Respiratory Event Detection by a Positive Airway Pressure Device

STUDY OBJECTIVES: Compare automatic event detection (AED) of respiratory events using a positive airway pressure (PAP) device with manual scoring of polysomnography (PSG) during PAP treatment of obstructive sleep apnea (OSA). DESIGN: Prospective PSGs of patients using a PAP device. SETTING: Six academic and private sleep disorders centers. PATIENTS: A total of 148 PSGs from 115 participants with OSA (apnea-hypopnea index AHI ≥ 15 events/hr) were analyzed. INTERVENTIONS: A signal generated by the PAP device identifying the AED of respiratory events based on airflow was recorded during PSG. MEASUREMENTS AND RESULTS: The PSGs were manually scored without visualization of the AED signal and scoring of a hypopnea required a ≥ 4% oxygen desaturation. The apnea index (AI), hypopnea index (HI), and AHI by manual score and PAP AED were compared. A customized computer program compared individual events by manual scoring and AED to determine the true positive, false positive, false negative, or true negative events and found a sensitivity of 0.58 and a specificity of 0.98. The AHI, AI, and HI by the two methods were highly correlated. Bland-Altman analysis showed better agreement for AI than HI. Using a manually scored AHI of ≥ 10 events/hr to denote inadequate treatment, an AED AHI ≥ 10 events/hr had a sensitivity of 0.58 and a specificity of 0.94. CONCLUSIONS: An AHI < 10 events/hr by PAP AED is usually associated with good treatment efficacy. Differences between manually scored and AED events were primarily due to different criteria for hypopnea detection.