International Consensus Statement on Obstructive Sleep Apnea

Abstract

Background:

Evaluation and interpretation of the literature on obstructive sleep apnea (OSA) allows for consolidation and determination of the key factors important for clinical management of the adult OSA patient. Toward this goal, an international collaborative of multidisciplinary experts in sleep apnea evaluation and treatment have produced the International Consensus statement on Obstructive Sleep Apnea (ICS:OSA).

Methods:

Using previously defined methodology, focal topics in OSA were assigned as literature review (LR), evidence-based review (EBR), or evidence-based review with recommendations (EBR-R) formats. Each topic incorporated the available and relevant evidence which was summarized and graded on study quality. Each topic and section underwent iterative review and the ICS:OSA was created and reviewed by all authors for consensus.

Results:

The ICS:OSA addresses OSA syndrome definitions, pathophysiology, epidemiology, risk factors for disease, screening methods, diagnostic testing types, multiple treatment modalities, and effects of OSA treatment on multiple OSA-associated comorbidities. Specific focus on outcomes with positive airway pressure (PAP) and surgical treatments were evaluated.

Conclusion:

This review of the literature consolidates the available knowledge and identifies the limitations of the current evidence on OSA. This effort aims to create a resource for OSA evidence-based practice and identify future research needs. Knowledge gaps and research opportunities include improving the metrics of OSA disease, determining the optimal OSA screening paradigms, developing strategies for PAP adherence and longitudinal care, enhancing selection of PAP alternatives and surgery, understanding health risk outcomes, and translating evidence into individualized approaches to therapy.

I |. INTRODUCTION

Obstructive sleep apnea (OSA) is a complex and multifaceted disease with multiple associated symptoms and comorbidities. Work performed within the last decade has contributed to expanding knowledge of disease incidence, new approaches to diagnosis, and novel improvements in therapeutic options. OSA was first defined in 1965.^1,2 For many years, the only therapy was tracheotomy tube placement to bypass upper airway (UA) obstruction. Continuous positive airway pressure (CPAP) therapy was introduced in 1981³ and marked a pivotal discovery in OSA treatment. Since that time, growth in the literature and understanding of OSA as a heterogeneous and complex chronic disease has been exponential. Our abilities to diagnose OSA and to determine its far-reaching consequences have advanced significantly. OSA is currently recognized as a common and important major health issue, imposing a large cost on health systems around the world.

This International Consensus Statement on Obstructive Sleep Apnea (ICS:OSA) was created to summarize the best available evidence into a format that allows clinicians to examine diagnosis and management options for adult OSA, to understand the quality of evidence, and to translate the findings and recommendations into evidence-based care. Contributions came from more than 130 international authors from various OSA specialties including neurology, pulmonology, sleep medicine, otolaryngology, oral-maxillofacial surgery, dentistry, anesthesiology, psychiatry, cardiology, and sleep physiology. The specialists contributing to this statement represent a diverse set of expertise that encompass the multidisciplinary approach necessary to understand and treat OSA. Topics on OSA were assigned to experts who utilized a structured review process to evaluate and interpret the evidence. The ICS:OSA recommendations for diagnosis and management of OSA rely directly on the reviewed evidence with delineations of the benefits, harms, and costs that were considered for each recommendation.

This document highlights the current understanding and impact of OSA in adult patients. The ICS:OSA utilizes an evidence-based format defined by the International Consensus Statement on Allergy and Rhinology: Rhinosinusitis in 2016 (ICAR-RS-2016),⁴ which was adapted from a framework⁵ that uses a blinded iterative review process. This method has been used for other subjects including allergic rhinitis (AR), skull base surgery, and olfaction. For ICS:OSA, experts in the fields of sleep medicine and sleep surgery contributed to its creation as both section authors and blinded reviewers of other sections. Each section attempts to emphasize published, peer-reviewed evidence where available and identify gaps in knowledge.

The ICS:OSA is not a clinical practice guideline (CPG) and does not employ the steps of a CPG creation. The ICS:OSA includes meta-analyses and other systematic reviews (SRs) when available for specific OSA topic areas but does not perform separate or new meta-analyses on the data summarized. This document aims to summarize and consolidate the best available knowledge on the diagnosis and treatment of OSA, to provide a standardized format to display the evidence, and to allow for alternative interpretations.

Recommendations exhibited in the ICS:OSA are based upon the best available evidence, but for many topics the level and quality of evidence are variable or weak. Summary recommendations should be assessed in the context of the evidence on which they are based and the populations of the studies themselves especially when attempting to translate the data for individualized recommendations. Recommendations in the ICS:OSA do not define standard of care or medical necessity and cannot dictate care of an individual patient. Variability in the presentation, symptoms, treatment responses, and tolerance levels of therapy is appreciated for all complex diseases and OSA is no exception.

The best evidence-based practice utilizes shareddecision making approaches that incorporate evidence with individual patient factors, values, expectations, and goals in creating individualized clinical decisions and recommendations. New and future research in OSA should aim to fill the knowledge gaps and strengthen the evidence that moves us toward optimal care of the OSA patient. As new and stronger evidence is examined, summary recommendations will require reevaluation and updates.

II |. METHODS

II.A |. Topic Development

The ICS:OSA document focused on incorporation and summarization of the published literature. The methodology for ICS:OSA followed that of prior International Consensus in Allergy and Rhinology documents,^4,6,7 which involved a process adapted by Rudmik and Smith.⁵ The approach aims to maximize impact of published evidence by systematically evaluating the literature, grading the evidence, and creating evidence-based recommendations.

The ICS:OSA was divided into over 150 topics, each topic was assigned to a senior author who is a recognized expert in care of OSA patients. Topic generation spanned definitions of respiratory events in polysomnogram testing, controversies in different scoring definitions, epidemiology of disease, economic burden, risk factors, contributory factors for pathogenesis of OSA, diagnosis and screening tools, diagnostic testing modalities, medical comorbidities, medical management, and surgical management for OSA. A focus of the ICS:OSA included the many cardiovascular (CV), cognitive, and metabolic comorbidities associated with OSA which impact OSA screening and management. Separate sections were created to examine the evidence on the effects of PAP and surgical therapy for improving OSA-related symptoms and comorbidity risks.

A few topics based on disease definition or background information were assigned as literature reviews (LRs). Certain topics were not appropriate or lacked sufficient evidence and were assigned as evidence-based reviews (EBRs). Other topics had evidence to inform clinical recommendations were assigned as evidence-based reviews with recommendations (EBR-R).

For each topic, authors were asked to perform an SR of the literature using Ovid MEDLINE (1947-December 2019), EMBASE (1974-December 2019), and Cochrane Review databases. The Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) standardized guidelines⁸ were used to inform the SRs. Randomized controlled trials (RCTs), meta-analyses, SRs, and published guidelines were first identified to provide the strongest level of evidence (LOE) if available. When these were not available, observational studies were used. Identified studies were included to ensure relevant studies were captured. The majority of the studies included were written in English. If the authors felt a non-English study should be included, the paper was translated and summarized. Important evolving topics did include papers after December 2019 if the paper significantly contributed new understanding or influenced the recommendations.

For EBR and EBR-R sections, relevant studies were displayed in a standardized format with quality of each study graded using the Oxford LOE (LOE Level 1a–5).⁹ Next an aggregate grade of evidence (A–D) was determined based on guidelines from the American Academy of Pediatrics Steering Committee on Quality Improvement and Management (AAP SCQIM)¹⁰ (see Table II.A.1). When appropriate, a recommendation was written using the AAP SCQIM (Table II.A.2). Each recommendation was based on the aggregate LOE along with an assessment of benefit, harm, and costs related to the specific topic.

TABLE II. A. 1.

Aggregate grade of evidence

Grade	Research quality
A	Well-designed RCTs
B	RCTs with minor limitations Overwhelming consistent evidence from observational studies
C	Observational studies (case–control and cohort design)
D	Expert opinion Case reports Reasoning from first principles

TABLE II.A.2.

AAP defined strategy for recommendation development¹⁰

Evidence quality	Preponderance of benefijit over harm	Balance of benefijit and harm	Preponderance of harm over benefijit
A. Well-designed RCT’s	Strong recommendation	Option	Strong recommendation against
B. RCT’s with minor limitations; Overwhelmingly consistent evidence from observational studies	Recommendation
C. Observational studies (case–control and cohort design)			Recommendation against
D. Expert opinion, case reports, Reasoning from fijirst principles	Option	No recommendation

II.B |. Iterative Review

Following the development of each ICS:OSA section in LR, EBR, or EBR-R formats, the section underwent a two-stage iterative review process using at least two independent reviewers. The purpose of the reviews was to evaluate the completeness of literature identification, determine accuracy of the grade of evidence, and ensure the recommendations were appropriate. Most sections were reviewed across disciplines. Following the review process, changes were agreed upon by both reviewers and initial author(s).

II.C |. ICS:OSA Statement Development

After review and edits were completed, the principal editors (J.L.C., A.N.G., and I.R.) synthesized all sections into the ICS:OSA statement. The document draft was then reviewed by all contributing authors. Once consensus was achieved on literature completeness and final recommendations, the final ICS:OSA statement was produced.

As each topic was authored by individuals, search results and level of evidence grading may vary and this document does not present every study published on every topic. Most sections do not include an exhaustive list of all studies ever performed and authors were given the ability to select the most relevant evidence. For certain topics, the literature is extensive and the section focused mainly on high-quality studies or SRs. The editors also made an effort to ensure recommendations aligned with published guidelines.

II.D |. Possible Adverse Effects of OSA Interventions

Throughout ICS:OSA, possible side effects or risks of testing or interventions were considered. In order to standardize a collection of these possible side effects and adverse effects, Table II.D defines typical adverse effects for a variety of OSA interventions for both the immediate and the long term. Risks for these effects should be considered when determining policy level recommendations. Each intervention has the additional risk of insufficient improvement of symptoms and continued OSA disease. Table II.D may not include all possible risks.

TABLE II.D.

Risks, side effects and adverse effects of common obstructive sleep apnea interventions^a

Intervention	Possible side effects and adverse effects
Home sleep study testing	False negative result, diagnosis delay, sensor discomfort or failure, skin irritation (from adhesives).
In-laboratory sleep study testing	Discomfort, skin irritation (from adhesives), sensor discomfort, or failure.
Continuous positive airway pressure therapy or automatic positive airway pressure therapy	Discomfort, nasal complaints, oral dryness, skin irritation, allergy to mask materials, poor sleep quality, aerophagia, claustrophobia, mask leak, epistaxis, dizziness.
Bilevel positive airway pressure therapy	Discomfort, nasal dryness, oral dryness, skin irritation, poor sleep quality, aerophagia, claustrophobia, mask leak, epistaxis, dizziness.
Oral appliance	Discomfort, occlusal changes, jaw or temporomandibular joint pain, tongue irritation, excessive salivation, mouth dryness, damage to teeth.
Drug-induced sleep endoscopy	Excessive sedation, desaturation, laryngospasm, risk of anesthesia,b loss of airway, need for intubation, epistaxis.
Tonsillectomy	Acute postoperative pain and swelling, severe pain, temporary dysphagia, postoperative bleeding, infection, risk of anesthesia.b
Soft palate surgery (i.e., UPPP, ESP,LP)	Acute postoperative pain and swelling, dysphagia, taste change, postoperative bleeding, infection, temporary or permanent velopharyngeal insufficiency, long term globus sensation, pharyngeal dryness, oronasal fistula, risk of anesthesiab, death. UPPP surgery associated with 1.6% major complication rate and 0.09% mortality rate.
Genioglossus advancement	Acute postoperative pain and swelling, mandibular fracture, neurosensory changes, mandibular fracture, dental injury or dental pulp necrosis, postoperative bleeding or floor of mouth hematoma, infection, risk of anesthesia.b
Hyoid suspension	Acute postoperative pain and swelling, temporary dysphagia, infection, postoperative bleeding, infection, risk of anesthesia.b
Base of tongue surgery or lingual tonsillectomy	Acute postoperative pain, postoperative bleeding, infection, dysphagia, globus sensation, taste loss or change, tongue numbness, pharyngeal scarring, risk of anesthesia.b
Epiglottis surgery	Acute postoperative pain, postoperative bleeding, infection, dysphagia, changes in speech, taste change aspiration, globus, risk of anesthesia.b
Nasal and sinus surgery	Acute postoperative pain, postoperative bleeding, infection, orbital injury, septal perforation, cosmetic changes, lacrimal system injury, hyposmia/anosmia, vision changes or blindness, intracranial injury, cerebrospinal fluid leak, risk of anesthesia.b
Jaw surgery	Occlusal changes, facial neurosensory deficits, cosmetic changes in facial profile and structure, infection, bleeding, subcutaneous emphysema, malunion or nonhealing of the mandible, temporomandibular related pain, risk of anesthesia.b
Hypoglossal nerve stimulation surgery	Acute postoperative pain, postoperative bleeding, infection, tongue stiffness, tongue abrasion, transient tongue hemiparesis, lip weakness, device malfunction requiring revision surgery, device migration, pneumothorax, discomfort with use, risk of anesthesia.b
Tracheostomy	Acute postoperative pain, postoperative bleeding, infection, tube obstruction, tube displacement, tracheoesophageal fistula, pneumothorax, tracheal stenosis, trachea-innominate fistula, thyroid injury, voice changes, risk of anesthesia.b
Bariatric surgery	Acute postoperative pain, postoperative bleeding, infection, nausea, vomiting, gastroesophageal reflux, mucosal ulcers, gallstones, anastomotic leak, chronic malnutrition, dumping syndrome, hernia, need for revision surgery, risk of anesthesia.b

^aMay not include all possible risks of listed interventions. Surgical risks include common expected postoperative symptoms and rare long-term risks. All interventions have the risk for insufficient improvement in snoring and OSA.

^bIncludes risk of sore throat, nausea, vomiting, damage to teeth, laryngospasm, aspiration, anaphylaxis, hypoxia, respiratory failure, cardiovascular collapse, or death.

III |. DEFINITIONS AND CLASSIFICATIONS RELATED TO OSA

III.A |. Sleep-Disordered Breathing

Sleep-disordered breathing (SDB) refers to a range of sleep-related breathing disorders that includes primary snoring, OSA, central sleep apnea (CSA), Cheyne-Stokes respiration, and sleep-related hypoventilation. The risk factors and underlying pathophysiological mechanisms for these disorders have overlapping as well as unique features. Each is associated with impaired ventilation during sleep and sleep disruption, although they differ in the underlying severity of gas exchange abnormalities, anatomic obstruction, and degree of underlying ventilatory control abnormality.

OSA is the most common sleep disorder in adults that is evaluated at sleep centers. It is defined on the basis of nighttime and daytime symptoms as well as objective data from a sleep study. Objective sleep testing, with the use of a home sleep apnea test (HSAT) or full polysomnography (PSG), records multiple channels of physiological data that allows identification of the key respiratory-related events that are used to assess SDB: apneas, hypopneas, and respiratory-event related arousals (RERAs), patterns of oxyhemoglobin saturation, sleep-state related parameters (in PSG only), and body position.

The American Academy of Sleep Medicine (AASM) publishes and regularly updates the guidelines that specify the technical requirements for sleep study data acquisition and scoring.¹¹ Since 1999, these have been updated to address advances in technology (e.g., use of nasal pressure signals for airflow detection) as well as changes in hypopnea definitions (reflecting both updates to the evidence and pragmatic issues in operationalizing alternative definitions).

III.A.1 |. Respiratory event definitions

In 2012, the AASM updated the 2007 respiratory event scoring rules,¹² which continue to be clarified.¹¹ Note that the scoring of some events (e.g., hypopneas, RERAs) remains controversial, resulting in Recommended and Acceptable definitions and Optional recommendations. Definitions of relevant respiratory event subtypes are largely based on expert consensus and are summarized.

III.A.1.a |. Apnea

A drop in peak signal excursion by ≥90% of pre-event baseline for ≥10 s using an oronasal thermal signal (recommended sensor), positive airway pressure (PAP) device flow, or an alternative apnea sensor. No requirement for a desaturation or an arousal.

III.A.1.b |. Hypopnea

Recommended definition (AASM definition): A drop in peak signal excursion by ≥30% of pre-event baseline for ≥10 s using nasal pressure (recommended sensor), PAP device flow, or an alternative hypopnea sensor, AND a ≥3% oxygen desaturation from the pre-event baseline OR the event is associated with an electroencephalogram (EEG, cortical) arousal.

Alternative definition: A drop in peak signal excursion by ≥30% of pre-event baseline for ≥10 s using nasal pressure (recommended sensor), PAP device flow, or an alternative hypopnea sensor, AND a ≥4% oxygen desaturation from the pre-event baseline. This is the current definition used by Centers for Medicaid and Medicare Services (CMS).

III.A.1.c |. Respiratory Effort Related Arousal (RERA)

Sequence of breaths lasting ≥10 s characterized by increasing respiratory effort or by flattening of the inspiratory portion of the flow signal leading to an arousal from sleep when the sequence of breaths does not meet criteria for an apnea or hypopnea. Scoring of this event type is considered optional. Of note, the revised recommended hypopnea rules allow hypopneas to be recognized if associated with an arousal, thus identifying many events that previously would have been missed without the RERA classification.

Note: Apneas and hypopneas require comparison of breathing amplitude relative to a “baseline.” The AASM 2007 and 2012 manuals operationalize baseline as: “mean amplitude of stable breathing and oxygenation in the 2 min preceding the onset of the event (in individuals who have a stable breathing pattern during sleep) or the mean amplitude of the three largest breaths in the 2 min preceding onset of the event (in individuals without a stable breathing pattern).”

III.A.2 |. Summary metrics

Summary metrics of OSA severity tabulate the frequency of breathing disturbances to (1) use as thresholds for defining disease; and (2) assess disease severity, with higher indices considered to reflect more severe disease.

III.A.2.a |. Apnea hypopnea index (AHI)

The AHI is calculated as the number of respiratory events (apneas, hypopneas) divided by the number of hours of sleep documented during a PSG study. It is important to distinguish if all respiratory events are included (central and obstructive) or only obstructive events. Ideally, a total AHI inclusive of central and obstructive events would be reported along with a separate total for central apnea index.

III.A.2.b |. Respiratory event index (REI)

The REI is calculated as the number of respiratory events divided by the number of hours of estimated sleep using a HSAT. The recommended approach for estimating sleep time is by editing from the total recording time the periods when the participant is likely awake, as evidenced by artifact, movement, and characteristic changes in heart rate and breathing.

III.A.2.c |. Respiratory disturbance index (RDI)

The RDI is calculated as the number of respiratory events (apneas, hypopneas, and RERAs) divided by the number of hours of sleep documented during a PSG study.

III.A.2.d |. Oxygen desaturation index (ODI)

This metric refers to the number of dips (3% or 4%; ODI3%, ODI4%) in oxygen saturation (SaO₂) relative toa local baseline per hour of sleep from an overnight sleep study. Unlike the AHI, the ODI is almost always automatically derived using analysis software. Software programs utilize various algorithms for averaging SaO₂ signals, defining local baseline SaO₂, requiring minimal durations of oxygen fall, and identifying and excluding artifacts.

III.A.3 |. Considerations in sleep study scoring and data interpretation

Event features:

There are three key features of event definitions that influence prevalence estimates of OSA: (1) Degree of associated oxygen desaturation, (2) Use of event-associated EEG arousal, and (3) Amplitude of breathing reduction (magnitude and duration). These dimensions relate to the prognostic importance of the features under analysis: oxygen desaturation captures the effect of reduced ventilation and increased work of breathing on blood oxygenation and resultant intermittent hypoxemia. Events with associated arousals provide information on whether the breathing disruption was sufficient to trigger central responses and fragment sleep. Magnitude of breathing reduction provides measures of airflow limitation amplitudes and airway collapsibility. Varying hypopnea definitions have mostly focused on the level of associated desaturation and presence/absence of an arousal and have not rigorously compared differences in approaches for quantifying flow limitation or ventilation.

Alternative hypopnea definitions impact AHI values:

All of the summary metrics are highly correlated with one another but can vary tremendously in the absolute numbers of events detected. In one of the first and largest studies, a 10-fold difference in OSA prevalence estimates was reported to result from use of different definitions of hypopnea and application of various disease-defining AHI thresholds.¹³ More recently, studies have estimated the reclassification of OSA severity that results from use of a “3% desaturation or arousal” hypopnea definition compared to a “4% desaturation” criterion. A series of calibration equations were published to allow imputation of AHI levels across definitions.¹⁴ This analysis highlighted that the largest differences in mean AHI occurred at lower OSA disease severity, with convergence at higher levels of disease severity, underscoring how different hypopnea definitions may reclassify individuals with mild to moderate OSA. A meta-analysis (MA) of 11 studies comparing these definitions calculated a sensitivity of 82.7% (95% confidence interval [CI] 0.72–0.90) and specificity of 93.2% (0.82–0.98) for the two definitions.¹⁵ The analysis estimated that an additional 20% of individuals would be classified with OSA using the “3% desaturation or arousal” hypopnea definition.

Comparisons of the prognostic utility of the various hypopnea definitions have not identified clear advantages to any single definition but have highlighted the need to adjust the thresholds used to define SDB disease severity (none, mild, moderate, and severe) according to the AHI (Table III.A.1). In a cross-sectional analysis of the community-based general population, HypnoLaus cohort, adjusted thresholds for defining moderate and severe OSA were shown to associate with both hypertension and diabetes. Specifically, this study suggested that the thresholds for defining disease severity need to be reduced by approximately one half for hypopnea definitions using a 4% desaturation criteria compared to the AASM hypopnea definition.¹⁶ In the Sleep Heart Health Study (SHHS), the application of the current AASM recommended definition (3% desaturation or arousal) resulted in an approximately doubling of the number of individuals classified with moderate OSA (AHI > 15). The group solely identified with use of the AASM definition had a high prevalence of hypertension as compared to the group classified using the more conservative 4% definition, supporting the importance of using more inclusive definitions for identifying individuals at risk for hypertension.¹⁷ It is important to recognize that OSA is associated with other comorbidities, in addition to hypertension and diabetes. The prognostic utility of different definitions may vary for sleepiness symptoms, other diseases, and mortality.

TABLE III.A.1.

Adult OSA severity classification

OSA severity classification	AHI (events/h)
None	<5
Mild	≥5 to <15
Moderate	≥15 to <30
Severe	≥30

Population characteristics, such as age, gender, and obesity, also may influence the accuracy and prognostic value of alternative event definitions, requiring care in selecting definitions most appropriate for given populations. For example, women tend to have shorter apneas and experience less desaturation than men,¹⁸ but with aging, may experience increased rates of death and incident heart failure (HF).¹⁹ A study that utilized estimates of AHI based on different hypopnea definitions (from calibration equations) showed greater variation in effect estimates for CV disease when arousals were used in the hypopnea definition in the overall population but not in women, where an arousal-based definition appeared appropriate.²⁰ Hypopnea definition and subsequent AHI scoring also have implications for insurance coverage for various OSA therapies.²¹

Scoring reliability:

Although there is active research developing automated tools for respiratory scoring, event identification is largely done by manual annotation by trained scorers. Accredited sleep laboratories need to document acceptable inter-scorer reliability using well-defined protocols. Consistency in scoring will vary according to the technician skill, quality of the underlying signals, and severity of the disorder. Apneas are generally considered easier to consistently score as their recognition requires identification of absent airflow. However, distinguishing event subtypes (obstructive, central, and mixed) is much more difficult, resulting in worse scorer reliability.²² Identifying subtle changes in ventilation can be difficult as is needed for hypopnea detection. One argument for requiring correlative data (desaturation, arousal) to identify a hypopnea is to provide additional signs of physiological disturbances over and beyond those identified through a non-calibrated estimate of breathing amplitude change, thus improving scoring reliability. In those cases, care is needed to ensure reliable arousal scoring, which is additionally dependent on the quality of the underlying EEG and electromyography (EMG) channels. For example, in the unattended PSGs in the SHHS, events were more reliably scored when desaturation criteria alone were used in comparison to inclusion of the arousal criteria.²³

ODI versus AHI:

Given that hypopneas utilize oxygen desaturation criteria for event identification, it is not surprising that the AHI and ODI often are highly correlated, but dependent on the specific oximeter and AHI scoring definition used.²⁴ The ODI is automatically derived using only a single sensor, providing objectivity, simplicity, and scalability. High diagnostic accuracy has been reported.²⁵ A number of epidemiological studies and clinical trials have utilized the ODI for defining SDB eligibility criteria (e.g., SAVE²⁶), or for characterizing risk of incident disease.²⁷ However, the ODI may be less appropriate as a screening or prognostic index in individuals less likely to desaturate with respiratory events, such as younger patients, non-obese individuals, and women. Conversely, the ODI may be particularly useful to identify hypoxia-related stresses, which have been related to metabolic disease.¹⁶

Time spent with oxygen saturation <90% (T90):

Measures of hypoxemia, such as SaO₂ nadir and time spent with arterial SaO₂ less than 90% (T90), may be important in assessments of OSA severity and health risk determinations.²⁸ It has been shown that more hypoxemia as measured by T90 and lower SaO₂ nadir in those with similar AHI is associated with more inflammation (measured by C-reactive protein [CRP], platelet count, and endothelial stiffness).^28–30 Recently a retrospective study found that having moderate to severe OSA and T90>20% of sleep time can be associated with a higher risk of hypertension, type 2 diabetes, and 5-year mortality compared to those with T90 < 20%.²⁸ In addition, an SaO₂ nadir of <75% correlated with increased risk of hypertension in this group.

III.B |. OSA and Subtype Definitions

OSA is a chronic disorder caused by repetitive collapse of the UA during sleep. Episodes of complete (apneas) and partial (hypopneas) cessation of airflow can lead to two main consequences: arousals from sleep and oxyhemoglobin desaturations. Apneas and hypopneas occur in all phases of sleep, but are more common in N1, N2 and rapid eye movement (REM) sleep stages than in N3 sleep.

III.B.1 |. Obstructive sleep apnea syndrome (OSAS)

OSAS diagnosis requires the patient to have³¹ symptoms of sleep-related breathing disturbances (snoring, snorting, gasping, or breathing pauses), excessive daytime sleepiness, or fatigue that occurs despite sufficient opportunity to sleep and is unexplained by other medical problems; and¹² five or more episodes of predominantly obstructive respiratory events (obstructive or mixed apneas, hypopneas, or RERAs) per hour of sleep (AHI or RDI ≥ 5).³¹ OSA also may be diagnosed in the absence of symptoms if the AHI is ≥15 episodes/h.

If presenting daytime and nighttime symptoms or cardiometabolic comorbidities are caused by OSA, the term OSAS is used. However, the terms OSA and OSAS are often used interchangeably in the medical literature.

OSAS is recognized to be a heterogeneous syndrome.³² The classification that is most commonly used in clinical practice is the one based on frequency of obstructive events based on AHI (Table III.A.1). Frequency of respiratory events influences several important clinical consequences (hypertension, stroke) in a dose dependent fashion.³³ Frequency-based OSA classification using AHI is utilized as an indication for therapy in current national insurance coverage guidelines in the United States (US).³⁴

However, the classification based on AHI alone poorly addresses various phenotypes of the disease. An alternative classification based on presenting symptoms has been proposed which identifies three clusters of patients: a group with sleep disturbance, a group with excessive daytime sleepiness, and a group with minimal symptoms.³⁵ Several other classifications, based on pathophysiology,³⁶ comorbid conditions,³⁷ and clinical outcomes,³⁸ have also been described. Current guidelines for the treatment of OSA typically take into account several factors including the AHI, presence or absence of symptoms, and associated comorbidities.³⁹

III.B.2 |. Positional OSA

Episodes of airway obstruction in OSA are more frequent and more severe in the supine compared to the nonsupine body position in nearly all patients.⁴⁰ OSA patients who have an increase in breathing abnormalities while in the supine versus lateral position exhibit positional OSA (POSA). In non-positional patients, respiratory events appear in all positions of sleep. With POSA, changes in sleep position effect the overall AHI on a sleep study and account for night-to-night variability in sleep study results. The predominant sleep position and time in each position on the night of the study can be considered when evaluating OSA severity.⁴¹ Patients with POSA have lower body mass index (BMI), smaller neck circumference, longer posterior airway space measurements, and smaller lateral pharyngeal wall tissue volumes.^41,42

Various definitions have been used to diagnose POSA. Cartwright’s definition is commonly used, which describes POSA when AHI in the supine position is greater than two times the AHI in non-supine sleep position. For a subset of patients with POSA, the airway only reaches critical collapsibility in the supine, but not in the nonsupine position.⁴³ These patients exhibit supine-isolated OSA, who have respiratory disturbances exclusively in the supine position without abnormalities when nonsupine. Supine-isolated positional patients represent 27% of patients with POSA.⁴⁴ POSA is associated with lower BMI and lower total AHI in males, lower AHI and higher sleepiness in premenopausal females, and lower AHI and lower Mallampati score in postmenopausal females.⁴⁵

Treatment of POSA takes into account differences in the critical closing pressure between supine and non-supine positions of sleep. These are usually reflected by differences in pressures that need to be generated by PAP devices in different positions of sleep. Although not universally accepted, (auto-titrating positive airway pressure [APAP] devices may provide a better treatment option for patients with POSA than constant-pressure PAP, as they may produce higher pressures in supine, and lower pressures in non-supine body positions.

Treatment of patients with POSA may comprise of enforcement of non-supine sleep. Supine sleep can be avoided by employing the older tennis-ball technique (TBT), which involves placement of a bulky object on the patient’s back, or newer generation sleep position trainers (SPT) that include small, battery-powered devices attached to the neck or chest that provide vibrotactile feedback when in the supine position⁴⁶ (see Section VIII.D.1).

TABLE IX.C.2.c.2.

Evidence for the association of DISE with improved outcomes

Study	Year	LOE	Study design	Study groups	Clinical endpoint	Conclusion
Golbin et al.1650	2016	4	Cohort	1) Palate surgery ± tonsillectomy without DISE (n = 40) 2) Palate surgery ± tonsillectomy ± partial glossectomy ± partial epiglottectomy with DISE (n = 64)	1) AHI 2) Epworth Sleepiness Scale score	Both groups showed a significant decrease in AHI and no change in the Epworth score. There were no differences between the groups in either endpoint.
Pang et al.1651	2020	2b	Cohort	1) Surgery without DISE (n = 156) 2) Surgery with DISE (n = 170)	1) AHI 2) Systolic and diastolic blood pressure	Surgical success rate was greater in group without DISE, as were the decreases in systolic and diastolic blood pressure.

III.B.3 |. REM-related OSA

A number of physiological effects uniquely predispose airways to collapse in REM versus non-rapid eye movement (NREM) sleep. In REM, the genioglossus muscle activity is lower,⁴⁷ the respiratory drive is lower,⁴⁸ and the autonomic drive is higher than in NREM.⁴⁹ These factors result in a longer duration of obstructive events and deeper hypoxemia in REM than in NREM.

The term “REM-related OSA” describes a condition where SDB is predominantly present in REM sleep. While definitions vary, most authors define REM-related OSA in terms of the frequency of SDB events (AHI) in REM and NREM sleep, with AHI in REM being at least double the AHI in NREM. Some groups add a qualifier of requiring an NREM AHI of <15 events/h.^50–53 Depending on the definition used and the population examined, the prevalence of REM-related OSA varies from 13.5% to 36.7%.^50,51,53

Clinically, REM-related OSA is more common in younger patients, in patients with mild to moderate OSA, and in women and African American patients.^50,54,55 Excessive daytime sleepiness is usually less prominent in REM-related OSA than in non-REM-related OSA.^56,57 Depression has been reported more commonly in patients with REM-related OSA than in non-REM-related OSA with combined significant impact on daytime sleepiness and quality of life (QOL).^58,59

The clinical significance of the “REM-related OSA” entity is unclear, and some treat it merely as a mild form of OSA. Similarly, large studies failed to detect significant differences in symptoms or clinical outcomes between patients with REM-related (or REM-predominant) versus non-REM-related OSA.^50,60–62 As a result, the current classification of sleep disorders does not list REM-related OSA as a separate entity.

III.B.4 |. Primary snoring

Isolated snoring, also known as primary snoring, describes a pattern of vibrational sounds related to narrowing of the UA during sleep without significant apnea or hypopnea events (AHI < 5) and without sleep-related daytime symptoms. Snoring is a reflection of partial collapse of the UA and increased airway resistance. The soft tissue that forms the UA is prone to collapse during sleep which is associated with turbulent airflow and pharyngeal tissue vibrations that create the sounds of snoring.

Airway collapsibility is higher in patients with sleep apnea than in primary snorers and healthy controls.⁶³ Isolated snoring, upper airway resistance syndrome (UARS), and OSA can thus be viewed as three manifestations of progressive intermittent UA obstruction that occur during sleep. With progressive degrees and frequency of airway collapse as well as other modifying factors present in a given individual (arousal threshold, preexisting hypoxemia), intermittent obstruction of the airway is clinically reflected as a spectrum of SDB that presents as: (1) isolated snoring, then (2) obstruction-related arousals or desaturations (upper airway resistance syndrome or UARS), and finally (3) partial or complete cessation of airflow through the UA associated with desaturations and/or arousals (OSA). What determines a given individual’s propensity to move along the path of progression from isolated snoring to UARS to OSA is complex, but is likely affected by several factors including weight gain, sex, race, genetic factors, and advancement of age.^64,65

Occasional snoring (snoring on three or less nights per week) is almost universal in the population with a prevalence of 38%–76% in different populations.^66–69 Habitual snoring, defined as snoring on most nights of the week, is present in as many as 12%–25% of the population.^66–69 Both occasional and habitual snoring are more frequent in men than in women.^66–68

Since snoring is the primary nocturnal symptom of OSA, its presence should prompt direct questioning about other symptoms of OSA including pauses in breathing at night, unrefreshing sleep, and daytime sleepiness, especially when the presence of snoring is identified in a setting of OSA-related comorbidities.

III.B.5 |. Upper airway resistance syndrome (UARS)

Some patients have significant fragmentation of sleep due to obstructive events that do not meet formal criteria of apneas or hypopneas. In a PSG tracing, these patients usually display respiratory effort-related arousals (RERAs). RERAs are defined as >10 s sequences of breaths with increasing respiratory effort or flattening of the inspiratory portion of the flow signal, leading to an arousal.⁷⁰ Guilleminault et al. have coined the term UARS⁷¹ to describe the clinical entity of nocturnal airway collapse leading to respiratory effort-related arousals that are associated with daytime symptoms, usually excessive daytime sleepiness. These events do not meet the criteria of OSA.

Most of the clinical and polysomnographic characteristics of UARS patients are based on a limited number of retrospective studies.^72–74 The prevalence of UARS among members of a military academy was found to be 8.6%.⁷⁵ Similar to OSA, the majority of patients with UARS are men, with M:F ratio of about 3:1.⁷⁴ Patients with UARS tend to have lower BMI than those with OSA and the amount of weight gain in the years prior to diagnosis is typically less in UARS than in OSA patients.⁷⁴ Daytime impairment in functioning tends to be worse in patients with UARS compared to those with mild OSA.^74,76 On PSG, patients with UARS may have less N1 and N2 sleep than patients with OSA, but more N3 sleep.⁷⁴ Several groups have reported an increased frequency of non-specific or functional somatic complaints in patients with UARS, compared to OSA patients including irritable bowel, insomnia, difficulty concentrating, cognitive problems, depression, and poor psychomotor performance.^77–79 These features may be apparent on the initial presentation or may appear over time. It has been postulated that some of these associated symptoms may result from lack of specific therapy for UARS, as many insurers do not include UARS definitions in coverage policies for PAP therapy.⁸⁰

The UARS is no longer recognized as a nosological entity by the current edition of International Classification of Sleep Disorders,⁸¹ as its pathophysiology is the same as OSA. As much as the term UARS has fallen out of favor, the value of this designation lies in underscoring the importance of respiratory-related sleep fragmentation in causing symptoms of sleepiness and impaired daytime functioning that may not be reflected by the AHI.

III.B.6 |. Obesity hypoventilation syndrome

Obesity hypoventilation syndrome (OHS) is defined by awake hypercapnia (PaCO₂ ≥ 45 mmHg) in obese individuals (BMI ≥30 kg/m²) that is not otherwise explained by restrictive lung disorders or neuromuscular disease. This disorder represents the extreme end of the SDB spectrum. The prevalence of OHS in the general population is unknown, but conservative estimates range between 0.15% and 0.4% amongst adults in the US.^82,83 However, in obese individuals referred to sleep clinics, prevalence rate increases to 10%–20%.^84,85 In approximately 70% of those with OHS, severe OSA (i.e., AHI ≥30/h) is present while in 10% the abnormality is non-apneic sleep-dependent hypoventilation only, particularly in REM sleep.⁸⁶ The exact mechanisms underlying why only some obese individuals with SDB hypoventilate are not fully understood, but appear to be the end result of a complex interplay between obesity, lung mechanics, respiratory drive, and neurohumoral factors.⁸⁷

IV |. EPIDEMIOLOGY AND CONTRIBUTING FACTORS IN OSA

OSA is a common disorder in adults; its prevalence has increased with rising rates of obesity. OSA is strongly associated with CV and metabolic comorbidities; it is highly prevalent in populations with diabetes, hypertension, heart disease, and stroke. Environmental factors also contribute to OSA risk. The epidemiology of OSA depends on the criteria used to define the disease including how respiratory events are defined, AHI cut offs, and the manner of testing. The 4% desaturation criteria for hypopneas (≥4% oxygen desaturation) may be associated with a more cohesive group epidemiologically – typically older, male, obese, and with underlying cardiac disease (e.g., HF, coronary artery disease [CAD], and atrial fibrillation [AF]) or diabetes. The AASM definition of hypopneas for OSA classification (≥3% desaturation or arousal) is more inclusive and captures more variable sleep disruption phenotypes of OSA, seen in younger, nonobese, and women patients who tend not to desaturate as substantially.³⁶

IV.A |. Prevalence

General population-based studies of OSA show a high prevalence of undiagnosed sleep apnea.⁸⁸ An SR of OSA prevalence studies worldwide found a wide range (13%–33% in men and 6%–19% in women), likely due to methodological heterogeneity such as differences in sampling of the population, AHI cut-off applied, sleep-study scoring criteria used, and test type performed.⁸⁹ In the US, the Wisconsin Sleep Cohort study⁹⁰ and SHHS⁹¹ sampled large non-clinical cohorts in the 1990s. These studies found a general prevalence of 24% in men, 9% in women using the alternative definition of hypopnea (≥4% oxygen desaturation). A smaller subset, 3%–7% in men and 2%–5% in women, had OSAS based on AHI ≥ 5 with excessive daytime sleepiness.^88,90,92 Population-based studies in Asia show similar prevalence to these estimates.^93,94 An updated estimate based on the Wisconsin data in 2013 indicated higher rates of OSA in 14% of men and 5% of women ages 30–70 in the US over time, attributed to changes in population levels of obesity.⁹⁵ Prevalence of OSAS was found to be even higher in a population study of Sao Paulo.⁹⁶ A more recent population-based study in Switzerland found a very high prevalence: 23% in women and 50% in men.⁹⁷ Using rates of obesity, race/ethnicity data, age, and regional prevalence data, it has been estimated that nearly one billion people have OSA worldwide, of which the vast majority is undiagnosed.⁹⁸

IV.A.1 |. OSA prevalence and risk factors

Consistent across all epidemiological studies, the prevalence of OSA is associated with sex, obesity, and age. Prevalence is higher in men by 2:1, but rates increase in women after menopause and become nearly equal.⁹⁹ As BMI increases, the risk of OSA rises with an increase in prevalence of AHI ≥ 15 from 3.6% in normal weight to 56% in those with BMI ≥40 among men aged 50–70 years.⁹⁵ OSA incidence increases with age through age 60, although there also may be subsequent increased risk associated with aging.¹⁰⁰ Other risk factors of OSA include a family history of OSA,¹⁰¹ and certain structural abnormalities of the viscerocranium, including reduced size of the mandible.¹⁰²

Minority populations have lower rates of clinical diagnosis of OSA despite a higher prevalence than white US populations and greater burden of obesity, diabetes, and CV disease.^103,104 Prevalence of OSA in relation to BMI differs by race/ethnicity in US studies, with Hispanic and Chinese American subjects having higher odds of OSA after adjustment for obesity.¹⁰⁵ Differences in prevalence by race/ethnicity may be related to a number of factors, many of which are not well characterized (e.g., environmental exposures). In addition, variation of craniofacial features by ancestry background may be associated with OSA risk; for example, some studies indicate that some Asian populations may be at increased risk of OSA despite low BMI levels due to shorter midface and shorter cranial base measurements.¹⁰⁶ Additionally, there are likely genetic factors that determine the amount of soft tissue in the UA that are responsible for family clustering of OSA.¹⁰⁷

IV.A.2 |. OSA and cardiovascular disease

The prevalence of OSA is higher in those with CV disease. A recent MA estimated 49% of subjects hospitalized with cardiac disease (HF, acute coronary syndrome, and arrhythmias) had previously undiagnosed OSA.¹⁰⁸ There is a very high prevalence of OSA in those with CAD, hypertension, HF, arrhythmias, and especially AF.^109–112 Among subjects with resistant hypertension, OSA is highly prevalent (up to 85% in one series) and the most common secondary cause of hypertension.¹¹³

OSA is associated with a multitude of CV outcomes in observational studies, especially in studies utilizing the CMS definition of OSA, as most CV outcomes are thought to be associated with oxygen desaturation.¹¹¹ Cohort studies have found severe OSA to be associated with higher odds of incident hypertension,^114,115 stroke,^116,117 and CV death.¹¹⁸ This is thought to be due to the increased nocturnal sympathetic activation, proinflammatory state due to cyclic hypoxemia, arousals, and negative intrathoracic pressures with occluded UA.^110,119 Stroke has been consistently identified as a devastating outcome of untreated severe OSA, especially in middle aged adults.¹²⁰

IV.A.3 |. OSA and metabolic disease

OSA is strongly correlated with diabetes, both of which are highly linked to obesity and increased waist/hip ratio. Obesity is the most consistent risk factor for both disorders, especially central adiposity.¹²¹ Trials enrolling obese diabetics have found that 86% have some degree of OSA.¹²² Weight loss trials show improvement in both diabetes outcomes and reduced sleep apnea severity.¹²³ Additional metabolic risk factors include excess testosterone, growth hormone, and hypothyroidism.

IV.A.4 |. Environmental factors and OSA

Recent studies examining residential factors demonstrate an association between neighborhood features and sleep apnea. High density of fast-food outlets and lower mixed land use may foster OSA through promoting obesity. Neighborhoods with more parks, higher street connectivity, and aesthetic features promote walking. Residing in less favorable walking environments has been associated with more severe sleep apnea especially in male and obese individuals.¹²⁴ Neighborhood disadvantage, living in poor residential areas, has been associated with childhood sleep apnea in several epidemiological studies.^125,126 A recent MA also identifies that the social gradient is associated with OSA.¹²⁷ This association may be related to obesogenic environments with limited options for recreation and healthy food¹²⁸ and greater pollution in disadvantaged neighborhoods.¹²⁹ Ambient air quality has also been associated with sleep apnea.^130,131 The odds of sleep apnea increased by 39% as nitrous dioxide levels increased by 10 parts per billion (ppb) in US study.¹³⁰ These environmental features may contribute to disparities in obesity and sleep health.

IV.B |. Economic Burden of OSA

OSA not only impacts health and functioning but also economic outcomes. Several economic topics related to OSA include: the costs of OSA due to its impact on health, attention, and productivity; the costs incurred in management; the cost effectiveness of diagnosis and treatment; and whether medical cost savings occur with treatment.

A 2016 AASM commissioned report estimated the economic impact of undiagnosed OSA in the US to be $149.6 billion in 2015 based on the interview of opinion leaders, the scientific literature, survey of patients, and CMS fee schedules.¹³² Direct economic costs included OSA-related comorbidities such as high blood pressure (BP), motor vehicle or workplace accidents, and compensating behaviors such as substance abuse. Indirect economic costs included decreased productivity at work, reduced QOL, and stress on interpersonal relationships. The largest economic impact was assigned to lost productivity ($86.9 billion) followed by medical comorbidities and reduced mental health ($30 billion) and motor vehicle accidents ($26.2 billion).

Patients with OSA have about twice the medical costs of controls prior to diagnosis.^133,134 For example, a study of health maintenance organization members found average total medical costs in OSA patients prior to diagnosis was $2720 versus $1384 in age and gender matched controls (in 1996 dollars).¹³⁴ In the SHHS, 256 subjects with OSA (AHI > 11 and sleepiness) had about 17% higher predicted healthcare utilization compared to subjects not meeting these OSA criteria after adjustment for age, gender, BMI, and study site.¹³⁵ Similarly, in the Outcomes of Sleep Disorders in Older Men (MrOS) study, among 1316 elderly men, mean annualized total healthcare costs were 11.6% higher in untreated subjects with moderate to severe SDB than in those without SDB.¹³⁶

With regards to OSA and productivity, data from the Danish National Patient Registry found lower employment (60% in OSA patients vs. 64% for age, sex, and region matched controls) and higher rate of public transfer payments (18% vs. 13% for controls). As a result, labor market income was 9.3% lower and social transfer payments were 18% higher in patients over the 8 years prior to and after diagnosis.¹³³

Data from the National Safety Council and an MA of studies investigating collisions and OSA indicated that 800,000 drivers were involved in OSA-related motor vehicle collisions at a cost of $15.9 billion and 1400 lives in the year 2000 in the US ($23.9 billion adjusted to 2020 dollars).¹³⁷

IV.B.1 |. Costs of treating OSA

The AASM commissioned-report estimated that in 2015, $12.4 billion was spent diagnosing and treating the 5.9 million US adults with diagnosed OSA.¹³² Costs related to OSA management vary based on whether laboratory or home-based diagnostic and management strategies are used. Based on data from a multicenter clinical trial using the payer perspective, the mean cost per patient over 3 months using home-based management was estimated at $1807 versus $2424 using lab-based management. Some of the cost advantage with use of home sleep testing ($167 vs. $782 for lab study) was offset by the need for additional testing in some patients who have failed or had non-diagnostic home studies.¹³⁸ Management costs are highest at the time of diagnosis and tend to diminish over time. In a recent Australian cost effectiveness analysis (CEA), the average annual cost over 5 years for diagnosis and treatment of OSA using lab-based study was estimated at $579 Australian dollars (AUD; $495 US dollars).¹³⁹

Cost effectiveness

CEA compares interventions based on the differences in costs needed to provide an additional quality-adjusted life year (QALY) gained. A number of CEA have evaluated the value of managing moderate to severe OSA from a payer perspective: the estimated cost per QALY over 5 years has varied considerably but generally support its value when compared to other accepted interventions (<$50,000 per QALY).¹⁴⁰ Some CEA have considered downstream medical cost (due to strokes, heart attacks, and motor vehicle collisions) that could be prevented from OSA therapy. For example, Guest et al. estimated that cost savings exceeded the costs of treating OSA after 13 years.¹⁴¹ More recently, Streatfeild et al. estimated that from a healthcare system perspective, the estimated cost to treat OSA was $12,495 AUD (US $8746) per disability adjusted life year (Equivalent to QALY gained).¹³⁹ They included the impact of estimated healthcare cost savings from reduced medical morbidity ($76 AUD/year). From a societal perspective, there was net cost saving of $326/year because of a projected $783/year cost saving due to the reduction of financial costs (due to gains in productivity, fewer accidents, and other savings).

Cost savings

Observational studies are equivocal regarding whether medical cost savings occur with OSA treatment and their magnitude. Concern has been expressed regarding the adequacy of methodology in some of these studies, because they do not include an appropriate control group, use a longitudinal cohort design, compare change in costs between groups (rather than relying only on costs in the posttreatment period), measure treatment adherence, adjust for potential confounders, and exclude costs of diagnosis and treatment (which naturally decline over time).¹⁴² A recent high-quality study in an older population found costs during the 2 years following diagnosis increased relative to pre-diagnosis levels in individuals regardless of OSA diagnosis or PAP adherence status.¹⁴³ Medical costs increased about 8% less for the group fully adherent to CPAP relative to sleep apnea patients who did not receive therapy. It is likely that the impact of therapy on medical costs depends on factors such as the population characteristics (OSA severity and comorbidity), health system, effectiveness of therapy, and duration of observation.

In summary, untreated OSA is associated with substantial economic cost related to reduced productivity, medical comorbidity, and motor vehicle and work accidents. Medical costs in clinically identified OSA patients are about twice that for age-sex matched controls prior to diagnosis. In contrast, when OSA is identified in community-based research cohorts, cost differences are more modest (10%–20% more than similar patient without OSA). Costs for managing OSA are substantial though the cost effectiveness for the treatment of moderate to severe OSA with CPAP is well established. It is reasonable to expect some cost savings long-term with CPAP therapy due to downstream benefits, though there is a need for additional high-quality studies to clarify this issue with regards to the magnitude of savings and variation with patient characteristics, models of OSA care, and treatment adherence.

IV.C |. Risk Factors for OSA

IV.C.1 |. Sex

Most population-based studies show a two- to threefold greater prevalence of OSA among men compared to women.^{18,45,90,93,96,99,104,144–156} These findings span across a range of diagnostic modalities, hypopnea definitions, and AHI cutoffs. These findings are also consistent across ethnicities.^{45,93,96,146,148,149,153,157} Even when matched for BMI and age, OSA occurs more commonly in men, and with greater severity.^155,156 Age and BMI are risk factors for OSA in both men and women, though the degrees of their impact vary between the sexes.¹⁴⁷ Weight loss or gain impacts AHI greater in men than in women.¹⁵⁵ The effect of BMI on OSA severity decreases with age for both sexes, with lesser impact for individuals older than age 60 years than in younger individuals.¹⁵⁶ In a prospective cohort population-based study, Tishler et al.¹⁵⁶ demonstrated that the AHI increases by 140% per 10-years in women (OR 2.41) but by only 15% in men (OR 1.15). This results in a narrowing of the sex difference in OSA risk with increasing age. Sex hormones are implicated in the differential risk between men and women. Post-menopausal women, and particularly those not on hormone-replacement therapy, are at increased risk for OSA compared to pre-menopausal women, even when correcting for age.^{99,144,147,154}

Clinic-based prevalence studies also show men to have greater AHI compared to BMI- and age-matched women.^{155,158–160} The risk of mild OSA (AHI ≥ 5/h) is two-fold greater in men than women, while the risk of severe OSA (AHI ≥ 30/h) is almost eight-fold greater. In a clinic population of 26,425 adults, 21–80 years old, age was less of a factor in AHI severity in obese men than in obese women.¹⁶¹

Polysomnographic features of OSA also differ between men and women. Men with OSA tend to have more frequent apneas (vs. hypopneas), longer duration of apneas, and more severe oxygen desaturations.^158,162 The distribution of apneas and hypopneas also differs between men and women. Women are more likely to have events during REM sleep.^{18,55,158,163–167} Several studies show that women tend to have less NREM events, but similar or greater REM events compared to age- and BMI-matched men. The difference in the prevalence and severity of OSA between men and women are affected by AHI definitions. For example, Won et al.¹⁸ showed defining respiratory events using lesser oxygen desaturation threshold levels and including arousals increased the relative proportion of women classified with OSA.

The reason for sex differences in the risk for OSA is not well understood but is thought to relate to several factors: (1) Facial anatomical distinctions, such as mandible position, impart different degrees of risk in men and women.^168,169 (2) UA anatomy and function differ between sexes, with men demonstrating longer and more collapsible UAs.^170–173 (3) Obesity¹⁷⁴ and fat distribution^175,176 differentially affect men and women’s propensity for OSA. Lim et al.¹⁷⁶ found that while waist-to-hip ratio predicted OSA in both men and women, neck circumference was only predictive of men with OSA. (4) Loop gain and other measures of ventilatory control have been measured in men and women, and in post- and pre-menopausal women, with some evidence to suggest sex differences in respiratory control.^177–182

In summary, there are significant sex differences in the prevalence, PSG features, and risk factors for OSA. Men are in general at greater risk for OSA during NREM sleep, while men and women share similar risk during REM sleep. More studies are needed to understand the sex and sleep stage-dependence of OSA. (Table IV.C.1a and IV.C.1b)

TABLE IV.C.1a.

Population studies on prevalence or incidence of OSA in men and women

Study	LOE	Study design	Cohort	Sample population	OSA diagnostics/criteria	Conclusion
Young90	2c	Cross-sectional	Wisconsin Sleep Cohort 30–60 years N = 602	Population-based	PSG AHI ≥ 5	Higher prevalence in men for all age groups Men 2–3.7 times greater prevalence
Bixler144	2c	Cross-sectional	20–100 years N = 1741	Population-based	PSG AHI > 15	Prevalence greater in post-menopausal women, and in those not taking hormone replacement therapy
Shahar154	2c	Cross-sectional	Sleep Heart Health Study Women >50 years N = 2852	Population-based	HST IV AHI > 15	Hormone use associated with less OSA, particularly among women 50–59 years old (adjusted OR 0.36)
Young99	2c	Cross-sectional	Wisconsin Sleep Cohort Study Women 30–60 years N = 539	Population-based	PSG AHI ≥ 5	Menopausal transition associated with OSA after controlling for age, body habitus, and several lifestyle factors Postmenopausal women 2.6 times more likely to have AHI ≥ 5, and 3.5 times more likely to have AHI ≥ 15, compared to premenopausal women
Quintana-Gallego183	3b	Case–control	Spain sleep clinic Women mean age 58 ± 10 years; men mean age 53 ± 11 years N = 1745	Clinic-based	PSG AHI ≥ 10 or HST III RDI ≥ 10	Prevalence: 4.9:1 men to women ratio
Gabbay161	3b	Case–control	Israel multiple sleep centers 21–80 years N = 26,425	Clinic-based	PSG AHI ≥ 10	AHI increased with increasing age for both non-obese men and women Obesity affected AHI in men aged 20–40 years, but AHI did not change after age 40 years in obese men
Huang147	2b	Prospective cohort	Nurses’ Health Study (NHS) Post-menopausal women who were free of known OSA N = 50,473 NHS N = 53,827 NHSII	Population-based	Medical record NHS: 12-year incidence NHSII: 20-year incidence	Surgical menopause had 26% higher risk of incident OSA compared to naturally post-menopause, adjusted for age at menopause and other OSA risk factors
Won18	2c	Cross-sectional	MESA cohort Mean 69 ± 9 years N = 2057	Population-based	PSG AHI ≥ 15	Prevalence: 41% men, 22% women

TABLE IV.C.1b.

