+ MODEL

Pediatrics and Neonatology xxx (xxxx) xxx

Available online at www.sciencedirect.com

ScienceDirect

journal homepage: http://www.pediatr-neonatol.com

Original Article

Lung ultrasound score as a predictor of

ventilator use in preterm infants with
dyspnea within 24 h after dhospitalization
Lihua Zhang, Jinnan Feng, Di Jin, Zekun Yu, Yangming Qu,
Meiyu Zheng, Hui Wu*

Department of Neonatology, The First Hospital of Jilin University, No. 71 Xinmin Street, Changchun
130021, China

Received May 30, 2022; received in revised form Aug 2, 2022; accepted Sep 21, 2022
Available online - - -

Key Words Background: Selecting the correct ventilation strategy is crucial for the survival of preterm in-
dyspnea; fants with dyspnea in NICU. Lung ultrasound score (LUSsc) is a potential predictor for respira-
lung ultrasound; tory support patterns in preterm infants.
lung ultrasound score Methods: We prospectively included 857 preterm infants. LUS was performed in the first 2 h af-
ter admission, and LUSsc was determined by two specialist sonographers. Participants were
divided into two categories according to gestational age (<32þ0 weeks and 32þ0e36þ6 weeks)
and randomly divided into a training set and a validation set. There were two main outcomes:
invasive and non-invasive respiratory support. In the training set, clinical factors were analyzed
to find the best cut-off value of LUSsc, and consistency was verified in the verification set. The
choice of invasive respiratory support was based on neonatal mechanical ventilation strategies.
Results: Preterm infants with invasive respiratory support had a higher LUSsc, greater use of
Pulmonary Surfactant（PS）, and lower Oxygenation Index（OI）、birth weight than those with
non-invasive support. In the <32 þ0 weeks group, the area under the curve (AUC) for the
receiver operating characteristic curve plotted with 2-h LUSsc was 0.749 (95% CI: 0.689
e0.809), the cut-off point of LUSsc was 8, and the sensitivity and specificity were 74.0% and
68.3%, respectively. In the 32þ0e36þ6 weeks group, the AUC was 0.863 (95% CI: 0.811
e0.911), with a cut-off point of 7. Sensitivity and specificity were 75.3% and 0.836%, respec-
tively. In the validation set, using the actual clinical respiratory support selection results for
verification, the validation results showed for the <32þ0 weeks group (Kappa value 0.660,
P < 0.05, McNemar test P > 0.05) for preterm 32 þ0 e36 þ6 weeks (Kappa value 0.779,
P < 0.05, McNemar test P > 0.05).
Conclusion: The LUSsc showed good reliability in predicting respiratory support mode for pre-
term infants with dyspnea.

Abbreviations: LUS, Lung ultrasound; LUSsc, Lung ultrasound score.

* Corresponding author.
E-mail address: [email protected] (H. Wu).

https://doi.org/10.1016/j.pedneo.2022.09.019
1875-9572/Copyright ª 2023, Taiwan Pediatric Association. Published by Elsevier Taiwan LLC. This is an open access article under the CC BY-
NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/).

Please cite this article as: L. Zhang, J. Feng, D. Jin et al., Lung ultrasound score as a predictor of ventilator use in preterm infants with
dyspnea within 24 h after dhospitalization, Pediatrics and Neonatology, https://doi.org/10.1016/j.pedneo.2022.09.019
 + MODEL
L. Zhang, J. Feng, D. Jin et al.

Registered at ClinicalTrials.gov (identifier: chiCTR1900023869).

Copyright ª 2023, Taiwan Pediatric Association. Published by Elsevier Taiwan LLC. This is an
open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/
by-nc-nd/4.0/).

