Factors associated with the development of neonatal acute respiratory distress syndrome in preterm infants: a retrospective cohort study

Abstract

Neonatal acute respiratory distress syndrome (NARDS) carries high morbidity and mortality. Preterm infants at risk need to be identified early so that treatment is not delayed. We aimed to identify factors associated with NARDS in preterm infants. Clinical data of 548 preterm infants admitted to the NICU between 2019 and 2023 were retrospectively analyzed. Preterm infants were categorized into the NARDS group (n = 275) and the control group (n = 273) according to the Montreux definition. Univariate analysis and multivariable logistic regression were used to identify factors independently associated with NARDS; the results are shown in a forest plot. The study was reported in accordance with the STROBE guidelines. Six factors were independently associated with NARDS in preterm infants: male sex (OR = 1.895, 95% CI 1.051–3.420), premature rupture of membranes (PROM) (OR = 6.784, 95% CI 3.415–13.478), chorioamnionitis (OR = 11.143, 95% CI 5.460–22.742), cesarean delivery (OR = 2.104, 95% CI 1.125–3.935), higher PaCO2 (OR = 1.188 per mmHg, 95% CI 1.143–1.234), and lower gestational age (OR = 0.715 per week, 95% CI 0.587–0.871).

Conclusion: Male sex, lower gestational age, PROM, chorioamnionitis, cesarean delivery, and elevated early postnatal PaCO2 were independently associated with NARDS in preterm infants. All six are available in routine practice and could help clinicians flag preterm infants who need closer monitoring.

What is Known: • NARDS is a severe, recently defined form of acute respiratory failure in neonates, and evidence on its risk factors in preterm infants remains limited.

What is New: • Male sex, lower gestational age, premature rupture of membranes, chorioamnionitis, cesarean delivery, and elevated PaCO2 within 15 min of birth were independently associated with NARDS in preterm infants.

Supplementary Information

The online version contains supplementary material available at https://doi.org/10.1007/s00431-026-07457-6.

Introduction

Acute respiratory distress syndrome (ARDS) has long been recognized beyond the neonatal period, but neonates had no specific definition until the recent International Neonatal ARDS Consensus. Neonatal acute respiratory distress syndrome (NARDS) is now defined as acute respiratory failure with acute onset, bilateral pulmonary infiltrates on chest imaging, severe hypoxemia, and non-cardiogenic pulmonary edema [1, 2]. It can be triggered by direct inflammatory insults such as pneumonia and aspiration, or by indirect ones such as sepsis, necrotizing enterocolitis, and chorioamnionitis.

NARDS must be distinguished from respiratory distress syndrome due to primary surfactant deficiency (RDS), the classical respiratory disorder of preterm infants. The two conditions overlap clinically, but the Montreux definition offers practical guidance for telling them apart: unlike NARDS, RDS responds quickly and completely to surfactant replacement or alveolar recruitment maneuvers [1]. This matters most in very preterm infants, in whom the two conditions are most likely to coexist. Recent data indicate that NARDS affects approximately 1–5% of the general NICU population, with reported mortality rates of 10–19% [3]. Preterm infants make up a large share of these cases, and their prognosis remains poor [4]. Early identification and treatment of at-risk preterm infants therefore matter for both survival and long-term outcome [6–8].

Previous studies on respiratory distress syndrome due to primary surfactant deficiency (RDS) have focused mainly on prediction models [9–11], with study populations primarily comprising term or late-preterm neonates [12, 13]. In contrast, evidence on factors associated with neonatal acute respiratory distress syndrome (NARDS) in preterm infants remains limited. We therefore reviewed clinical data from preterm infants admitted to the neonatal intensive care unit (NICU) to identify factors associated with NARDS, with the aim of supporting earlier monitoring and treatment.

Methods

Study design and population

This retrospective cohort study was conducted at the Department of Neonatology, Yinchuan First People’s Hospital (Ningxia, China), a tertiary referral center for the surrounding region. From an initial screening of 649 preterm infants (gestational age < 37 weeks) admitted between October 2019 and December 2023, 548 preterm infants who met the eligibility criteria were included in the final analysis. Based on the Montreux criteria [1], 275 preterm infants were diagnosed with NARDS (NARDS group), while the remaining 273 non-NARDS preterm infants admitted during the same period were consecutively enrolled as controls; the prevalence of NARDS among the screened preterm infants was 42.4% (275/649). Exclusion criteria were as follows: (1) major congenital malformations or cyanotic congenital heart disease, (2) incomplete medical records or missing key laboratory data (e.g., initial arterial blood gas analysis), and (3) infants who were transferred or died before a definitive diagnosis could be established. The study was conducted and is reported in accordance with the Strengthening the Reporting of Observational Studies in Epidemiology (STROBE) guidelines.

