| 足月新生儿死亡预测模型的研究 |
| 投稿时间：2024-10-16 修订日期：2024-12-02 点此下载全文 |
| 引用本文：张明明,张蒙,张心,周彬,侯利娜.足月新生儿死亡预测模型的研究[J].医学研究杂志,2025,54(6):138-142 |
| DOI：
10.11969/j.issn.1673-548X.2025.06.025 |
| 摘要点击次数: 15 |
| 全文下载次数: 9 |
|
| 基金项目:江苏省徐州市科技计划项目(KC21255) |
|
| 中文摘要:目的 开发足月新生儿死亡预测模型并进行内部验证。方法 选择2015年1月1日～2023年12月31日徐州市中心医院儿科死亡的足月新生儿86例,根据样本估量法,随机选择同时期住院的足月儿135例作为对照组,收集母亲孕产史和患儿出生史、患儿生后24h内血常规、凝血功能、二氧化碳分压(arterial carbon dioxide pressure,PaCO2)、血清总胆红素等临床资料。采用单因素分析、逐步回归法和多因素Logistic 回归分析筛选可能的预测因素,建立预测模型。应用受试者工作特征(receiver operating characteristic,ROC)曲线评价模型的区分度,采用Hosmer-Lemesshow检验评价模型的校准度,通过Bootstrap方法进行内部验证。结果 羊水污染(OR=3.818,95% CI:1.009~14.447,P=0.048)、凝血功能异常(OR=12.981,95% CI:3.732~45.152,P<0.001)、妇科炎性疾病(OR=7.203,95% CI:1.216~42.659,P=0.030)、机械通气(OR=54.451,95% CI:12.913~229.619,P<0.001)、PaCO2(OR=1.131,95% CI:1.055~1.212,P=0.001)、血清乳酸(OR=4.540,95% CI:2.561~8.046,P<0.001)是足月儿死亡的独立影响因素,可预示足月儿新生儿死亡的发生(敏感度为75.8%,特异性为87.0%,AUC=0.814)。Hosmer-Lemesshow检验显示该模型与临床实际足月新生儿死亡的发生率一致性较好(χ2=3.787,P=0.876)。经Bootstrap内部验证,该模型具有较好的区分度(AUC=0.849)。结论 通过母亲孕产史和足月儿生后24h内临床资料可建立死亡危险因素的预测模型,有助于提前做出临床决策。 |
| 中文关键词:足月儿 高危因素 病死率 预测模型 |
| |
| Research on the Prediction Model of Full-term Neonatal Mortality. |
|
|
| Abstract:Objective Develop a prediction model for full-term neonatal mortality and conduct internal validation.Methods 86 full-term newborns who died in the pediatric department of Xuzhou Central Hospital from January 1,2015 to December 31,2023 were selected. According to the sample estimation method, 135 full-term infants who were hospitalized during the same period were randomly selected as the control group. Clinical data such as maternal and child birth history, blood routine within 24hours after birth, coagulation function, arterial carbon dioxide pressure(PaCO2),and serum total bilirubin were collected.Using single factor analysis, stepwise regression, and multiple Logistic regression analysis to screen possible predictive factors and establish a predictive model. The receiver operating characteristic (ROC) curve was used to evaluate the discriminative power of the model, and Hosmer Lemeshow was used to test the calibration of the evaluation model. Bootstrap method was used for internal validation.Results Amniotic fluid contamination (OR=3.818,95% CI:1.009-14.447,P=0.048), coagulation dysfunction (OR=12.981,95% CI 3.732-45.152,P<0.001), gynecological inflammatory diseases (OR=7.203,95% CI:1.216-42.659, P=0.03), mechanical ventilation (OR=54.451,95% CI:12.913-229.619,P<0.001), PCO2 (OR=1.131,95% CI:1.055-1.212,P=0.001), and serum lactate (OR=4.540,95% CI:2.561-8.046,P<0.001) are independent influencing factors of full-term infant mortality, which can predict the occurrence of neonatal mortality in full-term infants (sensitivity 75.8%, specificity 87.0%, AUC=0.814).The Hosmer Lemeshow test showed that the model had good consistency with the actual occurrence probability of full-term neonatal mortality in clinical practice (χ2=3.787, P=0.876). After internal validation by Bootstrap, the model had good discrimination (AUC=0.849).Conclusion A predictive model for mortality risk factors can be established based on maternal pregnancy history and clinical data within 24hours after full-term birth, which helps to make clinical decisions in advance. |
| keywords:Term neonatal High risk factors Mortality Prediction model |
| 查看全文 查看/发表评论 下载PDF阅读器 |
|
|
|