1. Introduction inclusion criteria were as follows: 1) admission to the hos-

pital 4 h after birth and complete LUS examination within
Most premature infants have difficulty breathing after birth 2 h of admission; 2) symptoms of dyspnea after birth,
and require respiratory support and adjuvant treatment. including skin cyanosis, groaning, nasal flaring, and
Current treatments focus on lung-protective ventilation retraction. The exclusion criteria were as follows: 1) suf-
strategies, with the option of non-invasive respiratory focation and general anesthesia for cesarean section; 2)
support rather than invasive respiratory support.1 However, supplementation of PS before LUS examination; 3) chest
non-invasive respiratory support is insufficient for some wall deformities or abnormal respiratory mechanics caused
hospitalized preterm infants who require invasive me- by lung tumors, including severe arrhythmia caused by
chanical ventilation. For these patients, failure to treat in congenital heart disease, and organic diseases of the liver
time can delay recovery, so interventions that maximize and kidney. Included patients were divided into two cate-
survival while minimizing potential adverse effects, gories according to gestational age (<32þ0 weeks and
including Bronchopulmonary Dysplasia (BPD), are required. 32þ0e36þ6 weeks); preterm infants from each stratum
In infants who receive invasive ventilation inappropriately, were randomly divided into a training set and a validation
overtreatment may lead to lung injury and more sedation, set at a 4:1 ratio. This study was approved by the Ethics
as well as financial loss.2 Lung ultrasound (LUS) plays an Committee of the First Hospital of Jilin University, and all
active role in auxiliary treatment for lung disease, including guardians of patients signed an informed consent form. The
prognosis prediction, disease diagnosis, evaluation, and ethical batch number is 19K053-001.
monitoring.3e5 By dividing the lungs into several lung re-
gions and scoring the ultrasound images of each lung region 2.2. Research outcomes
according to a unified scoring standard, the total LUSsc can
be calculated.6 LUSsc is a comprehensive index that can The main outcomes of our study were non-invasive and
reflect the condition of lung ventilation and can more invasive respiratory support within 24 h of admission: 1) use
accurately assess the oxygenation status.7 LUSsc can be of continuous positive airway pressure (CPAP), non-invasive
used to evaluate disease severity, monitor the changes in positive airway pressure ventilation (NIPPV), and nasal
disease, and reflect the treatment effect sensitively.6,8e14 catheter oxygen inhalation were classified as non-invasive
Some correlation analyses between LUSsc and the clinical respiratory support; 2) use of synchronized intermittent
criticality of the child revealed that the more severe the pa- mandatory ventilation（SIMV）and high frequency oscilla-
tient’s condition, the higher was the LUSsc.6,11,15,16 A study tory ventilation (HFOV) were classified as invasive respira-
used modified LUSsc to predict ventilation requirements in tory support. Invasive respiratory support was selected
Neonatal Respiratory Distress Syndrome (RDS) and the mode of based on neonatal mechanical ventilation strategies17 and
respiratory support on day of life 3 (DOL 3).15 However, there specific ventilation mode combined with clinical experi-
have been no studies on the prediction of respiratory support ence. The investigator did not intervene.
patterns within 24 h of admission in preterm infants using
LUSsc. Selection of the correct respiratory support mode in
the early admission of premature infants with dyspnea can 2.3. Equipment
reduce the adverse consequences caused by delayed illness or
excessive treatment, so the early and reasonable selection of Mindray diagnostic equipment (Manufacturer: Mindray
respiratory support mode is crucial. We designed a prospec- Company, Shenzhen, China) was used for bedside ultrasonic
tive study in the Chinese population with the aim of predicting diagnosis; the probe model was the C1-5 Linear probe, and
respiratory support needs within 2 h of admission in preterm the frequency was 9.0 MHz.
infants. We planned to stratify premature infants by gesta-
tional age using 12-segment LUSscs and analyze other clinical 2.4. Calculation of lung ultrasound score
factors that affect the mode of respiratory support to further
to validate corresponding LUSscs in a validation set. Each lung was divided into three areas (front, middle, and
back) using the anterior and posterior axillary lines as the
2. Methods boundaries. Taking the midpoint of the two nipples as the
boundary, each lung was divided into upper and lower lung
fields. The left and right lungs were divided into the upper
2.1. Population front, lower front, upper axillary, mid-axillary, upper back,
and lower back regions, giving a total of 12 areas between
This was a prospective study that included premature in- the two lungs. In LUS examination, preterm infants were
fants with dyspnea admitted to Bethune First Hospital of placed in the supine, lateral or prone position, and US ex-
Jilin University from July 2019 to April 2021 (Fig. 1). The amination was performed in a quiet state, following the

2
 + MODEL
Pediatrics and Neonatology xxx (xxxx) xxx

Figure 1 Screening flowchart for premature infants.

