A meta-analysis of poststroke depression risk factors comparing depressive-related factors versus others

Abstract

Poststroke depression (PSD) is a public health issue, affecting one-third of stroke survivors, and is associated with multiple negative consequences. Reviews tried to identify PSD risk factors with discrepant results, highlighting the lack of comparability of the analyzed studies. We carried out a meta-analysis in order to identify clinical risk factors that can predict PSD. PubMed and Web of Science were searched for papers. Only papers with a strictly defined Diagnostic and Statistical Manual of Mental Disorders depression assessment, at least 2 weeks after stroke, were selected. Two authors independently evaluated potentially eligible studies that were identified by our search and independently extracted data using standardized spreadsheets. Analyses were performed using MetaWin®, the role of each variable being given as a risk ratio (RR). Eighteen studies were included in the meta-analysis. Identified risk factors for PSD with RR significantly above 1 were previous history of depression (RR 2.19, confidence interval (CI) 1.52-3.15), disability (RR 2.00, CI 1.58-2.52), previous history of stroke (RR 1.68, CI 1.06-2.66), aphasia (RR 1.47, CI 1.13-1.91), and female gender (RR 1.35, CI 1.14-1.61). Fixed effects model leads to identification of two more risk factors: early depressive symptoms with an RR of 2.32 (CI 1.43-3.79) and tobacco consumption (RR 1.40, CI 1.09-1.81). Time bias was found for alcohol consumption. Sample size was significantly involved to explain the role of "alcohol consumption" and "cognitive impairment." Five items were significantly predictive of PSD. It might be of clinical interest that depressive-related risk factors (such as past depressive episodes) were having the largest impact.