Loneliness and social networks have been extensively studied in relation to cognitive impairments, but how they interact with each other in relation to cognition is still unclear. This study aimed at exploring the interaction of loneliness and various types of social networks in relation to cognition in older adults.

a cross-sectional study.

face-to-face interview.

497 older adults with normal global cognition were interviewed.

Loneliness was assessed with Chinese 6-item De Jong Gierverg’s Loneliness Scale. Confiding network was defined as people who could share inner feelings with, whereas non-confiding network was computed by subtracting the confiding network from the total network size. Cognitive performance was expressed as a global composite z-score of Cantonese version of mini mental state examination (CMMSE), Categorical verbal fluency test (CVFT) and delayed recall. Linear regression was used to test the main effects of loneliness and the size of various networks, and their interaction on cognitive performance with the adjustment of sociodemographic, physical and psychological confounders.

Significant interaction was found between loneliness and non-confiding network on cognitive performance (B = .002, β = .092, t = 2.099, p = .036). Further analysis showed a significant interaction between loneliness and the number of family members in non-confiding network on cognition (B = .021, β = .119, t = 2.775, p = .006).

Results suggested that a non-confiding relationship with family members might put lonely older adults at risk of cognitive impairment. Our study might have implications on designing psychosocial intervention for those who are vulnerable to loneliness as an early prevention of neurocognitive impairments.