Depressive symptoms occurring after stroke by age, sex and socioeconomic status in six population-based studies: Longitudinal analyses and meta-analyses

Le Peng,Ji Zhang,Na Li,Yinzi Jin,Zhi-Jie Zheng
DOI: https://doi.org/10.1016/j.ajp.2022.103397
IF: 13.89
2023-01-01
Asian Journal of Psychiatry
Abstract:Aims: Population-based cohort studies that examined the effects of stroke on depressive symptoms in elderly participants with and without stroke in different countries/regions had yielded scant evidence. The objectives of this multi-cohort study were to identify the relationships between stroke and depressive symptoms and to assess the pooled risks of depressive symptoms in total populations and in stratified samples by age, sex, education levels, and total household income in large-scale studies.Methods: The associations in the overall participants in the six cohorts were examined using univariate and multivariable Cox proportional hazards models. The relationships in the population subgroups of the cohorts were determined using multivariable Cox regression models. Crude hazard ratios (HRs) were estimated, and adjusted or time-varying HRs were calculated and then pooled by meta-analyses.Results: During a median follow-up time of 3.8 (interquartile range [IQR]: 2.2-4.5) -7.9 (IQR: 4.0-8.0) years, 18267 participants out of 62150 participants developed incident cases of depressive symptoms. In the univariate Cox regression analyses, stroke was consistently linked to depressive symptoms across all studies, with HRs ranging from 1.39 (95% confidence interval [CI]: 1.04-1.85) to 1.84 (95% CI: 1.38-2.46). Adjusted HRs varied between 1.19 (95% CI: 0.89-1.59) and 1.48 (95% CI: 1.00-1.98), and time-varying HRs remained largely un-changed. According to the meta-analysis of overall HRs, stroke survivors were more likely than non-survivors to develop depressive symptoms (pooled HR: 1.35 (95% CI: 1.26-1.44)).Conclusions: In general, stroke significantly increased the risk of depressive symptoms in total and stratified participants in most cohorts. In the future, extensive research will be needed to assess the variations in associ-ations between various population subgroups.
What problem does this paper attempt to address?