母婴分离早产儿母亲泌乳Ⅱ期启动延迟风险预测评分模型的构建与验证

Establishment and validation of a risk-scoring model for predicting delayed onset of lactogenesis stage Ⅱ in preterm mothers during maternal separation

目的构建母婴分离早产儿母亲泌乳Ⅱ期启动延迟(delayed onset of lactogenesis stage Ⅱ,DOLⅡ)风险预测评分模型,并验证其预测效果。方法本研究为回顾性研究。(1)建模组:以2021年12月至2022年11月于无锡市妇幼保健院分娩的母婴分离早产儿母亲310例为建模组。按照是否发生DOLⅡ分为DOLⅡ组(144例)和非DOLⅡ组(166例)。根据多因素logistic回归分析结果,对各危险因素赋值,并建立风险预测评分模型。(2)验证组:以2022年12月至2023年3月于无锡市妇幼保健院分娩的130例母婴分离早产儿母亲为验证组。采用受试者工作特征(receiver operating characteristic,ROC)曲线下面积检验模型区分度,Hosmer-Lemeshow检验判断模型的拟合优度。采用χ^(2)检验(或Fisher精确概率法)或Wilcoxon秩和检验进行组间数据比较。结果该风险预测评分模型共纳入10个危险因素[产妇年龄≥35岁、妊娠期高血压疾病、妊娠合并贫血、妊娠期糖尿病、胎膜早破、开奶时间>6 h、母亲产后收住妇产科重症监护病房、剖宫产、爱丁堡产后抑郁量表得分≥9.5分、产后中性粒细胞与淋巴细胞比值≥4.369、产后疲劳量表-14得分≥7.5分、孕早期体重指数(body mass index,BMI)≥23.719 kg/m^(2)、产后BMI≥27.661 kg/m^(2)、孕期BMI增幅≥5.393 kg/m^(2)]。模型的ROC曲线下面积为0.838(95%CI:0.795~0.882,P<0.001),Hosmer-Lemeshow拟合优度检验χ^(2)=3.43,P=0.634,最大约登指数为0.526,特异度为0.825,灵敏度为0.701,最佳界值为4.5,四舍五入后以评分≥5分者为DOLⅡ高危组,<5分为低危组。以此模型预测,建模组310例中,高危组130例(41.9%),实际发生DOLⅡ101例(32.6%);低危组180例(58.1%),实际发生DOLⅡ43例(13.9%)。该风险预测评分模型的阳性预测值为77.7%,阴性预测值为76.1%,准确率为76.8%。验证组130例中,DOLⅡ高危组59例(45.4%),实际发生DOLⅡ39例(30.0%);低危组71例(54.6%),实际发生DOLⅡ19例(14.6%)。模型验证结果显示ROC曲线下面积为0.774(95%CI:0.693~0.855,P<0.001),Hosmer-Lemeshow检验χ^(2)=3.09,P=0.687,阳性预测值为66.1%,阴性预测值为73.2%,准确率为70.0%。结论本研究初步建立了母婴分离早产儿母亲DOLⅡ风险预测评分模型。该模型具有一定的预测效能,可为制定预见性泌乳支持措施提供参考。

Objective To construct and validate a risk-scoring model for predicting delayed onset of lactogenesis stage Ⅱ(DOLⅡ)in mothers separated from their premature infants.Methods This was a retrospective study.(1)Modeling group:This group enrolled 310 mothers who were separated from their premature infants after delivery at Wuxi Maternal and Child Health Hospital from December 2021 to November 2022.They were further divided into the DOLⅡ group(144 cases)and the non-DOLⅡ group(166 cases)according to whether they had DOLⅡ or not.Based on the results of multivariate logistic regression analysis,each risk factor was assigned a score,and a risk prediction scoring model was established.(2)Validation group:This group included 130 mothers of premature infants who experienced mother-infant separation after delivery at Wuxi Maternal and Child Health Hospital from December 2022 to March 2023.The area under the receiver operating characteristic(ROC)curve was used to evaluate the discrimination,and the Hosmer-Lemeshow test was used to assess the goodness of fit.The Chi-square test(or Fisher's exact probability test)or Wilcoxon rank sum test were used for inter-group data comparison.Results This risk prediction scoring model included 10 risk factors[maternal age≥35 years old,hypertensive disorders of pregnancy,anemia,gestational diabetes mellitus,preterm rupture of membrane,start breastfeeding>6 hours,postpartum admission of maternal intensive care unit,cesarean section,score of Edinburgh Postpartum Depression Scale>9.5,postpartum neutrophil-to-lymphocyte ratio≥4.369,Fatigue Scale-14≥7.5,body mass index in the first trimester≥23.719 kg/m^(2),postpartum BMI≥27.661 kg/m^(2),and increase of BMI during pregnancy≥5.393 kg/m^(2)],with an area under the ROC curve of 0.838(95%CI:0.795-0.882,P<0.001),a maximum Yoden index of 0.526,a specificity of 0.825,a sensitivity of 0.701,and an optimal threshold of 4.5.After rounding the score off to the nearest whole number,those with a score≥5 were defined as at high risk of DOLⅡ,while those with a score<5 were at low risk.Hosmer-Lemeshow test showed χ^(2)=3.43 and P=0.634.The positive predictive value,the negative predictive value,and the accuracy were 77.7%,76.1%,and 76.8%,respectively.In the modeling group,130 out of the 310 cases(41.9%)were predicted to be at high risk by the model with 101(32.6%)experiencing DOLⅡ,while 180 cases(58.1%)were predicted to be at low risk with 43(13.9%)experiencing DOLⅡ.Among the 130 cases in the validation group,59(45.4%)were predicted to be at high risk with 39(30.0%)experiencing DOLⅡ,while 71(54.6%)were predicted to be at low risk with 19(14.6%)experiencing DOLⅡ.The model validation results showed that the area under the ROC curve was 0.774(95%CI:0.693-0.855,P<0.001)and the Hosmer-Lemeshow test showed χ^(2)=3.09 and P=0.687,with the positive predictive value of 66.1%,the negative predictive value of 73.2%,and the accuracy of 70.0%.Conclusions This study preliminarily establishes a risk scoring model for predicting DOLⅡin mothers separated from their premature infants which is of certain predictive value and can provide a reference for developing predictive lactation support measures.