P035 Is the Grael oximetry averaging time interchangeable with a Masimo pulse oximeter algorithm in polysomnography?

Compumedics recording software (Grael V2) for polysomnography (PSG) calculates SpO2 values using a 3-heartbeat long averaging window. This is derived from the ECG and thus introduces variability in the averaging time that is dependent on the heart rate. Little is known about the effect this has on the common oximetry metrics used in PSG interpretation. This study explorer the interchangeability of the Grael V2 inbuilt 3-beat averaging algorithm with a short averaging window of 2 - 4 seconds using a Masimo Radical 7 pulse oximeter during a PSG. SpO2 data were collected from 2 oximeter probes (Grael and Radical 7) both attached to a patient’s fingers. After SpO2 artifacts were removed, the following SpO2 parameters from each oximeter were generated: mean sleep SpO2, oxygen desaturation index (ODI) using 2%, 3% and 4% drop in SpO2 in sleep, total sleep time (TST) with SpO2 < 90% and < 80% as well as time spent < SpO2 88% in minutes. 88 sleep studies were included in the data collection. For ODI2%, 3% and 4%, bias (95% limits of agreement) values were -0.75 events/hr (9.99 to -11.49 events/hr), -0.74 events/hr (10.00 - -11.49 events/hr) and -0.20 events/hr (8.45 - -8.86 events/hr) respectively. There was no significant difference between measurements except for the mean sleep SpO2 values, p < .001. Although no bias found between measurements, there was poor agreement between the algorithms as demonstrated by the wide 95% limits of agreement suggesting that the two oximeter devices are not interchangeable.