住院脑梗死患者抑郁相关因素分析及风险预警模型的构建

目的 构建脑梗死后抑郁风险预警模型，为医护人员早期、快速识别高危患者，及时行干预措施提供理论依据。方法 选取住院脑梗死患者142例，单因素分析相关资料；二分类logistic回归分析确定脑梗死后抑郁的独立危险因素并建立风险预警模型；ROC曲线判断各独立指标与风险预警模型诊断的准确性；交叉验证法检验风险预警模型效能。结果 脑梗死后抑郁的发生率为33.80%。单因素分析显示脑梗死后抑郁与NIHSS评分、Barthel指数、MRS、FSS、对支持的利用度、文化程度、家庭人均月收入、居住地、手术史、总胆固醇、甘油三酯有关（P＜0.05）。二分类logistic回归分析显示手术史、Barthel指数、FSS是脑梗死后抑郁的独立危险因素。手术史、Barthel指数、FSS及基于独立危险因素建立的风险预警模型ROC曲线下面积/灵敏度/特异度分别为0.599/72.9%/46.80%、0.744/64.6%/77.70%、0.829/68.8%/85.10%、0.928/72.90%/97.87%，交叉检验风险预警模型准确率为85.92%。结论 手术史、Barthel指数、FSS是脑梗死患者发生抑郁的独立危险因素，基于此建立的风险预警模型对脑梗死后抑郁预警效果良好。

Analysis of related factors of depression in hospitalized patients with cerebral infarction and construction of risk warning model

Objective To construct an early warning model of depression risk after cerebral infarction to provide theoretical basis for doctors and nurses to identify high-risk patients early and quickly and take intervention measures in time. Methods A total of 142 hospitalized patients with cerebral infarction were selected and the related data were analyzed by univariate analysis. Binary Logistic regression analysis was performed to determine the independent risk factors of depression after cerebral infarction and establish a risk early warning model. ROC curve was drawn to determine the accuracy of each independent index and risk early warning model diagnosis. Cross-validation method was used to test the effectiveness of the risk early warning model. Results The incidence of depression after cerebral infarction was 33.80%. Univariate analysis showed that depression after cerebral infarction was associated with National Institute of Health stroke scale(NIHSS) score,Barthel index, modified Rankin scale(MRS), fraction skill score(FSS), utilization of support, education level, family per capita monthly income, place of residence, history of surgery, total cholesterol and triglyceride(P<0.05). Binary Logistic regression analysis showed that operation history, Barthel index and FSS were independent risk factors for depression after cerebral infarction. The area/sensitivity/specificity under the ROC curve of operation history, Barthel index, FSS and risk early warning model based on independent risk factors were 0.599/72.90%/46.80%, 0.744/64.60%/77.70%, 0.829/68.80%/85.10%, 0.928/72.90%/97.87%, respectively. The accuracy of cross-validation test risk early warning model was 85.92%. Conclusion Surgical history, Barthel index, and FSS are independent risk factors for depression in patients with cerebral infarction. The risk warning model based on these factors is effective in early warning of depression after cerebral infarction.