The Accuracy of Repeated Sleep Studies in OSA A Longitudinal Observational Study With 14 Nights of Oxygen Saturation Monitoring

BACKGROUND: Strong evidence exists for clinically relevant night-to-night variability of respiratory events in patients with suspected OSA. RESEARCH QUESTION: How many sleep study nights are required to diagnose OSA accurately? STUDY DESIGN AND METHOD: Patients with suspected OSA underwent up to 14 nights of pulse oximetry (PO) at home and one night of in-hospital respiratory polygraphy (RP). The accuracy of each of the 13 sleep study nights was analyzed using the mean oxygen desaturation index 3% (ODI3%) of all 14 nights as a reference. Multiple regression analyses assessed possible predictors for night-to-night variability. RESULTS: One hundred three patients underwent in-hospital RP. Using only the results of the RP, 19.7% were misdiagnosed using an ODI3% cutoff of 15/h. One hundred eight patients underwent properly performed PO studies at home with a coefficient of variation (CV) of 31.5% (SD, 14.7%) across all nights. The first PO night demonstrated a sensitivity of 71.4% (95% CI, 55.4%-84.3%) and a specificity of 89.4% (95% CI, 79.4%-95.6%) to diagnose moderate OSA. Using only the first PO night, the negative predictive value was 83.1%. Adding a second recording night increased sensitivity up to 88.1% (95% CI, 74.4%-96.0%) with a slightly lower specificity of 85.9% (95% CI, 74.9%-93.4%). The ODI3% of the in-hospital RP showed an independent negative association to the log-transformed CV (exponentiated coefficient, 0.989; 95% CI, 0.984-0.995). INTERPRETATION: One single night of in-hospital RP may miss relevant OSA. Multiple study nights, for example, using ambulatory oxygen saturation monitoring, increase accuracy for diagnosing moderate OSA.