Reply to: More correlates – not predictors – of hypoglossal nerve stimulation effectiveness

Extract We thank R.C. Dedhia and B.T. Keenan for their thoughtful engagement with our work around what constitutes prediction in clinical research. In brief, our study demonstrated that patients with obstructive sleep apnoea (OSA) who have a particular pharyngeal site of collapse (oropharyngeal lateral wall collapse) – identified both by drug-induced sleep endoscopy and by our non-invasive airflow-shape method – have approximately a 20-percentage-point lower efficacy (percent reduction in apnoea–hypopnoea index) with hypoglossal nerve stimulation (HGNS) and three- to six-fold lower odds of achieving a successful treatment response [1]. Subgroup differences were mechanistically plausible, of clinically meaningful magnitude, and reproducible, supporting the potential for precision medicine.