Reasons for sex differences in OSA

Study	LOE	Study design	Study groups	Clinical end-point	Conclusion
Anatomy and anthropometric traits
Polesel184	3b	Cross-sectional	Brazil N = 552 women N = 450 men	Anthropometrics	Waist circumference and waist-to-height best predicted OSA in women Waist-to-height ratio and neck circumferences best predicted mild OSA in men, while BMI was associated with severe OSA
Cho168	3b	Cross-sectional	Korea Suspected OSA N = 2016	Anthropometrics Cephalometry	Men with OSA had increased soft palate length compared to controls Women with OSA had increased soft palate thickness and nasion angle than controls
Sutherland169	3b	Cross-sectional	Chinese and Australian sleep clinic N = 363 200 Chinese 163 Caucasian	Craniofacial features by face photography ethnicity	Women with OSA had different facial features compared to controls, such as greater face width to eye width ratio, and reduced mandibular plane Men with OSA had increased mandibular plane angle relative to controls
Perri185	3b	Cross-sectional	Caucasian of European origin Sleep clinic N = 104 OSA N = 85 healthy	Surface cephalometry	OSA patients had anthropometric and surface cephalometric phenotypes that differed in comparison with healthy subjects, however, sexual dimorphism levels (i.e., male–female ratios) were broadly similar in the two groups
Lim176	3b	Cross-sectional	Korean Clinic N = 151	Anthropometrics	Neck circumference predicted OSA in men only, while waist-to-hip ratio predicted OSA in both men and women
Dancey186	3b	Cross-sectional	Canada sleep clinic N = 3942	Neck circumference	Neck-to-height ratio predicted AHI, accounting for 19% of the variability, more significantly in men
Upper airway anatomy and function
Brooks173	3b	Cross-sectional	Healthy volunteers N = 77 men N = 98 women	Pharyngeal cross-sectional area during quiet breathing using the acoustic pulse technique	The men had a larger change in pharyngeal area with changing lung volume than the women
Segal172	2b	Cross-sectional	Chart review for those who have had PSG and CT neck N = 24 (15 men, nine women)	Upper airway length	Men with OSA were found to have longer upper airway length even when normalized to body height compared with OSA women Correlation between upper airway length and AHI
Mohsenin171	3b	Cross-sectional	Sleep Clinic AHI ≥ 5 N = 71	Upper airway dimensions using acoustic reflectance while lower jaw was in the resting and retrusive posture	Men tend to have a larger but more collapsible airway during mandibular movement than women
Eckert187	3b	Cross-sectional	12 healthy, eight OSA	Genioglossus electromyogram (GGEMG)	Reductions in GGEMG during REM sleep were not different between OSA patients and control subjects or between sexes
Malhotra170	3b	Case–control	Healthy volunteers age and BMI matched 19 men, 20 women	MRI	Pharyngeal airway length and cross-sectional area, soft palate area, and airway volume were greater in men compared with women
Obesity and fat distribution
Simpson188	3b	Cross-sectional	Western Australian Sleep Health Study Sleep clinic OSA patients (60 men, 36 women)	Obesity measures using dual-energy absorptiometry	In women, percentage of fat in the neck region and body mass index together explained 33% of the variance in AHI In men, percentage of fat in the abdominal region and neck-to-waist ratio together accounted for 37% of the variance in AHI
Huang174	3b	Cross-sectional	Taiwanese Sleep center >18 years N = 2345 patients (339 females)	BMI	AHI was associated with BMI more strongly in men
Harada175	3b	Cross-sectional	Japanese Sleep Clinic 271 men, 100 women with OSA	CT scan measured visceral fat area	AHI was independently associated with visceral fat area only in men
Ventilatory control
Sin177	3b	Cross-sectional	Canadian sleep center N = 219 (43 women, 176 men; 104 OSA, 115 no-OSA)	Hypercapnic ventilatory response test (HCVR)	Elevated carbon dioxide and older age were significantly correlated with low HCVR in men, while BMI was associated with HCVR in women
Jordan179	3b	Cross-sectional	Healthy volunteers12 men, 11 women at both their luteal and follicular phases	Post-stimulus ventilatory decline (PSVD)	There were no significant differences in PSVD between men and women in either luteal or follicular phases
Syed178	2b	RCT	10 men and seven women with OSA, with control 10 healthy men and 10 healthy women	Hypoxic ventilator response augmentation (HVRA) Ventilatory long-term facilitation (vLTF)	HVRA and vLTF were enhanced in the OSA group compared with control Sex did not impact any measures Exposure to intermittent hypoxia during sleep led to a corresponding increase in respiratory events compared with sham exposure, irrespective of sex
Pillar181	3b	Cross-sectional	8 normal women and eight age and BMI-matched men, during stable NREM	Upper airway mechanics Ventilation Activation of two dilator muscles, genioglossus and tensor palatini were monitored during loading	Men developed more severe hypopnea in response to identical applied external loads than did women Men and women had near identical minute ventilation responses to total load, implying no differences in central drive or load response There were no significant increases in genioglossus or tensor palatini activation in response to loading in either sex
Zhou182	3b	Cross-sectional	8 men (25–35 years) and 8 women in the midluteal phase of the menstrual cycle (21–43 years) Repeated studies in 6 women during the midfollicular phase	Apneic threshold	In women, induction of a central apnea required an increase in tidal volume by 155% ± 29%, compared to men with an increase in tidal volume by 142% ± 13% Similarly, women required greater reduction in PETCO2 compared to men, to general a central sleep apnea There was no difference in the apneic threshold between the follicular and the luteal phase in women
Jordan180	3b	Cross-sectional	11 men and 11 women were matched for severe OSA severity Second group of 12 men and 12 women were matched for body mass index	Loop gain (LG) measured using proportional assist ventilator Critical closing pressure of upper airway (Pcrit) by progressive CPAP drop technique	In the BMI-matched subgroup, women had less severe OSA during NREM sleep and lower Pcrit, but similar LG compared to men

IV.C.2 |. Obesity

Obesity is one of the most significant risk factors for SDB.^{95,156,189,190} It is estimated that approximately 58% of adult OSA cases are attributable to obesity.¹⁹¹ OSA is prevalent in 44.6% of men and 13.5% of women aged 30–49 years with BMI of 30–39.9 kg/m², compared to 7% of men and 1.4% of women whose BMI is less than 25 kg/m².⁹⁵ It is estimated that the prevalence of OSA in bariatric surgery patients is as much as 60%–83%,^192,193 with weight loss after surgery resulting in significant reduction in AHI.^192–195 Furthermore, it has been shown that 10% increase in body weight is associated with an increase in AHI of approximately 30%.¹⁸⁹ Another study showed an increase in BMI by 1 standard deviation (SD) was associated with a threefold increase in risk of OSA.⁹⁰ However, this association with BMI decreases with age, and after age 60 years may be less significant.⁹⁵

According to the US National Health and Nutrition Examination Survey (NHANES), the prevalence of ageadjusted obesity (BMI > 30 kg/m²) has increased from 30.5% to 42.4%, and of severe obesity (>40 kg/m²) from 4.7% to 9.2% since 1999 to 2017 (the prevalence of both obesity and severe obesity was highest in non-Hispanic black adults).¹⁹⁶ BMI correlates with body fat percentage, more so in women, but neither represents adipose tissue distribution nor differentiates between fat and lean tissue.¹⁹⁷ Fat distribution, particularly in the upper body rather than total body, is the most important factor contributing to OSA.¹⁹⁸

The mechanism by which obesity causes and progresses OSA is still unclear. Obesity, specifically central adiposity, increases pharyngeal collapsibility that causes recurrent episodes of UA obstruction in sleep apnea through both mechanical and neuromuscular effects.^199–203 Central adiposity increases the mechanical load on the UA and decreases the compensatory neuromuscular response. This is thought to be mediated by adipokines that directly impact central nervous system activity.¹⁹⁹ A study by Sands et al. showed that obese patients with OSA compared to those without OSA have worse pharyngeal collapsibility with less active pharyngeal muscle response.²⁰⁴ Despite the evidence of a link between obesity and central adiposity with OSA, there is variability in the prevalence and severity of OSA even in markedly obese patients suggesting there are likely other neuro-hormonal, anatomic, and genetic components involved.¹⁹⁹

Although it is unclear whether a predominant mechanism of OSA in obesity is increased para-pharyngeal tissue, most studies suggest obese patients with OSA have a smaller UA cross-sectional area. Some studies suggest this is due to retropalatal fat deposition, while others suggest it is parapharyngeal fat or increased pharyngeal wall thickness that determines the development of OSA in obese individuals.^{200,202,205–208} Recent studies evaluating dynamic changes have shown that parapharyngeal fat is associated with concentric obstruction in the retropalatal area. They have also shown that parapharyngeal fat deposition in the subglosso-supraglottic area is associated with increased OSA severity independent of BMI or neck circumference.^209,210

Investigators have also shown that caudal traction by the mediastinal and ribcage muscle attachments to the UA improves airway patency, suggesting UA patency is proportional to lung volumes.^211,212 Caudal traction improves airway caliber by reducing transmural pressure and reducing compliance (i.e., increasing stiffness) of the pharyngeal muscles. In obese individuals, lung volumes are often reduced, which results in reduction in caudal traction, thereby contributing to UA collapse.²¹¹

It is not understood why some obese patients also have hypoventilation. Shimura et al. showed higher levels of leptin in hypercapnic OSA patients compared to eucapnic OSA patients, even after correcting for BMI, fat distribution, AHI, and mean SaO₂. This suggests that leptin, an adipocyte derived hormone which is elevated in obesity, does not prevent hypoventilation in hypercapnic patients, despite it being a respiratory stimulant. It is possible that these obese individuals prone to hypercapnia have leptin resistance contributing to both obesity and hypoventilation.²¹³

Further evidence that obesity is pathogenic in OSA for some individuals, is that weight loss results in marked improvement of OSA.^122,214 Since obesity confers additional CV risk to OSA patients, weight loss directly benefits CV health.²¹⁵ (Table IV.C.2)

TABLE IV. C. 2.

Obesity as a contributing factor for OSA

Study	Year	LOE	Study design	Study groups/age/N	Clinical end-point	Conclusion
Hoffstein212	1984	3b	Case–Control	Obese subjects; OSA versus non-OSA N = 19 Age=28–68	1. Cross-sectional area of pharynx 2. TLC (total lung capacity) to RV (residual volume)	In obese patients with OSA, pharyngeal cross-sectional area is small and varies considerably with change in lung volume (and this change in size with change in lung volume was significantly different in the two groups)
Katz198	1990	3b	Cross-sectional	Canada Sleep clinic N = 123	PSG AHI ≥ 5	External, internal neck circumference and degree of obesity are important predictors of OSA
Mezzanotte201	1992	3b	Case–control	Denver Veterans Affairs Hospital OSA and normal controls N = 25 Age = 40–46 years	1. Supraglottic resistance 2. Genioglossal EMG 3. Minute ventilation 4. End tidal CO2	Neuromuscular compensation present during wakefulness from genioglossus may be lost during sleep in apneic patients
Shelton205	1993	3b	Case–control	Univ of Virginia Sleep clinic and control from community N = 30 Age = 23–65	MRI for adipose tissue volume	Adipose tissue is deposited in pharyngeal area in OSA patients, and the volume of this tissue is related to presence and degree of OSA
Young90	1993	2c	Cross-sectional	Wisconsin Sleep Cohort study N = 602	PSG AHI ≥ 5	1. Prevalence of OSA is 9% in women and 24% in men 2. Male sex and obesity strong risk factors 3. Increase in BMI by 1 SD is associated with three-fold increase in risk of OSA
Schwab200	1995	3b	Cross-sectional	Sleep Clinic at University of Pennsylvania N = 68	1. MRI 2. PSG RDI > 15	1. Lateral pharyngeal wall is larger in apneic patients 2. Wall thickness explains the largest part of variance in airway caliber
Peppard189	2000	2b	Prospective cohort	Population based N = 690	1. Percent change in AHI on PSG 2. Odds of developing moderate to severe SDB, with respect to change in weight	1. 10% weight gain predicted an approximate 32% increase in AHI, with six-fold increase in odds of developing moderate to severe SDB 2. 10% weight loss predicted a 26% decrease in AHI
Stanchina203	2002	3b	Cross-sectional	Healthy individuals in Boston N = 15 Age = 24–32	During NREM sleep: 1. Genioglossus EMG 2. Epiglottic pressure 3. Airflow under different conditions	Genioglossus muscle responds well during NREM sleep when hypercapnia is combined with resistive load, but is less responsive to either chemical stimuli (hypoxia, hypercapnia) or inspiratory resistive load alone
Young190	2002	2c	Cross-sectional	Sleep Heart Health Study N = 5615	In home PSG AHI ≥ 15	Male sex, age, BMI, neck girth, snoring, and repeated breathing pause frequency were independent, significant correlates of AHI ≥ 15
Schwab207	2003	3b	Case–Control	Penn Center for sleep disorder and control from community in same neighborhood N = 96 Age = 24–66	MRI of upper airway	Volume of lateral pharyngeal wall, total soft tissues and tongue larger in OSA than normal subjects
Tishler156	2003	2b	Prospective study	Cleveland family study N = 286	HST AHI ≥ 10	Five-year incidence of OSA – 7.5% for moderate SDB and 16% for mild to moderate
Shimura213	2005	3b	Cross-sectional	Japanese sleep clinic patients N = 185 Age = 22–72	1. PSG AHI ≥ 5 2. CT scan for visceral and subcutaneous fat accumulation 3. Lung function 4. Leptin levels	1. Location of body fat does not contribute to hypoventilation 2. Circulating leptin levels does not maintain alveolar hypoventilation in hypercapnic obese patients with OSA
Kairaitis 211	2007	5	Animal study	Male NZ white rabbits N = 20	Upper airway extraluminal tissue pressure in lateral and anterior pharyngeal walls	Decrease in upper airway collapsibility due to lung volume related caudal traction
Foster122	2009	1b	RCT	16 US centers (Overweight/obese with DM and OSA) N = 264	1. BMI 2. Waist and neck circumference 3. HbA1c	1. The intensive lifestyle intervention (ILI) group lost more weight at 1 year than Diabetes support and education 2. ILI was associated with an adjusted decrease in AHI of 9.7 events/h
Flegal197	2010	2c	Cross-sectional	NHANES (National Health and Nutrition Examination Survey) population N = 5555 Age > 20 years	BMI	In 2007–2008, the prevalence of obesity was 32.2% in men and 35.5 women
Ashrafian 192	2012	2a	SR(with heterogeneity) of cohort studies	Studies with metabolic interventions-33 Studies with lifestyle intervention-24	1. BMI 2. AHI	Metabolic surgeries offer significant reduction in symptoms and measures of OSA by both weight dependent and independent mechanisms
Dixon194	2012	1b	Individual RCT (Bariatric surgery vs. conventional weight loss)	Australian Hospital-Obese patients (BMI of 35–55) and <6 months diagnosis of OSA (AHI > 20) N = 60	1. PSG (baseline to 2 year change in AHI) 2.Weight 3. CPAP adherence 4. Functional status	In obese patients with OSA, bariatric surgery compared with conventional weight loss therapy did not result in a statistically greater reduction in AHI despite major differences in weight loss
Li202	2012	3b	Case-control	Chinese Han population (Otolaryngology Head & Neck Surgery Dept.) N = 28	1. MRI 2. Pharyngoscopy under general anesthesia	OSA patients have more fat tissue adjacent to pharyngeal cavity, and the fat-deposition correlated to collapsibility
Peppard95	2013	2c	Cross-sectional	Wisconsin Sleep Cohort Study N = 1520	PSG AHI ≥ 5 and ESS>10	Prevalence of SDB is increasing in the population (relative increases of between 14% and 55% depending on the group)
Chirinos 215	2014	1b	Individual RCT (Randomized to CPAP, weight loss and combined CPAP with weight loss)	Obese, moderate to severe OSA and CRP > 1 mg/L N = 181	1. CRP 2. Insulin sensitivity 3. Lipid levels 4. Blood pressure	1. No difference in CRP level reduction 2. Weight loss provided incremental reduction in insulin resistance and TG level when combined with CPAP
Jang209	2014	3b	Cross-sectional	OSA patients from Korean Sleep Center N = 33 Age = 31–54	1. Facial CT 2. DISE (drug-induced sleep endoscopy)	Parapharyngeal fat pad is associated with concentric narrowing of the retropalatal pharynx
Kim208	2014	3b	Case–control	University of Pennsylvania Center for Sleep and Circadian Neurobiology N = 121	MRI upper airway	Increased tongue volume and fat deposition at tongue base in OSA compared to controls
Pahkala206	2014	1b	RCT	Kuopio University Hospital, Finland N = 60	1. PSG 2. CT scan of upper airway	1. Pharyngeal fat pad area was significantly larger and hyoid bone to cervical spine area longer in OSA than habitual snorers 2. Weight loss by lifestyle intervention-based program led to improvement in OSA by reducing both central obesity and pharyngeal fat pad
Sands204	2014	3b	Individual case–control studies	Overweight/obese without apnea (AHI < 15/h), overweight/obese with OSA (AHI ≥ 15/h) and normal weight/nonapneic N = 54	1. Pcrit (Pharyngeal critical closing pressure) 2. Pharyngeal muscle (greater genioglossus) EMG	Overweight/obese without mod/severe OSA have increased (three times greater) pharyngeal muscle EMG activity during sleep responsiveness
Ashrafian195	2015	2a-	SR (with heterogeneity) of cohort studies (?but also has some RCT)	19 surgical (n = 525) and 20 non-surgical (n = 825) studies	BMI and AHI before and after intervention	Surgical patients achieved a significant 14 kg/m2 weighted decrease in BMI with a 29/h weighted decrease in AHI. Non-surgical patients achieved a significant weighted decrease in BMI of 3.1 kg/m2 with a weighted decrease in AHI of 11/h
Ng214	2015	1b	RCT	Prince of Wales Hospital, Hongkong OSA patients with AH ≥ 15/h N = 104	1. HST AHI 2. ESS 3. SF-36 (Short Form Health) survey 4. BMI	Lifestyle modification program (LMP) was more effective in reducing AHI from baseline (16.9% fewer events in LMP vs. 0.6% more events in control group with usual care)
Peroma193	2017	2b	Prospective cohort	Bariatric OSA patients who underwent bariatric surgery N = 132	1. PSG (12 months after surgery) 2. BMI 3. Neck and waist circumference	Prevalence of OSA decreased from 71% at baseline to 44% at 12 months after surgery (p < 0.001)
Chen210	2019	3b	Cross-sectional	Taiwan Hospital (Otorhinolaryn- gology Dept.) N = 41 Age = 34–48	1. PSG AHI ≥ 5 2. Drug induced Sleep CT 3. BMI 4. Neck circumference	Subglosso-supraglottic parapharyngeal fat pad area, independent of BMI, and neck circumference influenced severity of OSA
Hales196	2020	2c	Cross-sectional	US census from 2000	BMI	1. In 2017–2018, the age adjusted prevalence of obesity in adults was 42.4% 2. No significant differences between men and women 3. Severe obesity prevalence higher in women at 9.2%

Abbreviations: AHI, apnea hypopnea index; BMI, body mass index; CT, computed tomography; DM, diabetes mellitus; EMG, electromyography; MRI, magnetic resonance imaging; OSA, obstructive sleep apnea; Pcrit, Pharyngeal critical pressure; PSG, polysomnography; SD, standard deviation.

IV.C.3 |. Craniofacial anatomy

There are significant craniofacial differences among OSA subjects when compared to healthy controls. Craniofacial analysis within the literature has primarily been based on cephalometric measurements obtained through lateral plain film cephalograms. Jamieson et al. first brought craniofacial morphology into OSA pathophysiology by examining cephalometry on sagittal plain films among 155 OSA patients and 17 non-OSA controls showing that OSA patients had a more acute cranial base angle and greater retroposition of the mandible.²¹⁶ Numerous studies have since been published examining a variety of cephalometric measures among OSA patients obtained through lateral cephalograms as well as craniofacial measurements obtained through other modalities including computed tomography (CT) and magnetic resonance imaging (MRI).^217–228 An SR and MA of 25 studies comparing lateral cephalometric measurements in OSA²²⁹ showed the strongest cephalometric measurements associated with OSA with the least variability and heterogeneity among studies were (see Figure IV.C.3):

• Increased anterior facial height

• Inferiorly and posteriorly positioned hyoid

FIGURE IV.C.3.

Common sagittal cephalometric findings associated with obstructive sleep apnea (OSA): (A) decreased cranial base length, (B) decreased maxillary length, (C) decreased mandibular length, (D) decreased posterior facial height, (E, F) increased anterior lower facial height and total facial height, (G) decreased cranial base angle, (H) decreased gonial angle (posterior rotation of mandible), (I) inferior and posterior displacement of the hyoid bone.

Additionally, the following cephalometric measurements are suggested to be associated with OSA:

• Short anterior cranial base angle

• Decreased cranial base length

• Shorter maxillary length

• Shorter mandibular length

• Increased mandible rotation

• Retroposition of the mandible

• Decreased posterior facial height

However, there was greater variability and heterogeneity among studies for these measurements with some noting non-significant differences, while others did not measure all values in their cephalometry.²²⁹

In addition to lateral cephalometry, studies have also looked at transverse craniofacial features. Chi et al. compared both sagittal and coronal three-dimensional (3D) cephalometry obtained from MRI among OSA patients showing that both an inferiorly positioned hyoid as well as a smaller and shallower mandible was associated with OSA risk.²²⁴ Seto et al. measured interdental widths and showed that OSA patients had decreased inter-molar widths as well as higher palatal indexes which suggests a higher, deeper arched palate.²³⁰ Johal and Conaghan measured maxillary widths and found that OSA patients did not have shorter interdental widths but did show an association of OSA with increased palatal heights.²³¹ Kuzucu et al. reported that OSA patients have decreased distances between the pterygoid hamuli along the posterior maxilla.²³²

Common craniofacial physical exam findings that are associated with OSA risk include: retrognathia, micrognathia, high arched palate, larger thyromental angles, and shorter thyromental distances.^{219,230,233–236} Dental findings may also be suggestive of underlying craniofacial deficiency and include: open bite, overbite, overjet, and proclination of the mandibular incisors.^219,234,237

BMI is a confounding factor for craniofacial influences on OSA. OSA patients, when compared to BMI matched controls, showed craniofacial differences including a lower positioned hyoid, shorter mandibular body length, retroposition of the mandible, decreased mandibular ramus height, increased lower anterior facial height, and shorter cranial base length.^238–241 Furthermore, comparing OSA among different ethnicities, studies have shown that Asian subjects have shorter cranial base angles, shorter maxillary lengths, shorter mandibular body lengths, greater mandibular ramus lengths, greater maxillary widths, and shallower maxillary depths while Caucasian subjects tend to have higher BMI and larger tongue volumes.^151,242 In contrast, African Americans tend to have larger tongues than Caucasians and were less likely to be brachycephalic, a craniofacial form that is associated with reduced anterior–posterior facial dimensions.^243,244 These studies suggest that craniofacial dimensions are a risk factor for OSA even when controlling for BMI and that different ethnicities may carry different OSA risks based on the contributions of BMI and craniofacial form.

In summary, there is a large body of evidence suggesting there are associated craniofacial factors in OSA. However, there is heterogeneity among studies regarding which specific craniofacial measurements are associated with OSA. All studies were also skewed heavily toward male patients and none focused primarily on female patients, making gender-based differences less clear. Increased anterior facial height and lower hyoid position were the most strongly associated with OSA while there is more mixed evidence regarding cranial base, maxillary, and mandibular measurements. These measurements appear more pronounced on non-obese patients suggesting that craniofacial factors play a greater role in OSA among thin patients. (Table IV.C.3)

TABLE IV. C. 3.

Evidence for craniofacial anatomy as a contributor to OSA

Study	Year	LOE	Study design	Study groups	Clinical endpoints	Conclusions
Jamieson et al.216	1986	3b	Case–control study	Adult OSA patients (n = 155) and non-OSA controls (n = 17)	Lateral cephalometry	OSA patients had greater retroposition of the mandible and a more acute cranial base angle.
Hochban and Brandenburg217	1994	3b	Case–control study	Adult OSA patients (n = 403) and non-OSA controls (n = 88)	Lateral cephalometry	OSA patients had longer facial type, retroposition of the mandible, more dorsocaudally positioned hyoid bone, and an anteroposterior narrowing of the posterior airway space.
Frohberg et al.218	1995	3b	Case–control study	Adult OSA subjects (n = 30) and adult chronic snorers (n = 20)	Lateral cephalometry	OSA group had retroposition of maxilla and lower positions hyoid.
Lowe et al.219	1995	3b	Case–control study	Adult OSA subjects (n = 80) and non-OSA controls (n = 25)	Cephalometric measurements obtained from CT	OSA subjects have higher upper and lower facial heights, elongated maxillary and mandibular teeth, and proclined lower incisors.
Battagel et al.220	1996	3b	Case–control study	Adult OSA patients (n = 35) and healthy non-OSA controls (n = 24)	Lateral cephalometry	OSA patients had shorter cranial base length, shorter mandiblular body length, enlarged soft palate.
Tsai et al.221	2007	3b	Case–control study	Asian adult non-obese severe OSA patients (n = 46) and non-obese, snoring controls (n = 36)	Lateral cephalometry	Severe OSA patients had increased soft palate length and lower positioned hyoid bone compared to snorers.
Riha et al.222	2005	3b	Case–control study	Adult patients with OSA (n = 22) and non-OSA sibling controls (n = 22)	Lateral cephalometry	OSA patients had shorter mandibular body length and lower positioned hyoid compared to non-OSA siblings.
Johal et al.231	2007	3b	Case–control study	Adult OSA patients referred for oral appliance (n = 78) and gender matched healthy non-OSA controls (n = 78)	Lateral cephalometry	OSA patients the following differences compared to non-OSA controls: low set hyoid, retroposition of the mandible (SNB angle), increased soft palate length, thickness, and area.
Gungor et al.225	2013	3b	Case–control study	16 Turkish adults with OSA and 16 Turkish non-OSA controls	Lateral cephalometry	OSA subjects had shorter midface length, protrusion of upper incisors, and lower positioned hyoid bone.
Banhiran et al.226	2013	3b	Case–control study	Adult OSA patients divided into mild (n = 43) and moderate/severe (n = 98) categories compared to non-OSA controls (n = 47)	Lateral cephalometry	Moderate/severe apneics had inferiorly positioned hyoid, shorter anterior cranial base angle, and longer soft palate compared to controls. There were no statistically significant differences between mild OSA patients and non-OSA controls.
Costa et al.227	2013	4	Case series	Portuguese Adult OSA patients (n = 171)	Cephalometric measurements stratified by BMI	Caudally positioned hyoid, rotated mandible, and decreased hypopharyngeal airway caliber were associated with increasing AHI severity.
Sakat et al.228	2016	3b	Case–control study	Japanese adult severe OSA patients (n = 30) and BMI matched non-OSA controls (n = 10)	Craniofacial measurements on sagittal CT versus OSA	Retroposition of maxilla (SNA angle), retroposition of mandible (SNB angle), inferiorly positioned hyoid, shorter mandibular length were associated with OSA.
Neelapu et al.229	2017	3a	Systematic review and meta-analysis	Adult OSA patients and healthy non-OSA controls (n = 26 articles)	Lateral cephalometry	Significant weighted mean difference with insignificant heterogeneity was found for the following parameters between OSA patients and non-OSA controls: greater anterior lower facial height, lower position of hyoid bone, and decreased pharyngeal airway space. Significant weighted mean difference with significant heterogeneity was found for the following parameters between OSA patients and non-OSA controls: shorter cranial base and angle, decreased mandible, increased mandible rotation, decreased maxillary length, increased tongue area, increased soft palate area, and increased upper airway length.
Chi et al.224	2011	3b	Case–control study	Adult OSA patients(n = 55) and healthy non-OSA controls matched for age, sex, and race (n = 55)	Cephalometry based on MRI	Significant mean differences between OSA patients and non-OSA controls: Inferiorly and posteriorly displaced hyoid is associated with OSA.
Seto et al.230	2001	3b	Case–control study	Adult OSA patients (n = 29) and non-OSA controls (n = 21)	Lateral cephalometry, maxillary width, and height on dental exam	OSA patients had shorter mandibular body length and lower positioned hyoid compared to non-OSA patients. On maxillary width, OSA patients showed decreased inter-molar widths and increased palatal index.
Johal et al.223	2007	3b	Case–control study	Adult OSA patients referred for oral appliance (n = 78) and gender matched healthy non-OSA controls (n = 78)	Lateral cephalometry	OSA patients had the following differences compared to non-OSA controls: low set hyoid, retroposition of the mandible (SNB angle), increased soft palate length, thickness, and area.
Kuzucu et al.232	2020	4	Case series	89 Adult OSA subjects	Craniofacial measurements on CT	Narrower interpterygoid distance and smaller interpetrygoid to velopharyngeal length ratio was associated with increasing AHI.
Lam et al.236	2005	3b	Case–control study	Asian adults (n = 164) and White adults (n = 75)	Craniofacial measurements on physical exam obtained with patient sitting upright with head in neutral position	OSA subjects had; larger thyromental angle, neck circumference, BMI, and Mallampati scores.
Ferguson et al.238	1995	4	Case series	Adult OSA patients divided into three groups based on neck circumference (n = 161)	Lateral cephalometry	Larger neck circumference apneics had lower hyoid bone positions, smaller neck circumference patients had smaller maxillary and mandibular lengths.
Tangugsorn et al.239	2000	3b	Case–control study	Obese OSA patients (n = 57), non-obese OSA patients (n = 48), and healthy non-OSA controls (n = 36)	Lateral cephalometry	Significant differences observed between OSA patients and non-OSA controls for the following parameters: inferiorly positioned hyoid, shorter cranial base, increased gonial angle, increased lower anterior facial height, decreased posterior facial height, large tongue, and large soft palate associated with OSA compared to non-OSA controls. Significant differences observed between non-obese and obese OSA patients for the following parameters: shorter mandibles, greater posteriorly displaced and rotated mandibles, greater anterior facial height, decreased posterior facial height.
Paoli et al.240	2001	3b	Case–control study	Obese adult OSA patients (n = 46) and non-obese adult OSA patients (n = 39)	Lateral cephalometry versus obesity	Retroposition of mandible (SNB angle), shorter cranial base, shorter mandibular ramus in non-obese OSA patients versus obese OSA patients.
Yu et al.241	2002	3b	Case–control study	Asian adult obese n = (33) and non-obese OSA patients (n = 29) and non-obese, non-OSA snoring patients (n = 13)	Lateral cephalometry	OSA patients have longer soft palates and lower hyoid position compared to snorers.
Lee et al.106	2010	3b	Case–control study	Caucasian adult OSA patients (n = 74) and Chinese adult OSA patients (n = 76)	Lateral cephalometry	Chinese OSA patients had shorter cranial base angle, shorter maxillary length, and shorter mandibular bodies compared to Caucasians. Chinese OSA patients had greater severity of OSA and lower BMI compared to Caucasians.
Xu et al.242	2020	3b	Case–control study	Icelandic adults (n = 108) and Chinese adults (n = 57) with OSA defined by ODI > 10 matched for age, gender, and ODI	Craniofacial dimensions as measured on MRI	Chinese subjects had bigger ANB angle, shorter mandibular corpus length, longer mandibular ramus length, and a wider and shallower maxilla.
Cakirer et al.244	2000	3b	Case–control study	Caucasian adults (n = 364) and African American adults (n = 165) and BMI <32	Cranial and facial indices measured using calipers	Caucasians with AHI > 5 had lower cranial and facial indices compared to those with AHI <5 while African Americans had similar cranial and facial indices between the same groups.

IV.C.4 |. Genetics

It has been known for over two decades that OSA aggregates in families. This is shown in the Cleveland Family Study in the US²⁴⁵ and also in studies in Scotland,^246,247 Iceland,²⁴⁸ and Israel.²⁴⁹ The family aggregation is not explained by obesity.²⁴⁵ Individuals with a first-degree family member with OSA have approximately a two-fold increased risk of having the disorder.²⁴⁵ Structural risk factors for OSA – both soft tissues¹⁰⁷ and craniofacial^250,251 – also aggregate in families. Moreover, characteristics of the sleep study in OSA such as AHI, event duration, and measures of oxygen desaturation during sleep are heritable.²⁵² Heart rate response to arousal is also heritable²⁵³ as is cardiopulmonary coupling,²⁵⁴ a measure of sleep depth.²⁵⁵ Extreme phenotypes have been demonstrated.²⁵⁶ Individuals with high phenotype risk scores but not OSA are extreme controls, while individuals with low phenotype risk scores but severe OSA are extreme cases. This suggests that there are likely to be rare genetic variants leading to these extreme phenotypes.

While family aggregation has been shown, progress on identifying relevant gene variants has been slow. There are different pathways to disease, such that OSA is a heterogeneous disorder. Obesity plays a major role.⁹⁰ There are both soft tissue¹⁰⁷ and craniofacial^250,251 risk factors. There are also multiple physiological traits that determine risk for the disorder.^36,257,258 The initial approach to identifying genes was linkage analysis,^259,260 and most recently linkage has been used in conjunction with sequencing data to narrow windows of analysis to improve power,²⁶¹ and applied to quantitative traits related to overnight SaO₂. However, candidate gene studies have been underpowered and largely have not been replicated.²⁶² More recently, progress has been made using genome-wide association studies^263–265 or use of a special SNP panel.²⁶⁶

Genome-wide studies^263–265 have largely focused on quantitative variables obtained from the sleep study to characterize SDB that were shown to be heritable.²⁵² Based on primarily the Hispanic Community Health Study, a variant in GPR83 was found to be associated with AHI.²⁶³ However, the variant in GPR83 was specific to Caribbean background and data were unavailable for replication.

A larger genome-wide association study using data from multiple cohorts in both men and women looked for association with AHI in NREM sleep and REM sleep independently.²⁶⁵ While there were several suggestive associations, only one reached genome-wide significance, that is, a variant in RAI1 with NREM AHI in males (but not females) – a gene that has also been shown to show sexual dimorphism for cardiac disease and for adiposity traits in model organisms. Haploinsufficiency of the gene is implicated in the Smith–Magenis syndrome.²⁶⁷ Individuals with this syndrome have craniofacial abnormalities and circadian rhythm disturbances.²⁶⁸ Notably, significant sex by gene variants for sleep apnea was detected for multiple genes, supporting the importance of considering sex-specific genetic effects.

Studies have also focused on association with oxygen desaturation measures. Variants in the interleukin 18 receptor and hexokinase genes were associated with oxygen desaturation measures during sleep.²⁶⁹ This is based on a large multi-ethnic sample with replication. A later linkage-sequencing study also implicated multiple rare variants in the GTPase-activating protein DLC1²⁶¹ with oxygen desaturation measures, with findings replicated in independent samples and supported by multiple sources of information, including genetic variants, gene expression, and methylation.

Admixture mapping has also been applied.²⁷⁰ This is an analytical tool that is applied to recently admixed populations where ancestors came from isolated continents.²⁷¹ Novel variants were associated with AHI as well as oxygen desaturation measures. Notably, local African ancestry was shown to have a protective effect on the AHI and associated with shorter duration events. A region on chromosome 18q21 that included 20 variants on NARS/FECH was associated with both AHI and percentage time SaO₂ < 90%, with 12 associations that replicated in independent cohorts. Evidence for a causal role in OSA was supported by finding that FECH expression was increased in association with lower nocturnal SaO₂. These findings suggested a novel for iron metabolism in pathway influencing OSA phenotypes.

A genome-wide study that specifically examined both common and rare variants by whole genome sequencing²⁶⁴ identified novel rare variants associations with measures of desaturation, that is, ARMCX3, MRPS33, and C16orf90.

A Phenome-wide association study (PheWAS) study sought to determine support for previously identified variants.²⁷² A PheWAS seeks association between gene variants, the diagnosis of OSA in the electronic health record²⁷³ or association with quantitative traits from sleep studies done clinically.²⁷² For most of the previously described gene variants, there was no evidence of association with OSA diagnosis or with quantitative traits from the sleep study with findings from community-based cohorts which underwent standardized sleep apnea phenotyping.

Thus, studies of genetic variants in OSA are evolving. Recently, large biobanks have begun analyzing genetic associations with snoring and ICD-based OSA diagnosis.²⁷⁴ Future advances are expected as data from large clinical cohorts are combined with data from more deeply phenotyped research studies that also include data on gene function (expression, methylation).

V |. PATHOGENESIS OF OSA

V.A |. Contributory Factors for OSA: Anatomy

V.A.1 |. Nasal airway

The nasal cavity accounts for half of UA resistance.²⁷⁵ Several recent reviews indicate that nasal obstruction from septal deviation, inferior turbinate hypertrophy, nasal valve collapse, polyposis, and/or inflammation are common in patients with SDB, but the role of nasal obstruction as a contributing factor in the pathogenesis of OSA and SDB is controversial.^276–278

Pathophysiology

Nasal breathing serves important physiological functions, including humidification, heating, and filtration of inhaled air.²⁷⁵ Nasal breathing is often decreased during sleep due to increased nasal mucosal congestion when recumbent²⁷⁹ and circadian variation in nasal resistance.^280,281 Several physiologic mechanisms have been reported to explain the relationship between nasal airflow and breathing during sleep.

The first of these mechanisms is the Starling resistor model, which views the UA as a hollow tube^282,283 with a collapsible pharyngeal segment bounded by the nasal and tracheal airways. The flow of air through the collapsible pharynx is influenced by upstream nasal, downstream tracheal, and transluminal pressure gradients. For example, when the nasal airway is obstructed, nasal resistance increases. In order to maintain stable maximum airflow from the nasal passages through the pharynx, the pharyngeal pressure drops, leading to increased collapsibility at the pharynx. In addition, by Pouseille’s law, airway resistance is proportional to the length of the airway and inversely proportional to the fourth power of the radius.²⁸⁴ Thus, even a small change in nasal airway patency can significantly increase airway resistance.²⁸⁵

The second reported mechanism is high nasal resistance promoting mouth breathing, which can predispose patients to SDB. During sleep, airway resistance is higher when breathing through the mouth compared to through the nose.²⁸⁶ Mouth breathing during sleep has been associated with up to 2.5 times higher airway resistance.²⁸⁷ This increased airway resistance can narrow the pharyngeal lumen leading to more frequent obstructive events, as demonstrated by Fitzpatrick et al. who found worsening of OSA severity when mouth breathing compared to nose breathing.²⁸⁷ Similarly, another study demonstrated that jaw opening during sleep, associated with posterior tongue displacement, is greater in patients with OSA than those without OSA.²⁸⁸

A third proposed factor is the nasal ventilatory reflex, which states that decreased nasal airflow can result in decreased activation of nasal receptors leading to decreased pharyngeal muscle tone and central respiratory drive and worsening of apneic events.^289,290 This is supported by studies that have applied local anesthetic to nasal mucosa in healthy subjects resulting in significant increase in central and obstructive apnea events.^291,292

Finally, it is postulated that nitrous oxide (NO) may play a role in maintaining pharyngeal muscle tone, spontaneous breathing, and sleep regulation.²⁹³ NO is produced in the nasal cavity and sinuses, and the total inspired quantity of nitrous oxide varies with nasal airflow.²⁹⁴ Despite these studies, its role in OSA development and regulation is still not completely understood.²⁹⁵

Clinical relationship between nasal obstruction and SDB

Several studies demonstrate that patients with SDB have greater rates of nasal obstruction. A recent study by Magliulo et al.²⁷⁶ demonstrated that 70% of patients with OSA had nasal obstruction confirmed by clinical examination and rhinomanometry. Another study of 49 OSA patients found that 45% reported subjective nasal obstruction.²⁹⁶

Data are conflicting regarding patients with nasal obstruction and risk of SDB. A study of 541 snorers who underwent posterior rhinomanometry concluded that nasal obstruction is an independent risk factor for OSA.²⁹⁷ The Wisconsin Sleep Study showed that those with self-reported nocturnal congestion had a three-fold increased incidence of habitual snoring.⁶⁹ Conversely, a study by Miljeteig et al.²⁹⁸ divided 683 patients referred for PSG into three groups based on severity of nasal resistance and found no difference in apnea or snoring severity between groups.

Allergic rhinitis and SDB

A recent consensus statement on AR summarized the literature regarding the relationship between AR and sleep. The authors concluded that AR is associated with snoring, sleep fragmentation, and decreased sleep QOL, and successful treatment of AR can improve sleep quality. Many small studies with polysomnogram (PSG) data indicate that AR is associated with worsening PSG parameters, in terms of sleep efficiency, oxygen desaturation, and sleep architecture metrics, although results are mixed between studies.⁶

Nasal packing and SDB

Clinical studies and studies of normal subjects with artificially induced nasal obstruction have consistently demonstrated a greater predisposition to SDB.^299–304 These studies obstructed the nasal cavity via the following measures: inflated balloons,³⁰³ petroleum jelly soaked gauze,^{300–302,304} or tape around the nares.²⁹⁹ Likewise, many studies have shown that the application of nasal packs during the management of epistaxis and post-nasal surgery can result in worsening sleep quality and increased apneic episodes.^305–310

In summary, the literature suggests that nasal obstruction, either artificially or disease-induced, is associated with OSA-related symptoms such as snoring and poor quality sleep. Among those with SDB, nasal obstruction is highly prevalent. There does not appear to be a direct correlation between degree of nasal obstruction and SDB severity by AHI or oximetry.

V.A.2 |. Pharyngeal airway

Anatomic sites of airway obstruction during sleep have been assessed with a variety of techniques including pharyngeal pressure catheters placed at various sites in the UA, cine fluoroscopy, sleep endoscopy, CT, and MRI.³¹¹ Airway pressure monitoring studies have demonstrated that airway collapse is confined to or initiated in the oropharynx (OP) in the majority of patients.³¹²

Pharyngeal shape

Several studies have identified differences in the pharyngeal lumen shape between patients with SDB and those without. Compared with normal subjects, habitual snorers have a generalized narrowing of the pharyngeal region.³¹³ A normal pharyngeal lumen is elliptical in shape, with the long axis in the lateral dimension.³¹⁴ In contrast, the lumen of snorers and OSA patients is circular or elliptical, with the long axis in the anterior–posterior dimension.³¹⁴ This change in shape is likely a result of obstruction from the lateral pharyngeal walls. Studies have shown that the pharyngeal shape changes are most prominent at the retropalatal level²⁰⁰ and during sleep.³¹⁵

In addition to shape, the length of the pharynx may also play an important role in OSA. Cephalometric studies demonstrate that lengthening of the pharynx is related to the severity of OSA.³¹⁶ Similarly, increased rate of OSA in men may be primarily a result of a longer pharyngeal airway compared to female counterparts.³¹⁷

Soft tissue factors

Several soft tissue factors can lead to pharyngeal narrowing or pharyngeal shape changes in adults with OSA including hypertrophy, inflammation, and/or edema.³¹¹ Thickened and collapsible lateral pharyngeal walls are a well-recognized factor in the pathophysiology of adult OSA.^200,207,318 MRI studies confirm soft tissue enlargement in the UA leading to reduced size of the retropalatal airway, specifically, in OSA patients compared to control subjects.^200,207,319 Lateral pharyngeal wall edema and enlarged lateral pharyngeal fat pads have been identified via tissue specimens and imaging in OSA patients.^{200,205,320–322} Lateral pharyngeal wall edema has been hypothesized to result from multiple factors including vascular congestion, inflammation secondary to trauma from snoring vibration, or pulmonary hypertension (PH) from recurrent hypoxic pulmonary artery vasoconstriction.³²³ An MRI study demonstrated that thickness of the lateral pharyngeal wall musculature was the predominant anatomic factor causing lateral airway narrowing in OSA patients as opposed to enlarged pharyngeal fat pads.²⁰⁰

Soft palate changes have also been noted in OSA patients including thickening of the soft palate³²⁴ and enlargement of the uvula.³²⁵ Soft palate position may also be related to OSA severity^{311,316,318,326}

Many studies have assessed the relationship of tonsil size (both grade and volume) and OSA. The most recent study by Jara and Weaver studied an adult heterogeneous OSA population in the US and found that increasing tonsil grade was significantly associated with increasing OSA severity with adjusted data showing that every increase in tonsil grade as measured by the Brodsky classification system correlated to an increase in AHI of approximately 14 events/h.³²⁷ They also found that tonsil grade was more predictive of OSA than tonsil volume indicating that the actual tonsil volume may be less important than the proportion of space occupied within the OP.³²⁷ Other groups have studied the relationship between subjective (grade) or objective (volume/weight) tonsil evaluation and AHI and reported a trend toward significance³²⁸ or a significant association.^{318,329–334} To further support the role of enlarged tonsils in OSA, Camacho et al. performed an SR and MA of 17 studies evaluating tonsillectomy alone as treatment for adults with enlarged tonsils and OSA. They found a significant improvement in OSAS with a mean AHI decrease from 40.5 to 14.1 and a mean Epworth sleepiness scale (ESS) decrease from 11.6 to 6.1.³³⁵

Other groups have assessed a relationship between narrow fauces (posterior pillars of the tonsillar fossae) and OSA. Anatomically, the width of the fauces is generally defined as the diameter of the palatopharyngeal arch from oral view using a pharyngeal grading system.²³⁵ The grading system is based on the location of the palatopharyngeal arch intersection with the tongue. The presence of narrowing at the fauces has been previously defined as impingement of greater than 25% of the pharyngeal space by the peritonsillar tissues, excluding the tonsils.³¹⁸ Studies report that the presence of narrow fauces is predictive of the presence of OSA.^235,332,334 Woodson and Naganuma found that smaller endoscopically measured oropharyngeal spaces were correlated with elevated AHI.³³⁶

In summary, the OP is a common location of airway obstruction in subjects with SDB, and many anatomic factors have been identified as potential contributors to SDB including pharyngeal lumen size/shape, thickened lateral pharyngeal walls, elongated soft palate, enlarged uvula and/or tonsils, and narrow fauces.

V.A.3 |. Craniofacial structure

Craniofacial differences are present among OSA patients when compared to non-apneic controls.²²⁹ There are few studies examining how these anatomic factors contribute to the mechanism of UA collapse. A comprehensive review of the literature revealed five studies evaluating craniofacial measurements and physiologic factors that contribute to adult OSA. Sforza et al. measured critical closing pressures (upper airway critical closing pressure [Pcrit]) among 54 OSA patients and compared this to cephalometry. They found an association between greater Pcrit values (indicative of greater airway collapse) and longer soft palate length as well as with an inferiorly positioned hyoid.³³⁷ Genta et al. studied 34 Japanese-Brazilian men with OSA and also found an association between greater Pcrit values with an inferiorly and posteriorly positioned hyoid.³³⁸ Verin et al. compared UA resistance and cephalometry among OSA patients, snorers, and controls and found that an inferiorly and posteriorly positioned hyoid was associated with greater UA resistance among OSA patients.³³⁹ Watanabe et al. examined pharyngeal closing pressures at the velopharynx (VP) and OP among 54 Japanese OSA patients.³⁴⁰ Patients were stratified to either VP or VP + OP collapse which were the only two types of collapse patterns seen in their study. They found that patients with multilevel collapse (VP + OP) had a lower BMI, smaller maxillary and mandibular lengths, and a lower positioned hyoid compared to non-apneic controls while patients with only VP collapse had a larger BMI, and a lower positioned hyoid compared to non-apneic controls. No other cephalometric differences were seen among the groups. Comparing UA collapse among different ancestry background groups, Schorr et al. compared cephalometric measurements and Pcrit among Japanese-Brazilian OSA (n = 39) patients and Caucasian-Brazilians (n = 39).³⁴¹ They found both groups had similar Pcrit values but that Japanese-Brazilian group had smaller cranial base angles, mandibular and maxillary lengths while the Caucasians-Brazilian group had greater BMI, tongue volume, and tongue length.

The studies to date suggest an association between several cephalometric values including: low hyoid position, short maxillary and mandibular length, short anterior cranial base length, and acute cranial base angles with higher Pcrit values, increased UA collapsibility, and greater UA resistance. Craniofacial influences may also be associated with increased risk for multilevel UA collapse. Furthermore, comparing OSA patients of different ethnic backgrounds have shown similar Pcrit values despite differences in BMI, tongue size, and craniofacial metrics suggesting that all these factors play a role in UA collapse. However, most studies had relatively small sample sizes. Therefore, larger studies controlling for BMI, soft tissue size, are needed to better elucidate the contributions of craniofacial metrics to OSA pathophysiology.

V.A.4 |. Upper airway fat

Tongue fat appears to play an important role in linking obesity and OSA risk. The pattern of fat deposition rather than BMI may be more indicative of OSA risk. There are only a handful of studies examining this relationship. Brennick et al. showed that among obese rats, there was greater deposition of fat within the tongue compared to non-obese rats.³⁴² Kim et al. compared MRIs of the tongue between 90 obese OSA patients and compared them to 31 obese subjects without OSA. They showed that tongue fat volume calculated from the MRI was greater among the obese OSA patients versus obese subjects without apnea.²⁰⁸ Wang et al. compared tongue fat volumes among 67 obese OSA patients before and after a medical or surgical weight loss regimen.³⁴³ A reduction of tongue fat through weight loss correlated strongly with AHI reduction and was the primary UA mediator of the relationship between reductions in weight and AHI.

Although there is evidence that tongue fat is increased in obesity and that greater tongue fat is associated with OSA, how tongue fat influences the pathophysiology of OSA has not been well studied. There are no studies examining how tongue fat influences UA resistance, tongue muscle properties, and UA collapsibility. Future studies are needed to better understand the role of tongue fat in the pathogenesis of OSA.

V.B |. Contributory Factors for OSA: Physiology

V.B.1 |. Ventilatory stability/(loop gain)

“Loop gain” is an engineering term used to define the stability (“low” loop gain) or instability (“high” loop gain) of a negative feedback control system.^{180,344–347} Control of breathing is a negative feedback system in which chemoreceptors (e.g., in the carotid body; “controller”) and the lung (“plant”) try to maintain a PaCO₂ at roughly 40 mmHg. In this setting, a high loop gain leads to large fluctuations in carbon dioxide (CO₂). A useful analogy comes from consideration of room temperature in which a negative feedback control system tries to maintain a relatively constant temperature, for example, at 20 °C. If the thermostat (controller) were too sensitive, then any minor drop in room temperature would start the furnace (plant) and thus cause frequent fluctuations in room temperature. By analogy the CO₂ levels are expected to fluctuate markedly if the chemoreceptors are too sensitive to minor CO₂ perturbations. The temperature analogy is also helpful when one considers a plant that is too powerful. For example, if the room temperature were to fall to 19°C and in response the thermostat were to blast the temperature to 40°C, then the result would be major oscillations in room temperature. By analogy, if an individual were to respond to an increase in PaCO₂ from 40 to 45 mmHg with hyperventilation lowering the PaCO₂ down to 10 mmHg, then major fluctuations in CO₂ would occur.^348–350 Mathematically, loop gain is defined as the response (i.e., the increase in PaCO₂ or room temperature) divided by the disturbance (i.e., initial drop in PaCO₂ or room temperature): a major response to a minor disturbance (i.e., high loop gain) is considered destabilizing.

Loop gain has several components which are sometimes referred to as controller and plant gain. Controller gain can be thought of as chemoresponsiveness, which is the amount that ventilation changes for a given change in CO₂. High loop gain due to a high controller gain (i.e., overly sensitive thermostat in the analogy above) can be seen in clinical situations such as high altitude or heart failure.^351,352 Plant gain on the other hand is a measure of the efficiency of CO₂ excretion (i.e., too powerful furnace, or a very small room in our analogy), and is the amount that CO₂ changes for a given change in ventilation.^353,354 High loop gain due to high plant gain is clinically less common, but situations in which the lung volume decreases (e.g., supine sleep)³⁵⁵ can increase overall loop gain. Another component which is less often discussed is the so-called mixing gain which is a function of circulatory time and other factors.³⁵⁶ The overall loop gain is the product of the various gains, principally determined by the controller and the plant.

Clinically a high loop gain can express itself as periodic breathing, for example, at high altitude,³⁵⁷ or in congestive heart failure (CHF).^358,359 In OSA, the contribution of loop gain to OSA pathogenesis is less clear, but many believe it plays an important role.^{345–347,360,361} Several studies have shown an elevated loop gain in OSA compared to controls, although some debate is ongoing as to whether the observed abnormalities are a cause or effect of the disease.³⁶² In theory, a high loop gain could manifest as fluctuations in output from the central pattern generator in the brainstem. This output would affect the diaphragm as well as the UA muscles. When output from the central pattern generator is at its nadir, the UA would be vulnerable to collapse in those who are anatomically susceptible based on the low output to the UA dilator musculature. Of note, because loop gain is considered a response over a disturbance, a spontaneous respiratory event, for example, a hypopnea may be destabilizing if a patient experiences a marked response to such a respiratory disturbance. Some have coined the expression “apnea begets apnea” based on the self-perpetuating nature of the control system abnormalities.

From a surgical perspective, elevated loop gain may be important since it has been found to predict failure of sleep apnea surgeries, at least in some cases.^363–365 Given the variability in OSA pathogenesis, the strategy to improve pharyngeal anatomy/collapsibility surgically may be prone to failure in the subset of patients in whom OSA is primarily caused by abnormal ventilatory control.³⁶ In theory, interventions lowering loop gain (e.g., acetazolamide or oxygen)^366,367 may be considered rescue strategies for such patients who have residual OSA despite UA surgical intervention.³⁶⁸

It may be possible to quantify loop gain directly from inlab or home sleep studies.^369–371 More research is needed to test on reproducibility and if prospective measurements of loop gain allows identification of surgical responders, and if loop gain lowering interventions may be useful for (a subset) of patients with an incomplete response to sleep apnea surgeries.

V.B.2 |. Neuromuscular control

Patients with OSA typically have an anatomical compromise of the UA predisposing them to pharyngeal collapse.^207,372 Through protective UA dilator muscle reflexes, the activity of the muscles is increased during wakefulness in OSA compared to matched controls.²⁰¹ However, with sleep-onset there is loss of UA motor output leading to collapse of the vulnerable airway.³⁷³ Brainstem control of UA dilator muscles has been the subject of intense investigation.^374,375

The UA in humans includes 23 pairs of muscles which support its patency. The genioglossus is an important, major UA dilator muscle. The genioglossus is a large muscle comprising the substance of the tongue and maintains patency of the retroglossal airway.^350,354,376 Notably, the hypoglossal motor nucleus in the medulla of the brainstem provides the input for the genioglossus muscle. Additionally, it is a complex muscle which has both tonic activity (i.e., present throughout the respiratory cycle) as well as phasic activity (i.e., bursts with each inspiration).^377,378 The phasic activity of the genioglossus is thought to be representative of other phasic UA dilator muscles; thus, the study of genioglossus motor control may be reflective of other phasic muscles (e.g., palatoglossus and hyoglossus).³⁷⁹ The genioglossus has been shown to have state dependence, that is, has activity during wakefulness which is attenuated at sleep onset. Indeed, there is a marked fall in genioglossus activity at the alpha–theta transition which may be important in compromising pharyngeal mechanics.^373,380,381 Lastly, hypoglossal nerve stimulation (HNS) has been shown to be effective for treatment of OSA.^382–384 HNS likely acts via tongue protrusors from the medial branch of the hypoglossal nerve, that is, largely through genioglossal stimulation.^385–387

The genioglossus’ behavior is influenced by a number of important factors:

1. A negative pressure reflex (NPR) exists whereby a sub-atmospheric (negative inspiratory or suction) pressure leads to a robust activation of the genioglossus muscle.^388–390 This reflex is thought to be important in modulating UA patency since it serves to restore pharyngeal patency in the face of a collapsing perturbation. The NPR has been mapped using neurochemical techniques and is thought to be regulated by cholinergic systems in the brainstem, for example, the peri-obex region which is heavily cholinergic.³⁹¹ Pharmacological studies indicate manipulation of the cholinergic system may influence this reflex, recognizing the complex role of acetylcholine throughout the brain and systemically.³⁹² A recent pilot study showed potential benefits of the combination of oxybutynin and atomoxetine on the AHI.³⁹³ A larger multicenter trial recently completed enrollment, but the results are not yet reported (NCT03919955).

