Accuracy of oximetry with thermistor (OxiFlow) for diagnosis of obstructive sleep apnea and hypopnea

Objectives: To evaluate the diagnostic accuracy for obstructive sleep apnea and hypopnea (OSAH) of the OxiFlow (OF) device which combines oximetry wit h recording of thermistor airflow. Design & Setting: Patients scheduled for overnight diagnostic polysomnograp hy (PSG) were studied with OF either simultaneously during laboratory PSG ( L-OF, n=86), at home on a separate night (H-OF, n=66), or both (n=55). Patients: 97 patients with suspected OSAH, of whom 40 had OSAH defined as a n apnea-hypopnea index (AHI) of more than 15 events per hour of sleep on PS G. Interventions: NA Measurements & Results: The automated respiratory disturbance index (RDI) g enerated by the OF software considerably underestimated the AHI by PSG for both L-OF and H-OF. Altering the parameters for hypopnea identification by the software did not improve this. Visual inspection of the computerized OF tracings added considerable diagnostic information, but a manual count of RDI during visual review overestimated AHI. For the identification of cases vs. non-cases of OSAH, receiver operating characteristic area-under-the-cu rve statistics ranged from 0.77 - 0.90 for L-OF and from 0.71 - 0.77 for H- OF. Combining automated analysis with subsequent visual inspection of OF tr acings yielded an overall sensitivity of 86% and specificity of 74% for the diagnosis of OSAH during H-OF recordings. Analysis of potential technician time saved indicated a benefit from the use of OF. Conclusions: OF has diagnostic utility for the identification of OSAH. Howe ver, because of hardware and software limitations, it is unclear whether th is device is superior to oximetry alone.