Background

Obstructive sleep apnea (OSA) is common in the general population, with prevalence in the United States of 3–7% in men and 2–5% in women (Punjabi 2008). OSA is defined as a sleep-related breathing disorder (SBD) that results in decreased or complete cessation of airflow while the patient has ongoing breathing effort. The cessation of airflow results from upper airway obstruction due to inadequate motor tone of the tongue and/or airway dilator muscles in the oropharynx during sleep (Park et al. 2011). Various factors contribute to OSA, including high body mass index (BMI), craniofacial anomalies, excess neck and/or tongue tissue, enlarged palatine tonsils, or sleeping in the supine position (Epstein et al. 2009). Among persons 30–70 years of age, the prevalence of mild to severe SBD (apnea–hypopnea index [AHI] ≥ 5 to < 15) is estimated to be 26% and prevalence of moderate to severe SBD (AHI ≥ 15) is estimated at 10% (Peppard et al. 2013).

It is well documented that trunk position significantly affects the severity of OSA. In fact, 50–75% of individuals with an OSA diagnosis exhibit supine predominance or worsened AHI when sleeping in the supine position (Ravesloot et al. 2013). Positional OSA is defined as an AHI ≥ 5 with > 50% AHI reduction between the supine and non-supine positions and AHI that normalizes (AHI < 5) in the non-supine position (Peppard et al. 2013). Identification of positional OSA, especially in the mild and moderate OSA populations, has led to research and development on positional therapies (Cartwright et al. 1991, 1985), including physical barriers to lying supine, such as a tennis ball attached to the patient’s back (Ravesloot et al. 2013; Eijsvogel et al. 2015), and small vibrating devices attached to the neck or back that arouse the patient if they are sleeping supine (Ravesloot et al. 2013, 2017; Bignold et al. 2011; Levendowski et al. 2014; van Maanen et al. 2012).

Recent research regarding the effect of head position, rather than torso position, on OSA severity may help improve upon the currently available positional therapies. A study using dual head and trunk position sensors to determine head and trunk position dependence in OSA patients identified 6.5% of OSA patients with solely head supine position dependence (van Kesteren et al. 2011). The study also showed that head position significantly influenced AHI in 46.2% of patients with trunk supine position dependence. When these patients were sleeping in the supine position, rotation of the head from supine (90° as shown in 1A) to lateral resulted in a significantly decreased mean AHI in both male and female patients. The authors defined these patients’ OSA as head position-aggravated, trunk supine position-dependent OSA (van Kesteren et al. 2011). This study also showed that the average AHI when the head was supine (regardless of trunk position) was significantly higher than when the trunk was supine (regardless of head position) (van Kesteren et al. 2011). Furthermore, a study of 67 patients with positional OSA who completed drug-induced sleep endoscopy found that lateral head position was associated with decreased frequency of complete anteroposterior collapse at the velum (P < 0.01), tongue base (P < 0.01), and epiglottis (P < 0.01) levels compared with supine head position (Safiruddin et al. 2014).

While these studies support the impact of head position on positional OSA severity, they do not describe specific head rotation angles that must be achieved in order to obtain clinical improvements in AHI in positional OSA patients. Both studies describe lateral versus supine head positions determined either by sensors with a threshold angle of ± 10° from the 45°-position boundary (van Maanen et al. 2012) or through video of the sleeping patient (Gandotra et al. 2018). Therefore, neither could identify the specific head rotation angles needed to significantly improve positional OSA severity. These metrics are essential for designing clinical therapies to reduce apneas for these patients.

The primary aim of this preliminary study was to determine the head rotation angles that reduce AHI and oxygen desaturation severity in positional OSA patients. Ultimately, the goal is to identify a “safe zone” of head rotation that can be used to support development of a head positional therapy for positional OSA patients.

