This was an observational, cross-sectional, international, pathophysiological, bi-center study conducted between December 2019 and July 2021 in two academic referral intensive care (one neonatal and one adult) units (ICU). Two institutional review boards independently approved the study (n.9596/17/1513 and n.16/58) for adults and neonates, respectively. Informed consent was obtained from parents or guardians upon ICU admission of neonates. For ICU-admitted adults, informed consent was obtained as per local regulations, following the local Ethical Committee recommendations. The participation to the study did not change the routine clinical care. The study was conducted in accordance with the Helsinki declaration, was completely anonymous and respected local and European privacy regulations. The manuscript was prepared following STROBE guidelines .
Cases consisted of adult and newborn patients admitted to ICUs for restrictive respiratory failure. To be enrolled adults must have fulfilled all the following criteria: (1) age > 18 years; (2) diagnosis of ARDS according to the Berlin definition ; (3) need of invasive mechanical ventilation. Neonates could have been enrolled if they fulfil all the following criteria: (1) postnatal age ≤ 7 days; (2) diagnosis of neonatal ARDS (NARDS), or diagnosis of respiratory distress syndrome (RDS; i.e., hyaline membrane disease due to primary surfactant deficiency), both according to the criteria detailed in the Montreux definition ; (3) need of invasive mechanical ventilation. Both RDS and NARDS patients were enrolled, since these ideally represent examples of moderate and severe restrictive respiratory failure, respectively [16, 17].
Additionally, two other groups of subjects were enrolled as controls and consisted of adult and newborn patients admitted to the ICU with no lung disease (NLD). These patients fulfilled all the following criteria: (1) need of invasive ventilation for non-pulmonary reason; (2) no need for supplemental oxygen (i.e., ventilation in room air) as well as steadily normal blood gases and vital parameters; (3) normal chest clinical examination; (4) absence of thoracic trauma and any respiratory disorder in the last month or week, in adults and neonates, respectively.
Exclusion criteria for adults with and without ARDS were: acute or chronic obstructive conditions (i.e., chronic obstructive pulmonary disease, asthma), neuromuscular diseases, interstitial lung diseases, rib cage anomalies, home long-term oxygen therapy, need for ECMO. Exclusion criteria for neonates with and without NARDS or RDS were: major congenital malformations, genetic syndromes and chromosomopathies, neuromuscular diseases, congenital lung or rib cage anomalies, airway obstruction due to meconium plugging, need for rescue high frequency oscillatory ventilation, as previously described , or ECMO.
Lung mechanics measurements
Adult patients were on apneic sedation or paralyzed when clinically indicated, according to current best practice principles . Volume-controlled ventilation was provided with a tidal volume (Tv) of 6 mL/kg of predicted body weight; inspiratory time and flow were 60 L/min and 0.3 s, respectively. Respiratory rate was titrated to obtain pH between 7.35 and 7.45 with a maximal rate of 30 bpm. Positive end-inspiratory pressure (PEEP) was titrated according to current international guidelines  and inspired oxygen fraction (FiO2) was set to obtain a peripheral oxygen saturation (SpO2) no lower than 90%. Crs was measured in a “quasi-static” way using a standard 500 ms end-inspiratory pause to obtain plateau pressure (Pplat) and do the following calculation:
Crs = Tv/(Pplat − (PEEP + auto-PEEP)). Crs in adults was indexed to the ideal body weight. Respiratory system resistances (Rrs) were estimated as Rrs = (Peak inspiratory pressure − Pplat)/flow.