procedure from top to bottom, left to right, and one by one performed to find relationships between LUSsc and the
between the intercostals. The LUS score was assigned as need of invasive respiratory support. Then, its sensitivity
follows: 0 Z normal aeration; 1 Z moderate loss of aera- and specificity were calculated. The cut-off value was
tion: interstitial syndrome (defined by multiple spaced B verified using kappa consistency and the McNemar test in
lines), or localized pulmonary edema (defined by coales- the verification set.
cent B lines in less than 50% of the intercostal space
examined in the transversal plane or subpleural consolida-
tions); 2 Z severe loss of aeration: alveolar edema defined 3. Results
by diffused coalescent B lines occupying the whole inter-
costal space; and 3 Z complete loss of lung aeration: lung Overall, 857 cases of premature infants born at 23þ0e36þ6
consolidation defined as a tissue pattern with or without air weeks were included. Non-invasive respiratory support
bronchogram. LUSsc was calculated as the sum of the 12 modes included CPAP (442 cases) and NIPPV (126 cases).
regional scores.18 The lowest score was 0 points, and the Invasive respiratory support modes included SIMV (182
highest was 36 points. All enrolled patients completed the cases) and HFOV (107 cases). The training set included 459
initial LUS examination within 2 h after admission and cases in the non-invasive respiratory support group (104
before the use of PS. Two fixed sonographers scored cases, <32þ0 weeks; 355 cases 32þ0e36þ6 weeks), and 228
together, and the sonographers did not know the breathing cases in the invasive respiratory support group (147 cases,
pattern of the infants. <32þ0 weeks; 81 cases, 32þ0e36þ6 weeks). We compared
two groups of clinical data, including general patient in-
formation, perinatal factors, arterial blood gas within 2 h of
2.5. Statistical analysis admission, and LUSsc, OI, Lung ultrasound diagnosis, etc.
Gestational age stratification was analyzed separately.
Statistical analyses were conducted using IBM SPSS 25.0
statistical software. Data were tested for normality with
the KolmogoroveSmirnov test, then either the median and 3.1. Univariate analysis of clinical data following
interquartile range was calculated, or absolute numbers invasive and non-invasive respiratory support for
and percentages to describe the variables were measured. preterm infants (23D0e36D6 weeks) (Table 1)
To compare variables between groups, we used the Mann
Whitney U test or the chi-square test depending on the The training set for preterm infants born at <32þ0 weeks
specialty of the variables. Multivariate logistic regression included 104 infants in the non-invasive respiratory support
analysis was performed on the significant factors of single group and 147 infants in the invasive respiratory support
factor analysis. The receiver operating curve (ROC) was group. There were significant differences in the weight,

3
 + MODEL
L. Zhang, J. Feng, D. Jin et al.

