Comment on “Association between late‐life depression or depressive symptoms and stroke morbidity in elders: A systematic review and meta‐analysis of cohort studies”

With great interest, we have thoroughly reviewed a recent systematic review and meta-analysis conducted by Cai et al.,1 which effectively pooled the association between depressive symptoms and stroke in the elderly population. The authors have successfully identified compelling evidence that late-life depression or depressive symptoms significantly contribute to the risk of stroke among older individuals. Notably, this finding is particularly relevant as the prevalence of depression continues to rise alongside global aging trends. According to a report published by the World Health Organization, currently over 280 million patients diagnosed with depressive disorders worldwide. Additionally, the authors have conducted an extensive subgroup analysis that provides valuable insights into understanding the relationship between stroke and depression across various variables. However, it is important to address certain issues in relation to data synthesis within this article. First of all, in the data synthesis section of the article, it was observed that the authors did not provide a detailed description of the employed data transformation method for extracting effect sizes. Our previous experience in reading this article revealed four instances where relative risk (RR) values were extracted instead of adjusted hazard ratios (HR), as stated by the authors. These articles include “Kawamura et al.,”2 “Simons et al.”3 and “Wassertheil-Smoller et al.”4 Also, the authors may extracted the wrong number from “Liebetrau et al.,”5 which should be “HR = 2.7; 95% CI: 1.5–4.7.” Combining two different effect sizes for data analysis is deemed inappropriate. Therefore, we re-extracted and summarized the data from these articles separately. The pooled HR value obtained from our study was consistent with that reported in the original article, which was 1.38 (95% CI: 1.20–1.59). Additionally, only HR values, both our pooled estimate (1.46; 95% CI: 1.24–1.71) and RR estimate (1.27; 95% CI: 0 0.99–1.61) showed no statistically significant heterogeneity among them (p > 0 0.05). Hence, we suggest that these articles be carefully screened during data selection to ensure more reliable results. (Figure 1; Supplementary material: Table S1). Second, although the authors have conducted comprehensive subgroup analyses, we believe it would be inappropriate to overlook conducting subgroup analyses on the study population itself, as the selection of the cohort may also impact the association between stroke incidence and depressive symptoms. Consequently, subgroup analyses were performed revealing a pooled hazard ratio (HR) of 1.48 (95% CI: 1.24–1.76) in the hospital population and 1.30 (95% CI: 0.65–2.58) in the community population, indicating that both populations with depressive symptoms exhibited a significantly higher prevalence of stroke. In conclusion, while the authors have already conducted comprehensive subgroup analyses, it is crucial to also examine the study population itself to gain a more comprehensive understanding of the relationship between stroke incidence and depressive symptoms. (Supplementary material: Figure S1). Thirdly, the article reported that various tools and cut-off values were used to assess depression/depressive symptoms in the included studies, but the assessment criteria were not clearly demonstrated in either the article or supplementary documents. Therefore, we suggest that authors provide additional information to help readers gain a more comprehensive understanding of the characteristics of the included articles. In conclusion, Cai et al. systematically summarized the current evidence of the impact of depression on the incidence of stroke, which brings about more attention for clinicians to early intervene in this special population and ultimately improve their quality of life. We believe that fully figure out these mentioned problems can provide the true prognostic value of depression for stroke morbidity among older people. Tian-Chao Chen and Xin-Juan Wu is responsible for design and writing, Xiao-Ming Zhang and Yun-Feng Bai are responsible for data analysis, Xin-Yi Liu and Yue-Ying Feng is responsible for data extraction. The peer review history for this article is available at https://www.webofscience.com/api/gateway/wos/peer-review/10.1111/acps.13638. Figure S1: Subgroup analysis of different setting. Table S1: Extracted data of the studied included in the meta-analysis. Please note: The publisher is not responsible for the content or functionality of any supporting information supplied by the authors. Any queries (other than missing content) should be directed to the corresponding author for the article.