Physiological Determinants of Snore Loudness

Abstract

Rationale

The physiological factors modulating the severity of snoring have not been adequately described. Airway collapse or obstruction is generally the leading determinant of snore sound generation; however, we suspect that ventilatory drive is of equal importance.

Objective

To determine the relationship between airway obstruction and ventilatory drive on snore loudness.

Methods

In 40 patients with suspected or diagnosed obstructive sleep apnea (1–98 events/hr), airflow was recorded via a pneumotachometer attached to an oronasal mask, ventilatory drive was recorded using calibrated intraesophageal diaphragm electromyography, and snore loudness was recorded using a calibrated microphone attached over the trachea. “Obstruction” was taken as the ratio of ventilation to ventilatory drive and termed flow:drive, i.e., actual ventilation as a percentage of intended ventilation. Lower values reflect increased flow resistance. Using 165,063 breaths, mixed model analysis (quadratic regression) quantified snore loudness as a function of obstruction, ventilatory drive, and the presence of extreme obstruction (i.e., apneic occlusion).

Results

In the presence of obstruction (flow:drive = 50%, i.e., doubled resistance), snore loudness increased markedly with increased drive (+3.4 [95% confidence interval, 3.3–3.5] dB per standard deviation [SD] change in ventilatory drive). However, the effect of drive was profoundly attenuated without obstruction (at flow:drive = 100%: +0.23 [0.08–0.39] dB per SD change in drive). Similarly, snore loudness increased with increasing obstruction exclusively in the presence of increased drive (at drive = 200% of eupnea: +2.1 [2.0–2.2] dB per SD change in obstruction; at eupneic drive: +0.14 [–0.08 to 0.28] dB per SD change). Further, snore loudness decreased substantially with extreme obstruction, defined as flow:drive <20% (–9.9 [–3.3 to –6.6] dB vs. unobstructed eupneic breathing).

Conclusions

This study highlights that ventilatory drive, and not simply pharyngeal obstruction, modulates snore loudness. This new framework for characterizing the severity of snoring helps better understand the physiology of snoring and is important for the development of technologies that use snore sounds to characterize sleep-disordered breathing.

Snoring is a hallmark of obstructive sleep apnea (OSA), a key manifestation of this disorder that provides the motivation for patients to seek treatment and is easy to detect in the home environment. Although snoring and OSA are consequences of a collapsible pharyngeal airway, snoring is not a specific indicator of obstructive respiratory events. Notably, snoring is often seen in the absence of OSA, and the reasons why it appears and disappears are poorly understood, compromising its use as a diagnostic signal. For example, a particularly puzzling phenomenon is the observation of increased snoring during slow wave sleep, a stage at which the airway is paradoxically less collapsible (1–3), whereas snores are less frequent in rapid eye movement (REM) sleep (4–7), a stage at which the airway is most collapsible (1–3). These observations do not appear to be well explained by the prevailing view that the leading determinant of snore sound generation is pharyngeal airway obstruction, in which sleep-related pharyngeal dilator muscle atonia yields a narrowed pharyngeal airway, increased resistance to airflow, and vibration of the pharyngeal tissues as higher-velocity turbulent air passes through the narrowed compliant pharynx (7, 8).

In addition to pharyngeal obstruction, an underappreciated determinant of snoring is ventilatory effort or drive (9). Increasing drive and effort would increase driving pressure across the pharynx, leading to faster flow velocity and louder snoring independent of the severity of pharyngeal compromise. Indeed, Stoohs and Guilleminault reported that increased ventilatory effort, measured by esophageal pressure, is seen at the onset of snoring and increases further as snoring continues (9). However, the relative contributions of ventilatory drive and pharyngeal obstruction to snoring remain unclear. Of note, previous work examined esophageal pressure rather than neural ventilatory drive (i.e., using diaphragm electromyography [EMG]). This may be problematic because negative pressures can increase as a result of the presence of obstruction, making it challenging to interpret whether drive increased or pharyngeal obstruction increased when pressures became more negative.

