需要治疗的早产儿视网膜病变独立预测因素分析与列线图预测模型构建  

Independent factors analysis and prediction model development of treatment-requiring retinopathy of prematurity

作　　者：许瑜玲 孙伟 张夏茵 李静[2] 余洪华[1] 吴乔伟 Xu Yuling;Sun Wei;Zhang Xiayin;Li Jing;Yu Honghua;Wu Qiaowei(Guangdong Eye Institute,Southern Medical University,Department of Ophthalmology,Guangdong Provincial People's Hospital(Guangdong Academy of Medical Sciences),Guangzhou 510080,China;Department of Ophthalmology,Guangdong Women and Children Hospital,Guangzhou 511400,China)

机构地区：[1]南方医科大学广东省眼科研究所广东省人民医院(广东省医学科学院)眼科,广州510080 [2]广东省妇幼保健院眼科,广州511400

出　　处：《中华眼底病杂志》2024年第10期750-757,共8页Chinese Journal of Ocular Fundus Diseases

基　　金：国家自然科学基金(面上项目)(82171075);中国博士后科学基金(2021M700899);广州市重点研发计划(202206010092、202103000045);广东省人民医院人才引进基金(Y012018145);广东省人民医院自然科学基金启动基金(8217040546);精准医学研究与产业创新发展基金个体化医疗孵化项目(KH012023367);广东省人民医院“登峰计划”杰出青年培育人才项目(粤医[2019]10号)(KJ012019087);广东省人民医院博士后进站启动项目(BY012021057)。

摘　　要：目的分析需要治疗的早产儿视网膜病变(TR-ROP)的独立影响因素, 建立预测TR-ROP的列线图模型。方法回顾性队列研究。2012年1月1日至2022年3月31日于广东省妇幼保健院出生并接受早产儿视网膜病变(ROP)筛查的早产儿6 998名纳入研究。TR-ROP定义为1型ROP和急进型ROP。详细收集受检儿一般资料、母亲围产期情况、干预措施及新生儿疾病等22个临床因素。根据临床经验以8:2的比例将受检早产儿分为训练集、验证集, 分别为5 598、1 400名。组间定量资料比较行t检验, 计数资料比较行χ^(2)检验。将单因素分析有差异的指标进行多因素logistic回归分析, 利用R软件可视化回归分析结果建立列线图预测模型, 通过一致性指数(C-index)、受试者工作特征曲线(ROC曲线)验证列线图预测模型的准确性。结果 6 998名受检儿中, 男性4 069名, 女性2 920名;出生胎龄(33.69±3.19)周;出生体重(2 090± 660)g。TR-ROP 376例(5.4%, 376/6 998)。多因素logistic回归分析结果显示, 出生胎龄[比值比(OR)=0.63, 95%可信区间(CI)0.47~0.85, P=0.002]、宫内窘迫(OR=0.30, 95%CI 0.10~0.99, P=0.048)、支气管肺发育不良(OR=0.23, 95%CI 0.09~0.60, P=0.003)、缺氧缺血性脑病(OR=5.40, 95%CI 1.45~20.10, P=0.012)、输血史(OR=4.05, 95%CI 1.50~10.95, P=0.006)是TR-ROP发生的独立影响因素。基于此并结合出生体重建立列线图预测模型, 训练集、验证集C-index分别为0.940、0.885, ROC曲线下面积分别为0.945(95%CI 0.930~0.961)、0.931(95%CI 0.876~0.986), 灵敏度、特异性分别为86.2%、94.0%和83.2%、93.3%。结论出生胎龄、宫内窘迫、支气管肺发育不良、缺氧缺血性脑病、输血史是TR-ROP发生的独立影响因素。依据独立影响因素构建的TR-ROP列线图预测模型具有较高的灵敏度和特异性。Objective eTo analyze independent factors for treatment-requiring retinopathy of prematurity(TR-ROP)and establish a predictive nomogram model for TR-ROP.Method A retrospective cohort study.A total of 6998 preterm infants who were born at Guangdong Women's and Children's Hospital between January 1,2012 and March 31,2022 and were screened for retinopathy of prematurity(ROP)were included in the study.TR-ROP was defined as type 1 ROP and aggressive ROP;22 independent factors including general information,maternal perinatal conditions,interventions and neonatal diseases related to ROP were collected.The infants were divided at the level at an 8:2 ratio according to clinical experience,with 5598 in the training cohort and 1400 in the validation cohort.t test was used for comparison of quantitative data and x test was used for comparison of counting data between groups.Multivariate logistic regression analysis was carried out for the indicators with differences in the univariate analysis.The visualized regression analysis results of R software were used to obtain the histogram.The accuracy of the nomogram was verified by C-index and receiver operating characteristic curve(ROC curve).Results Among the 6998 children tested,4069 were males and 2920 were females.Gestational age was(33.69±3.19)weeks;birth weight was(2090±660)g.There were 376 cases of TR-ROP(5.4%,376/6998).The results of multivariate logistic regression analysis showed that gestational age[odds ratio(OR)=0.63,95%confidence interval(C)0.47-0.85,P=0.002],intrauterine distress(OR=0.30,95%CI 0.10-0.99,P=0.048),bronchopulmonary dysplasia(OR=0.23,95%CI 0.09-0.60,P=0.003),hypoxic-ischemic encephalopathy(OR=5.40,95%CI 1.45-20.10,P=0.012),blood transfusion history(OR=4.05,95%CI 1.50-10.95,P=0.006)were the independent influencing factors of TR-ROP.Based on this and combined with birth weight,a nomogram prediction model was established.The C-index of the training set and validation set were 0.940 and 0.885,respectively,and the area under ROC curve were 0.945(95%CI 0.9

关 键 词：早产儿 早产儿视网膜病变 预测因素 预测模型 列线图 

分 类 号：R774.1[医药卫生—眼科]