This study was a retrospective, single-center cohort study of preterm infants admitted to the neonatal intensive care unit (NICU) in the First Affiliated Hospital of Anhui Medical University between 1 July 2019 and 31 March 2021. All infants were born in the Obstetrics Department of the same hospital and transferred to NICU immediately after birth or recovery. Other inclusion criteria were as follows: (1) gestational age (GA) was less than 32 weeks, (2) birth weight (BW) was lower than 1,500 g, and (3) preterm infants had a survival time ≥28 days. The exclusion criteria were as follows: (1) infants had a congenital malformation of vital organs or genetic/metabolic diseases, (2) infants had culture-proven sepsis, (3) infants had stage IIA or higher NEC, and (4) the clinical information was incomplete. The Ethics and Research Committees of the First Affiliated Hospital of Anhui Medical University approved this study (Quick-PJ 2021-05-33).

Information on the demographics, perinatal characteristics, treatments during hospitalization, and complications after birth was collected from medical records. The demographics included GA, BW, gender, assisted reproduction, multiple births, cesarean delivery, maternal age, and score for clinical risk index for babies (CRIB)-II. The perinatal characteristics included pregnancy complications (gestational diabetes mellitus, hypertensive disorders of pregnancy, prolonged rupture of membranes, antenatal steroids, 5 min Apgar score, and intubation in the delivery room). Treatments during hospitalization included pulmonary surfactant, caffeine, and duration of mechanical ventilation. Complications after birth included BPD, RDS, and patent ductus arteriosus. Clinical characteristics required to calculate the nSOFA score included the peripheral oxygen saturation; fraction of inspired oxygen to achieve the peripheral oxygen saturation; intubation and mechanical ventilation; any requirement for glucocorticoid, inotropic, or vasoactive drugs; and the latest platelet count.

The nSOFA score was calculated after admission to the NICU within the first 72 h after delivery, and the peak value was used for the calculation. The nSOFA system uses categorical scores with a total score ranging from 0 as best to 15 as worst, which were used to objectively describe dynamic changes in respiratory deterioration (0–8), hemodynamic compromise (0–4), and hematologic dysfunction (0–3) (15) (Table 1).

The study outcome was the presence or absence of BPD. BPD was determined according to the National Institutes of Health consensus definition (18) and defined as required supplemental oxygen for ≥28 days.

Circulatory support was based on the potential pathophysiology of circulatory system damage, mean blood pressure, laboratory, clinical signs of hypoperfusion, and bedside echocardiography. Dopamine was the most common positive inotropic drug. Dobutamine was the most popular second-line treatment when the initial inotrope therapy failed. If the above treatment was ineffective, adrenaline or norepinephrine was used as the third-line treatment. When refractive hypotension or severe blood loss/asphyxia was present and affected adrenal perfusion, corticosteroids were used for the treatment (19, 20).

Given that infants were not randomly assigned to receive antibiotics, propensity scoring was used to control for potential confounding. BPD cases were matched with controls at 1:1 based on GA, BW, gender, assisted reproduction, multiple births, cesarean delivery, maternal age, and CRIB-II score. Patients were matched on the logit of propensity score using optimal matching with a caliper width of 0.02. Matching was performed without replacement.

Descriptive statistics were calculated for demographics and clinical characteristics between the original and matched populations. Values are presented as medians and interquartile ranges for continuous variables and as numbers and percentages for categorical variables. The continuous variables were compared using the Mann–Whitney test, and the categorical variables using the Chi-squared test.

The accuracy of the total nSOFA and component nSOFA scores in predicting BPD was calculated by the area under the receiver operating characteristic curve (AUROC). The cutoff value of the receiver operating characteristic (ROC) curve was determined by calculating the Youden index, which divided the population into the high nSOFA and low nSOFA groups.

Logistic regression models were used to evaluate the relationship between the nSOFA score and BPD in the original and matched populations. Model 1 was unadjusted, model 2 was adjusted for GA and BW, and model 3 was adjusted for GA, BW, CRIB-II score, 5 min Apgar score, intubation in the delivery room, and mechanical ventilation >7 days. A subgroup analysis was performed to examine the relationship after modifying for gender, mode of delivery, and conception.

Statistical analyses were performed using the statistical software packages R 3.3.2 (http://www.R-project.org, The R Foundation) and Free Statistics software version 1.4 (21). A two-sided P-value of <0.05 was considered statistically significant.

During the study period, 376 premature infants had GA less than 32 weeks and BW less than 1,500 g. Infants with congenital malformation of vital organs (n = 11), genetic or metabolic diseases (n = 10), and culture-proven sepsis (n = 46) were excluded. In addition, infants who had stage IIA or higher NEC (n = 18), had incomplete clinical data (n = 37), and died before 28 days after birth (n = 16) were also excluded. The remaining 238 infants were included in our analysis, of which 93 were diagnosed with BPD and 145 were free of BPD. Figure 1 shows the study flowchart. Before propensity score matching, compared with the low nSOFA group, the GA was smaller, the BW was lower, and the CRIB-II score was higher in infants with high nSOFA scores (all P < 0.001). After propensity score matching, there were no significant differences in the GA, BW, gender, multiple births, cesarean delivery, assisted reproduction, CRIB-II score, and maternal age (Table 2).

The AUROC of the nSOFA score in predicting BPD was 0.790 [95% confidence interval (CI): 0.731–0.849], which was higher than the predictive efficacy of the component score: the AUROC of the respiratory score was 0.751 (95% CI: 0.693–0.809), the AUROC of the cardiovascular score was 0.607 (95% CI: 0.557–0.656), and the AUROC of the hematologic score was 0.679 (95% CI: 0.618–0.741) (Figure 2). The optimal cutoff value of the nSOFA score for BPD in preterm infants was 1.5, which divided the population into the high nSOFA (n = 100) and low nSOFA (n = 138) groups.

A regression analysis was performed to identify factors associated with BPD. In the unadjusted model, the nSOFA score was significantly associated with the risk of BPD (OR = 1.9, 95% CI: 1.57–2.3). In the multivariate regression model, after adjusting for GA and BW, the odds ratio was 1.74 (95% CI: 1.43–2.12). After adjusting for GA, BW, CRIB-II score, 5 min Apgar score, intubation in the delivery room, and mechanical ventilation >7 days, the odds ratio was 2.09 (95% CI: 1.57–2.76). In addition, the high nSOFA score significantly increased the risk of BPD as compared with the low nSOFA score (aOR = 6.36, 95% CI: 2.73–14.79) (Table 3). Propensity score matching showed that the relationship between the nSOFA score and BPD remained stable (Table 3).

Stratified analyses were performed to evaluate whether the association between the nSOFA score and BPD varied by gender, mode of delivery, and conception. Results showed that none of the subgroups significantly affected this association (Figure 3).