Ioannis Koutsourelakis, MD, PhD, Medical School of Athens University, Center of Sleep Disorders, Evangelismos Hospital, 45–47 Ipsilandou Str, GR 106 75, Athens, Greece. Tel.: 0030-210-72-01-843; fax: 0030-210-72-01-843; e-mail: firstname.lastname@example.org
Snoring is considered one of the hallmarks of sleep-disordered breathing, but its determinants remain obscure in both obstructive sleep apnoea (apnoeic) and non-apnoeic snorers. We aimed to document positional dependency of snoring along with its association with clinical and polysomnographic variables. Seventy-seven apnoeic and 27 non-apnoeic snorers who complained for every-night loud snoring and slept in supine and lateral positions in all sleep stages during overnight polysomnography were included. Snoring (i.e. sound intensity > 40 dB) was quantified by measuring the mean and maximum sound intensity, and snoring frequency. In apnoeic and non-apnoeic snorers, mean snoring intensity and snoring frequency were higher in supine than in lateral positions irrespective of sleep stage, and were also usually higher in N3 in comparison to rapid eye movement and/or N2 sleep stage in any given position. Positional change in snoring intensity as expressed by the ratio of mean intensity in the supine to lateral positions was independently and positively correlated with body mass index, tonsils size and age in the total of patients. Snoring is more prominent in the supine position and in N3 sleep stage in apnoeic and non-apnoeic snorers. Snoring positional dependence is determined by body mass index, tonsils size and age.
In patients with OSA, body position during sleep is known to influence the frequency of obstructive respiratory events (American Academy of Sleep Medicine Task Force, 1999). In fact, it is documented that in one out of four patients with OSA who undergo polysomnography, apnoeas and hypopnoeas occur predominantly in the supine position (Mador et al., 2005). The positional dependence of OSA is explained by the fact that turning from the lateral to supine position can lead to a decrease and an eventual collapse of the upper airway (positional OSA patients; Pevernagie et al., 1995). In these patients positional therapy has proven to be an effective alternative to continuous positive airway pressure therapy (Permut et al., 2010).
Therefore, the aim of the present study was to investigate the effect of body position on snoring in a large sample of patients with sleep-disordered breathing, taking into account the sleep stage along with a variety of polysomnographic and clinical factors. It was hypothesized that snoring depends on sleep position, sleep stage and the presence of OSA.
Materials and Methods
All consecutive subjects who were referred to the Center of Sleep Disorders of ‘Evangelismos’ General Hospital of Athens for sleep-disordered breathing between January 2010 and September 2010 and complained for every-night loud snoring were recruited. Exclusion criteria were considered: (1) the presence of < 15 min of any sleep stage [N2, N3 and rapid eye movement (REM)] in supine or lateral sleep positions during polysomnography; (2) central apnoeas more than 50% of total apneas; and (3) the absence of snoring during polysomnography. The study protocol was approved by the hospital ethics committee, and all patients gave written informed consent.
Initial assessment included a full clinical history and physical examination. Full clinical history included information about chronic illnesses, alcohol use and smoking, as per the authors’ usual protocol (Koutsourelakis et al., 2008). Subsequently, nasal resistance was measured in the supine position using active anterior rhinomanometry (Clement and Gordts, 2005). Tonsils size was defined according to the standard criteria (Friedman et al., 2002). In brief, size 1 corresponded to tonsils hidden within the pillars; size 2 corresponded to tonsils extending to the pillars; size 3 corresponded to tonsils extending beyond the pillars but not to the midline; and size 4 corresponded to tonsils extending to the midline. Friedman soft palate position score was graded as follows: grade 1: visualization of the entire uvula and tonsils/pillars; grade 2: visualization of the uvula but not the tonsils; grade 3: visualization of the soft palate but not the uvula; and grade 4: visualization of the hard palate only (Friedman et al., 2002). Thereafter, patients underwent a full-night polysomnography with concomitant measurement of snoring sound intensity.
A full-night diagnostic polysomnography (EMBLA S7000; Medcare Flaga, Iceland) was performed in each patient. To determine the stages of sleep an electroencephalogram (C4-A1, C3-A2, O2-A1, O1-A2), an electro-oculogram and an electromyogram of the submentalis muscle were obtained. Arterial blood oxyhaemoglobin was recorded with the use of a finger pulse oximeter. Thoracoabdominal excursions were measured qualitatively by respiratory effort sensors placed over the rib cage and abdomen. Body posture was detected with a body position sensor, which differentiates between the following four positions: supine; right lateral; left lateral; and prone. Airflow was monitored using an oral thermistor and a nasal cannula/pressure transducer. All variables were recorded with a digital acquisition system (Somnologica 3.3; Medcare Flaga, Iceland).
