Discussion
We modified the LUS scanning method and developed a simplified
assessment system (based on ”high-risk” imaging patterns) to predict
respiratory support needs for the first time. To the best of our
knowledge, this system is practical and useful in obstetrics and
gynecology hospitals that need to identify infants with potential lung
diseases in the several hours immediately after birth. This approach can
help physicians identify these patients before their respiratory
symptoms deteriorate, and chest X-ray can be applied so that physicians
can implement NICU care earlier.
Findings and interpretation: This study has two clinically
relevant findings. (1) Four ”low-risk” patterns, and two ”high-risk”
patterns can be found in late preterm or term infants immediately after
birth. While two ”low-risk” patterns, the ”moderate discrete B-line” and
”abundant discrete B-line” patterns, were reported to be pathological in
previous studies27-28, our study shows that they can
be seen as strong evidence of healthy infants immediately after birth.
This discrepancy may be because of different extent of delay in lung
fluid clearance. This delay results in a small number of alveoli that
are uninflated and full of fluid29-31 but not enough
to cause TTN. (2) Two ”high-risk” patterns have high predictive accuracy
for respiratory support needs. These two patterns are also regarded as
evidence of other diseases, such RDS15,
MAS8 and pneumonia32. This
concordance indicates that our findings of ”high-risk” patterns are
highly likely to be an early stage of RDS or MAS, especially when
infants have only mild respiratory symptoms.
Because a total of ten scanning regions were evaluated, the number of
regions with ”high-risk” patterns was inversely related to those with
”low-risk” patterns. Thus, the ROC curve of ”low-risk” patterns can
support the predictive accuracy of ”high-risk” patterns.
Strengths and Limitations of our study: To our knowledge, this
is the most straightforward semiquantitative method to predict lung
diseases in late preterm and term infants. The assessment requires only
a count of the number of scanning regions on the chest wall with
”high-risk” patterns, and these patterns are easy to discern. More
importantly, finding more than two regions with ”high-risk” patterns
provides 87.10% sensitivity and 88.02% specificity. Coupled with LUS’s
radiationless and convenience, this method can be used as an effective
lung disease screening tool between the delivery room and NICU.
Nevertheless, there are some limitations to our study. Most
significantly, there was an insufficient number of patients who received
CPAP (23/74), MV (18/74), and PS (29/74). This insufficiency made it
difficult to draw a convincing conclusion to predict these individual
modes of advanced respiratory support. But considering our goal is to
discern potential lung diseases patients, and nearly all of their
treatment starts with a hood oxygen support need, it’s not necessary to
predict every advanced respiratory support need at such an early stage
of life. Besides, this limitation, if needed, can be improved in later
research containing more patients with severe respiratory diseases. The
second limitation is the possible inconsistency of image interpretation.
Our study used only one LUS interpreter due to the limited budget, so we
did not test consistency between interpreters. This may lead to a
variance in predictive accuracy. This drawback may be corrected in later
research by us or others. The last limitation is the concern about the
overuse of LUS. Our participants were all late preterm and term infants,
which means that they had a lower incidence of lung diseases than
smaller preterm infants. Of all infants who undergo LUS, only a small
part may develop lung diseases and need respiratory support. However,
because LUS is radiationless, easy to perform, and economical, we think
it is reasonable to perform LUS on every late preterm and term infant
with respiratory symptoms. Also, as smaller preterm infants will receive
more attention from physicians from birth, their respiratory issues are
less likely to be ignored than those in late preterm and term infants.
Comparison with other studies: Many studies have verified the
diagnostic value of LUS for neonatal lung diseases2,
28, 33-34. The difference between prior studies and ours is that we
focused on predictive value. As we found that these abnormal LUS
patterns identified immediately after birth do not have significant
specificity for certain diseases, we can use LUS to predict respiratory
needs for all kinds of severe lung diseases. Recently, some studies have
also paid attention to the predictive value of LUS in
neonatology14-16, 35. They evaluated the predictive
value of LUS for PS need in preterm infants, whereas we studied the need
for all kinds of respiratory support, including CPAP and MV, which are
closely related to severe lung disease. Therefore, we believe our study
is a good complement to current studies. Moreover, it is very useful
since it provides evidence to support early interventions in high-risk
infants.
Conclusion: Our assessment method allows a straightforward
semiquantitative use of LUS to discern infants with potential lung
diseases right after their birth. The LUS “high-risk” patterns show
good accuracy to predict respiratory support needs in late preterm and
term infants who manifest mild respiratory difficulty.
List of abbreviations
LUS: Lung ultrasound
STARD: Standards for the Reporting of Diagnostic Accuracy Studies
AUC: Area under the curve
ROC: Receiver operating characteristic curve
NICU: Neonate intensive care unit
RDS: Respiratory distress syndrome
TTN: Transient tachypnea of the neonate
MAS: Meconium aspiration syndrome
PTX: Pneumothorax
CPAP: Continuous positive airway pressure
SN: Serial number
MV: Mechanical ventilation
PS: Pulmonary surfactant
PEEP: Positive end-expiratory pressure
TcSO2: Transcutaneous oxygen saturation
GA: Gestational age
BW: Birth weight;
SGA: Small for gestational age;
SD: Standard deviation;
CI: Confidence interval;
DB: Discrete B-line
RR: Respiratory rate
HR: Heart rate;