The diagnosis of NARDS was established by the attending neonatologists according to the Montreux definition [1], which requires acute onset of respiratory distress (within 7 days of birth), bilateral pulmonary infiltrates on chest imaging, severe hypoxemia not fully explained by cardiac failure or fluid overload, and exclusion of respiratory distress syndrome due to primary surfactant deficiency and transient tachypnea of the newborn as the sole explanation for the respiratory distress. Because of the retrospective design, adjudication relied on the documented clinical, radiographic, and blood gas findings, and the assessors could not be blinded to the clinical exposures; the same criteria were applied to all infants to limit misclassification. Oxygenation indices at diagnosis (OI/OSI), the mild/moderate/severe distribution, and the use of surfactant and respiratory support were not retrievable for the entire cohort and are therefore not reported; this is acknowledged as a limitation.

This study was approved by the Ethics Committee of Yinchuan First People’s Hospital (Approval No.: 2025014). Written informed consent from the guardians of included neonates was waived by the Ethics Committee of Yinchuan First People’s Hospital due to the retrospective nature of the study, which posed no additional risks to the participants and used only de-identified clinical and laboratory data.

Data collection

Birth was defined as time zero; arterial blood gas samples were obtained within the first 15 min of life, and all clinical and laboratory data were extracted from the medical records. Chorioamnionitis was diagnosed clinically by the attending obstetricians according to standard criteria based on maternal intrapartum fever (≥ 38 °C) and associated signs of intra-amniotic infection (maternal or fetal tachycardia, purulent or malodorous amniotic fluid, or maternal leukocytosis) [31], as documented in the medical records; histopathological examination of the placenta was not routinely performed.

Statistical analysis

Continuous variables were assessed for normality using the Shapiro–Wilk test. Normally distributed data were expressed as mean ± standard deviation (SD) and compared between groups using the independent samples t-test; Levene’s test was employed to evaluate the homogeneity of variance, and Welch’s t-test was applied where variances were unequal. Non-normally distributed data were presented as median (interquartile range, IQR) and analyzed using the Mann–Whitney U test. Categorical variables were reported as frequencies (n) and percentages (%) and compared using the Chi-square test or Fisher’s exact test; for categorical variables, odds ratios (ORs) with 95% confidence intervals (CIs) were calculated. Variables with P < 0.05 in univariate analysis were entered into a multivariable logistic regression model, and backward stepwise selection was used to identify factors independently associated with NARDS. Multivariable results are reported as adjusted ORs with 95% CIs and displayed as forest plots. A two-sided P-value < 0.05 was considered statistically significant. All statistical analyses were performed using R software (version 4.3.1).

Results

During the study period, 548 preterm infants were enrolled: 275 with NARDS (NARDS group) and 273 without (control group). The demographic and clinical characteristics of the participants are summarized in Table 1. No significant differences were observed between the two groups regarding maternal age, gravidity, multiple pregnancy, fetal intrauterine distress, gestational diabetes mellitus (GDM), hypertensive disorders of pregnancy (HDP), or placental abruption (all P > 0.05). The groups differed significantly in infant sex, birth weight, 5-min Apgar score, neonatal asphyxia, cesarean delivery, chorioamnionitis, antenatal steroid (ANS) exposure, and premature rupture of membranes (PROM) (all P < 0.05; Table 1).

Table 1.