Snoring sound measurement
The same room with polysomnography was used for the sound measurement. Snoring was measured using a calibrated microphone-sound meter system. The two microphones were suspended at a distance of 1 m above the patient’s bed. This arrangement allowed a non-invasive measurement of snoring that simulated the distance between sleeping bed partners. Before every study the system was acoustically calibrated using a reference noise produced by a noise generator (86 dB). The signal was sent at a sampling rate of 12 kHz through an analogue–digital converter to a computer system for subsequent analysis. All digitized signals were recorded on a personal computer. A noise analyser (Praat; Boersma, 2001) was used for the intensity analysis of snoring sound (Boersma, 2001).
After manually scoring the sleep stages, all awake time was excluded (Iber et al., 2007). Loud breathing measured at a 1 m distance from the mouth is considered to register sound levels of up to 40 dB (Pevernagie et al., 2010); this value has been proposed as the threshold for transition from loud breathing to snoring (Pevernagie et al., 2010), and also was adopted in the present study. Indeed, spikes in sound intensity > 40 dB were always perceived as snores by the sleep technologist who was able to hear patient’s breathing. The noise analyser automatically erases the noise of the background, generates a histogram of sound intensity and provides a summary of statistics, which includes the mean and maximum sound intensity over the sampling period expressed in dB.
Definitions and analysis
Snoring was quantified by measuring snoring frequency/index (defined as the number of snores per hour of sleep – snores h−1) and sound intensity (mean and maximum – dB).
Sleep stage was scored manually in 30-s epochs and obstructive respiratory events were scored using standard criteria by an experienced technician (Iber et al., 2007). The number of episodes of apnoeas and hypopnoeas per hour of sleep is referred to as the apnoea–hypopnoea index (AHI – events h−1). OSA was diagnosed if AHI was > 5 (American Academy of Sleep Medicine Task Force, 1999). All measurements were analysed by a single investigator to ensure consistency, and all polysomnographies were scored by a single experienced sleep technologist and subsequently reviewed by an investigator, who was blinded to the patient’s clinical data.
Quantitive data are reported as mean ± SD. The minimum sample size was calculated (G*Power 3.0.10) to be 80 patients based on 80% power, a two-sided 0.05 significance level, and known standard deviations (Nakano et al., 2003). The normality of data distribution was assessed by the Kolmogorov–Smirnov test. Comparison of data between snorers and apnoeics was carried out using the unpaired t-test. In order to investigate the effect of body position, sleep stage and their interaction on snoring we performed repeated-measures two-way analysis of variance (anova) followed by the Scheffé test for post hoc analyses. Relationships between positional dependence of snoring and various variables were investigated by performing simple linear regression analysis for each variable separately. Multiple linear regression analysis was performed to identify the variables independently related to positional dependence of snoring. All variables that significantly correlated with positional dependence of snoring in simple linear regression analysis were the independent variables included in the model. The stepwise procedure was used to select the best model. A P-value of < 0.05 was considered to indicate statistical significance.
Of the 309 patients initially enrolled in the study, 104 were considered eligible for further analysis, while 205 patients were excluded because in 180 patients the duration of sleep stages was < 15 min in either the supine or lateral position (22 patients had < 15 min of N2 sleep, 102 patients had < 15 min of N3 sleep, and 56 patients had < 15 min of REM sleep), in six patients central apnoeas were more than 50% of total apnoeas, and in 19 patients no snoring was detected during polysomnography. Of the remaining 104 patients, who formed the study group for this report, and based on their polysomnographic studies, 27 were simple snorers and 77 were apnoeic snorers. Table 1 summarizes the anthropometric, sleep and snoring data of the two groups. Apnoeic snorers were older, and had higher (mean and maximum) snoring intensity and snoring frequency and lower N3 and REM sleep duration in comparison to simple snorers. In apnoeic snorers, AHI in supine and lateral positions was 32.8 ± 26.2 and 31.2 ± 27.1 events h−1, respectively (P = 0.61), whereas AHI in N3 (23.3 ± 19.6 events h−1) was lower than in N2 (30.2 ± 26.6 events h−1) and REM (34.5 ± 26.5 events h−1; P < 0.05 for both).
Table 1. Anthropometric, sleep and snoring data in simple snorers and apnoeics
*P < 0.05 versus simple snorers; **P < 0.01 versus simple snorers.
