Social buffering of stress in rats: A multisystem meta-analysis and translational framework for stress resilience (2026)

Sarapultsev, A., Komelkova, M., Gusev, E. et al.

Abstract

Background: Social context modulates stress physiology and resilience, yet preclinical rat paradigms vary widely in stressor type, timing of social exposure, contact modality, and endpoint definitions. We synthesized rat studies to quantify directional and, where feasible, standardized effect-size evidence for social buffering and to outline translational implications. Methods: PubMed, Scopus, and Web of Science were searched (2008–2025) for in vivo rat studies comparing conspecific (pair/group housing or conspecific presence) versus solitary conditions across validated stress and PTSD-like paradigms (e.g., fear conditioning/extinction, CUS/CMS, social defeat, predator threat). Data extraction and reporting followed PRISMA 2020 and SYRCLE guidance. Synthesis followed a two-tier approach: (i) all eligible contrasts were direction-coded as beneficial, neutral/mixed, or detrimental under conspecific conditions based strictly on reported statistical contrasts; and (ii) for domains with sufficient coded contrasts, the proportion of beneficial comparisons was estimated with exact binomial tests and 95% confidence intervals. Standardized mean-difference meta-analysis (Hedges’ g; random-effects REML) was conducted only for predefined outcomes with adequate numerical reporting. Results: Forty studies met inclusion criteria, yielding 89 extracted comparisons. Overall, 69/89 comparisons (≈78%) favored conspecific conditions. Domain-level directional syntheses supported predominance of beneficial outcomes for hormonal (0.72; 95% CI 0.50–1.00; p = 0.048) and neurotrophic/plasticity markers (0.89; 95% CI 0.57–1.00; p = 0.020), whereas inflammatory/oxidative outcomes were more variable (0.71; 95% CI 0.39–0.94; p = 0.227). For predefined behavioral endpoints with sufficient data, effect-size pooling showed a large reduction in conditioned fear (Hedges’ g = -1.22 [-1.53; -0.91], p < 0.0001). Conclusions: Social buffering is robust at behavioral and neuroendocrine levels and often aligns with neurotrophic/plasticity markers, while peripheral immune/redox readouts are more context-dependent.