The present study aims to elucidate the relationship between snore loudness and both pharyngeal obstruction and ventilatory drive. We hypothesized that snoring increases in loudness as 1) pharyngeal obstruction increases and 2) ventilatory drive increases. We also considered the caveat that severe obstruction (i.e., apnea) would yield quieter snores accompanying an absence of airflow. To test these hypotheses, we studied patients with suspected OSA during sleep by measuring snore sounds with a calibrated microphone attached over the trachea, along with direct measurements of ventilation and ventilatory drive. Primary analysis modeled snore loudness as a function of obstruction (i.e., a ventilation–to–ventilatory drive ratio termed “flow:drive” as a measure of resistance, i.e., actual ventilation as a percentage of intended ventilation), ventilatory drive, and the presence of severe obstruction (flow:drive <20%). We then apply our new framework to explain differences in snore loudness during the night. Specifically, we examine differences in snore loudness across: 1) different forms of sleep-disordered breathing, i.e., stable flow limitation (characterized by increased drive) and hypopneas (characterized by greater pharyngeal obstruction), and 2) different sleep stages, i.e., REM sleep (characterized by low drive [10]) and slow wave sleep (characterized by increased drive [11]).

Methods

Participants

Data on 46 participants with suspected or diagnosed OSA were used as an ancillary study to two parent studies (12, 13) to investigate the contributions of airway obstruction and ventilatory drive to snore loudness. Exclusion criteria to the parent studies included central sleep apnea, lung diseases or heart failure, use of respiratory simulants or depressants (e.g., opioids, benzodiazepines), and pregnancy. Of the 46 patients, one could not tolerate the intraesophageal catheter and five were excluded because the microphone disconnected from the patient, yielding poor-quality tracheal snore data. Periods with evidence of the microphone disconnecting from the patient were identified through detailed manual review. Specifically, signals were considered suspicious for disconnection based on unusually low snore loudness despite snoring visible in the airflow signal, absence of cardiac and breath sounds, and/or a high noise floor; suspicious periods were confirmed by listening to the raw audio for evidence of open-air snoring sound characteristics. This left 40 participants for the present analysis.

Protocol

Participants attended a single overnight PSG study with additional measurements of 1) airflow via a pneumotachometer (Hans-Rudolph) attached to a tightly fitting oronasal mask and 2) ventilatory drive with an intraesophageal diaphragm EMG catheter (Yinghui Medical Equipment Technology Co. Ltd.). Snore sounds (10 kHz) were also measured using a studio-quality omnidirectional microphone (ECM-77B; Sony) fit inside of a custom sealed microphone holder that facilitated attachment to the trachea using double-sided tape (14, 15). Raw data tracings from a representative patient are illustrated in Figure 1.

Figure 1.

Raw data tracings from a single representative patient illustrating differences in snore loudness (in dB), airflow, and diaphragm electromyography during apnea, hypopnea, and flow-limited breathing during stage N2 and N3 sleep. Each panel maintains identical scaling for all variables. Green, red, and purple lines represent technician-scored apneas, hypopneas, and arousals, respectively. EMGdia = diaphragm electromyography.

Breath-by-Breath Ventilation, Ventilatory Drive, and Snore Loudness

Breath-by-breath ventilation was calculated as the product of tidal volume (height of the volume signal above baseline) and respiratory rate (reciprocal of breath duration), expressed as a percentage local mean of ventilation (i.e., eupneic ventilation). To calculate ventilatory drive, swings in root mean squared diaphragm EMG (peak minus preinspiratory nadir; cardiogenic artifact removed) were scaled such that the average value during wakefulness was equal to the average value of ventilation during wakefulness. The result is a ventilatory drive signal in units of ventilation (L/min) that can be thought of as an “intended” ventilation. Ventilation and ventilatory drive were then both presented as a percentage local mean of ventilation (i.e., eupneic ventilation) as previously described (12, 16, 17). Breath-level mean snore loudness in decibels was computed for each inspiration. Processing involved filtering with a 20-Hz first-order high-pass filter (i.e., minimum audible range). The resulting signal was squared and smoothed (50-ms time constant) to produce a snore “energy” signal (P²; units of Pa²), which was converted to decibels with the formula dB = 20 × log10(P² / P_o²), where P₀ is the reference sound pressure (0.00002 Pa). Breath-level values of ventilation, ventilatory drive, and snore loudness were tabulated, along with other relevant sleep and breathing metrics such as presence of respiratory events, arousals, and sleep stages.

