Diagnostic accuracy of current glossoptosis classification systems: A nested cohort cross-sectional study.

OBJECTIVES/HYPOTHESIS: To assess the performance of endoscopic grading systems of glossoptosis in identifying severe clinical manifestations in children with Robin sequence (RS). STUDY DESIGN: Nested cohort cross-sectional study. METHODS: All RS patients diagnosed at Hospital Clinics of Porto Alegre from October 2012 to June 2016 were enrolled in this cohort. Patients underwent sleep endoscopy and were classified according to Yellon (Y) and de Sousa et al. (S) scales. Symptom severity evaluation was performed as defined by Cole et al. The outcome of interest was Cole's clinical classification grade 3. RESULTS: Eighty patients were eligible for analysis. Sensitivity (Y: 56.2%, S: 28.1%, P < .001) and specificity (Y: 85.4%, S: 93.8%, P = .038) in identifying severe clinical symptoms patients (i.e., Cole grade 3) were statistically different between Y and S classifications. A low but significant overall correlation was observed for both Y (rho = 0.372, P < .001) and S (rho = 0.439, P < .001) classifications when compared with Cole classification. Diagnostic odds ratio (DOR) for Y (DOR: 7.53, 95% confidence interval [CI]: 4.15-10.90) and S (DOR: 5.87, 95% CI: 1.86-9.87) were equivalent (P = .92). Also, receiver operating characteristic curves area under the curve were not significantly different between them. The positive likelihood ratio was 3.86 (95% CI: 1.82-8.16) and 4.50 (95% CI: 1.32-15.36) for Y and S, respectively. CONCLUSIONS: Y and S grading systems showed a low sensitivity and moderate to high specificity in detecting patients with severe clinical manifestations. Correlation between Y/S and Cole et al. grading were also considered low. Development of a more discriminative anatomic grading system is still needed for this specific disorder. LEVEL OF EVIDENCE: 2b. Laryngoscope, 128:502-508, 2018.