To ascertain the sensitivity and specificity of clinical diagnosis of obstructive sleep apnoea in children, and to determine if a published clinical algorithm identifies those at high risk of post-adenotonsillectomy complications.

Sixty-seven children aged three to eight years underwent clinical assessment and overnight polysomnography.

Polysomnography detected a significant apnoea–hypopnoea index (i.e. ≥5, indicating significant obstructive sleep apnoea) in 13 (43 per cent) children with a clinical diagnosis of obstructive sleep apnoea and in six (19 per cent) children with no such diagnosis. The sensitivity of clinical assessment was 68.4 per cent and the specificity 59.5 per cent. The post-operative risk algorithm failed to identify any high risk children, although in actuality seven had severe obstructive sleep apnoea confirmed by polysomnography.

This study of two English centres confirms that the clinical diagnostic process for obstructive sleep apnoea is reasonably insensitive and has low specificity. The studied algorithm discriminated poorly between children with and without severe obstructive sleep apnoea. Realistic diagnostic screening guidelines for paediatric sleep apnoea are overdue in the UK, where access to polysomnography is limited.