40.9 ± 12.9
55.0 ± 13.6**
Body mass index (kg m−2)
28.0 ± 4.3
30.1 ± 4.4
Male, n (%)
Nasal resistance supine (cmH2O L−1 s)
2.5 ± 0.9
2.7 ± 0.8
Friedman soft palate position score
1.92 ± 1.01
1.66 ± 0.98
1.00 ± 0.83
1.48 ± 0.97*
Smoke, n (%)
Alcohol, n (%)
Depression, n (%)
Asthma, n (%)
AHI (events h−1)
2.7 ± 1.0
31.7 ± 25.1**
Sleep efficiency (%)
86.5 ± 16.3
90.3 ± 10.7
192.6 ± 43.6
202.5 ± 46.8
61.6 ± 10.7
39.9 ± 8.5*
60.6 ± 19.5
38.2 ± 6.7*
Epworth Sleepiness Scale score
6.6 ± 3.9
8.6 ± 4.2
Snoring maximum intensity (dB)
64.1 ± 10.5
73.9 ± 9.4**
Snoring mean intensity (dB)
52.3 ± 8.0
58.9 ± 8.3**
Snoring frequency (snores h−1)
178 ± 61
203 ± 55**
In both simple and apnoeic snorers, mean snoring intensity in the supine position was higher than in the lateral position irrespective of sleep stage (P < 0.001; Fig. 1). In the supine position, mean snoring intensity of simple snorers was higher in N3 than in REM (P < 0.01; Fig. 1a), whereas in both supine and lateral positions the mean snoring intensity of apnoeic snorers was higher in N3 than in N2 and REM (P < 0.01; Fig. 1b).
In both simple and apnoeic snorers, snoring frequency in the supine position was higher than in the lateral position irrespective of sleep stage (P < 0.001; Fig. 2). In the supine position, the snoring frequency of simple snorers in REM was lower in comparison to N2 and N3 (P < 0.05 and P < 0.01, respectively; Fig. 2a). In the supine position, the snoring frequency of apnoeic snorers was higher in N3 than in REM and N2 (P < 0.01), and higher in N2 than in REM (P < 0.05; Fig. 2b). In the lateral position, the snoring frequency of apnoeic snorers was higher in N3 than in REM (P < 0.01; Fig. 2b).
The relationships between the ratio of mean snoring intensity in the supine to lateral positions and polysomnographic or anthropometric variables in the total of patients are shown in Table 2. As can be seen, the ratio of mean snoring intensity in the supine to lateral positions was positively correlated with body mass index, age, tonsils size, AHI, nasal resistance, male gender and Friedman soft palate position score, and negatively correlated with Epworth Sleepiness Scale score (Table 2).
Table 2. Simple linear regression analysis models for the ratio of mean snoring intensity in the supine to lateral positions in all patients
AHI, apnoea–hypopnoea index; B, parameter estimate; SE, standard error.
Body mass index (kg m−2)
AHI (events h−1)
Nasal resistance supine (cmH2O L−1 s)
Epworth Sleepiness Scale score
Gender (1 = female, 2 = male)
Friedman soft palate position score
The results of the forward-stepwise multiple linear regression analyses are summarized in Table 3. Backward-stepwise multiple linear regression analyses gave identical results. All variables significantly related to the ratio of mean snoring intensity in the supine to lateral positions in simple linear regression analysis (Table 2) were the independent variables included in the model. The ratio of mean snoring intensity in the supine to lateral positions was independently related to body mass index, tonsils size and age (higher in patients with increased values of these variables). Body mass index, tonsils size and age accounted for 49.3, 10.5 and 3.4%, respectively, of the variance explained by the model (R2 = 0.632; Table 3).
Table 3. Multiple regression analysis for the ratio of mean snoring intensity in the supine to lateral positions in all patients
Model R2 = 0.632
Selection of variables was made by the forward stepwise procedure.
B, parameter estimate; PC, partial correlation; R2, total variance explained by the model; ΔR2, partial contribution of each variable to the total variance explained by the model; SE, standard error.
Body mass index (kg m−2)
The main findings of the present study were: (1) in non-apnoeic and apnoeic snorers, mean snoring intensity and snoring frequency were higher in the supine than the lateral position, irrespective of sleep stage; (2) in non-apnoeic and mainly in apnoeic snorers, mean snoring intensity and snoring frequency were usually higher in N3 in comparison to REM and/or N2 sleep stage in any given position; and (3) the determinants of positional change in snoring intensity as expressed by the ratio of mean snoring intensity in the supine to lateral positions were body mass index, tonsils size and age.
