心脏再同步化治疗患者非计划再入院风险预测模型的构建及验证

Construction and validation of a risk prediction model for unplanned readmission of patients undergoing cardiac resynchronization therapy

目的构建心脏再同步化治疗(CRT)患者非计划性再入院风险预测模型,并验证模型的应用效果。方法采用便利抽样法,选取2017年7月—2020年7月在郑州大学第一附属医院心血管内科行CRT后出院的患者为建模组(n=279)和内部验证组(n=120);选取2021年8月—2022年8月因相同或相关疾病收治于郑州大学第一附属医院心血管内科的CRT患者为外部验证组(n=86)。采用多因素Logistic回归分析探讨CRT患者非计划再入院的影响因素并建立预测模型,通过Hosmer-Lemeshow检验和ROC曲线评估模型的拟合效果及区分度,基于R-4.1.2和Rstudio软件建立列线图。结果多因素Logistic回归分析结果显示,肌酐、左心房内径、肺动脉收缩压、NYHA分级、BMI是CRT患者非计划再入院的危险因素(P<0.05)。预测模型公式为:P=1/{1+exp〔-(0.792×肌酐+1.408×左心房内径+0.887×肺动脉收缩压+0.769×NYHA分级-0.970×BMI-2.266)〕}。ROC曲线下面积为0.874,约登指数最大值为0.636,最佳界值为0.256,灵敏度为0.826,特异度为0.810。内部验证和外部验证的准确率分别为90.00%和90.70%。结论构建的CRT患者非计划再入院预测模型预测效果较好,可视化的列线图提高了模型的实用性能,有助于医护人员早期识别CRT非计划再入院高风险人群,为制订不同风险人群护理策略提供依据。

Objective To construct a risk prediction model for unplanned readmission of patients undergoing cardiac resynchronization therapy(CRT)and verify the performance of the model.Methods Using convenience sampling,patients who underwent CRT at the Department of Cardiovascular of the First Affiliated Hospital of Zhengzhou University from July 2017 to July 2020 were selected as the modeling group(n=279)and the internal validation group(n=120).CRT patients admitted to the Department of Cardiovascular of the First Affiliated Hospital of Zhengzhou University from August 2021 to August 2022 due to the same or related diseases were selected as the external validation group(n=86).Multivariate Logistic regression was used to explore the influencing factors of unplanned readmission of CRT patients and establish the prediction model.The fitting effect and discrimination of the model were evaluated through the Hosmer-Lemeshow test and receiver operating characteristic(ROC)curve.The nomogram was established based on R-4.1.2 and Rstudio software.Results The multivariate Logistic regression analysis showed that creatinine,left atrial diameter,pulmonary artery systolic pressure,New York Heart Association(NYHA)classification,and body mass index(BMI)were risk factors for unplanned readmission in CRT patients,with statistically significant differences(P<0.05).The prediction model formula was:P=1/{1+exp[-(0.792×creatinine+1.408×left atrial inner diameter+0.887×pulmonary artery systolic pressure+0.769×NYHA classification-0.970×BMI-2.266)]}.The area under the ROC curve was 0.874,the maximum value of the Jordan index was 0.636,the optimal threshold was 0.256,the sensitivity was 0.826,and the specificity was 0.810.The accuracy of internal validation and external validation was 90.00%and 90.70%,respectively.Conclusions The constructed prediction model for unplanned readmission of CRT patients has good predictive performance,and the visualized nomogram improves the practical performance of the model.It helps medical and nursing staff identify high-risk groups of unplanned readmission of CRT patients in the early stage and provides a basis for formulating nursing strategies for different risk groups.