Comparison of general data between the two groups

	NARDS group (n = 275)	Control group (n = 273)	P	OR (95% CI)
Infant characteristics
Sex (male), n (%)	186 (67.6)	147 (53.8)	< 0.001	1.79 (1.27–2.53)
Birth weight (g), Median (IQR)	2060 (1530, 2500)	2485 (2050, 2820)	< 0.001	-
Gestational age (weeks), Median (IQR)	33.00 (31.00, 35.00)	35.00 (34.00, 36.00)	< 0.001	-
Apgar score at 5 min, Median (IQR)	9.00 (8.00, 10.00)	10.00 (9.00, 10.00)	< 0.001	-
Neonatal asphyxia, n (%)	95 (34.5)	43 (15.8)	< 0.001	2.823 (1.87–4.26)
Maternal characteristics
Maternal age (years), Median (IQR)	28.00 (24.00, 33.00)	28.86 (25.00, 33.00)	0.307	-
Gravidity, Median (IQR)	2.00 (2.00, 3.00)	2.00 (1.00, 3.00)	0.207	-
Delivery mode (cesarean), n (%)	174 (63.3)	148 (54.2)	0.039	1.455 (1.034–2.048)
Multiple pregnancy, n (%)	33 (12.0)	45 (16.5)	0.168	0.691 (0.426–1.121)
ANS, n (%)	89 (32.4)	65 (23.8)	0.033	1.531 (1.051–2.230)
PROM, n (%)	126 (45.8)	55 (20.1)	< 0.001	3.352 (2.292–4.848)
Fetal intrauterine distress, n (%)	45 (16.4)	33 (12.1)	0.190	1.423 (0.877–2.309)
GDM, n (%)	30 (10.9)	34 (12.5)	0.667	0.861 (0.511–1.451)
HDP, n (%)	19 (6.9)	24 (8.8)	0.509	0.770 (0.411–1.441)
Chorioamnionitis, n (%)	229 (83.3)	124 (45.4)	< 0.001	5.982 (4.025–8.891)
Placental abruption, n (%)	21 (7.6)	17 (6.2)	0.630	1.245 (0.642–2.415)

GA gestational age, ANS antenatal steroid, PROM premature rupture of membranes, GDM gestational diabetes mellitus, HDP hypertensive disorders of pregnancy, IQR interquartile range. Odds ratios (ORs) are only reported for categorical variables

Blood gas results differed between the two groups (Table 2). Acidemia (pH < 7.20) was more common in the NARDS group than in controls (28.4% vs. 5.1%, P < 0.001), and PaO2, PaCO2, hemoglobin, and base excess (BE) also differed significantly (all P < 0.05; Table 2). Platelet count and white blood cell count, routinely measured on admission as part of the complete blood count, were compared between groups to explore potential inflammatory differences; ORs are reported only for categorical variables, while continuous variables were compared using the Mann–Whitney U test or the t-test.

Table 2.

Blood gas analysis and routine blood test of NARDS neonates and controls

	NARDS group (n = 275)	Control group (n = 273)	P	OR (95% CI)
pH < 7.20, n (%)	78 (28.4)	14 (5.1)	< 0.001	7.325 (4.027–13.324)
PaO2 (mmHg), Median (IQR)	57.30 (46.00, 72.00)	65.40 (57.00, 77.30)	< 0.001	-
PaCO2 (mmHg), Median (IQR)	47.60 (41.00, 55.65)	34.50 (28.10, 40.60)	< 0.001	-
BE (mmol/L), Median (IQR)	− 6.00 (− 8.04, − 4.10)	− 4.80 (− 6.40, − 3.00)	< 0.001	-
PLT (× 10⁹/L), Median (IQR)	246.00 (206.00, 293.00)	254.00 (213.00, 303.00)	0.217	-
Hemoglobin (g/L), Median (IQR)	169.00 (155.00, 183.00)	177.00 (162.00, 190.00)	< 0.001	-
WBC, Median (IQR)	11.70 (8.60, 15.80)	11.70 (8.90, 16.10)	0.532	-

PaO2 partial pressure of oxygen, PaCO2 partial pressure of carbon dioxide, BE base excess, PLT platelet, WBC white blood cell count

Multivariable analysis

To adjust for potential confounders, multivariable logistic regression was performed. Fourteen variables with P < 0.05 in univariate analysis (Tables 1 and 2) were entered as candidates, and backward stepwise selection was applied. Male sex, gestational age, PROM, chorioamnionitis, cesarean delivery, and PaCO2 remained independently associated with NARDS (all P < 0.05; Table 3 and Fig. 1).

Table 3.

Multivariable logistic regression analysis of factors associated with NARDS in preterm infants

Variables	β	SE	P	OR (95% CI)
Sex (male)	0.639	0.301	0.034	1.895 (1.051–3.420)
PROM	1.915	0.350	< 0.001	6.784 (3.415–13.478)
Chorioamnionitis	2.411	0.364	< 0.001	11.143 (5.460–22.742)
Cesarean delivery	0.744	0.319	0.020	2.104 (1.125–3.935)
PaCO2 (per mmHg)	0.172	0.020	< 0.001	1.188 (1.143–1.234)
Gestational age (per week)	− 0.335	0.101	< 0.001	0.715 (0.587–0.871)

Fig. 1.

Forest plot of factors independently associated with NARDS

Discussion

We studied factors associated with neonatal acute respiratory distress syndrome (NARDS) in preterm infants born before 37 weeks of gestation. Our multivariable analysis identified male sex, gestational age, PROM, chorioamnionitis, cesarean delivery, and PaCO2 as factors independently associated with NARDS in this population. The results point to several contributing factors and may help identify the preterm infants most at risk.

