Objective: To explore the risk factors of pulmonary hypertension (PH) in premature infants with bronchopulmonary dysplasia (BPD), and to establish a prediction model for early PH. Methods: This was a retrospective cohort study. Data of 777 BPD preterm infants with the gestational age of <32 weeks were collected from 7 collaborative units of the Su Xinyun Neonatal Perinatal Collaboration Network platform in Jiangsu Province from January 2019 to December 2022. The subjects were randomly divided into a training cohort and a validation cohort at a ratio of 8∶2 by computer, and non-parametric test or χ2 test was used to examine the differences between the two retrospective cohorts. Univariate Logistic regression and multivariate logistic regression analyses were used in the training cohort to screen the risk factors affecting the PH associated with BPD. A nomogram model was constructed based on the severity of BPD and its risk factors,which was internally validated by the Bootstrap method. Finally, the differential, calibration and clinical applicability of the prediction model were evaluated using the training and verification queues. Results: A total of 130 among the 777 preterm infants with BPD had PH, with an incidence of 16.7%, and the gestational age was 28.7 (27.7, 30.0) weeks, including 454 males (58.4%) and 323 females (41.6%). There were 622 preterm infants in the training cohort, including 105 preterm infants in the PH group. A total of 155 patients were enrolled in the verification cohort, including 25 patients in the PH group. Multivariate Logistic regression analysis revealed that low 5 min Apgar score (OR=0.87, 95%CI 0.76-0.99), cesarean section (OR=1.97, 95%CI 1.13-3.43), small for gestational age (OR=9.30, 95%CI 4.30-20.13), hemodynamically significant patent ductus arteriosus (hsPDA) (OR=4.49, 95%CI 2.58-7.80), late-onset sepsis (LOS) (OR=3.52, 95%CI 1.94-6.38), and ventilator-associated pneumonia (VAP) (OR=8.67, 95%CI 3.98-18.91) were all independent risk factors for PH (all P<0.05). The independent risk factors and the severity of BPD were combined to construct a nomogram map model. The area under the receiver operating characteristic (ROC) curve of the nomogram model in the training cohort and the validation cohort were 0.83 (95%CI 0.79-0.88) and 0.87 (95%CI 0.79-0.95), respectively, and the calibration curve was close to the ideal diagonal. Conclusions: Risk of PH with BPD increases in preterm infants with low 5 minute Apgar score, cesarean section, small for gestational age, hamodynamically significant patent ductus arteriosus, late-onset sepsis, and ventilator-associated pneumonia. This nomogram model serves as a useful tool for predicting the risk of PH with BPD in premature infants, which may facilitate individualized early intervention.
目的： 探讨支气管肺发育不良（BPD）早产儿伴肺动脉高压（PH）的危险因素，并构建早期预测PH发生的列线图预测模型。 方法： 采用回顾性队列研究设计，分析2019年1月至2022年12月入住苏新云新生儿围生期协作网7家合作单位新生儿重症监护病房出生胎龄<32周且确诊BPD的777例早产儿的临床资料，包括出生体重、出生胎龄等围生期资料及住院期间临床指标。计算机随机按8∶2比例分为训练队列和验证队列，采用非参数检验或χ2检验比较两队列间差异；根据是否存在PH，分为PH组和非PH组。在训练队列中采用单因素Logistic回归和多因素Logistic回归分析筛选BPD伴发PH的高危因素，基于BPD严重程度和高危因素构建列线图预测模型，并采用Bootstrap法进行内部验证。最后，应用训练队列和验证队列对预测模型进行模型区分度准度、临床适用度的评估。 结果： 777例BPD早产儿中130例伴发PH，发生率为16.7%。777例BPD早产儿出生胎龄为28.7（27.7，30.0）周，其中男454例（58.4%），女323例（41.6%）。训练队列共622例，其中PH组105例、非PH组517例；验证队列共155例，其中PH 25例。训练队列多因素Logistic回归分析结果显示，5 min Apgar评分低（OR=0.87，95%CI 0.76~0.99），剖宫产出生（OR=1.97，95%CI 1.13~3.43），小于胎龄（OR=9.30，95%CI 4.30~20.13），有血流动力学意义的动脉导管未闭（OR=4.49，95%CI 2.58~7.80），晚发型败血症（OR=3.52，95%CI 1.94~6.38）和呼吸机相关性肺炎（OR=8.67，95%CI 3.98~18.91）均为BPD伴发PH的独立危险因素（均P<0.05）。基于独立危险因素和BPD严重程度构建列线图模型，训练队列和验证队列中列线图模型的受试者工作特征曲线的曲线下面积分别为0.83（95%CI 0.79~0.88）、0.87（95%CI 0.79~0.95），列线图校准曲线接近于理想的对角线，临床决策曲线高于两条极端曲线。 结论： 5 min Apgar评分低、剖宫产出生、小于胎龄、合并有血流动力学意义的动脉导管未闭、晚发型败血症、呼吸机相关性肺炎的BPD早产儿发PH的风险增加。据此及BPD严重程度建立的列线图模型为预测BPD伴发PH的发生风险提供一定的参考价值，有助于临床医生早期进行个体化干预。.