2. Chemoreflexes are also thought to be important since hypoxia and CO₂ may serve to activate the genioglossus muscle. CO₂ stimulation may have differential effects on the diaphragm versus the UA dilator muscles. The combination of negative pressure plus CO₂ may serve synergistically to activate the UA dilator muscles.^{203,317,394–396}

3. The arousal response also has a major impact on genioglossus activity. When an individual awakens from sleep, there is a robust activation of the genioglossus muscle which is thought to restore pharyngeal patency.^290,397,398 On the other hand, if sleep is maintained following UA collapse, then the accumulation of respiratory (CO₂, negative pressure) stimuli may activate pharyngeal dilator muscles and thus restore pharyngeal patency without the need for repetitive arousal from sleep. Thus, the arousal threshold has become an important therapeutic target, but its manipulation is a double-edged sword: drugs (hypnotics, sedatives) which delay arousal may allow some stabilization of breathing but may also lead to severe hypoxemia prior to arousal.^399–401 Consequently, patient selection becomes a critical factor in designing appropriate studies.^377,402 At present, no RCTs have shown improvements in hard clinical outcomes, despite some potential physiological benefits.

4. A number of neurochemical influences can affect hypoglossal motor control. Monosynaptic projections from various brain structures have been shown in animal models including locus coeruleus (adrenergic), lateral dorsal/pedunculopontine tegmentum (cholinergic), and hypothalamic (orexinergic) and raphe neurons (serotonergic) among others. These neurochemical targets may allow augmentation of hypoglossal motor output and effect genioglossus activity.^{375,403–405}

Other muscles are likely important, for example, the tensor palatini which receives its output from the mandibular branch of the trigeminal nerve.^406–409 Of note, the neurobiology and control of different motor nuclei differ substantially from the standpoint of premotor inputs and neuropharmacology. The tensor palatini has primarily tonic activity (i.e., constant activity throughout the respiratory cycle) and thus has less in the way of respiratory modulation than the genioglossus or other phasic muscles.

Neuromuscular control is an important factor in UA patency. It can potentially be manipulated either pharmacologically or via electrical stimulation. Further study regarding underlying mechanisms and clinical trials focused on hard outcomes would be encouraged.

V.B.3 |. Arousal threshold

The arousal threshold refers to the propensity to wake up from sleep. Some people have a low arousal threshold meaning they wake up easily – or with minimal stimulus – whereas other people have a high arousal threshold – meaning they require considerable stimulus to arouse.^397,410 The arousal threshold is thought to be important in OSA pathogenesis since roughly 1/3 of OSA patients are found to have a low arousal threshold and may wake up prematurely.^398,411 The accumulation of respiratory stimuli during stable sleep has been shown to activate pharyngeal dilator muscles, which in many patients is both necessary and sufficient to stabilize breathing.^394,396 Stanchina et al.²⁰³ showed a combination of CO₂ and negative intrapharyngeal pressure could lead to robust activation of the UA dilator muscles during stable sleep. Thus, patients with a low arousal threshold may not experience sufficient accumulation of respiratory stimuli to activate the dilator muscles and thus repetitive airflow limitation is predicted. In contrast, patients with a high arousal threshold could get sufficient magnitude of respiratory stimuli for adequate duration to activate pharyngeal dilator muscles and thus stabilize breathing. The observation that even severe OSA patients have some periods of spontaneously occurring stable breathing has yielded discussion regarding potential therapies to manipulate the arousal threshold.^398,412 Some view the arousal threshold as a double-edged sword. That is, therapies to increase the arousal threshold may be beneficial if this intervention allows dilator muscle activation and stabilization of breathing.⁴¹³ On the other hand, an agent which raises the arousal threshold may yield substantial hypoxemia and hypercapnia which could impact end organ function.⁴¹⁴ Thus, agents such as sedatives or hypnotics which can raise the arousal threshold may be beneficial at least in theory for select patients. However, the existing data suggest that any improvements in apnea which occur with these agents are relatively modest.^400,401 Consequently, combinations of therapy may well be required to eliminate apnea using this approach.⁴¹⁵ Another consideration is that these pharmacological agents have risks and benefits like all interventions and thus carefully performed outcome-based studies will be needed before any clinical recommendations can be made.

One strategy which has been discussed in the context of combination therapies is that of surgical rescue. Some data suggest that a low arousal threshold may be a risk factor for failure of UA surgery to achieve a surgical cure.^363,364,416 In theory, the elevation of the arousal threshold may be a therapeutic target whereby an agent (e.g., trazodone or eszopiclone) could be used to elevate the arousal threshold in patients who have residual apnea, for example, following uvulopalatopharyngoplasty (UPPP).³⁹⁹ Such strategies would need to be studied carefully in the context of patient reported outcomes and hard endpoints.

Regarding the assessment of the arousal threshold, several techniques have been employed. The “gold standard” measurement was considered either esophageal manometry or intrapharyngeal pressure catheter measurements.⁴¹⁰ However, Edwards et al. reported a regression formula which has considerable value in estimating the arousal threshold using clinically accessible data, such as the degree of hypoxemia, the arousal index, the AHI, and the occurrence of apneas versus hypopneas.⁴¹¹ Using this approach more than 60% of the variance in the arousal threshold can be predicted. In addition, Sands et al. developed a technique using signal processing of the polysomnographic recordings which can also estimate the arousal threshold using clinically available data.³⁷⁰

One important consideration is the fact that the arousal threshold is not a fixed trait but rather a dynamic phenomenon which changes with treatment.^417,418 For example, many patients with a high arousal threshold with sleep apnea will experience a lowering of arousal threshold over time on therapy. This observation leads to speculation that the elevated arousal threshold seen in some OSA patients may be an adaptive phenomenon whereby elevation in the arousal threshold may allow the accumulation of respiratory stimuli, which could ultimately help to improve sleep to some extent.⁴¹⁸ Another extension of this logic is that CPAP treatment in some patients may lead to insomnia since the lowering of arousal threshold may be associated with worsening of sleep quality.⁴¹⁹ The same argument could be made for non-CPAP therapies as well. Further study is clearly required to determine the importance of arousal threshold in OSA pathogenesis and its importance in treatment of OSA both adjunctively (e.g., with CPAP)⁴²⁰ and as a rescue strategy (e.g., following failed UPPP).

VI |. DIAGNOSING OSA

VI.A |. Questionnaires for OSA

In-laboratory PSG is the gold standard for diagnosis of OSA, but can be expensive, inconvenient, and difficult to access. This is particularly true when considering screening in the general population or perioperatively. Additionally, PSG may not be readily available to all clinicians. Validated questionnaires are easily administered in all clinical settings and offer a rapid point-of-care tool to risk-stratify patients. Equally important is the assessment of the QOL impairment in patients with OSA as physiologic sleep measures are poor descriptors of QOL.⁴²¹ Patient-reported outcome assessment is especially important when evaluating changes in QOL after treatment or over time. This review is based on more than 20 studies with reports ranging from Level 1a to 2b (overall grade C evidence) (Table VI.A.1).

TABLE VI.A.1.

Evidence for the role of validated screening, functional outcomes, and health-related QOL questionnaires for OSA

Study	Year	LOE	Study design	Study groups	Clinical endpoint	Conclusion
Nagappa423	2015	1a	Systematic review	1. Sleep, 2. Surgical, 3. General	Meta-analysis of STOP-BANG	STOP-BANG is adequate screening tool in sleep and surgical patients.
Abrishami429	2010	1a	Systematic review	1. Sleep, 2. Surgical, 3. General	Meta-analysis of OSA screening questionnaires	STOP-BANG and Berlin had similar sensitivities. Studies were heterogenous.
Amra430	2017	1a	Systematic review	1. Sleep, 2. Surgical, 3. General	Comparison of OSA screening questionnaires.	STOP-BANG had best sensitivity and specificity for moderate OSA.
Chiu431	2016	1a	Systematic review	1. Sleep, 2. Surgical, 3. General	Meta-analysis of STOP-BANG, Berlin, STOP, and ESS	STOP-BANG most accurate at detecting OSA at all severity cutoffs.
Senaratna422	2017	1a	Systematic review	1. Sleep, 2. Surgical, 3. General	Meta-analysis of Berlin questionnaire	Berlin is adequate screening tool in sleep and surgical patients. Studies were heterogenous.
Billings437	2014	1b	Cohort	Sleep clinic	Comparing psychometric properties of SAQLI and FOSQ	Both demonstrate responsiveness to CPAP treatment. Comparable reliability and validity.
Coutinho424	2019	1b	Cohort	Sleep clinic	Validate NoSAS as screening tool for OSA	NoSAS is an adequate screening tool for OSA.
Marti-Soler426	2016	1b	Cohort	General	NoSAS derivation and validation. Comparison to other screening questionnaires.	NoSAS adequate screening tool compared with Berlin and STOP-BANG.
Silva441	2016	1b	Cohort	1. OSA, 2. non-OSA	Comparison of SAQLI, FOSQ, SF-36 scores with OSA severity	Scores correlate w/OSA severity in females, but not males. All demonstrated convergent validity.
Abma444	2016	2a	Systematic review	1. Sleep, 2. Non-sleep	Review of quality of PROMs for OSA	Most PROMs not adequately assessed due to low quality studies.
Mollayeva 443	2016	2a	Systematic review	1. Sleep, 2. Non-sleep	Meta-analysis of psychometric properties of PSQI	Adequate reliability and validity as QOL tool.
Kendzerska436	2013	2a	Systematic review	1. Sleep, 2. Non-sleep	Overview of psychometric properties of ESS	Convergent validity lower than expected. Few high quality studies.
Rong428	2019	2b	Cohort	1. OSA, 2. Primary snorers	Comparison of NoSAS to STOP-BANG	NoSAS and STOP-BANG had comparable sensitivities and ROC.
Flemons439	2002	2b	Cohort	OSA	Evaluate psychometric properties of SAQLI	Adequate properties as a QOL instrument.
Flemons438	1998	2b	Cohort	1. OSA, 2. Primary snorers	Derivation of SAQLI	Excellent internal consistency and responsiveness is present.
Hong425	2018	2b	Cohort	Sleep clinic	Comparison of NoSAS to STOP-BANG, ESS, and Berlin	NoSAS is an adequate screening tool for OSA.
Lacasse440	2002	2b	Cohort	Sleep clinic	Validation of SAQLI	Adequate validity and responsiveness to treatment.
Peng427	2018	2b	Cohort	1. OSA, 2. Primary snorers	Comparison of screening questionnaires	NoSAS, Berlin, and STOP-BANG have comparable sensitivities.
Weaver442	1997	2b	Cohort	1. Sleep, 2. Non-sleep	Derivation of FOSQ	FOSQ had excellent reliability and demonstrated known-groups validity.

Abbreviations: ESS, Epworth Sleepiness Scale; FOSQ, functional outcomes sleep questionnaire; PROMs, patient reported outcomes measure; PSQI, Pittsburgh sleep quality index; SAQLI, sleep apnea quality of life index.

V1.A.1 |. Screening questionnaires

Clinical questionnaires validated to assess the risk of OSA include the Berlin,⁴²² STOP-BANG,⁴²³ and NoSAS^424–428 screening tools.^429–431 The Berlin questionnaire contains a total of 10 items divided into three categories: (1) daytime sleepiness, (2) BMI and hypertension, and (3) height and weight, where a positive score in two categories indicates high risk for OSA.⁴²² The STOP-BANG is an 8-question survey including four subjective and four objective items with each positive item contributing 1 point for a maximum score of 8 points.⁴²³ The NoSAS is a newer tool developed in 2016 in sleep clinic patients and includes five items assessing mostly objective data such as BMI and neck circumference with scores ranging from 0 to 17 and a score ≧8 denoting high risk for OSA.⁴²⁶

A clinically useful questionnaire that assesses OSA risk should be able to identify patients with clinically relevant OSA (AHI ≥ 15), where an increased risk of cardiopulmonary disease and mortality has been established (true positive)^432–434 and minimize the incorrect labeling of individuals without clinically relevant OSA as having the condition (false positive). The Berlin and STOP-BANG questionnaires have sensitivities of 82% and 94%, respectively,^422,423 while the NoSAS has a sensitivity of 65%–90% in identifying sleep clinic patients with clinically relevant OSA.^424–431 Additionally, the Berlin and STOP-BANG questionnaires have reported sensitivities of 79%–82% and 91% respectively among surgical patients and 89% and 88% respectively among general clinic patients.^422,423 The ROC curves for the STOP-BANG at different OSA severity classes showed a STOP-BANG cutoff score of 5 had the best combination of sensitivity and specificity for moderate-severe OSA (AHI ≥ 15) in sleep medicine (60%, 72%) and surgical populations (45%, 56%).⁴²³ At this cutoff score, the STOP-BANG also has the best combination of sensitivity and specificity for severe OSA (AHI ≥ 30) in sleep clinic patients (66%, 66%) and surgical patients (56%, 74%).⁴²³

V1.A.2 |. Functional status measures and health-related quality of life questionnaires

It is well documented that, in general, patients with sleep apnea have limitations in daily activities and poorer QOL than normal healthy subjects.⁴³⁵ Thus, assessment of functional status and QOL is highly important to evaluate in addition to physiologic sleep measures when assessing OSA patients. Validated functional status and QOL measures most used in the clinical and research settings include the ESS,⁴³⁶ Sleep Apnea Quality of Life Index (SAQLI),^437–441 Functional Outcomes of Sleep Questionnaire (FOSQ),^437,441,442 and the Pittsburgh Sleep Quality Index (PSQI).⁴⁴³

Questionnaires can be divided into functional status questionnaires, which measure the ability to perform daily activities, and health-related QOL questionnaires, which capture the unique personal response to not being able to perform routine activities. Functional status questionnaires include the PSQI, ESS, and FOSQ. The PSQI was developed to measure multiple components of sleep quality and includes questions on sleep propensity (likelihood of falling asleep) over the past 1 month.⁴⁴³ The ESS assesses likelihood of falling asleep across eight scenarios on a scale of 0–3 with a higher score indicating more sleepiness.⁴³⁶ The psychometric properties of the FOSQ suggest that it offers a unique self-report measure of functional status as it relates to the impact on daily activities.⁴⁴² The FOSQ has 30 questions in five domains: activity, vigilance, intimacy and sex, general productivity, and social outcome.^437,441 The SAQLI is a health-related QOL questionnaire that is diseasespecific and includes questions focused on evaluating functioning across daily aspects of life affected by poor sleep.^{437,439–441} The SAQLI includes 35 questions across four domains: daily functioning, social interactions, emotional functioning, and symptoms and can be combined with a fifth domain that assesses negative impacts of treatment.^{437,439–441}

When evaluating functional status and health-related QOL surveys, it is important to consider the reliability, validity, and responsiveness of these measures. Reliability of these questionnaires were reported in terms of internal consistency, the degree to which individual items on a survey are related as measured by Cronbach’s α, and test–retest reliability, the consistency of scores for the same patient over time measured by the intraclass correlation coefficient (ICC). All four of the QOL questionnaires demonstrated acceptable internal consistency (α ≥ 0.7)^436,443 while the FOSQ and SAQLI demonstrated excellent internal consistency (α ≥ 0.9).^{437,438,440,442} The FOSQ and SAQLI demonstrated excellent test–retest reliability (ICC ≥ 0.9)^439,440,442 while this test characteristic was not clearly assessed for the ESS and PSQI. Four studies assessed this factor in the ESS, but results of the test and retest populations are not comparable due to a long time interval between administrations or due to a change in test conditions.⁴³⁶ The data on test–retest reliability in the PSQI is limited to three studies that either did not include patients with a sleep disorder or had a long time interval between test administration.⁴⁴³ Measurement error, the error in a measurement that is not due to true differences in the construct being measured, was not reported for any questionnaire.

The validity of QOL and functional status instruments were measured in terms of convergent, divergent, and discriminative validity. Convergent validity measures the degree to which scores of a questionnaire correlate with other questionnaires with related constructs. The PSQI had strong correlations (r ≥ 0.7) with other questionnaires measuring sleep quality.⁴⁴³ The SAQLI showed moderate correlations (0.3 ≤ r < 0.7) with related domains of general health-status questionnaires such as the 36-item Short Form (SF-36) and the Global QOL Scale.^437–441 The FOSQ showed moderate correlations with related domains of the SF-36 and the Sickness Impact Profile.^437,441,442 The ESS showed a moderate correlation to the Maintenance Wakefulness Test and poor correlation (r ≤ 0.3) with the multiple sleep latency test (MSLT), both of which are objective daytime sleepiness measures.⁴³⁶ PSG measures (AHI, minimum O₂ saturation) were either poorly or not significantly correlated with any of the questionnaires.^{436,437,441,443,444}

Divergent validity assesses whether instruments with unrelated constructs in fact have low correlations with one another. Divergent validity was reported for the PSQI, which had non-significant correlations with unrelated instrument measures of bladder dysfunction and psychopathology.⁴⁴³ No other questionnaires reported on this measure. Discriminant validity refers to the ability of an instrument to distinguish between patients with sleep disorders and normal subjects. Discriminant validity was present for the ESS, PSQI, and FOSQ, but was not reported for the SAQLI.^436,442,443

The ability of an assessment to measure change over time is important especially if the instrument is used in treatment effectiveness studies. The SAQLI and FOSQ both demonstrated the ability to detect change over time after treatment initiation.^437–440 Both showed large effect sizes (d ≥ 0.8) after CPAP treatment in patients with OSA.^437,439 In one study, the FOSQ showed a larger improvement in scores in patients who used >4h of CPAP nightly while there was no difference in scores for the SAQLI.⁴³⁷ (Table VI.A.1) An important limitation of this review is the heterogeneity of studies reporting on both screening and QOL questionnaires. Definitions of hypopnea and method used to detect OSA varied between studies resulting in wider ranges of reported sensitivity and specificity. Additionally, several studies used translated versions of the questionnaires that were not previously validated in those languages. Another limitation is the lack of highquality validation studies for all QOL measures. Future research should focus on using uniform definitions of hypopnea/apnea and utilize the gold standard PSG in validating OSA questionnaires and to further validate QOL surveys to estimate their adequacy as an outcome measure.

Overall, validated clinical questionnaires may be used as a tool to identify patients at high risk for OSA, monitor response to treatment, or evaluate the functional status and health-related QOL of patients. To screen patients suspected of sleep apnea in the clinic setting or preoperatively before planned surgery, the STOP-BANG questionnaire with a cutoff score of 5 is recommended. To measure functional status related to OSA, the FOSQ is recommended as it is disease-specific, assesses the effect of sleepiness comprehensively across domains, and has the strongest evidence for all measurement properties. (Table VI.A.2)

TABLE VI.A.2.

Functional status and health-related QOL questionnaires

	Internal consistencya	Test–retest reliabilityb	Measurement error	Convergent validityc	Divergent validity	Discriminative validity	Responsiveness
ESS	Acceptable	**	–	Poor to moderate	–	Present	–
PSQI	Acceptable	**	–	Strong	Presentd	Present	–
SAQLI	Excellent	Excellent	–	Moderate	–	–	Present
FOSQ	Excellent	Excellent	–	Moderate	–	Present	Present

Abbreviations: – Not reported;

^**not clearly assessed. ESS, Epworth Sleepiness Scale; FOSQ, functional outcomes sleep questionnaire; PSQI, Pittsburgh sleep quality index; SAQLI, sleep apnea quality of life index.

^aBased on Cronbach’s α values: unacceptable (α < 0.5), poor (0.5 ≤ α < 0.6), questionable (0.6≤ α <0.7), acceptable (0.7 ≤ α < 0.8), good (0.8 ≤ α < 0.9), excellent (α ≥ 0.9).

^bBased on intraclass correlation values: poor (ICC < 0.5), moderate (0.5 ≤ ICC < 0.75), good (0.75 ≤ ICC < 0.9), excellent (ICC ≥ 0.9).

^cReported as degree of correlation with questionnaires measuring related constructs: poor (r < 0.3), moderate (0.3 ≤ r < 0.7), strong (r ≥ 0.7).

^dIndicates presence of poor or non-significant correlations with unrelated questionnaires.

VI.B |. Screening for OSA

VI.B.1 |. Primary care setting

Although a myriad of screening instruments exist for OSA, the majority of the literature centers around four main questionnaires: (1) Berlin, (2) STOP, (3) STOP-BANG, and (4) ESS. The Berlin questionnaire is composed of 10 items grouped into three domains to assess snoring severity, excessive daytime sleepiness, and history of hypertension or obesity.⁴⁴⁵ The STOP questionnaire is constructed in an even simpler format, with four yes or no questions corresponding to snoring, tiredness, observation (of apneas), and BP.⁴⁴⁶ The STOP-BANG questionnaire consists of the STOP items with an additional four questions for BMI, age, neck size, and gender.⁴⁴⁷ Lastly, the ESS is an 8-item questionnaire asking participants to rate their likelihood of falling asleep during various daytime activities.⁴⁴⁸ The convenience and low cost of questionnaires are well suited for the primary care setting.^449–452 The four questionnaires mentioned above are the most widely discussed with respect to OSA screening in primary care and have each been validated on multiple occasions across a wide range of demographics.

Several studies have evaluated the differences in sensitivity and specificity between screening questionnaires.^422,453,454 A recent multicenter prospective study enrolled 812 patients diagnosed with type 2 diabetes, obesity, HF, or resistant hypertension who completed multiple OSA screening questionnaires. Following a diagnostic sleep study, the STOP-BANG questionnaire (with “high-risk” set to 3 points or above) had a greater sensitivity than either the Berlin questionnaire or the OSA50 questionnaire (95%, 75%, and 88% sensitivity, respectively). After increasing the “high-risk” threshold to 5 or above on the STOP-BANG, it also had a greater specificity (69% compared to 38% for the Berlin and 21% for the OSA 50).⁴⁵⁵

The largest comparative study to date was a retrospective review by Silva et al. that included 4770 patients from the SHHS. For this cohort, the STOP-BANG had the highest sensitivity for moderate-to-severe OSA (87%). The 4-Variable screening tool (sex, BMI, BP, and snoring) had the highest specificity.⁴⁵⁶ A separate cohort study of 212 patients in Beijing similarly found the STOP-BANG questionnaire to be of superior predictive value than the ESS, Berlin, or STOP questionnaires.⁴⁵⁷ Additionally, a recent MA confirmed the STOP-BANG possessed the greatest pooled sensitivity. The ESS questionnaire, however, was shown to have markedly higher pooled specificity compared to the STOP, STOP-BANG, and Berlin.⁴³¹ Although the STOP-BANG is the most sensitive screening questionnaire, the evidence is unclear as to which questionnaire is the most specific. In the primary care setting, a screening test with higher sensitivity may be preferred to ensure adequate diagnostic investigation and mitigate sequelae of untreated OSA.

There is very little evidence on the benefits of widespread screening for OSA in asymptomatic adults without comorbidities.⁴²⁹ A 2017 SR supervised by the US Preventive Services Task Force analyzed 110 studies with over 46,000 patients and found no information relating OSA screening to health outcomes.⁴⁵⁸ To adequately assess the impacts of screening, a randomized trial assigning patients to screening and control groups – with adequate sample size and follow-up – would be required. With respect to costs, the general sentiment behind the creation of the STOP and STOP-BANG questionnaires was to simplify previous instruments and reduce the burden of measurement on survey participants.^447,457 Although there is no direct evidence describing the financial burden of implementing OSA screening, reports of other screening instruments have suggested minimal burden and disruption of the clinical workflow.^{446,459–461}

In summary, the literature with respect to OSA screening in the primary care setting remains ambiguous. Support exists for the superior sensitivity and performance of the STOP-BANG as a screening tool, but no evidence has shown the long-term health impact of screening compared with no screening. Although the harms associated with screening appear limited, the precise cost-utility tradeoff of screening in the primary care setting has yet to be assessed. (Table VI.B.1)

TABLE VI.B. 1.

Evidence for screening for OSA in the primary care setting

Study (authors)	Year	LOE (1a–5)	Study design	Study groups	Clinical endpoints	Conclusion
Jonas et al.458	2017	2a	SR	110 Studies (1994–2016)	Association of OSA screening with health outcomes	Insufficient evidence for OSA screening among asymptomatic adults; no studies were found comparing screening to no screening.
Abrishami et al.429	2012	2a	SR	10 studies (1988–2008): one retrospective review, nine prospective cohorts	Sensitivity and specificity values of the Berlin, STOP, STOP-BANG, Wisconsin, and SASDQ questionnaires	Reviewed OSA screening questionnaires show inconsistent predictive values.
Kee et al.455	2018	2b	Prospective multicenter cohort	Patients with diagnosed comorbidities (diabetes, obesity, resistant hypertension, and heart failure) and a baseline home sleep study who completed the STOP-BANG, Berlin, and OSA50 questionnaires	Sensitivity and specificity values	STOP-BANG had the best sensitivity for screening purposes in patients with diabetes, obesity, resistant hypertension, and heart failure.
Silva et al.456	2011	2b	Prospective multicenter cohort	Sleep Heart Health Study (SHHS) Participants who completed baseline PSG, 4-Variable Screening Tool, STOP, STOP-BANG, and ESS questionnaires	Sensitivity and specificity values	STOP-BANG had the highest sensitivity while the 4-Variable had the highest specificity for moderate-to-severe OSA.
Luo et al.457	2014	2b	Prospective single center cohort	Patients both with and without OSA who completed the STOP, STOP-BANG, Berlin, and ESS questionnaires	Sensitivity and specificity values	STOP-BANG had the highest sensitivity among the reviewed questionnaires.
Tan et al.454	2017	2b	Prospective multicenter cohort	Singapore Health Study 2012 (SH2012) participants who completed home sleep study and Berlin questionnaire	Sensitivity and specificity values	Berlin questionnaire possesses a high sensitivity and NPV for severe OSA.
Senaratna et al.422	2017	2a	SR and meta-analysis	35 Studies (2008–2016): 26 prospective cohorts, five retrospective reviews, four cross-sectional analyses	Pooled sensitivity and specificity values of the Berlin questionnaire	Limited evidence to suggest utility of Berlin questionnaire in general population.
Chiu et al.431	2017	2a	SR and meta-analysis	108 Studies (1999–2016)	Pooled sensitivity and specificity values of Berlin, STOP, STOP-BANG, and ESS questionnaires	The STOP-BANG questionnaire is the most accurate tool for detecting mild, moderate, and severe OSA.
Tan et al.453	2016	2b	Prospective multicenter cohort	Singapore Health Study 2012 (SH2012) participants who completed home-based sleep study and STOP-BANG questionnaire	Sensitivity and specificity values	STOP-BANG has moderate sensitivity and high NPV for moderate-to-severe OSA.

VI.B.2 |. Perioperative setting

The prevalence of OSA among surgical patients is 7%–10% and approximately 70% in patients undergoing bariatric surgery.^462–464 The diagnosis of OSA is associated with increased risk for postoperative complications in general and increases cardiac adverse events,^465–468 respiratory failure, oxygen desaturation, and intensive care admission.^{467,469–477} Moreover, these complications have increasingly resulted in malpractice lawsuits.^478,479 Data about postoperative mortality is mixed but generally shows no increased 30-day mortality, which is thought to be due to increased monitoring in this group of patients.^{468,472,480–485} Identifying at risk patients is critical to perioperative planning given the increased risk of postoperative cardiopulmonary complications with OSA.⁴⁸⁶

As most surgical patients with OSA are undiagnosed, the Society of Anesthesia and Sleep Medicine Guidelines on preoperative screening and preparation of patients with OSA strongly recommends screening for OSA in the preoperative period.⁴⁸⁶

Screening will prompt practitioners to create a safer postoperative management plan, such as extended monitoring, the use of CPAP therapy, using less opioids and sedatives, and more regional anesthesia techniques whenever applicable. It also helps in deciding on the eligibility for ambulatory procedures.⁴⁸⁷

Sleep study testing is the standard for OSA diagnosis. However, it is not cost-effective or practical to screen patients using PSG in the perioperative setting due to highcost and inconvenience. There are often logistical issues of coordinating sleep study testing and surgery scheduling timelines. However, the preoperative assessment is a cost-effective opportunity to screen for and select high risk patients who can then be offered sleep study and treatment for OSA.⁴⁸⁸ In order to facilitate OSA detection, several predictive screening tools have been proposed. Using the questionnaires with clinical criteria is easy, inexpensive, and has an acceptable sensitivity. However, they should not be used as a diagnostic tool without confirmatory sleep study testing given the low to moderate specificity and the potential for misdiagnosis.^{446,464,486,489}

Screening components

It is best if the patient is seen a few weeks before surgery, so a proper diagnosis and treatment are initiated. Patients can be screened by their primary care physician, by their surgeon, or by the anesthesia care provider in the preanesthesia clinic. Screening may take place in the hospital, or on the day of the procedure.

Preoperative screening

Preoperative evaluation of a patient for potential identification of OSA should include: medical history, screening questions, and physical exam. A focused review of history includes prior airway difficulty, hypertension, CV problems, craniofacial abnormalities, and any previous sleep studies. A physical examination should include evaluation of the airway, neck circumference, tonsil size, and tongue volume.⁴⁹⁰

Screening methods in the perioperative setting

Several pre-operative screening questionnaires are available. Only STOP-BANG, Berlin, American Society of Anesthesiologists (ASA) checklist, and perioperative sleep apnea prediction have been evaluated and validated in surgical populations.⁴⁹¹

STOP-BANG (SB) questionnaire

The questionnaire builds on the STOP questionnaire (snoring, tiredness, observed apnea, and high BP) and adds the BMI, age, neck circumference, and gender criteria (BANG). Each item in the questionnaire is score as (yes/no) and is scored as (0–1) for a total score of 8. The STOP-BANG questionnaire has a high sensitivity where a score of ≥3 score is associated with an increased risk of moderate-to-severe OSA,^470,492,493 and a higher rate of perioperative complications.^{469,471,473,475–477,485,486,494–496} On the other hand, the specificity ranges from 30% to 43%, which may yield high false positive rates and may increase unnecessary testing.⁴⁴⁶ The addition of serum HCO₃⁻ of at least 28 mmol/L to the STOP-BANG score improves the specificity but decreases its sensitivity.⁴⁴⁷

Perioperative Sleep Apnea Prediction Score (P-SAP)

The P-SAP score includes the following: snoring, thyromental distance <6 cm, type 2 diabetes mellitus (DM), high BP, Mallampati class III or IV, BMI > 30 kg/m², age > 43 years, neck circumstance >40 cm, and male gender. A diagnostic threshold P-SAP score of 2 or higher showed excellent sensitivity (0.939) but low specificity (0.323), whereas a P-SAP score of 6 or higher had low sensitivity (0.239) but excellent specificity (0.911).⁴⁶⁴

American Society of Anesthesiologists (ASA) checklist

ASA OSA scoring checklist combines physical characteristics, history of apparent airway obstruction, and somnolence.⁴⁹⁷ Physical characteristics include signs and symptoms suggestive of the diagnosis of OSA (BMI > 35, craniofacial abnormalities, and neck circumference). Symptoms of apparent airway obstruction include: loud snoring, frequent snoring, observed pauses in breathing during sleep, and awakening from sleep with choking sensation. Somnolence is present if the patient reports fatigue despite adequate “sleep” or falling asleep easily in a non-stimulating environment. The ASA checklist was validated in surgical population with a moderate sensitivity (72%–87%) which was comparable to both the STOP and Berlin questionnaires.⁴⁸⁹

The Berlin questionnaire⁴⁴⁵

The Berlin questionnaire focuses on some risk factors for OSA. It includes questions about snoring, daytime sleepiness, and fatigue. Patients are also asked to provide information on history of hypertension, age, weight, height, sex, neck circumference, and ethnicity. Its sensitivity is 69–87% in surgical patients.⁴⁸⁹

DES-OSA score

The DES-OSA is a morphologic OSA prediction score; it takes into account the Mallampati score (MP), thyroid chin distance (DTC), BMI, neck circumference, and sex.⁴⁹⁸ Patients with MP class II scores 2 points, and 3 points if they are class III or IV. They are awarded points based on their DTC, neck circumference, and BMI. Patients who score 5 or above are likely to have OSA diagnosis.

Oxygen Desaturation Index (ODI)

Using overnight oximetry has been found to have a very good correlation with AHI on PSG in surgical patients.⁴⁹⁹ ODI is defined as the number of (3% or 4%) desaturation episodes per hour. The ODI > 10 demonstrated a sensitivity of 93% and a specificity of 75% to detect moderate and severe OSA. (Table VI.B.2)

TABLE VI.B. 2.

Association of OSA with worse postoperative outcomes

Study	Year	LOE	Objective	Study design	Study groups	Clinical endpoints	Conclusions
Khiani500	2009	3b	If the diagnosis of OSA using Berlin questionnaire will result in postoperative oxygen desaturation that needs supplemental oxygen following sedation for gastrointestinal procedures.	Cross sectional study	Berlin Questionnaire in patients undergoing sedation for GI procedures.	Rates of transient hypoxia, defined as a pulse oximetry measurement less than 92% requiring an increase in supplemental oxygen.	No significant difference in rate of transient hypoxia between the high- and low-risk groups (odds ratio 1.48; 95% CI, 0.58–3.80). Suggests that the majority of patients with no OSA can undergo conscious sedation for routine endoscopic procedures with standard monitoring practices.
Vasu et al.477	2010	2b	To determine whether high risk -OSA (HR-OSA) ≥ 3 on preoperative STOP-BANG (SB) questionnaire correlated with a higher rate of postoperative complications.	Retrospective Cohort study	Adults undergoing elective surgery at a tertiary care center who were administered the SB questionnaire. 135 patients were identified, 56 (41.5%) had high risk scores for OSA.	Pulmonary complications: Hypoxemia, atelectasis, pulmonary embolism, or pneumonia. Cardiac complications: New-onset atrial fibrillation, systemic hypotension, or myocardial infarction.	Patients at high risk of OSA had a higher rate of postoperative complications compared with patients at low risk (19.6% vs. 1.3%; p < 0.001). The SB questionnaire is useful for preoperative identification of patients at higher risk for complications.
Coté et al.471	2010	2b	To de termine: The preoperative prevalence of OSA using SB questionnaire related to the need for airway maneuvers and sedation related complications.	Prospective Cohort study	231 consecutive patients undergoing advanced gastrointestinal procedures under sedation were identified by SB and were classified as high risk for OSA (score, ≥ 3 of 8; SB+) or low risk (SB<3).	-Airway Maneuvers (AM): - defined as a chin lift, modified mask ventilation, nasal airway, bag-mask ventilation, and endotracheal intubation. -Sedation-related complications (SRCs) were defined as any duration of pulse oximetry less than 90%, systolic blood pressure less than 90 mm Hg, apnea, or early procedure termination.	The prevalence of SB+ was 43.3%. The frequency of hypoxemia was significantly higher among patients with SB+ than SB− (12.0% vs. 5.2%; relative risk [RR], 1.83; 95% confidence interval [CI], 1.32–2.54). The rate of AMs was also significantly higher among SB+ (20.0%) compared with SB− (6.1%) patients (RR, 1.8; 95% CI, 1.3–2.4).
Pereira et al.475	2013	2b	SB score can predict the risk of a patient having OSA and to evaluate the incidence high risk OSA (HR-OSA) in surgical patients admitted to the Post-Anesthesia Care Unit (PACU).	Prospective Cohort study	340 adult patients after noncardiac and non-neurological surgery were admitted to PACU. 179 (52%) were considered HR-OSA. HR-OSA if SB score≥3 and Low-risk of OSA (LR-OSA) if SB score<3 (LR-OSA).	Postoperative respiratory complications; residual neuromuscular blockade (NMB); hospital length of stay.	HR-OSA had: 1. More frequent hypoxia in PACU (9% vs. 3%, p = 0.012). 2. Higher incidence of residual neuromuscular blockade (NMB) (20% vs. 16%, p = 0.035). 3. Longer hospital stays.
Chia et al.495	2013	2b	To evaluate if the HR-OSA by SB can predict difficult intubation	Prospective cohort study	200 patients undergoing surgery under general anesthesia. 83 with HR- OSA based on the SB score ≥ 3. Mallampati score and tonsil size, as well as demographic data, were recorded preoperatively for all patients.	Difficulty of intubation defined by Cormack & Lehane grade III or IV or the need of an intubation aid, or the need of three or more intubation attempts. 7% had difficulty of intubation.	Rate of difficult intubation was higher in HR- OSA patients than in LR-OSA patients. (13.3% vs. 2.6%. (p < 0.001). A SB ≥ 3 was seen more frequently in the difficult intubation patients (78.6% vs. 38.7%) (p = 0.009).
Acar494	2014	2b	To evaluate if the HR-OSA by SB can predict difficult intubation	Prospective cohort study	200 patients undergoing surgery under general anesthesia. 83 with HR- OSA based on the SB score ≥ 3. Mallampati score and tonsil size, as well as demographic data, were recorded preoperatively for all patients.	Difficulty of intubation defined by Cormack & Lehane grade III or IV or the need of an intubation aid, or the need of three or more intubation attempts. 7% had difficulty of intubation.	Rate of difficult intubation was higher in HR- OSA patients than in LR-OSA patients. (13.3% vs. 2.6%. (p < 0.001). A SB ≥ 3 was seen more frequently in the difficult intubation patients (78.6% vs. 38.7%) (p = 0.009).
Mehta et al.501	2014	2b	To identify the prevalence of OSA by using the SB questionnaire and subsequent risk factors for airway interventions (AI) and sedation related adverse events (SRAE)	Prospective cohort study	243 patients undergoing routine gastrointestinal procedures under sedation. HR-OSA = SB ≥ 3 score, low risk −OSA = SB score < 3	Airway interventions (AI): Chin lift, mask ventilation, placement of nasopharyngeal airway, bag mask ventilation, unplanned endotracheal intubation, hypoxia, hypotension, or early procedure termination.	48% with HR- OSA. An SB score ≥3 was found not to be associated with occurrence of AI (relative risk [RR] 1.07), or SRAE (RR 0.81, 95% CI, 0.53–1.2) after adjustment for propofol dose, BMI, smoking, and age. HR-OSA patients are not at higher risk for airway intervention or sedation-related adverse events SRAE.
Proczko et al.485	2014	2b	To determine if morbidly obese patients using CPAP have fewer and less severe perioperative complications and a shorter hospital stay than patients with at least three SB criteria and are not using CPAP therapy.	Retrospective Cohort Study	3 groups of morbidly obese patients undergoing bariatric surgery. Group A: 99 patients who were diagnosed with PSG and used CPAP therapy before and after surgery. Group B: 182 patients who met at least three SB criteria and did not use CPAP. Group C: 412 patients who scored 1–2 items on the SB.	Postoperative hospital stays and pulmonary complications.	Group B patients had a significantly higher rate of pulmonary complications, worse oxygen saturation, respiratory rates, and increased length of stay in hospital. There were also two cases of sudden death in this group.
Seet et al.476	2015	2b	Predict independent risk factors for intraoperative and early postoperative adverse events.	Retrospective cohort study	5432 patients who underwent elective surgery were analyzed.	Adverse events: hypoxia, failed intubation or multiple attempts, dental injury, laryngospasm, bronchospasm, arrhythmia, hypertension and hypotension, unanticipated surgical bleeding, hypothermia, nerve injury, drug adverse reaction, equipment failure, unplanned ICU admission, post-anaesthesia care unit (PACU) time >2 h.	7.4% had unexpected intraoperative and early postoperative adverse events. These events were greater in patients with SB scores ≥ 3 Patients with SB scores ≥ 5 had a fivefold increased risk of unexpected adverse events.
Chung	2016	2a	Does the diagnosis of OSA changes the postoperative outcome?	Systematic Review	Patients undergoing surgery either under general or neuraxial anesthesia or sedation	Pulmonary, desaturation, difficult intubation, atrial fibrillation, cardiac complications, resource utilization, mortality.	The presence of OSA negatively influences perioperative outcomes.
Gokay et al.473	2016	4	SB vs. Berlin OSA questionnaires for evaluating potential respiratory complications.	Prospective Cohort Study	126 patients who underwent laparoscopic cholecystectomy.	Perioperative respiratory complications.	Both questionnaires found statistically significant differences between the low- and high-risk groups.
Chudeau et al.469	2016	2b	To evaluate whether the SB is predictive of perioperative respiratory complications in urgent surgery	Prospective cohort study	The SB questionnaire was used. 104 patients were HR-OSA and 85 LR-OSA.	Perioperative complications: respiratory complications, cardiac complications, neurologic complications, hospital length of stay and mortality.	HR-OSA vs. LR-OSA had: Higher respiratory complications (21% vs. 6%, a prolonged length of hospital stay (6 [3–12] vs. 4 [2–7] days. SB score was independently associated with respiratory complications (OR 1.44 [1.03–2.03].
Setaro et al.502	2018	4	To determine if longer monitoring of patients with OSA in the PACU improves patient outcomes after general anesthesia	Retrospec- tive cohort study	602 patients were evaluated. 68 patients (11%) had a confirmed and a presumptive diagnosis of OSA on chart review and screening STOP > 1.	Oxygen desaturation <95%, PACU length of stay	Most patients (96.5%) did not experience oxygen desaturation regardless of OSA diagnosis or a positive STOP score. Patients with OSA did not experience a higher incidence of respiratory symptoms while in the PACU.
Diagnosis of OSA and postoperative morbidity
Memtsoudis et al.474	2011	2b	To analyze perioperative demographics and pulmonary outcomes of patients with OSA after orthopedic and general surgical procedures	Case–control study	2,610,441 entries for orthopedic and 3,441,262 for general surgical procedures performed between 1998 and 2007. Of those, 2.52% and 1.40%, respectively, carried a diagnosis of OSA.	Aspiration pneumonia, adult respiratory distress syndrome (ARDS), pulmonary embolism (PE), and the need for intubation and mechanical ventilation.	OSA was associated with a significantly higher adjusted OR of developing pulmonary complications with the exception of PE.
Kaw et al.467	2012	1b	OSA is often undiagnosed before elective surgery and may predispose patients to perioperative complications.	Systematic review	13 studies (n = 3942). Studies without controls, involving upper airway surgery, and with OSA diagnosed by ICD-9 codes alone were excluded.	The incidence of postoperative desaturation, acute respiratory failure (ARF), postoperative cardiac events, and ICU transfers.	OSA was associated with significantly higher risk of postoperative cardiac events odds ratio (OR) 2.07; ARF OR 2.43; desaturation OR 2.27, and ICU transfer OR 2.81.
Opperer et al.468	2016	2a	The diagnosis of OSA has an impact on postoperative outcomes.	Systematic review	413,304 OSA and 8,556,279 control patients.	Combined complications of Cardiac, pulmonary, airway, mortality complications and resource utilization. Length of hospital stay and ICU admissions.	OSA patients had worse outcomes for pulmonary and combined complications, in-hospital mortality varied among studies.
Chan et al.465	2019	2b	To determine the association between OSA and 30-day risk of cardiovascular complications after major noncardiac surgery.	Prospective cohort study	1364 patients recruited without prior diagnosis of OSA and undergoing major noncardiac surgery. Monitored with nocturnal pulse oximetry and measurement of cardiac troponin concentrations.	Primary outcome was a composite of myocardial injury, cardiac death, heart failure, thromboembolism, atrial fibrillation, and stroke within 30 days of surgery.	Rates for composite complications: 30.1% for severe OSA, 22.1% for moderate OSA, 19.0% for mild OSA, and 14.2% for no OSA. OSA and risk for complications was significant only among patients with severe OSA (adjusted HR, 2.23 [95% CI, 1.49–3.34]; and not among those with moderate or mild OSA.
Diagnosis of OSA and postoperative mortality
Lockhart et al,481	2013	2b	To determine whether a prior diagnosis of OSA, or a positive screen for OSA is associated with increased risk for 30 days and one year mortality.	Prospective cohort study	14,962 patients, of whom 1939 (12.9%) reported a history of OSA. All patients completed preoperative OSA screening combination of the Berlin and Flemons STOP, and SB scores.	30 day postoperative mortality and 1-year mortality.	Screening tools identified a high prevalence of undiagnosed patients at risk for OSA (9.5%–41.6%). Neither a prior diagnosis of OSA nor a positive screen for OSA risk was associated with increased 30-day or one-year postoperative mortality.
D’Apuzzo et al.480	2012	2b	To determine if patients with OSA havea higher likelihood of postoperative in-hospital complications or increased costs after revision arthroplasty.	Retrospective cohort study	Nationwide sample of 258, 455 patients who underwent revision total hip or knee arthroplasty. Of these patients, 16,608 (6.4%) had been diagnosed with OSA.	In-hospital mortality, pulmonary embolism, and wound complications	OSA was associated with increased in-hospital mortality (odds ratio, 1.9; Pulmonary embolism (odds ratio, 2.1) Wound hematomas or seromas (odds ratio, 1.36) Increased postoperative charges ($61,044 vs. $58,813).
Mokhlesi et al,484	2013	2b	To determine if sleep disordered Breathing (SDB) is associated with higher hospital mortality, longer hospital stay, higher cost, respiratory and cardiac complications in bariatric surgeries	Retrospective Cohort study	Nationwide Inpatient Sample database of 91,028 adult patients undergoing bariatric surgeries	In-hospital death, total charges and length of stay, respiratory and cardiac complications	SDB was independently associated with decreased mortality (OR = 0.34), total charges (−$869), and length of stay. SDB associated with significantly increased emergent endotracheal intubation, noninvasive ventilation, and atrial fibrillation.
Lyons PG,503	2016	2b	To investigate the association between OSA and in-hospital mortality rapid response team (RRT) activation, ICU admission, intubation, and cardiac arrest on the wards in a large cohort of surgical and nonsurgical ward patients.	∖Retrospective cohort	93,676 ward admissions from 53,150 adult patients’ records were screened for the end points. OSA was identified in 5625 (10.6%) patients.	Primary outcome is in-hospital mortality. Secondary outcomes included length of stay (LOS), RRT activation, transfer to the ICU, endotracheal intubation, cardiac arrest on the wards, and a composite outcome of RRT activation, ICU transfer, and death.	OSA patients had more frequent RRT activations (1.5% vs. 1.1%) and ICU admission (8% vs. 7%) than controls but a lower inpatient mortality rate (1.1% vs. 1.4%). OSA was not associated with clinical deterioration and was associated with decreased in-hospital mortality.

Abbreviations: HR-OSA, high risk-OSA; ICU, intensive care unit; LR-OSA, low risk-OSA; SB, STOP-BANG.

VI.C |. Symptoms Associated with OSA

VI.C.1 |. Snoring, gasping, excessive daytime sleepiness

OSA is characterized by repetitive episodes of partial and complete pharyngeal collapse. Accordingly, in patients with OSA, increased UA resistance from partial pharyngeal collapse may manifest as loud snoring. Complete pharyngeal collapse with cessation of airflow may result in witnessed apneas. Individuals with OSA may report nocturnal episodes of choking or gasping, and/or symptoms of daytime fatigue, sleepiness, tiredness, or even insomnia. Bedpartners may be affected.

Large community studies have demonstrated habitual snoring (occurring most nights of the week) in 34%–44% of the population, witnessed breathing pauses in 6%–29%, and daytime hypersomnolence in 18%–28% of individuals.^88,145,456 Recent data from a population sample in the United Kingdom has shown a significant increase in the rate of reported witnessed apneas in the community over the last 20 years concurrently with a rise in obesity.⁵⁰⁴ The frequency of these symptoms is higher in those with OSA. Among individuals diagnosed with OSA, 38%–80% report snoring, 10%–67% report witnessed apneas, and 32%–40% report excessive daytime sleepiness (ESS ≥ 11). In general, increasing frequency of each symptom occurs with increasing AHI and OSA severity.^88,456,505

Findings from several population-based and retrospective studies have demonstrated that habitual snoring and witnessed apneas are predictors for OSA.^{88,145,505,506} However, the most predictive symptom for identifying individuals with OSA may be nocturnal choking or gasping episodes.⁵⁰⁷ A large Canadian population study in 2009 found that 4.3% of the population reported awakening suddenly with gasping or choking rarely or sometimes, and 1.5% reported this symptom occurring once a week or more⁵⁰⁸; of individuals diagnosed with OSA, 14.4% reported nocturnal choking or gasping episodes rarely or sometimes (OR 3.52, 95% CI 1.92–6.46), and 11.2% reported this symptom once a week or more (OR 7.92, 95% CI 3.74–16.74). Furthermore, an SR by Myers et al. in 2013⁵⁰⁷ demonstrated that nocturnal choking or gasping episodes had a likelihood ratio (LR) of 3.3 (95% CI 2.1–4.6) for the diagnosis of OSA (AHI ≥ 10), while snoring was found to be less predictive (LR 1.1, 95% CI 1.0–1.1).

Despite being common in the general population, snoring or apneas may not be routinely screened for or reported in the primary care setting.⁵⁰⁹ Many validated questionnaires^445,446,510 include some or all of these symptoms and may help identify patients at risk for OSA.⁵⁰⁹ Sex differences in reported sleep symptoms have been reported, with men being more likely to report snoring (34% vs. 19%) and women being more likely to report hypersomnolence (22.6% vs. 15.5%).⁹⁰ Shepertycky et al.⁵¹¹ also demonstrated that women with OSA were more likely to report insomnia (OR 4.20, 95% CI 1.54–14.26) and less likely to report witnessed apneas (OR 0.66, 95% CI 0.38–1.12). These variations in OSA symptomatology are important to recognize not only for appropriate screening, but also may have broader implications for OSA treatment.⁴²⁰ (Table VI.C.1)

TABLE VI.C.1.

Evidence on the relationship between snoring, gasping, daytime sleepiness, and OSA

Study	Year	LOE (1a–5)	Study design	Study groups	Clinical end-point	Conclusion
Myers et al.507	2013	1a	Systematic review	Community-screened patients and patients referred for sleep evaluation	In-laboratory PSG	Nocturnal choking or gasping episodes most predictive symptom (LR 3.3, 95% CI 2.1–4.6) for the diagnosis of OSA (AHI ≥ 10). Snoring, reported apneas, and excessive daytime sleepiness were less predictive of the diagnosis (LR 1.1 [95% CI 1.0–1.1], 1.4 [1.2–1.5], and 1.3 [1.1–1.4], respectively).
Young et al.90	1993	1b	Prospective cohort	Wisconsin Sleep Cohort Study population (n = 602)	In-laboratory PSG	In individuals with OSA (AHI ≥ 5), more men than women reported habitual snoring ± breathing pauses (34% vs. 19%), and more women than men reported hypersomnolence (22.6% vs. 15.5%).
Johns et al.510	1993	1b	Cross sectional survey	Adult patients suspected of having OSAS by history of snoring (n = 273)	Questionnaire (ESS) and in-laboratory PSG	Patients found to have OSAS (RDI > 5) had significantly higher levels of daytime sleepiness (measured by ESS) than primary snorers; ESS scores and thus daytime sleepiness increased significantly with increasing OSAS severity.
Netzer et al.445	1999	1b	Cross sectional survey	Adult primary care patients (n = 100)	Questionnaire (Berlin), in-home PSG	High-risk patients (in ≥2 categories) were more likely to meet criteria for OSAS (RDI > 5). High-risk qualification by one symptom category did not predict RDI as well as composite score.
Young et al.88	2002	1b	Prospective cohort	Sleep heart health study population (n = 5615)	Questionnaire, clinical examination, in-home PSG	Habitual snoring (OR 2.87, 95% CI 2.10–3.91), and breathing pauses occurring <3 nights/week (OR 1.78, 95% CI 1.34–2.37), or 3–7 nights/week (OR 4.03, 95% CI 2.87–5.67) were associated with moderate to severe OSA (AHI ≥ 15).
Chung et al.446	2008	1b	Diagnostic accuracy	Adult pre-operative patients (n = 177)	Questionnaire (STOP), in-laboratory PSG	Patients identified as high risk for OSA (≥2 “yes” answers) demonstrated a sensitivity of 65.6%, specificity of 60%, positive predictive value (PPV) 78.4%, and NPV 44.0% for OSA (AHI > 5). The STOP questionnaire was more sensitive in detecting patients with moderate (AHI > 15) to severe (AHI > 30) OSA.
Silva et al.456	2011	1b	Prospective cohort	Sleep heart health study population (n = 4770)	In-home PSG	Among individuals with moderate-to-severe SBD (RDI ≥ 15, <30) and severe SBD (RDI ≥ 30): 55.9% and 69.6%, respectively snore often/every night; 38.9% and 66.9% have witnessed apneas, and 32% and 40% have excessive daytime sleepiness (ESS≥11).
Johnson et al.506	2020	1b	Prospective cohort	Jackson Heart Sleep Study (n = 719)	Questionnaire (Including, STOP-bang, ESS), in-home PSG	Among an adult African American cohort with moderate to severe OSA (REI ≥ 15), 49.3% snore ≥3 nights/week (vs. 31.1% with mild or no OSA), 6.6% had witnessed apneas ≥3 nights/week (vs. 3.1%), and 22.4% had sleepiness (ESS > 10) (vs. 19.5%).
Durán et al.145	2001	2b	Cross-sectional	Community sample (n = 555)	Structured interview, in-laboratory PSG	Habitual snoring (OR 5.45) and breathing pauses during sleep (OR 13.4) are risk factors for moderate-severe OSA (AHI ≥ 15). Habitual snoring (OR 3.36) is also risk factor for mild OSA (AHI ≥ 5–14.9). Breathing pauses not significantly associated with mild OSA (OR 4.63). Daytime hypersomnolence is not associated with mild OSA (OR 1.37) or moderate-severe OSA (OR 1.05).
Shepertycky et al.511	2005	2b	Retrospective cohort	Population of patients diagnosed with OSAS (n = 260)	Questionnaire, in-lab PSG	At the time of OSAS diagnosis, women were more likely to present with insomnia (OR 4.20, 95% CI 1.54–14.26) and less likely to report witnessed apneas (OR 0.66, 95% CI 0.38–1.12).
Ustun et al.505	2016	2b	Retrospective cohort	Clinical Sleep Laboratory Database (n = 1992)	In-lab PSG	Patients found to have OSA (AHI > 5 or RDI > 10) were significantly more likely to self-report snoring (79.7%) and witnessed apneas (43.8%) than individuals without OSA. A clinical prediction model for diagnosing OSA using medical history features was superior to model performance using symptoms alone.
Li et al.512	2019	2b	Cross-sectional observa-tional study	Adult patients with snoring and suspected OSAS (n = 134)	Questionnaire (ESS and Montreal Cognitive Assessment), in-lab PSG	Daytime sleepiness, as measured by ESS, worsens with increasing OSAS severity.
Epstein et al.513	2009	5	Guideline		Recommendations on the evaluation and management of adult patients with OSA	A comprehensive sleep history in a patient with suspected OSA should include evaluation for snoring, witnessed apneas, gasping/choking episodes, excessive sleepiness not explained by other factors (including assessment of sleepiness severity by ESS), total sleep amount, nocturia, morning headaches, sleep fragmentation/sleep maintenance insomnia, and decreased concentration and memory.