Quantifying Airflow Obstruction

The present study defined airflow obstruction as the mismatch between ventilation and ventilatory drive (i.e., “flow:drive” ratio [18]). Conceptually, this ratio is akin to the reciprocal of the respiratory system resistance or impedance. For example, in a completely open airway, actual ventilation will match ventilatory drive and produce a flow:drive ratio of 100%. A flow:drive ratio of 50% represents an obstructed airway in which the achieved ventilation is only half of the intended level based on ventilatory drive. We note that the measure captures greater obstruction caused by an increase in the collapsibility of the airway (reduced peak inspiratory flow and therefore ventilation) but also can show a greater degree of obstruction without a change in collapsibility per se. For example, increasing drive across a flow-limited pharyngeal airway with Starling resistor–like behavior increases pharyngeal narrowing and thereby the overall resistance, i.e., greater pharyngeal obstruction. Thus, airflow obstruction (flow:drive) is not a fixed property of the pharyngeal airway.

Definitions of Sleep-Disordered Breathing Types

Sleep, arousals, and apneas and hypopneas were scored from the PSG study by a registered PSG technologist according to the recommended American Academy of Sleep Medicine guidelines (19). Respiratory events were at least 10 seconds in duration. Hypopneas were defined by ⩾30% flow reduction with ⩾3% desaturation or arousal. Apneas were defined by a ⩾90% flow reduction; unscored breaths with ventilation <10% of eupnea were pooled with apnea breaths. Arousals were based on at least 3 seconds of an abrupt shift in higher-frequency electroencephalographic activity (excluding spindles) per American Academy of Sleep Medicine criteria. Stable breathing was defined as a consecutive period of at least 3 minutes without a respiratory event or arousal from sleep. Stable breathing was divided into flow-limited and non–flow-limited breaths based on flow:drive ratios <86% or ⩾86%, respectively (i.e., the cutoff at which airflow visually appears flow-limited [20]).

Statistical Analysis

The present study tested the hypotheses that greater snore loudness is associated with increased ventilatory drive and greater airflow obstruction. To test these hypotheses quantitatively, our primary analysis employed multivariable mixed model regression to evaluate the association of snore loudness (dependent variable) with obstruction (flow:drive) and ventilatory drive in mutually adjusted analysis with subject included as a random effect. Quadratic terms (fixed effects) were included based on observed nonlinear associations and interaction effects. We also included a fixed effect variable representing severe obstruction (flow:drive <20%) to account for the observed unique reversal in snore loudness at this obstruction threshold (Figure 2). Finally, we included the interaction between the squared ventilatory drive and apnea because this provided optimal model fit to the rapid decrease in snore loudness with severe obstruction (Figure 2, left). Models were built using all breath-level data during scored sleep and arousals (i.e., wake data removed).

Figure 2.

There was an inverted U-shaped relationship between snore loudness and percent airway obstruction whereby snores became louder with increasing obstruction (i.e., the ratio of ventilation to ventilatory drive [flow:drive], or actual ventilation as a percentage of intended ventilation), but only up to a threshold of obstruction; thereafter, snores become quieter. The left panel illustrates that the model (black) was able to capture the inverted U-shaped relationship between snore loudness and obstruction observed in the actual data (gray) at increased ventilatory drive (solid line, drive between 175% and 225% of eupnea) and a normal ventilatory drive range (dashed line, drive between 75% and 125% of eupnea). The right panel shows changes in modeled snore loudness with obstruction at each level of ventilatory drive. The change in line colors from “cold” (black) to “hot” (bright yellow) represents increasing levels of ventilatory drive. Snore loudness was expressed as a change from 82 dB, which was the estimated tracheal sound loudness value for unobstructed eupneic breathing (drive = 100%, flow:drive = 100%).

To evaluate the differences in snore loudness across sleep-disordered breathing categories, linear mixed effect analysis described snore loudness (dependent variable) across apneas, hypopneas, arousals, and stable flow-limited breathing (fixed effects) versus stable unobstructed breathing (reference). Subject was included as a random effect. Each patient provided multiple breaths for each condition.

To evaluate the differences in snore loudness across sleep stages, a similar model described snore loudness as a function of stages N1 and N3 and REM sleep as fixed effects (stage N2 was used as the reference). Each patient contributed multiple breaths for each sleep stage.

