A proposed classification of peristomal tracheal pathology in children with long-term tracheostomies.

OBJECTIVE: There is currently no classification module for predicting decannulation and comparing and choosing among various treatment modalities for peristomal tracheal pathologies. Our aim is to propose such a classification for application in clinical practice and to evaluate its interobserver and intraobserver reliability. METHODS: Videos of 28 rigid tracheal endoscopies of children with long-standing tracheostomies were included in this retrospective study. Ten repeat videos were also included. The 38 videos were then viewed individually by 6 pediatric otolaryngologists from 4 referral centers who scored them with the proposed scoring system. There were 5 scoring domains: 1. Estimate of overall obstruction, 2. Granulation tissue involvement, 3. Tracheal stenosis involvement, 4. Tracheomalacia, and 5. Lateral wall involvement. RESULTS: Kendall's coefficient for concordance between raters was very good (0.847, 95% CI: 0.787-0.901) for Percentage of obstruction estimate (condition 1), moderate (0.575, 95% CI: 0.514-0.647) for Granulation tissue involvement (condition 2) and fair for Tracheal stenosis involvement, Tracheomalacia component involvement and Lateral wall involvement (conditions 3, 4, and 5). The intraobserver agreement for conditions 1, 2 and 5 was very good (0.866, 0.857 and 0.805, respectively), good for condition 3 (0.704) and only moderate for condition 4 (0.576). CONCLUSION: Agreement was excellent for overall airway obstruction and, for granulation tissue involvement, moderate between observers but very good within observers. Agreement for tracheomalacia, tracheal stenosis and lateral wall involvement was only fair, and these domains need refinement. Prospective outcome studies are required to test predictive value. LEVEL OF EVIDENCE: 2b.