Table 1 A
Comparison of clinical factors of invasive and non-invasive respiratory support in preterm infants (<32þ0 weeks).
Basic Information non-invasive respiratory invasive respiratory X2/Z P-value
support group support group
(n Z 104) (n Z 147)
Gestational age (weeks) (IQR) 30 (28, 30) 30 (28，31) 0.252 0.801*
Weight (g) 1695 (1530, 1907) 1220 (1000, 1450) 8.846 ＜0.001 *
Male (n) 60 (57.7%) 77 (52.4%) 0.693 0.405y
cesarean section (n) 70 (67.3%) 76 (51.7%) 6.097 0.009y
Apgar score at 1min(n) 7 (7，8) 6 (5，7) 6.062 ＜0.001*
Apgar score at 5 min(n) 8 (8，9) 8 (7，8) 6.965 ＜0.001*
RDS (n) 48 (46.1%) 106 (72.1%) 20.120 ＜0.001y
TTN (n) 32 (30.8%) 27 (18.4%) 15.751 ＜0.001y
Use PS after LUS inspection (n) 9 (8.7%) 122 (82.9%) 186.094 ＜0.001y
PROM (n) 38 (36.5%) 40 (27.2%) 2.371 0.076y
Amniotic fluid fecal stain (n) 2 (1.9%) 10 (6.8%) 2.097 0.074y
Gestational diabetes (n) 16 (15.4%) 19 (12.9%) 2.306 0.354y
Gestational hypertension (n) 31 (29.8%) 32 (21.8%) 3.680 0.097y
Prenatal use of Corticosteroids (n) 46 (44.2%) 68 (65.3%) 1.073 0.425y
PH 7.32 (7.24，7.38) 7.3 (7.25，7.36) 0.765 0.444 *
PCO2 (mmHg)(IQR) 45 (36，56) 44 (37，54) 0.200 0.842*
OI (mmHg)(IQR) 326 (219，395) 200 (132，264) 7.373 ＜0.001*
BE (mmol/L)(IQR) 3.6 (-5.3，-1.7) 4.4 (-5.9，-2.6) 2.270 0.023*
lac (mmol/L)(IQR) 1.7 (1.3，2.3) 2.3 (1.7，3.4) 4.667 ＜0.001*
PEEP (cmH2O)(IQR) 6 (6，7) 7 (6，8) 6.424 ＜0.001*
PIP (cmH2O)(IQR) 16 (15.17) 17 (16.18) 3.616 ＜0.001 *
LUSsc (point)(IQR) 4 (2，8) 12 (5，18) 6.735 ＜0.001*
B
Comparison of clinical factors of invasive and non-invasive respiratory support in preterm infants (32þ0-36þ6 weeks).
Basic Information non-invasive respiratory Invasive respiratory X2/Z P-value
support group support group
(n Z 355) (n Z 81)
Gestational age (weeks) (IQR) 34 (33, 35) 33 (32, 34) 6.728 ＜0.001a
Weight (g) (IQR) 2150 (1790, 2490) 2000 (1570, 2330) 2.730 0.006a
Male (n) 184 (51.8%) 44 (54.3%) 0.164 0.390y
cesarean section (n) 258 (72.7%) 60 (74.0%) 0.065 0.459y
Apgar score at 1min(n) (IQR) 8 (7，8) 7 (6, 8) 5.391 ＜0.001a
Apgar score at 5 min(n) (IQR) 9 (8, 9) 8 (7.9) 5.432 ＜0.001a
RDS (n) 92 (25.9%) 66 (81.5%) 17.230 ＜0.001y
TTN (n) 95 (26.7%) 9 (11.1%) 14.351 ＜0.001y
Use PS after LUS inspection (n) 8 (2.3%) 47 (58.0%) 186.094 ＜0.001y
PROM (n) 98 (27.6%) 16 (19.8%) 0.499 2.371y
Amniotic fluid fecal stain (n) 9 (2.5%) 0 (0.0%) 0.154 2.097y
Gestational diabetes (n) 59 (16.1%) 8 (9.9%) 0.085 2.306y
Gestational hypertension (n) 103 (29.0%) 15 (18.5%) 0.035 3.680y
Prenatal use of Corticosteroids (n) 102 (28.7%) 28 (34.6%) 0.183 1.073y
PH (IQR) 7.33 (7.28，7.38) 7.31 (7.23，7.40) 1.379 0.168a
PCO2 (mmHg) (IQR) 42 (36，50) 44 (34，55) 1.147 0.251a
OI (mmHg) (IQR) 352 (276，428) 188 (123，262) 10.079 ＜0.001a
BE (mmol/L) (IQR) 3.8 (-5.4，-2.0) 4.4 (-6.7，-2.0) 1.641 0.101a
lac (mmol/L) (IQR) 1.8 (1.3，2.5) 2.1 (1.5，2.8) 2.182 0.029a
PEEP (cmH2O) (IQR) 5 (5，6) 7 (6，8) 10.879 ＜0.001a
PIP (cmH2O) (IQR) 16 (15.17) 17 (16.18) 4.407 ＜0.001a
LUSsc (point) (IQR) 2 (0，5) 14 (6.5，20) 10.463 ＜0.001a
IQR Z interquartile range; LUS: Lung Ultrasound; RDS: Respiratory Distress Syndrome; TTN: Transient Tachypnea of the New-born;
PROM: Premature Rupture of Membranes; PS: Pulmonary Surfactant; PH: Potential of Hydrogen; PCO2:Partial Pressure of Carbon Di-
oxide; OI: Oxygenation Index(PaO2/FiO2); BE: Base Excess; Lac: Lactic Acid; PEEP: Positive End Expiratory Pressure; LUSsc: Lung Ul-
trasound Score; Arterial blood gas analysis is completed within 2 h of admission; RDS and TTN diagnosis follow the standards of lung
ultrasound diagnosis of diseases; The application of PS follows the standard for supplementation of Pulmonary surfactant in premature
infants，PEEP and other ventilator parameters are based on the highest level of respiratory support required within 24 h of admission.
a
Mann-Whitney U test, yc2 test.