Case presentation

Study design and participants

In this preliminary single-arm, non-randomized study, all participants completed overnight polysomnography with recording of head angles during sleep at the Mass Lung & Allergy, P.C. Sleep Disorders Center (Worcester, MA) between November 2016 and November 2017. The study was conducted in accordance with the Declaration of Helsinki and International Conference on Harmonisation (ICH) guidelines for Good Clinical Practice, and the study protocol was reviewed and approved by the New England IRB (Newton, MA). Participants were informed of the purpose of the study and provided written informed consent. Eligible participants were 21–60 years of age with a diagnosis of positional OSA from recent polysomnography results using the AASM guidelines of positional OSA defined as a lower AHI in the non-supine position than in supine (de Vries et al. 2015). Participants self-reported daytime drowsiness that persisted with use of continuous positive airway pressure (CPAP) therapy. They had to show compliance with CPAP use in the week prior to the screening visit, follow directions during an overnight sleep study, and discontinue CPAP therapy for two consecutive nights. Exclusion criteria included documented diagnosis of insomnia, chronic ear infections, persistent neck pain, persistent chronic posture physical issues, previous C-spine fusion, history of cardiac arrythmia, history of seizures, allergy to standard tape used in sleep centers, non-English speaking, hospitalization within the previous 4 weeks, use of antibiotics or steroids within the previous 4 weeks, any major uncontrolled disease or condition, such as congestive heart failure, malignancy, end-stage heart disease, amyotrophic lateral sclerosis (ALS), or severe stroke, or other condition as deemed appropriate by investigator, history of severe osteoporosis, excessive alcohol intake (> 6 oz hard liquor, 48 oz beer or 20 oz wine daily), or illicit drug use, and daily use of prescribed narcotics (> 30 mg morphine equivalent).

Sleep study and head rotation measurement

A proprietary sensor patch (Sleep Systems, LLC, Bedford, NH, USA) was developed for this trial and integrated with a dual axis inclinometer (SignalQuest, Lebanon, NH, USA), which was then packaged by Sleep Systems, LLC into a bedside unit fitted to the Embla communication system for the polysomnography montage. Before sleep, the sensor patch was attached to the participant’s forehead with adhesive and tape. The sensor was calibrated for each participant for head rotation angles 0° to 180° and head incline angles of 0° to 90°. Head rotation of 90° corresponds to lying supine with the head facing the ceiling; head rotation of 0° and 180° correspond to the head rotated fully to the right or left horizon, respectively (Fig. 1A). Using Fowler’s positioning model, head incline of 0° corresponds to lying supine with head facing the ceiling and 90° represents sitting upright (Fig. 1B).

Fig. 1
figure 1

Head rotation angle definitions (A). Head incline angles based on Fowler’s positioning model (B)

Full size image

Each participant was asked to fall asleep in their usual torso and head position for sleeping. Once asleep, data were recorded for all head and torso positions at as many of these positions as allowed. Each participant was coached in all sleeping positions for at least 30 min to collect data. Standard polysomnogram data were collected during the overnight study by a sleep research technologist, including 6-channel electroencephalogram (EEG), left and right electrooculogram (EOG), three chin electromyography (EMG), snoring sensor, airflow measured by oronasal thermistor and nasal pressure transducer, respiratory effort measured by thoracic and abdominal respiratory inductance plethysmography, pulse oximetry, left and right anterior tibialis EMG, electrocardiogram, body position and integrated digital audio and video recordings.

Sleep research technologists, who were blinded to head position associated with each sleep epoch, performed sleep and event scoring in accordance with The American Academy of Sleep Medicine Manual for the Scoring of Sleep and Events Version 2.0.3 (The and Manual for the Scoring of Sleep and Associated Events: Rules, Terminology, and Technical Specifications.Darien, IL:American Academy of Sleep Medicine 2014). In addition to standard polysomnogram data collection, AHI (total number of apnea and hypopnea episodes per hour of sleep), SpO2, torso position, and head rotation angle and head incline angle as determined by the forehead sensor, were recorded for each sleep epoch (defined as a period of sleep that maintained head angle with angular consistency ≤ 2°). The peak AHI was calculated for every epoch as the highest AHI of each head rotational subgroup.

The polysomnogram montages were specifically altered to include investigational channels for head rotation and head incline angles. A custom interface was developed to maintain compatibility with specific bedside polysomnography recorder auxiliary inputs. Small sponge wedges were added to compensate for individual forehead shape to keep angular errors ≤ 2°.

Body and head positioning during sleep

During the overnight sleep test, participants were coached by sleep research technologists to sleep with head rotation angles of ≤ 20°, 30–150°, and ≥ 160°, where horizon refers to head rotation fully to the left (180°) or right (0°). Each rotation angle was measured and recorded by the participant’s forehead sensor. Each head position was attempted with the torso in both supine and non-supine positions. The sleep research technologist-initiated changes in head position from right to left at specific degrees after a minimum sleep epoch duration of 10 min (goal 30 min) by placement of pillow wedges to keep the head stable at each angle. However, participants were able to change position while they slept. All positions, both coached and natural, were recorded for each sleep epoch.

