PURPOSE: Continuous positive airway pressure (CPAP) is indicated for treatment of obstructive sleep apnea (OSA). Titration is manual during level 1 polysomnography or by automatic CPAP devices (autoCPAP) where fixed CPAP is chosen from autoCPAP data, provided the residual apnea-hypopnea index (R-AHI) is acceptable. Accuracy of R-AHI from autoCPAP has not been well-studied in unmonitored conditions, yet these titrations are typically unattended. This study compares R-AHI scored by Respironics Remstar AutoCPAP to R-AHI scored manually on simultaneous level 3 polysomnography in OSA.
METHODS: A rural center affiliated with the University of Manitoba sleep laboratory utilizes level 3 diagnostic monitoring and single-night autoCPAP titrations with simultaneous level 3 monitoring for OSA patients. Patients referred in 2009-2010 with baseline testing and autoCPAP titrations were included. R-AHI scored by AASM criteria on level 3 study was compared to auto-CPAP-detected R-AHI using Pearson correlation and Bland-Altman Limits of Agreement. Assuming R-AHI <10/hr indicates acceptable treatment, frequency of patients incorrectly identified as "controlled" by the autoCPAP was calculated.
RESULTS: Eighty-six patients were included. Baseline characteristics: 65% males, mean age 53.7 years, mean BMI 33.4kg/m2, mean ESS 11.5, mean baseline AHI 27.6/hr (range 1.4-126.7/hr). AutoCPAP R-AHI and level 3 R-AHI were highly correlated (Pearson coefficient 0.75). Bland-Altman plot demonstrates low bias 0.11 with limits of agreement -11.24 to 11.45. The plot shows variation depends on magnitude of measurements. For R-AHI >15/hr, the methods did not agree consistently. Six patients (7%) had R-AHI<10/hr detected by the autoCPAP when simultaneous level 3 showed R-AHI>10/hr. In this situation, physicians may falsely believe OSA is well-controlled when further investigations or treatment adjustments may be needed.
CONCLUSIONS: In unattended CPAP titrations, R-AHI detected by Respironics Remstar AutoCPAP correlates highly with R-AHI detected by simultaneous level 3 monitoring. Bland-Altman analysis indicates the agreement of these measures is more accurate for R-AHI <15/hr.
CLINICAL IMPLICATIONS: Ensuring accuracy of R-AHI by autoCPAP is vital for clinicians who use these devices for make treatment decisions in OSA.
DISCLOSURE: The following authors have nothing to disclose: Nancy Porhownik, Stephen Corne
No Product/Research Disclosure Information