Opportunity for invasive mechanical ventilation in NRDS: a retrospective cohort study in China

Abstract Purpose Mechanical ventilation, as a critical breathing support, plays a critical role in the treatment of neonatal respiratory distress syndrome (NRDS). We aim to describe the clinical characteristics of NRDS and give suggestions about when to start mechanical ventilation. Methods We conducted a retrospective cohort study, enrolling 95 neonates between December 2016 and October 2021. Diagnosis of NRDS was according to the Berlin definition. Spearman’s and ROC analysis was used to determine the variables correlated with hospital stay and optimal opportunity for mechanical ventilation. Results Ninety-five subjects with NRDS were included. Lower PaO2 and higher PaCO2 in arterial blood gas prompt longer discharge time after mechanical ventilation and total in-hospital stay (p < .05), in which significant correlations were identified in Spearman’s analysis. ROC analysis illustrated that mechanical ventilation starting when PaO2 was 52.5 mmHg contributed to the shortest discharge time and in-hospital stay. PaCO2 of 45.4 mmHg was another optimal cut-off value for the initiation of mechanical ventilation with an AUC of 0.636 (sensitivity 91.5%, specificity 29.2%, p = .022). Conclusion PaO2 and PaCO2 were significantly correlated with discharge time and in-hospital stays. When PaO2 was reduced to 52.5 mmHg or PaCO2 increased to 45.5 mmHg, mechanical ventilation was strongly recommended.


Introduction
Acute respiratory distress syndrome (ARDS) is a type of respiratory failure characterized by the rapid onset of widespread inflammation in the lungs [1]. Symptoms include dyspnea, tachypnea, and cyanosis. The underlying mechanism involves diffuse injury to the cells that form the barrier of the microscopic air sacs of the lungs, surfactant dysfunction, activation of the immune system, and dysfunction of the body's regulation of blood clotting [2]. In effect, ARDS impairs the lungs' ability to exchange oxygen and carbon dioxide [1]. Diagnosis of ARDS in adults according to the Berlin definition is based on respiratory failure within 1 week of a known clinical insult or new or worsening respiratory symptoms that cannot be fully explained by cardiac failure or fluid overload, a PaO 2 /FiO 2 ratio of <300 mmHg despite a positive endexpiratory pressure (PEEP) of >5 cm H 2 O, and bilateral opacities not fully explained by effusions, lobar/lung collapse, or nodules in chest imaging [3].
Neonatal respiratory distress syndrome (NRDS), also called infantile respiratory distress syndrome (or increasingly called surfactant deficiency disorder) [4] and previously called hyaline membrane disease, is a syndrome in premature infants caused by developmental insufficiency of pulmonary surfactant production and structural immaturity in the lungs. It can also be a consequence of neonatal infection and can result from a genetic problem with the production of surfactant-associated proteins [5,6]. NRDS manifests as rapid breathing (>60 breaths per minute), a rapid heart rate, chest wall retractions, expiratory grunting, nasal flaring, and blue discoloration of the skin during breathing efforts. The diagnosis is made by the clinical picture and chest X-ray examination, which demonstrates decreased lung volumes; absence of the thymus; a small, discrete, and uniform infiltrate that involves all lobes of the lung; and air bronchograms.
Mechanical ventilation is a critical breathing support measure and treatment for all patients with ARDS, including children and neonates. However, there are still no clear standards about when neonates with NRDS should be given mechanical ventilation. We performed the present study to examine the clinical characteristics of NRDS and determine the optimal time to treat NRDS by mechanical ventilation.

Study population
This single-center retrospective observational cohort study was conducted from December 2016 to October 2021 in the neonatal intensive care unit of the Affiliated Hospital of Jining Medical University. Neonates aged <28 d were considered eligible for inclusion in this study if they fulfilled the Berlin definition of ARDS and were anticipated to require mechanical ventilation. Neonates were excluded if they met at least one of the following criteria: neonatal pneumonia (defined according to the following criteria: Cough, dyspnea, and dry and wet rales of the lung, symptoms of pneumonia indicated by lung X-ray and positive sputum culture), congenital cardiac disease, or persistent pulmonary hypertension (PPHN, diagnosed as a chronically elevated mean pulmonary arterial pressure of !20 mmHg at rest with increased pulmonary vascular resistance by echocardiography); thoracic deformations; severe hemodynamic instability, defined as an increase in vasoactive drug dosages in the last 6 h to increase the mean arterial pressure or cardiac index; immunodeficiency or immunosuppressive drug therapy other than glucocorticoids; and technique accuracy issues. The study was approved by the Medical Research Ethics Committee of Affiliated Hospital of Jining Medical University. Because of the retrospective study design, the requirement for informed consent was waived. The study was performed in accordance with the guidelines of the Declaration of Helsinki.

