目的:本研究旨在探讨慢性肾脏病(CKD)患者肺部感染的相关危险因素。主要目标是开发一种预测模型,该模型可以预测CKD患者住院期间肺部感染的可能性。
方法:这项回顾性队列研究在两家著名的三级教学医院进行。采用三种不同的方法制定了三种不同的模型:(1)统计驱动模型,(2)临床知识驱动模型,(3)决策树模型。通过比较它们的预测能力,获得了最简单和最有效的模型,稳定性,和实用性。
结果:本研究共涉及971名患者,388名个体组成建模组,583名个体组成验证组。三种不同的模型,即模型A,B,C,被利用,导致七个人的身份,四,和11个预测因子,分别。最终,选择了统计知识驱动模型,表现出0.891(0.855-0.927)的C统计量和0.012的Brier评分。此外,Hosmer-Lemeshow测试表明该模型具有良好的校准效果。此外,模型A在外部验证期间显示出令人满意的C统计量0.883(0.856-0.911)。统计驱动模型,称为A-C2GH2S风险评分(其中包括白蛋白、C2[既往COPD病史,血钙],随机静脉血糖,H2[血红蛋白,高密度脂蛋白],吸烟),用于确定CKD患者肺部感染发生率的风险评分。研究结果表明,肺部感染的发生率逐渐增加,A-C2GH2S风险评分≤6的个体为1.84%,A-C2GH2S风险评分≥18.5的个体为93.96%。
结论:建立了一个包含7个预测因子的预测模型来预测CKD患者的肺部感染。该模型的特点是简单,实用性,验证后具有良好的特异性和敏感性。
OBJECTIVE: The objective of this
study is to investigate the associated risk factors of pulmonary infection in individuals diagnosed with chronic kidney disease (CKD). The primary goal is to develop a predictive model that can anticipate the likelihood of pulmonary infection during hospitalization among CKD patients.
METHODS: This retrospective cohort
study was conducted at two prominent tertiary teaching hospitals. Three distinct models were formulated employing three different approaches: (1) the statistics-driven model, (2) the clinical knowledge-driven model, and (3) the decision tree model. The simplest and most efficient model was obtained by comparing their predictive power, stability, and practicability.
RESULTS: This
study involved a total of 971 patients, with 388 individuals comprising the modeling group and 583 individuals comprising the validation group. Three different models, namely Models A, B, and C, were utilized, resulting in the identification of seven, four, and eleven predictors, respectively. Ultimately, a statistical knowledge-driven model was selected, which exhibited a C-statistic of 0.891 (0.855-0.927) and a Brier score of 0.012. Furthermore, the Hosmer-Lemeshow test indicated that the model demonstrated good calibration. Additionally, Model A displayed a satisfactory C-statistic of 0.883 (0.856-0.911) during external validation. The statistical-driven model, known as the A-C2GH2S risk score (which incorporates factors such as albumin, C2 [previous COPD history, blood calcium], random venous blood glucose, H2 [hemoglobin, high-density lipoprotein], and smoking), was utilized to determine the risk score for the incidence rate of lung infection in patients with CKD. The findings revealed a gradual increase in the occurrence of pulmonary infections, ranging from 1.84% for individuals with an A-C2GH2S Risk Score ≤ 6, to 93.96% for those with an A-C2GH2S Risk Score ≥ 18.5.
CONCLUSIONS: A predictive model comprising seven predictors was developed to forecast pulmonary infection in patients with CKD. This model is characterized by its simplicity, practicality, and it also has good specificity and sensitivity after verification.