A new radiographic scoring system for bronchopulmonary dysplasia. Newborn Lung Project

Abstract

Objectives: To develop a simple, clinically meaningful radiographic score for bronchopulmonary dysplasia (BPD). To investigate its reliability, validity, and usefulness and to compare it to the Edwards score.

Working hypothesis: Our radiographic scoring of BPD is reliable, correlates with respiratory support, and provides a necessary standardization in comparing severity of respiratory disease between hospitals.

Study design: Prospective cohort study.

Patient selection: The study included all neonates (n = 366) with birth weight below 1501 g admitted to 7 neonatal intensive care units, who had chest radiographs taken at age 25-35 days.

Methodology: A simple radiographic scoring system was developed. Scores ranging from 0 to 6 were assigned based on standard radiographs and descriptors of degree of abnormality. All radiographs taken between days 25 and 35 of age (n = 1087) were graded by a radiologist and a neonatologist. Radiographs from a stratified random sample of 37 neonates (10%) were also scored by the method of Edwards (n = 128 radiographs). A respiratory support index was constructed for days 25-35 and correlated with the radiographic score.

Results: Between-reader correlation was r = 0.87 for our score and r = 0.88 for the Edwards score. The two scores correlated with each other at r = 0.94. The respiratory support index correlated with our radiographic score at r = 0.75 overall, and r = 0.56 to 0.88 within hospitals. Higher postnatal corticosteroid use was found at the hospitals with the lower correlations.

Conclusions: Our radiographic scoring is reliable, valid, and gives results similar to the Edwards score. Radiographs play a standardizing role in assessing severity of BPD between hospitals.