Data collection
The patients' electronic medical records were retrospectively reviewed, and basic information on demographics, neonatal birth conditions, maternal gestation history, and medical comorbidities were extracted at enrollment. The diagnosis of ARDS was confirmed according to the Berlin definition [3]. After enrollment, clinical data including vital signs and laboratory results were recorded daily, among which oxygenation was evaluated by arterial blood gas and SpO 2 . The gestational age of the patients included in the study was determined by ultrasound. The reliability of radiographic findings was independently confirmed by two pediatric radiologists who were blinded to the study. Vasopressor usage was recorded for any doses. The patients were classified into mild, moderate, and severe groups according to the Pediatric Acute Lung Injury Consensus Conference (PALICC) oxygenation index criteria at the time of diagnosis [7]. To assess the severity of ARDS, the Acute Physiology and Chronic Health Evaluation III (APACHE III) score and the Simplified Acute Physiology Score II (SAPS II) were recorded during the first 24 h after diagnosis of ARDS [8,9]. The modes and duration of mechanical ventilation and the ventilator setting parameters, including FiO 2 , PEEP, peak inspiratory pressure, tidal volume, and other data, were recorded throughout the patients' stay in the neonatal intensive care unit. The PaO 2 /FiO 2 ratio was calculated from the mechanical ventilation-related variables. The total hospital stay, hospital stay after mechanical ventilation, and inhospital mortality were recorded as well. The total hospital stay was defined as the duration from admission to discharge, and the discharge time was defined as the duration from the end of mechanical ventilation to discharge. The manufacturer of an invasive ventilator was a SLE5000 infant ventilator with high-frequency oscillation from UK, while noninvasive ventilations were operated by German Siemens Maquet 3090 (MedinCNO@) and American Carefusion Vyaire Infant Flow SiPAP.

Statistical analysis
All patients were stratified according to the median discharge time after invasive ventilation ( 9 and >9 d) and the total in-hospital stay ( 11 and >11 d). Statistical analyses were performed using SPSS statistics version 24.0 (IBM Corp., Armonk, NY, USA). Variables were tested to determine whether they had a normal distribution using the Shapiro-Wilk test and the Kolmogorov-Smirnov test. Continuous variables are expressed as mean ± standard deviation or (for non-normally distributed data) median (range) and were compared between groups using Student's t test or the nonparametric Mann-Whitney U test. Categorical variables are presented as numbers and percentages, and clinically relevant differences were evaluated by the chi-square test or Fischer's exact test, as appropriate. Correlations between the discharge time after invasive ventilation as well as the total inhospital stay and some relevant variables in clinical practice were determined using Spearman's correlation coefficient. Receiver operating characteristics (ROC) analysis was used to determine the cutoff value of relevant variables for a reduced hospital stay. A p-value of <.05 was considered statistically significant.

Demographic characteristics
Ninety-five neonates were finally included, and their demographic and other basic information are presented in Table 1. The median time after birth was 0.8 (0.2-72.0) h, and there was an overall majority (66.3%) of male neonates. The mean birth height and weight were 48.6 ± 2.6 cm and 2.9 ± 0.5 kg, respectively. No significant difference in the Apgar score was found between groups (p > .05). In total, 82 of 86 (95.3%) patients had neonatal complications, such as neonatal sepsis, asphyxia, and other conditions. The mean maternal age was markedly older in those with a longer length of stay after mechanical ventilation (31.6 ± 4.9 vs. 29.1 ± 5.8 years, p ¼ .028). No significant difference was found between groups concerning abnormalities in the maternal history of pregnancy and delivery, including cesarean delivery, pregnancy complications, abnormal pregnancy, and pregnancy complications.

Comparison of clinical features between groups
Neonates with a shorter in-hospital stay had a significantly higher heart rate at the diagnosis of ARDS than neonates with a longer in-hospital stay (145.0 ± 10.5 vs. 140.6 ± 11.3 beats/min, p ¼ .043). Lower PaO 2 and higher PaCO 2 in the arterial blood gas analysis were associated with a longer discharge time after mechanical ventilation and longer total in-hospital length of stay (p < .05). Twenty-one (22.1%) patients were diagnosed with mild ARDS, 59 (62.1%) with moderate ARDS, and 15 (15.8%) with severe ARDS. In association with a longer stay after mechanical ventilation, there was a significant increasing trend of the APACHE III score (25 vs. 24, p ¼ .021) and simultaneously higher usage of surfactant and vasoactive agents (72.3% vs. 43.8%, p ¼ .005 and 41.9% vs. 19.1%, p ¼ .019) ( Table 2).
Eighty-five (89.5%) patients underwent noninvasive mechanical ventilation before invasive mechanical ventilation, and noninvasive high-frequency oscillatory ventilation was the most widely used mode (61.2%). The median time between ARDS onset and the start of mechanical ventilation was 5.0 h. High-frequency oscillatory ventilation mode was selected in 85 of 95 (89.5%) patients. Most mechanical ventilation-related variables (e.g. PaO 2 /FiO 2 , PEEP, tidal volume, and ventilation mode), other than the plateau pressure, did not differ significantly between groups. Eighty-five of 95 (89.5%) patients' conditions were complicated with other severe diseases such as neonatal myocardial damage, intracranial hemorrhage, sepsis, and others; however, no significant differences were observed between the groups ( Table 3).

