Association between cerebrovascular disease and perioperative neurocognitive disorders: a retrospective cohort study.

Abstract

Previous studies have shown that patients with cerebrovascular disease (CVD) have a significantly increased risk of cognitive decline or dementia; however, the association between preoperative CVD and perioperative neurocognitive disorders (PNDs) remains unclear. This study aimed to explore the correlation between preoperative CVD and PNDs, as well as combine logistic regression and ROC curves to construct a clinical prediction PND model. This retrospective cohort study evaluated 13,899 surgical patients of a large-scale comprehensive hospital between January 2021 and January 2022 to explore the association between preoperative CVD and PNDs, with follow-up to monitor postoperative survival until February 28, 2023, unless the patient died. The study participants comprised all inpatients from the Bone and Joint Surgery, Spine Surgery, Urology, Hepatobiliary Surgery, Gastrointestinal Surgery, and Thoracic Surgery departments. Patients were classified into two groups: the CVD group with a confirmed diagnosis and the non-cerebrovascular disease (NCVD) group. The incidence of PNDs was measured, and potential associations between patient demographic information, preoperative comorbidities, and CVD, as well as the correlation between preoperative CVD and PNDs, were investigated by multivariate logistic regression analysis. Next, we constructed a clinical prediction PND model by drawing the receiver operating characteristic (ROC) curve. The postoperative survival of all patients was tracked, and a survival curve was constructed and incorporated into the Cox proportional hazard regression model to analyze the relationship between preoperative CVD and overall postoperative survival rate. Of the included 13,899 patients, propensity score matching yielded 1006 patient pairs. Multivariate logistic regression analysis revealed that CVD was an independent risk factor for PNDs [Odds ratio: 10.193; 95% confidence interval: 7.454-13.938; P<0.001]. Subsequently, we developed a clinical prediction model for PNDs by multivariate logistic regression analysis. The area under the ROC curve was 0.798 (95% CI: 0.765-0.830). The survival of 11,702 patients was followed up. Multivariate Cox hazard ratio regression analysis revealed that CVD affected the overall postoperative survival rate (Hazard ratio, 1.398; 95% confidence interval: 1.112-1.758; P＜0.001). CVD was an independent risk factor for PNDs and affected the overall postoperative survival rate of surgical patients with preoperative CVD.