This retrospective observational study was conducted in the third level (NICU) at Fondazione IRCCS Ca’ Granda Ospedale Maggiore Policlinico, Milan, Italy.
All procedures were performed in accordance with the Helsinki Declaration. The study was approved by the local Ethics Committee (Milan Area 2, Italy) with approval number 951_2022. Due to the study’s retrospective nature, our Institutional Review Board has waived the need for informed consent from the parents.
We retrospectively collected data from all premature infants born under 32 weeks of GA and/or less than 1500 g of BW hospitalized in our NICU between January 2010 and December 2020. We excluded patients with incomplete data on hyperbilirubinemia or ROP status.
Through the electronic patient charts NeoCare® (GPI S.p.A., Trento, Italy), we collected the following data on the prenatal, perinatal, and neonatal period: GA, BW, sex, delivery characteristics, antenatal steroid administration, maternal comorbidities, APGAR score, resuscitation, death, RDS, severe IVH (greater than grade II), BPD, NEC, hemodynamically significant patent ductus arteriosus (HsPDA), early-onset sepsis (EOS), late-onset sepsis (LOS), ventilatory management, oxygen supplementation, number of red blood cell (RBC) transfusion and exchange transfusions, presence and length of hyperbilirubinemia, use and length of phototherapy, and ROP.
Chorioamnionitis was defined as maternal fever ≥ 38° C, maternal/fetal tachycardia (> 100/160 bpm), uterine tenderness, and purulent fluid from the cervical os [30]. Preeclampsia was identified by new gestational hypertension after 20 weeks gestation, with systolic pressure ≥ 140 mmHg or diastolic pressure ≥ 90 mmHg twice, 4 hours apart, or severe spikes (systolic ≥ 160 mmHg, diastolic ≥ 110 mmHg) [31]. Small for gestational age (SGA) was defined as a birth weight below the 10th percentile for GA and sex [32]. RDS was defined by the combination of clinical and radiographic findings [33]. IVH was diagnosed by brain ultrasound and classified according to Papile et al. [34]. BPD was defined and classified according to Jobe and Bancalari [35]. NEC was diagnosed by clinical and radiographic findings and categorized by Bell staging [36; 37]. HsPDA was defined by clinical and echocardiographic signs [38–41]. EOS and LOS were defined as the positivity of blood cultures with clinical and laboratory signs of infection, respectively, before and after 72 hours of life [42–45]. Hyperbilirubinemia was diagnosed when the total serum bilirubin (TSB) level was above the threshold for phototherapy treatment, according to the National Institute for Health and Care Excellence (NICE) nomograms [46]. All infants in the study underwent ROP screening, and ROP was classified according to ICROP. Non-severe ROP was defined as stages 1 and 2, while severe ROP was defined as stage 3 or higher [47].
Statistical analysis
Demographic characteristics of all infants were presented with descriptive statistics. Mean (standard deviation [SD]) and median (interquartile range [IQR]) were used for normal and non-normal continuous variables, and the two groups were compared using an independent t-test and Mann-Whitney U test, respectively. Absolute frequency (percentage) and χ2 test were used to summarize and compare categorical variables. Mann-Whitney U test was used to compare the distribution of days of hyperbilirubinemia in the no ROP and any ROP groups. Kruskal-Wallis test was used to compare the three ROP groups (no ROP, non-severe, and severe ROP) with post hoc pairwise multiple comparisons performed using Dunn’s test. The association between ROP and length of hyperbilirubinemia was studied using the Poisson regression model with a robust error variance. Crude and adjusted risk ratio (RR), 95% confidence intervals (CI), and p-values were shown. Considered adjustment confounding variables were GA, sex, and SGA.
Additionally, data were analyzed for the overall study population, and sub-analyses were performed for the subgroup that only included GA < 27 weeks and BW < 1000 g. R statistical software version 4.2.0 was used for all analyses [48]. All p-values < 0.05 were considered statistically significant.