Evaluating the diagnostic accuracy of smartphone video clips against polysomnography for paediatric obstructive sleep apnoea.

Abstract

Background: The gold standard of paediatric obstructive sleep apnoea syndrome (OSAS) diagnosis is overnight polysomnography (PSG). Validated screening methods for paediatric OSAS are limited. Increasingly, concerned parents record videos that capture behaviours indicative of OSAS. We aimed to compare the diagnostic accuracy of a standardised sleep video analysis tool (Videosomnography Analysis for Paediatric Sleep Apnoea (VAPS)) against PSG for paediatric OSAS.

Methods: Children aged 2-12 years consecutively referred for suspected OSAS underwent PSG. Parents recorded two 1 min videos of their child's sleep during PSG, and for two nights at home post-PSG. Videos were scored for the presence of nine behaviours indicative of OSAS. PSG was scored independently and compared with VAPS scores.

Results: Total VAPS Score had good discriminatory ability between PSG-diagnosed clinically non-significant (n=63) and clinically significant (n=48) OSAS, with an area under the receiver operating characteristic (ROC) curve of 81.99%. Scores of ≥3 had a sensitivity of 92% (specificity=49%) and scores of ≥7 had a specificity of 89% (sensitivity=40%). The area under the ROC curve rose to 85.45% when parents confirmed they were able to capture their child's worst breathing. For this subset, VAPS Score of ≥3 had a sensitivity of 96% (specificity=44%) and scores≥7 had a specificity of 89% (sensitivity=50%).

Discussion: VAPS Score may be useful in triaging patients suspected of paediatric OSAS to the appropriate physiological sleep investigations or treatment options, where a score of ≤2 is considered low risk, 3-6 medium risk and ≥7 high risk, for paediatric OSAS.