术前预测感染性肾结石列线图模型的构建及验证  

Construction and validation of a nomogram model for predicting infectious kidney stones before surgery

作　　者：郑鹏 吴赛萍[1] 谢秀璋 史庆丰 Peng Zheng;Saiping Wu;Xiuzhang Xie;Qingfeng Shi(Department of Infection Management,Wujin Hospital Affiliated to Jiangsu University(Wujin Clinical College of Xuzhou Medical University),Changzhou 213000,China;Department of Infection Management,Zhongshan Hospital,Fudan University,Shanghai 200032,China)

机构地区：[1]江苏大学附属武进医院(徐州医科大学武进临床学院)感染管理科,常州市213000 [2]复旦大学附属中山医院感染管理科,上海200032

出　　处：《中华实验和临床感染病杂志（电子版）》2023年第5期299-306,共8页Chinese Journal of Experimental and Clinical Infectious Diseases(Electronic Edition)

基　　金：2019年度上海市卫生健康委员会卫生行业临床研究专项项目(No.201940413)。

摘　　要：目的建立一款用于术前预测感染性肾结石的列线图模型并进行验证。方法回顾性总结2020年2月至2023年2月江苏大学附属武进医院(徐州医科大学武进临床学院)诊断肾结石患者共350例,按7︰3随机分为建模集(245例)和验证集(105例)。建模集中感染性肾结石91例和无感染性肾结石154例,验证集中感染性肾结石患者39例和无感染性肾结石患者66例。比较建模集中感染性肾结石组与无感染性肾结石组患者的临床资料,采用最小绝对值收敛和选择算子回归(Lasso)模型和多因素Logistic回归模型筛选感染性肾结石的危险因素,R软件建立列线图模型,采用1000次自我重复采样进行验证。结果单因素分析发现,建模集患者中感染性肾结石组女性、复发性肾结石和鹿角形结石、结石面积较无感染性肾结石组显著增加,而结石Hounsfield单位(HU)显著降低;术前膀胱尿液培养阳性(PBUC)、尿白细胞计数(WBC)和细菌量、尿蛋白阳性、尿亚硝酸盐阳性、尿白细胞酯酶阳性(ULE)、尿pH值升高、尿液浊度阳性例数占比较无感染性肾结石组显著增多,而尿比重显著下降;血尿酸水平较无感染性肾结石组降低,但血磷和血镁升高,差异均有统计学意义(P均<0.05)。Lasso回归分析筛选出8个最具差异性的指标,即女性、复发性肾结石、结石面积≥601 mm^(2)、HU值<1000、阳性PBUC、阳性ULE、尿pH和尿液浊度阳性。Logistic回归分析显示,女性(OR=1.568、95%CI:1.231~1.902、P<0.001)、复发性肾结石(OR=3.023、95%CI:2.568~3.467、P<0.001)、结石面积≥601 mm^(2)(OR=2.123、95%CI:1.756~2.569、P<0.001)、HU值<1000(OR=3.856、95%CI:3.456~4.325、P<0.001)、阳性PBUC(OR=1.895、95%CI:1.623~2.325、P<0.001)、阳性ULE(OR=1.754、95%CI:1.326~2.124、P<0.001)、尿pH>6.5(OR=1.323、95%CI:1.102~1.889、P<0.001)和尿液浊度阳性(OR=1.602、95%CI:1.314~1.956、P<0.001)均为感染性肾结石的危险因素。通过R软件建立列线�Objective To establish a nomogram model for preoperative prediction of infectious kidney stones.Methods Total of 350 patients with kidney stones diagnosed in Wujin Hospital Affiliated to Jiangsu University(Wujin Clinical College of Xuzhou Medical University)from February 2020 to February 2023 were summarized,retrospectively,and were randomly divided into a modeling set(245 cases)and a validation set(105 cases)according to 7︰3.The modeling focused on 91 cases with infectious kidney stones and 154 cases with non-infectious kidney stones,and verified 39 cases of concentrated infectious kidney stones and 66 cases of non-infectious kidney stones.The clinical data of patients in infective kidney stone group and non-infective kidney stone group were compared in the modeling set.The minimum absolute convergence and selection operator regression(Lasso)model and multi-factor Logistic regression model were used to screen the risk factors of infective kidney stone.The nomographic model was established and verified by 1000 self-repeated samples through R software.Results Univariate comparison showed that female,recurrent kidney stones and staghorn stones in the infectious kidney stones group were more,the stone area was larger,while the Hounsfield unit(HU)of stones was significantly fewer;positive preoperative bladder urine culture(PBUC),urine white blood cell count(WBC)and bacterial count,urine protein positive,urine nitrite positive,positive urine leukocyte esterase(ULE),urine pH value,and urine turbidity positive were significantly higher,while urine specific gravity was significantly lower;blood uric acid was lower,while blood phosphorus and magnesium were higher(all P<0.05).Lasso screened 8 most differential indicators,namely female,recurrent kidney stones,stone area≥601 mm^(2),HU value<1000,positive PBUC,positive ULE,urine pH and urine turbidity positive.Logistic regression showed that female(OR=1.568,95%CI:1.231-1.902,P<0.001),recurrent kidney stones(OR=3.023,95%CI:2.568-3.467,P<0.001),stone area≥601 mm^(2)(OR=2.

关 键 词：感染性肾结石 列线图模型 术前膀胱尿液培养 尿白细胞酯酶 

分 类 号：R692[医药卫生—泌尿科学]