Role of overnight oximetry in assessing the severity of obstructive sleep apnoea in typically developing children: a multicentre study

Cardiorespiratory polygraphy (CRP) is the predominant technology used to diagnose obstructive sleep apnoea (OSA) in tertiary centres in the UK. Nocturnal pulse oximetry (NPO) is, however, cheaper and more accessible. This study evaluated the ability of NPO indices to predict OSA in typically developing (TD) children.Indices from simultaneous NPO and CRP recordings were compared in TD children (aged 1-16 years) referred to evaluate OSA in three tertiary centres. OSA was defined as an obstructive apnoea-hypopnoea index (OAHI) ≥1 event/hour. Receiver operating characteristic curves assessed the diagnostic accuracy of NPO indices including ODI3 (3% Oxygen Desaturation Index, ODI4 (4% Oxygen Desaturation Index), delta 12 s index and minimum oxygen saturation. Two-by-two tables were generated to determine the sensitivities and specificities of whole number cut-off values for predicting OAHIs ≥1, 5 and 10 events/hour.Recordings from 322 TD children, 197 male (61.2%), median age 4.9 years (range 1.1-15.6), were reviewed. OAHI was ≥1/hour in 144 (44.7%), ≥5/hour in 61 (18.9%) and ≥10/hour in 28 (8.7%) cases. ODI3 and ODI4 had the best diagnostic accuracy. ODI3 ≥7/hour and ODI4 ≥4/hour predicted OSA in TD children with sensitivities/specificities of 57.6%/85.4% and 46.2%/91.6%, respectively. ODI3 ≥8/hour was the best predictor of OAHI ≥5/hour (sensitivity 82.0%, specificity 84.3%).Raised ODI3 and ODI4 predict OSA in TD children with high specificity but variable sensitivity. NPO may be an alternative to diagnose moderate-severe OSA if access to CRP is limited. Low sensitivities to detect mild OSA mean that confirmatory CRP is needed if NPO is normal.