2010 Volume 129 Pages 41-47
We compared pulse oxymeters and SleepStrips in sleep apnea syndrome (SAS) screening. We conducted pulse oxymeter (Pulsox-M24 and Me300, Teijin Inc., Tokyo, Japan) and SleepStrip (SLP Ltd., Tel Aviv, Israel) simultaneously at the homes of 85 subjects with suspected sleep apnea syndrome recruited from August 2006 to August 2007. Polysomnography (PSG) was later conducted with Alice4 (Respironics Inc., Murrysville, PA, USA) at the hospital. The 3% oxygen desaturation index (ODI3) with pulse oxymeter and the SleepStrip score based on the respiratory disturbance index (RDI) were calculated by automatic analysis. The apnea hypopnea index (AHI) in PSG was calculated by manual scoring. The correlation coefficient of ODI3 and RDI to PSG AHI was 0.802 (p<0.0001) and 0.557 (p<0.0001). A difference was seen in the Bland-Altman plot between ODI3 and AHI, and the ratio of ODI3 to AHI was underestimated. The area under the curve (AUC) of the SleepStrip was <0.8 and that of the pulse oxymeter was ≥0.9, at each cutoff. Agreement rate, over-diagnosis and under-diagnosis of these to PSG were 45.9% vs 70.6%, 34.1% vs 25.9%, and 20.0% vs 3.5%. Although the best cutoff from receiver operating characteristic curves (ROC) of SleepStrip are RDI≥25, AUC<0.8, the positive and negative likelihood ratios (LR+, LR-) were 3.2 and 0.4. These results indicate that SleepStrips are not effective in SAS screening. The best pulse oxymeter cutoff was 7.2, sensitivity was 0.914, and specificity was 0.815. Positive LR was 4.93 and negative LR 0.106, so pulse oxymetry can be used effectively to rule out SAS. Although pulse oxymetry was more effective in screening than SleepStrip, its use involved a difference and underestimation between ODI3 and AHI. We concluded that both devices were inadequate in home screening for SAS.
