输尿管结石致尿脓毒血症的预测模型建立与评估

Establishment of predictive model for the risk of renal sepsis caused by ureteral calculi

摘要:

目的分析输尿管结石致尿脓毒血症的相关危险因素，建立预测输尿管结石发生尿脓毒血症风险的预测模型。方法选取2013年6月-2015年12月该科收治747例输尿管结石患者的临床资料。选取输尿管结石并发尿脓毒血症患者62 例作为病例组研究对象，随机抽取同期本院住院的无尿脓毒血症的输尿管结石患者685 例作为对照组研究对象。通过单因素分析和多因素Logistic回归分析评价输尿管结石患者发生尿脓毒血症的危险因素，根据回归系数建立相应的预测模型，并绘制受试者工作特征曲线。结果单因素分析结果显示，性别、年龄、糖尿病、高血压、既往结石手术史、结石长径、结石短径、结石平均CT 值、合并同侧肾结石、肾积水平均CT 值、尿培养、尿常规白细胞、尿亚硝酸盐和功能性孤立肾等危险因素，差异有统计学意义（P <0.05）。多因素Logistic 回归分析结果显示，性别、肾积水平均CT值、尿培养、尿常规WBC、尿亚硝酸盐及功能性孤立肾6个因素是输尿管结石并发尿脓毒血症的危险因素（P <0.05）。受试者工作特征曲线下面积为（0.934±0.012）。Hosmer and Lemeshow 检验提示，预测模型拟合（χ2=5.357，P =0.616）。结论女性患者、肾积水平均CT 值高、尿培养阳性、尿常规WBC 阳性、尿亚硝酸盐阳性及功能性孤立肾可增加输尿管结石患者尿脓毒血症发生率，根据患者临床资料，应用预测模型可有助于提高输尿管结石尿脓毒血症高危患者的筛选和识别能力。

Abstract:

Objective To investigate the risk factors for renal sepsis caused by ureteral calculi, and to establish a regression model for predicting occurrence of renal sepsis. Methods A total of 747 patients diagnosed with ureteral calculi were retrospectively analyzed. Of which 62 patients developed renal sepsis, and 685 cases without renal sepsis served as control group. Single factor analysis and logistic regression analysis were performed for establishment of a predictive model for development of renal sepsis. Receiver operative characteristic curve was utilized to verify the diagnostic value of the novel model. Results Single factor analysis showed that gender, age, diabetes mellitus, hypertension, operation history of urinary calculi, calculi length and width，mean CT attenuation value of calculi, ipsilateral renal calculi, mean CT attenuation value of hedronephrosis, urine routine WBC, urine culture, urine routine nitrite and functional solitary kidney were significantly different between the two groups (P < 0.05). Logistic regression analysis showed that gender, mean CT attenuation value of hydro-nephrosis, urine culture, urine routine WBC, urine nitrite and functional solitary kidney were the independent risk factors for development of renal sepsis induced by ureteral calculi ( P<0.05). The area under the receiver operating characteristic curve of the predictive model was (0.934 ±0.012).Hosmer and Lemeshow Test indicated a good fitting of the predictive model (χ2 = 5.357, P= 0.616). Conclusions Female patients, high mean CT attenuation value of hydro -nephrosis, positive urine culture,positive urine routine WBC, positive urine nitrite, and functional solitary kidney are independent risk factors to predict occurrence of renal sepsis induced by ureteral calculi, which is informative for early diagnosis.