4
 + MODEL
Pediatrics and Neonatology xxx (xxxx) xxx

cesarean section, Apgar score at 1 min, Apgar score at For late preterm infants born within 32þ0e36þ6 weeks, the
5 min, ultrasound indications of lung disease conforming to area under the ROC curve was 0.863 (95% CI: 0811e0.915,
RDS, Transient Tachypnea of the Newborn (TTN), the use of P < 0.05). The best cut-off value was 7 points. These
PS after LUS examination, Lactic Acid (Lac), Positive End findings indicate that LUSsc has predictive value in the
Expiratory Pressure (PEEP), Peak Inspiratory Pressure (PIP), early respiratory support mode of premature infants with
LUSsc, OI (PaO2/FiO2), and BE between the two groups of dyspnea.
infants (P < 0.05).
Among preterm infants born at 32þ0e36þ6 weeks, 355 3.4. Verification（Table 3）
infants were included in the non-invasive respiratory sup-
port group and 81 infants were included in the invasive
Sixty-two early preterm infants (<32þ0 weeks) were
respiratory support group. There were significant differ-
included in the randomly selected validation set. The pre-
ences in the gestational age, both weight, Apgar score at
diction results were verified with the clinical respiratory
1 min, Apgar score at 5 min, ultrasound indications of lung
support model; the coincidence rate of two results was
disease conforming to RDS, TTN, use of PS after LUS ex-
83.9%, and the non-conformity rate was 16.1%. The LUSsc
amination, Lac, PEEP, PIP, LUSsc, and OI between the two
prediction results were in good agreement with the clinical
groups (P < 0.05).
results (Kappa value, 0.660; P < 0.05, McNemar test
P > 0.05). The late preterm infant (32þ0e36þ6 weeks)
3.2. Multivariate analysis of clinical predictors of group included 108 infants. The prediction results were
invasive respiratory support（Table 2） verified with the clinical respiratory support model. The
conformity rate of the two results was 92.6%, and the non-
Multiple logistic regression analysis was performed by conformity rate of the two results was 7.4%. The LUSsc
selecting indicators with clinical predictive value from the prediction results were very consistent with the clinical
factors with statistical differences in single factor analysis; results (Kappa value, 0.779; P < 0.05, McNemar test
among the factors included, for preterm infants born at P > 0.05).
<32þ0 weeks with dyspnea, infants with high LUSsc, low
oxygenation index, low birth weight, low Apgar score at 4. Discussion
5 min, and a more negative BE were more likely to receive
invasive respiratory support. For preterm infants born at
With the development of LUS technology in the field of
32þ0e36þ6 weeks with dyspnea, infants with high LUSsc,
neonatal medicine, we can diagnose lung diseases following
low oxygenation index, and low birth weight were more
the “Guidelines for the Ultrasonic Diagnosis of Newborn
likely to receive invasive respiratory support.
Lungs”.19 Additionally, using LUS to guide the use of me-
chanical ventilation and the timing of weaning,20 we can
3.3. ROC curve (Fig. 2) predict disease occurrence, establish a differential diag-
nosis, assess the severity of the disease, and guide the use
For early preterm infants born at <32þ0 weeks, the area of PS. Research has shown that LUS can be used as a pre-
under the ROC curve was 0.749 (95% CI: 0.689e0.809, dictor of invasive respiratory support with high accuracy;
P < 0.05). The best cut-off value was 8 points; the sensi- the sensitivity of 95% and specificity of 82.5% are highly
tivity and specificity were 74.0% and 68.3%, respectively. consistent with those of X-ray examinations (k Z 0.9; 95%

Table 2 A Logistic regression analysis of influencing factors between non-invasive respiratory support group and invasive
respiratory support group for ＜32þ0 weeks preterm.
Influencing factors b OR（95%CI） P
Use PS after LUS inspection 3.478 32.401（10.457e100.397） ＜0.001
OI 1.060 1.346（1.208e1.578） ＜0.001
LUSsc 0.113 1.119（1.037e1.209） 0.004
Birth weight 0.004 0.996（0.995e0.998） ＜0.001
Apgar score at 5 min 0.595 0.552（0.312e0.976） 0.041
BE 0.111 0.896 (0.778e1.030) 0.008
Constant 6.013 408.717 0.105
B
Logistic regression analysis of influencing factors between non-invasive respiratory support group and invasive respiratory
support group for 32þ0-36þ6 weeks preterm.
Influencing factors b OR（95%CI） P-value
Use PS after LUS inspection 4.500 89.988（8.796e920.686） ＜0.001
OI 0.900 2.304（1.409e3.768） <0.001
LUSsc 0.107 1.112（1.038e1.192） ＜0.001
Birth weight 0.002 1.002（1.001e1.003） 0.017
Constant 1.245 0.288 0.787

5
 + MODEL
L. Zhang, J. Feng, D. Jin et al.

