Mandibular Distraction Osteogenesis as Paradigm Shift for Treatment of Tongue-Based Obstructive Apnea: Results of a 16-Year, Hospital-Wide Intervention Trial.

Han, Nicholas A; Tolley, Philip D; Massenburg, Benjamin B; Ryan, Isabel A; Hu, Allison C; Bartlett, Scott P; Taylor, Jesse A; Napoli, Joseph A et al. · Plast Reconstr Surg · 2026

retrospective_cohort · Level III

Where this comes from

Abstract

Mandibular distraction osteogenesis (MDO) has emerged as the preferred surgical treatment for neonatal tongue-based airway obstruction (TBAO), yet comprehensive outcomes data comparing surgical and non-surgical approaches remain limited. We conducted a retrospective cohort study of 579 patients with congenital micrognathia from 2007-2023 at a single tertiary pediatric care institution according to STROBE guidelines, with 344 patients having confirmed TBAO managed in the neonatal period. Patient demographics, treatment modalities, and outcomes were analyzed. A Random Forest machine learning model was developed to predict MDO failure, defined as subsequent tracheostomy or death. Of 344 patients with TBAO, 267 (77.6%) underwent surgical intervention: 189 (70.8%) MDO, 53 (19.9%) tracheostomy, and 25 (9.4%) tongue-lip adhesion. Seventy-seven patients (22.4%) were managed non-surgically, primarily with continuous positive airway pressure (CPAP) therapy. Following institutional implementation of MDO in 2011, tracheostomy rates decreased significantly from 39.6% to 15.7% (p<0.001). Among non-surgical patients receiving CPAP, obstructive apnea-hypopnea index normalized from 26.3±26.6 to 4.5±4.5 events/hr (p<0.001), though treatment duration averaged 570 days. MDO failure occurred in 21 patients (11.1%), with neurologic disorders (OR 7.61, p=0.001) and lower airway pathology (OR 11.72, p<0.001) as strongest predictors. The Random Forest model predicted MDO success with superior accuracy compared to the GILLS score (90.0% vs 70.0%). MDO implementation achieved high success rates while reducing tracheostomy rates. Non-surgical management with CPAP achieved airway normalization in selected patients despite prolonged treatment duration. Machine learning models outperformed traditional scoring systems, identifying neurologic disorders and lower airway pathology as key failure predictors to guide surgical decision-making.