Primary and secondary outcomes

The primary outcome was change in AHI severity with head rotation ≤ 20° above the horizon compared with > 20° above the horizon. The secondary outcome was change in SpO2 associated with head rotation ≤ 20° above the horizon vs > 20° above the horizon. To determine the effect of torso position on AHI, the peak AHI was calculated for sleep epochs when the participant was sleeping with torso supine compared with sleep epochs with torso non-supine (i.e., sleeping on right or left side or prone).

Statistical analysis

Formal power calculations were not performed for this exploratory analysis. All analyses were two-sided with a significance level at P < 0.05 and reported with 95% confidence levels. Data were analyzed for AHI and SpO2 dependence on head rotation ≤ 20° and > 20° above the horizon. Mean AHI was also calculated for supine versus non-supine torso positions overall and within the head rotation angle subgroups of ≤ 20° and > 20° above the horizon.

Results

Of the ten participants who enrolled and completed the overnight polysomnogram, data from two participants was excluded due to no apneic episodes observed in one patient and central apnea observed in another patient. Of the eight participants included in the analysis, six were female and two were male, with an average age of 48.5 years and average of BMI 32.5 kg/m2. Three participants were found to be torso position sensitive with supine to non-supine AHI ratio > 2. The remaining five participants had an AHI ratio < 1.6.

As shown in Table 1, participants spent more of their sleep time with head rotation > 20° above the horizon than ≤ 20° above the horizon. Peak AHI improved markedly for all patients when head rotation was ≤ 20° above the horizon compared with > 20° above the horizon (mean improvement: 66%, range: 18% to 88%). Seventy-five percent of participants had peak AHI improvement of at least 70% when head rotation was ≤ 20°% above the horizon. Of note, both the peak and the minimum AHI improvements were observed in patients with the highest BMI (49.2 kg/m2 and 42.5 kg/m2, respectively). The minimum SpO2 with head position ≤ 20° above the horizon was significantly higher than the minimum SpO2 with head position > 20° above the horizon (mean: 90.6% vs 84.3%, P = 0.03). Figure 2 shows peak AHI and minimum SpO2 for each patient for each 20° head rotation segment. With respect to torso position, six of the eight patients spent more of their sleep time in the non-supine position, and peak AHI was lower in the non-supine position for seven of eight participants (Table 2).

Fig. 2
figure 2

Individual participants’ peak apnea-hypopnea index (AHI) and minimum oxygen saturation (SpO2) by head rotation angle

Full size image
Table 1 Individual patient AHI and SpO2 values by head rotation angle > 20° and ≤ 20° off the horizon
Full size table
Table 2 Polysomnographic results and peak AHI in the torso supine vs non-supine versus torso positions
Full size table

When peak AHI was analyzed for sleep epochs by 20° segments of rotation, head rotation angles 61–80° and 101–120° had the highest average peak AHI values (Fig. 3A). When rotation was 81–100°, many patients experienced gasping and inability to sleep due to discomfort, as observed by the sleep research technologist. As a result, there were fewer sleep epochs with head rotation near 90°, resulting in less opportunity for apneas (Fig. 4A). The average of peak AHI values during the 20 sleep epochs in which head rotation was ≤ 20° above the horizon was significantly lower than during the 52 sleep epochs in which head rotation was > 20° above the horizon (20.0 vs 45.3, P = 0.002) (Fig. 3B). While the average of peak AHI values was lower for sleep epochs in the torso non-supine compared with supine position, regardless of head position, this difference was not statistically significant (47.5 vs 32.0, P = 0.05) (Fig. 3C). In the supine position, the average of peak AHI values was significantly lower when head rotation was ≤ 20° compared with > 20° (7.1 vs 52.1, P < 0.001) (Fig. 3D). In the non-supine position, the lower average of peak AHI values for head rotation ≤ 20° was not statistically significant (22.3 vs 38.4, P = 0.09) (Fig. 3D).