was the LUSsc score.8,16,23 In our study, we demonstrated

the significant predictive ability of the 2 h LUSsc in pre-
dicting need for mechanical ventilation within 24 h of
admission. Thus, our findings may have significant practical
implications. Preterm infants born at <32þ0 weeks have a
LUSsc greater than 8, and preterm infants born at
32þ0e36þ6 weeks have a LUSsc greater than 7. This allows
for early identification of the most vulnerable infants,
which may lead to earlier decisions regarding invasive
mechanical ventilation.
In our study, the 12-segment score was chosen instead
of the commonly used 6-segment score.6 Although some
studies have found that the use of complex LUSscs is of
little significance,24 other studies that affirmed that the
12-zone score is superior to the simple score in diagnosing
certain diseases; the 12-segment score takes into account
the posterior part of the lung and it can be used to eval-
uate the lung condition more comprehensively than the
simple and more commonly used 6-segment score.4,15,18
However, complex LUS scoring means more operations
are required. We were careful about the implementation
process and kept the examination time short (approxi-
mately 30 s); the procedure did not cause problems or
Figure 2 a ROC curve of Lung ultrasound scores in preterm destabilize the patient’s condition. All neonates with
with dyspnea less than 32þ0weeks b ROC curve of Lung ultra- dyspnea in our clinical center undergo comprehensive
sound scores in preterm with dyspnea within 32þ0-36þ6 weeks. bedside LUS examinations in a timely manner. The study
did not cause additional harm to the patients and there
was no delay in rescue treatment.
CI: 0.83e1).15,21,22 However, there are few studies on using Early preterm infants have a higher cut-off value for
LUSsc to predict the use of ventilator mode in preterm in- invasive respiratory support than late preterm infants. This
fants. A study using modified LUSsc predicted ventilation may be related to the immature lung development in pre-
requirements and the mode of respiratory support on DOL mature infants and the atypical clinical manifestations.
3.15 In a clinical setting, it is particularly important to Therefore, when there are clinical manifestations and
select the respiratory support mode quickly, accurately, laboratory examination results of the same severity, pre-
and reasonably after admission; however, previous studies term infants with a low gestational age often have more
have not addressed the prediction of respiratory support severe lung changes, so the cut-off value corresponding to
mode within 24 h of admission. Our study aimed to establish invasive respiratory support is higher.6
this. However some researchers still believe that the LUSsc has
It is widely accepted that LUSsc is a comprehensive some definite shortcomings and that it may not be able to
index that can reflect the oxygen content and ventilation of assess the severity of certain pulmonary diseases in neonates
the lungs. In studies where LUS images were quantified, it accurately. For example, in cases of pulmonary Langerhans
was shown that the more critically ill the infant, the higher cell histiocytosis, lung interstitial syndrome, diffuse

Table 3 A Comparison of lung ultrasound scores in the validation set for preterm less than 32þ0 weeks predicted respiratory
support mode and clinical actual respiratory support (Kappa value 0.660, P < 0.05, McNemar test P > 0.05).
Actual respiratory support
Non-invasive Invasive Total
Lung ultrasound score predicts respiratory support Non-invasive 19 5 24
Invasive 5 33 38
Total 24 38 62
B
Comparison of lung ultrasound scores in the validation set for preterm 32þ0-36þ6 weeks predicted respiratory support
mode and clinical actual respiratory support. (Kappa value 0.779, P < 0.05, McNemar test P > 0.05).
Actual respiratory support
Non-invasive Invasive Total
Lung ultrasound score predicts respiratory support Non-invasive 81 4 85
Invasive 4 19 23
Total 85 23 108