Results

Baseline characteristics of the 40 included subjects are summarized in Table 1.

Table 1.

Patient characteristics (n = 40)

Characteristic	Value
Age, yr	55 (37–60)
Male sex	30
BMI, kg/m2	30.1 (28.5–35.9)
AHI per hour	23.1 (12.0–42.8)
Stable flow limitation, % of night	12.4 (1.2–23.0)
Stage N1, %	29.0 (20.7–53.3)
Stage N2, %	48.7 (37.7–65.3)
Stage N3, %	8.4 (0.5–18.0)
REM	6.6 (3.1–16.8)
Total sleep time, min	278.0 (206.2–345.8)

Definition of abbreviations: AHI = apnea-hypopnea index; BMI = body mass index; REM = rapid eye movement.

Data presented as median (interquartile range) where applicable.

Effect of Drive and Obstruction

As hypothesized, increased snore loudness was associated with increased ventilatory drive and greater pharyngeal obstruction (Figure 3). However, there was a substantial interaction effect between drive and obstruction. Specifically, increasing drive was strongly associated with louder snoring when in the presence of obstruction. For example, when the airway was 50% obstructed, snore loudness increased by 3.4 (95% confidence interval, 3.3–3.5) dB for every standard deviation (SD) change in ventilatory drive (SD = 85.8% of eupnea). On the contrary, when the airway was unobstructed (completely patent, flow:drive = 100%), snore loudness increased only 0.23 (0.08–0.39) dB for every SD change in ventilatory drive (Figure 4, left). Similarly, worsening pharyngeal obstruction was associated with louder snoring, but only in the presence of increased ventilatory drive. For instance, when ventilatory drive was increased to 200% of eupnea, snore loudness increased by 2.1 (2.0–2.2) dB per SD change in obstruction (SD = 30.6%). Conversely, at eupneic drive, snore loudness increased by only 0.14 (–0.08 to 0.28) dB per SD change in obstruction (Figure 4, right). Accordingly, snoring was loudest in the presence of both increased drive and increased obstruction. Interestingly, however, snoring became quieter when the airway reached >80% obstruction (flow:drive = 20%), suggesting that a certain amount of airflow is necessary to generate the snoring sound (the detailed model plot describing snore loudness as a function of obstruction and ventilatory drive is shown in Figures 1 and 2; the full model is summarized in Table E1 in the data supplement). As a result, extremely obstructed breaths (flow:drive <20%) were associated with a clear reduction in snore loudness (–9.9 [–13.3 to –6.6] dB) versus unobstructed eupneic breathing (see Table E1).

Figure 3.

Heat map describing the mixed model estimated effects of obstruction (ratio of ventilation to ventilatory drive [flow:drive]) and ventilatory drive on snore loudness (numbers within cells, in dB). Snores get louder (represented by “hotter” colors, yellow and white) with increasing drive and increasing obstruction (lower flow:drive). Cell outlines in black represent unobstructed eupneic breaths to which snore loudness is referenced. Note that extremely obstructed breathing, e.g., flow:drive <20%, is particularly quiet (i.e., −11 to −6 dB vs. unobstructed eupneic breathing) and that partially obstructed breathing at a higher drive is particularly loud (i.e., +1 to +12 dB). Model R² = 0.69.

Figure 4.

Increase in snore loudness in decibels per standard deviation (SD) change in ventilatory drive at varying levels of obstruction (left) and per SD change in obstruction at varying levels of drive (right). Estimates summarize the modeled effects of obstruction and ventilatory drive on snore loudness (Figure 3). SD values for ventilatory drive and the ratio of ventilation to ventilatory drive are 85.8% and 30.6%, respectively. flow:drive = ratio of ventilation to ventilatory drive.

Effect of Sleep-Disordered Breathing Type

Compared with reference breaths (unobstructed eupneic breathing during sleep), snoring was loudest during stable flow-limited breathing (+4.7 [4.5–4.9] dB), followed by hypopneas (+1.0 [0.78–1.3] dB), and quietest during apnea breaths (–14.1 [–14.4 to –13.7] dB; Table 2). The increased loudness during stable flow limitation versus hypopneas (+3.7 [3.5–3.9] dB) was accompanied by greater ventilatory drive (+0.55 [0.53–0.57] SD), as observed despite similar obstruction (+0.10 [0.09–0.11] SD) (Table E2).