Fig. 3
figure 3

Mean apnea-hypopnea index (AHI) at all head rotation angles (A), head rotation ≤ 20° vs > 20° from horizon (B), torso position supine vs non-supine (C), and combination of head rotation angle and torso position (D)

Full size image
Fig. 4
figure 4

Force of gravity acting on oropharyngeal tissue to compress upper airway when head is supine compared to laterally rotated head (A). Proposed “Safe Sleep Region” of head rotation angles to reduce apnea - hypopnea index in positional OSA patients (B)

Full size image

Mean head incline was between 1° and 21° for all participants and maximum head incline was between 13° and 41° (Table 1). Changes in AHI and SpO2 were not observed across head incline angles (data not shown), likely owing to a narrow range of values that was < 35° in all but one patient who had a maximum head incline of 41°. According to the equation Fairway = Fgrav*sin [head rotation angle]* cos [head incline angle], at head incline < 25°, 90% of gravitational forces would be applied to the structures of the upper airway, resulting in upper airway collapse.

Discussion

This preliminary proof-of-concept study was conducted to evaluate the feasibility and acceptability of the association of head position with apnea severity. The outcome of this study has provided preliminary validation data and knowledge to refine the methodology and design for future main studies including participant enrollment, instrumentation, data collection and participant positioning procedures. This preliminary study demonstrates that AHI and SpO2 de-saturation severity can be significantly improved by maintaining head rotation ≤ 20° above the horizon in participants with positional OSA. It is hypothesized that this is due to a significant reduction in the compressive force of oral cavity and oropharyngeal tissue on the upper airway as the head rotates away from the 90° position. Due to the perpendicular force of gravity on the jaw, tongue and soft palate, which are positioned directly anterior to the upper airway when a patient is sleeping with his or her head at the 90° position, 100% of the gravitational force is acting on this tissue to compress and, therefore, obstruct the airway (Fig. 4A)  (Marques et al. 2017). As a patient rolls his or her head to the right or left, the position of this tissue changes such that the gravitational force results in progressively less compression of the upper airway. In a study of OSA patients who underwent upper airway endoscopy during natural sleep, epiglottic collapse was virtually abolished and ventilation increased by 45% with lateral positioning (i.e. lying on right side with the head in a lateral neutral position) compared to supine positioning (Marques et  al. 2017). Our preliminary study expands on the current understanding of anatomical collapse of the upper airway by analyzing the influence of specific head rotation angles on AHI and SpO2.

As the head is rotated away from 90°, gravitational force also pushes the tongue and soft palate tissue laterally in the mouth and oropharynx rather than completely posteriorly towards the upper airway. The more the gravitational force acts on these tissues laterally, the less the tissues are pushed into the upper airway where they can cause apnea or hypopnea events. When the patient’s head position is not 90°, tongue and soft palate tissue act as an object on an incline (Fig. 4A). When on this incline, which is defined by the head rotation angle, the perpendicular force of gravity (Fairway), which acts to compress the airway, is less than the total gravitational force (Fgrav) (Fairway = Fgrav*sin [head rotation angle]) (Fig. 4A). These calculations suggest that the airway collapsibility when head rotation is 20° above the horizon is 34% of the airway collapsibility when head rotation is at 90°. Such a reduction in airway collapsibility would be expected to significantly reduce the number of apnea and hypopnea events, which we observed in our study for head rotation ≤ 20° above the horizon. In a similar study, 26 participants underwent an overnight polysomnography calculating AHI with all trunk and head positions (supine and lateral) compared with the complete supine position (i.e., head and trunk supine) (Zhu et al. 2017). Lateral rotation of the head to the right or left with the trunk supine resulted in a significant reduction in mean AHI from 36.0 (22.5) to 25.8 (16.6) (P = 0.008), and an AHI decrease of 10 in 27% of patients. The trunk lateral-head lateral position resulted in a more dramatic reduction in mean AHI from 31.6 (20.2) to 4.1 (4.1) (Zhu et al. 2017).

Future studies should include a significantly larger sample size and open the inclusion criteria to all subjects diagnosed with OSA since most subjects showed a higher positional sensitivity ratio of AHI when monitoring head rotation as compared to torso rotation. Head rotation angles greater than 45° above the horizon should be minimized since most subjects experience maximum apnea severity within this range. The primary focus of future studies should show that AHI and SpO2 desaturation can be consistently minimized by limiting head rotation to angles less than 20° above the horizon.

Conclusion

In conclusion, this preliminary study has provided evidence that severity of AHI and SpO2 desaturation can be significantly minimized in positional OSA patients by positioning the head at angles ≤ 20° above the horizon when the torso is both supine and non-supine. Based on these results, we propose a “Safe Sleep Region” for OSA patients that encompasses all head positions ≤ 20° from the horizon to reduce AHI, thereby reducing severity of OSA. These results reinforce the influence of the head position on AHI.