作者：兰怡昕1 ，邱小芩2 ，黄兰青1 ，韦晓静3 ，韩瑞林4 ，彭婉琳1 ，姜晓冬5 ，蓝春晗5

单位：

1.530000广西壮族自治区南宁市，广西中医药大学研究生院　2.530022广西壮族自治区南宁市，中山大学 附属第一医院广西医院护理部　3.533000广西壮族自治区百色市，右江民族医学院研究生学院　4.710021陕西省西安 市中医医院心血管内科　5.530021广西壮族自治区南宁市，广西壮族自治区人民医院心血管内科

Units：

1.Graduate College of Guangxi University of Chinese Medicine, Nanning 530000, China 2.Department of Nursing, Guangxi Hospital Division of the First Affiliated Hospital, Sun Yat-sen University, Nanning 530022, China 3.Graduate School, Youjiang Medical University for Nationalities, Baise 533000, China 4.Department of Cardiovascular Medicine, Xi'an Hospital of Traditional Chinese Medicine, Xi'an 710021, China 5.Department of Cardiovascular Medicine, the People's Hospital of Guangxi Zhuang Autonomous Region, Nanning 530021, China

关键词：

心力衰竭；低蛋白血症；老年人；影响因素分析；列线图

Keywords：

Heart failure; Hypoproteinemia; Aged; Root cause analysis; Nomograms

CLC：

R 541.62 R 591.2

DOI：

10.12114/j.issn.1008-5971.2024.00.107

Funds：

广西医疗卫生适宜技术开发与推广应用项目（S2023003，S201644）；广西中医药大学研究生教育创新计划项目 （YCSY2023052）

摘要：

目的　分析老年慢性心力衰竭（CHF）患者发生低蛋白血症的影响因素，构建并验证其风险预测列 线图模型。方法　选取2022年12月—2023年5月在广西壮族自治区人民医院心血管内科住院治疗的老年CHF患者355 例，按照7∶3的比例将其随机分为建模组249例和内部验证组106例。选取2023年6—8月于中山大学附属第一医院广 西医院心血管内科住院治疗的老年CHF患者60例为外部验证组。收集患者临床资料，采用多因素Logistic回归分析探 讨建模组老年CHF患者发生低蛋白血症的影响因素；基于多因素Logistic回归分析结果，构建老年CHF患者发生低蛋 白血症的风险预测列线图模型；采用Hosmer-Lemeshow检验、校准曲线评估该列线图模型的拟合程度，采用ROC曲 线评估该列线图模型的预测效能。结果　根据是否发生低蛋白血症，将建模组老年CHF患者分为非低蛋白血症亚组 133例和低蛋白血症亚组116例。两组BMI、住院时间、纽约心脏病协会（NYHA）分级、冠心病发生率、合并水肿者 占比、日常生活活动能力（ADL）分级、利尿剂口服频率、强心药口服频率、血清钙、总胆红素、血浆D-二聚体、 红细胞计数异常者占比、C反应蛋白、肌酐、丙氨酸氨基转移酶、天冬氨酸氨基转移酶、三酰甘油比较，差异有统计 学意义（P＜0.05）。多因素Logistic回归分析结果显示，住院时间、合并水肿、总胆红素、血浆D-二聚体、红细胞 计数异常、C反应蛋白是建模组老年CHF患者发生低蛋白血症的影响因素（P＜0.05）。基于多因素Logistic回归分析 结果，构建老年CHF患者发生低蛋白血症的风险预测列线图模型。Hosmer-Lemeshow 检验及校准曲线分析结果显示， 该列线图模型在建模组、内部验证组、外部验证组中的拟合情况较好（χ 2 值分别为4.416 、7.671、5.812，P值分别为 0.818、0.715、0.342）。ROC曲线分析结果显示，该列线图模型预测建模组、内部验证组、外部验证组老年CHF患者 发生低蛋白血症的AUC分别为0.808〔95%CI（0.755～0.862）〕、0.702〔95%CI（0.596～0.808）〕、0.748〔95%CI （0.545～0.951）〕。结论　住院时间延长、合并水肿、总胆红素≥26 μmol/L、血浆D-二聚体≥0.5 mg/L、红细胞计 数异常、C反应蛋白≥8 mg/L是老年CHF患者发生低蛋白血症的危险因素，本研究基于上述影响因素构建的老年CHF患 者发生低蛋白血症的风险预测列线图模型拟合情况较好，且其对老年CHF患者发生低蛋白血症有一定预测价值。

