Acoustic Snoring Analysis Can Provide Important Information in OSA Diagnosis

The electroencephalogram (EEG) was first reported from a human scalp in 1929, by Hans Berger.¹ The importance of eye movements and electrooculogram (EOG) recording, as well as the discovery of REM sleep in 1952, led to an exponential interest in sleep studies, and eventually to the manual by Rechtschaffen and Kales published in 1968, providing rules for scoring polysomnographic sleep studies. Since then, for about 4 decades, the science and clinical aspects of sleep and sleep disturbances have been based on this manual, with only minor modifications. Starting some 15 years ago, researchers as well as the American Academy of Sleep Medicine (AASM) started to question the accuracy and clinical relevance of the 1968 paradigms. This resulted in the AASM's publishing of a new scoring manual in 2007.² However, this manual, although providing novel scoring guidelines, is still based largely on the traditional biosignals, predominantly EEG, EOG, EMG, respiratory channels (flow and effort), and ECG, with some distinction between flow assessed by thermistors as compared to that assessed by nasal pressure.

During the last decade, researchers have searched for novel informative biosignals. Sophisticated ECG analyses, predominantly of heart rate variability (HRV), peripheral arterial tone (PAT), and pulse transit time (PTT) have shown the potential importance of information gleaned by recording autonomic nervous system (ANS) output. Although it is generally agreed that sleep architecture can be detected solely based on central nervous system output, in recent years it has increasingly been shown that sophisticated analyses of the ANS can lead to accurate sleep architecture determination.^3–5 Furthermore, quantification of ANS output can also accurately detect sleep fragmentation^6,7 as well as sleep disordered breathing events,^8,9 since it has been well documented that the termination of any SDB event is associated with substantial, abrupt, and immediate sympathetic activation.^10,11

Interest in breath sounds and the acoustic characteristics of snores began almost two decades ago.¹² Since then, there have been several reports attempting to shed light on respiratory problems based on analyses of acoustic signals recorded from various points on the chest wall and trachea. The study by Ben Israel and colleagues in this issue of SLEEP¹³ is the first time that analysis of snoring sounds alone have been shown to be accurate in quantifying the severity of SDB, expressed as the apnea hypopnea index (AHI). The authors studied 90 consecutive patients evaluated at the sleep lab for suspected obstructive sleep apnea (OSA) and simultaneously recorded their sleep by both regular PSG channels and by snoring sound investigation. The first 60 patients served as a training set in order to develop the algorithm for scoring apnea and hypopnea, while the remaining 30 patients were blindly analyzed by both methods to validate the algorithm for quantifying AHI. Results showed good agreement between both methods, as expressed by correlation analyses, Bland Altman analyses, and White Westbrook analyses. The strengths of the study by Ben Israel et al. are the relatively high number of participants and the novelty of sophisticated analyses based solely on snoring sounds. The major weakness of the study is that comparisons were based on whole night recording rather than on single events analyses. Thus, it is possible that some events were overscored and some were underscored with a reasonable overnight average, but it tells us less about event-by-event comparisons. However, the concept of detecting SDB events based solely on snoring signal recording from 1 meter above the patient's bed is interesting and may be potentially applicable for in-home detection of SDB.

Many recent studies and devices have raised the possibility of shifting from in-lab PSG sleep studies to ambulatory home studies for clinical purposes. This shift has the advantage of studying the patients in their own environment, as well as cutting down waiting lists and costs. Furthermore, these methods are easier than in-lab PSG recording and increase accessibility to sleep studies especially in rural regions far from hospitals with sleep labs, and they can be applied in extreme conditions.¹⁴ The disadvantages are that these are unattended studies (potentially associated with higher failure rate) and with limited channels recorded. Eleven years ago, Reuveni et al. predicted that home monitoring would not be cost-effective unless the classical electrode attachment system was changed.¹⁵ Recent studies investigating the entire process of both diagnosis and treatment of OSA with a lab arm on one side and a home arm on the other have shown that home recordings demonstrate comparable outcomes to the lab recordings while reducing costs.^16,17 Thus, the study by Ben Israel and colleagues in this issue of SLEEP adds another potential signal analysis that further expands the horizon for easily recorded informative data in the home, and that may potentially be incorporated into ambulatory devices or even serve as a single channel in a future novel home recorder. One unique aspect of snoring sound analysis is that no wires are attached to the patient' This may be very relevant and a breakthrough for ambulatory studies in children, although the technique would have to be validated in children. Children move a lot and are unlikely to cooperate with electrode attachment, especially in the 2- to 4-year age group, when the prevalence of OSA is the highest, and to date there are no good ambulatory devices for pediatric OSA assessment. Currently this requires in-lab attended studies, in order to adjust and re-attach electrodes that have been removed by the child. However, if snoring signal analysis can indeed accurately quantify OSA in children as well (children were not studied by Ben Israel et al.¹³), this could be a profoundly important contribution to the field of pediatric sleep medicine.

In summary, it appears that sophisticated analyses of signals of autonomic nervous system output and/or snoring sound may accurately quantify SDB severity. More research and development are required to clarify whether snoring sound analyses can be developed into a reliable method to improve OSA diagnosis especially in the home environment. With additional signals and better analyses, eventually most cases of OSA in the clinical setting will be diagnosed and treated in the home. This is in concert with other aspects of medical care that are being shifted from hospital to home and matches world demographic trends of increased population size with a reduced ratio of hospital beds per person.

CITATION

Pillar G; Etzioni T; Katz N. Acoustic snoring analysis can provide important information in OSA diagnosis. SLEEP 2012;35(9):1195-1196.

DISCLOSURE STATEMENT

The authors have indicated no financial conflicts of interest.