摘要
目的:分析子痫前期孕妇并发产后焦虑的危险因素,初步构建子痫前期孕妇并发产后焦虑的临床风险预测模型。方法:分两次前瞻性收集子痫前期孕妇纳入队列,其中2017年1月至2019年1月收集264例作为训练样本构建预测模型,2019年3月至2020年1月收集100例作为验证样本对预测模型进行临床验证。研究对象被纳入队列后,按标准操作流程收集其基本资料,实验室检查获得各项生理生化指标等。以产后6周是否存在产后焦虑为研究终点。先用2017年1月至2019年1月队列建立预测模型,采用逐步回归法进行多因素logistic回归分析筛选子痫前期孕妇并发产后焦虑的独立危险因素,并构建预测模型;然后采用Hosmer-Lemeshow检验模型的校准度,采用ROC曲线下面积(AUC)评估模型区分判别能力。再用2019年3月至2020年1月队列进行前瞻性验证该预测模型,以评估该模型的临床实用性。结果:264例训练样本因失访、不良结局原因剔除23例,共241例完成研究,其中有63例出现产后焦虑,占比26.14%。对241例子痫前期孕妇基线资料分析结果显示,夫妻感情、家庭成员有无性别歧视、血细胞比容(Hct)、雌二醇(E 2)、白细胞介素6(IL-6)可能是子痫前期孕妇并发产后焦虑的影响因素(P<0.05~P<0.01)。logistic回归分析结果显示:夫妻感情不好、家庭成员存在性别歧视、Hct≥53.70%、E 2≤45.80 pg/mL、IL-6≥58.66 pg/mL是子痫前期孕妇并发产后焦虑的危险因素(P<0.05~P<0.01)。构建子痫前期孕妇并发产后焦虑的预测模型:Logit(P)=0.725×夫妻感情+0.752×家庭成员存在性别歧视+1.277×Hct+1.657×E 2+0.787×(IL-6)-3.670。该预测模型AUC为0.829(95%CI:0.768~0.890),通过最大约登指数(0.521)得出该模型的阈值为0.411,对应的灵敏度为0.821,特异度为0.800;采用拟合优度检验评价该预测模型的校准度,显示预测模型拟合度好,准确性高(χ2=5.26,P>0.05)。临床验证该模型预测子痫前期孕妇并发产后焦虑的灵敏度为84.00%,特异度为78.79%,准确率为80.22%。结论:夫妻感情不好、家庭成员存在性别歧视、Hct≥53.70%、E 2≤45.80 pg/mL、IL-6≥58.66 pg/mL是子痫前期孕妇并发产后焦虑的危险因素,据此建立的预测模型区分能力良好,校准度高,准确度高,可操作性强,具有较高的临床价值。
Objective:To analyze the risk factors of postpartum anxiety in preeclampsia pregnant women,and to initially construct a clinical risk prediction model for preeclampsia pregnant women complicated with postpartum anxiety.Methods:Two prospective collections of preeclampsia pregnant women were enrolled in the cohort.Among them,264 cases were collected from January 2017 to January 2019 as training samples to construct a prediction model,and 100 cases collected from March 2019 to January 2020 were used as validation samples for clinical validation of the prediction model.After the research subjects were included in the cohort,their basic data were collected according to standard operating procedures,and various physiological and biochemical indicators were obtained by laboratory examinations.The end point of the study was whether there was postpartum anxiety at 6 weeks postpartum.Firstly,the cohort from January 2017 to January 2019 was established with a prediction model,and a stepwise regression method was used to perform multivariate logistic regression analysis to screen independent risk factors for preeclampsia pregnant women complicated with postpartum anxiety,and to construct a prediction model.Then,Hosmer-Lemeshow was used to test the calibration of the model,and the area under the ROC curve was used to evaluate the discriminative ability of the model.Lastly,the cohort from March 2019 to January 2020 was used to prospectively validate the prediction model to evaluate the clinical applicability of the model. Results: Twenty-three cases of 264 training samples were excluded due to loss of follow-up and poor outcome,and 241 cases completed the study,of which 63 cases developed postpartum anxiety,accounting for 26.14% (63/241).Analysis of baseline data of 241 pregnant women complicated with preeclampsia showed that the relationship between couples,whether family members had sex discrimination,hematocrit (Hct),estradiol (E 2),and interleukin 6 (IL-6) might be influencing factors of preeclampsia pregnant women complicated with postpartum anxiety ( P <0.05 to P <0.01).Logistic regression analysis showed that poor relationship between couples,gender discrimination in family members,Hct ≥53.70%,E 2≤45.80 pg/mL,IL-6 ≥58.66 pg/mL were risk factors for preeclampsia pregnant women complicated with postpartum anxiety ( P <0.05 to P <0.01).A prediction model of postpartum anxiety in preeclampsia pregnant women was constructed:Logit (P)=0.725×couple feelings+0.752×sex discrimination in family members + 1.277×Hct+1.657×E2+0.787 × (IL-6)-3.670.The area under the ROC curve of this prediction model was 0.829 (95% CI :0.768-0.890),the threshold of the model was 0.411 through the maximum Youden index (0.521),the corresponding sensitivity was 0.821,and the specificity was 0.800.The goodness of fit test was used to evaluate the calibration of the prediction model,which showed that the prediction model had good fitting degree and high accuracy ( χ 2=5.26, P >0.05).It was clinically verified that the sensitivity of this model to predict postpartum anxiety in pregnant women with preeclampsia was 84.00%,the specificity was 78.79%,and the accuracy was 80.22%. Conclusions: Poor relationship between couples,gender discrimination in family members,Hct≥53.70%,E 2≤45.80 pg/mL,IL-6≥58.66 pg/mL are risk factors for preeclampsia pregnant women complicated with postpartum anxiety.The prediction model based on those has good differentiation ability,high calibration,high accuracy,strong operability and high clinical value.
作者
赵琼
胡秋文
蒋雪玲
陈贵娟
莫力
朱光美
ZHAO Qiong;HU Qiu-wen;JIANG Xue-ling;CHEN Gui-juan;MO Li;ZHU Guang-mei(Department of Obstetrics,The People′s Hospital of Guangxi Zhuang Autonomous Region,Nanning Guangxi 530021,China)
出处
《蚌埠医学院学报》
CAS
2023年第10期1431-1436,共6页
Journal of Bengbu Medical College
基金
广西壮族自治区卫生健康委员会自筹经费科研课题(Z20190855,Z20191088,Z20190785,Z20201051)。
关键词
子痫前期
产后焦虑
危险因素
预测模型
preeclampsia
postpartum anxiety
risk factors
prediction model