Abstract：

Objective To analyze the influencing factors of hypoproteinemia in elderly patients with chronic heart failure (CHF) , and establish and verify its risk prediction nomogram model. Methods A total of 355 elderly CHF patients hospitalized in the Department of Cardiovascular Medicine, the People's Hospital of Guangxi Zhuang Autonomous Region from December 2022 to May 2023 were selected and randomly divided into modeling group (249 cases) and internal validation group (106 cases) according to a ratio of 7∶3. Sixty elderly patients with CHF who were hospitalized in the Cardiovascular Department of Guangxi Hospital Division of the First Affiliated Hospital, Sun Yat-sen University from June to August 2023 were selected as the external validation group. The clinical data of the patients were collected, and the influencing factors of hypoproteinemia in elderly CHF patients in modeling group were investigated by multivariate Logistic regression analysis. Based on the results of multivariate Logistic regression analysis, the risk prediction nomogram model of hypoproteinemia in elderly CHF patients was constructed. Hosmer-Lemeshow test and calibration curve were used to evaluate the fitting degree of the nomogram model. ROC curve was used to evaluate the prediction efficiency of the nomogram model. Results According to the occurrence of hypoproteinemia, the elderly CHF patients in modeling group were divided into non-hypoproteinemia subgroup (133 cases) and hypoproteinemia snbgroup (116 cases) . There was statistical significance in BMI, length of stay, New York Heart Association (NYHA) grade, incidence of coronary heart disease, proportion of patients with edema, activity of daily living (ADL) grade, oral frequency of diuretics, oral frequency of cardiac drugs, serum calcium, total bilirubin, plasma D-dimer, proportion of patients with abnormal red blood cell count, C-reactive protein, creatinine, alanine aminotransferase, aspartate aminotransferase, and triacylglycerol between the two groups (P < 0.05) . Multivariate Logistic regression analysis showed that length of stay, edema, total bilirubin, plasma D-dimer, abnormal red blood cell count and C-reactive protein were the influencing factors for hypoproteinemia in elderly CHF patients in modeling group (P < 0.05) . Based on the results of multivariate Logistic regression analysis, the risk prediction nomogram model of hypoproteinemia in elderly CHF patients was constructed. Hosmer-Lemeshow test and calibration curve analysis results showed that the nomogram model fitted well in the modeling group, internal validation group, and external validation group (the χ 2 values were 4.416, 7.671 and 5.812 respectively, the P values were 0.818, 0.715 and 0.342 respectively) . ROC curve analysis results showed that, the AUC of the modeling group in predicting hypoproteinemia in elderly CHF patients in modeling group, internal validation group and external validation group was 0.808 [95 %CI (0.755-0.862) ] , 0.702 [95 %CI (0.596-0.808) ] and 0.748 [95 %CI (0.545-0.951) ] , respectively. Conclusion Prolonged length of stay, edema, total bilirubin ≥ 26 μmol/L, plasma D-dimer ≥ 0.5 mg/L, abnormal red blood cell count, C-reactive protein ≥ 8 mg/L are the risk factors for hypoproteinemia in elderly CHF patients. In this study, the risk prediction nomogram model of hypoproteinemia in elderly CHF patients constructed based on the above influencing factors has a good degree of fitting, and it has certain predictive value for the occurrence of hypoproteinemia in elderly CHF patients.

ReferenceList：