6
 + MODEL
Pediatrics and Neonatology xxx (xxxx) xxx

parenchymal lung disease, lymphangioleiomyomatosis etc, LUSsc as an important reference method and judge by
high-resolution CT examination is better than LUS, and pul- integrating other influencing factors to choose respiratory
monary hemorrhage and pneumothorax LUSsc cannot be support appropriately.
used to judge the severity of the disease accurately.25,26 Regarding limitations and prospects, our research
Therefore, we excluded cases with these diseases as much included samples from a single source, and clinicians in a
as possible. The main clinical diagnosis of mechanically single center have a limited choice of ventilator modes.
ventilated infants in our study was performed using ultra- Therefore, multi-center and large-sample studies are
sound RDS and TTN. LUSsc can be used to effectively assess required to further verify and modify the LUSsc. Besides the
disease severity.27 LUSsc, there are other indicators that can predict the res-
In our study, we also explored clinical factors affecting piratory support pattern of premature infants, but our
mechanical ventilation. In both gestational age categories, experiment did not include all such indicators. Further
LUSsc, birth weight, OI, and use of PS after LUS inspection research should be conducted to establish a complete
were predictors of mechanical ventilation. Body weight was prediction system for assisting in the selection of ventilator
an indicator of fetal growth and development. The lower modes in premature infants with dyspnea after birth. Non-
birth weight, the greater the probability was that infants invasive and invasive respiratory support modes include a
required invasive respiratory support. In a research report variety of options such as CPAP, NIPPV, SIMV, and HFOV. In
on mechanical ventilation after congenital heart disease, the future, we can expand the data and use the LUSsc to
children with lower body weight had longer mechanical guide the selection of specific ventilator modes appropri-
ventilation times than children with higher body weight.28 ately. The method of oxygen supply, such as mechanical
Our research also shows that low body weight is a risk ventilator or CPAP, may have influenced scores because we
factor for invasive respiratory support. OI was calculated as did not exclude patients who were receiving any kind of
the PtcO2 to FiO2 ratio, referring to the oxygen exchange respiratory support. The variation trend of LUSsc may be
efficiency of the lungs. The normal range is explored at different time points after respiratory support
400e500 mmHg; OI below 300 mmHg indicates a risk of lung in the future research. This study focused on premature
respiratory dysfunction, so OI can be used an effective infants. In the future, we can conduct relevant research on
predictor of invasive respiratory support. Research has full-term infants who have difficulty breathing and need
shown that the LUSsc was significantly correlated with the respiratory support after admission. The large size and low
OI and that the correlation remained significant after frequency of the linear array 9.0-MHz high-frequency probe
adjustment for GA.11 Since OI is greatly affected by the may affect the accuracy of readings from each lung area.
inspired oxygen concentration at the time of blood gas For each lung area and each probe, an appropriate LUSsc
analysis, the time when LUS should be performed and time cut off value should be initially calculated. Future studies
when OI should be measured are not completely consistent. should try to use different probes to verify the consistency
Our study did not combine the LUSsc with the OI to quantify of the results.
the predictive value of respiratory support. Follow-up
research should be carried out in this direction on the
basis of rational design, resulting in more reasonable and 5. Conclusion
scientific predictions.
When PS is lacking due to congenital or pulmonary dis- LUSsc is an independent predictor of the respiratory sup-
ease, the patient may experience alveolar collapse and port pattern in premature infants with dyspnea within 24 h
exudation with progressively worsening dyspnea. Treat- after admission. For preterm infants born at <32þ0 weeks,
ment requires PS to be supplemented in time. After use of an LUSsc greater than 8 indicated that invasive mechanical
PS, most infants with RDS are immediately relieved of their ventilation was suitable; for preterm infants born at
dyspnea, the parameters of the ventilator are lowered, and 32þ0e36þ6 weeks, LUSsc greater than 7 shows infants
some infants are moved from the invasive respiratory sup- require invasive ventilation. This technique had high
port mode to the non-invasive respiratory support mode. In sensitivity and specificity, and the validation results were
the experimental data inclusion criteria, the infants who good.
received PS before the initial LUS examination were
excluded. The results showed that the use of PS was
significantly related to invasive respiratory support. The
application of clinical PS is often based on FIO2 and the
Funding
clinical situation of the infants; PS is used only in premature
infants with obvious dyspnea, and such patients often The study was supported by grants from the Scientific
require early invasive respiratory support to survive the Research Foundation of the department of science and
dangerous period. technology of Jilin Province, China (Grant Number:
In summary, we obtained the prediction scores using 20190701050 GH).
LUSsc for use of early ventilator mode in premature infants
with dyspnea. The LUSsc prediction score had high sensi-
tivity and specificity, and the consistency of verification Ethics approval
results was good, so it can be used for clinical reference.
The results are also useful in predicting the factors of the This study was approved by the Ethics Committee of the
choice of ventilator mode; and birth weight, OI, and use of First Hospital of Jilin University. The ethical batch number
PS also had predictive value. Clinicians should consider is 19K053-001.

7
 + MODEL
L. Zhang, J. Feng, D. Jin et al.