VI.C.2 |. Nocturia

Nocturia, the need to urinate more than once per night,⁵¹⁴ affects greater than 60% of adults aged 70 and older and is more prevalent in women.^515,516 The prevalence of high risk for OSA (based on results from the Berlin questionnaire) studied in a sample of female patients, in an urogynecology clinic was greater in patients with nocturia (61.7%) compared to those without (24.1%) (OR 2.9, 95% CI 1.29–6.52).⁵¹⁷ A recent MA demonstrated an association between OSA and risk of nocturia 1.41 (95% CI 1.26–1.59). Furthermore the authors’ report that patients with severe OSA had a higher incidence of nocturnal urination in comparison to patients with mild or moderate disease (relative risk [RR] 0.82; 95% CI 0.72–0.94). In men there was a statistically significant association between OSA and risk of nocturia (RR 1.487, 95% CI 1.087–2.034), which was not the case in women.⁵¹⁸ To support the link between severity of OSA and nocturia, in a study comparing patients with and without nocturia, an increased ODI was associated with an increased probability of nocturia, in the multivariable model (OR = 1.03; 95% CI = 1.01–1.06).⁵¹⁹ The higher the ODI score, the greater the probability of nocturia.

A chain of physiologic effects leads to diuresis and nocturia in the setting of OSA. Increased intra-thoracic negative pressures (needed to overcome a closed UA) lead to atrial stretch which triggers a signal of excessive fluid volume and the subsequent release of atrial natriuretic peptide induces diuresis.^520,521

Nocturia has significant effects on QOL and in particular sleep quality, impacting total sleep time (TST) and sleep efficacy. In comparison to patients with OSA without nocturia, mean SaO₂, time spent in desaturation below 90%, and ODI were worse in OSA patients with nocturia.⁵¹⁶ Treatment with CPAP has been shown to improve nocturia in patients with OSA.⁵²²

VI.C.3 |. Caffeine intake

Excessive daytime sleepiness is a common symptom of OSA. One of the reasons to consume caffeinated products is to counteract sleepiness. Caffeine is a natural psychoactive substance which blocks adenosine receptors in the brain 30–60 min after consumption. As adenosine builds up during the day, the sleepier we become. Caffeine blocks this process, and as a consequence we remain alert and vigilant.

Bearing this is mind, assessing caffeine use as a marker of sleepiness is recommended when taking patient history in patients suspected of OSA. Studies on patients with untreated OSA have demonstrated that increased caffeine intake is associated with less cognitive impairment⁵²³ thereby indicating a potential therapeutic effect of caffeine and evidence for self-treatment of OSA. On the other hand, caffeine can delay onset of sleep and can interfere with our circadian rhythm.

Despite its common use, the prevalence of caffeine use in patients at risk for OSA is difficult to estimate. The SHHS provides the most comprehensive assessment of caffeine use in OSA patients.⁵²⁴ Women with SDB have the strongest correlations between caffeinated soda and OSA. There was only an association between caffeinated soda and severe SDB in men.⁵²⁴ Coffee and tea consumption were not associated with SDB.

Several trials were created to evaluate interactions among caffeine use, OSA, and CV risks. Bardwell et al. demonstrated that OSA patients consumed significantly greater caffeine than normal controls (295 vs. 103 mg/day of caffeine).⁵²⁵ Robinson et al. measured caffeine consumption before and after 4 weeks of CPAP. Although sleepiness improved after CPAP, caffeine intake did not change.⁵²⁶ Such research has contributed to the concept that OSA has independent effects on hypertension and other CV markers which are separate from the impact of caffeine use.

VI.D |. Physical Exam Findings Related to OSA

Evaluation of patients with OSA includes standard physical exam that evaluates for age, BMI, neck circumference, assessment of craniofacial structures, and standard UA examination of the nasal cavity, oral cavity, and OP. The ability of an individual physical exam finding to predict OSA alone is limited, but the entire physical exam can be considered in the context of the patient’s history to provide a better assessment of OSA risk.

Some goals of the physical exam for OSA aim to: (1) incorporate exam findings for use in OSA screening tools in conjunction with patient symptoms; (2) determine potential causes of airway obstruction, (3) identify potential anatomic concerns that may limit PAP therapy, and (4) aid in determining potential therapeutic targets for sleep surgery (see Section IX.C.1).

This section evaluates the components of the physical exam relevant to OSA evaluation and summarizes the available evidence that associates particular exam findings with OSA. Evidence tables are provided for specific exam findings that have adequate data.

Not all components of the physical exam are routinely performed, and the examination should be tailored to each individual patient after completing a thorough history and symptom review.

VI.D.1 |. BMI

Obesity is a well-established risk factor for the development and progression of OSA. A BMI greater than 30 increases risk for OSA³⁵⁰ and weight reduction is a treatment option that can improve OSA severity.¹⁸⁹ BMI level also has implications for surgical outcomes with higher BMI associated with persistent OSA after surgery.⁵²⁷ BMI level should be included in the evaluation of OSA patients.

VI.D.2 |. Neck circumference

Neck circumference (NC) or collar size is associated with OSA in adults.^{184,528–534} It is measured as the circumference of the neck at the superior border of the cricothyroid membrane. Larger neck circumference is related to elevated BMI. Fat deposition in the tissues of the neck results in a smaller and more collapsible UA, thus increasing the likelihood of OSA.

An SR and MA of facial phenotype in adult OSA found that adults with OSA had significantly larger NC compared to controls.⁵³⁵ Neck circumference is associated with OSA severity.^138,528 Specifically, NC > 40 cm (16 inch) is associated with snoring and OSA.^{528,529,536–538} Although it is positively correlated with OSA, NC alone is a poor predictor of OSA.⁵³⁹ The sensitivity of NC alone is insufficient to identify patients with sleep apnea, but increases when combined with other clinical, anthropometric, or cephalometric measures.^{537,540–542} One screening tool that incorporates NC is the STOP-BANG which is used to identify patients at high risk for OSA⁴⁴⁶ and uses NC or shirt collar size estimates with thresholds of 43 cm (17 inch) for males and 41 cm (16 inch) for females. (Table VI.D)

TABLE VI.D.

Neck circumference (NC) measurement in the diagnosis of OSA

Study	Year		Study design	Study groups	Clinical endpoint	Conclusion
Agha et al.535	2017	3a	Systematic review and meta-analysis of predominantly case–control studies	Not applicable	Not applicable.	Adults with OSA had an increased weighted mean difference in NC compared with controls, in five case–control studies. The pooled mean change was 1.26 mm (0.64–1.88), with large heterogeneity found between studies.
Kim et al.138	2015	3b	Cross-sectional retrospective study	Cohort of snoring Asian patients	Neck circumference in Asian patients with OSA compared to individuals with simple snoring.	NC predicted OSA presence and severity.
Kushida et al.537	1997	4	Prospective cohort study	Consecutive patients referred for evaluation of sleep disorders	Evaluated measurements of the oral cavity with body mass index and NC, in predicting OSA.	A NC ≥ 40 cm is associated with OSA sensitivity of 61% and specificity 93%.
Stradling et al.533	1990	4	Prospective study	Men 35–65 years old GP registry	Registry study.	NC was an independent predictor of nighttime hypoxia events.
Adult Obstructive Sleep Apnea Task Force of the American Academy of Sleep Medicine513	2009	5	Guideline	Not applicable	Not applicable.	Features that may suggest the presence of OSA include increased NC (> 17 inch in men and > 16 inch in women).
Ahbab et al.528	2013	4	Cohort study	OSA adults	Evaluate NC and metabolic syndrome parameters in OSA patients.	NC in severe OSA was significantly higher than in mild–moderate OSA. NC was an independent risk factor for severe OSA (odds ratio [OR]: 1.55. 95% confidence intervals [CI]: 1.09–2.21, p = 0.01).
Chang et al.529	2013	4	Prospective cohort	Cohort of Chinese patients with snoring complaints in sitting position	Evaluate if snoring in a sitting position is a predictor of OSA in patients.	NC ≥40 cm as a predictor for OSA had sensitivity 48.6%, specificity 87.8%, and likelihood ratio 3.98.
Park et al.532	2014	4	Retrospective cohort	Non-obese Asian patients who underwent PSG	Evaluate association between body weight and obstructive sleep apnea even in patients who are not obese.	AHI found to be positively correlated with the NC. NC is independently associated with OSA.
Davies et al.530	1990	4	Prospective study	Cohort of patients referred for investigation of sleep disorders	Importance of NC, obesity, and radiographic pharyngeal dimensions for OSA.	NC of 42.5 cm (16.75 inch) for a 1.78 m (5 feet 10 inch) man) is 77% sensitive and 82% specific for significant OSA with positive predictive value (PPV) of 75%.
Yildirim et al.534	2015	4	Prospective	Cohort of patients with a BMI ≥ 30 and OSA symptoms	Evaluate anthropometric measurements with OSA.	Significant positive correlation between AHI and neck circumference.
Mirrakhimov et al.531	2013	3a	Systematic review	Not applicable	Not applicable.	Neck circumference is associated with OSA.
Santos et al.542	2019	4	Cohort study	Cohort civil servants 35–74 years	Compare BMI, several surrogate markers of body fat (in isolation or combined) and validated questionnaires for screening OSA.	Age and gender-adjusted NC (AUC = 0.733 [0.711–0.755]) had moderate accuracy as predictor for OSA.
Yusoff et al.538	2010	4	Prospective	Cohort randomly selected express bus drivers	Identifying factors or conditions related with OSA.	NC (OR = 1.31, 95% CI 1.18–1.46) was significantly associated with OSA status.

VI.D.3 |. Nasal exam

Nasal obstruction is associated with sleep apnea and treatment of obstruction improves sleep quality and CPAP tolerance²⁹⁶ (see Section V.A.1). A thorough nasal examination includes evaluation of external nasal structures and visualization of the nasal cavity using a nasal speculum and/or nasal endoscope. The examiner should identify any external nasal deformity, internal septal deviation, inferior turbinate hypertrophy, nasal valve collapse, nasal cavity inflammation, and adenoid hypertrophy. The Cottle and Modified Cottle Maneuvers may be beneficial in assessing for internal nasal valve collapse.⁵⁴³ The nasal exam is focused and based on patient symptoms of nasal obstruction. Nasal obstruction has implications in PAP tolerance⁵⁴⁴ and OSA-related symptoms.⁵⁴⁵

VI.D.4 |. Oral cavity and oropharynx exam

Examination of the oral cavity and OP is an important component of the OSA physical exam. Thickened and collapsible lateral pharyngeal walls and an enlarged uvula are well-recognized physical exam finding in adults with OSA.^324,325 Several studies also demonstrate that a narrow palatopharyngeal arch, or fauces, is predictive of the presence and severity of OSA.^{235,332,334,336}

Tonsil size is commonly reported using the Brodsky tonsil grading scale: 0 = surgically absent; 1+= 0%–25% OP obstruction; 2+= 25%–50% OP obstruction; 3+= 50%–75% OP obstruction; and 4+ = 75%–100% OP obstruction.⁵⁴⁶ Tonsil size may predict severity of OSA, and a study of adults with OSA demonstrated that each increase in tonsillar size correlates to an increase in AHI of approximately 14 events/h.³²⁷ Identification of enlarged tonsils in patients with OSA also has implications in treatment decisions.³³⁵

VI.D.5 |. Mallampati classification and Friedman tongue position

Common classification schema used for grading the oropharyngeal exam include the Mallampati classification (MC) and Friedman tongue position (FTP).

The MC was developed as a clinical indicator for difficult intubation and is performed by examining the seated patient with the mouth open and tongue protruded. Based on the visualized oropharyngeal structures (palate edge, uvula, and tonsils), a classification from I to IV is assigned. Class I represents complete visualization of the soft palate and uvula, class II includes partial visualization of the uvula and complete view of the soft palate, class III involves view of the base of the uvula, and class IV views of the hard palate only.

The MC has since been modified several times most notably as the FTP, in which the airway is assessed with the tongue remaining inside the oral cavity instead of protruded with the original MC.^547,548 FTP was designed to grade the relative size of the tongue base and was used as a predictor for surgical success after UPPP. The modified Mallampati (MMP) also examines the oral cavity with the tongue at rest and is a variation of FTP.

Both the MC and FTP have been examined as predictors for the presence and severity of OSA, but conflicting evidence lends to uncertainty of their predictive values.^547,549 MC is associated with AHI but not among women.⁵⁵⁰ Mallampati class IV is associated significantly with OSA.⁵⁵¹ FTP and AHI are positively correlated with higher FTP grade associated with higher AHI^547,552,553; however, low inter-examiner agreement in FTP scoring has been suggested.⁵⁵⁴

A recent review article concludes that MC and FTP may have limited predictive values for OSA when used independently, but they may play a role when incorporated into the overall patient assessment.⁵⁴⁸ (Table VI.D.5)

TABLE VI.D.5.

Evidence for the association between Mallampati classification, Friedman tongue position, and OSA

Study	Year	LOE	Study design	Study groups	Clinical endpoint	Conclusion
Friedman et al.547	2013	3a	Systematic review	Adults with OSA	Systematic review and meta-analysis to assess the association between Mallampati and Friedman tongue position for OSA severity	Friedman tongue position and Mallampati classification were significantly associated with obstructive sleep apnea severity
Bins et al.549	2011	3a	Systematic review	Adults with OSA	Systematic review to evaluate the diagnostic value of the Mallampati score in patients suspected of OSAS	No evidence for diagnostic value of Mallampati score in patients suspected of having obstructive sleep apnea syndrome
Ruangsri et al.551	2016	3b	Case–control study	Group 1: OSA (n = 78); Group 2: no-OSA (n = 78)	Multivariate logistic regression analysis was used to identify the factors associated with OSA	Mallampati class IV was found to be significantly associated with OSA (adjusted odds ratio 5.040 (1.655, 15.358))
Amra et al.550	2019	3b	Cross sectional study	Adults with OSA	Evaluated anthropometric data and the Mallampati classification for predicting the severity of OSA	Mallampati classification was found to be associated with AHI indices, but not among women patients
Friedman et al.547	2013	3a	Systematic review	Adults with OSA	Systematic review and meta-analysis to assess the association between Mallampati and Friedman tongue position for OSA severity	Friedman tongue position and Mallampati classification were significantly associated with obstructive sleep apnea severity
Banhiran et al.555	2014	3b	Cross sectional study	Group1 no OSA (n = 66); Group 2: OSA (AHI > 15/h) (n = 217)	Identify physical findings that may predict the presence of moderate to severe OSA in snoring patients	No significant difference was found in FTP between moderate to severe OSA and control patients (AHI < 5/h)
Lin et al.552	2019	4	Retrospective cohort	Adult habitual snorers and adults with OSA	Identify standard clinical parameters that may predict AHI and OSA severity	FTP was found to bea reliable predictor of OSA (r = 0.504, 95% CI: 0.417–0.580)
Subramanian et al.553	2011	4	Prospective cohort	Adult patients being screened for OSA	Develop a novel screening tool for the detection of OSA, combining self-reported historical factors with physical exam findings	FTP was found to bea reliable predictor of OSA (r = 0.14, p = 0.0019)

VI.D.6 |. Laryngoscopy

Laryngoscopy may be included in the OSA physical exam if indicated based on patient history. Laryngoscopy can be used to identify lingual tonsillar hypertrophy, to diagnose masses or lesions within the UA, and to determine the shape, size and position of the epiglottis. While lingual tonsil grade as defined by Friedman et al. does not have a direct correlation with OSA, its presence may play a role in surgical planning.^556,557 A recent study demonstrated that presence of a mega-epiglottis on laryngoscopy is an independent predictor of moderate and severe OSA. This study also demonstrated that a modified Cormack-Lehane score of 2 or more as measured on laryngoscopy may predict OSA.⁵⁵⁸

Physical exam summary

Specific physical exam findings may correlate to OSA risk and severity of disease however physical exam findings alone are insufficient for OSA diagnosis and severity estimations. Sleep study testing is required to achieve a diagnosis of OSA. The physical exam has a role in the evaluation for OSA specifically to: (1) incorporate exam findings into OSA screening tools for risk stratification; (2) identify potential anatomic concerns that may limit PAP therapy, and (3) aid in determining potential therapeutic targets for sleep surgery.

VI.E |. Imaging Findings Associated with OSA

VI.E.1 |. Lateral cephalometry

X-ray cephalometry is a widely available and inexpensive imaging modality that assesses the craniofacial skeleton and its association with soft tissue of the UA. It was introduced in the 1980s as part of routine sleep surgical evaluation where cephalometric measurement of the length of the soft palate and the position of the hyoid bone were associated with severe OSA.⁵⁵⁹ In an SR and MA, the following parameters significantly correlated with OSA: (1) a mandibular body length as measured from gonion to gnathion of less than 80 mm, (2) a sella, nasion, A point angle (SNA) of less than 75°, (3) an increased anterior lower facial height measured from the anterior nasal spine to the gnathion greater than 85 mm, and (4) hyoid bone more than 18 mm below the mandibular plane correlate strongly with OSA.²²⁹ While this study provides the highest LOE, caution must be exercised in generalizing its conclusions. There are methodological differences in standardization of image acquisition, magnification of cephalograms, matching of controls, article selection bias, and quality of the constituent studies. Studies generally did not account for ethnic differences in craniofacial structures. Additional limitations to cephalograms include lack of 3D assessment and low soft tissue detail. And despite the anatomic correlates, cephalograms can neither confirm nor exclude OSA. An MA linking adult OSA with cephalometric parameters failed to show a statistically significant difference when subjects with the disorder are compared to healthy controls.⁵⁶⁰ Moreover, because they are done awake and in the upright position, they do not account for changes in the airway during sleep.

Correlation of cephalometric parameters with the AHI on overnight PSG has been inconsistent in the literature. Analysis of preoperative clinical screening tests for OSA revealed that cephalometry alone may miss a significant proportion of patients with OSA in low-risk populations.⁵⁶¹ Therefore, it cannot be relied upon for OSA screening and should be considered as only an investigative modality for OSA evaluation. The main benefits of cephalogram are its low cost and low-risk nature in providing anatomic assessment for OSA. It is most useful in treatment planning for orthognathic surgery.

VI.E.2 |. Cone-beam CT

Cone beam CT (CBCT) has the benefits of 3D volumetric assessment, which is more expensive than cephalogram, but increasingly available. High quality evidence supporting its use for OSA airway assessment is lacking. An SR of observational studies by Guijarro-Martínez and Swennen noted inconsistencies and discrepancies in the technique of imaging acquisition.⁵⁶² Most studies did not control respiratory phase, mandibular position, and/or tongue position, which influence airway dimensions. Also, the most widely available CBCT scanners acquire images in upright position, which impacts airway analysis for OSA patients. Conclusions from upright CBCT airway dimensions may not apply to supine cross-sectional imaging as the gravitational effect in response to the postural change leads to a smaller cross-sectional area of UA due to movement of the hyoid bone, the mandible, the tongue, and UA muscle. Hsu and Wu showed that in the upright position, there is a significantly greater distance between the mandibular plane and the hyoid bone when compared to CBCT derived cephalometric images in the supine position among normal test subjects.⁵⁶³ Chen et al. performed an SR of the most relevant UA anatomical parameters related to OSA by CBCT, conventional CT, and MRI.⁵⁶⁴ On sagittal cross-sections, a soft palate measuring more than 38 mm in length and 10 mm in width, an oropharyngeal length measuring more than 70 mm, and tongue length more than 65 mm are correlated with OSA. That corresponds to a retropalatal cross-sectional area of less than 100 mm² and a retroglossal cross-sectional area of less than 150 mm² on axial views. Similar to cephalogram, CBCT can be of value as an adjunct in the anatomic evaluation of OSA and for surgical planning in orthognathic surgery.

VI.E.3 |. Other imaging types

Cine CT (ultra-fast CT) provides dynamic airway examination, an advantage over CBCT and conventional CT. It does not reliably distinguish between patients with OSA and primary snorers. Somnofluroscopy can distinguish snoring from apneas, but high radiation exposure and poor anatomical detail limit its utility. Sleep MRI provides superior soft tissue anatomical detail. Using sleep MRI, Liu et al. showed in a multivariate analysis for subjects matched for age and BMI that severity of OSA can be predicted by the presence of lateral pharyngeal wall collapse and low hyoid bone position.⁵⁶⁵ Limitations for the use of sleep MRI include lack of widespread availability, expense, and challenges involved in promoting sleep in an MRI scanner discomfort. Awake ultrasonography is a promising imaging modality in evaluating OSA. A recent MA by Singh et al. showed a number of parameters with moderate to good correlation with OSA.⁵⁶⁶ However, these parameters remain to be validated.

In the literature, there are only a few publications dealing with the evaluation of soft tissue and skeletal anatomy using MRI and lateral cephalometry with both control and OSA subjects. These studies utilized dynamic or ultrafast MRI sequences and cephalometric measures, but did not discriminate OSA from control subjects by sleep study or validated questionnaires.

The soft tissue landmarks in cephalometry are influenced by the superposition of all the structures present in the same plane, which makes some of the landmarks difficult to accurately and reliably identify. MRI provides unparalleled definition of UA soft tissue structures, without exposing patients to the ionizing radiation. However, MRI offers a less precise bony contour definition. Because the enhanced soft tissue resolution of MRI affords greater measurement accuracy and allows for the determination of additional airway measurements, which cannot be achieved through cephalometry (e.g., tongue volume, PAS area), MRI may be superior to radiocephalometry for the assessment of anatomic measures in OSA patients.⁵⁶⁷ (Table VI.E)

TABLE VI.E.

Evidence on imaging and OSA diagnosis

Study	Year	LOE	Study design	Study groups	Clinical endpoint	Conclusion
Riley et al.559	1983	2b	Retrospective study	15 adult OSA patients, 10 controls	Cephalogram	Length of the soft palate and the position of the hyoid bone serve as important cephalometric landmarks in OSA patients.
Neelapu et al.229	2017	2a	Systematic review, meta- analysis	25 Studies	Cephalogram	Significant variables in OSA patients include increase in total anterior facial height, inferior hyoid bone position, and reduced pharyngeal airway space.
Armalaite et al.560	2015	2a	Systematic review, meta- analysis	11 Studies	Cephalogram	The most reliable parameters in OSA subjects are MP-H and upper posterior pharyngeal space. However, the diagnosis of OSA cannot be made using cephalograms as the sole investigative modality.
Ramachandran et al.561	2009	2a	Systematic review, meta- analysis	26 Studies	Multiple screening questionnaires, multiple history and physical examination elements, oximetry, cephalometry, and morphometry	No ideal preoperative screening tool exists for OSA in the surgical population. Preoperative screening tools including cephalometry may not identify a significant proportion of patients with OSA.
Guijarro- Martinez and Swennen562	2011	2a	Systematic review	5 studies (relationship between upper airway and OSA)	CBCT	Statistically significant difference between OSA and non-OSA patients in the ratio of the airway cross-section area/square area.
Hsu and Wu563	2019	2b	Retrospective study	21 normal subjects (no OSA)	Cephalogram, CBCT	Cephalograms in the upright position had a significantly larger MP-H when compared to CBCT derived cephalometric images in the supine position.
Chen et al.564	2016	2a	Systematic review	8 studies	CBCT	The most significant anatomical characteristic related to the pathogenesis of OSA is the small cross-sectional area of the airway (CSAmin). It is unclear how soft tissue structures interact to control upper airway dimensions.
Liu et al.565	2015	2b	Case–control study	15 mild adult OSA patients, 15 adult severe OSA patients	Sleep MRI	Based on sleep MRI, lateral pharyngeal wall collapse and low hyoid position are significantly associated with severe OSA.
Singh et al.566	2019	2a	Systematic review, meta-analysis	21 Studies (3339 patients)	Ultrasound	Airway parameters correlated with OSA: neck circumference and retropalatal diameter shortening during MM, tongue base thickness during MM, resting tongue base thickness, tongue base width, and LPW thickening. Non-airway: carotid plaque presence and carotid intimal thickening.

VI.F |. Diagnostic Testing for OSA

VI.F.1 |. Types of sleep studies

Traditionally, the standard way to diagnose OSA has been with PSG which is an overnight test conducted when one is asleep in a sleep lab with a sleep technologist in attendance. PSG uses ≥7 channels to monitor sleep, respiration, and muscle activity. Monitoring of brain activity started in 1875 and evolved to the Sleep Research Society developing a standardization for the scoring of sleep in 1968 using the Rechtschaffen and Kales (R and K) manual.⁵⁶⁸ R and K was the standard to score PSG until 2007 when the AASM published a scoring manual⁵⁶⁹ for scoring sleep studies that is followed by all accredited sleep centers and updated on an approximately annual basis. It describes types of sensors to be used, their placement, and how to score sleep, respiratory events, cardiac events, and limb movements. A standard PSG includes electroencephalography (EEG) with frontal, central, and occipital leads placed using the 10–20 international system, electro-oculography (EOG), chin electromyogram (EMG), lead II electrocardiogram (ECG), airflow monitoring via oral thermistor and a nasal pressure transducer, respiratory effort with inductance plethysmography, SaO₂ via a pulse oximetry, and limb EMG.⁵⁶⁸ Other monitoring that is typically done is snoring via a snore microphone, body position sensor, and video monitoring. A PSG report will contain information about the patient’s sleep, limb movements, and the presence of SDB based on the AHI calculated during the PSG.

In general there are four types of sleep monitoring devices (I–IV) (Table VI.F.1). Type I is a PSG described above and types II–IV are portable sleep monitoring (PM) or HSAT.⁵⁶⁸ A type II study is portable monitoring done using the same channels as a type I device only it is unattended – meaning there is not a sleep technologist or sleep trained personnel with the patient while they are having the study done. Type III HSAT uses four to seven channels and type IV uses one to two channels with at least one being a pulse oximeter.^568,570 Most HSATs do not directly record sleep. Therefore, as opposed to an AHI, HSATs report an REI which is the number of apneas and hypopneas per hours of recording time.⁵⁷¹ The AASM recommends that HSATs record airflow, respiratory effort, and blood oxygenation.^570,572 Most LRs and recommendations focus on type III devices. Type II devices are not typically conducted in the clinical setting and not much data exists on their efficacy. Of note, the naming of sleep monitoring tests (type I–IV) is non-descriptive and reflects a descending order of complexity/required channels. A separate mode of classifying sleep tests, the SCOPER criteria, have been proposed, but are not yet formally adopted. The Scoper criteria classify sleep tests by the parameters being assessed within a given individual test and is a reasonable alternative to the type I–IV classification.⁵⁷³

TABLE VI.F.1.

Overview of the types of sleep testing, strengths, and associated limitations

	Type I	Type II	Type III	Type IV
Specifications	Full-attended polysomnography (≥7 channels) EEG, EOG, EMG, ECG, airflow, effort, oximetry	Full-unattended polysomnography (≥7 channels) EEG, EOG, EMG, ECG, airflow, effort, oximetry	Limited channel devices (four to seven channels) ECG, airflow, effort, oximetry	One or two channels with one typically being oximetry Oximetry and other
Common examples	Routine polysomnography	NA	Alice PDX Apnealink PAT	Pulse oximetry
Strengths	Gold-standard sleep staging and event scoring	Done at home Diagnostic accuracy to rule in/out OSA	Done at home Sufficient to rule in OSA	Done at home Simple set up
Limitations	Requires subject to stay overnight in the sleep lab and has greater associated costs	Requires technical expertise to set up and administer appropriately Not appropriate for complex patients	High level of pre-test probability required in order to accurately diagnose OSA	Very limited data collected Additional testing often warranted

A type III device that is FDA approved uses peripheral arterial tone (PAT) in combination with actigraphy, and pulse oximetry to diagnose OSA as opposed to the traditional combination of airflow, respiratory effort, and pulse oximetry used in other type III devices.⁵⁷⁴ The algorithm correlates low SaO₂ with sympathetic tone (determined by the PAT) to determine if low oxygen is due to an obstructive event (event with decreased SaO₂ associated with high sympathetic tone) or central event (events with decreased SaO₂ associated with low sympathetic tone).^574–576 It also uses an actigraph to identify which potential events occurred in wakefulness or sleep. The device is worn around the wrist and has two finger probes. The actigraph is in the body of the device and detects movement of the arm to approximate sleep and wake. Obstructive events cause increased sympathetic activity which will lead to vasoconstriction of digital blood vessels resulting in attenuation of the PAT signal.^574,577 The device will report estimates of the AHI, sleep time, lowest SaO₂, REM percentage, and oxygen percentage.⁵⁷⁴ Studies comparing PAT technology to PSG found a strong correlation of AHI and lowest SaO₂ between the two study devices as well as reliable reproducibility with the PAT device results.^574–576

VI.F.2 |. Home sleep studies

An HSAT is recommended for the evaluation of patients with high clinical probability of OSA along with a comprehensive sleep evaluation.⁵⁷⁸ SRs and meta-analyses of well-designed RCTs demonstrate that treatment outcomes and adherence are not statistically different between subjects diagnosed with HSAT or PSG.^573,579 Current guidelines recommend using PSG or home sleep apnea testing with a technically adequate device for OSA diagnosis.⁵⁷⁸ These devices include type I–IV monitors (see section Types of Sleep Studies). There is increasing evidence supporting the use of some home-based type III and IV type sleep studies to “rule-in” but not “rule-out”⁵⁸⁰ moderate to severe OSA. Such devices may therefore prove useful in populations where there is high prevalence of OSA or when combined with validated sleep questionnaire(s) that enhance the pre-test probability of moderate to severe OSA.⁵⁸¹ An 18% false negative rate for HSAT in high-risk patients has been observed⁵⁷²; therefore, if a single HSAT is negative, inconclusive, or technically inadequate, an in-laboratory attended ^573,579 PSG should be performed. In addition, an MA of laboratory versus HSAT concluded that they both provide similar diagnostic information, but HSAT may underestimate severity of AHI by around 10%.⁵⁸²

Meta-analyses have been conducted comparing type III portable studies and type I (PSG) studies in uncomplicated patients at risk for OSA.⁵⁸³ The type III portable studies measured airflow, thoracoabdominal effort, SaO₂, and body position. The results showed that type III devices are both sensitive and specific for the diagnosis of OSA. In addition, as disease severity increases, there is increased specificity and decreased sensitivity. The technical failure rate is also higher for portable monitoring done at home as opposed to in the lab.⁵⁸³ Therefore, evidence supports that type III studies are useful to diagnose OSA in those with a high pre-test probability of having moderate to severe OSA.^{138,573,579,582} MA comparing one channel, two channel, and four channel HSATs showed that the sensitivity is increased for devices using four channels compared to one or two channels.^584,585 Also sensitivity decreases and specificity increases when the AHI cut off was moved from 5 to 15 events/h. It has been concluded that using type IV (one to two channels) devices may lead to a higher number of false positives and negatives compared to type III devices but may have a role in screening for OSA in areas where access to type I and III studies are limited.

Subjects with significant cardiorespiratory disease, potential respiratory muscle weakness due to neuromuscular conditions, awake hypoventilation or suspicion of sleep-related hypoventilation, chronic opioid medication use, history of stroke, or severe insomnia were excluded from most studies, and therefore a PSG is recommended over HSAT in these cases.⁵⁷⁹ However, there is emerging data from small RCTs that HSAT can be feasibly implemented after stroke or transient ischemic attack (TIA) or even patients in a stroke rehabilitation unit.^586,587 Therefore, this population could be considered for testing with HSAT.

On a practical level, the patient’s insurance coverage may be the major determinant for which test is utilized. We recommend that if a single HSAT is negative, inconclusive, or technically inadequate, in-lab PSG be performed for the diagnosis of OSA.¹³⁸ More robust evidence-based economic evaluation is needed to guide decision-makers about the cost and effectiveness of home-based testing compared to PSG. The gold standard diagnostic tool is PSG, yet the test is expensive, labor intensive, and timeconsuming. Home-based testing can broaden access to diagnostic services, in hopes to reduce the substantial economic burden related to OSA and provide consistent results signifying similar effectiveness.¹³⁸

We recommend against home sleep testing for the routine assessment of isolated insomnia, restless legs syndrome (RLSs), or uncomplicated parasomnias if one of these conditions is considered the likely primary abnormality. HSAT may be considered if there is a high suspicion of overlapping OSA, as insomnia and OSA may co-exist in up to 30% of sleep clinic populations.⁵⁸⁸ We recommend that if an HSAT is used to rule in OSA, there are clearly defined pathways for assessing the pre-test probability and co-morbidities. We recommend that if HSAT confirms diagnosis of OSA, there is no further need to confirm with in-lab PSG.^458,579 (Table VI.F.2) the recent AASM CPG. Namely, it addresses new literature assessing the utility of PSG in patients with significant cardiorespiratory disease, potential respiratory muscle weakness due to neuromuscular condition, awake hypoventilation, suspicion of sleep related hypoventilation, chronic opioid medication use, history of stroke, or severe insomnia.⁵⁷²

TABLE VI.F. 2.

Home sleep study testing compared to PSG

Study	Year	LOE	Study design	Study groups	Clinical endpoint	Conclusion
Kushida et al.578	2005	1a	Guideline	3464 studies	Standards of practice: AASM indications for PSG for sleep disorders.	For patients with high-pretest-probability, type III sleep study may be an acceptable alternative to full-night PSG. In the unattended setting, or in patients without high pretest probability stratification, the data does not support the use of these devices.
Collop et al.579	2007	1a	Systematic review	70 studies	Clinical Guidelines for the use of unattended portable monitors in diagnosing OSA.	Portable monitoring for the diagnosis of OSA should be performed only in conjunction with a comprehensive sleep evaluation. In the absence of a comprehensive sleep evaluation, there is no indication for portable monitoring.
Collop et al.573	2011	1a	Systematic review	27 RCTs	OOC testing with devices that measure Sleep, Cardiovascular, Oximetry, Position, Effort, Respiratory (SCOPER) parameters.	The literature is currently inadequate to state that a thermistor alone without any effort sensor is adequate to diagnose OSA.
Masa et al.580	2011	1b	Multicenter, randomized, blinded crossover study	366 patients suspected of OSA	Diagnostic and cost effectiveness of home testing compared with in-hospital PSG.	Home respiratory polygraphy is a cost-effective alternative to polysomnography for sleep apnea/hypopnea diagnosis. Telematic procedures may help patients with limited mobility and those who live a long way from the sleep center.
Masa et al.589	2013	1b	Randomized control blinded trial	348 patients with suspected OSA	Diagnostic cost-effectiveness of a sequential HRP scoring protocol compared with manual HRP scoring, and with in-hospital PSG.	Manual HRP scoring had better agreement than automatic HRP scoring; The sequential HRP protocol is a cost-effective alternative to PSG; and the cost savings of the sequential HRP protocol is low in comparison to the manual HRP protocol.
Kim et al.138	2015	1a	Economic analysis of RCT-home PAP	373 at risk for moderate to severe OSA	Cost-minimization analysis of home vs. lab sleep study.	Per subject costs for the in-lab testing were $1840 compared to $1575 for home testing for the payer. Costs for the laboratory arm were $1697 compared to $1736 in the home arm for the provider.
Jonas et al.458	2017	1a	Randomeffects meta-analyses	110 studies	Review primary care–relevant evidence on screening adults for OSA, test accuracy, and treatment of OSA.	There is uncertainty about the clinical utility of all potential screening tools. Although screening with multivariable apnea prediction, followed by home PM testing may have promise for distinguishing persons in the general population who are more or less likely to have OSA, current evidence is limited.
Abrahamyan et al.585	2018	1a	Systematic review and meta-analysis	24 full-text articles for final review	Systematically review the evidence on diagnostic ability of type IV PMs compared to PSG for OSA diagnosis.	Use of Level 4 PMs in clinical practice can potentially widen access to diagnosis and treatment of OSA, but evidence is not strong.
Corral et al.590	2017	1b	Multicentric noninferiority RCT with two open parallel arms and a cost-effectiveness analysis	430 screened patients with sleep apnea suspicion	Long-term effectiveness of home versus lab PSG in patients with intermediate-to-high sleep apnea suspicion.	The home testing protocol was noninferior to the PSG protocol based on the Epworth scale. Home testing was the Most cost-effective protocol, with a lower per-patient cost of 416.7€.
Douglas et al.591	2017	1a	Guideline/position paper utilizing well designed RCTs		Consensus statement on the indications and performance of sleep studies in adults.	There is increasing evidence supporting the use of some home-based type III and IV type sleep studies to “rule-in” moderate to severe obstructive sleep apnea in high prevalence obstructive sleep apnea, and should be used under the supervision of an accredited sleep physician.
Kapur et al.572	2017	1a	Systematic review and meta-analysis	98 studies included in evidence-based recommendations and 86 included in meta-analysis Clinical practice guideline (AASM).	Guidelines on appropriate and effective diagnosis of OSA.	Polysomnography, or home sleep apnea testing with a technically adequate device, is used for the diagnosis of OSA in uncomplicated adult patients presenting with signs and symptoms that indicate an increased risk of moderate to severe OSA. If a single home sleep apnea test is negative, PSG be performed for the diagnosis of OSA.
El Shayeb et al.583	2014	1a	Systematic review and meta-analysis	From 59 studies, 19 studies were included in the meta-analysis	Assess the diagnostic accuracy of Level 3 testing compared with Level 1 testing and to identify the appropriate patient population for each test.	Level 3 sleep studies are safe and convenient for diagnosing OSA in patients with a high pretest probability of moderate to severe forms of the condition without substantial comorbidities.
Rosen et al.592	2012	1b	Randomized, open-label, parallel group, unblinded, multicenter clinical trial	7 AASM accredited centers recruited adults with high probability of OSA and ESS > 12 Home PAP study	To test the utility of an integrated clinical pathway for OSA diagnosis and CPAP treatment using portable monitoring devices.	A home-based strategy for diagnosis and treatment compared with in-laboratory PSG was not inferior in terms of acceptance, adherence, time to treatment, and functional improvements.
Gabriela et al.593.	2019	2b	Randomized, prospective, cross-over and single blind clinical trial	251 patients	Automatic validation of a new HRP system.	The automatic analysis of the HRP BTI-APNiA software presents a high validity in comparison to the AHI results measured by PSG. HRP BTI-APNiA is a valid alternative to PSG.
Flemons et al.594	2003	1a	Systematic review	51 studies	To assess the utility of portable monitors in diagnosing sleep apnea in adults.	High-quality studies of type III monitors in the sleep laboratory attended setting had low false-positive rates; most studies found a threshold that distinguished patients with sleep apnea from those without.
Chesson et al.595	1997	2a	Systematic review	MEDLINE search; January 1966–April 1996	Indications for PSG.	22% failure rate of home-based study to diagnose OSA.
Garcia-Diaz et al.596	2007	1b	Prospective randomized study with blinded analysis	62 patients with suspected OSA included	Utility and reliability of a respiratory polygraphy (RP) device with actigraphy in the diagnosis of sleep apnea-hypopnea syndrome.	HPR is an effective and reliable technique for the diagnosis of SAHS, although it is less sensitive than LRP. Wrist actigraphy improves the results of HRP only slightly.
Ayappa et al.597	2008	2b	Prospective study with blinded analysis	102 subjects recruited. 96 returned to lab	Validity of the Apnea Risk Evaluation System (ARES) Unicorder, for the evaluation of sleep disordered breathing.	ARES Unicorder provides acceptably accurate estimates of SDB indices compared to conventional laboratory NPSG for both the simultaneous and in-home ARES data.
Garg et al.581	2014	1b	RCT crossover	75 urban African Americans with high pre-test probability of OSA, identified with the Berlin questionnaire	Feasibility of home PM in an urban population at risk for OSA compared to in-laboratory polysomnography (PSG) and patient preference with respect to home PM versus PSG.	Home PM for diagnosis of OSA in a high-risk urban population is feasible, accurate, and preferred by patients. As home PM may improve access to care, the cost-effectiveness of this diagnostic strategy for OSA should be examined in underserved urban and rural populations.
Skomro et al.598	2010	1b	Randomized trial	102 subjects	Compared subjective sleepiness, sleep quality, quality of life, BP, and CPAP adherence after 4 weeks of CPAP therapy in subjects diagnosed via lab versus home testing.	Compared with the home-based protocol, diagnosis and treatment of OSA in the sleep laboratory does not lead to superior 4-week outcomes in sleepiness scores, sleep quality, quality of life, BP, and CPAP adherence.
Abraham et al.584	2006	2b	Prospective multi-center study	50 patients with NYHAIII CHF (SEARCH study)	Validity and clinical utility of home testing for SDB and arrhythmias in HF.	With CPS compared in in-lab PSG, the diagnostic accuracy was between 83% and 87% for OSA based on AHI of 5, 10, and 15.
Bravata et al.599	2017	1a	Randomized controlled intervention trial	225 randomized patients	Evaluate whether the intervention strategy improved sleep apnea detection and treatment, and hypertension control among patients with chronic cerebrovascular disease and hypertension.	The use of portable polysomnography, and auto-titrating CPAP in the patients’ homes, improved both the diagnosis and the treatment for sleep apnea compared with usual care but did not lower blood pressure.
Kotzian et al.586	2018	1b	Single-blind, single center, randomized controlled trial	55 patients (HOPES study)	Determine whether PAP adherence in patients who had a stroke with OSA can be improved by a PAP training strategy during in hospital rehabilitation combined with a telemedicine monitoring system after discharge.	Pre-results to our clinical experience, a severe SA in the screening PG remains a severe SA also in PSG. The diagnosis will not Change. We think that patients who had severely affected stroke need a quick access to therapy.
Fitzpatrick et al.600	2020	1b	Randomized, parallel, multicenter, single-blind, pragmatic controlled trial	233 patients (SIESTA Trial)	To evaluate the accuracy of the clinical diagnosis of OSA informed by the home sleep study with a type IV portable monitor versus type I polysomnography	Home testing with portable devices plays a valuable role for diagnosing of OSA in a variety of settings.
Hui et al.601	2017	1b	Prospective, randomized controlled CPAP parallel study	316 patients	Comparisons of home-based versus hospital-based approach in managing patients with suspected obstructive sleep apnea.	Home-based approach is non-inferior to hospital-based approach in managing patients with suspected OSAS, with shorter waiting time, and substantial cost savings.
Mulgrew et al.602	2007	1b	Randomized, controlled, open-label trial	68 patients	To test the utility of a diagnostic algorithm in conjunction with ambulatory CPAP titration in initial management of obstructive sleep apnea.	In patients with a high probability of obstructive sleep apnea, PSG confers no advantage over the ambulatory approach in terms of diagnosis and CPAP titration. The ambulatory approach may improve adherence to treatment. When access to PSG is inadequate, the ambulatory approach can expedite management of patients in need of treatment.
Guerrero et al.603	2014	1b	Randomized, blinded, crossover study	56 patients	Evaluate three night portable monitoring for OSA diagnosis.	Three consecutive nights of portable monitoring at home evaluated by a qualified sleep specialist is useful for the management of patients without high pretest probability of obstructive sleep apnea or with comorbidities.
Ferber et al.604	1994	2a	Systematic review	Literature review; MEDLINE (1966–1994)	Usefulness of portable Recording in the assessment of OSA.	No clear guidance for who is appropriate for home testing.
Morales et al.605	2012	2a	Prospective cohort study	452 participants	Assess utility of home testing for the elderly with the complaint of daytime sleepiness.	Unattended, self-assembled, in-home sleep studies recording airflow and respiratory effort are most useful along with a comprehensive sleep history, is accurate in identifying severe OSAS in older adults.
Pietzsch et al.606	2011	1b	Decision- analytic Markov model using Tree-Age Pro 2009 Suite	Comparison of clinical health guidelines and health- economic studies	Benefits and cost-effectiveness of diagnostic tests.	For payers, a home-based diagnostic pathway for obstructive sleep apnea with robust patient support incurs fewer costs than a laboratory-based pathway. For providers, costs are comparable if not higher, resulting in a negative operating margin.

VI.F.3 |. In-lab studies

Traditionally, in-lab PSG has been the testing modality of choice to diagnose and assess the severity of OSA.⁵⁷⁸ The 2017 AASM CPG recommended that PSG remains the diagnostic testing modality for patients suspected of having OSA who had significant comorbidities or who failed an initial HSAT, but that HSAT or PSG could be used to diagnose OSA for uncomplicated patients considered to be at risk of moderate to severe OSA.⁵⁷² This diagnostic testing section reviewed the literature and updated the indications for PSG in the context of the recent AASM CPG. Namely, it addresses new literature assessing the utility of PSG in patients with significant cardiorespiratory disease, potential respiratory muscle weakness due to neuromuscular condition, awake hypoventilation, suspicion of sleep related hypoventilation, chronic opioid medication use, history of stroke, or severe insomnia.⁵⁷²

Heart failure with reduced ejection fraction (hFrEF or congestive heart failure [CHF]):

hFrEF patients are at high risk for SDB, including both OSA and CSA. HSATs have been evaluated for the diagnosis of SDB in multiple studies of patients with hFrEF, including three Level 1b studies, one Level 2b study, and one Level 3b study.^{584,607–610} In a study of 100 hFrEF patients undergoing simultaneous home PSG and a two channel version of the ApneaLink, accuracy for an AHI > 5 and >15 events/hour was reported at 94%, though the PSG required a 4% desaturation to qualify as a respiratory event and the HSAT could not distinguish between obstructive and central events.⁶⁰⁹ Aurora et al. reported that adding an effort belt to the IT (ApneaLink Plus) yielded similar accuracy and could distinguish obstructive versus central events (obstructive AHI r = 0.91, central AHI r = 0.998), though the study included only 57 inpatients with decompensated hFrEF.⁶⁰⁸ In a prospective multicenter cross-sectional study, stable hFrEF patients (N = 47 in the validation group) underwent simultaneous SleepMinder (a novel two channel touchless HSAT) and PSG.⁶¹⁰ The SleepMinder required a 50% reduction in airflow only for a respiratory event, while the PSG required 30% drop in flow and 4% desaturations. The SleepMinder had an area under the curve (AUC) of 0.85 but misclassified 19% of patients using an AHI cutoff of 15 to define SDB and was unable to distinguish obstructive versus central events. The non-Level 1 studies in this group were downgraded due to small numbers and/or antiquated criteria used to define respiratory events.^584,607

Chronic obstructive pulmonary disease (COPD):

Patients with COPD are at risk for OSA, but also hypoventilation and nocturnal hypoxemia. Limited data is available to assess the utility of HSATs in the diagnosis of SDB in patients with COPD (one Level 1b study and one Level 3b study).^611,612 In prospective single center cross-sectional study of 90 outpatients with stable COPD and symptoms of OSA, Chang et al. evaluated one night with an HSAT device (Nox-T3) and then an in lab PSG with simultaneous Nox-T3 within 1 week.⁶¹¹ The best agreement was found when the studies were done on the same night, used a 4% desaturation criteria for respiratory events, and used an AHI cutoff of 15 to define SDB. Kappa values ranged from as low as 0.48 (different night, 3% desaturation criteria, AHI cutoff of 5) to 0.93 (same night, 4% desaturation criteria, AHI cutoff of 15). A small (N = 33) prospective single center cross-sectional study of patients with stable Gold II and III COPD and symptoms of OSA compared an HSAT (WatchPAT) to in lab PSG and found good sensitivity (92%–96%) with moderate specificities (55%–65%), and a concordance of 79% between the tests for an AHI > 15.⁶¹² However, of note, this later study required only a 50% drop in airflow (no associated desaturation) to define a respiratory event on the PSG.

Post-stroke:

Patients who have had a stroke have a roughly 60%–70% risk of having OSA.⁶¹³ A single Level 2b study compared an HSAT device (Embletta) to simultaneous PSG within 3 days of an ischemic stroke.⁶¹⁴ In a prospective single center cross-sectional study, Chernyshev et al. found HSAT was accurate for diagnosing OSA (accuracy 1.0 and 0.95 for AHI cutoff’s of 5 and 15). Both HSAT and PSG events were scored according to the 4% desaturation criteria. The main limitation to the study was that it included only 21 participants.

Technically inadequate or normal initial HSAT:

A Level 2b study performed as a retrospective single center review of patients with a high pretest probability of OSA who had a technically inadequate (N = 111) or normal HSAT (N = 127) but then a subsequent PSG, found OSA in 71% of the patients with technically inadequate studies and 24% of the patients with normal HSATs.⁶¹⁵ Older age and lower ESS were associated with OSA on the subsequent PSG. (Table VI.F.3)

TABLE VI.F. 3.

Diagnostic testing: indications for in-lab polysomnography

Study	Year	LOE	Study design	Study groups	Clinical end point	Conclusion
CHF
Abraham et al.584	2006	2b	Prospective multicenter study comparing diagnostic accuracy of ClearPath System (CPS) vs. PSG. Stable CHF with EF ≤35%	Night 1 did both in the lab and Night 2 (<3 nights apart) was home CPS. N = 50	RDI as determined by 3% desaturations for respiratory events and “flow change.”	1) Same night: RDI>5 sens 92%, spec 52%, acc 73%, AUC NR RDI>15 sens 67%, spec 78%, acc 75%, AUC NR 2) Different night: RDI>5 acc 73% RDI>15 acc 77%
de Vries et al.609	2015	1b	Prospective single center cross-sectional study in patients with chronic heart failure (stable CHF by clinical judgment).	Subjects underwent the ApneaLink (2 channel – airflow + oximetry) and simultaneous PSG at home. N = 100	AHI as determined by respiratory events required 30% drop in airflow and 4% desaturations.	AHI > 5 sens 98%, spec 60%, acc NR, AUC 0.94 AHI>15 sens 92.9%, spec 91.9%, acc NR, AUC 0.94 ICC 0.85 for categories 3) 29% CSA, 19% OSA, 13% mixed 4) Best accuracy at AHI 15. 5) Cannot differentiate CSA vs. OSA
Araújo et al.607	2018	3b	Prospective single center cross-sectional study in patients with chronic heart failure (stable CHF by clinical judgment).	Subjects were studied by the ApneaLink (2 channel – airflow + oximetry) and PSG simultaneously during in the sleep laboratory. N = 35	AHI as determined by respiratory events required 50% drop in airflow and 3% desaturations.	AHI > 5 sens 81.8%, spec 61.5%, acc 74.2, AUC 0.85 AHI>15 sens 83.3%, spec 91.3%, acc 88.6, AUC 0.93 AHI correlation r = 0.79 Highest accuracy at AHI 15. Cannot differentiate CSA vs. OSA Downgraded due to small number and EF50%.
Aurora et al.608	2018	1b	Prospective single center cross sectional study in hospitalized patients with decompensated heart failure (stabilized at the time of testing).	Subjects underwent concurrent ApneaLink Plus (3 channel – airflow, oximetry and effort belt) and PSG. Recordings blindly scored for OSA and CSA. N = 57	AHI as determined by respiratory events that required a 50% drop in airflow and 3% desaturations (PSG allowed arousals).	AHI > 5 sens 95.8%, spec 80.0%, acc NR, AUC NR Central AHI>5 sens 90.9%, spec 100%, acc NR, AUC NR 3) ICC 0.89 for categories, Obs AHI r = 0.91, Central AHI r = 0.99 58.5% central AHI > 5 4) Higher accuracy for central vs. obstructive. 5) Cannot differentiate CSA vs. OSA.
Savage et al.610	2016	1b	Prospective multicenter cross-sectional study. Stable CHF patients with an EF<45%.	Subjects underwent simultaneous Sleep Minder (airflow + movement via electromagnetic signals) and PSG. Development (D) (n = 28) and validation (V) (n = 47) groups were studied. N = 75	AHI as determined by Sleep Minder (SM) required 50% reduction in airflow only. PSG required 30% drop in flow and 4% desaturations.	AHI > 5 sens NR, spec NR, acc NR, AUC NR AHI>15 sens 70%, spec 89%, acc NR, AUC 0.85. AUC 0.85 for all, 19% misclassified (> <AHI 15) by SM. 25%/34% (D/V) with OSA, 11%/9% with CSA. Best accuracy at AHI >30. Cannot differentiate CSA vs. OSA.
COPD
Chang et al.611	2019	1b	Prospective single center cross-sectional study in outpatients with stable COPD and symptoms of OSA.	Subjects did 1 night with Nox-T3 and then an in-lab PSG with simultaneous Nox-T3 within 1 week. N = 90	AHI as determined by 2 different definitions for respiratory events: 1) Nox-T3 and PSG respiratory events required 30% drop in airflow and 4% desaturation, and 2) Nox-T3 and PSG respiratory events required 30% drop in airflow and 3% desaturation or, for PSG, an arousal.	Same night 4%: AHI>5 sens 96%, spec 84%, kappa 0.82 AHI>15 sens 95%, spec 98%, kappa 0.93 Different night 4%: AHI>5 sens 95%, spec 78%, kappa 0.75 AHI>15 sens 74%, spec 98%, kappa 0.74 Same night 3% kappa AHI>5 0.58, kappa AHI>15 0.88 Different night 3% kappa AHI>5 0.48, kappa AHI>15 0.70 Nox-T3 more often found hypoxemia (15 vs. 5 with AHI < 5 and > 5 min with sats ≤88%.)
Jen et al.612	2020	3b	Prospective single center cross-sectional study in outpatients > 40 years old with stable Gold II and III COPD and OSA symptoms.	In random order, subjects underwent 1 night PSG and WatchPAT device in the lab and another night with the WatchPAT at home. N = 33	AHI as determined by WatchPAT and PSG. WatchPAT uses PAT with either a 3% with “arousal” or 4% desaturation. PSG required 50% drop in airflow without desaturation or arousals	AHI>5 sens 96%, spec 55.6% AHI>15 sens 92.3%, spec 65.0% 3)Intraindividual AHI difference 78.8% concordance (AHI>15 and difference <10)
CVA
Chernyshev et al.614	2015	2b	Prospective single center cross-sectional study of inpatients admitted for acute ischemic stroke. Studied within 72 h of the CVA.	Subjects underwent a simultaneous PSG and HSAT within 3 days of their CVA. N = 21 BMI 33.1 (OSA) vs. 23.8 (no OSA), 66.6% had an AHI > 5, 48% had OSA, 19% had CSA	AHI as determined by HSAT and PSG. Respiratory events required 30% drop and 4% desaturation.	For OSA only: AHI>5 sens 100%, spec 85.7%, acc NR, AUC 1.0 AHI>15 sens 100%, spec 83.9%, acc NR, AUC 0.95 Intraindividual AHI difference 1.5 Downgraded due to small number.
Failed HSAT
Zeidler et al.615	2015	2b	Retrospective single center review of patients referred for OSA evaluation who had a technically inadequate (N = 111) or normal HSAT (N = 127) but then had a subsequent PSG.	All had a high pretest probability of OSA. N = 238	Respiratory events scoring criteria not defined for HSAT or PSG. OSA defined by an AHI > 5.	1) Technically inadequate HSAT: 71% with OSA: 38.7% mild, 32.4% moderate/severe 2) Normal HSATs: 24% OSA: 18.1% mild, 5.5% moderate/severe Older age and lower ESS were associated with OSA on PSG.

