摘要
目的 探讨妊娠期糖尿病患者不良分娩结局的影响因素,并构建不良分娩结局的列线图预测模型.方法 回顾性选取2018年5月至2023年12月在张家港市第一人民医院产科住院且入院诊断为妊娠期糖尿病的患者484例为研究对象,依据其分娩结局将不良分娩结局的患者纳入不良分娩结局组(n=126),正常分娩结局的患者纳入正常分娩妊娠组(n=358).回顾性分析两组患者的临床资料[年龄、确诊孕周、妊娠前超重肥胖、孕次、产次、不良孕产史、糖尿病家族史和脐带异常、口服葡萄糖耐量试验(OGTT)血糖异常指标、胰岛素、糖化血红蛋白(GHb)、甘油三酯、总胆固醇、高密度脂蛋白胆固醇(HDL-C)、低密度脂蛋白胆固醇(LDL-C)、血清白蛋白、血清总蛋白、贫血、低蛋白血症、稳态模型胰岛素抵抗指数(HOMA-IR)、组织蛋白酶S(CatS)、单核细胞趋化蛋白-1(MCP-1)、胰高血糖素样肽1(GLP-1)水平];采用多因素Logistic回归模型分析不良分娩结局的影响因素;采用R语言建立风险列线图预测模型,采用受试者操作特征(ROC)曲线评价预测模型的内部效能,校正曲线评价模型的预测概率与实测概率的一致性.采用临床决策曲线评估列线图预测模型的临床实用性.结果 不良分娩结局组和正常分娩妊娠组的年龄、确诊孕周、是否妊娠前超重肥胖、是否有不良孕产史、是否有糖尿病家族史、OGTT血糖异常指标、GHb≥5.5％患者占比、低蛋白血症患者占比及CatS、MCP-1、GLP-1水平比较,差异均有统计学意义(P＜0.05).年龄、妊娠前超重肥胖、不良孕产史、糖尿病家族史、OGTT血糖异常指标3项异常、GHb≥5.5％、低蛋白血症、CatS、MCP-1和GLP-1是妊娠期糖尿病患者不良分娩结局的影响因素(P＜0.05).基于上述因素构建的预测妊娠期糖尿病患者不良分娩结局的列线图预测模型ROC曲线下面积为0.850(95％CI:0.793～0.907),具有较好的区分度;预测妊娠期糖尿病患者不良分娩结局的校准曲线结果显示预测概率与实测概率基本一致;且Hosmer-Lemeshow拟合优度检验x 2=3.366,P=0.909,一致性较好;决策曲线结果显示,阈值概率在0.4～1.0之间,该模型具有良好的分辨力.结论 年龄、妊娠前超重肥胖、不良孕产史、糖尿病家族史、OGTT血糖异常指标3项异常、GHb≥5.5％、低蛋白血症、CatS、MCP-1和GLP-1是妊娠期糖尿病患者不良分娩结局的影响因素,根据影响因素构建的不良分娩结局的列线图预测模型预测价值较为重要.
Abstract
Objective To explore the influencing factors of adverse delivery outcomes in diabetes mellitus and constructed the nemato-graphic prediction model of gestational diabetes mellitus.Methods A total of 484 patients with gestational diabetes admitted to the Department of Obstetrics,Zhangjiagang First Peoples Hospital from May 2018 to December 2023 were retrospectively selected as the study objects.According to their delivery outcomes,patients with adverse delivery outcomes were included in the adverse delivery outcome group(n=126),and patients with normal delivery outcomes were included in the normal delivery pregnancy outcome group(n=358).The clinical data[age,gestational age at diagnosis,pre-pregnancy overweight and obesity,gravidity,parity,history of adverse pregnancy,family history of diabetes and umbilical cord abnormalities,oral glucose tolerance test(OGTT)blood glucose abnormalities,insulin,glycosylated hemoglobin(GHb),triglyceride,total cho-lesterol,high-density lipoprotein cholesterol(HDL-C),low-density lipoprotein cholesterol(LDL-C),serum albumin,serum total protein,anemia,hypoproteinemia,Homeostasis model assessment of insulin resistance(HOMA-1R),cathepsin S(CatS),monocyte chemotactic protein-1(MCP-1),glucagon-like peptide-1(GLP-1)levels]of the two groups were retrospectively analyzed.The factors affecting adverse de-livery outcomes in patients with gestational diabetes were analyzed by multivariate Logistic regression model.The risk nomogram prediction model for adverse delivery outcomes in patients with gestational diabetes was established by using R language,and the internal efficacy of the prediction model was evaluated by receiver operating characteristic(ROC)curve.The consistency between the prediction probability and the measured prob-ability of the model is evaluated by the correction curve.Clinical decision curve was used to evaluate the clinical practicability of the nomogram model.Results There were statistically significant differences in age,gestational age at diagnosis,overweight and obesity before pregnancy,his-tory of adverse pregnancy and delivery,family history of diabetes,abnormal glucose index of OGTT,proportion of patients with GHb≥5.5％,pro-portion of patients with hypoproteinemia and levels of CatS,MCP-1 and GLP-1 between the adverse delivery outcome group and the normal de-livery pregnancy group(P＜0.05).Age,overweight and obesity before pregnancy,adverse pregnancy history,family history of diabetes,three abnormal OGTT blood glucose indicators,GHb≥5.5％,hypoproteinemia,CatS,MCP-1 and GLP-1 were the influential factors for adverse de-livery outcomes in patients with gestational diabetes(P＜0.05).Based on the above factors,the area under ROC curve of the nomogram model for predicting adverse delivery outcomes of patients with gestational diabetes was 0.850(95％CI:0.793-0.907),showing good differentiation.The results of calibration curve for patients with gestational diabetes mellitus show that the predicted probability is basically consistent with the meas-ured probability.Hosmer-Lemeshow goodness of fit testx2=3.366,P=0.909,good consistency;the decision curve results show that the thresh-old probability is between 0.4-1.0,indicating that the model has good resolution.Conclusion Age,pre-pregnancy overweight and obesity,ad-verse pregnancy history,family history of diabetes,abnormal OGTT blood glucose indicators,GHb≥ 5.5％,hypoproteinemia,CatS,MCP-1 and GLP-1 were the influencing factors of adverse delivery outcomes in patients with gestational diabetes.The nomogram prediction model of adverse de-livery outcomes in diabetes mellitus constructed in this study has a very important predictive value for the delivery outcomes in this population.
NETL
NSTL
万方数据