Male sex was independently associated with NARDS in our cohort (OR 1.895, 95% CI 1.051–3.420), consistent with earlier reports. A propensity score-matched analysis found higher rates of RDS and moderate-to-severe BPD among very low birth weight male infants after adjustment for confounders [14]. Male preterm infants also needed more respiratory support than female infants, including conventional and high-frequency ventilation, post-CPAP intubation, inhaled nitric oxide, and rescue surfactant therapy [15]. Female sex has been associated with intact survival in extremely preterm neonates with early hypoxic respiratory failure [16], while male sex independently predicted mortality even among tracheostomized infants [17]. Recent longitudinal data from the UK also link male sex to worsening lung function trajectories [18]. Biological sex clearly influences respiratory outcomes in preterm infants. Sex differences in organ development extend to the lung, where fetal lung maturation and surfactant synthesis in particular follow sex-specific patterns [19]. Gonadal hormones appear to mediate these differences: androgens delay surfactant production and secretion, whereas estrogens accelerate lung maturation. This is a plausible mechanism for the male predominance seen in respiratory distress syndromes. The association we observed between male sex and NARDS may reflect these underlying developmental differences. In practice, male preterm infants may warrant closer respiratory monitoring.

Chorioamnionitis was the strongest factor we identified (OR 11.143, 95% CI 5.460–22.742); it was present in 83.3% of NARDS cases versus 45.4% of controls. The prevalence of chorioamnionitis in our cohort is higher than that reported in unselected populations, which may reflect the case mix of our referral NICU as well as possible misclassification, as the diagnosis was based on clinical criteria (maternal intrapartum fever and associated signs of intra-amniotic infection) rather than placental histopathology; this should be considered when interpreting the magnitude of the association. Chorioamnionitis at ≥ 34 weeks’ gestation has been associated with a 3- to fivefold increase in adverse neonatal outcomes and higher rates of respiratory distress [20], suggesting that intra-amniotic inflammation drives NARDS. Chorioamnionitis triggers systemic inflammation with increased endothelial permeability, which lets inflammatory cells and proteins leak into the alveolar spaces [21]. This vascular leakage worsens interstitial and alveolar edema and impairs gas exchange. Endothelial dysfunction also disturbs the balance between vasodilators and vasoconstrictors and creates a proinflammatory environment, with upregulated adhesion molecules, generation of reactive oxygen species (ROS), and transendothelial migration of immune cells. Cytokines such as IL-1, IL-6, and tumor necrosis factor-α are released in greater amounts and extend the lung injury [22]. These mechanisms all feed into the pathways thought to underlie NARDS [23–25]. Because PROM and chorioamnionitis lie on the same causal pathway, mutual adjustment may induce over-adjustment; however, the standard errors of both estimates remained small relative to their coefficients (SE 0.350 for PROM and 0.364 for chorioamnionitis), arguing against important multicollinearity, and both variables were retained given their independent contributions. Preterm infants exposed to chorioamnionitis should therefore be considered at high risk of NARDS and monitored closely.

PROM was also independently associated with NARDS in our multivariable analysis, which points to events upstream of the syndrome. PROM is a major route to intra-amniotic infection and complicates 5–10% of pregnancies [26]. It is easy to diagnose and rarely missed, so it could serve as a practical early warning sign for NARDS and allow antenatal intervention.

Elective cesarean delivery without signs of labor has been shown to increase the risk of RDS [27], and we also identified cesarean delivery as an independent risk factor in preterm infants. Cesarean delivery requires anesthesia, which can cause maternal hypotension, reduce uteroplacental perfusion, and lead to fetal hypoxia in utero [28]. Spontaneous labor also clears fetal lung fluid and promotes lung maturation. Infants born by cesarean without labor or passage through the birth canal miss the thoracic squeeze needed for pulmonary adaptation, which leaves fluid retained, impairs lung aeration, and favors fibrin deposition [29], predisposing preterm infants to NARDS.