VI.F.4 |. Oximetry

General population:

In the general population, two Level 1b studies,^499,616 three Level 2b studies,^617–619 and one Level 3b⁶²⁰ found reasonably good correlations between oximetry ODI and PSG AHIs, though primarily when the 4% desaturation criteria was used to score respiratory events. For example, in a study by Pataka et al., utilizing a 4% desaturation definition, there was good correlation between the ODI4% and the PSG ODI (r = 0.95), and for AHI > 15, ODI4% had a sensitivity, a specificity and an AUC of 82%, 94%, and 90%.⁶¹⁶ However, a number of these studies were downgraded for using the oximetry off the PSG as the comparison to the PSG derived AHI,^617,618 or for using highly selective populations.⁶¹⁹

Two studies used acoustic devices to diagnosis OSA, one Level 2b study⁶²¹ and one Level 4 study.⁶²² In an exploratory study by Erdenebayar et al., there was moderate accuracy between in-lab PSG AHI and the piezoelectric sensor snoring index when worn at the same time as the PSG.⁶²¹

Heart failure with reduced ejection fraction (hFrEF or CHF):

The accuracy of oximetry in patients with hFrEF was assessed by two Level 1b studies (Sharma et al., 2017 and Ward et al.)^619,623 and one Level 3b study (Sharma et al., 2015).⁶²⁴ The study by Sharma et al., 2017, was a prospective single center, controlled trial of 105 inpatients with acute CHF, where simultaneous HSAT with ApneaLink and a high-resolution pulse oximetry (HRPO) was performed for a single night.⁶¹⁹ The HSAT-derived REI was compared to the HRPO-derived ODI using 30% drop in flow with 4% oxygen desaturations. The receiver operating curve (ROC) analysis showed an area under the ROC curve (AUC) of 0.89 for REI > 5 events/h with 88% of the REI in the moderate–severe category being correctly classified. However, HRPO was unable to reliably differentiate between central and obstructive respiratory events. The second Level 1b study by Ward et al. was also a prospective, single center study of 173 CHF patients with simultaneous recording of unattended PSG, ambulatory electrocardiography and overnight pulse oximetry performed at the home or hospital per patient preference.⁶²³ The results demonstrated that at the cutoff of >7.5 desaturations/h, the ODI3% had a sensitivity 97%, a specificity 32%, a negative likelihood ratio 0.08, and a positive likelihood ratio 1.42. The diagnostic accuracy increased using a cutoff of 12.5 desaturations/h, with a sensitivity 93% and a specificity 73%. The ODI3% had an AUC of 0.92 (95% C.I. 0.88–0.96) for the detection of SDB in CHF, at the cutoff of >7.5 desaturations/h. The third study, by Sharma et al. (2015), a Level 3b study, was a prospective, single center study, in patients admitted with acute decompensated HF with a high clinical suspicion of SDB and an ODI4% ≥5 on overnight photoplethysmographic signal recording who underwent PSG.⁶²⁴ Among 86 patients who had oximetry, 68 underwent outpatient PSG within 4 weeks of discharge. Utilizing an AHI ≥ 5 to define OSA, the AUC was 0.82.

COPD:

One each of Level 1b, 2b, and 3b studies met criteria for inclusion.^625–627 The study by Andrés-Blanco et al., a level 1b study, was a prospective single center in 193 COPD patients, where supervised portable oximetry was compared simultaneously to in-hospital PSG as well as to unsupervised portable oximetry at home.⁶²⁵ A control group of 100 non-COPD patients was also included. An ODI4% cutoff was used for the oximeter. There were no significant differences between COPD and non-COPD groups in both settings, particularly for severe OSAS. In a Level 2b study, Scott et al. compared overnight oximetry results to PSG in 50 COPD patients on long-term oxygen therapy.⁶²⁷ Pulse oximetry tracing interpretation had a modest diagnostic value in identifying OSA in patients with moderate to severe COPD; the AUC was 0.57–0.64.

Atrial fibrillation (AF):

One Level 2b study by Linz et al. met criteria for inclusion.²⁵ It was a large prospective single center study in patients with documented AF (62% paroxysmal AF) who underwent PSG. Overnight oximetry from the PSG was used to determine the ODI and this was validated against the PSG AHI. In 439 patients with AF, the median AHI was 9.5 [3.6–21.0]/h and the prevalence of moderate (AHI 15–29/h) and severe SDB (AHI ≥ 30/h) was 17.3% and 16.6%, respectively. The ODI4% was able to detect moderate-to-severe SDB with an AUC of 0.95 and only severe SDB with AUC 0.93. An ODI4% of 7.6/h yielded a sensitivity and specificity for AHI ≥ 30/h of 89% and 83%, respectively. However, oximetry was unable to distinguish between central or obstructive respiratory events.

Post-stroke:

There were only two Level 2b oximetry studies and one Level 3b sonography study that were included in this analysis.^628–630 Lin et al. performed a retrospective study in 254 acute stroke patients who underwent HSAT.⁶²⁹ The ODI3% from the pulse oximetry channel was compared to REI obtained from HSAT devices. Nocturnal pulse oximetry had a high diagnostic accuracy in predicting moderate to severe SDB in patients with acute stroke and the ODI3% was accurate in predicting SDB at different REI thresholds (REI ≥ 5, REI ≥ 15, and REI ≥ 30 events/h) with AUCs of 0.965, 0.974, and 0.951, respectively. Aaronson et al. also performed retrospective analysis of data obtained from 56 stroke patients who underwent nocturnal polygraphy and oximetry.⁶²⁸ Sensitivity, specificity, and positive and negative predictive values (NPVs) for the ODI4% ≥15 were, 77%, 100%, 100%, and 83%, respectively. Ryan et al. reported on 23 patients who were post-stroke and had SDB.⁶³⁰ Using a cutoff AHI of ≥15 by PSG to diagnose OSA, in-lab use of the BresoDx (single channel acoustic device) had a sensitivity of 90.0%, a specificity of 84.6%, and an overall accuracy of 87.0%. Bland–Altman plot showed close agreement, although a tendency for the BresoDx device to slightly overestimate the AHI. The AUCs for PSG diagnostic AHI thresholds of ≥5, ≥10, and ≥15 were 0.90, 0.91, and 1.00, respectively. Comparing the home use of BresoDx versus in-lab PSG, using an AHI threshold of ≥15, the BresoDx had a sensitivity of 100%, a specificity of 85.7%, and an overall accuracy was 91.3% (Table VI.F.4).

TABLE VI.F. 4.

Evidence for oximetry for diagnosis of OSA

Study	Year	LOE	Study design	Study groups	Clinical end point	Conclusion
CHF
Sharma et al.619	2017	1b	Prospective single center, controlled trial of patients admitted with CHF.	Simultaneous measurement of apnea link and high-resolution pulse oximetry (HRPO) for a single night. N = 105 61 (58%) M	HRPO-derived ODI (oxygen desaturation index) was compared with PM-derived respiratory event index (REI) using receiver operator characteristic (ROC) curve analysis and a Bland–Altman plot.	1) ROC area under curve (AUC) was 0.89 for REI > 5 events/h. AUC ranged from 0.84 (REI ≤ 10 events/h) to 0.89 (REI ≤ 5 events/h and REI ≤ 20 events/h). 2) The Bland–Altman plot had good agreement. 3) 88% of the REI in moderate–severe category were correctly classified. 4) Cannot differentiate CSA vs. OSA.
Sharma et al.624	2015	3b	Prospective, single center cohort of consecutively admitted acute decompensated heart failure patients with high clinical suspicion of SDB.	Overnight (ON) inpatient oximetry (photoplethysmography) compared with outpatient PSG apnea hypopnea index (AHI). N = 105 subjects had ON oximetry and 68 underwent outpatient PSG within 4 weeks of discharge.	PSG defined hypopneas as some drop in flow with 4% oxygen desaturation compared with ON oximetryODI of 4% desaturation using ROC analysis and Bland–Altman plot.	1) ODI correlated with AHI with AUC of 0.82 on ROC for AHI ≥5. 2) The Bland–Altman plot had no major bias.
Ward et al.623	2012	1b	Prospective, single center cohort of CHF patients from cardiology clinics.	Simultaneous unattended PSG, ambulatory electrocardiography, and ON pulse oximetry at home or hospital N = 173 86% M.	Compared oximetry % ODI cutoff of >7.5 desaturations/h to PSG AHI >15/h. ODI used 3% desaturation. PSG defined hypopneas as 50% drop in flow with 3% desaturation or arousal.	1) At a cutoff of >7.5 desaturations/h, the ODI3% had sensitivity 97%, specificity 32%, negative likelihood ratio (LR) 0.08, and positive LR 1.42. 2) At a cut-off of 12.5 desaturations/h, ODI3% sensitivity was 93% and specificity was 73%. 3) The 3% ODI had an AUC under ROC curve of 0.92 for detection of SDB in CHF, at the cutoff of >7.5 desaturations/h.
COPD
Andrés-Blanco et al.625	2017	1b	Prospective single center cohorts.	Simultaneous portable ON oximetry at home and in-hospital PSG; and unsupervised portable ON oximetry at home. Two independent validation datasets were analyzed: COPD versusnon-COPD. N = 110 non-COPD test set (69% M) and 68 COPD test group (88% M).	A regression-based multilayer perceptron (MLP) artificial neural network (ANN) was trained to estimate AHI from portable oximetry recordings. Two independent validation datasets were analyzed: COPD vs. non-COPD.	1. Portable ON oximetry-based ANN reached similar ICC values between the estimated and actual AHI for the non-COPD and the COPD groups either in the hospital (non-COPD: 0.937, COPD: 0.936) and at home (non-COPD: 0.731, COPD: 0.788) setting. 2. No significant differences in ROC between COPD and non-COPD groups in both settings.
Lajoie et al.626	2020	3b	Prospective cohort recruited from an ongoing multicenter trial.	Compared home ON oximetry and laboratory-based PSG in patients with moderate-to-severe COPD. N = 90 45 had OSA, 71% M 45 did not have OSA, 87% M	ODI3% used for oximetry. AHI hypopnea definition not stated.	1. Oxygen desaturation indices obtained with nocturnal oximetry and during PSG were not correlated (r = –0.27; p = 0.1). 2. Diagnosis of OSA in COPD should not be based solely on oximetry. OSA was confirmed in only 50% of subjects with oximetry tracings suggestive of OSA.
Scott et al.627	2014	2b	Consecutive chart review of the inpatient pulmonary rehabilitation service.	Subjects with moderate–severe COPD who were clinically prescribed oximetry and PSG. N = 59 46% M	Criteria consisted of visually identified desaturation “events” (sustained desaturation ≥4%, 1 h time scale), “patterns” (≥3 similar desaturation/saturation cycles, 15 min time scale) and the automated oxygen desaturation index. Compared using AUC.	1) Thirty-five were correctly identified as having OSA/no OSA with accuracy of 59%, a sensitivity and specificity of 59% and 60%, respectively; AUC 0.57. 2) Using software-computed desaturation events (hypoxemia ≥4% for ≥10 s) indexed at ≥15 events/h of sleep as diagnostic criteria, sensitivity was 60%, specificity was 63%, and the AUC 0.64.
Atrial Fibrillation
Linz et al.25	2018	2b	Prospectively single center cohort in patients with atrial fibrillation (AF) who underwent PSG.	Subjects with documented AF. N = 439 69% M	ON oximetry from the PSG was used to determine the ODI. ODI was validated against PSG AHI. ODI4% used for oximetry, which came off PSG. PSG hypopnea definition was a 30% drop in flow for 10 s with either a 3% desaturation or an arousal.	1) ODI was able to detect moderate-to-severe SDB (AHI ≥ 15/h) AUC: 0.951; severe SDB (AHI ≥ 30/h) AUC 0.932. 2) An ODI cut-off of 4.1/h had 91% sensitivity and 83% specificity in patients with and without AHI ≥ 15/h. 3) An ODI of 7.6/h yielded a sensitivity and specificity for AHI ≥ 30/h of 89% and 83%, respectively. 4) Cannot differentiate CSA vs. OSA.
CVA
Lin et al.629	2018	2b	Retrospective chart analysis.	Subjects with acute stroke or TIA underwent ON oximetry and HSAT. N = 254 50.7% M 232 (91.3%) were ischemic or TIA.	ODI from pulse oximetry channel was compared to respiratory event index (REI) obtained from HSAT devices. ODI3% used for oximetry. REI3% used for HSAT.	1) ODI3% had correlation (r = 0.902) and agreement with REI3%. 2) ODI3% was accurate in predicting SDB at different REI thresholds (REI ≥ 5, REI ≥ 15, and REI ≥ 30 events/h) with AUC of 0.965, 0.974, and 0.951, respectively. 3) An ODI3% ≥ 5 events/h rules in the presence of SDB (specificity 91.7%, PPV 96.3%). 4) An ODI3% ≥ 15 events/h rules in moderate to severe SDB (specificity 96.4%, PPV 95%) and an ODI3% < 5 events/h rules out moderate to severe SDB (sensitivity 100%, NPV 100%).
Aaronson et al.628	2012	2b	Retrospective study of stroke patients.	Compared polygraphy and oximetry from HSAT in stroke subjects. N = 56 62% male, 46% of the stroke patients had OSA. 69% with OSA were ischemic strokes.	Compared REI to ODI. REI hypopneas defined as 50% drop in flow with a 4% desaturation. ODI used 4% desaturation.	1) Sensitivity, specificity, and PPV and NPV for the ODI4% ≥15 were, respectively, 77%, 100%, 100%, and 83%. 2) ODI4% predicted 87% of the variance in the REI. 3) Given a 46% prevalence of OSA in stroke, the PPV of oximetry was 100% with an NPV of 83%
Ryan et al.630	2017	3b	Prospective cohort of patients with acute stroke in a stroke rehabilitation unit (SRU).	Compared testing with BresoDx – a portable single-channel acoustic device – both simultaneously during attended PSG in lab and unattended on the SRU. N = 23 48% M 78% had OSA (defined by AHI ≥15) on PSG. 74% of subjects were ischemic strokes.	Compared PSG AHI to BresoDx AHI. PSG hypopneas defined by a 30% drop in flow with 3% desaturation or arousal. Determined AUC and Bland–Altman plot.	1) Using cutoff AHI of ≥15 by PSG to diagnose OSA in-lab BresoDx had sensitivity of 90.0%, specificity of 84.6%, and accuracy of 87.0%. 2) Bland–Altman plot: good agreement, but BresoDx overestimated AHI by 4.4. 3) The AUCs for AHI in lab Breos vs. in-lab PSG at thresholds of ≥5, ≥10, and ≥15 were 0.90, 0.91, and 1.00, respectively. 4) For home BresoDx vs. in-lab PSG, at an AHI threshold of ≥15 had a sensitivity of 100%, specificity of 85.7%, and accuracy of 91.3%.
General
Pataka et al.616	2019	1b	Prospective study in a sleep clinic.	Compared sleep questionnaires STOP-BANG (SB), Berlin (BQ), Epworth Sleepiness Scale (ESS) completed by subjects with home oximetry and in laboratory PSG, to determine predictive value of test for CPAP initiation. N = 204 77.5% M	Determine correlations and accuracy. Compared PSG and oximetry values as well. PSG hypopneas defined by a 30% drop in flow with 4% desaturation or arousal. ODI used 4% desaturations.	1. Good correlation between oximetry ODI (ODIox) and PSG ODI (r = 0.95, p < 0.0001) and between ODIox and AHI (r = 0.811, p < 0.0001). ODIox ≥ 15 had sensitivity 89.3%, specificity 83.5%, PPV 87%, and NPV 86.4% for CPAP initiation. 2) Among questionnaires, ESS had highest specificity (68.6%) and PPV (68.6%) and SB had the highest sensitivity (98%) and NPV (80%) but the lowest specificity (11%) for CPAP initiation. Oximetry was superior to questionnaires for predicting CPAP treatment initiation.
Christensson et al.617	2018	2b	Prospective, observational multicenter trial of sleep clinic patients.	Subjects underwent HSAT (Nox-T3), ON oximetry, and STOP-BANG (SB) questionnaires. N = 449 subjects with suspected OSA. 61.5% M	Compared REI to ODI. Compared REI to SB questionnaire scores. HSAT hypopneas defined by 30% drop in flow and 30% desaturation. ODI used 3% for oximetry.	1) Strong correlation between REI and ODI3%, Spearman 0.96. 2) Positive correlation between SB score and ODI3%, Spearman ρ 0.50; An SB score of <2 almost excludes moderate to severe OSA, whereas nearly all OSA patients with an SB score ≥6 had OSA.
Sharma et al.619	2017	3b	Retrospective review of a large database of hospitalized inpatients. Only those high ODI on ON oximetry were offered PSG.	Compared in-hospital ON HRPO to PSG post-discharge. N = 1410 underwent in-hospital HRPO with 1092 having and ODI4% ≥ 5. Of these, 680 underwent PSG post-discharge. 54% M (of HRPO group).	Determined accuracy, AUC, and Bland–Altman plot of HRPO-determined ODI vs. AHI. ODI used 4% for oximetry. PSG hypopneas defined by a 30% drop in flow with 4% desaturation.	1) ODI4% ≥5 had sensitivity 0.89 and pecificity 0.48. 2) ODI4% ≥ 15 had a sensitivity 0.65 and specificity 0.90. 3) ODI4% ≥5 had an AUC of 0.83 for an AHI ≥5 and 0.76 for an AHI ≥15. 4) Bland–Altman plot showed no significant bias when using ODI vs. AHI to define SDB.
Hang et al.618	2015	2b	Prospective study of sleep clinic patients undergoing PSG for suspected OSA.	Oximeter from PSG was used for ODI calculation without considering other PSG information. N = 699 (though only analyzed 544 with adequate TST and acceptable PSG signals) 77.1% M. PSG results: 20.6% had an AHI 5–15, 21.4% had an AHI >15–30, 46.3% had an AHI >30.	Compared accuracy and AUC of ODI from PSG oximetry to AHI from same PSG. ODI used 3% and 4% desaturations. PSG defined hypopneas as 30% drop in flow with a 4% desaturation.	1) For AHI ≥ 15, ODI3% had sensitivity, specificity, and accuracy of 86.1%, 92.4%, and 89.5%. 2) For AHI ≥15, ODI4% had sensitivity, specificity, and accuracy of 85.7%, 89.7%, and 87.8%. 3) AUC for severe OSA: 0.953–0.957; AUC of 0.921–0.924 for moderate to severe OSA patients. 4) Limitation due to removal of those with low TST on PSG.
Chung et al.499	2012	1b	Prospective study of patients presenting to presurgical clinic for elective surgery.	Subjects underwent unattended PSG and ON oximetry on the same night. N = 475 45.7% M	Compared PSG AHI and ON oximetry ODI. Hypopnea definition was 30% drop in flow and 4% desaturation. ODI used 4% desaturation.	1) ODI4% > 5 had a sensitivity, specificity, accuracy of 0.96, 0.67, 87% for an AHI >5; and 0.99, 0.39, 61.7% for AHI >15. 2) ODI4%>15 had a sensitivity, specificity, accuracy of 0.45, 0.98, 62.1% for an AHI >5, and 0.70, 0.93, 84% for AHI >15. 3) The AUC for ODI to predict AHI >5, AHI >15, and AHI >30 was 0.908, 0.931, and 0.958, respectively.
del Campo et al.620	2006	2b	Prospective study of cohort of patients undergoing PSG for suspected OSA.	Oximetry and PSG done at the same time. Approximate entropy (ApEn) (a mathematical tool) was calculated off oximetry and compared with PSG data. N = 187 (22.5% had COPD) 79% M	Determined accuracy between PSG and ApEN. PSG hypopneas defined by 30% drop in flow with a 3% desaturation. ODI used 3% and 4% desaturations.	1) AHI correlated with ApEn (r = 0.607; p < 0.001). 2) For AHI > 10, ApEn at 0.679 had sensitivity, specificity, PPV, and NPV of 88.3%, 82.9%, 88.3%, and 82.9%, respectively.
Erdenebayar et al.621	2017	2b	Prospective cross-sectional study of patients referred to a sleep clinic.	Subjects underwent an in-lab PSG and piezo-electric sensor at the same time. The piezo-electric sensor detected snoring and heartbeat information, and snoring index (SI) and features based on pulse rate variability (PRV) analysis. A support vector machine (SVM) was used as a classifier to detect OSA events. N = 45 70% M	Compared accuracy of piezo-electric sensor with PSG. PSG scored per “AASM standards” but not defined further.	1) Mild OSA detection: sensitivity, specificity, and accuracy of 72.5%, 74.2%, and 71.5%; moderate OSA detection: 85.8%, 80.5%, and 80.0%; and severe OSA: 70.3%, 77.1%, and 71.9%. 2) Automatic snoring detection had sensitivity, specificity, and accuracy of 88.5%, 96.1%, and 95.6%. 3) Heartbeat detection had sensitivity and PPV of 94.3% and 87.1%, all respectively.
Alakuijala et al.622	2016	4	Prospective cross-sectional study of patients referred to a sleep clinic.	Subjects underwent a HSAT (Nox T3) at home. Periodic snoring data was collected from the same HSAT. N = 211 61% M There was no separate validation group.	Analyzed the percentage of periodic snoring during HSAT and compared to the AHI from the HSAT. Correlations and Bland–Altman plot were analyzed. The HSAT defined hypopneas by 3% desaturations.	1) AHI ranged from 0.1 to 116 events/h, and % of periodic snoring from 1% to 97%. 2) Positive correlation (r = 0.727, p < 0.001) between periodic snoring and AHI. 3) Sensitivity was 93.3%, specificity 35.1%, and NPV 75.0%. 4) Bland–Altman plot showed that periodic snoring percentage, and AHI agreed within range of various grades of OSA.
Neuromuscular disease
	2018	2b	Prospective cross-sectional study of patients followed for chronic respiratory failure due to neuromuscular disease, treated with chronic noninvasive ventilation (NIV).	All patients underwent the screening test panel (clinical evaluation, daytime arterial blood gas [ABG], nocturnal pulse oximetry [SpO2], and data from ventilator software), HSAT (Embletta Gold) and nocturnal transcutaneous CO2 (while on their NIV). N = 67	Compared accuracy among the tests. HSAT used 4% desaturation criteria. ODI3% used for oximetry.	1. Nocturnal SpO2 and daytime ABG all failed to accurately detect nocturnal hypoventilation (NH). 2. ODI3% had a high sensitivity but low specificity for identifying obstructive events on NIV.

Summary

This evidence-based review updated the review of the literature in the areas of diagnostic testing for patients with comorbid conditions and a moderate to high risk for OSA. While new data was found evaluating alternative diagnostic approaches, the weight of the evidence for each comorbidity did not support changing the 2017 CPG recommendations that in-lab PSG remains the diagnostic testing approach of choice for these patients.

Future research should account for several considerations. First and foremost, consideration should be given to the different definitions used to determine hypopneas as these differences will significantly impact the accuracy of testing that does not measure sleep and arousals, and as a result may affect long-term health outcomes in patients with OSA not associated with significant hypoxemia. Future work should also focus on the ability to predict patients likely to have false-negative HSAT or overnight oximetry testing results that may warrant follow-up testing, and what the impact of missed diagnosis may have on outcomes. Certain patient populations are at risk for complicated breathing disorders, including CSA and hypoventilation, and research is needed to advance technology for alternative testing devices that should take these factors into consideration. And finally, there is a lack of knowledge regarding the financial analysis of different approaches to diagnosing OSA in these specific high-risk patient populations.

VII |. COMORBIDITIES ASSOCIATED WITH OSA

VII.A |. Comorbidities Associated with OSA: Cardiovascular Disease

OSA is highly prevalent in the general population and in individuals with CV disease. OSA is characterized by repeated episodes of UA collapse during sleep, resulting in intermittent hypoxemia and arousals.^632–635 The accompanying increase in sympathetic activity, inflammation, endothelial dysfunction, and elevated BP is associated with increased risk for CV morbidity and mortality.^{634,636–638}

Many observational studies have demonstrated an association between OSA and incident CV disease, such as hypertension, AF, CAD, CHF, myocardial infarction (MI), stroke, and all-cause and CV mortality.

VII.A.1 |. Cardiovascular and all-cause mortality

A recent MA by Fu et al. examined the relationship between OSA and all-cause and CV mortality in 27 cohort studies. OSA increased risk for both all-cause mortality (HR 1.86, 95% CI = 1.81–1.91) and CV mortality (HR 2.36, 95% CI = 1.22–4.57). However, when OSA was stratified by severity, there was no significant association between mild OSA or moderate OSA and all-cause and CV mortality. Only severe OSA was an independent risk factor for both all-cause and CV mortality.⁶³⁹ On cluster analysis of the SantOSA cohort with moderate or severe OSA, the excessive sleepiness subtype was associated with an increased risk of incident CV mortality.⁶⁴⁰

VII.A.2 |. Cardiovascular disease

Compared to the general population, OSA is highly prevalent (38%–65%) in patients with CAD.⁶³⁵ Historically, evidence supports a significant association between OSA and CAD.

A large, population-based study was conducted to examine the cross-sectional relationship between OSA and cardiovascular disease (CVD). A cohort of 6424 subjects aged 40 and older from the SHHS underwent an unattended polysomnogram and categorized into quartiles of AHI. The first, second, third, and fourth AHI quartile ranges were 0–1.3, 1.4–4.4, 4.5–11.0, >11.0, respectively. Of the 6424 subjects, a total of 1023 subjects reported at least one CAD outcome, as defined as MI, angina, coronary revascularization procedure, HF, or stroke. Compared to the first and lowest quartile of AHI, the OR of prevalent CVD for the second, third and fourth highest quartiles were 0.98 (95% CI = 0.77–1.24), 1.28 (95% CI = 1.02–1.61), and 1.42 (95% CI = 1.13–1.78), respectively, indicating that severe OSA is independently associated with CAD events with a dose response relationship after adjusting for demographic variables, tobacco use, cholesterol, and hypertension variables.⁶³⁸

A secondary analysis of the SHHS cohort examined the prospective association of OSA and CAD. A subset from the SHHS cohort without CAD and HF at enrollment, and adequate data for analysis (4422 subjects, 56.4% women), were followed for a median of 8.7 years for incident CV disease. After adjusting for other risk factors, OSA was significantly associated and predicted incident CAD events, defined as MI, revascularization procedure, or coronary heart disease related death, however this was only observed in men aged 70 or younger (HR 1.10, 95% CI = 1.00–1.21 per 10-unit increase in AHI). Severe OSA (as defined by AHI ≥ 30 events/h) conferred a stronger increased risk of developing symptomatic CAD (HR 1.68, 95% CI = 1.02–2.76) again seen only in men aged 70 or younger.⁶⁴¹ However, another analysis of the full SHHS cohort has shown severe OSA to be an independent predictor of death, and in particular death related to CAD. While this relationship was again strongest in men under 70 years of age (adjusted HR 2.09; 95% CI: 1.31–3.33), it was nonetheless seen across all patients with severe OSA in the study population (adjusted HR 1.46; 95% CI: 1.14–1.86).⁶⁴²

In addition to evidence supporting an independent link between OSA and incident CAD, there is evidence that OSA may be associated with recurrence of CAD events including restenosis after percutaneous coronary dilation and death among individuals with CAD and OSA.⁶⁴³ Nakashima et al. conducted a prospective cohort study to determine whether moderate to severe OSA was associated with an increased risk of adverse CV events in patients who underwent primary percutaneous coronary intervention (PCI). The cohort was comprised of 272 patients who were admitted to the Nagasaki Citizens Hospital with acute MI. Patients with moderate to severe OSA had independently and significantly increased acute coronary syndrome recurrence and major adverse cardiac events compared to patients with mild or no OSA.⁶⁴⁴

Participants in the other sentinel community-based prospective observational study of OSA, the Wisconsin sleep cohort, were significantly younger than those in the SHHS. This may provide at least a partial explanation for the much stronger association seen between OSA and CAD in the former, wherein an AHI of ≥30 conferred a greater than two-fold risk of incident CAD and HF events (adjusted HR 2.63; 95% CI: 1.13–6.10) over a period of approximately 18,000 person years.⁶⁴⁵ The fully adjusted model for incident CAD only was not statistically significant (HR 2.4, CI 0.99–6.0).

The inevitable meta-analyses suggest that OSA confers an increased risk of incident clinically overt CAD in men, with an apparent weaker relationship between OSA and CAD in women.^633,635,639 An SR of untreated OSA and long-term adverse outcomes suggested that any negative effect of significant OSA on CV events was attenuated by female gender, age, a lack of daytime sleepiness, and obesity.⁶³⁷ Overall, there is relatively strong, but not uniform, evidence from clinical and population studies to support an important role for OSA in promoting the evolution of CAD, particularly in younger male patients. (Table VII.A.2)

TABLE VII.A.2.

Evidence on coronary artery disease and OSA

Study	Year	LOE	Study design	Study groups	Clinical endp oints	Conclusion
Bauters et al.632	2015	2b	Review	Review to determine the association between OSA and cardiovascular disease and impact of CPAP treatment on cardiovascular risk factors and outcomes.	Hypertension, stroke, ischemic heart disease, heart failure, atrial fibrillation, and cardiovascular mortality.	Moderate to severe OSA is independently associated with various forms of CV disease.
Bouzerda643	2018	2a	Systematic review	Review of pathophysiological mechanisms between OSA and cardiovascular disease and to determine prevalence of OSA in general population.	Systemic and pulmonary arterial hypertension, heart rhythm disorders, coronary heart disease, heart failure, and stroke.	Recommend screening for clinical symptoms of OSA in patients with cardiovascular disease.
Campos-Rodriguez et al.646	2012	2b	Prospective cohort study	Cohort of 116 women consecutively referred to two sleep clinics in Spain for suspected OSA between 1998 and 2007 categorized into five groups: 1) control without OSA 2) CPAP treated group with mild-moderate OSA 3) CPAP treated group with severe OSA 4) untreated group with mild-moderate OSA 5) untreated group with severe OSA	Cardiovascular mortality, including death from stroke, myocardial infarction, heart failure, or arrhythmia.	Severe OSA is associated with cardiovascular mortality in women. Treatment with CPAP may reduce this risk.
Catalan-Serra et al.647	2019	2b	Prospective cohort study	Cohort of 1005 consecutive patients referred to sleep units at two Spanish University hospitals who were ≥65 years. They were categorized into four groups based on AHI values from sleep studies and CPAP adherence: 1. control group 2. untreated mild-moderate OSA 3. untreated severe OSA 4. CPAP-treated OSA	Incidence of coronary heart disease and incidence of stroke after adjusting for OSA group, age, BMI, HTN, sex, smoking, ESS, and afib.	The incidence of stroke, but not coronary heart disease, is increased in elderly patients with untreated severe OSA.
Drager et al.636	2015	2b	Review	Review of pathogenesis and association between OSA and cardiovascular disease.	Cardiovascular disease including hypertension, autonomic dysfunction, insulin resistance, vascular dysfunction, lipid metabolism impairment, atherosclerosis.	OSA activates multiple intermediate pathways that lead to cardiovascular disease.
Fu et al.639	2016	2b	Meta-analysis	27 cohort studies with 3,162,083 participants who were diagnosed with OSA by PSG.	All-cause or cardiovascular mortality.	Severe OSA is an independent risk factor for all-cause and cardiovascular mortality. CPAP use significantly reduced both all-cause and cardiovascular mortality in patients with OSA.
Ge et al.633	2013	2b	Meta-analysis	6 cohort studies with 9165 adults who had been diagnosed with OSA, of any severity, confirmed by using a standardized polysomnography.	Cardiovascular mortality (defined as death from stroke, heart failure, myocardial infarction, or arrhythmia), and all-cause mortality.	Severe OSA is a strong independent predictor for future cardiovascular and all-cause mortality. CPAP treatment was associated with reduced risk of cardiovascular mortality.
Gonzanga et al.648	2015	2b	Review	Impact of OSA on cardiovascular disease and cardiovascular benefits of CPAP treatment.	Prevalence of OSA in patients with hypertension, coronary artery disease, atrial fibrillation, stroke, and heart failure.	OSA is highly prevalent in the general population and those with cardiovascular disease. Screening and treatment are needed to decrease cardiovascular risk.
Gottlieb et al.641	2010	2b	Prospective cohort study	Cohort of 4422 men and women who are ≥40 years old and free of coronary heart disease and heart failure followed for incident CHD and HF from the Sleep Heart Health Study.	Incident CHD (myocardial infarction, CHD death, or coronary revascularization procedure) and incident heart failure.	OSA is associated with increased risk of CV outcomes in community-dwelling middle-aged and older men.
Hla et al.645	2015	2b	Prospective cohort study	1280 men and women from the Wisconsin Sleep Cohort who were free of CHD or HF at baseline and followed for 24 years.	Incident CHD or heart failure.	Untreated severe OSA was associated with incident coronary heart disease or heart failure.
Javaheri et al.634	2016	2b	Prospective cohort study	Cohort of 2865 older men from Osteoporotic Fractures in Men Study.	Incidence of heart failure.	Older men with elevated central sleep apnea index or Cheyne-Stokes breathing had an increased risk for incident HF. However, OSA was not significantly associated with incident HF.
Kasai et al.649	2010	2b	Review	Review of association between OSA and heart failure.	Cardiovascular function and autonomic function.	OSA has adverse cardiovascular effects and is associated with reduced survival in patients with HF.
Loke et al.650	2012	2a	Systematic Review	Association between OSA and incident cardiovascular events.	Stroke incidence and cardiovascular mortality.	OSA is an independent risk factor for stroke and cardiovascular mortality.
Martinez-Garcia et al.651	2012	2b	Prospective Cohort Study	Cohort of 943 elderly patients (≥65 years) with mild/mod OSA or severe OSA between 1998 and 2007 categorized into four groups: 1) control group (AHI<15) 2) untreated mild/mod OSA without CPAP 3) untreated severe OSA 4) OSA treated with CPAP	Cardiovascular mortality defined as death from stroke, heart failure, or myocardial infarction.	Untreated severe OSA is significantly associated with cardiovascular mortality in the elderly.
Nakashima et al.644	2015	2b	Prospective cohort study	Cohort of 272 patients with acute myocardial infarction who underwent primary percutaneous coronary intervention within 12 h of onset.	MACE (cardiovascular mortality, acute coronary syndrome recurrence, and readmission for heart failure).	Pt with untreated moderate–severe OSA had an increased the risk of acute coronary syndrome recurrence. Moderate–severe OSA also increased risk for MI related percutaneous coronary intervention progressive lesions.
Punjabi et al.642	2009	2b	Prospective cohort study	Cohort of 6441 men and women from the Sleep Heart Health Study.	All-cause and cardiovascular mortality.	OSA is independently and significantly associated with all-cause and cardiovascular mortality with a stronger association in men 40–70 years old with severe OSA.
Rosen et al.637	2014	2b	Review	Review of the pathophysiology of OSA and heart failure and review of treatment options for OSA in patients with heart failure.	Incident heart failure and mortality.	Untreated OSA is an independent risk factor for increased mortality in heart failure patients.
Shah et al.652	2010	2b	Prospective cohort study	Cohort of 1436 patients ≥50 years of age who were referred during 1997–2001 to the Yale Center for Sleep Medicine for suspected sleep disordered breathing.	Cardiovascular outcomes myocardial infarction, coronary artery revascularization procedures, or cardiovascular mortality.	OSA increases the risk of coronary events or cardiovascular mortality.
Shahar et al.638	2001	2c	Cross-sectional study	Cohort of 6424 participants from the Sleep Heart Health Study.	Self-reported cardiovascular disease outcomes defined as MI, angina, coronary revascularization procedure, heart failure, or stroke.	OSA is associated CVD outcomes, but more strongly associated with self-reported heart failure and stroke than coronary heart disease.
Wang et al.635	2018	2a	Systematic review and meta-analysis	Review of nine studies (two RCTs and seven observational studies) for 1430 patients with CAD and CVD.	Adverse CVD event (MACE) -all-cause or cardiovascular death, myocardial infarction, stroke, repeat revascularization, or hospitalization for heart failure.	CPAP may prevent incident cardiovascular events in patients with CAD and OSA. However, this was only shown in observational studies, not in RCTs.
Yeboah et al.653	2011	3b	Nested case–control study	5338 men and women from MESA study cohort.	Incident cardiovascular events as defined by myocardial infarction, angina, resuscitated cardiac arrest, stroke, stroke mortality, CHD mortality, or other CVD death as defined by the MESA protocol.	OSA, but not habitual snoring, was associated with incident CV events and all-cause mortality in adults without CVD.
Yu et al.654	2017	1b	Systematic review and meta-analysis	Review of 10 RCTs with 7266 patients to determine association of CPAP compared with standard care or sham PAP among adults with OSA or central sleep apnea.	Acute coronary syndrome events, stroke, or vascular events or death, major adverse cardiovascular events.	CPAP use was not significantly associated with reduced risk of cardiovascular outcomes or mortality for OSA patients.

VII.A.3 |. Myocardial ischemia

OSA has been associated with numerous CV conditions, including CAD and myocardial ischemia. A number of possible pathophysiologic mechanisms have been implicated including sympathetic nervous system hyperactivity, hypertension, endothelial dysfunction, metabolic dysregulation, insulin resistance, and hypercoagulable state.^{641,655–663} In addition, OSA results in repetitive hypoxia and re-oxygenation and this has been associated with increased oxidative stress and systemic inflammation. These mechanisms may contribute to the increased risk of atherosclerosis and myocardial ischemia in patients with OSA.

OSA has been associated with coronary artery calcification,⁶⁶⁴ plaque instability,⁶⁶⁵ and vulnerability.⁶⁴⁴ During obstructive apneas, increased adrenergic tone and hypoxemia may increase the risk of myocardial ischemia.^666,667 Interestingly, a temporal relationship between hypoxia and the development of ST changes and chest pain has been reported by Franklin et al.⁶⁶⁸ The severity of hypoxemia also appears to be a determinant of ST depression during sleep.⁶⁶⁹

Shah et al.⁶⁵² have reported that in an observational cohort of over 1400 patients, OSA was associated with a two-fold increased risk of CV events or death even after adjustment for traditional risk factors. This suggests that OSA might independently increase the risk of coronary events. In patients with ST-segment elevation myocardial infarction (STEMI), the prevalence of undiagnosed OSA is almost 40%.⁶⁷⁰ In addition, the onset of MI is more likely to be during the nighttime.⁶⁶⁶ Patients who have had prior STEMI are more likely than the general population to have OSA, and have worse event free survival compared to STEMI patients without OSA.^666,670

Major adverse CV events (defined as a composite of CV mortality, non-fatal MI, non-fatal stroke, and unplanned revascularization) after PCI also appear to be worse in patients with OSA.^671,672 It is controversial however whether treatment of OSA with CPAP may reduce the risk of repeat revascularization after PCI.^673,674 It has been reported that CPAP use of >4 versus <4 h per night may be associated with a significant reduction of CV risk.⁶⁷³ (Table VII.A.3)

TABLE VII.A.3.

Association between myocardial ischemia, coronary artery disease, and OSA

Study	Year	LOE	Study design	Study groups	Clinical endpoints	Conclusion
Sorajia et al.664	2008	2b	Cross-sectional	202 consecutive patients with no history of coronary disease who underwent electron-beam CT within 3 years of polysomnography between March 1991 and December 2003 were included. OSA was defined by an apnea hypopnea index (AHI)>or =5 events/h, and patients were grouped by quartiles of AHI severity.	Evaluated the association between OSA and the presence of subclinical coronary disease assessed by coronary artery calcification (CAC).	In patients without clinical coronary disease, the presence and severity of OSA is independently associated with the presence and extent of CAC.
Franklin et al.668	1995	2b	Prospective cohort	10 consecutive men with severely disabling angina pectoris and nocturnal angina on at least two nights per week for at least 1 month.	Overnight recordings of nasal and oral airflow, abdominal and chest movements, oxygen saturation, body position, electrocardiogram, electro-oculograms, electroencephalograms, and submental electromyogram were registered.	Results show a relation between nocturnal angina pectoris and sleep apnea.
Mooe et al.669	2000	3b	Observational study	132 men and 94 women	Overnight sleep study and Holter recording to examine disordered breathing (oxyhemoglobin desaturations > or = 4% and apnea-hypopneas), heart rates, and ST-segment depressions (> or = 1 mm, > or = 1 min).	Episodes of nocturnal myocardial ischemia are common in patients with angina pectoris. However, a temporal relationship between sleep-disordered breathing and myocardial ischemia is present only in a minority of the patients, but occurs more frequently in men and in more severely disordered breathing.
Kuniyoshi et al.666	2008	2b	Prospective cohort	92 patients with MI for which the time of onset of chest pain was clearly identified. The presence of OSA was determined by overnight polysomnography.	Evaluate the day-night variation of acute myocardial infarction in patients with obstructive sleep.	The diurnal variation in the onset of MI in OSA patients is strikingly different from the diurnal variation in non-OSA patients. Patients with nocturnal onset of MI have a high likelihood of having OSA. These findings suggest that OSA may be a trigger for MI.
Xie et al.675	2016	2b	Prospective study	112 patients without a prior diagnosis of sleep apnea underwent comprehensive polysomnography within a median of 7 days after MI. Patients were followed up at 6-monthly intervals (±2 weeks) for 48 months.	Investigate the effects of OSA on prognosis after MI, and to determine which specific measures of OSA severity best predicted outcomes.	Nocturnal hypoxemia in OSA is an important predictor of poor prognosis for patients after MI.
Nakashima et al.644	2015	2b	Prospective cohort	Patients with acute MI and followed over time.	The main study outcome measures were cardiac death, recurrence of ACS, and re-admission for heart failure. Major adverse cardiac events (MACEs) were defined as composite end points of individual clinical outcomes.	Moderate-to-severe OSA increased the risk of ACS and the incidence of PCI for progressive lesions. Increased plaque vulnerability might be related to these clinical manifestations.
Shah et al.652	2010	3b	Observational cohort study	1436 enrolled patients, 1024 (71%) had an apnea hypopnea index > or = 5.	Study aimed to determine whether obstructive sleep apnea independently increases the risk of coronary events, including death from cardiovascular causes.	Obstructive sleep apnea increases the risk of coronary events or death from cardiovascular causes.
Lee et al.670	2011	2b	Prospective cohort	120 patients underwent an overnight sleep study during index admission for STEMI.	Determine the effect of severe obstructive sleep apnea (OSA) on long-term outcomes after myocardial infarction.	42% of the patients admitted with STEMI have undiagnosed severe OSA. Severe OSA carries a negative prognostic impact for this group of patients. It is associated with a lower event-free survival rate at 18-month follow-up.
Lee et al.671	2016	2b	Prospective study	Between December 2011 and April 2014, 1748 eligible patients were prospectively enrolled. The 1311 patients who completed a sleep study within 7 days of percutaneous coronary intervention formed the cohort for this analysis. Median follow-up of 1.9 years.	Hypothesized that OSA predicts subsequent major adverse cardiac and cerebrovascular events (MACCEs) in patients undergoing percutaneous coronary intervention.	OSA is independently associated with subsequent MACCEs in patients undergoing percutaneous coronary intervention.
Qu et al.672	2018	1a	Meta-analysis of prospective cohort studies	7 studies including 2465 patients.	Followed up with patients after PCI, and evaluated their overnight sleep patterns within 1 month for major adverse cardiac events (MACEs) as primary outcomes including cardiac death, non-fatal myocardial infarction (MI), and coronary revascularization and secondary outcomes including re-admission for heart failure and stroke.	In patients after PCI, OSA appears to increase the risk of cardiac death, non-fatal MI, and coronary revascularization.
Xie et al.676	2018	2b	Prospective study	112 post-myocardial infarction patients.	Investigate whether excessive daytime sleepiness (EDS) would be an independent prognostic factor after myocardial infarction.	EDS may be an independent prognostic factor of adverse outcome in post-myocardial infarction patients with moderate to severe sleep-disordered breathing.
Gottlieb et al.641	2010	1b	Prospective longitudinal epidemiological study	1927 men and 2495 women > or = 40 years of age and free of coronary heart disease and heart failure at the time of baseline polysomnography were followed up for a median of 8.7 years.	Objective of this study was to assess the relation of obstructive sleep apnea to incident coronary heart disease and heart failure in a general community sample of adult men and women.	Obstructive sleep apnea is associated with an increased risk of incident heart failure in community-dwelling middle-aged and older men; its association with incident coronary heart disease in this sample is equivocal.
Schwarz et al.663	2015	1b	Systematic review and meta-analysis	Systematic review and meta-analysis evaluating RCTs on the effect of CPAP on endothelial function in OSA, assessed by flow-mediated dilatation (FMD) and other validated techniques. Eight RCTs comparing the effects of therapeutic CPAP versus subtherapeutic CPAP (or no intervention) on endothelial function involving 245 OSA patients were included in the systematic review.	Assess the effect CPAP therapy on endothelial function in patients with OSA.	In patients with OSA, CPAP therapy improves endothelial function significantly and to a clinically important extent.
Phillips et al.661	2012	2b	Randomized crossover trial	28 patients received therapeutic or placebo CPAP, each for 2 months with a 1 month washout between treatments. After each treatment period, a 24 h coagulation study was conducted.	Plasminogen activator inhibitor-1 (PAI-1), D-dimer, fibrinogen, von Willebrand Factor (vWF), factor VIII (FVIII), factor VII (FVII), and factor V (FV) were determined at seven time points over the day and night.	CPAP may reduce cardiovascular in OSA, in part through reducing risk of thrombosis.
Milleron et al.658	2004	1a	Long-term prospective study	Studied 54 patients (mean age 57.3 ± 10.1 years) with both CAD (> or = 70% coronary artery stenosis) and OSA (apnea-hypopnoea index > or = 15). In 25 patients, OSA was treated with continuous positive airway pressure (n = 21) or upper airway surgery (n = 4); the remaining 29 patients declined treatment for their OSA.	A composite of cardiovascular death, acute coronary syndrome, hospitalization for heart failure, or need for coronary revascularization.	Treatment of OSA in CAD patients is associated with a decrease in the occurrence of new cardiovascular events, and an increase in the time to such events.
Wu et al.674	2015	2b	Prospective study	390 patients with OSA who had undergone PCI.	Impact of OSA treatment with CPAP on percutaneous coronary intervention (PCI) outcomes.	Untreated moderate–severe OSA was independently associated with a significant increased risk of repeat revascularization after PCI. CPAP treatment reduced this risk.
Peker et al.673	2016	1b	Randomized controlled trial	Consecutive patients with newly revascularized CAD and OSA (apnea hypopnea index ≥15/h) without daytime sleepiness (Epworth Sleepiness Scale score <10) were randomized to auto-titrating CPAP (n = 122) or no positive airway pressure (n = 122).	Determine the effects of CPAP on long-term adverse cardiovascular outcome risk in patients with CAD with nonsleepy OSA.	Routine prescription of CPAP to patients with CAD with nonsleepy OSA did not significantly reduce long-term adverse cardiovascular outcomes in the intention-to-treat population. There was a significant reduction after adjustment for baseline comorbidities and compliance with the treatment.

VII.A.4 |. Hypertension

OSA and hypertension (HTN) are highly prevalent conditions in the general population.^95,97,677 Both conditions are highly comorbid, with 50% of subjects with OSA noted to have HTN⁶⁷⁸ and 50% of those with HTN found to have OSA.⁶⁷⁹ While OSA has been associated with several adverse CV consequences, the evidence linking OSA and HTN is the most robust.^680,681 OSA and HTN have shared risk factors such as obesity,⁶⁸² but the presence of OSA has been found to be an independent risk factor for prevalent as well as incident pre-HTN and HTN.^{115,660,683,684} More recent evidence suggests that untreated rapid eye movement (REM)-related OSA, a subtype of OSA where SDB events are predominantly confined to REM sleep, is significantly associated with the development of HTN.⁶⁸⁵

There is a dose–-response relationship between OSA and HTN, in that greater severity of OSA appears to confer a higher risk of HTN.^115,660,684 Furthermore, the effects of OSA on BP seem to be more pronounced in subjects with subjective and objective daytime sleepiness^686,687; these individuals may have a greater degree of desaturation and higher diastolic BP following SDB events.⁶⁸⁸ Finally, the temporal distribution, night-to-night variability, and degree of desaturation and/or autonomic response associated with SDB events may ultimately affect an individual’s risk of developing HTN.⁶⁸¹

The prevalence of nocturnal non-dipping BP at night, an adverse CV prognostic risk factor, is high in subjects with OSA, in the range of 50%–80%.^689,690 While the severity of respiratory abnormalities appears to be associated with a nocturnal non-dipping BP pattern in younger individuals, in older subjects, the severity of sleep disruption seems to correlate with nocturnal non-dipping status.⁶⁹¹

Resistant HTN is defined by the use of three or more antihypertensive medications; refractory HTN refers to treatment with five or more medications.^692,693 OSA is a secondary cause of HTN,⁶⁹⁴ and is particularly common in cases of resistant and refractory HTN.^695–697 Over 50% of subjects with resistant HTN have underlying OSA.^698–700 Furthermore, those with severe untreated OSA appear to have a significantly higher risk of having resistant HTN when compared to subjects with moderate OSA.⁷⁰¹ A recent study demonstrated a two-fold increase in risk of resistant HTN in African-Americans with severe OSA, and it is notable that the burden of undiagnosed OSA is high in this population.⁷⁰² (Table VII.A.4)

TABLE VII.A.4.

Association between OSA and refractory hypertension

Study	Year	LOE	Study design	Study groups	Clinical end-point	Conclusion
Pratt-Ubunama et al.699	2007	3b	Prospective, single-center, case–control study; all subjects underwent PSG (n = 46).	1) RHTN (n = 17). 2) Controls with suspected OSA (n = 29).	1. AHI. 2. PAC, renin concentration.	OSA was common in subjects with RHTN (85%) and was more common and severe in men vs. women. PAC (but not renin concentration) correlated with AHI.
Pedrosa et al.679	2011	3b	Prospective cohort, study conducted at two outpatient HTN units (n = 125).	Consecutive patients with RHTN.	Evaluation for secondary caiuses of HTN.	Moderate–severe OSA was the most common condition associated with RHTN, seen in 64% of subjects.
Walia et al.703	2014	3b	Secondary analysis of baseline data from the Heart Biomarker Evaluation in Apnea Treatment (HeartBEAT) RCT (n = 284).	Subjects with HTN; 64% had OSA, 10% had RHTN.	Association between RHTN and OSA.	RHTN was more prevalent in subjects with severe vs. moderate OSA (58.3% vs. 28.6%, p = 0.01) who were prescribed antihypertensives; those with severe OSA had four-fold higher adjusted odds of RHTN (OR 4.1, 95% CI: 1.7–10.2).
Johnson et al.702	2018	3b	Analysis of data from the Jackson Heart Study, community cohort (n = 913)	Subjects with HTN (n = 664); 25.7% with OSA (untreated in the vast majority), 14% with RHTN; majority female.	Association between OSA and RHTN in blacks.	Subjects with moderate–severe OSA had a two-fold higher odds of RHTN (OR 2.0; 95% CI: 1.14–3.67) after accounting for confounders. OSA and oxyhemoglobin saturation <90% were not associated with uncontrolled BP.
Martinez-Garcia et al.696	2018	3b	Cross-sectional study (n = 229)	Consecutive subjects with RHTN, 18% met criteria for refractory HTN.	Association between refractory HTN and presence/severity of OSA.	Subjects with refractory HTN had a two-fold higher risk of OSA (prevalence of moderate OSA was 95% and severe OSA was 64%) and greater OSA severity (AHI 41.8 vs. 33.8; p = 0.026) compared to those with RHTN.
Sapina-Beltran et al.700	2019	3b	Multicenter cohort study (n = 284)	Consecutive subjects with RHTN.	Prevalence of OSA in subjects with RHTN and association of OSA with BP control.	83.5% had OSA, 25.7% moderate, and 26.1% severe OSA. Those with severe OSA had higher BP than those with mild/no OSA, with a greater effect noted on nighttime BP vs. those with no OSA. The prevalence of severe OSA was higher in those with uncontrolled BP (not statistically significant).

Abbreviations: AHI, apnea hypopnea index, events/hour; BP, blood pressure; CI, confidence intervals; DBP, diastolic blood pressure; OSA, obstructive sleep apnea; PAC, plasma aldosterone concentration; PSG, polysomnography; RCT, randomized controlled trial; RHTN, resistant hypertension; SBP, systolic blood pressure.

VII.A.5 |. Atrial fibrillation

Patients with AF have a high prevalence of OSA. Direct comparison of studies regarding the concurrence of OSA and AF is complicated by varying respiratory event definitions and by inclusion in some studies of both OSA and CSA. Despite these limitations, studies that systematically evaluated AF patients using home or laboratory sleep testing report a prevalence of OSA (AHI ≥ 5) of 43%–85%, with prevalence >75% in most studies.^704–710 Moderate or severe OSA (AHI ≥ 15) is reported in 20%–62%.^{330,704,706,710–715} When a comparison group of patients without AF was included, the prevalence of OSA was higher in those with AF than in controls in most^707,715,716 but not all⁷¹⁷ studies. Community-based cohort studies support an independent association of OSA with the development of AF. In the Multi-Ethnic Study of Atherosclerosis, the prevalence of AF increased from 4.0% in those without OSA to 7.5% in those with severe OSA.⁷¹⁸ In two cohorts, the presence of AF was identified from the polysomnographic recording. In the SHHS, compared to those with AHI <5, participants with AHI ≥ 30 had an age-, sex-, BMI-, and prevalent CV disease-adjusted odds ratio (aOR) of 4.0 (95% CI 1.0–15.7) for the presence of AF,⁷¹⁹ while in the Outcomes of Sleep Disorders in Older Men Study there was a dose-dependent association of OSA with prevalent AF.⁷²⁰

Longitudinal studies have yielded equivocal results regarding the association of OSA with incident AF. While studies using administrative claims data⁷²¹ or patientreported diagnosis of OSA⁷²² suggest an association of OSA with incident AF, these studies have a high risk of bias. Three large retrospective studies of incident AF in cohorts of patients referred for diagnostic PSG have been reported. In one, patients with AHI ≥ 5 were twice as likely to develop AF as those with AHI < 5, an effect that was limited to those under age 65; however, adjusting for age, sex, BMI, and prevalent CV disease, mean nocturnal SaO₂ but not AHI was an independent predictor of incident AF.⁴¹⁴ In another, the incidence of hospitalized AF increased progressively with increasing severity of OSA, although after multivariate adjustment that included BMI, neither AHI nor time at saturation <90% was a significant predictor of incident AF, except in those with >30% of sleep time at saturation <90%.⁷²³ It is unclear whether these measures of SaO₂ reflect OSA or other conditions, such as reduced pulmonary or cardiac function, that might cause AF. In the third, AHI ≥ 5 was associated with an aOR of 1.55 (95% CI 1.21–2.00) for incident hospitalized AF after extensive covariate adjustment, with AF incidence increasing with greater severity of OSA.⁷²⁴ In contrast, two community-based cohort studies, each of which had shown a cross-sectional association of OSA with prevalent AF, found that CSA but not OSA was independently associated with increased incidence of AF.^725,726

Data are more consistent regarding the association of OSA with recurrent AF. In six studies that systematically assessed the presence of OSA prior to treatment of AF, OSA was associated with an approximately two- to three-fold increased risk of recurrent AF following electrical cardioversion⁷¹⁴ or pulmonary vein isolation (catheter ablation) procedures.^{708,709,711,713,727} A similar increased risk of AF recurrence was reported in several studies that did not systematically screen for OSA, but compared the risk of AF recurrence following ablation procedures in patients with a documented prior diagnosis of OSA to those without known OSA.^728–730 The presence of severe OSA also predicts failure of antiarrhythmic drugs to suppress AF.⁷³¹ These studies generally included a mix of patients with paroxysmal and persistent AF, and the finding of increased risk of recurrence in those with OSA does not appear to be limited to either group.

