摘要
目的探索妊娠期高血压疾病(HDP)不良妊娠结局的危险因素,并构建列线图风险预测模型。方法回顾性分析2016年1月至2023年6月在汕头大学医学院第二附属医院产科分娩的720例HDP患者的临床资料。将数据集按照7∶3的比例随机分为建模的训练集(504例)和验证的测试集(216例)。根据是否发生不良妊娠结局,将训练集和测试集的HDP患者分为不良结局组和对照组,对训练集病例资料进行单因素和多因素Logistic回归分析,筛选出HDP患者不良妊娠结局的独立影响因素,采用R软件制作列线图,用测试集中样本进行验证。模型区分度采用受试者工作特征(ROC)曲线下面积(AUC)评估,校准度采用Hosmer-Lemeshow检验和校准图评估。结果720例HDP患者不良结局者87例(12.08%)。训练集504例患者中不良结局者55例(10.91%)。单因素及多因素Logistic回归分析结果显示高血压家族史[OR=2.620,95%CI(1.185,5.791),P=0.017]、缺铁性贫血[OR=2.193,95%CI(1.192,4.034),P=0.012]、重度子痫前期[OR=3.300,95%CI(1.526,7.113),P=0.002]、血清同型半胱氨酸[OR=1.477,95%CI(1.361,1.604),P<0.001]和血清白蛋白水平[OR=2.758,95%CI(1.158,6.565),P=0.022]是HDP患者不良妊娠结局的独立影响因素。模型的AUC为0.894[95%CI(0.849,0.939)],灵敏度和特异度分别为93.42%和72.44%。测试集中验证的AUC为0.912[95%CI(0.831,0.945)]。模型Hosmer-Lemeshow显示χ^(2)=9.513、P=0.326,校准图中偏差校正曲线与理想曲线重合度较高,模型校准度较好。结论本研究构建的HDP患者不良妊娠结局的列线图风险预测模型区分度与校准度较好,可为临床上评估不良妊娠结局风险、制定个性化预防及治疗方案提供参考。
Objective:To explore the risk factors associated with adverse pregnancy outcomes in hypertensive disorders of pregnancy(HDP)and to develop a nomogram-based risk prediction model.Methods:A retrospective analysis was conducted on 720 HDP patients who delivered at the Second Affiliated Hospital of Shantou University Medical College from January 2016 to June 2023.The dataset was randomly divided into a training set(for model development,504 cases)and a test set(for validation,216 cases)according to a ratio of 7∶3.The HDP patients in both the training set and the test set were further categorized into an adverse outcome group and a control group based on the occurrence of adverse pregnancy outcomes.The univariate and multivariate logistic regression analyses were performed to identify independent predictors of adverse outcomes in the training set.A nomogram was constructed using R software and validated with the test set samples.The model’s discrimination was assessed by the area under the receiver operating characteristic(ROC)curve(AUC),and the calibration was evaluated using the Hosmer-Lemeshow test and calibration plots.Results:Eighty-seven(12.08%)out of 720 HDP patients and 55(10.91%)out of 504 samples in the training set experienced adverse outcomes.The results of univariate and multivariate logistic regression analyses revealed that a family history of hypertension[OR=2.620,95%CI(1.185,5.791),P=0.017],iron deficiency anemia[OR=2.193,95%CI(1.192,4.034),P=0.012],severe preeclampsia[OR=3.300,95%CI(1.526,7.113),P=0.002],serum homocysteine levels[OR=1.477,95%CI(1.361,1.604),P<0.001],and serum albumin levels[OR=2.758,95%CI(1.158,6.565),P=0.022]were independent influencing factors for adverse pregnancy outcomes in HDP patients.The model’s AUC was 0.894[95%CI(0.849,0.939)]with a sensitivity of 93.42%and a specificity of 72.44%.The AUC for validation in the test set was 0.912[95%CI(0.831,0.945)].The Hosmer-Lemeshow test showed aχ^(2)of 9.513 with a P-value of 0.326,indicating good calibration as depicted in the calibration plots closely aligning with the ideal curve.Conclusions:The nomogram-based risk prediction model for adverse pregnancy outcomes in HDP patients demonstrated good discrimination and calibration.This model can serve as a valuable tool for clinical assessment of risk and for the development of personalized prevention and treatment strategies.
作者
李丹妍
刘琴
谢莉莉
LI Dan-yan;LIU Qin;XIE Li-li(Department of Obstetrics,The Second Affiliated Hospital of Shantou University Medical College,Shantou 515000)
出处
《生殖医学杂志》
2025年第8期1056-1062,共7页
Journal of Reproductive Medicine