Investigation of correlations and ROC analysis
Spearman's correlation analysis was performed for variables with remarkable differences between groups, including maternal age, highest heart rate, lowest PaO 2, and highest PaCO 2 , APACHE III score, use of surfactant and vasoactive agents, and plateau pressures in mechanical ventilation. For all variables except PaO 2 and PaCO 2 , no significant correlations were observed. A graphic representation of Spearman's correlation coefficient between PaO 2 /PaCO 2 and the discharge time after mechanical ventilation as well as the total length of stay is presented in Figure 1(A-D). Spearman's correlation coefficient for the suspected correlations was À0.365 (p < .001) for the correlation between PaO 2 and the discharge time, À0.326 (p ¼ .001) for that between PaO 2 and the in-hospital stay, 0.324 (p ¼ .001) for that between PaCO 2 and the discharge time, and 0.297 (p ¼ .003) for that between PaCO 2 and the total length of stay. As is shown by the ROC curve in Figure 2(A,B), initiation of mechanical ventilation when PaO 2 was 52.5 mmHg contributed to the shortest discharge time and in-hospital length of stay, the areas under the curve (AUCs) of which were 0.676 (sensitivity, 50.0%; specificity, 76.6%; p ¼ .003) and 0.620 (sensitivity, 46.9%; specificity, 73.9%; p ¼ .043), respectively. Figure  2(C) shows the ROC analysis for PaCO 2 and the discharge time, in which the cutoff value was 45.4 mmHg and the AUC was 0.636 (sensitivity, 91.5%; specificity, 29.2%; p ¼ .022). The in-hospital stay was shortest when mechanical ventilation was started at a PaCO 2 of 55.5 mmHg, of which the AUC was 0.662 (sensitivity, 45.7%; specificity, 77.6%; p ¼ .006) (Figure 2(D)).