Table 2.

Snore loudness across different manifestations of sleep-disordered breathing and sleep stages

Independent Variable	Estimate (95% CI)	P Value
Sleep-disordered breathing
Intercept	2.4 (‒1.19 to 5.9)
Apnea	‒14.1 (‒14.4 to ‒13.7)	<10‒9
Hypopnea	1.0 (0.78 to 1.3)	<10‒9
Stable flow limitation	4.7 (4.5 to 4.9)	<10‒9
Arousal	‒2.4 (‒2.6 to ‒2.1)	<10‒9
Sleep stage
Intercept	0.04 (‒3.5 to 3.6)
N1	‒4.1 (‒4.2 to ‒3.9)	<10‒9
N3	3.7 (3.5 to 3.9)	<10‒9
REM	‒5.4 (‒5.6 to ‒5.2)	<10‒9

Definition of abbreviations: CI = confidence interval; REM = rapid eye movement.

Snore loudness was expressed as a change from 82 dB, which was the estimated tracheal sound loudness value for unobstructed eupneic breathing. The model intercept represents snore loudness during conditions of stable non–flow-limited breathing (top part of table) and stage N2 sleep (bottom part). Estimates represent decibel change in snore loudness relative to the intercept condition.

Effect of Sleep Stage

Compared with breaths during stage N2 sleep (reference), snoring was quietest in REM sleep (–5.4 [–5.6 to –5.2] dB) and stage N1 (–4.1 [–3.9 to –3.2] dB), when drive is known to be reduced, and it was loudest in stage N3 (+3.7 [3.5–3.8] dB), when drive is known to be increased. Indeed, the quieter snoring in REM sleep and stage N1 were accompanied by lower drive (–0.52 [–0.54 to –0.52] and –0.32 [–0.33 to –0.30] SD changes in drive, respectively), whereas the louder snoring in stage N3 was accompanied by modestly increased drive (+0.13 [0.11–0.15]). These differences were observed despite no meaningful change in obstruction across stages (<0.05 SD changes in flow:drive).

Discussion

The present study demonstrates that the amplitude of snore sounds is influenced by ventilatory drive and the degree of pharyngeal obstruction. In particular, the intensity of snoring increases with higher ventilatory drive levels, with an 86% eupnea increase in drive (1 SD) yielding a 3.4-dB increase in loudness (at 50% obstruction). As expected, snoring intensity also increased with increasing pharyngeal obstruction, with a 31% increase in obstruction (1 SD) yielding a 2.1-dB increase in loudness (at 200% of baseline, or eupneic, ventilatory drive). Importantly, drive and obstruction interact to determine the snoring intensity. Specifically, snoring is at its loudest in the presence of greater drive and greater obstruction, but then decreases in the presence of extreme obstruction. Accordingly, increasing drive in the absence of obstruction had little impact on snore loudness. Likewise, worsening obstruction at normal drive had a minimal effect on snoring. These findings highlight that ventilatory drive has an equal role to pharyngeal obstruction (if not a greater one) in generating loud snoring. Underscoring the key role of drive in snoring intensity, we found that snoring is loudest during periods characterized by increased ventilatory drive (and similar levels of obstruction), i.e., stable flow-limited breathing versus hypopneas and stage N3 versus stage N2 sleep. Likewise, snoring is quietest during REM and stage N1 sleep, when ventilatory drive is reduced, compared with stage N2 sleep.

Novel Physiological Insight

Snoring and respiratory drive/effort

Previous studies have established that periods of snoring are accompanied by more negative esophageal pressures (presumably with greater ventilatory drive) compared with nonsnoring periods (−17.8 vs. −8.4 cm H₂O) (9, 21). Notably, the previous work did not quantify obstruction or evaluate the interaction between obstruction and ventilatory drive on snore loudness. Further, these studies are limited by the use of esophageal pressure as a surrogate measure of ventilatory drive, as pressure swings tend to exaggerate ventilatory drive estimation in the presence of obstruction (22). Our study builds on their findings by separately quantifying the degree to which snore loudness is independently modulated by drive (as measured by diaphragm EMG), controlling for pharyngeal obstruction. Using a direct measure of neural ventilatory drive, we confirm that snores get louder with increased ventilatory drive in the presence of obstruction, such that changes in drive, in addition to changes in obstruction severity, can explain why snoring appears or disappears under different conditions throughout the night.

