Bronchopulmonary dysplasia predictor scale validation in preterm newborns in two neonatal units at 2600 m above sea level.

Introduction: Bronchopulmonary dysplasia (BPD) is common in premature babies. It is difficult to predict the risk of developing this disease and its definition is not well characterized in high altitude cities.

Objective: To determine the operating characteristics of a BPD predictive scale for in preterm infants based on a classical BPD definition and one adjusted for altitude.

Materials And Method: Observational, analytical, longitudinal study with infants gestational age ≤32 weeks admitted at two University Hospital's in the city of Bogotá, Colombia between June 2010 and December 2016. The 2001 NIH consensus definition for BPD was used and a definition with an altitude adjustment. Perinatal data, and the Romagnoli scale variables for the 3rd and 5th days were described. Score operational characteristics in relation to BPD frequency are measured using the two definitions.

Results: 335 patients were included. The median birth weight was 1335 g and GA was 31 weeks. For the BPD classical definition, the incidence was 85%, whereas after adjusting for altitude it was 20%. The scale score showed good overall discrimination (ROC curve, AUC 0.81-0.86). The sensitivity for the classical definition with a high cut point of 4 was between 93-98% and the adjusted for altitude was 28-37%. The specificity for the same cut was 28-32% with the classical definition and 96% with the adjusted one.

Conclusions: The scale using the classical definition has high sensitivity, but it is not specific; adjusting the definition for altitude decreases sensitivity but increases specificity. New cutoffs points are needed on the scale or a change in the weight for the variables included in the model in order to be used in our high-altitude population.