Whether a social or medical problem, snoring is very common and frequently sufficiently bothersome to the bed partner to impel medical attention. Its positional dependency has been addressed twice previously with contradicting results. Indeed, Braver et al. (1994) in a group of 20 male non-apnoeic and apnoeic snorers compared the number of snores per hour of sleep before and after positional therapy, and suggested that snoring index is not sensitive to changes of sleep position. Conversely, Nakano et al. (2003) documented a positional dependency of snoring in a group of 21 non-apnoeic snorers; apnoeic snorers did not demonstrate such a positional snoring dependency. In this study (Nakano et al., 2003) only seven simple and seven apnoeic snorers in whom all sleep stages were observed in lateral and supine positions were analysed for snoring changes in different sleep stages. However, besides the small number of patients recruited, the amount of time spent in different sleep stages was relatively small (e.g. N3 accounted for only 0.4 ± 0.7% of total sleep time spent in the supine position in apnoeics), thus making any conclusion drawn rather uncertain (Kryger et al., 2005). The present investigation adds to the literature by studying a respectable number of consecutive patients, both non-apnoeic and apnoeic snorers, therefore being adequately powered. Moreover, patients were included only if the time spent in every sleep stage in the supine and lateral positions exceeded 15 min, giving in that way a representative overview of their sleep staging (Kryger et al., 2005). Stage N1 was not included in the analysis because its amount was limited (mean time spent at N1 was 4.2 ± 2.2 min). In general, stage N1 is considered a transition between wakefulness and sleep, and usually accounts for only 2–5% of total sleep time, rendering its contribution in snoring relatively weak (Kryger et al., 2005).
The results of the present study show that snoring positional dependence is not only present in non-apnoeic snorers (Nakano et al., 2003), but also in apnoeics irrespective of sleep stage. This finding might be explained by the fact that in the supine position upper airway dimensions present increased propensity of collapse (Walsh et al., 2008). In this regard, it is not surprising that positional therapy has proven to be equivalent to continuous positive airway pressure at normalizing AHI in patients with positional OSA (Permut et al., 2010). However, even in patients with non-positional OSA, as were the apnoeic patients of the current study, positional dependence of snoring appears to persist.
In both non-apnoeic and apnoeic snorers, mean snoring intensity and snoring frequency proved to be usually higher in N3 than in REM and/or N2 sleep stage in a given position. This finding was observed in non-apnoeic snorers only in the supine position, whereas in apnoeics in both the supine and lateral positions (Figs 1 and 2). These results are compatible with the reports of previous trials, which also documented that snoring was more prominent in slow-wave sleep (Nakano et al., 2003; Perez-Padilla et al., 1987), and that apnoeics presented more pronounced differences (Hoffstein et al., 1991). However, the aforementioned studies either did not take into account the sleep position of patients (Hoffstein et al., 1991; Perez-Padilla et al., 1987) or did not recruit patients with adequate amounts of different sleep stages (Nakano et al., 2003). Consequently, the results of the present study make the findings of previous trials more conclusive in the same direction.
The determinants of positional change in snoring intensity as expressed by the ratio of mean snoring intensity in the supine to lateral positions proved to be body mass index, tonsils size and age. In particular, obesity along with facial and paraphayngeal fat volume have been associated previously with upper airway size changes in patients with OSA (Sutherland et al., 2011), and this might explain the role of body mass index in the current study. Additionally, the role of tonsils size as documented in the current study is consistent with the findings of Soga et al. (2009), who showed that the width of the fauces in the pharynx was greater among responders than among non-responders to positional therapy. Lastly, the contribution of age to positional variability of snoring intensity could potentially account for the inconsistent results of positional change between our and a previous study (Nakano et al., 2003), which both included patients with mean age ∼ 48 years and showed positional dependency of snoring, and another trial (Braver et al., 1994), which included younger patients (mean age 42 years) and demonstrated that snoring is not sensitive to changes of sleep position.
The present study has some potential limitations that should be addressed. The most important is that, although many consecutive patients were recruited, it was a single-centre study on a selected sample of snorers who were included only if they had spent adequate time in both the supine and lateral positions in N2, N3 and REM sleep stages during polysomnography. Consequently, the results obtained cannot be safely extrapolated to the general population of snorers. Additionally, despite the fact that all patients who were initially enrolled in this study complained of snoring, there were 19 patients who were excluded because they did not snore during polysomnography. Because there may be a significant night-to-night variability in snoring, one-night polysomnography might be unreliable to detect snoring in some of these patients, and they might have been included in the analysis if they were studied another night (Cathcart et al., 2010).
In conclusion, the results of the present study suggest that in non-apnoeic and apnoeic snorers snoring intensity and frequency were higher in the supine position irrespective of sleep stage, and were also usually higher in N3 in comparison to REM and/or N2 sleep stage in a given position. Positional dependence of snoring intensity was determined by body mass index, tonsils size and age. It is thus plausible to suggest that positional therapy might be more effective in more obese, with bigger tonsils and older patients. Further studies are warranted to prospectively test these findings in positional therapy.
Conflict of Interest
None of the authors has any financial support or conflict of interest to disclose.