Snoring and pharyngeal obstruction

It is a well-established observation that snoring occurs during periods of high resistance to airflow, which implies that snores are produced and modulated by airway obstruction. Most of the evidence supporting this is based on the physics of turbulent flow over a flexible structure (i.e., pharyngeal tissues), which produces oscillations that generate sound (i.e., snores) (8, 23, 24). To the best of our knowledge, there are no direct in vivo assessments indicating that snores are louder with increased pharyngeal obstruction. An indirect in vivo observation that a narrower airway produces louder snoring was shown in a study that captured images of the airway during awake computed tomography and correlated those images with snore loudness from a questionnaire (25). Despite methodological limitations in determining snore loudness, the authors demonstrated that the pharyngeal narrowing ratio, i.e., the ratio between the nasal area (at the hard palate level) and the narrowest pharyngeal cross-section (behind the soft palate) was higher (i.e., more narrowed) in snorers than in nonsnorers and higher in loud snorers than in moderate snorers. Furthermore, Maimon and Hanly also showed that snore loudness was positively correlated, albeit weakly, with OSA severity (26), implying that increased pharyngeal obstruction accompanying more severe respiratory events produces louder snores. However, it is known that more severe events are also accompanied by increased ventilatory drive, particularly at the end of the events (27). Therefore, from this study, we cannot rule out that increased ventilatory drive explains louder snoring in patients with more severe OSA. The present study provides compelling evidence that an obstructed airway, independent of ventilatory drive, contributes to loud snoring. More precisely, at all levels of ventilatory drive above eupnea, snores get louder with worsening obstruction until the airway becomes almost completely closed. Moreover, the effect of obstruction on snore loudness is stronger as ventilatory drive increases.

Reduced snoring during apnea and severe obstruction

The present study detailed the reversal of snoring intensity during very high levels of obstruction (i.e., pharynx >80% obstructed). Breaths during this level of obstruction were primarily scored apneas (75% of breaths) but also included severely obstructed hypopneas and unscored breaths (25% of breaths). Silence during apnea breaths is commonly observed in sleep studies, and, indeed, the present findings determined that scored apneas are 14 dB quieter than stable unobstructed breathing (Table 2). However, we point out that snores become quieter at a nonzero threshold of obstruction (flow:drive = 20%), which includes nonapnea breaths, indicating that this reversal in snore loudness is not simply explained by complete airway occlusion during apnea, when there is zero flow. We consider two potential mechanisms contributing to this reversal: 1) reduced airflow velocity as the airway transitions toward apnea and 2) airway folding, which may stiffen the pharyngeal walls. More specifically, as the airway transitions from unobstructed to severely obstructed, there is typically a profound reduction in airway cross-sectional area (e.g., approximately 90%) accompanied by a more modest decrease in flow rate (e.g., approximately 70%), in which case airflow velocity (namely flow rate per cross-sectional area) must increase and is expected to generate louder snoring (7, 8, 22). Subsequently, as the airway transitions from severe obstruction toward apnea, the reduction in airflow (i.e., the final 30%) is more substantial than the reduction in airway area (i.e., the final 10%), resulting in a reduction in airflow velocity, which is expected to contribute to a decrease in the loudness of snoring. Contemporaneously, folding of the pharyngeal walls is expected to accompany severe obstruction in the transition to apnea, imparting stiffness to the pharynx, which, in principle, should resist vibration and snoring. Previous studies have demonstrated the phenomenon of folding in the upper airway imparting stiffness. Amatoury and coworkers subjected a Penrose tube, with air flowing through it, to various (external) chamber pressures to produce wall strain and simulate flow limitation (28). They demonstrated that, as the airway area approached zero and folding of the airway occurred, tubes became stiffer (i.e., less strain generated per increase in chamber pressure), especially in a trifold pattern. Similarly, in sleeping humans with OSA, Genta and colleagues showed the same pattern of increased airway stiffness as airway area approached zero (i.e., reduced airway strain per absolute increase in driving pressure) (29) and noted that airway folding was evident in all patients studied. Regardless of the precise mechanism, in the context of obstruction, when snoring becomes markedly attenuated, the airway does not necessarily have to be completely occluded.

