1、临床超声医学杂志2023年3月第25卷第3期J Clin Ultrasound in Med,March 2023,Vol.25,No.3 临床研究 基于产前超声指标和妊娠期高血压的列线图模型预测低体质量儿不良结局的价值姚雨柳阳冯梅刘建新摘要目的探讨基于产前超声指标和妊娠期高血压(PIH)构建的列线图模型在预测低体质量儿不良结局中的价值。方法选取在我院行产前超声检查且分娩的PIH孕妇102例(PIH组)和正常孕妇121例(对照组),比较两组临床及产前超声检查资料的差异;应用单因素及多因素Logistic回归分析筛选低体质量儿不良结局的影响因素,并建立预测低体质量儿不良结局的列线图模型。采用Bo
2、otstrap重抽样法进行内部验证,绘制受试者工作特征(ROC)曲线分析模型的诊断效能;Hosmer-Lemeshow 检验并绘制校准曲线评估模型的拟合优度及校准度;决策曲线分析模型的临床获益。结果PIH组检出低体质量儿33例,对照组检出低体质量儿6例,两组低体质量儿占比比较差异有统计学意义(P0.01)。单因素及多因素Logistic回归分析显示,PIH、晚孕期胎儿体质量、妊娠32周后脐动脉收缩末期峰值流速与舒张末期峰值流速比值均是低体质量儿不良结局的独立影响因素(OR=6.162、0.997、16.712,均P0.05)。ROC曲线分析显示,列线图模型预测低体质量儿不良结局的曲线下面积为0
3、.852(95%可信区间:0.7890.914);Hosmer-Lemeshow检验显示,列线图模型预测概率与实际概率比较差异无统计学意义(P=0.788);校准曲线图显示,标准曲线与校准曲线贴合良好(内部验证结果显示C指数为0.841);临床决策曲线图显示该模型的潜在临床获益可观。结论 基于产前超声指标和PIH构建的列线图模型在预测低体质量儿不良结局方面有较高的临床应用价值,可指导不良妊娠结局的早期干预。关键词超声检查,产前;妊娠期高血压;低体质量儿;列线图中图法分类号R445.1文献标识码 AValue of a nomogram model based on prenatal ultra
4、sound indexes andpregnancy-induced-hypertension in predicting adverse outcomes inlow birth weight infantsYAO Yu,LIU Yang,FENG Mei,LIU JianxinDepartment of Ultrasound Medicine,the Central Hospital of Wuhan,Tongji Medical College,Huazhong University ofScience and Technology,Wuhan 430014,ChinaABSTRACTO
5、bjectiveTo explore the value of a nomogram model based on prenatal ultrasound indexes andpregnancy-induced-hypertension(PIH)in predicting adverse outcomes in low birth weight infants.MethodsA total of 102pregnant women(PIH group)and 121 cases of normal pregnant women(control group)who underwent pren
6、atal ultrasonographyand delivery in our hospital were selected.The clinical and prenatal ultrasonic data of the two groups were retrospectivelyanalyzed,and the influencing factors of adverse outcomes of low birth weight infants were screened by univariate and multivariateLogistic regression,and a no
7、mogram model for predicting adverse outcomes of low birth weight infants was established.Bootstrapresampling was used for internal validation,the receiver operating characteristic(ROC)curve was drawn to analyze thediagnostic efficiency of the model.Hosmer-Lemeshow test and calibration curves were dr
8、awn to assess the goodness of fit andcalibration of the model,and decision curves were used to analyze the clinical benefit of the model.ResultsA total of 33 casesof low birth weight infants were detected in PIH group and 6 cases in control group,the proportion of low birth weight infants wasstatist
