Department of Biomedical Science and Engineering, Gwangju Institute of Science and Technology, Gwangju, Korea.
Department of Urology, School of Medicine, Kyungpook National University, Kyungpook National University Hospital, Daegu, Korea.
Medicine (Baltimore). 2024 Aug 2;103(31):e38950. doi: 10.1097/MD.0000000000038950.
To identify the best combination of potential predictors of septic shock in patients with obstructive acute pyelonephritis associated with ureteral stones (OAPN-US) according to Sepsis-3 criteria. Patients who underwent percutaneous nephrostomy (PCN) with OAPN-US were retrospectively evaluated. Recursive feature elimination (RFE) was applied to patients with and without septic shock to identify factors associated with the prediction of progression to septic shock. We compared combinations of the selected features based on area under the receiver operating curve (AUROC) to determine which combination was most effective. This study included 81 patients who were treated with PCN due to OAPN-US. A comparison was made between 37 patients with septic shock (SS) and 44 patients without septic shock (NSS). SS group had a higher age, poorer Eastern Cooperative Oncology Group status, and significantly higher levels of positivity in urine cultures and blood cultures. There were also differences in laboratory tests between the 2 groups. Procalcitonin (PCT), international normalized ratio (INR), and absolute lymphocyte count (ALC) were selected based on RFE. We compared the predictive power for SS when each marker was used alone, when 2 markers were combined, and when all 3 markers were combined. Among these combinations, using all 3 variables together yielded the highest AUROC of 0.942. Of the 3 variables, PCT had the highest Gini importance score, indicating that it was the most influential factor. Clinical characteristics were different between the SS and the NSS groups. In patients with OAPN-US, the combination of PCT, ALC, and INR was an excellent predictor of septic shock.
为了根据 Sepsis-3 标准确定患有与输尿管结石相关的梗阻性急性肾盂肾炎(OAPN-US)并发脓毒症性休克的患者中脓毒症性休克的最佳潜在预测因子组合。回顾性评估了接受经皮肾造口术(PCN)治疗的 OAPN-US 患者。对有和没有脓毒症性休克的患者应用递归特征消除(RFE)来识别与预测进展为脓毒症性休克相关的因素。我们比较了基于接收者操作特征曲线下面积(AUROC)的选定特征组合,以确定哪种组合最有效。这项研究包括 81 名因 OAPN-US 接受 PCN 治疗的患者。比较了 37 例脓毒症性休克(SS)患者和 44 例无脓毒症性休克(NSS)患者。SS 组年龄较大,东部合作肿瘤组(ECOG)状态较差,尿液培养和血液培养阳性率显著较高。两组实验室检查也存在差异。基于 RFE,选择降钙素原(PCT)、国际标准化比值(INR)和绝对淋巴细胞计数(ALC)。我们比较了单独使用每种标志物、组合使用两种标志物和组合使用三种标志物时对 SS 的预测能力。在这些组合中,同时使用所有三种变量的 AUROC 最高,为 0.942。在这三个变量中,PCT 的基尼重要性得分最高,表明其是最具影响力的因素。SS 组和 NSS 组之间的临床特征不同。在患有 OAPN-US 的患者中,PCT、ALC 和 INR 的组合是脓毒症性休克的优秀预测因子。