Objective  To analyze influences of social isolation and loneliness on frailty and paths of the influences among community elderly people and to provide evidences for developing intervention on elderly frailty.

  Methods  Using stratified random cluster sampling, we recruited 1 130 permanent residents age 60 years and above in 16 urban communities/rural villages in Ji′nan municipality of Shandong province and conducted a survey among the residents during December 2019. A self-designed questionnaire on general information, Lubben Social Network Scale – 6, UCLA Loneliness Scale – 6, and Tilburg Frailty Indicator were adopted in the study and the model 4 in Process macro program of SPSS 25.0 statistical software was used in multiple regression analyses on impacts of social isolation and loneliness on frailty and paths of the impacts.

  Results  Among 1 091 participants with complete data, 245 (22.5%), 257 (23.6%), and 238 (21.8%) were identified as having social, family, and friend isolation; 373 (34.2%) were found having frailty. Compared to the scores among the participants without the three isolations, significantly higher scores of loneliness and frailty were observed among the participants with social isolation (11.55 ± 4.17 vs. 10.29 ± 3.94 and 4.73 ± 3.24 vs. 3.83 ± 2.75), family isolation (11.38 ± 4.08 vs. 10.32 ± 3.97 and 4.67 ± 3.18 vs. 3.83 ± 2.76), and friend isolation (11.55 ± 4.17 vs. 10.37 ± 3.95 and 4.73 ± 3.24 vs. 3.79 ± 2.70), respectively (all P < 0.01). After adjusting for confounding factors such as gender, age, education, marital status, monthly average income, residence, and prevalence of chronic diseases, the results of multiple regression analysis revealed that social isolation was significantly associated positively with loneliness (β = 1.149, P < 0.01), and social isolation and loneliness were associated positively with frailty (β = 0.487 and β = 0.171, both P < 0.05). The results of bias correction-based non-parameter percentile Bootstrap test showed that the direct effect value of social isolation on frailty was 0.487 (95% confidence interval 95% CI: 0.110 – 0.865), the indirect effect value of social isolation on frailty was 0.197 (95% CI: 0.094 – 0.314), and the total effect value was 0.684 (95% CI: 0.297 – 1.071). The direct and the indirect effect accounted for 71.2% and 28.8% of total variance, respectively.

  Conclusion  Among urban and rural community elderly in Ji′nan municipality, social isolation could affect frailty condition, not only directly but also indirectly with the loneliness as an intermediate factor.