Establishment and Verification of Risk Prediction Model for Silent Brain Infarction in Maintenance Hemodialysis Patients: a Multicenter Study

Background Maintenance hemodialysis (MHD) patients have a high incidence of silent brain infarction (SBI) and are in the preclinical stage of symptomatic stroke and vascular dementia. Therefore, there is a great need to explore the risk of SBI in patients with MHD for early detection and reduction of poor prognosis. Objective To explore the risk factors for the occurrence of SBI in MHD patients, a predictive model was constructed and its performance was evaluated. Methods 486 MHD patients from 4 centers (Nanchong Central Hospital Affiliated to North Sichuan Medical College, Guangyuan Central Hospital, Suining Central Hospital, and Pengan County People's Hospital) from January 2017 to October 2022 were included. Patients with MHD were divided into an SBI group (n=102) and a non-SBI group (n=384) using the presence or absence of SBI as the outcome event, and the baseline characteristics of the two study groups were compared. Patients were randomized in a 7∶3 ratio to the modeling set (n=340) and the validation set (n=146). The predictor variables were identified through LASSO regression and multifactorial Logistic regression analyses, and a risk prediction model for the occurrence of SBI in patients with MHD was constructed and presented as a nomographic chart. The predictive performance, accuracy, and clinical utility of the model were evaluated using area under the ROC curve, calibration curve, and decision curve analysis. Results In the modeling set, 70 cases (20.6%) of MHD patients experienced SBI, while in the validation set, 32 cases (21.9%) of patients experienced SBI. The results of LASSO regression combined with multifactor logistic regression analysis showed that age (OR=1.027, 95%CI=1.005-1.050), history of alcohol consumption (OR=4.487, 95%CI=2.075-9.706), BMI (OR=1.082, 95%CI=1.011-1.156), insufficient sleep or excessive sleep (OR=6.286, 95%CI=3.560-11.282), history of chronic disease (chronic obstructive pulmonary disease, diabetes, chronic hepatitis B) (OR=1.873, 95%CI=1.067-3.347), serum lactate level (OR=1.452, 95%CI=1.152-1.897), urea reduction ratio (URR) (OR=0.922, 95%CI=0.875-0.970), and history of antiplatelet medication (OR=0.149, 95%CI=0.030-0.490) were independent influences on the occurrence of SBI in MHD patients (P<0.05). A predictive model incorporating the aforementioned 8 influencing factors was constructed, and a nomographic chart was developed. The area under the ROC curve of the predictive model in the modeling set and validation set were 0.816 (95%CI=0.759-0.873) and 0.808 (95%CI=0.723-0.893), respectively, and the calibration curves show good consistency. DCA curve suggested that this model could provide maximum clinical benefit to patients. Conclusion A prediction model for the risk of SBI in MHD patients based on age, history of alcohol consumption, BMI, insufficient sleep or excessive sleep, history of chronic disease (chronic obstructive pulmonary disease, diabetes, chronic hepatitis B), serum lactate level, URR, and history of antiplatelet medication demonstrated good predictive performance and clinical utility. It is expected to accurately and individually assess the risk of SBI in MHD patients and implement early interventions to reduce the incidence rate.