Snoring and different forms of sleep-disordered breathing

The present study demonstrated that the form of sleep-disordered breathing with the loudest snoring was (stable) flow limitation, which was 3.7 dB louder than during hypopneas. No studies have specifically described within-patient differences in snore loudness between respiratory events and stable flow-limited breathing. Given our new framework for understanding snore loudness, we believe loud snoring during stable breathing is likely explained by the increased drive in the presence of pharyngeal obstruction, which is enabling patients to resist any reduction in ventilation, thereby preventing respiratory events, yielding stable breathing (30, 31). Indeed, in the data supplement, we show that drive during stable flow-limited breathing was increased 71% above eupneic drive. By comparison, drive during hypopnea was increased only 28% above eupneic drive. Perhaps unexpectedly, the severity of obstruction (per flow:drive), although increased, is similar during hypopneas and stable flow limitation. Taken together, despite similar levels of pharyngeal obstruction during stable flow limitation and hypopneas, snores are louder during stable flow limitation as a result of the presence of increased ventilatory drive.

Snoring and sleep stage

Previous research on the relationship between snoring severity and sleep stages aligns with the present study, reporting increased snore frequency during stage N3 and lower snore frequency during REM sleep (6, 7). It may seem counterintuitive that snore frequency would be greater during stage N3 sleep, when the dilator muscles are relatively more active, than during REM sleep, when the muscles are more hypotonic (1–3). However, in the present study, we show that it is the differences in ventilatory drive between these two sleep stages that explain their respective differences in the severity of snoring. Specifically, we demonstrate that REM sleep (and stage N1 sleep), characterized by an overall reduction in ventilatory drive (13), is associated with marked reduction in snore loudness compared with stage N2 sleep. Conversely, stage N3 sleep, a sleep state characterized by augmented ventilatory drive (11), produced markedly louder snoring. In the data supplement, we confirm that, relative to stage N2 sleep, ventilatory drive is markedly lower during stage N1 and REM sleep and modestly increased during stage N3 sleep despite minimal change in obstruction across stages (<0.05 SD change), supporting the concept that the differences in snore loudness across sleep states is likely modulated by ventilatory drive.

Individual differences in snore loudness

Even after considering differences in drive and pharyngeal obstruction, our study revealed substantial unexplained variability in snore loudness across subjects, with an 11-dB SD in the estimated snore loudness (Figure E1). Thus, although increasing drive and obstruction may yield predictable changes in snore loudness within a patient, “baseline” snoring levels (e.g., at normal drive and increased obstruction) vary substantially across patients. This finding raises doubts about the use of snore loudness levels across patients to provide insight into flow limitation because a cutoff suitable for those with loud baseline levels will fail to detect pharyngeal compromise in others with quieter snoring patterns. This is a significant clinical concern because snoring has been linked to adverse outcomes in adults (32, 33) and children (34). Quiet snoring that goes undetected could mask the risks of these outcomes. Further research is needed to determine whether the risk of adverse outcomes is related to absolute snore volume or changes in snore loudness relative to a patient-specific landmark, such as apnea or normal breathing. If the latter is true, snore detection should be quantified based on normalization to such a landmark within each patient.

Conclusions

Here we provide a framework for understanding the physiological factors explaining snore loudness. The framework highlights that ventilatory drive, and not simply pharyngeal obstruction, is an important modulator of snore loudness. In fact, variability in ventilatory drive tends to explain changes in snore loudness through different sleep stages (e.g., loudest in stage N3 and quietest in stage N1 and REM sleep) and forms of sleep-disordered breathing (e.g., loudest during stable flow-limited breathing). Finally, we found that some patients appear to inherently snore louder than others after controlling for physiological variables (i.e., drive and obstruction), which has clinical implications for defining loudness thresholds that identify pathophysiological airway obstruction. This new framework for understanding the severity of snoring is important in simply understanding the physiology, and for the development of technologies that use snore sounds to characterize sleep-disordered breathing.