Putative pathophysiological mechanisms linking OSA to AF include both chronic cardiac structural changes and acute arrhythmogenic effects of obstructive events, and has been recently reviewed.⁷³² Chronic changes include increased atrial dimension and slowed atrial conduction, while acute triggers to AF likely include atrial distension due to intrathoracic pressure swings along with hypoxemia, hypercapnia, and the associated acute elevation of sympathetic nervous system activity. In the SHHS, paroxysmal AF events were markedly more likely to occur during the 90 s following an obstructive apnea or hypopnea than during periods without obstructive events (OR 17.9, 95% CI 2.2–144.2).⁷³³ Similarly, in patients with paroxysmal AF, night-to-night variation in OSA severity predicts changes in AF burden.⁷³⁴ (Table VII.A.5)

TABLE VII.A.5.

Association between atrial fibrillation and OSA

Study	Year	LOE	Study design	Study groups	Clinical endpoints	Conclusion
Gami et al.716	2004	3b	Case–control study	1) 151 consecutive patients undergoing electrocardioversion for AF 2) 312 consecutive patients without AF referred to a general cardiology practice	Presence of OSA identified based on a Berlin Questionnaire score of 2–3	Adjusted OR of 2.19 (95% CI 1.40–3.42) for the association of AF and OSA
Porthan et al.717	2004	4	Case–control study	1) 59 patients with AF identified from hospital records, free of known causes of AF 2) 56 age- and sex-matched controls from general population registry, free of AF or known causes of AF	Presence of “sleep apnea syndrome” based on AHI from in-lab cardiorespiratory polygraphy and sleep apnea symptoms	No significant difference between groups in prevalence of sleep apnea syndrome (32% in AF versus 29% in control)
Mehra et al.719	2006	3b	Nested case–control study within a community-based cohort study	Participants in the Sleep Heart Health Study 1) 228 subjects with AHI ≥30 on home polysomnography 2) 338 subjects with AHI <5, frequency-matched on age, sex, race/ethnicity, BMI	Presence of arrhythmias on bipolar lead I ECG recorded on home polysomnography	Atrial fibrillation was present in 4.8% of severe OSA group and 0.9% of controls, with adjusted OR 4.02 (95% CI 1.03–15.74) for the association of AF and severe OSA
Stevenson et al.715	2008	3b	Case–control study	1) 90 patients with paroxysmal or persistent AF 2) 45 patients referred to the same tertiary care center without AF, frequency match for age, and sex	AHI >15 on home polysomnography	Mean AHI was higher in the AF group than the control group (23.2 [SD 19.3 vs. 14.7 [SD 12.4]), with adjusted OR 3.04 (95% CI1.24–7.46) for the association of AF and OSA
Braga et al.707	2009	3b	Case–control study	1) 57 consecutive patients with chronic persistent AF in Sao Paulo, Brazil 2) 32 age-, sex-, and BMI-similar subjects from the Sao Paolo general population participating in the Epidemiologic Sleep Study (EPISONO)	From in-lab polysomnography: 1) AHI 2) Sleep time with SaO2 <90%	No significant difference between groups in mean AHI (24.3 [SD 16.5] vs. 19.1 [SD 15.3]), but higher prevalence of AHI ≥10 (81.6% vs. 60%, p = 0.03) and longer time with SaO2 <90% in the AF group
Bitter et al.706	2009	4	Case series	150 patients with persistent AF and normal left ventricular ejection fraction	AHI from cardiorespiratory polygraphy, classified as central or obstructive based on the predominant event type	OSA was present in 42.7% (18% mild, 12% moderate, 12.7% severe); an additional 31.3% had central sleep apnea
Mehra et al.720	2009	3b	Community-based cross-sectional cohort study	2911 participants in the Outcomes of Sleep Disorders in Older Men Study	Association of nocturnal arrhythmias present on home polysomnography with total AHI, obstructive AHI (obstructive apneas plus all hypopneas), and central apnea index	There was a progressive increase in prevalence of AF with increasing quartile of total AHI, from 3.2% to 7.3%. The adjusted OR in the highest AHI quartile was 2.15 (95% CI 1.19–3.89). This effect was stronger for the central apnea index, and was non-significant for the obstructive AHI in adjusted analyses
Pathak et al.330	2014	4	Case series	149 patients referred for pulmonary vein isolation procedure for AF	Presence of severe OSA (AHI ≥30) on in-lab polysomnography	86 of 149 patients (57.7%) had severe OSA
Kwon et al.718	2015	3b	Community-based cross-sectional cohort study	2048 participants in the Multi-Ethnic Study of Atherosclerosis Sleep Study	Association of AHI from in-home polysomnography at MESA exam 5 with AF based on International Classification of Disease codes or ECG recordings over the preceding approximately 10 years (n = 92) or present on the polysomnogram ECG (an additional n = 8)	There was an increase in AF prevalence from 4.0% in those with none to mild OSA, 6.0% in those with moderate OSA, and 7.5% in those with severe OSA. The adjusted OR was 1.22 (95% CI 0.99–1.49) for each 1 SD increase in AHI.
Abuammar et al.704	2018	4	Case series	100 consecutive patients with AF without prior diagnosis of OSA recruited from arrhythmia clinics	Presence and severity of OSA from mean of 2 nights of home polysomnography	OSA was present in 85% of subjects (38% mild, 23% moderate, and 24% severe)
Traaen et al.710	2019	4	Case series	579 patients with paroxysmal AF admitted for pulmonary vein isolation procedure, without known OSA	Presence and severity of sleep apnea from mean of 2 nights of cardiorespiratory polygraphy	Sleep apnea was present in 83% of subjects (41% mild, 30% moderate, and 12% severe); this was OSA in 97.5% of those with sleep apnea, central sleep apnea in 2.5%
Gami et al.414	2007	2b	Retrospective cohort study	3542 adults without history of AF referred for diagnostic polysomnography 1) OSA defined as AHI ≥5 (n = 2626), with group mean AHI 36 (SD 32) 2) No OSA	Incidence of AF based on electronic medical record review, with mean follow-up of 4.7 years	AF occurred in 2.1% of those without OSA, 4.3% of those with OSA, with unadjusted HR 2.18 (95% CI 1.34–3.54). This effect was restricted to those under age 65. In multivariate analysis, the difference between awake and asleep oxygen saturation was a strong predictor of incident AF. It is not clear in this study whether measures of OSA per se are independently associated with incident AF
Cadby et al.724	2015	2b	Retrospective cohort study	6841 adults without history of AF referred for diagnostic polysomnography 1) OSA defined as AHI 0≥5 (n = 4352) 2) No OSA	Incidence of hospitalization for AF based on review of Western Australia Hospital Morbidity and Mortality Data, with median follow-up of 11.9 years	AF occurred in 8.6% of those with OSA and 3.3% of those without OSA, with adjusted HR 1.55 (95% CI 1.21–2.00) for the association of OSA with incident AF. There was a modest dose–response relationship, with adjusted HR increasing from 1.48 in those with mild OSA to 1.73 in those with severe OSA. Both AHI and time at SaO2 <90% were associated with incident AF
May et al.725	2016	2b	Community-based cohort study	Of 2316 participants in the Outcomes of Sleep Disorders in Older Men Study who were free of AF at baseline, 852 had a follow-up sleep study after approximately 6 years and 843 of these had outcome data and were analyzed	Association of AHI from baseline in-home polysomnography with incident adjudicated or self-reported AF (the proportion of adjudicated versus self-report AF is not stated)	AF occurred in 10.0% of those with AHI <15 and 14.2% of those with AHI ≥15 (adjusted OR 1.15 [95% CI 0.72–1.84]). There was no association of AF with obstructive AHI; however, a central apnea index ≥5 was present in 48 participants at baseline and the incidence of AF in this group was 22.9% (adjusted OR 2.34 [95% CI 1.14–4.77])
Tung et al.726	2017	2b	Community-based cohort study	Of 6441 participants in the Sleep Heart Health Study, 3346 had repeat polysomnography after approximately 5 years. Of these 388 were excluded due to restrictions on data use and 46 due to prevalent AF at baseline, yielding an analytic sample of 2912 participants.	Association of OSA and central sleep apnea with incident AF based on parent cohort adjudication or 12-lead ECG at the time of the second polysomnogram; median follow-up was 5.3 years; OSA was classified based on the obstructive AHI (OAHI, obstructive apneas plus all hypopneas per hour of sleep)	AF incidence increased with OSA severity from 9.6% of those with OAHI <5% to 16.4% with AHI ≥30; however, adjusting for age, sex, race and BMI there was no positive association of OSA with AF. Central apnea index ≥5 was present in 74 participants at baseline, and the incidence of AF in this group was 25.7% (adjusted OR 1.71, 95% CI 0.89–3.30, increasing to 3.00, 95% CI 1.40–6.44, with further adjustment for hypertension, diabetes, and cardiovascular disease)
Kendzerska et al.723	2018	2b	Retrospective cohort study	8256 patients without prior history of arrhythmia referred for diagnostic polysomnography 1) 2263 with AHI < 5 2) 2260 with 5 ≤ AHI < 15 3) 1823 with 15 ≤ AHI ≤ 30 4) 2263 with AHI > 30	Time to first hospitalization with a diagnosis of AF or atrial flutter, using Ontario Provincial Health Administrative Administrative Data, with median follow-up of 10 years	In unadjusted analyses, incident hospitalized AF increased with increasing severity of OSA based on either AHI or percent time at SaO2 < 90%; however, in multivariate models that included BMI these associations were not significant, with hazard ratios < 1.0 for each OSA severity compared to AHI <5. Only when comparing those with more than 30% of sleep time at SaO2 < 90% (n = 463) to those with less severe hypoxemia, was the adjusted HR significant (1.77, 95% CI 1.15–2.74).
Mazza et al.714	2009	1b	Clinic-based cohort study	158 consecutive patients admitted for electrocardioversion for AF; all had polysomnography the night prior to cardioversion 1) AHI ≥15 (n = 49) 2) AHI <15 (n = 109)	Recurrence of AF over 1-year follow-up	AF recurred in 69% of patients with AHI ≥15 and in 43% of those with AHI <15 (adjusted OR 3.04, 95% CI 1.45–6.36)
Matiello et al.727	2010	1b	Clinic-based cohort study	174 consecutive patients undergoing pulmonary vein isolation procedure for AF; all screened with Berlin Questionnaire, and the 51 with score of 2–3 underwent home cardiorespiratory polygraphy 1) Low risk Berlin Questionnaire or AHI <10 on polygraphy (n = 132) 2) AHI 10–<30 (n = 17) 3) AHI ≥30 (n = 25)	Recurrence of AF over mean follow-up of 17 (SD 11.5) months	Estimated 1-year recurrence free survival was 48.5% in the low-risk group, 30.4% in those with AHI 10–<30, and 14.3% in those with AHI ≥30. Adjusted OR for recurrence was 1.57 (95% CI 0.83–3.00) for AHI 10–<30 and 1.87 (95% CI 1.11–3.16) for AHI ≥30
Bitter et al.711	2012	1b	Clinic-based cohort study	82 consecutive patients undergoing pulmonary vein isolation procedure for AF, 75 evaluable; all underwent in-hospital cardiorespiratory polygraphy 1) AHI <15 (n = 53) 2) AHI ≥15 (n = 22, of which 15 were predominantly obstructive and seven were predominantly central)	Recurrence of AF over a median follow-up of 12 months	AF recurred in 45.5% of those with AHI ≥15 and 24.5% of those with AHI <15. The adjusted HR for AF recurrence in those with AHI ≥15 was 3.20 (95% CI 1.14–8.95) compared to those with AHI <15. Not analyzed separately by obstructive versus central pattern
Szymanski et al.709	2015	1b	Clinic-based cohort study	290 consecutive patients admitted for pulmonary vein isolation procedure for AF; in-hospital cardiorespiratory polygraphy the night prior to the procedure; polygraphy was inadequate in 22 patients and ablation was not performed in 14; 3 patients with central sleep apnea excluded 1) AHI <5 (n = 136) 2) AHI ≥5 (n = 115)	Recurrence of AF over a mean follow-up of 30 months	AF recurred in 65.2% of those with AHI ≥5 and 45.6% of those with AHI <5. AF recurrence increased progressively with more severe OSA, to 81.8% in those with AHI ≥30. The adjusted OR for AF recurrence in those with AHI ≥5 was 2.58 (95% CI 1.91–4.10)
Kawakami et al.713	2016	1b	Clinic-based cohort study	133 consecutive patients admitted for pulmonary vein isolation procedure for AF; in-hospital cardiorespiratory polygraphy the night prior to the procedure 1) AHI <15 (n = 69) 2) AHI 15–<30 (n = 39) 3) AHI ≥30 (n = 16)	Recurrence of AF over a mean follow-up of 13 (SD 7) months	AF recurred in 25% of those with AHI <15, 49% of those with AHI 15–<30, and 69% of those with AHI ≥30. However, in multivariate analysis, the association of AHI with recurrent AF was significant only in the subgroup with paroxysmal AF (adjusted HR 1.04, 95% CI 1.002–1.065)

VII.A.6 |. Congestive heart failure

HF is highly prevalent in the US with 6.5 million adults affected.⁷³⁵ HF is associated with significant mortality and is identified as the cause of one in every eight deaths in the US.⁷³⁶ Failure of the left ventricle is the most common subtype and can manifest as impairment of either systolic (hFrEF) or diastolic (HF with preserved ejection fraction [hFpEF]) function.⁷³⁷ OSA is more common in the hFrEF population, with prevalence estimates varying between 12% and 53%.⁷³⁸ A study of 700 patients with chronic stable hFrEF with ejection fraction (EF) ≤ 40% found a 36% prevalence of OSA, including 19% with severe OSA.⁷³⁹

In addition to shared risk factors like age and obesity, another proposed mechanism to explain the high prevalence of OSA in HF relates to nocturnal fluid shifting throughout the body. Excess interstitial fluid from the lower extremities redistributes during sleep when in the supine position. This rostral fluid shift correlates with an increase in neck circumference, decrease in diameter of the lumen of the pharynx, and increased propensity for obstruction as demonstrated by increase in AHI.⁷⁴⁰

Identification of OSA may be especially important in the setting of hFrEF, where it is an independent risk factor for mortality.^741,742 Although large confirmatory studies are lacking, there is interest in how OSA may adversely affect outcomes in patients with HF and a number of mechanisms have been proposed. HF is a state of sympathetic over-activity, so the autonomic imbalances associated with OSA in response to hypoxemia and repetitive arousals from sleep may generate concomitant physiologic stress. Additionally, large swings in intrathoracic pressure due to inspiratory effort against a closed UA exert transmural pressures across the heart and great vessels, leading to increasing afterload, reduced stroke volume and increased myocardial oxygen consumption.⁶⁸⁰

Considering these underlying pathophysiologic mechanisms, PAP therapy would be expected to have beneficial clinical outcomes in patients with HF. In addition to relieving obstruction of the UA in OSA and reducing work of breathing, PAP decreases venous return (preload), and may blunt sympathetic activity. These physiologic benefits, however, have not yet translated to improvements in clinical outcomes in patients with OSA and HF, which are limited to small, short-term RCTs. Some have shown improvements in physiologic parameters that are surrogates for CV outcomes, but results have not been consistent and it is uncertain if these improvements translate into a meaningful clinical benefit.

In small and select groups of patients with OSA and hFrEF, CPAP has been associated with reduced systolic BP^743,744 and heart rate,⁷⁴³ small but inconsistent changes in left ventricular EF (LVEF),^743–747 decreased overnight urinary norepinephrine excretion,⁷⁴⁶ improved QOL,^743,746 improved mean sleep-related SaO₂,⁷⁴⁵ and decreased pulmonary artery systolic pressure.⁷⁴⁷ There are a number of studies that have reported echocardiographic measurements of LVEF before and after CPAP. However, LVEF is but one assessment in the clinical diagnosis of hFrEF, and some would argue a minor determinant. Patil et al. performed an MA of five RCTs measuring LVEF by echocardiography or radionuclide ventriculography to compare the efficacy of CPAP against control conditions in patients with hFrEF. While some studies showed modest improvements in the absolute value of LVEF associated with CPAP, most reported changes (typically less than 5%) that would not be considered clinically significant.³⁹

Although most available studies focus on hFrEF, OSA appears to also be prevalent in hFpEF. Presence of OSA in hFpEF has been associated with increased brain natriuretic peptide (BNP) levels, which may serve as a marker for reduced cardiac function.⁷⁴⁸ Severity of OSA has also been associated with increasing severity of diastolic dysfunction.⁷⁴⁹ One small RCT reported improvement in diastolic function in hFpEF with nasal CPAP but there was no difference in heart rate, systolic function, ventricular structure, BP, or urinary catecholamines after 12 weeks of CPAP use.⁷⁵⁰

In terms of other treatment modalities, a recent randomized study of patients with hFrEF and OSA demonstrated benefits of exercise (three months of aerobic and strength training) as a stand-alone intervention and as an adjunctive therapy to CPAP. AHI decreased significantly in the exercise alone group (28 ± 17 to 18±12, p < 0.007) and the exercise + CPAP group (25 ± 15 to 10 ± 16, p < 0.007) compared to controls. Both exercise and CPAP improved NYHA functional class and daytime sleepiness as measured by the ESS. QOL improvements were maximal in the exercise groups.⁷⁵¹

Prospective data analyzing the association between OSA and HF and high-quality data regarding benefits of treatment in this population are lacking (Table VII.A.6a).^654,752 Several societies have made recommendations regarding the testing and treatment of those with hFrEF and OSA but these recommendations are largely based on low-quality evidence (Table VII.A.6b). Current evidence does not support the routine treatment of non-symptomatic OSA with CPAP in patients with hFrEF as a means to improve CV outcomes. Those with symptoms of OSA, such as excessive daytime sleepiness, should be offered treatment, in accordance with joint guidelines from national HF societies and the AASM.⁷⁵³ (Table VII.A.6a and VII.A.6b)

TABLE VII.A.6a.

Association between OSA and heart failure

Study	Year	LOE	Study design	Study group	Relevant clinical endpoint	Conclusion
Gottlieb et al.641	2010	2b	Prospective cohort 4422 subjects Followed for a median of 8.7 years	Participants had OSA on PSG and no coronary heart disease or heart failure at baseline	Incident CHF CHF-free survival	Incidence of CHF increased with increasing severity of OSA After adjustment for age, race, smoking, and BMI, AHI was associated with CHF in men but not women
Hla et al.645	2015	2b	Prospective cohort of people in South-Central Wisconsin 1546 participants	Participants had OSA on PSG and no documented CHD or CHF at baseline CHD and CHF was self-reported	Composite outcome of incident CHD or heart failure	After adjustment for age, sex, smoking, and BMI, OSA was associated with increased incidence of CHD or heart failure Association of OSA with incident CHD or CHF was attenuated when participants who reported using CPAP were included in the analysis
Kasai et al.754	2008	2b	Prospective cohort in Tokyo, Japan 88 patients followed for mean of 25.3 ± 15.3 months	Moderate to severe OSA with AHI≥15 LVEF<50% and NYHA class II or greater symptoms 65 were treated with CPAP and 23 were untreated	Event-free survival	Cumulative event-free survival was higher in patients w/OSA on CPAP vs. patients w/OSA not on CPAP Cumulative event-free survival was significantly lower in patients with poor compliance than in those with good compliance
Wang et al.742	2007	2b	Prospective, single-center, observational cohort Patients referred to Mount Sinai Hospital in Toronto 218 subjects were followed for a mean of 2.9 ± 2.2 years 164 had complete data	Heart failure for at least 6 months LVEF ≤ 45% at rest NYHA class II-IV dyspnea All underwent PSG and were divided into categories: AHI < 15 and AHI ≥ 15 Those with central sleep apnea were excluded Patients with OSA were divided into “treated” and “untreated” groups	Cumulative rate of death	Mortality was higher in those with untreated OSA compared to treated OSA after adjusting for LVEF, NYHA class, and age
Kaneko et al.743	2003	1b	RCT 24 patients	Heart failure for at least 6 months LVEF≤45% at rest NYHA class II-IV dyspnea No exacerbations within 3 months Optimal medical therapy OSA with AHI of ≥20 with >50% obstructive events Half of the patients were treated with CPAP, the other half were treated only with medical therapy	Cardiovascular physiologic parameters as measured by trans-thoracic echocardiography: LVEF, LVEDV, LVESV	Nocturnal CPAP improves daytime left ventricular systolic function in patients with heart failure and OSA

Abbreviations: CHD, coronary heart disease; CHF, congestive heart failure; LVEDV, left ventricular end diastolic volume; LVEF, left ventricular ejection fraction; LVESV, left ventricular end systolic volume; OSA, obstructive sleep apnea; PSG, polysomnogram.

TABLE VII.A.6b.

Guidelines for testing and treatment of OSA in hFrEF

Author	Question	Statement	Class	Level of evidence
AHA/ACC/HFSA	Should people with	A formal sleep assessment	IIa (moderate	C-LD (Limited data,
Guidelines on Management of Heart Failure753	hFrEF be tested for OSA?	is reasonable for people with suspicion of SDB or EDS	recommendation)	randomized or non-randomized observational, or registry studies with limitations of design or execution, meta-analyses of such studies, physiological, or mechanistic studies in human subjects)
	Should Patients with hFrEF and OSA be treated with CPAP?	In patients with CVD and OSA, CPAP may be reasonable to improve sleep quality and daytime sleepiness	IIb (weak recommendation)	B-R (moderate quality evidence from one or more RCTs/meta-analysis of moderate quality RCTs)
AASM Guidelines for Evaluation and management of OSA513; AASM Clinical Practice Guideline for OSA572; AASM guidelines for treatment of adult OSA with PAP39	Who should be evaluated for OSA?	Patients with comorbidities considered “high risk” for OSA including HF	N/A	N/A
	Where should patients with suspected OSA and significant comorbidities (including HF) be tested?	Consider initiating PAP using an in-lab strategy	Strong Recommendation	N/A
	Should PAP be recommended or withheld in non-sleepy OSA patients to reduce cardiovascular events or mortality?	Insufficient evidence	No recommendation	N/A
	Initiation of PAP therapy	In patients with significant comorbidities, PAP initiation using an in-lab strategy should be considered	N/A	N/A

Abbreviations: AASM, American Academy of Sleep Medicine; ACC, American College of Cardiology; AHA, American Heart Association; ASV, adaptive servoventilation; CSA, central sleep apnea; CVD, cardiovascular disease; EDS, excessive daytime sleepiness; HFSA, Heart Failure Society of America; NYHA, New York Heart Association; PAP, positive airway pressure; SDB, sleep-disordered breathing.

VII.A.7 |. Cerebrovascular disease

Cerebrovascular accident (CVA) or stroke is one of the leading causes of death and disability globally.^755,756 OSA is highly prevalent (55%) among the stroke population and significantly increases the risk for incident ischemic stroke.⁷⁵⁶

Johnson and Johnson conducted an MA of 29 studies consisting of 2343 patients with ischemic or hemorrhagic stroke and TIA patients. They found OSA was present in 65% of ischemic/hemorrhagic stroke and TIA patients, with greater prevalence in male patients, patients with recurrent strokes, and patients with strokes of unknown etiology.⁶¹³

Most strokes result from a reduction in cerebral blood flow to a specific region of the brain. The subsequent brain injury from blood–brain barrier dysfunction starts a series of inflammation, oxidative stress, excitotoxicity, and apoptosis.⁷⁵⁵

OSA increases the risk for stroke through a variety of factors leading to vascular damage in the brain. The repeated hypoxia can lead to endothelium damage and release of pro-inflammatory factors, such as plasma cytokines, tumor necrosis factor-alpha, and interleukin-6. This may ultimately cause vascular dysfunction by increasing endothelin, neurovascular oxidative stress, and increasing susceptibility to injury. Moreover, the associated large negative intrathoracic pressure swings could result in mechanical stress on the heart and valves. This can result in nocturnal apneic-related right to left shunting through patent foramen ovale and consequently increase risk of embolism and stroke.³³

A cohort of 5422 participants without a history of stroke and untreated OSA were followed for a median of 8.7 years for incident stroke. Men with moderately severe OSA were at increased risk for incident ischemic stroke (hazard ratio [HR] 2.86; 95% CI 1.10–7.39) after adjusting for demographic variables (age, race) and CV risk factors (BMI, smoking, systolic BP, use of antihypertensive medications, and diabetes). Furthermore, there seemed to be a dose response relationship in which the risk of stroke increased 6% with every one-unit increase in AHI, suggesting as severity of OSA increased, risk of stroke also increased. In women, stroke risk was not significantly associated with obstructive apnea-hypopnea index (OAHI) quartiles or desaturation levels, but increased risk was seen at OAHI levels greater than 25. After a minimum threshold of 25 obstructive events/h is met, with unit increase, there is a 2% increase in stroke HRs. Interestingly, arousal index was a significant negative predictor of incident stroke in women. Women with arousal index >12 was associated with decreased hazard rate of ischemic stroke compared to women with lower arousal index.³³ Future studies should further explore the protective role arousal index may play in incident stroke in women.

An MA was conducted of 12 prospective cohort studies that followed a total of 25,760 participants for major CV events, fatal and nonfatal stroke, coronary heart disease, and all-cause mortality. It was found that severe OSA is independently associated with an increased risk of stroke, CV disease, and all-cause mortality. Three of the 12 studies examined the relationship between severe OSA and risk of stroke. Patients with severe OSA had an increased risk for stroke (combined RR = 2.15, 95% CI: 1.42, 3.24).⁷⁵⁷

Another MA assessed the risk of cerebrovascular events among OSA patients in 15 prospective studies and 43 nonprospective cohort studies (includes cross-sectional studies, case–control studies, and prospective observational studies) to ascertain the prevalence of OSA among patients with cerebrovascular (CV) disease. OSA was found to significantly increase the risk of fatal and non-fatal cerebrovascular disease (pooled HR = 1.94, 95% CI: 1.31–2.89) after adjusting for confounders. Furthermore, OSA was shown to be highly prevalent (58.8%) among patients with cerebrovascular disease and there was greater prevalence with increasing age.⁷⁵⁸

It has been reported that CVA patients with OSA experience longer hospitalization and rehabilitation admissions and higher mortality rates compared to CVA patients without OSA. A prospective cohort study examined the relationship and prevalence of OSA in patients with acute ischemic stroke. Of the 174 patients with acute ischemic stroke, only seven had a past medical history of OSA. Those patients diagnosed with OSA prior to acute ischemic stroke experienced significantly worse functional outcome as measured by lower modified Rankin scale (mRS) scores at hospital discharge after adjusting for age and stroke severity.⁷⁵⁹ Untreated OSA can cause impaired cognitive function, decreased concentration, and excessive daytime sleepiness, which ultimately prolongs the hospitalization stay and compromises rehabilitation participation.^760,761

Therefore, it is recommended to screen for OSA in all patients presenting with TIA or ischemic or hemorrhagic stroke regardless of whether they are symptomatic or asymptomatic. Earlier diagnosis of OSA and early treatment could improve overall health and cognitive status and reduce the risk of recurrent stroke and stroke mortality. (Table VII.A.7)

TABLE VII.A.7.

Association between OSA and cerebrovascular disease

Study	Year	LOE	Study design	Study groups	Clinical endpoints	Conclusion
Birkbak et al.762	2014	2a	Systematic review	10 studies with 1203 stroke and TIA patients were reviewed to examine relationship between sleep disordered breathing and recurrent stroke and mortality.	All-cause mortality, recurrent vascular events, and case fatality in stroke/TIA patients.	Obstructive sleep disordered breathing is a risk factor for recurrent vascular events and all-cause mortality in stroke/TIA patients.
Culebras and Anwar763	2018	2a	Systematic review	Review of cerebrovascular complications of OSA.	Factors contributing to CVD.	OSA is highly prevalent in the stroke population. OSA increases the risk of hypertension, stroke, myocardial infarction, and afib and is closely linked to vascular dementia. OSA may have worse neurological outcomes in acute stroke patients and have worse recovery in stroke rehab.
Dong et al.764	2018	2a	Systematic review	37 studies with 3242 patients were reviewed to determine prevalence of OSA.	Prevalence of OSA in patients with CVD.	OSA is highly prevalent in patients with CVD. Prevalence of AHI > 5 was 70.4% and prevalence of AHI > 10 was 61.9%.
Johnson and Johnson613	2010	2a	Meta-analysis	29 studies with 2343 ischemic or hemorrhagic stroke and TIA patients were reviewed.	Prevalence of sleep disordered breathing (SDB) in ischemic and hemorrhagic stroke and TIA patients.	SDB is very common in stroke patients regardless of type of stroke or timing after stroke and is usually obstructive in nature.
Li et al.765	2014	2a	Meta-analysis	10 cohort studies investigating the effect of OSA on incident ischemic and hemorrhagic stroke.	Incident ischemic and hemorrhagic stroke.	There was a significant association between OSA and the risk of fatal or non-fatal stroke after adjustment of established cardiovascular risk factors.
Mansu-khani et al.759	2011	2b	Prospective cohort study	174 consecutive patients presenting with acute ischemic stroke in Saint Mary’s Hospital ED between June 2007 and March 2008.	Functional outcomes measured using modified Rankin scale at discharge.	Previous diagnosis of OSA was an independent predictor of worse functional outcome. Patients with definitive diagnosis of OSA before stroke are at increased risk of death within the first month after an acute ischemic stroke.
McDermott and Brown766	2020	2a	Systematic review	Assessing the directional relationship between the association between OSA and stroke.	Prevalence of OSA in the poststroke patient population.	Sleep apnea is an independent risk factor for stroke. OSA is highly prevalent poststroke and is associated with worse outcomes after stroke.
Mohammad et al.756	2019	3a	Case–control study	107 patients admitted with acute ischemic stroke who were categorized into two groups: those who woke up with stroke symptoms (WUS) and those whose stroke occurred while awake (NWUS).	Risk for OSA assessed using the Berlin questionnaire.	Risk for OSA is high among the stroke population, with a greater prevalence among the WUS group. OSA is an important risk factor for ischemic stroke during sleep.
Redline et al.33	2010	2a	Prospective cohort study	Community-based sample of 5422 male and female participants in the Sleep Heart Health study without a history of stroke and untreated for OSA were followed for a median of 8.7 years.	Incidence of ischemic stroke.	There is a strong association between ischemic stroke and obstructive AHI in community-dwelling men with mild to moderate OSA. In women, stroke was not significantly associated with obstructive apnea-hypopnea index (OAHI) quartiles or desaturation levels, but an increased risk of stroke was observed in women with OAHI levels greater than 25.
Seiler et al.767	2019	2a	Systematic review and meta-analysis	29 studies with 7096 patients with ischemic/hemorrhagic stroke or TIA were reviewed to determine prevalence of sleep disordered breathing.	Severity and prevalence of SDB.	Sleep disordered breathing is highly prevalent in patients after stroke and TIA. 70% of patients have an AHI > 5/h and 33% of patients have AHI > 30/h.
Wang et al.757	2013	2a	Meta-analysis	12 prospective cohort studies with 25,760 participants were reviewed.	Incident fatal and non-fatal coronary heart disease, incident fatal and non-fatal stroke, and all-cause mortality.	Severe OSA significantly increases CVD risk, stroke, and all-cause mortality. A positive association with CVD was observed for moderate OSA but not for mild OSA.
Wu et al.758	2017	2a	Meta-analysis	58 studies to determine prevalence of OSA among CV patients and risk for cerebrovascular events.	Prevalence of OSA	OSA is prevalent (58%) in patients with CV disease. As age increases, the prevalence of OSA increases. OSA is significantly increases risk for fatal or non-fatal CV disease.

VII.B |. Comorbidities Associated with OSA: Pulmonary Disorders

VII.B.1 |. Primary pulmonary hypertension

PH, defined as a mean pulmonary artery pressure (mPAP) >20 mmHg,⁷⁶⁸ is a pathophysiological disorder⁷⁶⁹ that may be of primary origin or a sequalae of clinical conditions most notably CV and respiratory diseases.

The relationship between OSA and PH was first noted in 1976 by Tilkian et al.⁷⁷⁰ Few studies have detailed the link between World Health Organization (WHO) group 1 pulmonary arterial hypertension (PAH) and OSA, or the effects of treatment of OSA on hemodynamic variables in adult patients^{747,771–777} (Table VII.B.1). Although the mechanism behind PH associated with OSA is not entirely understood, it is postulated to be due to a combination of factors including pulmonary arteriolar remodeling, susceptibility to hypoxia, and underlying left heart disease.⁷⁷³ A study of WHO group I PAH patients showed there was no significant difference in mortality in patients with and without OSA; however, mortality was significantly higher in patients with nocturnal hypoxemia, defined as an average SpO₂ < 90%, suggesting that duration and severity of nocturnal oxygen desaturation, well known to occur in OSA patients, is an important risk factor for development of PAH.⁷⁷⁶

TABLE VII.B.1.

Association between OSA and pulmonary hypertension

Study	Year	LOE	Study design	Study groups	Clinical endpoint	Conclusion
Minic et al.775	2014	2b	Retrospective cross-sectional	Patients with group 1 PAH (n = 52)	Prevalence and clinical predictors of SDB and impact on outcomes	SDB present in 71% of PAH patients; 60% had significant hypoxemia (>10% of total sleep time at <90% oxyhemoglobin saturation).
Minai et al.774	2009	2b	Prospective cohort	OSA patients undergoing right heart catheterization for suspected PH	1) Predictors of PAH in OSA patients 2) Survival in patients with OSA and PAH	Female gender, younger age, obesity, and nocturnal desaturation were associated with development of PH and OSA patients. PH increases mortality in patients with OSA.
Imran et al.772	2016	2a	Systematic review	Patients with PH and OSA on CPAP (n = 222)	Mean pulmonary artery pressure	CPAP therapy is associated with a significantly lower PA pressure in patients with isolated OSA and PH (13.3 mmHg; 95% CI 12.7–14.0).
Sun et al.777	2016	2a	Systematic review	Patients with OSA on CPAP (n = 181)	Mean pulmonary artery pressure	CPAP therapy was associated with a statistically significant reduction in pulmonary artery pressure in patients with OSA
Arias et al.771	2006	2b	RCT (cross over)	1) Severe OSA (n = 23) 2) Healthy controls (n = 10)	Pulmonary artery systolic pressure	CPAP therapy was associated with a decrease in PASP in patients with PAH and OSA (28.9 + 8.6 to 24.0+5.8 mmHg; p < 0.0001).
Sharma et al.747	2019	2b	RCT	1) Patients with OSA and heart failure (n = 11) 2) Patients with OSA and heart failure on CPAP (n = 10)	Pulmonary artery systolic pressure	CPAP therapy was associated with a decrease in PASP in patients with PAH and OSA (58.6 ± 2.5 to 42.8 ± 2.7 mmHg; p = 0.025).

Abbreviations: AHI, apnea hypopnea index; CPAP, continuous positive airway pressure; OSA, obstructive sleep apnea; PA, pulmonary artery; PAH, pulmonary arterial hypertension; PASP, pulmonary artery systolic pressure; PH, pulmonary hypertension; PSG, polysomnography; RCT, randomized controlled trial; RHC, right heart catheterization; SDB, sleep disordered breathing.

The prevalence of OSA in PAH and vice versa is not well elucidated. A study by Minic et al. showed that SDB (OSA, CSA, hypoventilation) was present in the majority of patients with PAH, with more than half of patients having significant nocturnal hypoxemia.⁷⁷⁵ Other reported risk factors including female gender, younger age, obesity, and nocturnal desaturations are thought to increase the risk of developing PAH in OSA patients.⁷⁷⁴

Although data on the effects of CPAP, the preferred treatment of OSA, on hemodynamic variables have been inconsistent, two SRs reported that CPAP therapy is associated with a reduction in mPAP in patients with OSA and PH.^772,777 Another study showed that CPAP induced significant improvements in echocardiographic parameters, most notably pulmonary artery systolic pressure (PASP), further highlighting CPAP use and its positive effects on hemodynamic variables.⁷⁷¹ Lastly, it was found that in patients with PH admitted for acute decompensated HF, the addition of 48 h of CPAP to standard care improved LVEF and significantly reduced PASP.⁷⁴⁷

In summary, there is some indication that survival for OSA patients with PH may be lower than that for OSA patients without PH.⁷⁷⁴ Based on limited studies, CPAP appears to improve hemodynamic variables including mPAP and PASP in patients with combined OSA and PAH; however, how these positive effects translate to meaningful clinical outcomes are still unclear. (Table VII.B.1)

VII.B.2 |. Chronic obstructive pulmonary disease (COPD)

The concurrent diagnosis of OSA and COPD in an individual is known as overlap syndrome (OS).⁷⁷⁸ OS is associated with a more severe clinical course compared to either disease alone.^779–782 Shawon et al. conducted an SR to determine the prevalence and clinical outcomes of OS and reported a significantly higher prevalence of OS in patients with either OSA or COPD compared to the general population.⁷⁸³ OS patients experienced greater degree of nocturnal oxygen desaturation (SpO₂ < 90%) and lower sleep efficiency compared to OSA patients. OS was also associated with increased CV complications (PH, AF, right ventricular dysfunction), COPD exacerbations, hospitalizations, and poorer QOL compared to either disease alone. (Table VII.B.2a)

TABLE VII.B.2a.

Prevalence and outcomes in OSA with coexisting COPD (overlap syndrome)

Study	Year	LOE	Study design	Study groups	Clinical endpoint	Conclusion
Shawon et al.783	2016	2a	Systematic review	1) General and hospital populations 1. COPD 2. OSA (n = 29,341; age 40–90)	1) Prevalence of overlap syndrome 2) Clinical endpoints of overlap: cardiovascular disease, pulmonary hypertension, resistant hypertension, exacerbation of COPD, mortality, quality of life, hospitalization rate	Overlap syndrome: High prevalence of overlap in either OSA (7.6%–55.7%) or COPD (2.9%–65.9%) population compared to general population (1.0%–3.6%). Overlap syndrome has greater nocturnal oxygen desaturations, reduced mean peripheral capillary oxygen saturation (SpO2) and increased sleep time spent with SpO2 < 90% (T90) and worse sleep quality than patients with only OSA. Overlap syndrome is associated with more frequent cardiovascular morbidity, poorer quality of life, more frequent COPD exacerbations and increased medical costs.
Donovan et al.779	2019	2b	Prospective cohort study	Participants from LOTT study (Long-term Oxygen Treatment Trial) with COPD assessed for OSA risk by STOP-BANG questionnaire Intermediate-to-high risk (score > 3) Low risk (score < 3)	Composite of death First hospitalization Overall incidence of COPD exacerbations	High percentage of COPD patients were at intermediate to high risk of undiagnosed OSA based on modified STOP-BANG criteria and were associated with greater risk of poor outcomes.
Sanders et al.781	2003	2b	Prospective cohort study	Participants from Sleep Heart Health Study (community based), using polysomnography and spirometry results	1. Determine association between OAD and SAH 2. Identify predictors of nocturnal oxyhemoglobin desaturation in OAD with and without SAH	1. No association between mild OAD and SAH. 2. Independent of SAH, FEV1/FVC < 65% is associated with more severe nocturnal oxygen desaturation. 3. Nocturnal oxygen desaturation is greater in combined OAD and SAH compared with either disease alone. 4. Mild OAD patients without OSA have minimally disturbed sleep.
Du et al.780	2018	2b	Cross-sectional study	NHANES database (2005–2008) No COPD or OSA OSA COPD Overlap Syndrome	Evaluate factors associated with all-cause mortality	Overlap syndrome and COPD were associated with higher all-cause mortality compared to neither disease and OSA alone.
Starr et al.782	2019	4	Retrospective cohort study	Enrollment and claims data 2004–2013 from a 5% national sample of Medicare beneficiaries with COPD and Overlap Syndrome	Examine the diagnosed prevalence, trend, and patient characteristics of overlap syndrome	Prevalence: 11.0% of the COPD cohort had coexisting OSA compared with patients with COPD alone. Trend: younger age and male gender have a higher number of comorbid conditions and have more complex COPD. Over the 10-year study period, overlap syndrome increased from 4.04% in 2004 to 17.80% in 2013 based on visit for OSA.

Abbreviations: COPD, chronic obstructive pulmonary disease; OAD, obstructive airway disease; OSA, obstructive sleep apnea; SAH, sleep apnea hypopnea.

VII.C |. Comorbidities Associated with OSA: Obesity

Obesity is the most common risk factor for OSA, diagnosed in more than 70% of OSA patients.⁷⁸⁴ Weight gain has been associated with greater SDB severity in both observational and intervention studies.^189,191,785 The literature linking obesity with OSA is challenged by the sheer number of confounders among patients with metabolic disease, including behavioral variables (activity level, dietary indiscretion, and medication compliance), access to care, and socioeconomic status.

The relationship between excess weight and OSA is likely reciprocal. OSA fragments sleep, leading to chronic insufficient sleep and excessive daytime sleepiness. These factors lead to reduced activity levels as well as increased appetite, with resultant weight gain.⁷⁸⁶ Patients with OSA appear to be more susceptible to weight gain compared to similarly matched obese counterparts with significant weight gain reported in the year leading up to a diagnosis of OSA.^787,788 Dysregulation of leptin, with adverse implications for weight loss, body fat storage, and distribution, is well documented in obese patients with OSA.^788,789 Among a cohort of subjects with morbid obesity and OSA who underwent bariatric surgery, long-term weight loss outcomes were worse among those that did not continue to use CPAP postoperatively.⁷⁹⁰

Although untreated OSA is associated with weight gain, CPAP adherence has not been associated with weight loss and may even lead to weight gain.^791–794 A number of theories have been put forth to explain the phenomena of weight gain following initiation of CPAP. Tachikawa et al. proposed that CPAP leads to a small reduction in basal metabolic rate, likely by eliminating additional energy expenditure due to breathing against an obstructive airway.⁷⁹⁵ Stenlof et al. have also found a reduction in energy expenditure with CPAP therapy.⁷⁹⁶ It is unclear if, and how, these reductions in energy expenditure translate to weight gain or inability to lose weight. Some studies have shown that those who gained weight with CPAP failed to record an increase in daytime activity and had more disordered eating behaviors.^797,798 Others postulate that a resurgence of slow wave sleep following treatment of OSA may lead to increases in growth hormone and subsequent weight gain.^795,799,800 These explanations are all speculative, and the best use of the information at present is tempering messages to patients that treatment of OSA will lead to a reduction in weight. While there are a number of meaningful outcomes from CPAP, including improvements in mood and QOL, a linear relationship with weight is unclear.

Impact of obesity on the management of OSA

Obesity exhibits both a mechanical and neurophysiologic effect on UA patency. Adipose deposition in the UA leads to a smaller lumen and increased collapsibility, both of which predispose to apnea.⁸⁰¹ In addition, fat deposits around the thorax and abdomen reduce chest wall compliance and diminish functional residual capacity, which both increases oxygen demand and contributes to greater nocturnal hypoxia, especially among apneic patients.⁸⁰²

The critical closing pressure, Pcrit, is determined by mechanical and neural factors that regulate pharyngeal collapsibility. In obese patients an elevated Pcrit is due to UA adipose deposition.⁸⁰³ Obesity can also lead to reduced lung volumes, further compromised during supine sleep. These factors make OSA more common in obese individuals, contributing to increased severity of disease, longer apneic episodes, and more hypoxemia. These factors also reduce the efficacy of OSA-specific therapies.

PAP therapy remains the most commonly prescribed treatment for OSA. While efficacious, the treatment response may be reduced in obese patients, who tend to have greater residual AHIs compared with their non-obese counterparts. In addition, therapeutic pressures required to maintain UA patency are generally higher in obese patients.⁸⁰⁴ Not surprisingly, weight loss can lead to a reduction in required PAP pressures.⁸⁰⁵

Intuitively, the increased pressure requirements and diminished treatment response seen in obese patients may be expected to contribute to poor adherence. Interestingly, adherence rates are not lower among those with obesity, with some studies even demonstrating superior adherence in this population.^793,806

Oral appliances (OAs) can provide both efficacious and effective treatment for patients with OSA, including those who are obese.^807–810 However, similar to PAP, the therapeutic effect and likelihood of achieving successful therapy are decreased in obesity. The efficacy of OAs, as measured by the reduction in AHI, decreases with increasing weight in a non-linear relationship. Evidence shows that OAs are more efficacious and have a greater responder rate among those with lower BMIs.^808,810,811 Likewise, the odds of having a high residual AHI are greater in those with increased weight, particularly among those with morbid obesity. As such, many consider morbid obesity to be a relative contraindication for OA therapy, although the use of OA as adjunct therapy requires further evaluation.

Surgical procedures specifically intended to reduce or eliminate SDB can present a treatment option for patients with OSA, particularly those unresponsive or intolerant to PAP therapy. The increased deposition of adipose tissue and greater collapsibility of the UAs commonly seen in obese patients may decrease the probability of a successful surgical outcome. Both the treatment effect and probability of successful therapy with UPPP decrease with increasing BMI.⁸¹² In addition, higher BMI is associated with greater risk for postoperative complications.⁸¹³ Increased BMI may be accompanied by increased adiposity in the tongue base, leading to multilevel collapse of the UA necessitating multiple or staged surgical interventions. While this decreased probability for successful treatment does not necessarily negate surgery as a treatment option for obese patients, it does need to be weighed against the increased associated perioperative risks and higher chance for eventual recurrence of OSA. UA staging systems have been developed to help develop individualized approaches.⁸¹⁴ Like other UA surgical procedures for OSA, HNS can provide a treatment option for those intolerant of PAP therapy. However, this therapy is only approved for a limited BMI range.⁸¹⁵

A determination of whether untreated OSA leads to obesity is complicated by factors which are difficult to study in a controlled fashion, notably patient behaviors and adherence to interventions. The current LOE precludes a determination of whether OSA causes obesity and whether OSA therapies improve obesity.

VII.D |. Comorbidities Associated with OSA: Insulin Resistance

Understanding the relationship between OSA and metabolic disorders is of paramount importance for global health. OSA and metabolic disease are common bedfellows with obesity and associated comorbidities. The global impact is striking as 400–700 million individuals have diabetes or metabolic risk factors for the future development of diabetes, with healthcare costs in the tens of billions of dollars. These rates are arguably on track to double in the next 20 years.^816,817

SDB is highly prevalent among individuals with impaired insulin sensitivity and diabetes, although underlying mechanisms are nebulous. Many studies control for variables such as age, gender, BMI, waist circumference, and race.^657,818,819 Unfortunately, other relevant factors are generally ignored (activity level, exercise, diet, socioeconomic status, access to healthcare, and adherence with medical treatment). Potential pathophysiological mechanisms for a relationship between OSA and metabolic disease tend to be based on animal studies or uncontrolled cohorts. As a result, although a role for OSA in diabetes is compelling, the rigor of available evidence is not available for developing evidence-based conclusions or recommendations.

Among patients with OSA the available literature has shown a broad prevalence of prediabetes (20%–67%), based on impaired fasting glucose and impaired glucose tolerance testing.⁸²⁰ A linear correlation between OSA severity and insulin resistance has been documented with greater severity at diagnosis predicting the risk of incident diabetes.^821,822 In a historical cohort of over 8000 patients, Kendzerska et al. found that those with an AHI of greater than or equal to 30 events/h had a 30% higher hazard of developing diabetes compared to those with AHI < 5 events/h.⁸²² Aside from overall severity, severity in REM and time with an SaO₂ less than 90% also increased risk of diabetes.

The mechanisms by which OSA may increase insulin resistance and impair glucose tolerance remain unclear. Both animal and human studies have linked intermittent hypoxia to insulin resistance.^823,824 An MA by Iftikhar et al. evaluated 16 case–controlled studies to evaluate the association between OSA and insulin resistance. This study found a significant relationship for increasing HOMA-IR with baseline BMI, but not with age, AHI, or gender.⁶⁵⁶ Other studies have shown independent correlations between hypoxemia (increased ODI, increased time spent with SpO₂ < 90%) and sleep fragmentation to higher glucose and insulin concentrations and insulin sensitivity.^816,825

Ongoing research continues to evaluate the role of hypoxia and a pro-inflammatory state, focusing on intermittent hypoxia and increased adipose tissue lipolysis. Murphy et al.⁸²⁶ provided evidence in mice that intermittent hypoxia led to a pro-inflammatory phenotype of adipose tissue. In humans, OSA was associated with increased spontaneous lipolysis that correlated with the severity of OSA as well as insulin resistance and impaired insulin secretion in patients with type 2 DM.⁸¹⁷ In a large-scale cross-sectional study, a significant positive interaction was observed between the severity of OSA and decreased lipoprotein(a) concentration.⁸²⁷ Several authors have outlined a role for episodic hypoxia during apneic events promoting increased adipose lipolysis and levels of free fatty acids systemically which have a pro-inflammatory effects leading to dysregulated glucose homeostasis and reduced insulin sensitivity.⁸¹⁷ Building a link between underlying pathophysiologic mechanisms between OSA and metabolic disease may be confounded by heterogeneity in how metabolic disorders present clinically and deciphering the impact of OSA as opposed to the role of sleepiness, age, gender, and obesity.⁸²⁸ At this time, potential mechanisms for a causative relationship between OSA and metabolic disease are only hypothesis generating. There is not published evidence to date that firmly establishes a role for untreated OSA as a contributor to diabetes.

VII.E |. Comorbidities Associated with OSA: Cognitive Impairment and Dementia

VII.E.1 |. OSA and cognitive performance

Several quantitative reviews evaluating clinical studies have found that OSA is associated with deficits in cognitive performance. Two meta-reviews have examined the quantitative reviews, and both identified associations between OSA and cognitive deficits across many domains.^829,830 Cognitive domains commonly found to be negatively impacted by OSA include attention, vigilance, executive function, processing speed, and subdomains of memory. Although the majority of meta-analyses have demonstrated cognitive deficits in patients with OSA compared to healthy controls, these results have been variable and somewhat inconsistent (Table VII.E).

TABLE VII.E.

