Can standard deviation of overnight pulse oximetry be used to screen childhood obstructive sleep apnea

摘要： Objectives: Pulse oximetry (PO) has been frequently used as an alternative test to polysomnography (PSG) in children. We conducted this study to determine which statistical parameters obtained from overnight PO monitoring would be most suitable and to evaluate its diagnostic performance. Methods: We prospectively recruited children with snoring referred for PSG. Subjects were monitored with PO while performing PSG. Eight statistical parameters of SpO2 data were analyzed to identify which had the best diagnostic performance as assessed by the area under the receiver-operating characteristic curve (AUC). To validate this parameter (which was found to be the standard deviation, SD) in a larger population, we retrospectively extracted raw data of SpO2 from our previous PSG records, calculated the SD of each patient, and assessed its AUC. Results: A total of 166 children were recruited in the first phase. SD of SpO2 was found to have the largest AUC. In the second phase, raw data of 457 patients were extracted. SD of SpO2 correlated well with the apnea/hypopnea index (AHI) (r=0.6, P<0.001). For diagnosis of moderate to severe obstructive sleep apnea (OSA) (AHI ≥5 events/h), AUC was 0.74. SD of SpO2 ≥1.06 representing mean + 2SD of normal to mild OSA (AHI <5) provided 97% specificity and 92% positive predictive value. The positive likelihood ratio was 11. Conclusion: Calculating the SD of SpO2, which quantifies the amount of dispersion of SpO2 values, is a useful initial investigation in childhood OSA. An SD ≥1.06 can predict moderate to severe OSA with confidence. This parameter is simple, practical, and readily accessible.