Development and validation of a bronchoscopically defined bronchitis scoring tool in children
Kah Peng Eg, Rahul J. Thomas, Ian Brent Masters, Margaret S. McElrea, Julie M. Marchant, Anne B. Chang
Abstract
INTRODUCTION/AIM: A validated tool for scoring bronchitis during flexible bronchoscopy (FB) is potentially useful for clinical practice and research. We aimed to develop a bronchoscopically defined bronchitis scoring system in children (BScore) based on our pilot study. METHODS: Children undergoing FB were prospectively enrolled. Their FB was digitally recorded and assessed (two clinicians blinded to each other and clinical history) for six features: secretion amount (six-point scale), secretion color (BronkoTest, 0-8), mucosal oedema (0-3), ridging (0-3), erythema (0-3), and pallor (0-3) based on pre-determined criteria. We correlated (Spearman's rho) each feature with bronchoalveolar lavage (BAL) neutrophil percentage (neutrophil%). BScore was then derived using models with combinations of the six features that best related to airway BAL neutrophil%. The various models of BScore were plotted against BAL neutrophil% using receiver operating characteristic (ROC) curves. RESULTS: We analyzed 142 out of 150 children enrolled. Eight children were excluded for unavailability of BAL cytology or FB recordings. Chronic/recurrent cough was the commonest indication for FB (75%). The median age was 3 years (IQR, 1.5-5.3 years). Secretion amount (r = 0.42) and color (r = 0.46), mucosal oedema (r = 0.42), and erythema (r = 0.30) significantly correlated with BAL neutrophil%, P < .0001. The highest area under ROC (aROC) was obtained by the addition of the scores of all features excluding pallor (aROC = 0.84; 95% CI, 0.76-0.90) with airway neutrophilia (defined as BAL neutrophil% of >10%). CONCLUSION: This prospective study has developed the first validated bronchitis scoring tool in children based on bronchoscopic visual inspection of airways. Further validation in other cohorts is however required.