Association between OSA and cognitive impairment

Study	Year	LOE	Study design	Study groups	Clinical endpoint	Conclusion
Olaithe et al.830	2018	2a	Meta-review	7 OSA meta-analyses (2003–2016): Individuals with OSA and healthy controls. Also included were 5 COPD studies, 2 insomnia studies, 4 sleep deprivation studies.	Cognitive deficits in OSA, COPD, insomnia, or sleep deprivation.	Cognitive deficits in OSA include attention, memory, executive function, psychomotor function, visuospatial, and language abilities. Small to large effects were found in most studies across most domains.
Bucks et al.829	2013	2a	Meta-review	5 quantitative reviews (2001–2012): older adults (≥50 years) with OSA vs. controls/norms.	Domains of cognitive function in OSA.	OSA is associated with deficits in attention/vigilance, delayed verbal and visual long-term memory, visuospatial/constructional abilities, and executive function, but does not appear to affect language ability or psychomotor function. The data are equivocal for working memory, short-term memory, and global cognitive functioning. Treatment with CPAP appears to improve executive dysfunction, delayed long-term verbal and visual memory, attention/vigilance, and global cognitive functioning.
Bubu et al.832	2020	2a	Systematic review	68 studies (cross-sectional, longitudinal, and RCTs, 1985–2019). Studies were stratified based on age: young/middle-aged (30–60 years) or older (>60 years) adults.	1. OSA and cognitive performance/impairment. 2. OSA and subsequent development of mild cognitive impairment/dementia. 3. OSA and biomarkers of Alzheimer disease.	1. OSA and cognition: In young and middle-aged adults, OSA is often associated with cognitive impairment, although unclear whether OSA precedes cognitive decline. CPAP treatment improves certain cognitive domains. In older adults, the OSA-cognition associations are variable and depend on study type and setting. 2. OSA and MCI/dementia: lack of longitudinal studies for young/middle-aged adults and more research is needed. In older adults, OSA is often associated with the development of MCI/AD. CPAP treatment improved sleep quality and cognitive function in AD patients with OSA. 3. OSA and biomarkers of AD: in middle-aged and older adults, there is an association between OSA and biomarkers of AD pathology in cross-sectional studies. Limited prospective studies show contrasting associations.
Wang et al.841	2020	2a	Systematic review/review	40 studies (cohort, case study, RCT).	1. Cognitive impairments associated with OSA. 2. OSA and Alzheimer disease pathological markers. 3. Cognitive performance with CPAP treatment.	1. Individuals with OSA had worse cognitive performance than controls in domains of attention, executive function, intelligence, memory, psychomotor speed, and alertness. 2. AD pathological markers appear to be increased in OSA. 3. CPAP therapy could improve cognitive impairments in OSA, particularly if duration of therapy is >4 weeks.
Mubashir et al.834	2019	3a	Systematic review	5 studies (4 cross-sectional, 1 retrospective cohort, 2011–2015) reporting the prevalence of OSA among adult patients (>18 years) with mild cognitive impairment (MCI).	1. Prevalence of OSA in MCI. 2. Risk of OSA in MCI.	1. There is a highly variable prevalence of OSA in MCI population (11%–71%), compared to 4%–70% in included control groups. 2. One study recruiting from memory clinics showed >3-fold risk of OSA in patients with MCI compared to control group. There were no differences in risk of OSA between MCI and control groups in the other included studies with control populations.
Zhu and Zhao831	2018	2a	Meta-analysis	6 cohort studies (2011–2016): 19,940 adults ≥18 years, with and without sleep-disordered breathing.	Association between sleep-disordered breathing and the incidence of cognitive decline.	Baseline SDB is independently associated with risk of cognitive decline (higher risk of mild cognitive impairment compared with dementia). Stratified analyses suggest gender difference (higher risk for incidence of cognitive decline in females, not males).
Cross et al.842	2017	3a	Systematic review and meta-analysis	13 studies (cross-sectional, case–control, 1985–2016): 5104 adults, ≥50 years, with or without OSA.	Effect of OSA on neuropsychological performance.	There was a small negative association between OSA and all combined neuropsychological outcomes, but this association may be due to publication bias (calculations accounting for bias resulted in a null association). Association between OSA and cognition in older age is highly variable and depends on type/setting of study.
Stranks et al.843	2016	3a	Meta-analysis	19 studies (1987–2012) of adult OSA patients vs. healthy controls.	Cognitive function of OSA patients.	For OSA patients, statistically significant negative effect sizes were found in the domains of non-verbal memory, concept formation, psychomotor speed, construction, executive functioning, perception, motor control and performance, attention, speed of processing, working and verbal memory, verbal functioning, and verbal reasoning. The domain of perception was not impaired. Non-verbal memory, concept formation, and psychomotor speed were most impaired.
Emamian et al.833	2016	3a	Meta-analysis	5 cross-sectional studies (1983–1989) of Alzheimer disease patients vs. healthy controls.	Odds ratio for OSA in Alzheimer disease vs. healthy control.	Patients with AD had a 5× higher chance of presenting with OSA than healthy age-matched individuals (aggregate odds ratio was 5.05 and homogeneous).
Vaessen et al.844	2015	3a	Systematic review	12 studies (9 case–control, 1 longitudinal uncontrolled, 1 descriptive uncontrolled, 1 cross-sectional) with adult patients with OSA vs. controls or non-OSA population.	Cognitive complaints in untreated OSA.	Concentration complaints were increased in untreated OSA patients compared to primary snorers and healthy controls. Memory and executive function may be similarly increased, however insufficient data precluded firm conclusions. Cognitive complaints may be related to higher levels of subjective sleepiness. The authors stress the importance of the difference between subjective cognitive complaints and objective cognitive impairment.
Kilpinen et al.845	2014	3a	Systematic review	44 studies: comparison of OSA patients to healthy controls.	Information processing speed.	Information processing speed was reduced in OSA patients in half of the studies. Reduced information processing speed was seen 75% of the time when compared to norm-referenced data. CPAP treatment improved processing speed marginally when compared to placebo/conservative treatment.
Wallace and Bucks846	2013	2a	Meta-analysis	42 studies: 2294 adults with untreated OSA, and 1364 healthy controls.	Tasks associated with episodic memory (immediate recall, delayed recall, learning, and/or recognition memory).	Compared to healthy controls, adults with OSA had significant impairment in verbal episodic memory (immediate recall, delayed recall, learning, and recognition) and visuo-spatial episodic memory (immediate and delayed recall), but not visual immediate recall or visuo-spatial learning. Compared to norms, adults with OSA had significant impairment only in verbal immediate and delayed recall.
Saunamäki and Jehkonen847	2007	2a	Systematic review	40 studies assessing executive function in OSA.	Domains of executive function.	Executive function domains that were most adversely affected in OSA include working memory, phonological fluency, cognitive flexibility, and planning. CPAP improved performance times, cognitive flexibility, and planning, but did not significantly improve deficits in working memory and phonological fluency.
Aloia et al.848	2004	2a	Systematic review	37 studies (1985–2002)	Pattern of cognitive deficits in OSA, cognitive domains improved by treatment, possible mechanisms of cognitive dysfunction.	Findings were equivocal for most cognitive domains. Attention/vigilance, executive function, and memory were impaired in the majority of reviewed studies (6/8, 6/9, and 7/11 studies, respectively). Treatment improved attention/vigilance in most studies and did not improve constructional abilities or psychomotor function.
Beebe et al.849	2003	2a	Meta-analysis	25 studies: 1092 patients with OSA and 899 healthy controls.	Neuropsychological outcome domains in untreated OSA.	Untreated OSA had significant impact on vigilance and executive functioning, and negligible impact on intellectual and verbal functioning. Data were mixed for visual and motor functioning, and memory functioning.
Fulda and Schulz850	2003	3a	Meta-analysis	54 studies (1985–2000): 1635 SRBD patients compared with 1737 controls.	Cognitive dysfunction in SRBD.	Compared with controls, SRBD patients demonstrated: Moderate to large reductions in mental flexibility, visual delayed-memory retrieval, and driving simulation performance. Small to moderate reductions in focused and sustained attention, verbal delayed-memory retrieval, verbal fluency, and composite measures of general intellectual functioning. No difference in divided attention, concept formation and reasoning, and verbal or visual immediate-memory performance. Other domains not assessed due to insufficient data or between-study heterogeneity.

VII.E.2 |. OSA and mild cognitive impairment (MCI)/dementia

The association between OSA and MCI/dementia is a growing area of research. A large MA of 19,940 patients has shown that SDB at baseline is associated with risk of cognitive decline.⁸³¹ A recent SR of 68 studies found that OSA is often associated with the development of MCI and Alzheimer’s disease (AD) in older adults,⁸³² and that OSA is often associated with cognitive impairment in young and middle-aged adults. An MA showed that patients with AD had a 5× higher chance of presenting with OSA than healthy age-matched individuals.⁸³³ In MCI patients, the association is less clear, with a highly variable prevalence of OSA (11%–71%) based on OSA diagnostic methods and patient recruitment locations⁸³⁴ (Table VII.E).

Several cross-sectional, cohort, and case–control studies examined the relationship between OSA and dementia. Cross-sectional studies showed high rates of OSA among dementia patients.^835,836 One recent study found a very high prevalence of OSA in patients with mild to moderate AD (116 of 128 patients, or 90.6%).⁸³⁵ In a retrospective cohort study, patients with SDB were 1.58 times more likely to develop AD than those without SDB.⁸³⁷ Large cohort studies focusing on OSA and dementia appear inconsistent. A prospective cohort study from the SHHS found that elderly cognitively-normal (mean age = 76.9 years, n = 208) individuals with OSA were found to have increased risk of dementia compared to those without OSA, independent of age, gender, and ApoE4 allele status.⁸³⁸ In contrast, the Atherosclerosis Risk in Communities Study, a 15-year prospective cohort study, found that sleep apnea severity and nocturnal hypoxemia were not associated with cognitive decline.⁸³⁹ In another analysis of the same cohort, OSA was not associated with risk of incident dementia; however, when adjudicated outcomes were used, severe OSA was associated with risk of dementia in later life.⁸⁴⁰

Overall, most evidence suggests that OSA may represent a modifiable risk factor for dementia, but conflicting and inconsistent data exist. (Table VII.E)

VII.F |. Comorbidities Associated with OSA: Cancer

The epidemiological association between OSA and cancer was first published in 2012, and since then, a total of 21 studies have been published [PubMed]. The consensus is that higher OSA severity (e.g., more hypoxic burden) is associated with an increase in cancer risk.

Two population-based epidemiological studies with long-term (>20 years) follow-up reported a significant association between OSA severity and all cancers.^851,852 Five epidemiologic studies with a shorter (<20 years) follow-up also reported a significant correlation between OSA severity and all cancers. These include increased cancer incidence⁸⁵³ and increased cancer mortality in patients <65 years old with severe OSA⁸⁵⁴; an increased cancer incidence in patients <45 years old with severe OSA⁸⁵⁵; an increased cancer prevalence in women with OSA, but not in men⁸⁵⁶; and a higher overall incidence of cancer in a veteran population with OSA.⁸⁵⁷ Two shorter follow-up studies reported no association between OSA and all cancers.^858,859 However, when epidemiological studies assessed an association of OSA and specific cancers, results were heterogeneous but support a higher incidence of breast cancer,⁸⁵⁸ primary central nervous system cancers,⁸⁶⁰ nasal and prostate cancer,⁸⁶¹ colorectal cancer,⁸⁶² and melanoma, kidney, breast, and uterine cancer.⁸⁶³

VII.F.1 |. Risk factors

VII.F.1.a |. Changes in upper airway anatomy

There is a strong increased incidence of OSA in patients with head and neck cancer – both on presentation⁸⁶⁴ and after radiation therapy,⁸⁶⁵ likely related to changes in craniofacial anatomy.

VII.F.1.b |. Increased hypoxic burden

Increased hypoxic burden, specifically, cyclical intermittent hypoxia (CIH) is hypothesized as the primary mechanism on OSA and increased cancer risk. Higher OSA severity as measured by PSG, specifically those related to an increased hypoxic burden (e.g., T90), has been associated with melanoma aggressiveness,⁸⁶⁶ pancreatic cancer progression,⁸⁶⁷ lung cancer prevalence,⁸⁶⁸ and cancer mortality.⁸⁶⁹ In addition, potential mechanisms that associate hypoxic burden with increased cancer risk include: significantly reduced levels of circulating invariant natural killer T cells and function (in OSA patients without cancer),⁸⁷⁰ alterations of molecules related to HIF1 (in OSA patients with lung cancer),⁸⁶⁸ increased microRNAs involved with some cancer types (miR-1254 and miR-320e in OSA patients without cancer),⁸⁷¹ and upregulation of circulating TGF-β1, VEGF, and Foxp3 + Tregs (in OSA patients with lung cancer).⁸⁷² To date, there have not been studies specifically assessing the effect of CPAP modifying cancer risk.

VII.F.2 |. Potential mechanisms

In vitro models have provided two main findings: (1) CIH results in changes that are distinct from sustained hypoxia; (2) CIH results in molecular changes that promote cancer progression or metastases. Most CIH mechanisms that have been explored center around stabilizing HIF1⁸⁷³ and measuring downstream effects. Examples of these downstream effects are conversion to cancer stem cell properties,⁸⁷⁴ increased NOX1 subunit of NADPH oxidase,⁸⁷⁵ increased monocyte polarization toward a tumor-promoting phenotype,⁸⁷⁶ glycolysis,⁸⁷³ and extracellular matrix remodeling.⁸⁷⁷ Other studies focused on why cancer cells exposed to CIH may be more resistant to therapy. These mechanisms include a decrease in topo alpha mRNA and protein,⁸⁷⁸ changes in proteasome function,⁸⁷⁹ and small interfering RNA.⁸⁸⁰ Lastly, CIH may increase expression of genes associated with metastases,⁸⁸¹ and cancer cell migration.⁸⁸²

VII.F.2.a |. Animal models

Animal models have demonstrated that CIH increases cancer progression.⁸⁸³ Although most mouse models placed the cancer in the flank (considered to be inherently more hypoxic⁸⁸⁴ than typical primary sites such as the lung) important findings have emerged: CIH accelerates tumor growth,⁸⁸⁵ increases tumor angiogenesis,⁸⁸⁶ increases metastases,⁸⁸⁷ reduces immunosurveillance,⁸⁸⁸ increases spontaneous tumorigenesis,⁸⁸⁹ and increases molecular markers of tumor aggression including PD-L1.⁸⁹⁰

These exciting areas of clinical, in vitro, and animal studies demonstrate an association of CIH, one of the common features of OSA, to cancer risk. This association could lead to an OSA subtype that would benefit from aggressive screening and treatment, and may identify novel cancer targets.

VII.G |. Comorbidities Associated with OSA: Nasal Disorders

Nasal obstruction or congestion can be caused by both inflammatory and structural factors. Common inflammatory diseases causing nasal obstruction include rhinitis, sinusitis, and nasal polyps. Structurally, obstruction may occur at the level of the external or internal nasal valves, septum, and inferior turbinates. Nasal obstruction, AR, and chronic sinusitis have repeatedly been demonstrated as significant risk factors for habitual snoring, chronic excessive daytime sleepiness, and SDB.^69,891–893

VII.G.1 |. Allergic rhinitis

Several studies have specifically investigated the relationship between AR and SDB and have demonstrated that AR negatively impacts sleep quality and successful treatment with nasal corticosteroids and/or montelukast improves sleep disturbance. In a population-based study of individuals enrolled in the Wisconsin Sleep Cohort Study, chronic severe nasal congestion was identified as an independent risk factor for habitual snoring and participants with nasal congestion due to allergy were 1.8 times more likely to suffer from moderate to severe SDB.^69,893 In a cohort study of AR and non-AR patients, seasonal AR was associated with increased daytime sleepiness and QOL impairment.⁸⁹² While data on the effects of AR on polysomnogram (PSG) parameters remains mixed, medical management of AR has been demonstrated to improve PSG parameters such as improved O₂ nadir, and supine AHI levels.⁸⁹⁴ Intranasal corticosteroids have been shown to improve congestion and daytime somnolence in patients with perennial rhinitis. Pooled data from three double-blind crossover RCTs of budesonide, flunisolide, and fluticasone demonstrated significantly decreased nasal congestion and sleepiness in treated patients.⁸⁹¹ Kiely et al. performed a double-blind crossover RCT on the effect of intranasal steroid on SDB, assessed with PSG and nasal resistance at baseline and after 4-week treatment, and found a significant but small reduction in average AHI in patients with OSA using fluticasone.⁸⁹⁵ (Table VII.G.1)

TABLE VII.G. 1.

Allergic rhinitis and OSA

Study	Year	LOE	Study design	Study groups	Clinical endpoint	Conclusions
Young et al.893	1997	1b	Prospective population-based cohort study	Population-based sample of individuals (n = 911, response rate 50%) enrolled in Wisconsin Sleep Cohort Study	Relationship between nasal obstruction and sleep-disordered breathing assessed via polysomnography, rhinometry, and questionnaires	Participants with nighttime symptoms of rhinitis (5+ nights/month) were significantly more likely to report habitual snoring (3–7 nights/week), chronic excessive daytime sleepiness, or nonrestorative sleep than those who rarely had symptoms (p < 0.0001). Those with nasal congestion due to allergy were 1.8× more likely to have moderate to severe sleep-disordered breathing.
Craig et al.891	2005	1a	Pooled data from three double-blind crossover RCT	69 patients with allergic rhinitis without OSA	Effect of intranasal corticosteroids in patients with perennial rhinitis on reduction of congestion and daytime somnolence	Pooled data of budesonide, flunisolide, and fluticasone demonstrated significantly decreased nasal congestion and sleepiness in treated patients. A correlation between reduction in nasal congestion and an improvement in sleep (p < 0.01) and daytime somnolence (p = 0.01) was demonstrated.
Stuck et al.892	2004	1b	Prospective, cohort study	25 patients with seasonal allergic rhinitis and 25 healthy volunteers assessed with questionnaires (ESS, SF-36) and polysomnography performed before and during pollen season	Effect of seasonal allergic rhinitis on subjective and objective sleep patterns	Seasonal allergic rhinitis leads to increased daytime sleepiness (p = 0.006) and quality of life impairment (p < 0.0001).
Santos et al.896	2008	1b	Double blind crossover RCT	31 patients with persistent allergic rhinitis and sleep disturbances treated with montelukast or placebo	Effect of rhinitis therapy on patient-reported sleep quality and symptoms of daytime sleepiness	Montelukast treatment demonstrated significant improvement in reported daytime somnolence (p = 0.0089) and daytime fatigue (p = 0.0087).
Golden et al.897	2010	1b	Double blind crossover RCT	24 patients with perennial AR received azelastine vs. saline	Effect of topical nasal antihistamines on symptoms of rhinitis, sleep, and daytime somnolence	Azelastine reduced rhinorrhea (p = 0.03) and improved subjective sleep quality (p = 0.04) but did not reduce daytime somnolence.
Lavigne et al.894	2013	2b	Prospective cohort study	21 patients with OSA and allergic rhinitis and 34 patients with OSA without allergic rhinitis	Effect of corticosteroid treatment on disease severity with sleep studies and biopsies obtained from the inferior turbinate, nasopharynx, and uvula	Improved O2 nadir, supine AHI, and daytime somnolence in AR group (p = 0.05).
Kiely et al.895	2004	1b	RCT	OSA with AHI > 10 (n = 13) vs. patients without OSA. Intervention: fluticasone spray for 4 weeks	AHI, ESS, snoring	Reduction in AHI in patients with OSA and AR compared to placebo (23.3 vs. 30.3). Most patients continued to have significant OSA.

VII.G.2 |. Nasal obstruction

Nasal obstruction is a significant risk factor for SDB. While data remains limited regarding objective measures of SDB following treatment for nasal obstruction, literature has shown that sleep quality is compromised in patients with chronic nasal obstruction, with significant improvement in subjective sleep quality after medical and surgical treatment. Anatomic nasal obstruction, including septal deviation, internal/external valve collapse, and inferior turbinate hypertrophy, has been repeatedly identified as common findings in patients with SDB.^6,898

The effect of medical management of nasal obstruction on SDB has been studied extensively. Topical nasal decongestion and/or external dilator strip for chronic nasal obstruction has been shown to decrease snoring, but has not demonstrated a significant, persistent change in AHI or improvement in sleep quality.^{895–897,899–904} In a double-blind crossover RCT investigating the effects of topical nasal decongestant on nasal conductance, symptom scores, and PSG findings, there was a significant decrease in AHI at time of maximal decongestion but no significant change in overall AHI or sleep quality.⁹⁰⁰ Djupesland et al. conducted a double-blind crossover RCT on the effects of external nasal dilation on SDB, assessed by PSG, acoustic rhinometry, and questionnaire. While nasal dimensions increased significantly with external dilator compared to placebo, there was no significant decrease in AHI. Kerr et al. studied the effect of nasal resistance reduction by application of topical vasoconstrictor and insertion of vestibular stents in patients with OSA. On posterior rhinomanometry and polysomnogram, it was found that while reduction of nasal resistance resulted in no significant change in AHI, significant improvement in subjective sleep quality was reported.⁹⁰² Sinonasal surgery continues to play a key role in the management of OSA given demonstrated improvement in CPAP tolerance. Further study of the effect of treatment of nasal obstruction on objective polysomnographic parameters is reviewed elsewhere. (Table VII.G.2)

TABLE VII.G. 2.

Association between nasal obstruction and OSA

Study	Year	LOE	Study design	Study groups	Clinical endpoint	Conclusions
Young et al.69	2001	1b	Prospective population-based cohort study	Population-based sample of individuals enrolled in ongoing Wisconsin Sleep Cohort Study. In-laboratory polysomnography was performed on subset (n = 1032).	Relationship between nasal congestion and snoring.	Chronic severe nasal congestion is an independent risk factor for habitual snoring at baseline and 5-year follow-up (OR 4.9; 95% confidence interval, 2.8–8.8).
An et al.899	2019	1b	Double blind crossover RCT	15 patients with OSA and chronic nasal obstruction without obvious pharyngeal narrowing completed two overnight polysomnograms (randomly applying oxymetazoline or placebo).	Effects of nasal patency on sleep architecture in nasal obstruction- predominant obstructive sleep apnea patients by applying nasal decongestant.	Oxymetazoline resulted in significant increase in REM sleep (p = 0.027) and reduction of stage 1 sleep (p = 0.004). AHI in supine position was significantly reduced (p = 0.001).
Kerr et al.902	1992	1b	Single-blind crossover study	10 patients with OSA receiving oxymetazoline and nasal dilator vs. placebo.	Effect of nasal resistance reduction on OSA and nasal airflow assessed via posterior rhinomanometry and PSG.	Reduction of nasal resistance resulted in no change in AHI but improved subjective sleep quality (p < 0.001).
Stradling et al.533	1991	2b	Cross sectional study	1001 men surveyed regarding sleep quality and underwent polysomnography.	Independent predictors of snoring and obstructive sleep apnea.	Multiple linear regression identified nasal stuffiness as a significant independent predictor of snoring but not OSA.
Vidigal et al.905	2012	2c	Case–control study	47 with moderate/severe OSAS and 20 matched controls.	To evaluate nasal obstruction in patients with OSA compared to controls via questionnaires, physical exam, rhinoscopy, nasal inspiratory peak flow (NIPF), and acoustic rhinometry (AR).	OSA group had a significantly higher score on the nasal symptoms scale (p < 0.01) and higher frequency of nasal alterations including presence of septal deviation (p = 0.01) and inferior nasal turbinate hypertrophy (p < 0.01). NIPF and AR parameters were not significantly different.
McLean et al.903	2005	1b	Single blind crossover RCT	10 patients with nasal obstruction and OSA.	Effect of topical decongestant and external dilator strip on nasal resistance, mouth breathing during sleep, and OSA severity.	While AHI reduced by mean 12 (range 3–22) with treatment, no correlation identified between nasal resistance change and AHI change (r2 = 0.001; p = 0.92).
Clarenbach et al.900	2008	1b	Double blind crossover RCT	12 patients with chronic nasal congestion and OSA treated with nasal xylometazoline or placebo for 1 week.	Effects of topical nasal decongestant on nasal conductance, symptom scores, polysomnography.	While there was a significant decrease in AHI at time of maximal decongestion (p < 0.05), there was no significant change in overall AHI or sleep quality.
Djupesland et al.901	2001	1b	Double blind crossover RCT	18 heavy snorers without severe OSA (mean AHI 9.3) reporting nocturnal nasal obstruction.	Effects of external nasal dilation (Breathe Right) assessed by polysomnography, acoustic rhinometry, and questionnaire.	Nasal dimensions increased significantly (p < 0.001) with active dilator compared to placebo. In habitual snorers (AHI < 10) with severe morning obstruction, no significant reduction in AHI (p = 0.06).
Ishii et al.545	2015	2a	Meta-analysis	10 studies meeting criteria with 320 patients: two RCTs, seven prospective studies, and one retrospective study.	Effect of isolated nasal surgery on OSA.	Isolated nasal surgery for patients with nasal obstruction and OSA improved ESS by 3.53 (95% CI [0.64, 6.23]) and RDI by 11.06 (95% CI [5.92, 16.19]), but had no significant AHI improvement (95% CI [–1.6, 11.62]).
Li et al.906	2011	2a	Meta-analysis	11 prospective noncontrolled clinical trials on outcomes of nasal surgery for OSA.	Effect of nasal surgery on daytime sleepiness, snoring, and polysomnogram.	Mean AHI decreased from 35.2 ± 22.6 to 33.5 ± 23.8 after nasal surgery (p = 0.69). ESS scores decreased from 10.6 ± 3.9 to 7.1 ± 3.7 (p < 0.001). Significant improvement in snoring assessed by questionnaires and visual analog scale (p < 0.05).
Yamasaki et al.907	2020	2b	Prospective cohort study	Patients undergoing nasal surgery surveyed through 24 months postoperatively.	Longitudinal snoring symptoms and nasal obstruction after functional nasal surgery for patients with SDB.	OSA patients achieved clinically significant improvement on Snoring Outcome Survey through 24 months (p > 0.05).
Bosco et al.908	2020	2b	Prospective cohort study	34 patients with AHI > 15 and septal deviation treated with nasal surgery.	Effects of nasal surgery on upper airway collapse, assessed using drug-induced sleep endoscopy.	Before nasal surgery, 74% of the patients demonstrated multilevel obstruction. After nasal surgery, 50% patients showed multilevel collapse (p < 0.05) with significant improvement shown in hypopharyngeal collapse.
Li et al.909	2009	2b	Prospective, controlled nonrandomized study	66 patients with OSA (AHI > 5, BMI < 33) and chronic nasal obstruction (surgical, n = 44; control, n = 22).	Effect of septoplasty with inferior turbinate reduction versus medical therapy (steroid or decongestant spray, saline lavage, or oral antihistamine) on snoring, daytime sleepiness, and respiratory adverse events assessed via questionnaire, rhinomanometry, and polysomnogram at baseline and 3 months.	Significantly improved nasal resistance found in the surgical group (p < 0.001). Both groups revealed insignificant changes in polysomnographic parameters. In subgroup analysis, a significant effect of nasal surgery on AHI was found when patients were divided by Friedman tongue position (FTP) into “low” (FTP 1 or 2) and “high” (FTP 3 or 4) (p = 0.007).
Nakata et al.910	2008	2b	Prospective cohort study	49 OSA patients with symptomatic nasal obstruction.	Effect of nasal surgery on nasal resistance, sleep apnea, and sleep quality in patients with OSA assessed via polysomnography before and after surgery.	While there was no significant change in AHI, nasal surgery decreased nasal resistance (p < 0.001), ameliorated sleep-disordered breathing (increased nadir oxygen saturation, p < 0.01; shortened apnea–hypopnea duration, p < 0.05), and improving sleep quality and daytime sleepiness in OSAS (ESS scores, p < 0.001).
Shuaib et al.911	2015	2c	Retrospective cohort study	26 patients with chief complaint of nasal obstruction found to have septal and nasal valve obstruction on examination, who subsequently underwent functional rhinoplasty.	Effect of nasal surgery on nasal resistance, sleep apnea, and sleep quality in patients with OSA assessed via polysomnography before and after functional septorhinoplasty.	Mean AHI preoperatively was 24.7, which dropped to mean postoperative AHI 16, a reduction of 35% (p = 0.013). Among patients with BMI < 30 resulted in 57% mean AHI reduction, from 22.5 to 9.6 (p < 0.01).
Hisamatsu et al.912	2015	2c	Retrospective cohort study	45 patients with moderate or severe OSA and high nasal resistance assessed by rhinomanometry underwent compound nasal surgery (septoplasty, turbinate reduction, and submucosal resection of the posterior nasal nerve).	Effects of nasal surgery on OSA assessed using polysomnography at 3 months, daytime sleepiness, nasal allergy symptoms, and health-related QOL.	Postoperative improvement was demonstrated in at least one polysomnography parameter in 57% and 75% patients with moderate or severe OSA, respectively. Quality of life measures were also significantly improved.
Silvoniemi et al.898	1997	2c	Cross-sectional study	46 patients with severe nasal obstruction due to septal deviation.	Sleep-disordered breathing as assessed by rhinomanometry and whole night sleep recording.	Thirty-one patients (67%) had also heavy disturbing snoring, and apnea periods during sleep were reported by 10 cases.
Lenders et al.913	1991	2c	Case–control study	45 habitual snorers and 22 patients with OSA examined by PSG, rhinomanometry, and acoustic rhinometry.	Association between anatomic nasal obstruction and sleep-disordered breathing.	In 97% of these patients, inferior turbinate hypertrophy was found by acoustic rhinometry, while increased nasal resistance of various degrees was measured in 93% of all patients by active anterior rhinomanometry.

VII.G.3 |. Chronic rhinosinusitis

Sleep quality is compromised in patients with chronic sinusitis (chronic rhinosinusitis CRS), with significant improvement in subjective sleep quality after medical and surgical treatment.^914–916 Among World Trade Center responders, CRS was an independent risk factor for OSA.⁹¹⁷ Data is limited regarding objective measures of sleep following treatment for CRS. In a prospective cohort study of 405 patients undergoing endoscopic sinus surgery for medically refractory CRS 15% of CRS patients were found to have a history of comorbid OSA based on chart review prior to treatment.⁹¹⁴ PSG-confirmed OSA was identified in approximately 65% of CRS patients undergoing surgery in another prospective cohort study.⁹¹⁸ Preliminary research has suggested that immune mediators associated with CRS may contribute to SDB.⁹¹⁸ (Table VII.G.3)

TABLE VII.G. 3.

Association of chronic rhinosinusitis and OSA

Study	Year	LOE	Study design	Study groups	Clinical endpoints	Conclusions
Quinn et al.916	2017	2a	Meta-analysis	6 prospective cohort studies; two retrospective case series	Relationship between sleep-disordered breathing and CRS.	Sleep quality is compromised in CRS patients with significant improvement in subjective sleep quality after medical and surgical treatment. Data is limited regarding objective measures of sleep following treatment for CRS.
Alt et al.915	2019	2b	Case–control study	52 patients with CRS and 56 controls	Objective sleep measures in CRS patients assessed by home polysomnogram.	Patients with CRS were found to have an increased number of awakenings (p = 0.004), lower average overnight oxygen saturation (p = 0.042), increased REM latency (p = 0.016), and increased snoring at >40 dB (p = 0.034).
Alt et al.914	2015	2b	Prospective cohort study	405 patients undergoing endoscopic sinus surgery for medically refractory chronic rhinosinusitis without nasal polyposis	Impact of comorbid OSA on sleep dysfunction in patients with CRS following functional endoscopic sinus surgery as assessed by Pittsburgh Sleep Quality Index (PSQI).	15% of CRS patients had comorbid OSA and had substantial disease-specific QOL improvements following endoscopic sinus surgery (p < 0.05).
Rotenberg et al.919	2015	2b	Prospective cohort study	53 patients undergoing endoscopic sinus surgery for medically refractory chronic rhinosinusitis without nasal polyposis	Sleep quality following sinus surgery recorded at baseline and 6 months after surgery assessed with Epworth Sleepiness Scale (ESS) and the Pittsburgh Sleep Quality Index (PSQI).	Sinus surgery for CRS is associated with improved patient-reported sleep quality (p < 0.01).
Tosun et al.920	2009	2b	Prospective cohort study	27 patients with nasal polyposis underwent endoscopic sinus surgery with polypectomy	Effect of endoscopic sinus surgery on sleep quality in patients with chronic nasal obstruction resulting from nasal polyposis evaluated by questionnaire and polysomnography before and 3 months after the surgery.	While endoscopic sinus surgery with polypectomy significantly improved snoring and daytime sleepiness in patients with chronic nasal obstruction due to nasal polyposis (p < 0.01), it did not demonstrate a significant change in AHI (p = 0.55).
Yalaman-chali et al.921	2014	2c	Retrospective cohort study	56 patients with OSA and chronic rhinosinusitis who underwent septoplasty with bilateral submucosal inferior turbinate reduction and concurrent endoscopic sinus surgery	Impact of combined nasal surgery and endoscopic sinus surgery on OSA assessed by polysomnography before and 3 months after the surgery.	AHI improved significantly after combined nasal surgery and endoscopic sinus surgery in patients with moderate OSA (p = 0.023) and severe OSA (p = 0.034), while patients with mild OSA did not have significant changes in AHI.
Sunderram et al.917	2019	2b	Cross sectional study	601 subjects enrolled in the WTC Health Program underwent two nights of home sleep testing	Association between CRS and OSA.	CRS is an independent risk factor for OSA (odds ratio of 1.80; p 5.006), even after adjusting for age, BMI, sex, GERD, and EtOH use.
Jiang et al.918	2016	2b	Cross sectional study	139 CRS patients who underwent functional endoscopic sinus surgery	Impact of CRS on sleep-disordered breathing as assessed by PSG and ESS.	OSA was diagnosed in 64.7% of patients with CRS, but there was no correlation with the severity of rhinosinusitis. The study did not assess for change in PSG or ESS outcomes after treatment.
Alt et al.922	2014	2b	Cross sectional study	20 CRS subjects completed disease-specific QOL and olfactory instruments	Association between immune mediators that have been associated with CRS and sleep-disordered breathing as assessed by Pittsburgh Sleep Quality Index (PSQI).	Increased expression of TGF-β (r = −0.443; p = 0.050) and IL-4 (r = −0.548; p = 0.012) correlated with sleep dysfunction, whereas IL-13 expression was linearly associated with worse sleep quality (PSQI scores r = −0.417; p = 0.075).

VII.H |. Comorbidities Associated with OSA: Gastroesophageal Reflux Disease

The relationship between OSA and gastroesophageal reflux disease (GERD) is complex and our understanding of a causative relationship and effects of treatment continues to evolve. Numerous studies have established a high incidence of GERD among patients with SDB.^923–936 Population-based studies have demonstrated that greater than 75% of patients with OSA report nocturnal reflux symptoms, which have been associated with sleep fragmentation and overall poor sleep quality.^927,930,936 Laryngopharyngeal reflux (LPR), which occurs when gastric contents breach both the lower esophageal sphincter (LES) and upper esophageal sphincter (UES), presents with symptoms such as chronic cough, hoarseness, and throat clearing, and has also been associated with SDB.^{923,924,926–930,932,933,936,937} While the causal relationship between obstructive events and nocturnal reflux has yet to be established, a cycle of increases in respiratory effort with negative intrathoracic pressure leading to reflux of gastric contents at the level of the LES during sleep has been suggested. However, in a prospective cohort study of 27 adults with reflux confirmed by two validated instruments (Reflex Symptom Index (RSI) ≥ 13 and Reflux Finding Score (RFS) ≥ 7) and OSA, no temporal association between reflux and obstructive respiratory events was identified on PSG with concomitant multichannel intraluminal impedance (MCII)-pH esophageal monitoring.⁹³⁶

While the high incidence of nocturnal reflux among patients with OSA has been repeatedly demonstrated, the correlation between severity of OSA and GERD remains controversial with studies demonstrating variable findings. Though the gold standard for identifying reflux events consists of pH-monitoring studies and outpatient MCII, these tests are costly and most retrospective cohort studies rely on clinical diagnoses based on endoscopic findings and validated patient questionnaires such as the RSI and the RFS. Caparroz et al. found that patients with LPR based on endoscopic findings and elevated RSI were more likely to have a higher mean AHI and higher percentage of sleep time with oxyhemoglobin saturation below 90%.⁹²⁴ Elhennawi et al. studied the relationship between OSA and LPR using 24-h pH monitoring and found that while the number of nocturnal reflux episodes and total duration of reflux during sleep were significantly correlated to the degree of OSA, daytime reflux was not related to the degree of OSA.⁹²⁶ Ultimately, the severity of OSA has not been shown to consistently correlate with gastroesophageal reflux.^926,932,933

While patients with OSA and gastroesophageal reflux have been shown to have a higher BMI than patients with OSA alone, the exact relationship between OSA, obesity, and gastroesophageal reflux has yet to be elucidated.^{924,933,938–941} Obesity is a known risk factor for both OSA and GERD and weight loss has been shown to improve both OSA and nocturnal reflux symptoms in obese patients.^{930,937,939,940} In a recent prospective cohort study on the influence of OSA on GERD in obesity, obese patients without OSA, nonobese patients with OSA, and obese patients with OSA underwent simultaneous PSG and esophageal manometry and pH monitoring. There was no significant difference in the total number of objective reflux events among obese patients with and without OSA. However, obese patients with OSA had a significantly greater number of reflux events compared to the non-obese group with OSA. In multivariate analysis, BMI was a significant predictor of the number of objective reflux events, but AHI showed no significant association with GERD severity.⁹⁴¹

Treatment of GERD has demonstrated significant reduction in snoring and patient-reported daytime sleepiness and improved sleep quality.^927,940 However, treatment of gastroesophageal reflux has not been shown to improve OSA severity based on objective polysomnographic parameters.⁹⁴⁰ Treatment for OSA with CPAP therapy and/or multilevel surgery (MLS) has been shown to significantly improve subjective parameters of reflux.^928,929 In a study of 73 patients who underwent MLS for OSA, mean RSI score decreased significantly after surgery.⁹²⁸ In a study of 44 patients with symptoms of SDB and reflux, subjects underwent 24 h pH monitoring simultaneously with PSG at 0 and 3 months after CPAP treatment and showed significant improvement of subjective parameters of reflux, such as RSI and RFS.⁹²⁷ Improvement in objective parameters of gastroesophageal reflux with treatment for OSA has yet to be demonstrated.

Ultimately, the relationship between OSA and GERD is multifactorial and closely tied to obesity. Treatment for both OSA and reflux results in subjective improvement in reflux and sleep quality, but objective improvement demonstrated through polysomnogram and/or pH monitoring is lacking. Further study of the causal relationship between reflux and SDB using objective parameters is needed. (Table VII.H)

TABLE VII.H.

Association between OSA and GERD

Study	Year	LOE	Study design	Study groups	Clinical endpoint	Conclusion
Teklu et al.935	2020	2c	Cross-sectional study	101 adult patients who underwent polysomnogram	Relationship between reported reflux symptoms and OSA	Patients with OSA have worse symptoms of laryngopharyngeal reflux based on RSI (p = 0.003).
Morse et al.931	2004	2b	Cross sectional study	136 patients referred for polysomnogram	Relationship between OSA and GERD in a large group of patients with OSA	Subjective reports of sleep quality were affected by GERD severity, but no significant correlation between OSA and GERD.
Sundar et al.934	2010	2b	Cross sectional study	75 patients with isolated chronic cough	Rates of OSA among patients with chronic cough	37% of patients with chronic cough had a diagnosis of GERD. 44% were found to have OSA. 93% of the patients that had interventions to optimize their sleep-disordered breathing reported improvement in cough.
Chan et al.925	2010	2b	Cross sectional study	108 consecutive patients referred for evaluation of SDB	Prevalence and severity of chronic cough in patients with sleep-disordered breathing	33% of patients with SDB reported a chronic cough. Patients with SDB and chronic cough were predominantly females (p < 0.001), reported nocturnal heartburn (p = 0.03), and rhinitis (p = 0.02) compared to those without SDB.
Caparroz et al.937	2019	2b	Cross sectional study	56 patients with moderate or severe OSA	Prevalence of laryngopharyngeal reflux in patients with moderate and severe OSA by reflux symptom index questionnaire and indirect videolaryngoscopy	64.3% of patients with moderate or severe OSA had laryngopharyngeal reflux (positive RSI and/or positive endolaryngeal reflux finding score). BMI was a predictor of reflux presence. In patients with positive score for endoscopic findings and reflux symptom index, there was a trend toward significance for a higher mean AHI and a percentage of sleep time with oxyhemoglobin saturation below 90% (p = 0.05).
Lee et al.929	2018	3b	Case–control study	19 OSA patients with unilevel complete obstruction and 69 OSA patients with multilevel complete obstruction on drug-induced sleep endoscopy (DISE)	Relationship between level of obstruction determined by DISE, and laryngopharyngeal reflux (LPR)-related clinical parameters	No significant correlation between OSA severity and LPR-related symptoms. Multilevel complete obstruction on DISE did not affect the LPR-related clinical parameters (p > 0.05).
Kim et al.928	2017	2b	Cross sectional study	73 patients who underwent multilevel surgery for OSA	Effect of multilevel surgery for OSA on symptoms of laryngopharyngeal reflux (LPR) and polysomnogram	Treatment for OSA using multilevel surgery reduced symptoms of LPR; mean RSI score decreased from 11.48 ± 7.95 to 4.95 ± 6.19 after surgery (p < 0.001).
Altintas et al.923	2017	2b	Cross sectional study	62 patients with AHI > 5	Relationship between presence of LPR and level of depression and anxiety in patients with OSA assessed via questionnaire and laryngeal examination	There were significantly higher levels of depression and anxiety in patients with LPR and OSA (p =0.016). A positive correlation was found between RSI and AHI scores (r = 0.338; p = 0.007).
Elhennawi et al.926	2016	2b	Cross sectional study	62 patients with OSA	Relationship between OSA and LPR assessed with ambulatory 24-h pH monitoring	LPR is common in patients with OSA (66%). Patients with severe OSA have significantly higher nocturnal LPR (p < 0.05). Number of reflux episodes and total duration of reflux during sleep are significantly correlated to degree of OSA (p < 0.05). Daytime reflux was not related to degree of OSA (p > 0.05).
Qu et al.932	2015	2b	Case–control study	36 OSA patients and 10 healthy controls underwent 24-h double-probed combined esophageal multichannel intraluminal impedance and pH monitoring simultaneously with polysomnography	Esophageal functional changes observed in OSA	63.9% of patients had both OSA and LPR by pH monitoring and polysomnogram. Significant differences were found in the onset velocity of liquid swallows (p = 0.029) and percent relaxation of the lower esophageal sphincter (LES) during viscous swallows (p = 0.049) between patients with OSA versus healthy controls.
Rodrigues et al.933	2014	2b	Cross sectional study	A total of OSA patients divided into obese group (n = 39) and non-obese patients (n = 66)	Relationship between obesity on LPR and OSA	In the obese group, mean RSI was 6.7 in patients with mild OSA and 11.53 in patients with moderate to severe OSA (p < 0.05). No correlation between OSA severity and RSI in non-obese group.
Xavier et al.936	2019	1b	Cross sectional study	27 adults with LPR confirmed by two validated instruments (RSI ≥ 13 and reflux finding score ≥ 7) and OSA underwent full polysomnography with concomitant multichannel intraluminal impedance-pH esophageal monitoring	Temporal correlation between reflux episodes and respiratory events in patients with LPR and OSA	Among patients with well-established laryngopharyngeal reflux and OSA, there is no temporal association between reflux and obstructive respiratory events.
Eryılmaz et al.927	2012	1b	Prospective cohort study	44 patients underwent double probed 24 h pH monitoring simultaneously with polysomnography due to the complaints of SDB and reflux, at 0 and 3 months	Effect of OSA therapy on LPR parameters	OSA and LPR coexist frequently. LPR treatment did not improve the polysomnographic parameters, but significantly reduced ESS (p = 0.02) and snoring (p = 0.007). Although CPAP treatment significantly improved subjective parameters of reflux, such as RSI and RFS (p = 0.016 for both), there was no significant improvement in objective parameters of 24-h pH monitoring.
Magliulo et al.930	2018	2a	Meta-analysis	10 papers studying LPR in OSA were included with 870 identified OSA patients	Incidence of LPR in OSA patients	There is a high incidence of LPR (45.2%) among OSA patients. AHI severity did not correlate with presence of laryngopharyngeal reflux (p = 0.3). OSA patients with LPR had higher BMI compared with LPR patients (p = 0.001).
Caparroz et al.924	2019	2b	Cross sectional study	70 patients with moderate or severe OSA underwent validated questionnaires, laryngoscopy to calculate the Reflux Finding Score (RFS), and fiber-optic endoscopic evaluation of swallowing (FEES)	Association between presence of dysphagia with signs and symptoms suggestive of LPR in patients with moderate and severe OSA	Although 17.9% of patients presented with findings suggestive of concomitant LPR and dysphagia, there was no statistically significant association between these two conditions.
Kim et al.939	2018	1b	Cross sectional study	216 patients underwent both PSG and EGD	Relationship between OSA and GERD	Endoscopically proven GERD was associated with more severe OSA (p = 0.01). GERD symptoms were also associated with worse sleep quality (p = 0.03).
Rassamee- hiran et al.940	2016	2a	Meta-analysis	2 randomized trials and 4 prospective cohort studies on the effect of treatment for GERD on OSA	Association between PPI treatment for GERD and improvement in OSA	No differences in AHI before and after treatment with PPIs (SMD 0.21; 95% CI [−0.11, 0.54]).
Gilani et al.938	2016	2b	Retrospective study	Adults with OSA and GERD and potentially confounding conditions were identified in the National Ambulatory Medical Care Survey and National Hospital Ambulatory Medical Care Survey	Association between GERD and OSA controlling for concurrent confounders in a national database	Significant positive association between GERD and OSA was observed, which persisted after adjustment for age, sex, race/ethnicity, sino-nasal obstruction, inflammatory disorders, obesity, asthma, and lung disorders (OR 1.94, 95% CI [1.07–3.54], p = 0.030).
Shepherd et al.941	2016	1b	Prospective cohort study	20 obese individuals (BMI > 30), nine non-obese individuals (BMI < 30) with moderate-to-severe OSA, and 17 obese control subjects (BMI > 30) underwent high-resolution esophageal manometry, 24-h esophageal pH-impedance monitoring, and in-laboratory polysomnography	Effect of OSA on GERD independent of obesity	The two obese groups did not show any significant differences in the total number of acidic reflux events (41 ± 20 vs. 28 ± 16); however, the obese OSA group had a greater number of acidic reflux events compared to the non-obese OSA group (22 ± 12 events, p < 0.05). In multivariate analysis, BMI significantly predicted number of acidic reflux events (r(2) = 0.16, p = 0.01). However, AHI showed no significant association with GERD severity.
Xavier et al.942	2013	2c	Cross-sectional study	74 patients with OSA	Prevalence of symptoms of reflux in OSA patients assessed with questionnaire	98% of the subjects with suspected OSA had symptoms suggestive of LPR; prevalence was significantly higher among obese individuals (p = 0.002).

VII.I |. Comorbidities Associated with OSA: Other Sleep Disorders

VII.I.1 |. Insomnia

The available literature addressing the prevalence of insomnia in patients with OSA (Table VII.I.1)^{588,943–956} suffers from several significant shortcomings. Variable definitions are used for OSA and insomnia and different studies focus on “insomnia symptoms” versus “insomnia disorder.” The former is likely to be more common in OSA patients as it requires the presence of less significant symptoms. Additionally, very few studies examined cohorts that are representative of the general population. Only three studies were available with representative population samples^946,948,951 while 10 studies were convenience samples. Thus, the estimates of prevalence may not generalize to the broader population.

TABLE VII.I.1.

Prevalence of insomnia in patients with OSA

Study	Year	LOE	Study design	Study groups	Clinical endpoint	Conclusion
Zhang et al.956	2019	4	Systematic review and meta-analysis	31 studies assessing the prevalence of insomnia symptoms in OSA; four studies assessing the prevalence of OSA in patients with insomnia; two studies assessing prevalence of both insomnia in OSA and OSA in insomnia	Presence of insomnia symptoms or diagnosis, PSG	38% of patients with OSA had comorbid insomnia. In patients with insomnia, the pooled prevalence rates of OSA were 35% for AHI ≥ 5 events/h and 29% for AHI ≥ 15 events/h.
Luyster et al.588	2010	4	Systematic review	4 studies assessing the prevalence of insomnia symptoms in OSA; three studies assessing the prevalence of OSA in patients with insomnia	Presence of insomnia symptoms or diagnosis, PSG	39%–58% of patients with OSA had comorbid insomnia symptoms. 29%–67% of patients with insomnia had an AHI of >5 events/h.
Gupta and Knapp946	2014	2c	Retrospective, cross-sectional, case–control study of epidemiologic databases	62,253,910 ± 5,274,747 (unweighted count = 7234) patient visits with diagnosis of OSA from 1995–2010	ICD-9 coded diagnosis of sleep apnea or insomnia	6.4% ± 0.9% of patient visits for OSA also contained a diagnosis of insomnia. Visits with both OSA and insomnia were significantly more likely to also list essential hypertension (OR = 1.83, 95% CI 1.27–2.65).
Sivertsen et al.951	2013	2c	Historical cohort case–control design	6892 participants aged40–45 years	Karolinska Sleep Questionnaire, Norwegian official registry data, Health information from the Hordaland Health Study in western Norway (1997–1999)	8.4% of the population had OSA and 0.6% had comorbid insomnia.
Lang et al.948	2017	2b	Prospective, general population, cohort study	837 randomly selected men without prior diagnosis of OSA	At-home unattended PSG; 36-item Short Form (SF-36) vitality scale, Beck Depression Inventory	46% of the population had OSA; of these, 14.5% had comorbid insomnia.
Stelzer et al.953	2020	4	Prospective, cross-sectional case series	238 patients with OSA	ICSD-3, PSG-3, Beck Depression, and Anxiety Inventories	29% of patients diagnosed with OSA had comorbid insomnia. Depression and anxiety were higher in the group with insomnia.
Wallace and Wohlgemuth955	2019	4	Prospective, cross-sectional, case series	630 veterans with a new diagnosis of OSA	Insomnia Severity Index (ISI), PSG	74% of patients with OSA had moderate or severe insomnia.
Cho et al.945	2018	4	Prospective, cross-sectional case series	476 patients with OSA seen at two tertiary hospital sleep centers	ISI, PSG, SF-36, PSQI, Beck Depression Inventory	29.2% of patients with OSA had significant insomnia. Patients with both OSA and insomnia had higher rates of heart disease, lower quality of life, and worse depression than those with OSA only. No significant differences in CPAP adherence between those with vs. without insomnia.
Taskaban et al.954	2018	4	Prospective, cross-sectional case series	197 consecutive patients with OSA seen in a sleep laboratory	PSG, WHO Quality of Life form	18.8% of patients with OSA had comorbid insomnia. Patients with both OSA and insomnia had lower physical and social scores than those with OSA only.
Bjorvatn et al.944	2014	4	Prospective, cross-sectional, general population sample case series	1502 randomly selected participants from the general population	Bergen Insomnia Scale. Self- or spouse-reported OSA symptoms	6.2% of participants were diagnosed with OSA; of these, 57.6% reported comorbid insomnia.
Mysliwiec et al.949	2013	4	Prospective cross-sectional case series	110 military personnel who returned from combat with sleep disturbances	ICSD-2, PSG	24.5% diagnosed with OSA alone and 38.2% with comorbid OSA and insomnia. Patients with both OSA and insomnia were significantly more likely to meet criteria for depression.
Bjornsdottir et al.943	2013	4	Prospective cross-sectional case series	705 adults with OSA	Basic Nordic Sleep Questionnaire, HSAT	68.3% had comorbid insomnia, 15.5% exhibited symptoms of initial insomnia, 59.3% had symptoms of middle insomnia, and 27.7% exhibited symptoms of late insomnia, with overlap between groups.
Nguyen et al.950	2010	4	Prospective, cross-sectional, case series	148 consecutive patients with OSA treated in a sleep clinic	ISI, PSG	50% of patients had moderate to severe comorbid insomnia. Presence of insomnia did not significantly affect CPAP adherence.
Smith et al.952	2004	4	Prospective, cross-sectional, case series	105 patients with OSA	ISI, PSG, sleep diary, reported daytime consequences of insomnia, Depression Anxiety Scale-21	35% of patients had comorbid insomnia. Patients with both OSA and insomnia had increased levels of depression, anxiety, and stress.
Krakow et al.947	2001	4	Retrospective cross-sectional case series	231 patients with OSA	Reported insomnia symptoms, sleep diary; reported psychiatric disorders; reported use of sedative and psychotropic medications	50% reported insomnia complaints. OSA + insomina had significantly more psychiatric disorders and physical/mental symptoms that disrupted sleep. OSA + insomnia also reported greater use of sedative and psychotropic medications.

Insomnia symptoms and insomnia disorder are quite common in OSA patients; however, the above-noted limitations are reflected in the wide-range of prevalence estimates found. The reported rates of an insomnia disorder diagnosis in patients with OSA ranged from 6.4% to 74%. Notably, the majority of studies found rates that were clustered in the range of 29%–50%. For insomnia symptoms, the prevalence varied from 7% to 68% with most estimates being in the range of 36%–58%. Notably, the studies with the lowest prevalence rates tended to be the general population studies.

Despite significant limitations in the literature, it is clear that insomnia symptoms and insomnia disorder are seen commonly in OSA patients. Co-morbid insomnia and sleep apnea (COMISA) is associated with greater impairment of sleep quality, reduced QOL, complex diagnostic decisions, and reduced acceptance and response rates to treatments that require adherence (i.e., PAP, mandibular advancement device [MAD], and HNS therapy).^955,957 Effective management of patients with both insomnia and OSA requires targeted therapies for both disorders.⁹⁵⁸ Treatment reviewed in Sections VIII.A.5.i and VIII.B.9. (Table VII.I.1)

VII.I.2 |. Sleep movement disorders

In patients with OSA, specific considerations are warranted in the presence of comorbid RLSs, periodic limb movements (PLMs), REM sleep behavior disorder (RBD), and sleep-related movement disorders. While the evidence supporting specific recommendations is limited, the available data provide indications regarding clinicallyimportant aspects of OSA management in these patient populations.

RBD is an REM-related parasomnia marked by dreamenactment behavior and absence of atonia during REM sleep.⁸¹ RBD may be idiopathic, that is, occurring in the absence of clear etiology, or may be secondary to underlying neurodegenerative synuclein-mediated disorders,⁹⁵⁹ structural lesions, or narcolepsy.⁹⁶⁰ Screening for OSA is indicated in patients with history of dream-enactment behavior, as untreated SDB can mimic RBD.⁹⁶¹ The presence of OSA may be associated with reduced response to RBD treatments such as clonazepam.⁹⁶² On the other hand, treatment of OSA in patients with RBD may improve RBD manifestations.⁹⁶³ Individuals with a neurodegenerative disorder who are diagnosed with RBD may be particularly at risk for OSA.⁹⁶⁴

RLS is a marked by sensory symptoms, often in the limbs, combined with urge to move the limbs that temporarily improves with limb movement and exhibits diurnal variation.⁸¹ OSA may be comorbid with RLS, and treatment of OSA is indicated in patients with RLS and can improve RLS severity.^965,966

PLMs may emerge after initiation of CPAP for the treatment of OSA, with risk factors including older age⁹⁶⁷ and female sex.⁹⁶⁸ PLMs cormorbid with OSA may be associated with increased REM latency and reduced stage 3 sleep.⁹⁶⁹ There may not be a significant impact of treatment-emergent PLMs on daytime sleepiness.^970,971 Nevertheless, presence of PLMs has important clinical implications as they may indicate greater risk of AF and other CV diseases,^972–975 perhaps related to increased sympathetic activation.⁹⁷⁴ Patients with OSA and treatmentemergent PLMs thus warrant age- and comorbidityappropriate monitoring for cardiac disease.^972–975

As for other sleep-related movement disorders, screening for OSA may be indicated in select patients with bruxism though a clear relationship between sleep bruxism and OSA has not been established.^976,977 Finally, limited data indicate that in select adults with rhythmic movement disorder it may be appropriate to perform PSG to identify OSA.⁹⁷⁸ (Table VII.I.2)

TABLE VII.I.2.

