Simplified lung ultrasound method to predict the respiratory support needs of neonates whose mothers had pregnancy complications: a diagnostic accuracy study

Objective:Some pregnancy complications are associated with respiratory problems in infants.Early identification and implementation of respiratory support are needed for these infants.This study aimed to(1)use lung ultrasound(LU)for predicting the respiratory support requirements of infants and(2)simplify the examination method to facilitate its clinical application.Methods:Neonates underwent LU examinations within 1 hour of birth.The infants were divided into a pregnancy complication(PC)group and a control group according to their mothers'clinical data.The primary outcome measure was the need for respiratory support.Predictive models were established based on the differences in LU patterns between the two groups.The predictive value of the models was compared with that of the LU score(LUS)system,which has been widely used in other studies.Simplified models were established by analyzing the predictive effects of LU patterns in different parts of the lungs.Predictive reliability was tested using the receiver operating characteristic curve analysis.Results:PC and control groups had 29 and 117 neonates,respectively;a total of 22 neonates required respiratory support.The PC and control groups showed differences in respiratory support requirements,A-lines,and coalesced B-lines.The percentages of A-lines and coalesced B-lines were associated with respiratory support requirements.The optimal cutoff values for the percentages of the A-lines and coalesced B-lines were 55.7%and 24.5%,respectively.Similar effects were obtained by calculating the percentage of coalesced B-lines in four areas(R3,R4,L3,and L4).These methods were judged as more visual and convenient than LUS assessment.conclusion:The percentages of A-line and coalesced B-lines on LU images can be used to predict the respiratory support needs of neonates whose mothers had pregnancy complications.Only four areas(R3,R4,L3,and L4)had similar predictive values.