness effects (USA, UK) were robust. Feasibility and safety: Retention below target; no adverse events. Mean acts completed: USA 3.01; UK 2.78; Australia 2.58. Acceptability ratings >6/10 across domains.

The KIND Challenge, a low-intensity, low-cost, community-delivered kindness intervention, produced small but significant reductions in loneliness in the USA and UK and improvements in related social and neighbourhood indicators across countries over four weeks. These effects, while modest, are comparable to those seen in more intensive, resource-dependent interventions, supporting acts of kindness as a scalable public health strategy to address social disconnection. Cross-country variability likely reflects differences in compliance, platform familiarity, and COVID-19 restriction stringency, which may have constrained opportunities for kind acts (notably in Australia). Findings are consistent with positive psychology literature showing benefits of kindness and support provision for the provider’s well-being and extend them to loneliness and social connection. Potential mechanisms include enhanced perceived support, strengthened relational ties, and reciprocity. The intervention’s flexibility, minimal resource requirements, and safety make it applicable across age groups and settings, including underserved contexts. The pandemic context underscores the significance of an approach that can yield benefits despite elevated population-level stress and isolation. Future work should examine mechanisms (e.g., reciprocity, purpose, belonging), assess benefits to recipients and broader community spillover, and evaluate durability and optimal dose of kindness behaviors.

This international set of parallel RCTs provides the first large-scale evidence that encouraging small, self-selected acts of kindness toward neighbours can reduce loneliness (USA, UK), lessen social isolation and social anxiety (USA), reduce stress (Australia), and improve neighbourhood relationships. The intervention is scalable, low-cost, and safe, offering a promising public health approach to address widespread loneliness and social isolation. Future research should assess long-term effects, optimal frequency/intensity (“dose”) of kindness, mechanisms of change (including reciprocity and social capital), potential impacts on recipients and communities, and implementation in diverse contexts beyond the pandemic period.

- Short duration (4 weeks) prevents assessment of long-term sustainability of effects. - Participants could not be blinded to condition; potential expectancy effects. - Conducted during the COVID-19 pandemic; varying social restrictions may have constrained or altered the nature/dose of kind acts and generalizability. - Sample skewed toward older adults and individuals already interested in community connection (Nextdoor users), limiting generalizability. - High attrition (~40–50%) from baseline to follow-up; although missingness was addressed (assumed MAR; FIML procedures), retention feasibility criteria were not met. - No correction for multiple comparisons across secondary outcomes; results should be interpreted cautiously. - Possible selection bias due to recruitment methods and participant characteristics. - Dose may have been insufficient; authors note higher doses might yield larger effects.