455 - A mode of perinatal factors to predict neonatal resuscitation at birth

Poster Number: 455
Publication Number: 455.130

Silu Wang, Shanghai First Maternity And Infant Hospital, Shanghai, Shanghai, China (People's Republic); Ming Zhou, Shanghai first maternity and infant hospital, shanghai, Shanghai, China (People's Republic); Jingyi ni, n/a, Shanghai, Shanghai, China (People's Republic); Zhen Li, Clinical research center, Shanghai, Shanghai, China (People's Republic); Jiang-Qin Liu, Tongji University School of Medicine, Shanghai, Shanghai, China (People's Republic)

Background: Neonatal asphyxia is one of the major causes of neonatal death. Neonatal resuscitation(NR)should be started within 1 minute after birth. However, having at least one well-trained medical staff member with NR skills on hand at the delivery site remains a big challenge. It is especially vital to predict NR in advance and help to assign the medical personnel reasonably.

Objective: To analyze the perinatal risk factors of NR, and establish a multivariable model predicting NR requirement in infants above 34 weeks gestational age.

Design/Methods: A total of 78,370 infants born in Shanghai First Maternity and Infant Hospital, China, from 2014 to 2018 were enrolled in this retrospective cohort study. The NR was defined according to the 1-minute Apgar score: NR (0-7 of 1-minute Apgar score) and no NR requirement (8-10 of 1-minute Apgar score). Random forest and multivariable logistic regression were used to analyze the association between NR requirement and the perinatal factors. All data were randomly split into training (70%) and validation (30%) sets, and the receiver operating characteristic curve was used to establish a multivariate model for predicting NR requirement.

Results: The participants included 6.1% preterm infants and 93.9% term infants. 0.8% received NR treatment including bag-mask or intubated ventilation, chest compression, and medications. Multivariable logistic regression was applied for the final modeling. The risk factors included pregnancy-induced hypertension (OR: 2.984, 95%CI: 2.172-4.120), maternal fever (OR: 4.930, 95%CI: 3.787-6.448), multiple pregnancy (OR: 3.712, 95%CI: 2.482-5.624), obesity (OR: 4.619, 95%CI: 3.155-6.822), intrahepatic cholestasis of pregnancy (OR: 2.969, 95%CI: 1.631-5.505), forceps delivery (OR: 4.719, 95%CI: 2.809-8.167), umbilical cord compression (OR: 2.897, 95%CI: 2.311-3.637), placenta diseases (OR: 2.999, 95%CI: 2.273-3.971), meconium-stained (OR: 4.506, 95%CI: 3.476-5.863), poly/oligohydramnios (OR: 3.796, 95%CI: 2.782-5.203), fetal distress(OR: 6.070, 95%CI: 4.369-8.534), and preterm birth (OR: 3.838, 95%CI: 2.764-5.355). The area under curve in the training set was 0.859, and in the validation set was 0.786. In predicting NR requirement at birth, the sensitivity and the specificity were 0.93 and 0.52, 0.42 and 0.96 corresponding to a cut-off score 0.2 and 0.8, respectively.Conclusion(s): Perinatal risk factors predicted the NR requirement at birth with moderate accuracy in this model. It might help to allocate the medical personnel in delivery room. Further prospective studies are needed to validate and to evaluate the performance of this model.
Table2 Logistic regression analysis of perinatal factors in neonatal resuscitationOR: Odds ratio; CI: confidence interval; PROM: Premature rupture of membranes; ICP: Intrahepatic cholestasis of pregnancy; TTTS: Twin-to-twin transfusion syndrome
Figure 1: Receiver operating characteristic curve for the perinatal factors of model to predict neonatal resuscitation