新生儿坏死性小肠结肠炎手术时机临床分析及其预测模型的构建
Clinical Analysis of Surgical Opportunity and Construction of Predictive Model for Neonatal Necrotizing Enterocolitis

目的：根据Bell分期对新生儿坏死性小肠结肠炎(necrotizing enterocolitis, NEC)手术患儿进行分组及临床相关数据分析，比较得出治疗效果最佳时机，并探究构建Bell-II期至Bell-III期阶段的患儿手术时机预测模型，将手术时机的选择量化，为手术治疗NEC提供一定依据。方法：回顾性分析2006年6月至2021年6月于青岛大学附属医院小儿外科进行手术治疗的59例NEC患儿以及经过禁饮食胃肠减压等保守治疗72 h后病情好转的61名患儿临床资料，根据Bell分期将接受手术患儿分为Bell-II期手术组(n = 27)与Bell-III期手术组(n = 32)，对照组为非手术组患儿。第一部分选取手术组与非手术组两组患儿入院诊断后治疗3天的临床数据如一般情况、生命体征和辅助检查等，以及对两手术组患儿围手术期临床数据的比较，选取敏感炎症指标降钙素原(PCT)、C反应蛋白(CRP)以及患儿住院时长和并发症情况，应用SPSS 25.0软件分析得出最佳手术时机；第二部分将两手术组数据采用R软件进行单因素和多因素Logistic回归筛选特异性指标，综合分析各项数据，建立手术时机预测模型，通过受试者工作特征(receiver operating characteristic, ROC)曲线下面积评估模型拟合效果并拟合优度验证。结果：第一部分数据显示相对照于非手术组患儿，手术组患儿病情指标进展有统计学意义，并且手术前和手术后Bell-III期手术组的 PCT、CRP水平均高于Bell-II期手术组(P < 0.05)，且Bell-III期手术组的住院时间及并发症发生率水平均高于Bell-II期手术组(P < 0.05)；Logistic单因素及多因素回归分析后显示术前血小板减少、中性粒细胞减少、酸中毒、腹胀和影像学Duke评分为对Bell-II期和Bell-III期手术时机差异产生影响的独立因素(P < 0.05)；该手术时机预测模型ROC曲线下面积为0.9；拟合优度检验，评估预测模型准确度，得到？2 = 3.444，P = 0.9；绘制列线图的校准曲线为斜率接近1的直线。结论：Bell-II期手术患儿较Bell-III期手术患儿术后恢复好，并发症少，康复时间短；手术时机预测模型列线图有较好的准确性，对于已进入Bell-II期至Bell-III期阶段的NEC患儿在手术时机的量化选择上具有一定的临床参考价值。
Objective: According to Bell staging, the patients with necrotizing enterocolitis (NEC) were divided into groups and clinical data were analyzed, and the best treatment time was obtained. The predic-tion model of surgical opportunity from Bell-stage II to Bell-stage III was constructed, and the choice of surgical opportunity was quantified so as to provide a basis for surgical treatment of NEC. Methods: The clinical data of 59 children with NEC who underwent surgery in the Department of Pediatric surgery of the Affiliated Hospital of Qingdao University from June 2006 to June 2021 and 61 children who improved after 72 hours of conservative treatment such as fasting gastrointestinal decompression were retrospectively analyzed. According to Bell stage, the children were divided into Bell-stage II operation group (n = 27) and Bell-stage III operation group (n = 32), and the con-trol group was non-operation group. In the first part, the clinical data such as general condition, vi-tal signs and auxiliary examination 3 days after admission between the operation group and the non-operation group were selected, and the perioperative clinical data of the two groups were compared, and the sensitive inflammatory indexes such as procalcitonin (PCT), C-reactive protein (CRP), length of stay and complications were selected, and the best operation time was obtained by SPSS 25.0 software analysis. In the second part, the two groups of data were screened by sin-gle-factor and multi-factor Logistic regression with R software, and the surgical