The influence of head and torso posture on obstructive sleep apnoea severity

Abstract

Obstructive sleep apnoea (OSA) is a common sleep disorder wherein the upper airway repeatedly collapses during sleep, leading to excessive daytime sleepiness, increased risk of fatal motor vehicle accidents and cardiovascular disease. As such, there is substantial motivation to identify the specific pathophysiological factors that contribute to the presence and severity of disease; and to develop methods of practically measuring these in the clinical environment. While it is well established that torso posture during sleep is associated with the severity of obstructive sleep apnoea, it is currently measured as a four-valued categorical value; and higher resolution measurements may identify more subtle, but potentially clinically important relationships between body posture and OSA severity. Furthermore, studies of anaesthetized patients suggest head flexion (chin down tilt) worsens airway patency; and that head rotation slightly improves patency. However, neither are measured routinely during sleep and head flexion has never been measured during sleep.Thus, the overarching aim of this thesis was to investigate the relationship between head and torso postures as continuous variables and the severity of obstructive sleep apnoea.Before the commencement of this thesis, a customised device to measure torso posture and head flexion/rotation from a torso supine posture was developed utilising tri-axial accelerometry (whereby one sensor unit is attached to the forehead and another to the chest). The specific aims of this project were to validate the device in bench-top testing and to apply the novel measurement device to a cohort of suspected apnoea patients in order to: (1) establish whether the dynamics of sleeping torso posture are adequately represented by the traditional four-posture sensor, (2) determine the impact of head flexion and head rotation during supine sleep on the severity of obstructive sleep apnoea and, (3) identify anthropometric and physiological predictors of head-posture dependent OSA.Bench-top testing revealed that the novel device had high single-axis performance (R2 > 0.99). During compound movements when the magnitude of flexion and rotation was less than 15°, the total error on any axis was less than 1.5°. Larger values of flexion and rotation was associated with increasingly greater error. Individuals attending the sleep laboratory for routine diagnostic polysomnography for suspected OSA were invited to participate in this study. 43 participants provided informed consent and were enrolled in the study. Polysomnographic recordings, including the novel head and torso posture data, from 28 individuals were of sufficient technical quality for inclusion in subsequent analysis.In the 28 participants, 44.8±28.2% of sleep time was spent with the torso outside of ±15° of the standard four-posture categorization. A novel polar-histogram based visualisation was developed to communicate sleep time and OSA severity at all torso postures. This tool allowed the identification of subtle torso-postural phenotypes that could not be elucidated using a standard 4-position posture sensor.No studies have previously been able to simultaneously measure and investigate head flexion and head rotation during sleep. In this analysis, the overnight data for the 28 participants was investigated by stratifying supine sleep into four categories by two binary variables: head rotated or not (>45°); and head flexed or not (>15°). 24 participants had >15 minutes sleep time in at least one of the four supine head postures. Linear mixed-effects analysis revealed head rotation >45° is associated with an 11.0 (95% CI: 0.3–21.6) events/hr improvement in the AHI; and head flexion >15° is associated with a worsening of apnoea by 12.9 (95% CI: 3.7 – 22.1) events/hr.In the final analysis, our investigation was stratified on a 30-second epoch-by-epoch basis to facilitate investigation of head postures as continuous variables; and to identify anthropometric and physiological associations with head-posture dependent OSA. Each epoch was considered a single observation and gender, BMI, age and sleep state are explanatory variables and were modelled to interact with head rotation and head flexion. The outcome variable is the binary presence of an obstructive event within the epoch. 25 participants had supine epochs that met inclusion criteria. Generalised logistic mixed-effects modelling revealed a variety of effects. We identified a non-linear relationship between head rotation and event probabilities and found a strong interaction effect between REM sleep and flexion (odds ratio of 5° flexion during REM sleep: 1.5, 95% CI: 1.1 – 1.9).In conclusion, the key findings of this thesis were: (1) The novel torso posture sensor elucidates varied torso-positional phenotypes previously unidentifiable with a categorical posture sensor; and a visualization was developed to quickly identify these in the clinical environment, (2) head flexion >15° significantly worsens apnoea severity, while head rotation > 45° improves it, consequently novel therapies which restrict head flexion and/or promote rotation may have utility in well selected patients and, (3) head flexion has a strong impact during rapid-eye movement sleep where neuromuscular responses are inhibited, which in turn implicates unfavourable passive anatomy as the primary cause of head flexion related OSA sensitivity.