9、ically significant(P0.01).Univariate and multivariate Logistic regression analysis showed that PIH,fetal weight in latepregnancy and S/D of the umbilical artery after 32 weeks of gestation were independent influencing factors for adverse outcomes oflow birth weight infants(OR=6.162,0.997,16.712,all
10、P0.05).ROC curve analysis showed that the area under the curve of thenomogram model for predicting adverse outcomes of low birth weight infants was 0.852(95%confidence interval:0.7890.914).作者单位:430014武汉市,华中科技大学同济医学院附属武汉中心医院超声医学科通讯作者:刘建新,Email: 172DOI:10.16245/ki.issn1008-6978.2023.03.004临床超声医学杂志2023
11、年3月第25卷第3期J Clin Ultrasound in Med,March 2023,Vol.25,No.3低体质量儿(出生体质量2500 g的新生儿)结局是临床长期关注的问题,全球每年约有250万新生儿死亡,其中80%的新生儿出生体质量过低,存活的低体质量儿成长过程中也可能面临生长发育迟缓、慢性病患病风险增加等问题1。妊娠期高血压(pregnancy-induced-hypertension,PIH)是低体质量儿不良结局的重要影响因素2-3,同时也是孕产妇死亡的重要原因之一,临床上对PIH孕妇多采取及时阴道试产或剖宫产的干预手段,但该方法会进一步增加低体质量儿不良结局的风险。目前临床主
12、要依赖产科医师的主观判断来平衡孕妇的生命安全和新生儿低体质量的发生风险,缺乏一个有效的综合判断工具。基于此,本研究尝试通过临床资料和产前超声指标构建预测低体质量儿不良结局的列线图模型,以指导产科医师在合适的妊娠时间进行综合判定,为临床诊断提供一个有效的决策工具。资料与方法一、研究对象选取2019年1月至2021年12月在我院行产前超声检查且分娩的PIH孕妇102例(PIH组),年龄2040岁,平均(30.24.6)岁,分娩孕周 30.041.0 周,平均孕(38.51.8)周;均符合 PIH 诊断标准4。排除标准:多胎妊娠;合并严重肝、肾、心血管疾病;合并胎膜早破、胎盘异常、胎儿呼吸窘迫综合征
13、;仅在我院行产前超声检查者;有妊娠期毒物、放射线及禁忌药物接触史;合并产前感染。另选同期正常孕妇121例为对照组,年龄2040岁,平均(29.44.2)岁,分娩孕周32.141.1周,平均孕(38.91.2)周。本研究经我院医学伦理委员会批准,所有孕妇及家属均知情同意。二、仪器与方法使用 GE Voluson E 8彩色多普勒超声诊断仪,探头频率2.55.0 MHz。产前超声检查依据2021年国际妇产超声学会实践指南(更新版)5。孕妇取仰卧位,先测量胎儿双顶径、头围、腹围及股骨长径,仪器自动计算胎儿体质量;然后在羊水中选取一游离段脐带,超声束平行于血管,在脐带静止状态下获取5个连续脐动脉频谱,
14、频谱形态清晰一致,计算收缩末期峰值流速与舒张末期峰值流速比值(S/D)、阻力指数(RI)。获得胎儿大脑轴向切面(包括丘脑及蝶骨翼)并放大,应用CDFI显示Willis环和大脑中动脉(MCA)近段,于MCA近段1/3处取样,测量时应避免对胎头进行不必要加压,超声束尽量与血流方向平行,记录5个连续波形频谱,获得峰值流速和RI。以上操作由同一经过系统培训且经验丰富的产前超声诊断医师完成。三、统计学处理应用 SPSS 23.0及 R 3.5.1 统计软件,符合正态分布的计量资料以xs表示,两组比较采用独立样本t检验;非正态分布的计量资料以中位数(四分位距)表示,两组比较采用秩和检验;计数资料以例或率表
15、示,两组比较采用2检验。先应用单因素Logistic回归分析筛选低体质量儿不良结局的可能预测因子,将 P0.01的变量纳入多因素Logistic回归分析,筛选低体质量儿不良结局的独立影响因素,构建预测低体质量儿不良结局的列线图模型。采用 Bootstrap重抽样法进行内部验证,绘制受试者工作特征(ROC)曲线分析列线图模型的诊断效能;Hosmer-Lemeshow 检验并绘制校准曲线评估模型的拟合优度及校准度;决策曲线分析模型的临床获益。P0.05为差异有统计学意义。结果一、两组一般资料及超声参数比较PIH组检出低体质量儿33例,对照组检出低体质量儿6例,两组低体质量儿占比比较差异有统计学意义
16、(P0.01)。PIH组孕妇年龄35岁、低蛋白血症、低甲状腺素血症占比及妊娠32周后胎儿脐动脉S/D、RI均高于对照组,差异均有统计学意义(均P0.01);两组其余一般资料比较差异均无统计学意义。见表1,2和图1,2。二、单因素Logistic回归分析结果Hosmer-Lemeshow test showed that there was no significant difference between the predicted probability of the nomogrammodel and the actual probability(P=0.788).The calibration curve showed that the standard curve fits well with the calibrationcurve(internal verification results show that the C index was 0.841).The clinical decision curve showed that the nomogram modelha