Reliability of an adherence monitoring sensor embedded in an oral appliance used for treatment of obstructive sleep apnoea

The aim of this study was to test whether digitally registered use of a mandibular advancement device (MAD) by a built-in thermal sensor was reliable compared to a self-reported diary of MAD use. Eighty consecutive patients referred to a specialist outpatient sleep medicine clinic (HUS) were recruited. Patients of both genders, aged from 25 to 70 years with a diagnosis of mild, moderate or severe, were included. All participants signed a written informed consent when they received the MAD. For the purpose of this reliability study, we found it sufficient to include the first 30 nights of MAD use in the reliability analysis. At the 30th night follow-up visit, the self-reported diary with duration of MAD use was returned and data on the duration of MAD use with the built-in sensor were retrieved. From a total of 2400 nights, complete data from both methods were retrieved for 2108 nights (84.6%). Missing data were largely a result of missing self-reported diaries, whereas technical failure occurred in 6 nights (0.002%). The relative reliability was very high with ICC₃,₁ 0.847, and the absolute reliability for digitally registered MAD usage was calculated to -0.17 (95% CI: 1.47 to -1.81) hours in decimal conversion. Objectively collected data from built-in thermal sensors in MADs are as reliable as those of the self-report assessments. This opens new possibilities for more accurate measurements of MAD adherence.