妊娠期糖尿病早期风险预测模型建立与评价
Development and evaluation of an early risk prediction model for gestational diabetes mellitus
李树华 1岳中航 1左群 1解春燕 2申月红 3霍晓桐 4赵建伟 5吴淑堃 1高章圈6
作者信息
- 1. 河北大学公共卫生学院 河北省公共卫生安全重点实验室,河北保定 071000
- 2. 河北省保定市徐水区妇幼保健院
- 3. 河北省定州市妇幼保健院
- 4. 河北省涞水县妇幼保健院
- 5. 河北省安国市妇幼保健院
- 6. 河北省生殖健康科学技术研究院河北省生殖医院国家卫生健康委员会计划生育与优生重点实验室 河北省生殖医学重点实验室
- 折叠
摘要
目的 分析妊娠期糖尿病(gestational diabetes mellitus,GDM)的危险因素,建立预测模型并进行初步验证,为防治GDM的发生提供指导.方法 随机抽取河北省5个市(县)妇幼计生服务中心,收集2020年6月—2022年6月符合纳入标准、信息完整的1 738名孕妇资料,随机分为模型构建组和验证组.采用单因素分析和多因素logistic回归分析筛选GDM的独立危险因素,并拟合最佳预测模型.采用Homer-Lemeshow拟合优度检验和受试者工作特征(receiver operating characteristic,ROC)曲线对该模型进行评估.结果 模型构建组共1 390名孕妇,年龄＞35岁(OR=1.82,95％CI:1.27～4.13)、孕前体质量指数超过24.0kg/m2(OR=2.26,95％CI:1.19～3.98)、孕早期空腹血糖升高(OR=2.47,95％CI:1.62～3.70)、一级亲属糖尿病史(OR=2.39,95％CI:1.09～4.42)、孕早期贫血(OR=1.22,95％CI:1.14～1.36)均为影响GDM发生的独立危险因素(P＜0.05).所建立的logistic模型拟合优度较好(P=0.796),ROC曲线下面积为0.82.以0.451为阈值时,该模型在验证组的灵敏度为71.93％,特异度为86.25％.结论 基于上述危险因素构建的预测模型拟合度好、预测价值高,能有效地筛选高危人群并提前纳入管理,减少GDM发病率.
Abstract
Objective To analyze the risk factors of gestational diabetes mellitus(GDM),with which to develop and preliminarily evaluate a risk prediction model,so as to provide guidance for the prevention and treatment of GDM.Methods Five maternal and child health service centers in Hebei Province were randomly selected,from which a total of 1 738 pregnant women with complete information were enrolled and randomly divided into model development group and model validation group.Univariate and multivariate logistic regression models were used to select the independent risk factors of GDM and to fit the prediction model with the best performance.Homer-Lemeshow goodness-of-fit test and receiver operating characteristic(ROC)curve were used to evaluate the performance of the model.Results In the model development group comprised of 1 390 pregnant women,age over 35 years old(OR=1.82,95％CI:1.27-4.13),pre-pregnancy body mass index(BMI)being over 24.0 kg/m2(OR=2.26,95％CI:1.19-3.98),elevated fasting blood glucose during early pregnancy(OR=2.47,95％CI:1.62-3.70),history of diabetes among first-degree relatives(OR=2.39,95％CI:1.09-4.42),and anemia during early pregnancy(OR=1.22,95％CI:1.14-1.36)were independent risk factors for GDM(P＜0.05).The goodness-of-fit test of the selected logistic model incorporating the above risk factors showed P=0.796,the area under the ROC curve was 0.82.With a cut-off value of 0.451,the sensitivity of the model in the model validation group was 71.93％,and the specificity was 86.25％.Conclusions The prediction model based on the five identified risk factors demonstrated good fitting and high predictive performance,which may facilitate identifying high-risk populations for early management to reduce the incidence of GDM.
关键词
妊娠期糖尿病/风险预测模型/模型评价/队列研究Key words
Gestational diabetes mellitus/Risk prediction model/Model evaluation/Cohort study引用本文复制引用
基金项目
河北省医学科学研究课题(20210147)
河北省重点研发计划(22377722D)
出版年
2024
NETL
NSTL
万方数据