Abstract
Introduction and ObjectivesMandibular movement (MM) monitoring with automated analysis has been shown to be a convenient home diagnostic test for obstructive sleep apnoea (OSA), comparable to both in-lab and home polysomnography.1 2 We aimed to evaluate the performance of a MM monitoring system (Sunrise) against polygraphy in two UK clinical settings: a geographically dispersed region- Highlands of Scotland and a densely populated region- inner London. We present an interim analysis from this study (SOSAT trial; NCT05204004).MethodsAn ongoing prospective, randomised, blinded pilot study comparing MM monitoring and home polygraphy in adults undergoing investigation for suspected OSA. Forty patients will be recruited with a body mass index (BMI)>28kg/m2 and Epworth sleepiness scale (ESS)>12. Participants wear MM monitoring (Sunrise, Sunrise SA, Belgium) and polygraphy (Apnoealink-Air, ResMed, Australia) simultaneously overnight. MM analysis is automated; polygraphy is manually reviewed. Primary outcome of the SOSAT trial is time to treatment decision. Secondary outcomes include accuracy of MM monitoring and agreement of the treatment decisions based on each device. In this interim analysis we have compared the apnoea-hypopnoea indices (AHI) from both devices using Bland-Altman analysis.Results17 participants have completed the trial to date, including one technical failure (both devices in the same participant). Data are, therefore, presented for n=16 participants (57% male, mean ± SD age 41.6 ± 9.7 years, BMI 39.2 ± 8.2 kg/m2, ESS 15.5 ± 3.8). Bland Altman analysis showed that MM monitoring underestimated AHI, compared to polygraphy (mean bias: -1.84 (95%CI -37.4 to +33.7) events/hour, figure 1). The variance was greater with higher AHIs (>15 events/hour). However, in 15 of 16 participants, the clinical treatment decision was the same irrespective of the diagnostic device used.Abstract S121 Figure 1Bland-Altman plot of the difference (Mandibular Movement estimated AHI minus Polygraphy AHI) vs average AHI of both devices (N=15). The dotted black line in the middle indicates the mean bias 1.84 (SD 18.1) events/hour. The 95% limits of agreement are marked by the upper and lower dotted lines. Shapes are based on the Polygraphy AHI thresholds; no OSA (< 5 events/hr), mild OSA (5–15) and moderate-severe OSA (>15 events/hr)ConclusionsThis interim analysis indicates that MM monitoring can be used for the clinical diagnosis of OSA. This novel diagnostic monitor could improve access to diagnosis in different geographical clinical settings.ReferencesPépin JL, et al. Assessment of mandibular movement monitoring with machine learning analysis for the diagnosis of OSA. JAMA NetwOpen 2020;3(1):e1919657.Kelly JL, et al. Diagnosis of sleep apnoea using a mandibular monitor and machine learning analysis: one-night agreement compared to in-home polysomnography. FrontNeurosci 2022;16:726880.Please refer to page A212 for declarations of interest related to this abstract.