Multivariable regression analysis identified PaCO2 as a significant independent factor associated with NARDS in preterm infants. The association likely reflects severe ventilatory impairment and increased physiological dead space early in the course. PaCO2 is both a hallmark of established respiratory failure and, in our analysis, an early associated factor, and this overlap deserves comment. In this study, arterial blood gas samples were obtained within the first 15 min of life, before the full clinical picture required by the Montreux criteria had necessarily developed. Immediate postnatal PaCO2 may therefore represent an early marker of lung immaturity or surfactant deficiency rather than established NARDS, and it correlates closely with the subsequent illness trajectory. Liu et al. reported significant positive associations between elevated PaCO2 and prolonged hospitalization as well as delayed post-ventilatory recovery, and derived a PaCO2 threshold of 45.5 mmHg as the optimal cutoff for initiating invasive mechanical ventilation [30]. Recognizing these ventilatory abnormalities early may help clinicians identify preterm infants at higher risk and escalate monitoring before the syndrome fully develops.

Gestational age was an independent risk factor for NARDS in our multivariable model (OR = 0.715, 95% CI 0.587–0.871). This finding is consistent with the physiological timeline of fetal lung development. During the canalicular and saccular stages of lung growth, alveolar type II epithelial cells are structurally and functionally immature, so synthesis, storage, and secretion of pulmonary surfactant (PS) are deficient. PS production only begins to rise substantially after 28–32 weeks of gestation. A lower gestational age therefore means higher alveolar surface tension and lower lung compliance, which raises the risk of alveolar collapse and NARDS. Our data are consistent with the view that each additional week of gestation is protective, as the surfactant system matures and the gas-exchange surface expands. Of note, antenatal steroid exposure was more frequent in the NARDS group (32.4% vs. 23.8%); this difference most likely reflects confounding by gestational age and illness severity, and the association should be interpreted with caution.

Multivariable analysis identified six factors associated with NARDS: male sex, gestational age, PROM, chorioamnionitis, cesarean delivery, and PaCO2. All six are available in routine practice and may help clinicians recognize preterm infants at higher risk and decide on monitoring, stabilization, or referral.

Several limitations should be noted. First, this is a retrospective single-center study; despite the consecutive enrollment of controls, selection bias cannot be excluded, and the findings may reflect the clinical practices and population demographics of our institution. Second, the prevalence of NARDS was 50.2% (275/548) among the included preterm infants and 42.4% (275/649) among all screened preterm infants, substantially higher than the 1–5% reported in the general NICU population [3]; this reflects the case mix of our referral NICU, which admits a large proportion of severely ill preterm infants. Consequently, the odds ratios reported here describe associations within our study population and should not be interpreted as individual absolute risks. Third, although the diagnosis of NARDS was made according to the Montreux definition by experienced neonatologists, the retrospective design precluded standardized, blinded adjudication, and some misclassification between NARDS and RDS due to primary surfactant deficiency cannot be excluded, particularly in very preterm infants in whom the two conditions may coexist. Fourth, 101 of the 649 screened infants were excluded, mainly because of incomplete records; although the principal reasons for exclusion are unlikely to be strongly related to the exposures studied, complete-case analysis may have introduced bias. Fifth, PaCO2 was measured at a single time point within 15 min of birth; the dynamic changes in blood gas parameters during the first hours of life might provide additional information but were not captured. Sixth, no severity-stratified analysis (mild, moderate, or severe NARDS according to oxygenation indices) was performed, and the long-term neurodevelopmental and pulmonary outcomes of the infants were not assessed. Prospective multicenter studies with standardized adjudication and severity stratification are warranted to confirm these associations.

Conclusion

In conclusion, this study identified six factors independently associated with NARDS in preterm infants: male sex, lower gestational age, PROM, chorioamnionitis, cesarean delivery, and elevated early postnatal PaCO2. All six are available in routine clinical practice, which may help clinicians identify preterm infants at higher risk of NARDS earlier and act sooner. Prospective multicenter studies are needed to confirm these associations.

Supplementary Information

Below is the link to the electronic supplementary material.

Authors’ contributions

Conceptualization: Lie HUANG, Xi XU Data curation: Xi TIAN Formal analysis: Xi Xu Resources:Xi TIAN Writing – original draft: Kehua HUANG, Yanling ZENG Writing – review & editing:Lie HUANG.

Funding

None.

Data availability

The datasets generated during and/or analyzed during the current study are available from the corresponding author upon reasonable request.

Declarations

Ethics approval and consent to participate

Ethics approval was obtained from the Ethics Committee of Yinchuan First People’s Hospital (Approval No.: 2025014). Given the retrospective design of this study, which involved the analysis of existing de-identified data without any intervention or additional risk to the participants, written informed consent from the guardians of the included neonates was formally waived by the Ethics Committee. All procedures were performed in accordance with the Declaration of Helsinki.

Consent for publication

Not applicable.

Competing interests

The authors declare no competing interests.