Discussion
ARDS was first reported in 1967 by Ashbaugh et al. who described sudden respiratory failure due to non-cardiogenic pulmonary edema [10]. It has been recognized as a rare disease with an incidence of 2 to 12 per 100,000 people per year, but it may occur at any age [11]. However, the definition of ARDS has undergone multiple changes and revisions, and studies aiming to improve the clinical outcomes of ARDS were mainly focused on adult patients. The concept of pediatric ARDS was not put forward until the PALICC in 2015 [7,12], and age-specific definitions are now available, with the PALICC definition and Montreux definition for pediatric ARDS and NRDS, respectively, used for children beyond the first month of age and for neonates from birth until 4 weeks of age (44 weeks post-menstrual age if born before 40 weeks' gestation) [13]. ARDS is characterized by inflammation of lung tissue, alveolar and/or endothelial damage, and complex surfactant injury. Despite several similarities, pediatric and neonatal patients with ARDS differ from adult patients in terms of etiology, epidemiology, pathophysiology, triggers, clinical course, clinical approach, and outcomes. The relative rarity, population diversity, and high morbidity and mortality have hindered the progression of evidence-based management; thus, clinical practice of pediatric and neonatal ARDS has remained challenging and bound to be performed with reference to studies of adult ARDS. Systematic data and studies involving age-related subgroups remain insufficient.
Dysregulation of various inflammatory cells (such as macrophages, neutrophils, and lymphocytes) and cytokine mediator signaling pathways is known to produce an inflammatory microenvironment in the lungs, thus contributing to the development of ARDS [14], for which intravenous glucocorticoid therapy has been widely used. However, despite improvements in multiple other treatments, including nitric oxide, surfactant, prostaglandins, fluid balance, and high-frequency ventilation, which are common in clinical practice, ARDS mortality remains high in most existing observational studies [15], while oxygen supplementation and lung-protective ventilation strategies remain crucial for ARDS treatment.
Nevertheless, the diagnostic criteria of NRDS are still not generally acknowledged. Few cross-sectional or cohort studies focusing on the clinical manifestations of NRDS or the therapeutic discrepancy between adults and neonates have been reported. Thus, our study aims to fill the gaps in this field through a retrospective analysis of the clinical characteristics of neonatal ARDS and identify the optimal opportunity for mechanical ventilation. In the present study, we examined a cohort of newborn patients with ARDS and provided a more detailed integration of this specific age-related subgroup of NRDS. Maternal age was significantly different between neonates with a short and long discharge time but not between those with a short and long in-hospital stay, and there was no linear correlation between maternal age and the discharge time or in-hospital stay. These findings demonstrate the significance of maternal age in influencing the recovery time of NRDS after mechanical ventilation, but the correlation was not strong enough to affect the whole disease course. Conversely, the highest heart rate was significantly different between neonates with a short and long in-hospital stay but not between those with a short and long discharge time, and there was no linear correlation between the highest heart rate and the discharge time or in-hospital stay. These findings illustrate the significance of the highest heart rate in the whole disease course, but the correlation was not strong enough to affect the recovery of newborns with ARDS. The heart rate was also associated with the anaerobic condition: a worse anaerobic condition was associated with a higher heart rate and longer in-hospital stay.
We chose PaO 2 and PaCO 2 as the most important indices because they showed significant differences in both discharge time groups and in-hospital stay groups, and they were both linearly correlated with the recovery time of NRDS after mechanical ventilation and the whole disease course. The levels of PaO 2 and PaCO 2 have already been proven important in adult ARDS. A decreased PaO 2 level and/or increased PaCO 2 level intuitively reflects the severity of ARDS, forming the basis for the classification of respiratory failure [16]. However, there is still a lack of a quantitative standard with which to measure the severity of NRDS and to guide further treatments for newborns. Therefore, we evaluated the proper levels of PaO 2 and PaCO 2 for the optimal initiation time of mechanical ventilation with the goal of informing clinicians of the prognostic benefits that are obtained for newborn patients by shortening the average length of stay and reducing adverse outcomes.
In the present study, we detected a linear correlation of the PaO 2 and PaCO 2 levels with the discharge time and in-hospital stay. We then analyzed and calculated the best cutoff level of PaO 2 (52.5 mmHg) and PaCO 2 (45.5 mmHg). This is the first study to determine specific cutoff levels of PaO 2 and PaCO 2 for newborn patients with ARDS and to give direct suggestions regarding the proper time to initiate mechanical ventilation for better recovery. If the PaO 2 level of a newborn patient decreases to 52.5 mmHg or the PaCO 2 level increases to 45.5 mmHg, mechanical ventilation is strongly recommended. However, if the newborn patient's condition becomes worse than this anaerobic condition, mechanical ventilation should be required. These suggestions will guide clinicians to improve the treatment of patients with NRDS and to reduce the deterioration of the disease. Improvements in treatment will reduce not only the in-hospital and discharge times but also the hospitalization expenses and even the anxiety level of the patients' parents. The present study may also promote the development of guidelines for the standard treatment of NRDS. The data collected and analyzed in this study may be able to provide necessary support and evidence when composing related guidelines and to help more patients attain better outcomes.
In the present study, the APACHE III score showed some between-group differences. The APACHE III score facilitates a complex and comprehensive assessment of many disease-related parameters. This score has mostly been used for adult patients, and it is not highly accurate when applied to newborn patients. The differences in surfactant treatment, vasoactive agents, and a plateau pressure of invasive ventilation indicate that patients who require a longer time for recovery need more medication and better supportive treatments.
The present study was limited by its small sample size and lack of long-term follow-up data. We plan to continue our work in newborn ARDS and perform more studies, especially prospective studies in multiple medical centers, to obtain more clinical data and evidence and to improve treatment for newborn patients with ARDS. Besides, since permissive hypercapnia is well tolerated nowadays in the early treatment of severe ARDS to avoid excessive mechanical ventilation, the cutoff values of PaO 2 and PaCO 2 obtained in this study should not be the solitary criteria for deciding whether to start invasive mechanical ventilation but should be incorporated into a comprehensive algorithm. In addition, socio-economic factors including the economic situation of the infants' family and the management of bed use, as one of the confounders that can affect the length of hospitalization, were hard to be accurately quantified and afterward led to inevitable biases.

Conclusions
PaO 2 and PaCO 2 were closely related to discharge time and in-hospital stays in NRDS patients. The recommended cut-off value of PaO 2 and PaCO 2 for invasive mechanical ventilation were 52.5 and 45.5 mmHg. Invasive mechanical ventilation has been recommended in previous literatures. This study provided more detailed evidence for future practical guidelines. These suggestions will guide therapeutic improvement and may help promote the overall prognosis for NRDS.