Association between sleep movement disorders and OSA

Study	Year	LOE	Study design	Study groups	Clinical endpoints	Conclusion
REM behavior disorder
Li et al.962	2016	4	Case series	39 iRBD patients assessed pre- and post-treatment with clonazepam	Modified RBD questionnaire, ESS, PSG measures	Clonazepam reduced subjective measures of sleep related injury, nightmares, limb movements, and objective measures of REM sleep tonic mentalis surface EMG activity. Residual symptoms were common. Presence of OSA was associated with reduced response to clonazepam.
Iranzo et al.961	2005	3b	Case–control study	16 patients with dream enactment behavior and witnessed apneas were compared, via VPSG, with 20 healthy controls	Describe VPSG features of patients with OSA that mimic behaviors of RBD	Severe OSA may mimic the symptoms of RBD. A clinical history of dream enactment behavior but normal atonia during REM sleep should suggest this RBD mimic VPSG is mandatory to establish the diagnosis of RBD, and identify or exclude other causes of dream-enacting behaviors.
Gabryelska et al.963	2017	4	Cross-sectional survey study	72 out of 120 patients previously diagnosed with RBD and OSA responded to a mailed questionnaire: 27 patients reported use of CPAP. 45 patients reported noncompliance with CPAP	Determine the prevalence of obstructive sleep apnea (OSA) in RBD patients and determine whether continuous positive airway pressure (CPAP) therapy improved RBD symptoms	OSA is a common comorbidity of RBD, being self-reported in 89% of patients with diagnosed RBD. CPAP therapy might improve self-reported RBD symptoms.
Bugalho et al.979	2017	3b	Case–control study	10 patients with RBD and AHI > 14 were compared to 22 with RBD with AHI ≤14	To understand the influence of OSA on symptoms in RBD	RBD patients with OSA had less O2 desaturation compared to those without OSA.
Koo et al.964	2018	3b	Case–control study	35 patients with idiopathic RBD compared to 42 patients with RBD plus a neurodegenerative Parkinsonian syndrome (25 with PD and 17 with MSA)	Compare sleep parameters in RBD patients with vs. without neurodegenerative Parkinsonian syndromes	RBD patients with diagnosed neurodegenerative Parkinsonian syndromes may be more prone to OSA than patients with idiopathic RBD.
Restless leg syndrome
Rodrigues et al.965	2006	4	Case series	17 patients with OSA and RLS or PLMs underwent CPAP titration and follow-up assessment 3 months later	Determine if CPAP therapy improves RLS, as measured by the IRLS, in patients with OSA	IRLS severity decreased significantly post-CPAP compared to pre-CPAP levels.
Silva et al.966	2017	4	Case series	28 patients with RLS and OSA were reviewed for evolution of RLS symptoms following CPAP therapy	Determine if CPAP therapy for OSA is associated with improved RLS symptoms	RLS symptoms improved following CPAP initiation and allowed for reductions in medications for RLS.
Lakshmanan et al.980	2019	3b	Case–control study	IRLS scores were compared in 325 individuals with OSA who received PAP therapy vs. 109 controls, and participants were screened for RLS using a single question	Determine if CPAP therapy for OSA is associated with improved RLS symptoms, as measured with IRLS	OSA patients with RLS who reported adherence to PAP had significant improvements in RLS symptoms compared to those who reported non-adherence.
PLMS/PLMD
Wu et al.974	2018	3b	Case–control	30 patients with OSA and treatment-emergent PLMs vs. 30 patients with OSA without treatment-emergent PLMs	HRV spectral analysis (FFT) applied to investigate basal autonomic regulation reflecting cardiovascular risk in patients with OSA with treatment-emergent PLMs. Parameters examined included RMSSD, LF, HF, and LF/HF ratio	PLMS emerging after CPAP therapy was associated with a decreased HRV during leg movement-free intervals and a shift toward sympathetic predominance in basal autonomic regulation.
Ren et al.968	2016	3b	Case–control	(1) 182 females with OSA matched for severity and age (2) 182 males with OSA matched for severity and age	Differences in PLMS in males vs. females with OSA	In age groups of 55 or less, females with OSA are significantly more likely to have PLMS (PLMI ≥ 15) than males with OSA of similar severity (OR 2.48; 95% CI 1.06–5.79).
Haba-Rubio et al.971	2005	4	Case series	In 57 patients diagnosed with OSA on CPAP: (1) 22 patients with PLMS (PLM index ≥ 5/h) (2) 35 patients without PLMS	Examine the role of PLMS on objective and subjective parameters of sleepiness before and after CPAP use in OSA patients	The presence of PLMS showed no correlation with increased sleepiness as measured by MSLT or by ESS after a 1 year follow-up on CPAP.
Al-Alawi et al.970	2006	3b	Case series	In data of 795 patients who underwent polysomnography: (3) 351 patients had PLMS and OSA (4) 26 patients had PLMS without OSA	Examine the prevalence of PLM arousals in OSA patients Examine relationship of PLM arousal index with risk factors	There was a higher PLMI in OSA subjects (44% had both). Patient with PLM arousal index ≥ 5/h had higher relationships with predisposing conditions (depression, Fibomyalgia, and DM), older age, more predisposing medications, obesity, and more likely to have OSA. There was no difference in the relationship with ESS or hypertension between patient with OSA only or OSA and PLM with arousals.
Xie et al.975	2017	2b	Cross-sectional study	In 15,414 patients who attended a PSG, 50.8% had PLMI>15/h, 36.1% had PLM>30/h, and 13.1% had PLMI between 15 and 30/h. The prevalence of atrial fibrillation was 15.3%	Examine whether PLMS is associated with the prevalence of atrial fibrillation in a group of patients with Sleep Disordered Breathing (SDB)	In a multivariate adjustment model, only mild SDB patients with PLMI > 30/h had 1.21 times higher odds for atrial fibrillation compared with patients with PLMI < 15/h (OR 1.21, 95% CI 1.00–1.47, p-value = 0.048). Similarly, PLMAI > 5/h had higher odds compared to PLMAI < 1 for atrial fibrillation in mild SDB patients (OR 1.27, 95% CI 1.03–1.56, p-value = 0.024).
Iriarte et al.969	2009	4	Cross-sectional study	In a referral sleep center for sleep symptoms, patients were recruited and grouped: (1) 35 patients with no diagnosis for OSA and PLMS (2) 94 patients with OSA only (3) 37 patients with PLMS only (4) 64 patients with PLMS and OSA	Examine the importance of PLMS on sleep quality in patient referred for sleep symptoms in patients with or without OSA	In a multivariate analysis, adjusting for age, sex, and AHI, PLMS was associated significantly with an increase in REM latency of 38 min (CI 95% CI 4.4–71.6; p-value = 0.03) and a decrease stage NREM 3 and 4 of 3.7% (CI 95% CI 0.2–7.2; p-value = 0.04), when comparing all four groups. The effects of PLMS on sleep architecture parameters were worse in the PLMS only group as compared to OSA only or combined OSA and PLMS group.
Loewen et al.981	2009	3b	Case–control	(1) 12 patients with OSA and end-stage renal disease (2) 18 patients with OSA and normal renal function	Differences in effect of PLMS on subjective and objective measures of sleep quality
Aritake-Okada et al.967	2012	3b	Case–control	997 patients diagnosed with OSA: 67 in persistent group (PLMI ≥ 15/h in both diagnositic and CPAP titration study), 80 in CPAP-emergent group (PLMI < 15/h in diagnostic, PLMI ≥ 15/h in CPAP titration study), 40 CPAP-disappeared group (PLMI ≥ 15/h in diagnostic study, PLMI < 15/h in CPAP titration study), non-PLMS group (PLMI < 15/h in diagnostic and CPAP titration study).	Examine the change in prevalence of PLMS in diagnostic and CPAP titration study in OSA patients (AHI > 20 events/h). Examine the associated factors on PLMS groups	PLMS were significantly increased from diagnostic to CPAP titration study in the persistent group, CPAP-emergent group, and non-PLMS group and decreased in the CPAP-disappeared group (all p-value < 0.001). In a multivariate regression analysis, CPAP-emergent PLMS group was associated with older age (age > 47 years: OR: 1.69, 95% CI 1.05–2.71, p-value = 0.03)) and higher AHI (diagnostic AHI > 30/h: OR: 2.19, CI: 1.16–4.11, p-value: 0.015).
Drakatos et al.972	2016	3b	Case–control	49 males without comorbid cardiovascular disease: eight controls, 13 with PLMs (defined as PLMI > 15/h), 17 OSA (defined as AHI > 10/h) and 11 OSA + PLMs	Compare stiffness index derived from the digital volume pulse (SI-DVP) among the groups	Patients with PLMS had higher arterial stiffness measures compared to controls. The OSA/PLMS group had the highest SI-DVP.
Xie et al.982	2019	3b	Case–control	14,444 PSGs conducted over a 4-year period were examined. 314 patients with CSA completed the study with CPAP titration and in cases of persistent CSA, ASV titration	To examine changes in PLMS in response to ASV for CSA	In the age group >68, presence of heart failure was associated with increases in PLMI and PLMAI, even after adjustment for age and severity of HF.
Murase et al.973	2014	3b	Case–control study	46 patients with OSA and PLMs were compared to 208 patients with OSA without PLMS	Compare inflammatory markers in OSA patients with vs. without PLMs	The OSa and PLMS group had higher CRP and fibrinogen levels compared to the OSA group without PLMS.
SRRMD; Aggregate level of evidence: D; (Level 4: one study)
Chiaro et al.978	2017	4	Case Series	5 patients with diagnosed SRRMD with RMEs seen following OSA events	Investigate the role of sleep apnea as a trigger for rhythmic motor events as a respiratory related arousal mechanism	SRRMD in adult patients may imply sleep-disordered breathing, possibly associated with longer respiratory events.
Bruxism; Aggregate level of evidence: B; (Level 2a: two studies)
Lopes977	2019	2a	Systematic review	Systematic review; 200 articles identified after initial search and seven included in the qualitative synthesis	Examine the association between sleep bruxism and OSA	Well-designed studies are lacking but based on available evidence, OSA patients do not experience sleep bruxism significantly more than controls. A subtype of patients with OSA may have sleep bruxism. Occurrence of sleep bruxism has been proposed as a protective mechanism for respiratory events.
Jokubauskas976	2017	2a	Systematic review	Systematic review; 691 articles identified after initial search and three included in the synthesis	Examine the association between sleep bruxism and OSA	There are insufficient data to establish an association between OSA and sleep bruxism. Sleep bruxism events occur during microarousal events resulting from apneas/hypopneas and most often are temporally related to the termination of an apnea/hypopnea.

VII.I.3 |. Narcolepsy

Narcolepsy is a chronic neurologic disorder characterized by excessive daytime sleepiness and abnormal regulation of REM sleep. Narcolepsy type 1 (NT1) refers to narcolepsy with cataplexy and NT2 to narcolepsy without cataplexy. NT1 occurs with a prevalence of 25–50 per 100,000 people.⁹⁸³ It is associated with a deficiency of hypocretin, also called orexin, which is a neuropeptide hormone secreted by a small group of cells located in the lateral hypothalamus. NT1 is thought to be autoimmune in nature; though no antibody has been detected, there is a strong HLA association to HLA-DQB1*06:02.⁹⁸⁴ The primary diagnostic test for NT1 and NT2 is the MSLT, consisting of overnight PSG followed by a series of five daytime naps at 2-h intervals and assessing average sleep latency across the naps and the number of sleep-onset REM periods (SOREMPs), that is, REM during the naps.

Patients with narcolepsy have an increased risk of multiple comorbidities compared to age matched controls including obesity, diabetes, depression, thyroid disease, and hypertension; however, OSA is the most common comorbidity, both at time of diagnosis and follow-up.⁹⁸⁵ The incidence of OSA is approximately 25% in those with narcolepsy and nearly half of these patients have moderate to severe OSA.^986,987 At the time of the initial diagnosis of narcolepsy, the odds of having a diagnosis of OSA compared to matched controls is markedly increased, ranging from 18 to 69.^985,988,989

There is on average a 6–10-year delay to diagnosis for those with narcolepsy,^986,990 which underscores the need to be alert to the frequency with which OSA and narcolepsy can co-occur. If treatment of OSA results in no or minimal improvement in excessive daytime sleepiness, the provider should consider the possibility of another sleep disorder, such as narcolepsy or other central disorder of hypersomnolence.⁹⁸⁷ The presence of cataplexy, unique to NT1, can also assist in this differentiation. If pursuing PSG and MSLT to quantify sleepiness in a patient with treated OSA and persistent excessive daytime sleepiness, the test result validity is maximized when OSA treatment is used during the polysomnogram and during MSLT naps.

It is also possible for a patient to be misdiagnosed with narcolepsy when the etiology of sleepiness is OSA alone, as polysomnographic findings similar to narcolepsy can result from untreated or undertreated OSA.⁹⁹¹ Five percent of those with sleep related breathing disorders can have two or more SOREMPs on an MSLT, suggesting REM periods during naps are less specific to narcolepsy than initially believed.⁹⁹¹ Male sex, sleepiness, short REM latency on nocturnal polysomnogram, and minimal SaO₂ are risk factors for two or more SOREMPs in a patient with OSA.

Among the treatments for narcolepsy are daytime stimulant medications and sodium oxybate. While sodium oxybate can improve excessive daytime sleepiness and cataplexy, it is also known to be a respiratory depressant at higher doses, raising concern that it could worsen sleep apnea. Though current data is mixed regarding this possibility,^992–995 the use of sodium oxybate in a patient with narcolepsy could potentially contribute to the development or worsening of OSA and repeat screening for this may pursued in certain clinical contexts.

In summary, OSA is a common comorbidity in patients with narcolepsy. OSA is a more common explanation for excessive daytime sleepiness than narcolepsy based on population prevalence. However, in patients with excessive daytime sleepiness, it is imperative to confirm efficacy of OSA treatment before ruling out narcolepsy. (Table VII.I.3)

TABLE VII.I.3.

Narcolepsy and prevalence of comorbid sleep apnea

Study	Year	LOE	Study design	N/age/cohort	Sample population	Conclusion
Sansa et al., 2010987	2010	2c	Cross-sectional	N = 133; ages 11–80	Single university hospital sleep clinic	AHI > 10 events/h in 24.8% of those with narcolepsy, 30% of these led to a delay in diagnosis of 6.1 ± 7.8 years. Excessive daytime sleepiness improved in 21% of those treated with CPAP.
Frauscher et al., 2013986	2013	2c	Cross-sectional	N = 100, ages 16–78; Innsbruck narcolepsy cohort	Single center, academic facility, tertiary referral center	24% of those with narcolepsy had sleep-related breathing disorders: 14 had mild sleep apnea, eight moderate sleep apnea, and two had severe sleep apnea. The majority had obstructive sleep apnea syndrome (21/24); 2/24 had mixed sleep apnea syndrome; and 1/24 central sleep apnea syndrome.
Jennum et al., 2013996	2013	3b	Case–control	N = 757; age <20 to >80; Danish National patient registry	Patients diagnosed with narcolepsy in Denmark	Prior to narcolepsy diagnosis OR for sleep apnea diagnosis 44.5, 95% CI 13.1–151.3; after the narcolepsy diagnosis OR for sleep apnea 19.2, 95% CI 7.7–48.3.
Pizza et al., 2013997	2013	2c	Cross–sectional	N = 35; mean age 40 ± 16	Narcolepsy with cataplexy patients seen at a university hospital	31% of patient had sleep-disordered breathing; Mild SDB was diagnosed in 4 (11%) subjects, moderate SDB in 4 (11%), and severe SDB in 3 (9%). In five out of 11 (46%) patients with SDB, apneas were mostly obstructive, in one it was central. The remaining five patients (46%) showed mixed, obstructive and central apneas.
Black et al., 2017988	2017	3b	Case–control	N = 9321; mean age 46.1 ± 13.3	US medical claims data (private insurance and Medicare)	Sleep apnea was present in 51.4% of those with narcolepsy and 5.8% of matched controls, 45.6% excess prevalence, p < 0.0001, OR 18.7, 95% CI (17.5–20.0).
Jennum et al., 2017989	2017	3b	Case–control	N = 339; 20–50 years old; Danish National Patient Registry	Patients diagnosed with narcolepsy in Denmark	Prior to narcolepsy diagnosis OR for sleep apnea 34.6, 95% CI 18–66.5; after narcolepsy diagnosis OR 35.2 for sleep apnea, 95% CI 19.4–63.9.
Cohen et al., 2018985	2018	2b	Prospective cohort	N = 68; ages 5–74; Rochester Epidemiology Project	Population-based in Olmsted Count, MN	At time of narcolepsy diagnosis OR of having OSA 69.25, 95% CI 9.26–517.99; OR of having OSA after observation period 13.55, 95% CI 5.08–36.14.
Filardi et al., 2020998	2020	3b	Case–control	N = 38; ages 2–18	Children and adolescents with narcolepsy with cataplexy at academic medical center	No difference in prevalence of sleep-disordered breathing between narcolepsy and controls.

VII.J |. Conditions Associated with OSA: Pregnancy

Due to a number of physiologic and hormonal changes, pregnancy is a time of unique vulnerability among women for SDB. High estrogen levels can lead to nasal congestion (i.e., rhinitis of pregnancy)⁹⁹⁹ and may contribute to UA narrowing.¹⁰⁰⁰ While high progesterone levels increase ventilatory drive, whether this effect promotes or protects against respiratory instability and SDB during sleep is uncertain.¹⁰⁰¹ Diaphragmatic elevation leads to decreased functional residual capacity, which in turn contributes to reduced oxygen reserve; these changes become more pronounced in the supine position and during late pregnancy.¹⁰⁰¹

Both subjective and objective measures of SDB increase with advancing pregnancy.^1002–1004 By the end of the second trimester, women report increases in the frequency of snoring, gasping or snorting, and witnessed apneas compared to the first trimester¹⁰⁰²; these symptoms generally further increase until delivery. Recently, when more than 3300 women underwent home sleep testing in the prospective, multicenter SDB substudy of the Nulliparous Pregnancy Outcomes Study (nuMoM2b), the prevalence of SDB (AHI ≥ 5 events/h) increased from 3.6% in early pregnancy to 8.3% in mid-pregnancy.¹⁰⁰³ The vast majority of SDB cases in the nuMOM2b cohort were mild,¹⁰⁰³ a similar finding to other studies.¹⁰⁰¹ Older age, higher BMI, larger neck circumference, non-Hispanic black race, smoking, and chronic hypertension were all associated with increased AHI.¹⁰⁰³ Other studies have identified similar risk factors for gestational OSA.^1005,1006

An accumulating body of data demonstrates an increased risk of cardiometabolic complications among women with gestational SDB.¹⁰⁰¹ Women in the nuMOM2b study with SDB in early or mid-pregnancy were nearly twice as likely to develop preeclampsia compared to women without SDB. The odds for hypertensive disorders of pregnancy (gestational hypertension or preeclampsia) were increased only among women with mid-pregnancy SDB, and not among women with early pregnancy SDB. The nuMoM2b study also observed an approximately three-fold increase in the odds for gestational DM among women with early and mid-pregnancy SDB, compared to women without SDB.

A recent metanalysis examined associations between subjectively and objectively measured SDB and multiple maternal and fetal outcomes, with preeclampsia, gestational hypertension, and gestational diabetes mellitus (GDM) having the largest numbers of studies available for inclusion (i.e., 15–20 individual studies).¹⁰⁰⁷ In these analyses, the risks for gestational hypertension, preeclampsia, and GDM were all increased among women reporting SDB symptoms.¹⁰⁰⁷ Among women with objectively diagnosed OSA, the risk for preeclampsia more than doubled, whereas risks for gestational hypertension and GDM nearly doubled.¹⁰⁰⁷ The risk for cesarean section also increased significantly among women with either subjective or objective SDB.¹⁰⁰⁷ Other studies have found increased risks for severe maternal morbidity including eclampsia, cardiomyopathy, pulmonary embolism, and in-hospital mortality.^1008,1009

Studies examining fetal outcomes have largely focused on preterm birth and growth-related outcomes, especially small for gestational age (SGA) births.¹⁰⁰⁷ A recent metanalysis found that the risk for preterm birth increased approximately 50% among women who reported subjective SDB symptoms or had been diagnosed with OSA.¹⁰⁰⁷ While some individual studies have observed increased risks for SGA infants among women with SDB, two meta-analyses did not find an increased risk for SGA births among women with SDB symptoms or objectively documented OSA.^1007,1010

Case reports and case series suggest that treatment of SDB during pregnancy improves maternal-fetal outcomes.^1011,1012 However, there is a paucity of controlled trial data. Several trials examining the effect of CPAP therapy on outcomes including BP, glycemic control, and CV risk are in progress.^1001,1013 The current approach to treatment is extrapolated from recommendations and data in the general population, as there are no pregnancy-specific guidelines. CPAP therapy is generally considered first-line treatment for gestational OSA, as it is widely effective and can be initiated quickly. Furthermore, auto-adjusting CPAP can accommodate changes in SDB severity during and after pregnancy. Women with preexisting OSA may also benefit from switching at least temporarily from fixed CPAP or OAs to auto-adjusting CPAP therapy during pregnancy to accommodate fluctuations in OSA severity. For women with positional (i.e., supine) SDB, the left lateral sleep position may be sufficient to avoid apneic events and maximize venous return via the inferior vena cava.¹⁰¹⁴

While pregnancies in women with OSA are widely considered high risk,¹⁰¹⁵ clinical practice guidelines specifically for management of pregnant women with OSA have not been offered. The use of continuous pulse oximetry during labor and the postpartum period can identify hypoxic episodes and guide management.¹⁰¹⁵ Early consultation with the anesthesia service can help to avoid the use of general anesthesia, which poses increased risks to individuals with OSA,¹⁰¹⁶ should delivery by cesarean section be indicated.

Postpartum, women receiving treatment for OSA should resume therapy as soon as is feasible. The use of opioids should be carefully considered, given increased risks for morbidity and mortality among OSA patients taking opioids.^1015,1017 In a group of women in the immediate postpartum period (within 48 h after delivery), elevating the upper body to 45° during sleep reduced the AHI significantly compared to sleeping in a non-elevated position.¹⁰¹⁸ This low technology intervention was well tolerated and can be easily recommended to patients.

While the severity of gestational OSA generally improves after delivery,^1019–1021 women diagnosed with or suspected of having OSA during pregnancy should be followed postpartum to determine whether OSA has resolved or requires further treatment.^1015,1017 Little is known about how frequently SDB during pregnancy persists or recurs afterwards, or about the effect of gestational SDB on risk later in life for adverse cardiometabolic outcomes such as hypertension and DM.

VIII |. MEDICAL TREATMENT FOR OSA

VIII.A |. Medical Management of OSA: Positive Airway Pressure (PAP) Therapy

VIII.A.1 |. Types of PAP for OSA

VIII.A.1.a |. Types of PAP: Continuous positive airway pressure (CPAP)

CPAP, the gold standard OSA treatment, acts as a pneumatic splint to maintain UA patency. Typically, after OSA diagnosis, CPAP is initiated with in-lab titration polysomnogram. An SR by Patil et al.³⁹ demonstrated a clinically significant reduction in disease severity as evidenced by complete resolution or near resolution of the AHI; improvement in subjective sleepiness (reduction in ESS score of 2.4 points based on metanalysis of 38 RCTs, and ability to maintain wakefulness based on the Maintenance of Wakefulness Test [MWT]); sleep related QOL; and BP. Two additional SRs^1022,1023 that studied the effect CPAP therapy had on QOL showed similar results. The randomized clinical trial by Ponce et al.¹⁰²² compared the use of CPAP to no CPAP in 154 patients and demonstrated that the use of CPAP decreased subjective sleepiness based on a reduction in the ESS by 2.6 points. A positive correlation between duration of CPAP use and reduction in ESS score was also seen. The Zhao study¹⁰²³ was an SR and demonstrated that the use of CPAP therapy in non-sleepy individuals with OSA improved health-related quality of life (HRQOL) by using the SF-36 questionnaire. Improvements were specifically seen in vitality, general health, bodily pain, and physical health. Given the extent of conflicting data about effect of CPAP use on CV outcomes, the SR by Patil et al.³⁹ noted that there is “insufficient and inconclusive evidence to either recommend or withhold PAP to treat non-sleepy adults with OSA as a means to reduce CV events or mortality.”

An SR of empirical studies with observational or experimental designs with monetized health economic outcomes of OSA treatments based on comparisons showed that, compared with no treatment, PAP was associated with favorable economic outcomes; these outcomes were greater with adherence to PAP therapy.¹⁰²⁴ In 2015, the average cost of CPAP inclusive of testing, appointments, treatment devices, and surgery, if necessary, was estimated at $2105 per patient per year.¹³² (Table VIII.A.1.a)

TABLE VIII.A.1.a.

Evidence for the use of CPAP to treat OSA

Study	Year		LOE	Study design	Study groups	Clinical endpoints		Conclusion
Wang et al.841	2019		2a	SR	1. CPAP 2. no therapy	Cognitive decline		CPAP likely improves memory impairment in patients with OSA and cognitive decline.
Ponce et al.1022	2019		1b	RCT	145 patients 1. CPAP 2. no CPAP	ESS, QSQ, sleep related symptoms, anxiety, and depression		CPAP therapy improved HRQOL measures in patients over the age of 70.
Patil et al.39	2019		1a	SR	336 studies that met inclusion criteria; 184 studies provided data suitable for meta-analyses	Multiple outcomes of sleep apnea		CPAP reduces disease severity, sleepiness, blood pressure, and motor vehicle accidents.
Zheng et al.1025	2019		1a	SR	17 studies comparing 1. CPAP, 2. control/placebo/sham CPAP in treatment of anxiety and depression in patients with OSA	Various measurements of anxiety and depression		CPAP reduced symptoms of depression.
Ning et al.1026	2019		1a	SR	14 RCTs evaluating effect of CPAP therapy on various cardiac biomarkers	Cardiac biomarkers		CPAP therapy improves levels of inflammatory biomarkers.
Khan et al.1027	2018		1a	SR	7 RCTs studying the use of CPAP in reducing MACE in patients with moderate to severe sleep apnea	MACE: MI, CV mortality		CPAP therapy of greater than 4 h a night significantly reduced MACE.
Wang et al.635	2018		2a	SR	9 studies (seven observational and two RCTs) studying cardiovascular outcomes in patients with coronary artery disease and OSA	MACE: cardiovascular death, stroke, hospitalization for heart failure		CPAP may decrease cardiovascular events in patients with CAD and OSA.
Labarca et al.1028	2018		1a	SR	6 RCTs studying effects of CPAP therapy in OSA and type II diabetes	HbA1c, fasting glucose		CPAP therapy does not significantly reduce HbA1c or fasting glucose levels.
Gupta et al.1029	2019		1b	RCT	70 patients with OSA and recent stroke randomized to 1. CPAP, 2. non-CPAP	New vascular events		CPAP therapy reduces the presence of new vascular events in patients with a stroke.
Congrete et al.1030	2018		2a	SR	7 studies reviewing the risk of recurrent atrial fibrillation after catheter ablation in patients with obstructive sleep apnea	Recurrent episodes of atrial fibrillation		Treatment of OSA with CPAP decreased the incidence of recurrent atrial fibrillation.
Joyerux-Faure et al.1031	2018		1b	RCT	37 patients with obstructive sleep apnea and resistant hypertension randomized to 1. CPAP 2. Sham CPAP	Blood pressure		Treatment with CPAP lowered nocturnal systolic blood pressure.
Zhao et al.1023		2017	1b	RCT	169 Patients with moderate to severe OSA and CVD randomized to 1. Medical therapy 2. Medical therapy + sham CPAP 3. Medical therapy + CPAP 4. Medical therapy + CPAP + motivational enhancement.	Health-related quality of life	CPAP improves health related QOL metrics.
Abuzaid et al.1032		2017	1a	SR	4 RCTs studying cardiovascular outcomes in patients with moderate to severe OSA treated with CPAP	MACE	CPAP therapy did not reduce the number of cardiovascular events.
Campos-Rodriguez et al.1033		2017	1b	RCT	307 women with moderate to severe OSA randomized to 1. CPAP 2. Conservative therapy	Blood pressure	CPAP therapy significantly decreases diastolic blood pressure.
Wickwire et al.1024		2019	1a	SR	Empirical studies with observational or experimental designs with monetized health economic outcomes of OSA treatments based on comparisons	Health economic outcomes	Relative to no treatment, PAP was associated with favorable economic outcomes. PAP adherence was positively associated with greater economic outcomes.
Hoyos et al.1034		2015	1a	SR	7 RCTs studying the effect of CPAP therapy on blood pressure	Blood pressure	CPAP therapy reduces nocturnal systolic and diastolic blood pressures.

Abbreviations: APAP, auto-titrating CPAP; BPAP, bilevel PAP; CAD, coronary artery disease; CPAP, continuous PAP; CV, cardiovascular; ESS, Epworth Sleepiness Scale; HbA1c, glycated hemoglobin; HRQOL, health-related quality of life; MACE, major adverse cardiovascular events; MI, myocardial infarction; OSA, obstructive sleep apnea; PAP, positive airway pressure; QOL, quality of life; QSQ, Quebec Sleep Questionnaire; RCT, randomized control trial.

VIII.A.1.b |. Types of PAP: Auto-titrating positive airway pressure (APAP)

Two SRs^39,1035 evaluated the differences between APAP and CPAP devices. Both modalities resulted in a clinically significant reduction of the AHI, sleepiness as measured by the ESS, and QOL. There was no significant difference between the two treatment modalities. The randomized equivalence trial of 208 sleepy patients (mean ESS 13.1) by Bloch et al.¹⁰³⁶ comparing APAP to CPAP demonstrated reductions of ESS scores of 6.3 and 6.2, respectively. The systemic review by Ip et al.¹⁰³⁵ demonstrated that APAP reduced ESS score by 0.5 points when compared to CPAP, though this increased benefit is likely not clinically relevant. One of the presumed benefits of APAP therapy is that it will lead to increased compliance due to its ability to automatically adjust pressures based on patient need. The SR by Patil et al.³⁹ did not demonstrate any difference in PAP adherence as measured by hours used, nights used, or nights used with greater than 4 h of usage per night. While the SR by Ip et al.¹⁰³⁵ did show an 11-min increase in PAP usage per day with APAP therapy, this is likely to be clinically insignificant. These studies did not evaluate peak or mean pressures achieved during APAP use. In addition, an RCT demonstrated that APAP was not found to lower mask leak rates when compared to CPAP.¹⁰³⁷ Additional RCTs^1036,1038 demonstrated similar findings in the reduction of sleepiness¹⁰³⁶ and improvement in QOL measures.¹⁰³⁹

APAP was shown to have no added benefit in BP reduction when compared to CPAP. The large meta review performed by Ip et al.¹⁰³⁵ included three RCTs that evaluated the effect of APAP and CPAP on BP and found that there was no significant difference between the two modalities. The RCT by Pepin et al.¹⁰³⁸ was a double-blind RCT, which randomized 322 patients with OSA to receive either APAP or CPAP, found reductions in diastolic BP with CPAP but not APAP in the intention to treat analysis.

Patil et al.³⁹ examined the comparison between in-lab titration of CPAP and home APAP (without lab titration) for initiation of PAP therapy. Analysis showed similar effects on OSA severity, sleepiness, and adherence with APAP initiated at home. Patients value preferences and resource utilization are considerations when choosing in-home APAP initiation versus CPAP titration. CPAP or APAP for the treatment of OSA is recommended. (Table VIII.A.1.b)

TABLE VIII.A.1.b.

Evidence for APAP for OSA

Study	Year	LOE	Study design	Study groups	Clinical endpoints	Conclusion
Wimms et al.1038	2019	1b	RCT	301 patients with mild OSA randomized to 1. APAP + standard care, 2. standard care	QOL from SF-36	APAP usage improved QOL in patients with mild obstructive sleep apnea.
Patil et al.39	2019	1a	SR	26 RCTs comparing APAP to CPAP	OSA severity, adherence, sleepiness, QOL, neurocognitive function	No difference between APAP and CPAP.
Lebret et al.1037	2019	1b	RCT	276 patients with OSA randomized to 1. CPAP, 2. APAP	Mask leak	PAP modality did not affect level of mask leak.
Bloch et al.1036	2018	1b	RCT	208 patients randomized to 1. CPAP, 2. APAP(5–15 cmH2O)	Subjective and objective sleepiness	APAP is effective in treating EDS.
Pepin et al.1039	2016	1b	RCT	322 patients with OSA randomized to 1. CPAP, 2. APAP	Difference in office based systolic blood pressure after 4 months of therapy	APAP can lower office based systolic blood pressure.
Ip et al.1035	2012	1a	SR	24 RCTs comparing CPAP to APAP	Adherence	Eleven-minute increase in PAP adherence with APAP.

Abbreviations: APAP, auto-titrating CPAP; BPAP, bilevel PAP; CPAP, continuous PAP; OSA, obstructive sleep apnea; PAP, positive airway pressure; QOL, quality of life; RCT, randomized control trial.

VIII.A.1.c |. Types of PAP: Bilevel PAP

Few studies included in the SR studied bilevel PAP (BPAP). An SR of PAP treatments for OSA found no benefit for BPAP over CPAP.³⁹ Review of studies showed no significant difference between BPAP, CPAP, and APAP for the improvement in OSA severity, sleepiness, and QOL. BPAP also confers no significant advantage over CPAP or APAP in improved adherence, except as a potential therapy option for patients nonadherent to CPAP.³⁹ Patients who require high PAP levels greater than what CPAP devices can deliver may benefit from the use of BPAP. For patients with routine OSA, BPAP devices may cost more and may not provide sufficient reduction in the AHI if the expiratory pressure setting is too low, thus BPAP should not be considered in these patients.

A single center RCT that studied an enhanced education support program for patients with OSA reported no difference in PAP adherence, nightly duration of PAP use or reduction in subjective daytime sleepiness between use of BPAP, CPAP, or APAP.¹⁰⁴⁰

BPAP may be indicated when CPAP is not tolerated. Ballard et al.¹⁰⁴¹ examined 100 patients who were persistently noncompliant with CPAP after mask optimization, humidification, and education measures and performed a double-blind randomized trial of standard CPAP versus BPAP use and showed more participants in the BPAP group achieved compliance of PAP use more than 4 h per night. (Table VIII.A.1.c)

TABLE VIII.A.1.c.

Evidence for use of BPAP for OSA

Study	Year	LOE	Study design	Study groups	Clinical endpoints	Conclusion
Patil et al.39	2019	1a	SR	5 RCTs comparing BPAP or auto-BPAP to CPAP.	OSA severity, treatment adherence, sleepiness, QOL	No difference between BPAP or auto-BPAP and CPAP for OSA severity, sleepiness, QOL.
Sarac et al.1040	2016	1b	RCT	115 patients with OSA randomized to standard care or enhanced educational support.	PAP adherence rates, PAP hours/night, reduction in Epworth Sleepiness Scale Score	No stated difference between BPAP, CPAP, and APAP.
Ballard et al.1041	2007	1b	RCT	104 OSA patients persistently noncompliant with CPAP: double blind randomized CPAP vs. flexible BPAP.	Compliance with PAP	More on BPAP 25/51 (49%) achieved compliance vs. 15/53 (28%) randomized to CPAP.

Abbreviations: APAP, auto-titrating CPAP; BPAP, bilevel PAP; CPAP, continuous PAP; OSA, obstructive sleep apnea; PAP, positive airway pressure; RCT, randomized control trial; QOL, quality of life.

VIII.A.1.d |. Types of PAP: Adaptive servo-ventilation

ASV is a form of auto-adjustable bilevel positive pressure support with a backup rate which is used to treat central sleep apnea syndromes (CSAS). ASV devices provide expiratory positive airway pressure (EPAP) to achieve and maintain UA patency during sleep as well as variable inspiratory pressure support (IPS) and auto-backup rate (BUR) in order to stabilize the respiratory control circuit and alleviate cyclical episodes of hyperand hypoventilation. The EPAP supplied by ASV devices is either titrated to the critical airway opening pressure needed to achieve UA patency during an attended PSG or auto-adjusted by the device (if equipped with auto-EPAP capabilities).^1042,1043 IPS and the BUR are dynamically auto-adjusted, in relation to the patient’s native respiratory efforts and device-specific minute-ventilation-associated targets, with the goal of preventing episodes of respiratory instability.^1044,1045 Importantly, the proprietary algorithms used for dynamic IPS and BUR adjustment have been shown to vary by device,^1045,1046 which could potentially lead to difficulties achieving therapeutic targets.¹⁰⁴⁵

ASV therapy is not recommended for uncomplicated OSA.³⁹ However, there is a limited body of literature to support its use in cases of persistent treatment-emergent central sleep apnea (TECSA).^1047–1055 TECSA is defined by the initial presence of OSA (obstructive AHI ≥ 5 events/h in at least 2 h of recorded sleep) followed by resolution of obstructive events with CPAP titration, and the subsequent appearance of central respiratory events with ongoing CPAP therapy (residual central AHI [CAHI] ≥ 5/h in at least 2 h of recorded sleep time).^81,1042 TECSA does not include patients with CHF-associated CSAS. ASV can be used in CSAS with concurrent HF and opioid-associated CSAS.¹⁰⁵⁶ Updated guidelines examined the increased risk for cardiac mortality in patients with an LVEF less than 45% and moderate to severe CSA predominant SDB. Guidelines recommend against the use of ASV in patients with LVEF ≤ 45%.¹⁰⁵⁷

VIII.A.2 |. PAP use: mask interfaces

CPAP therapy was originally designed for use with a nasal interface that would deliver positive pressure through the nares and act as a pneumatic splint, pushing the soft palate and tongue forward and away from the posterior oropharyngeal wall to prevent UA occlusion.³ Over time, numerous nasal masks, oronasal masks, nasal pillow interfaces, and even oral masks have been developed for use with CPAP, with the goal of maximizing mask tolerability, a major determinant of CPAP adherence.¹⁰⁵⁸ Nevertheless, questions about which types of CPAP interfaces are associated with the best adherence and about the effectiveness of oronasal and oral interfaces persist.

A 2006 Cochrane review of CPAP delivery interfaces for OSA identified only four randomized, controlled studies that met the stringent criteria for inclusion and concluded that the optimal form of CPAP delivery interface remained unclear.¹⁰⁵⁹ The authors indicated that nasal pillows or the Oracle oral mask could be useful alternatives for patients unable to tolerate conventional nasal masks. Similarly, they suggested that oronasal masks should not be recommended for initial use but could be considered among individuals whose nasal symptoms limited nasal mask use. Subsequently, an MA that included five randomized and eight non-randomized trials found that oronasal masks were associated with a significantly higher CPAP level, higher residual AHI, and poorer adherence compared to nasal masks.¹⁰⁶⁰ While data from a total of 4563 individuals with OSA was included, the majority of the data came from a single study.¹⁰⁶¹ Limitations including moderate to high heterogeneity between studies with regards to CPAP level and adherence were noted.

Most recently, as part of an SR focused on the use of PAP for treatment of OSA in adults, the AASM published a series of meta-analyses evaluating the effects of different PAP interfaces on outcomes including reducing AHI, improving adherence to PAP therapy and sleepiness.³⁹ These analyses utilized data from eight RCTs and three observational studies. Residual AHI was higher using oronasal masks compared to nasal interfaces, though the difference was not clinically significant. Adherence to CPAP was similar when intranasal (nasal pillows) and nasal interfaces were compared; however, adherence was significantly higher with nasal interfaces compared to oronasal interfaces. No significant differences in sleepiness were observed in comparisons of oronasal and nasal masks, nor in comparisons between nasal pillows and nasal masks. Data were insufficient to perform analyses examining differences in QOL or interface-associated side effects.

The current review identified eleven small (n = 14–85) randomized trials of different mask interfaces that examined outcomes including optimal CPAP pressure, residual AHI on CPAP, CPAP adherence (duration or frequency of use), ESS score, side effects, and patient preference.^1062–1072 Most were crossover trials in which participants were assigned to use CPAP with nasal masks and oronasal masks in random order^{1063,1065,1068,1069,1071}; two trials included a third arm using nasal pillows.^1064,1065

In one trial, oronasal mask use required significantly higher final CPAP pressures to treat OSA compared to use of nasal masks or nasal pillows.¹⁰⁶⁴ Another found no difference between oronasal and nasal masks in optimal CPAP pressures.¹⁰⁷¹ Four trials found significantly higher residual AHIs using oronasal masks compared to nasal masks, despite delivery of the same or similar CPAP pressures during both study arms,^{1063,1065,1069,1071} but differences were not clinically meaningful.

While some studies have observed significantly longer durations of CPAP use with nasal masks compared to oronasal masks,^1065,1068 others have found no significant differences.^1063,1069 Comparisons of objective and subjective sleep quality and daytime sleepiness have demonstrated more slow wave and TST during CPAP titration using nasal masks compared to oronasal masks,¹⁰⁷¹ better sleep quality using CPAP with nasal masks compared to oronasal masks,¹⁰⁶⁹ and less sleepiness (i.e., lower ESS scores) using nasal masks compared to oronasal masks.¹⁰⁶⁸ Patients also expressed an overall preference for nasal masks compared to oronasal masks.^{1068,1069,1071}

When nasal masks and nasal pillows were compared, no significant differences were observed in 95th percentile auto-titrating pressures.¹⁰⁷² Three randomized crossover trials observed no significant differences in mean nightly duration of CPAP use or residual AHI when comparing nasal masks to nasal pillows.^{1067,1070,1072} Patients reported similar or fewer adverse effects with nasal pillow use compared to nasal mask use.^1067,1070 Sleep quality,¹⁰⁷² ESS scores, and QOL^1067,1070 generally did not differ during nasal pillow use compared to nasal mask use. Patients did not express a clear preference for nasal pillows or nasal masks in these studies.^1070,1072

Oral masks have been compared to nasal masks in two randomized crossover studies.^1062,1066 Significant differences between the interfaces were not observed when effective CPAP pressure,¹⁰⁶² average duration of CPAP use^1062,1066 residual AHI during PAP titration,¹⁰⁶² and sleepiness¹⁰⁶² were compared. The side effect profiles differed: while oral mask use was associated with more dry mouth and throat, gum discomfort, and excessive salivation, nasal mask use was associated with more complaints of nasal congestion and mask leak.^1062,1066

Four observational studies comparing CPAP mask interfaces were identified.^{1061,1073–1075} The largest was a prospective cohort study of 2311 newly diagnosed OSA patients who were given a choice of nasal masks, oronasal masks, or nasal pillows.¹⁰⁶¹ Users of oronasal masks and nasal pillows were more likely to be non-adherent to CPAP (mean use <4 h/night) at follow-up compared to nasal mask users. Oronasal mask users were more likely than the other groups to report side effects and to find CPAP inconvenient. Two other studies have observed that effective CPAP pressure and residual AHI were higher with oronasal mask use compared to nasal masks¹⁰⁷⁴ and/or nasal pillows.¹⁰⁷⁵

In general, the data suggest that nasal interfaces should be utilized initially in patients starting CPAP therapy. The use of oronasal masks is associated with the need for higher levels of CPAP pressure, and with higher residual AHI. Compared to nasal masks, adherence in several studies was lower with oronasal mask use. Mostly, outcomes including CPAP adherence and residual AHI were similar when nasal pillows and nasal masks were compared. Data on use of oral masks was particularly sparse. Oronasal and oral masks may be appropriate for a select group of patients who are predominantly mouth breathers, or who have large air leaks during sleep due to mouth opening, but these studies have not been performed. (Table VIII.A.2)

TABLE VIII.A.2.

Summary of the evidence for PAP mask interfaces

Study	Year	LoE	Study design	Study groups	Clinical endpoint	Conclusion
Patil et al.39	2019	1a	Meta-analyses	8 RCTs, three observational studies	Residual AHI, CPAP adherence, sleepiness, QOL, side effects	No significant differences between NP and NM in residual AHI or adherence. Higher residual AHI and lower CPAP adherence using ONM compared to NM. No differences in sleepiness between NM and NP, or between NM and ONM.
Goh et al.1065	2019	2b	Randomized unblinded crossover	Newly diagnosed moderate-severe OSA (n = 85) 1. Nasal masks 2. Oronasal masks 3. Nasal pillows	CPAP adherence, residual AHI	Significantly better adherence with NM compared to ONM and NP. Significantly higher residual AHI with ONM compared to NM and NP.
Rowland et al.1069	2018	2b	Randomized unblinded crossover	Moderate-severe OSA (n = 48) 1. Nasal masks 2. Nasal masks plus chinstrap 3. Oronasal masks	CPAP adherence, residual AHI, sleepiness, patient satisfaction, sleep quality, side effects, patient preference	No significant different in CPAP adherence. Significantly higher residual AHI with ONM compared to NM or NM plus chin strap. No significant differences in sleepiness. Greater comfort, better sleep quality and overall preference for NM compared to ONM.
Andrade et al.1060	2018	1b	Meta-analysis	5 randomized, eight nonrandomized trials (n = 4563)	CPAP level, residual AHI, CPAP adherence	ONM associated with higher CPAP level, higher residual AHI, lower adherence than nasal masks.
Deshpande et al.1075	2016	4	Retrospective observational	OSA patients having CPAP titration studies (n = 358) 1. Nasal masks (34.6%) 2. Oronasal masks (46.1%) 3. Nasal pillows (19.3%)	CPAP therapeutic level, residual AHI	CPAP level higher for ONM compared with NP and NM. Residual AHI higher for ONM than NM and NP.
Bettinzoli et al.1074	2014	4	Retrospective observational	OSA patients with AHI ≥ 15 (n = 109) 1. Nasal masks (61.5%) 2. Oronasal masks (38.5%)	CPAP therapeutic level, residual AHI, mask preference	Patients using ONM had significantly higher therapeutic level and higher residual AHI.
Ebben et al.1063	2014	2b	Randomized unblinded crossover	Moderate-severe OSA naïve to CPAP (n = 14) 1. Nasal masks 2. Oronasal mask	Residual AHI, CPAP adherence	Residual AHI was significantly higher using ONM compared to NM. No significant difference in CPAP adherence.
Borel et al.1061	2013	2b	Prospective observational cohort	OSA newly prescribed CPAP (n = 2311) 1. Nasal masks (62.4%) 2. Oronasal masks (26.2%) 3. Nasal pillows (11.4%)	CPAP adherence, CPAP level, CPAP-related side effects	Both ONM and NP associated with higher risk of non-adherence than NM. ONM associated with higher pressures than NM or NP. Proportion reporting side effects significantly higher with ONM than NM.
Zhu et al.1072	2013	2b	Randomized unblinded crossover	OSA using CPAP with NM for ≥6 months (n = 20) 1. Nasal masks 2. Nasal pillows	CPAP daily usage, residual AHI, CPAP level, mask leak, mask performance/preference	No significant differences in average daily usage, pressure levels, mask leak, or residual AHI. Minor differences in side effects, no differences in patient preference.
Ebben et al.1064	2012	2b	Randomized unblinded trial	OSA with AHI >5/h (n = 55) 1. Nasal masks 2. Oronasal masks 3. Nasal pillows	Final pressure levels, AHI on CPAP, final mask leak from CPAP titration study	Patients titrated using ONM required significantly higher final pressure levels compared to NM or NP. Final pressures not significantly different between NM and NP. Final AHIs not significantly different. Mask leak greater using ONM compared to NP but not NM.
Ryan at al.1070	2011	2b	Randomized unblinded crossover	OSA (AHI ≥ 10/h) naïve to CPAP (n = 21) 1. Nasal masks 2. Nasal pillows	CPAP adherence, residual AHI, sleepiness, side effects, QOL, preference	No significant differences between NM and NP in adherence, residual AHI, sleepiness, side effects, QOL, or interface preference.
Teo et al.1071	2011	2b	Randomized single-blind crossover	OSA (RDI > 15/h) naïve to CPAP (n = 24) 1. Nasal masks 2. Oronasal mask	PSG titration study CPAP level, residual RDI, mask leak, SWS, TST	CPAP levels not significantly different between NM and ONM. Residual AHI, arousals, mask leak all significantly greater with ONM than NM. Significantly less SWS and TST with ONM. Greater satisfaction with NM than ONM.
Chai-Coetzer et al.1059	2006	1a	Systematic review	4 RCTs 1. 2 studies comparing NM vs. OM 2. 1 study comparing NM vs. NP 3. 1 study comparing NM vs. ONM	CPAP compliance, sleep physiological parameters, ESS, OSA symptoms, adverse effects, interface satisfaction	Optimum CPAP delivery interface remains unclear given the limited number of available studies.
Anderson et al.1062	2003	2b	Randomized single-blind crossover, intention to treat analysis	OSA (AHI > 20/h) naïve to CPAP (n = 21) 1. Nasal masks 2. Oral masks	PSG variables, questionnaires, compliance	No significant differences in PSG variables during CPAP titration between OM and NM. No differences in residual AHI, CPAP compliance, CPAP pressure, sleepiness, overall side effects, or mask preference.
Beecroft et al.1073	2003	4	Prospective observational	OSA (AHI > 5/h) naïve to CPAP (n = 98) 1. Nasal masks (66%) 2. Oronasal mask (7%) 3. Oral masks (23%)	Optimal CPAP level, residual AHI, self-reported adherence, mask comfort, side effects	No significant difference in optimal CPAP levels, residual AHI, self-reported usage, or satisfaction between NM, ONM, and OM users.
Khanna et al.1066	2003	2b	Randomized unblinded trial	Patients with OSA (RDI > 15/h, n = 38) 1. Nasal masks 2. Oral masks	CPAP compliance, overall satisfaction, side effects	No significant differences in CPAP compliance between NM and OM users. Overall satisfaction similar between groups, types of side effects differed by mask type. Dropout rates similar between groups.
Massie et al.1067	2003	2b	Randomized unblinded crossover	Patients with OSA (n = 39) 1. Nasal masks 2. Nasal pillows	CPAP compliance, residual AHI, sleepiness, QOL, side effects, overall satisfaction	Higher percentage of days using CPAP for NP compared to NM but no differences in overall adherence. No significant differences in residual AHI or QOL. Better sleep quality and overall satisfaction with nasal pillows.
Mortimore et al.1068	1998	2b	Randomized unblinded crossover	New OSA patients naïve to CPAP (n = 20) 1. Nasal masks 2. Oronasal mask	Compliance, symptoms on CPAP, side effects	Higher CPAP compliance and less sleepiness using NM compared to ONM. Significantly fewer side effects with NM compared to ONM.

Abbreviations: AHI, apnea hypopnea index; CPAP, continuous positive airway pressure; ESS, Epworth Sleepiness Scale; NM, nasal masks; NP, nasal pillows; ONM, oronasal masks; OM, oral masks; OSA, obstructive sleep apnea; PSG, polysomnogram; QOL, quality of life; RCT, randomized controlled trial; RDI, respiratory disturbance index; SWS, slow wave sleep; TST, total sleep time.

VIII.A.3 |. PAP titration paradigms

The immediate goal of an efficacious PAP titration is to determine the pressure required to eliminate apneas, hypopneas, and inspiratory flow limitation (IFL)^39,1077,1078 whilst preventing arousal and central apnea events. The midterm and longer-term goals are for patient acceptance of PAP therapy, continued effectiveness in eliminating respiratory events during sleep, ongoing patient adherence to therapy, and improvement in symptoms and outcomes related to SDB. PAP titration paradigms can be viewed along three dimensions: method of pressure changes (manual versus automatic), PAP titration setting (in-lab versus at-home), and time spent titrating pressure (partial or full night, or multiple nights).

The AASM has published several guidelines detailing practice parameters and guidelines for PAP titration^39,1079,1080 with the goal being to identify a pressure that reduces AHI to <5 events/h in all sleep positions and sleep stages. The gold standard PAP titration is an in-laboratory technician-guided, manual titration with continuous monitoring of airflow, leak, sleep, and SaO₂.¹⁰⁷⁹ This approach assumes a single PAP can be determined that maintains airway patency despite changes in sleep stage, sleep position, and other factors that change UA collapsibility such as weight change, seasonal allergies, use of medication, and alcohol. With the introduction of APAP devices this provided an alternative approach to identifying a therapeutic pressure where 90% or 95% of all titration pressures fall.¹⁰⁸⁰ The effective pressure values identified by manual titration versus auto-algorithms have not demonstrated systematic differences. Titration of BPAP is not addressed here as BPAP is not recommended over CPAP or APAP for routine treatment of OSA in the absence of sleep-related breathing disorders associated with hypercapnia.³⁹

An SR and MA of 10 RCTs comparing APAP at-home and in-lab PAP titration for initiation of PAP demonstrates high grade evidence that both approaches are effective in reducing OSA severity (AHI), daytime sleepiness, and for PAP adherence and QOL measures.³⁹ However, study participants were predominately those with moderate to high OSA severity and excluded patients with significant comorbidities. Of importance, these comparisons assume equivalent levels of patient CPAP education and mask fitting prior to titration plus equal opportunity for therapy support post-initiation.

Split-night titration, where the first couple of hours of the PSG are for diagnostic purposes and the remainder of the study involves PAP titration, compared to a fullnight titration, is not significantly different in terms of outcomes such as subjective sleepiness or adherence over time.^1081–1083 However, certain criteria must be met to initiate a split-night PSG, that is, moderate–severe OSA and least 3 h available for CPAP titration.⁵⁷² This results in the exclusion of those with mild OSA from the research comparing full night and split titration paradigms and contributes to the weak recommendation regarding the appropriateness of split night PSGs for all patients.⁵⁷² Additionally, there is recent evidence that the prevalence of TECSA tends to be higher for split-night compared to full night titration studies.^1042,1084 Nevertheless, there is no evidence that treatment-emergence central sleep apnea poses any harm to the patient.³⁹

Determination of effective pressures using equations that combine anthropomorphic and OSA severity variables have been suggested as alternatives to manual titration.¹⁰⁸⁵ Although these equations may provide a starting pressure for titration, their use as the sole determinant of a patient’s therapeutic pressure has not been accepted as they differ from manually determined pressures by up to 5 cmH₂O.^{1075,1086,1087}

From the perspective of the treating physician, avoidance of delays to therapy initiation is a pertinent driver for the chosen PAP titration paradigm. The only RCT assessing patient preference demonstrated a preference for home management⁶⁰² compared to an in-lab titration. As the necessity for increased telehealth technology grows, we will likely see a concomitant increase in telemonitoring approaches to PAP titration, perhaps including patientdirected titration paradigms.^1088–1090 Based on the above considerations that all show no clinically significant differences in outcomes between titration paradigms, the choice of strategy should be based on patient preferences and abilities, judgment of the sleep clinician, and availability of resources.

VIII.A.4 |. PAP adherence

VIII.A.4.a |. Measurement of PAP adherence

The use of PAP is objectively measured by the device and defined as the number of hours PAP is used at the prescribed pressure. This is most often reported as a single summary measure of the amount of time PAP was used across the 24-h day. While the term “use” is the most concise descriptor, treatment compliance and adherence are terms that are often utilized for prescribed medical interventions. Treatment adherence, implying a partnership between provider and patient, is the more commonly used term and is more consistent with the broader chronic illness medical literature.

CPAP adherence is in large part a behavioral measurement because the act of putting on a mask at night prior to sleep requires voluntary action. PAP adherence is therefore a measure of the extent to which an individual uses their prescribed medical device and can be categorized into 0 h of use, >0 h of use, or missing data. Missing data represents problems with data measurement (e.g., internal CPAP device problems) or transmission (e.g., poor cellular coverage). In these cases, the only way to know if PAP was used is to ask the patient directly. Some studies include a CPAP use questionnaire to determine if the patient had a medical reason to not use CPAP for some amount of time. In this case, a true 0 may be reclassified as missing data.

The quantity of PAP use as a marker of adherence requires some clarification. Measurement of the amount of CPAP use does not take into account the amount of sleep during a defined 24-h period. Because the CPAP device does not measure sleep, it is not known whether CPAP is used during wake or sleep periods. CPAP is often used during times of attempted sleep. To increase accuracy of measurement, one should consider combining CPAP with a reliable and valid measure of sleep duration to determine the percentage of the actual sleep period that CPAP is used to account for individual differences in sleep and wake. CPAP devices also do not measure respiratory effort, and unless equipped with an oxygen probe, do not monitor oxygenation.

The medical prescription for PAP includes a pressure mode and pressure level(s). Pressure modes include fixed and auto-adjusting continuous single and bilevel pressure modes. The amount of PAP use and the efficacy of PAP in controlling SDB can be affected by pressure mode, pressure level, and/or mask type and fit. In addition, it should be stated that the CPAP device provides measures of treatment efficacy, including residual AHI and mask leak.

Given the single measure of time a PAP device is used in a given 24 h period is the foundation of defining adherence, it is important to understand additional limitations that exist when objective device-reported measurements are not available. In this case, estimated CPAP use provided by patients (or their bedpartners) is the next best surrogate. The mean difference between subjectively and objectively measured CPAP adherence is 0.96 h (see Table VIII.A.4a), meaning that patients tend to overestimate their PAP use by approximately 1 h per night.

TABLE VIII.A.4a.

Subjective versus Objective measures of CPAP adherence

Study	Year	n	LOE	Study design	Clinical end-pointa	Conclusion
Kribbs et al.1091	1993	35	2b	Prospective cohort	SR: 6.3 ± 1.47; Obj: 5.1 ± 1.87	Subjective was 1.2 h more than objective
Rauscher et al.