Neonates were sedated and time-cycled, pressure-regulated, assisted-controlled ventilation was provided as previously described ; no muscle relaxants were used. The same blood gas values of adult patients were targeted, but FiO2 was as low as possible to guarantee pre-ductal SpO2 between 90 and 95%. They were on continuous flow neonatal ventilators equipped with a low dead-space, hot-wire anemometers coupled with pressure sensors at the Y-piece to maximize the accuracy of lung mechanics assessment . Sensors underwent serial technical quality controls  and were calibrated before each use, following manufacturers’ recommendations. Lung mechanics measurements were performed as previously described : Crs and Rrs were dynamically estimated by breath-to-breath analysis shown by the ventilator software but measurements were considered only after airway suctioning, once neonates were stable and using our previously described technique to reduce ventilatory drive and increase measurement precision . In detail, spontaneous breathing was temporally avoided by increasing the mechanical rate, whereas the flow was decreased to 5 L/min to reach a “quasi-static” situation. Under these conditions, when leaks were < 5% and pressure/flow-volume loops were steady, Crs and Rrs were averaged on 10 mechanical breaths. In neonates, Crs was indexed to the birth weight and was measured before surfactant administration, if any. Lung mechanics was assessed in all adults and neonates upon ICU admission as per our clinical routine and all patients were supine during the assessment.
Lung ultrasound was performed upon ICU admission as per our routine clinical protocols and anyway within 1 h from the lung mechanics measurements. Lung ultrasound exams were conducted with convex [5 MHz, using Xario-200® (Toshiba, Tokyo, Japan)] or micro-linear, “hockey-stick” (15 MHz, using CX-50®, Philips, Eindhoven, Netherlands) probes in adults and neonates, respectively. In both adults and neonates, ultrasound setting was as follows: gain was automatically adjusted with the dedicated software function, depth and focus were adjusted according to patients’ size and the sign of interest and no harmonics was used. Ultrasound exams were always performed by attending physicians with at least 2 years of training with daily use of lung ultrasound, or by residents in training under the supervision of the former . Lung ultrasound is the first-line lung imaging technique in both recruiting ICUs, which are known to use lung ultrasound on a daily basis integrated in the clinical routine.
LUS was calculated using a standardized simple protocol already validated for both adults and neonates [5, 26]. In detail, a total of six lung areas (3 for each lung) were examined, with both transverse and longitudinal scans, while patients were supine; a score from 0 to 3 was assigned to each lung area and LUS could range from 0 to 18 (the higher the score, the worse the lung aeration). The scoring system was based on classical lung ultrasound semiology: 0 indicated a normal lung with presence of lung sliding, visible A-lines with less than three B-lines per intercostal space; 1 was given to mild alveolo-interstitial pattern, depicted by at least three B-lines or presence of multiple subpleural consolidations (with a maximal size ≤ 1 cm); 2 was attributed to a severe alveolo-interstitial pattern, represented by multiple, crowded and coalescent B-lines (i.e., “white” lung) and/or multiple subpleural consolidations separated by thickened or irregular pleura; 3 was given to severe loss of lung aeration represented by consolidations (i.e., subpleural echo-poor or tissue-like echotexture zones with size > 1 cm and irregular borders, which may also have bronchogram as mixed hypo- and hyperechogenic zones).
A formal sample size calculation was unfeasible since only one case series of 10 adult patients with respiratory failure reported the relationship between ultrasound-assessed lung aeration and Crs, without investigating the effect of age . Thus, we decided to enroll a larger population and we choose a convenience sample size of at least 40 adult and 40 newborn patients.
Continuous variables were expressed as means (standard deviation). Categorical variables were presented as number (%). LUS and Crs were compared between adults and neonates with different respiratory conditions using one way-ANOVA followed by post hoc Sidak test, if appropriate. LUS and Crs were studied with correlation analysis using Pearson correlation coefficient (r) and its 95% confidence interval (CI), estimated with Fisher method. Results were also graphically shown using scatter plots analyzed with local (smoother) regression with 95% Epanechnikov kernel. The relationship between Crs and LUS was finally investigated with multivariable models built by multivariate linear regressions with backward-stepwise method. Covariates inserted in the models were: BMI, patient age (or gestational age for neonates) and the respiratory condition (considered as NLD = 0, RDS = 1, NARDS = 2 for neonates and NLD = 0, ARDS = 1 for adults). These covariates were chosen as they are known to potentially influence the severity of respiratory failure [27,28,29,30,31]. Multi-collinearity was evaluated considering the variance inflation factor, as previously described . Birth weight was not considered as it is known to have significant multi-collinearity with the gestational age. Analyses were performed with SPSS 28 and p < 0.05 was considered statistically significant.