P036 Evaluation of position sensors used during sleep studies

Abstract Introduction/Background Accurate body position measurement is used for diagnosing and managing obstructive sleep apnoea (OSA). Typically, position sensor data are corrected by experts using video recordings, but this is unavailable during unattended polysomnography (PSG). It is thus important to understand the accuracy of position sensors. Aim To evaluate the correction rates of position sensor measurements during PSG in a clinical laboratory setting. Methods We analysed 30 consecutive PSGs, all with video recordings, conducted in adults being assessed for OSA. The studies used Compumedics Ltd. (Abbotsford, Victoria, Australia) Grael™ PSG equipment with five-state (upright, left-lateral, right-lateral, prone, supine) body position sensors. After each study was completed the correct body position was determined in each thirty second epoch by viewing the video. These were then compared with the body position recorded by the position sensors for each epoch. Results A total of 28,377 30-second epochs from 30 PSGs (average: 946 epochs per study, range: 743-1249) were analysed. On average, across all studies 18.3% of epochs had the incorrect body position recorded by the position sensors. The degree of error within a study varied from 0.5% to 78% of epochs. Discussion/Conclusion Significant discrepancies in body position were found between the sensors and what was seen on video recordings of the studies. It is thus not possible to rely on position sensors when reporting sleep related OSA in sleep studies.