Associations of pandemic-era telework with cognition and depressive symptoms among middle-aged and older nonagricultural workers in China: a longitudinal CHARLS study

Abstract

Telework expanded rapidly during the COVID-19 pandemic, but evidence on its cognitive and mental-health correlates among middle-aged and older workers remains limited. We examined longitudinal associations of pandemic-era telework with cognition and depressive symptoms among middle-aged and older employed nonagricultural workers in China. We used China Health and Retirement Longitudinal Study (CHARLS) data, with 2018 as the pre-pandemic baseline and 2020 as follow-up. Primary samples included 4,184 participants for cognition (270 teleworkers) and 4,262 for depressive symptoms (278 teleworkers), across 436 communities. Co-primary outcomes were a harmonized cognition score (0–10) and continuous CES-D-10 (0–30). Missing baseline covariates were addressed using 20 multiply imputed datasets. Propensity-score overlap weights were incorporated into baseline-outcome-adjusted analysis of covariance models with community-clustered covariance; estimates were pooled using Rubin’s rules with Barnard-Rubin degrees of freedom, and the two primary P values were Holm-adjusted. A sensitivity analysis restricted each outcome-specific sample to communities containing both teleworkers and nonteleworkers and re-estimated the propensity score and overlap weights within the restricted sample. After overlap weighting, measured baseline covariates were well balanced. Telework was associated with a modestly higher cognition score (β = 0.278, 95% CI 0.040 to 0.515; Holm-adjusted P =.021953) and a lower CES-D-10 score (β=-0.792, 95% CI -1.317 to -0.266; Holm-adjusted P =.006430). In mixed-exposure communities, the cognition estimate was smaller than in the primary analysis (β = 0.226, 95% CI -0.025 to 0.478), whereas the CES-D-10 association remained negative (β=-0.870, 95% CI -1.424 to -0.315). Telework-by-sex interactions were not supported for cognition ( P =.351891) or CES-D-10 ( P =.666872). The depressive-symptom association was directionally consistent across major sensitivity and longitudinal analyses; the cognition finding was less robust. Among middle-aged and older employed nonagricultural workers in China, pandemic-era telework was associated with lower depressive-symptom scores after accounting for pre-pandemic symptoms and measured baseline differences. This association remained negative in mixed-exposure communities. A small positive association with cognition was observed in the primary analysis but was smaller and statistically uncertain in mixed-exposure communities and was less consistently supported by longitudinal analyses. There was no clear evidence that either association differed by sex. The findings should be interpreted as observational associations rather than causal effects.