Abstract
BACKGROUND: Post-stroke depression (PSD) refers to a depressive state that appears after stroke onset and is one of the most common complications in ischemic stroke patients. The occurrence of PSD exacerbates the risk of disability and increases the mortality of patients. Current diagnosis of PSD is severely underdiagnosed. METHODS: Patients hospitalized for acute ischemic stroke between December 2019 and November 2022 in the Department of Neurology of Sinopharm Gezhouba Central Hospital were retrospectively collected. Patients' basic clinical information, test data and related questionnaire scores were collected. They were divided into PSD group and NPSD group. Multivariate regression was used to analyze the risk factors of post-stroke depression and establish a risk prediction model to draw nomograms. Receiver Operating Characteristic Curve (ROC), Calibration curve and Decision Curve Analysis (DCA) decision curve were drawn using R language to assess the clinical efficacy and clinical utility of the model. RESULTS: Post-stroke depression in Acute Ischemic Stroke (AIS) patients was associated with single factors such as hypertension, living alone, education level, homocysteine level, National Institute of Health Stroke Scale (NIHSS) score, lymphocyte count, neutrophil count (P < 0.05). Among them, living alone, CRP level, hypertension, homocysteine level, education level, systemic immune inflammation index (SII), and NIHSS score were independent risk factors for post-stroke depression in AIS patients (P < 0.05). The seven selected variables were used to construct a risk prediction model, nomograms were drawn, and ROC curves were used to assess model discrimination, AUROC = 0.881. Calibration curve is used to evaluate the consistency of the model, DCA decision curve is used to evaluate the practicability of the model, and this model has good discrimination ability, calibration and clinical practicability. CONCLUSION: The probability of PSD in AIS patients in this study was 26.51%. Independent risk factors for developing PSD, including CRP level, living alone, history of hypertension, homocysteine level, education level, SII, NIHSS score to establish risk prediction model and draw nomograms. The model was demonstrated to have good discrimination, calibration and clinical utility by internal validation.