0498 Home automated OSA quantification of night-to-night short term variability: analysis of the associated conditions

Abstract

Abstract Introduction Single night in-lab polysomnography (PSG) is still the conventional procedure for making obstructive sleep apnea (OSA) diagnosis, or alternatively a home sleep apnea test (HSAT) in patients free from comorbidity. Recent publications drew attention to considerable internight within-subject variability in OSA clinical status, notably but not only with HSAT. Taking benefit of multiple nights recording in home-based setting using measurements of mandibular jaw movement (MJM) combined with automated machine learning analysis, this study aimed to investigate the effect of change in sleep parameters on the internight variability of the apnea hypopnea index (VAHI). Methods 160 adults suspected of having OSA successfully completed 3 sleep tests at home with the Sunrise device (Sunrise, Namur, Belgium). Data analysis implied generalized estimating equation (GEE) regression. Results Even though the internight variability of AHI was not statistically significant at the population level with a mean VAHI of -1.1 ± 7.8 events/h, such variability could lead to misdiagnosis and wrong therapeutic decisions based on a single night. After adjusting for effects of gender, body mass index (BMI), age, total sleep time (TST) and time interval between sleep tests, regression models indicated that sleeping in supine position was positively and independently associated with VAHI, the AHI increased by 0.95 event/h for each 10% of increase in sleeping time in supine position. VAHI was negatively and significantly (p < 0.01) associated with sleep efficiency and deep sleep; VAHI decreased of 0.16 event/h (0.04 to 0.28) and 0.29 event/h (0.17 to 0.42) for each % of increase in sleep efficiency and proportion of deep sleep, respectively. Conclusion At the individual level, position, sleep efficiency and deep sleep are associated with short-term internight variability in AHI. Clinical decision based on a single-night testing may lead to errors in OSA diagnosis with delayed or wrong therapeutic decisions. Support (if any)