Hour-to-hour variability of oxygen saturation in sleep apnea

Study objectives: Methods used to express the severity of oxygen desaturation during polysomnography include the average oxygen saturation (AO(2)), lowest oxygen saturation (LO2), and the percent of the total time with oxygen saturation level lower than 90% (T<90%). We wanted to determine which one of these methods is least variable during different hours of monitoring. Design: Prospective, observational study. Setting: Sleep center at a medical university. Patients: One hundred fifty patients with apnea-hypopnea index from 5 to 130. Measurements: AO(2), LO2, and T<90% were calculated during each of the 8 h of polysomnography. Data for each hour were compared and the Cronbach alpha coefficients were calculated. Results: There was a high degree of correlation among the three methods as well as between each method and the severity of sleep apnea. The mean+/-SD values for each method were as follows: AO(2), 92.7+/-5.6; LO2, 68.5+/-19.3; and T<90%, 15.7+/-24.2. The alpha coefficients for these methods were AO(2), 0.98; LO2, 0.88; and T<90%, 0.98. In all methods, the data of the first hour were significantly different from the data of the subsequent hours. Conclusion: Both AO(2) and T<90% methods show less hour to hour variability compared with LO2, and there is more variability in the first hour. Since the AO(2) values >90% may not convey the severity of O-2 desaturation, T<90% may be the best method of expressing oxygen saturation changes during polysomnography.