Diagnostic performance of nocturnal oximetry in the detection of obstructive sleep apnea syndrome: a Brazilian study.

OBJECTIVE: Due to the increasing prevalence of obstructive sleep apnea (OSA), more practical diagnostic methods than polysomnography (PSG) have become necessary. This research aims to analyze the performance of nocturnal oximetry (NO) in the diagnosis of OSA. METHODS: In this cross-sectional study, we analyzed 41 variables provided by the oximetry of all PSG performed by the LabSono of University Hospital Gaffrée and Guinle, a total of 83 exams. We evaluated the correlation coefficients (Spearman) between these data and the Apnea/Hypopnea Index (AHI) and then calculated the diagnostics performances, by the area under the curve (ROC) (AUC), of the best correlated variables and their respective cutoffs, in the identification of an AHI ≥ 15/h. RESULTS: Virtually all oximetric data showed good correlations with AHI, except for some temporal data. We chose 5 of them and calculated their diagnostic performances. T < 90% shows AUC of 0.904 (0.835-0.972) and, at cutoff > 19 min, a sensitivity (Sens.) of 75.68% and specificity (Spec.) of 95.65%. DO3/10Total, AUC 0.936 (0.888-0.989), and at the cutoff > 51 has Sens. 97.3% and Spec. 76.09%. ODI3/10/h has AUC 0.932 (0.884-0.988), at the cutoff > 7/h, Sens. 97.3% and Spec. 78.26%. DO4/5Total AUC 0.932 (0.882-0.981), at the cutoff > 64 has Sens. 86.49% and Spec. 82.61% and ODI4/5/h has AUC 0.930 (0.880-0.981), the cutoff > 5.69/h Sens. 97.3% and Spec. 73.91%. CONCLUSIONS: Our study concludes that NO is accurate in identifying AHI > 15/h, and provides reliable information on PSG replacement, which could make the diagnosis of OSA cheaper and more comfortable.