3 Results
Six hundred seventy-two eligible newborns were identified in the Neonatal Database from 2016 to 2020. After analyzing the database, 371 newborns met the inclusion criteria.
Subsequently, 129 patients were excluded because they were discharged before 14 days of life, and one neonate was excluded due to pulmonary agenesis.
During the analysis of medical records, 13 patients were excluded from the outcome analysis and equation formulation due to their passing before 28 days of life. Additionally, there were no discharges recorded before the 28-day mark.
The mean birth weight of the sample was 922.6 grams (standard deviation (SD): 185 g), with patients with moderate BPD having 872 grams (SD: 175.83 g) and those with severe BPD having 802.6 grams (SD: 180.8 g), while the mean gestational age was 27.6 weeks (SD: 1.76), and those with moderate BPD having 27.2 weeks (SD: 1.47) and severe BPD 26.5 weeks (SD: 1.71).
Females were present in 54.8% of the total sample. Rates of antenatal corticosteroid use were observed in 87.9% and surfactant use in 66.8%.
At 28 days of life, among the 228 who survived, 31.1% did not require respiratory support, considered without bronchopulmonary dysplasia, 26.7% were considered with mild bronchopulmonary dysplasia, 18.8% with moderate bronchopulmonary dysplasia, and 23.2% with severe bronchopulmonary dysplasia.
Table 2 shows patients’ antenatal and postnatal characteristics according to BPD and their classifications and relative risks.
After analyzing the variables, we elaborated prediction equations for BPD using 70% of the sample, with 160 patients (training sample), in which four strategies were evaluated (Table 1). The equations were validated using the remaining 30% of the sample (68 patients), and the instrument currently used in that institution was also applied to these patients.
The risks in three patients could not be obtained in the analysis with the instrument currently used, as this calculator does not offer resources to calculate BPD in SGA patients. In this case, two patients had a birth weight of less than 501 g, and one was SGA (birth weight of 630 g and a GA of 29 weeks and four days).
Equation 4 failed to predict any positive results in the sample. However, the other equations showed specificities and positive predictive values superior to those of the instrument currently used in that institution, standing out Equation 2, with a specificity of 98% and a positive predictive value of 93% (Table 3).
The calculated Kappa indices showed low agreement between the developed equations and the current foreign instrument (Table 4).