Consent for publication tant need in extremely preterm neonates. Pediatrics 2018;
142:e20180463.
12. Raimondi F, De Winter JP, De Luca D. Lung ultrasound-guided
Patients’ parents signed informed consent regarding pub- surfactant administration: time for a personalized,
lishing their data. physiology-driven therapy. Eur J Pediatr 2020;179:1909e11.
13. Rodriguez-Fanjul J, Jordan I, Balaguer M, Batista-Muñoz A,
Ramon M, Bobillo-Perez S. Early surfactant replacement guided
Conflict of interest by lung ultrasound in preterm newborns with RDS: the
ULTRASURF randomised controlled trial. Eur J Pediatr 2020;
179:1913e20.
The authors declare no competing interests.
14. Perri A, Tana M, Riccardi R, Iannotta R, Giordano L,
Rubortone SA, et al. Neonatal lung ultrasonography score after
surfactant in preterm infants: a prospective observational
Acknowledgements
study. Pediatr Pulmonol 2020;55:116e21.
15. Szyman
ski P, Kruczek P, Hoz_ ejowski R, Wais P. Modified lung
The authors are grateful to 2 senior sonographers Shuyu Si ultrasound score predicts ventilation requirements in neonatal
and Yiyi Guo for their analysis of the images and to the respiratory distress syndrome. BMC Pediatr 2021;21:17.
patients who participated in this study. 16. Abdelmawla M, Louis D, Narvey M, Elsayed Y. A lung ultrasound
severity score predicts chronic lung disease in preterm infants.
Am J Perinatol 2019;36:1357e61.
References 17. Keszler M. Mechanical ventilation strategies. Semin Fetal
Neonatal Med 2017;22:267e74.
1. Sweet DG, Carnielli V, Greisen G, Hallman M, Ozek E, Te Pas A, 18. Rouby JJ, Arbelot C, Gao Y, Zhang M, Lv J, An Y, et al. Training
et al. European consensus guidelines on the management of for lung ultrasound score measurement in critically ill Patients.
respiratory distress syndrome - 2019 update. Neonatology Am J Respir Crit Care Med 2018;198:398e401.
2019;115:432e50. 19. Kurepa D, Zaghloul N, Watkins L, Liu J. Neonatal lung ultra-
2. Jensen EA, DeMauro SB, Kornhauser M, Aghai ZH, sound exam guidelines. J Perinatol 2018;38:11e22.
Greenspan JS, Dysart KC. Effects of multiple ventilation cour- 20. Abdel Rahman DA, Saber S, El-Maghraby A. Diaphragm and lung
ses and duration of mechanical ventilation on respiratory ultrasound indices in prediction of outcome of weaning from
outcomes in extremely-low-birth-weight infants. JAMA Pediatr mechanical ventilation in pediatric intensive care unit. Indian
2015;169:1011e7. J Pediatr 2020;87:413e20.
3. Liu J. Lung ultrasonography for the diagnosis of neonatal lung 21. Rodrı́guez-Fanjul J, Balcells C, Aldecoa-Bilbao V, Moreno J,
disease. J Matern Fetal Neonatal Med 2014;27:856e61. Iriondo M. Lung ultrasound as a predictor of mechanical ventilation
4. Liu X, Lv X, Jin D, Li H, Wu H. Lung ultrasound predicts the in neonates older than 32 Weeks. Neonatology 2016;110:198e203.
development of bronchopulmonary dysplasia: a prospective 22. Raimondi F, Migliaro F, Sodano A, Ferrara T, Lama S, Vallone G,
observational diagnostic accuracy study. Eur J Pediatr 2021; Capasso L. Use of neonatal chest ultrasound to predict nonin-
180:2781e9. vasive ventilation failure. Pediatrics 2014;134:e1089e94.
5. Raimondi F, Rodriguez Fanjul J, Aversa S, Chirico G, Yousef N, 23. Raimondi F, Migliaro F, Corsini I, Meneghin F, Dolce P, Pierri L,
De Luca D, et al. Lung ultrasound for diagnosing pneumothorax et al. Lung ultrasound score progress in neonatal respiratory
in the critically ill neonate. J Pediatr 2016;175:74e8. distress syndrome. J Pediatrics 2021;147:e2020030528.
6. Brat R, Yousef N, Klifa R, Reynaud S, Shankar Aguilera S, De 24. Kayki G, Yigit S, Tandircioglu UA, Celik HT, Yurdakok M. Lung
Luca D. Lung ultrasonography score to evaluate oxygenation ultrasound (LUS) and surfactant treatment: looking for the best
and surfactant need in neonates treated with continuous pos- predictive moment. J Perinatol 2021;41:1669e74.
itive airway pressure. JAMA Pediatr 2015;169:e151797. 25. Liu J. The lung ultrasound score cannot accurately evaluate
7. Copetti R, Cattarossi L, Macagno F, Violino M, Furlan R. Lung the severity of neonatal lung disease. J Ultrasound Med 2020;
ultrasound in respiratory distress syndrome: a useful tool for 39:1015e20.
early diagnosis. Neonatology 2008;94:52e9. 26. Davidsen JR, Bendstrup E, Henriksen DP, Graumann O,
8. Pang H, Zhang B, Shi J, Zang J, Qiu L. Diagnostic value of lung Laursen CB. Lung ultrasound has limited diagnostic value in
ultrasound in evaluating the severity of neonatal respiratory rare cystic lung diseases: a cross-sectional study. Eur Clin
distress syndrome. Eur J Radiol 2019;116:186e91. Respir J 2017;4:1330111.
9. Corsini I, Parri N, Ficial B, Dani C. Lung ultrasound in the 27. Razak A, Faden M. Neonatal lung ultrasonography to evaluate
neonatal intensive care unit: review of the literature and need for surfactant or mechanical ventilation: a systematic
future perspectives. Pediatr Pulmonol 2020;55:1550e62. review and meta-analysis. Arch Dis Child Fetal Neonatal Ed
10. Vergine M, Copetti R, Brusa G, Cattarossi L. Lung ultrasound 2020;105:164e71.
accuracy in respiratory distress syndrome and transient 28. Alrddadi SM, Morsy MM, Albakri JK, Mohammed MA, Alnajjar GA,
tachypnea of the newborn. Neonatology 2014;106:87e93. Fawaz MM, et al. Risk factors for prolonged mechanical venti-
11. De Martino L, Yousef N, Ben-Ammar R, Raimondi F, Shankar- lation after surgical repair of congenital heart disease. Experi-
Aguilera S, De Luca D. Lung ultrasound score predicts surfac- ence from a single cardiac center. Saudi Med J 2019;40:367e71.