Latent profiles of perceived digital convenience and their associations with depressive symptoms among Chinese empty-nest older adults: evidence from the 2023 CLASS

Abstract

Digital technology is increasingly discussed as a potential resource for supporting mental health in later life, yet empirical findings remain mixed. Prior studies have largely relied on variable-centered approaches that may obscure how perceived convenience of digital activities co-occurs across life domains. Person-centered evidence among Chinese empty-nest older adults—a population marked by reduced intergenerational co-residence—remains limited. This study identified latent profiles of perceived digital convenience and examined their associations with depressive symptom scores. Data were from the 2023 China Longitudinal Aging Social Survey (CLASS; n = 8,190 empty-nest adults aged ≥ 60). Latent Profile Analysis identified profiles from ten perceived-convenience indicators. Associations with depressive symptoms (CES-D 9) were examined using multivariable OLS regression with heteroskedasticity-robust standard errors. Heterogeneity by residence, gender, and chronic disease was tested via interaction terms with multiple-testing corrections. Sensitivity analyses used the BCH method to account for classification uncertainty and inverse probability weighting to address observed covariate imbalance. Four profiles were identified: Digital Disengagement (15.45%), Socially-Oriented (17.53%), Moderate Integrators (26.37%), and Versatile Users (40.65%). Compared with Digital Disengagement, Versatile Users (B = − 1.15, 95% CI: −1.36 to − 0.95) and Moderate Integrators (B = − 0.32, 95% CI: −0.53 to − 0.11) had lower depressive symptom scores; the Socially-Oriented profile was not significantly associated with scores. Effect modification was supported for residence and chronic disease but not gender. Sensitivity analyses produced substantively consistent results. It should be noted that the incremental variance explained by profile membership beyond covariates was modest (ΔR² = 0.020), and the primary contrast corresponded to roughly one-third of a standard deviation on the CES-D 9 scale, without an established clinical anchor. Distinct profiles of perceived digital convenience were associated with depressive symptom scores among Chinese empty-nest older adults. Because the design is cross-sectional and the indicators capture perceived convenience rather than observed digital behavior, findings should be interpreted as associations rather than causal effects. Longitudinal studies incorporating behavioral measures and psychological mechanisms are warranted.