INTRODUCTION — Sleep-disordered breathing consists of several separate disorders, including obstructive sleep apnea (OSA), central sleep apnea (CSA), both with and without Cheyne-Stokes respiration, and sleep-related hypoventilation, in which abnormal breathing events during sleep are associated with adverse clinical outcomes. While each of these disorders has a characteristic presentation, history and physical examination alone are insufficient to make a definitive diagnosis, and formal sleep testing is required for diagnosis [1,2].
The current reference standard for formal evaluation of sleep-disordered breathing is attended polysomnography (PSG). Rules governing the performance and scoring of PSG are published and reviewed on an ongoing basis by the American Academy of Sleep Medicine (AASM), and accredited sleep laboratories are required to follow these guidelines [3].
This topic will review the technical specifications, scoring, and summary terms used to characterize sleep-disordered breathing in adults undergoing attended PSG. The clinical features and diagnosis of sleep-related breathing disorders are reviewed separately. (See "Clinical presentation and diagnosis of obstructive sleep apnea in adults" and "Central sleep apnea: Risk factors, clinical presentation, and diagnosis" and "Clinical manifestations and diagnosis of obesity hypoventilation syndrome".)
Home sleep apnea testing and the role of PSG for the evaluation of nonrespiratory sleep disturbances are also reviewed separately. (See "Home sleep apnea testing for obstructive sleep apnea in adults" and "Polysomnography in the evaluation of abnormal movements during sleep" and "Polysomnography in the evaluation of parasomnias and epilepsy".)
RECORDED SIGNALS — The performance of polysomnography (PSG) requires monitoring of a number of physiologic signals, all of which are relevant in the assessment of sleep-disordered breathing (table 1). These signals are typically displayed in an organized montage for ease of viewing during study review (waveform 1).
Sleep staging — Electroencephalography (EEG), chin electromyography (EMG), and electrooculography (EOG) monitoring are required to stage sleep and identify arousals. At the minimum, staging sleep requires three EEG leads: one frontal, one central, and one occipital lead, although bilateral leads are commonly collected to allow for redundancy in the case of lead failure. (See "Stages and architecture of normal sleep", section on 'Sleep staging'.)
Sleep staging is helpful in evaluating sleep-disordered breathing. It allows the number of respiratory events to be adjusted based upon total sleep time in order to reliably determine the severity of disease. Sleep staging also permits the identification of respiratory events associated with EEG arousals (abrupt shifts in brain wave frequency representing a transient awakening from sleep). (See 'Arousals' below.)
Respiratory airflow — Respiratory airflow is typically monitored nonquantitatively using two methods: oronasal thermistry and nasal pressure transduction. The two methods provide complementary information, and the American Academy of Sleep Medicine (AASM) recommends use of both monitors for routine PSG [3]. Quantitative monitoring of airflow is possible using a pneumotachometer with a tight-fitting mask, but this approach is not used in clinical studies because it may be uncomfortable and interfere with sleep.
●Oronasal thermistry – Oronasal thermistry is used to identify and score apneas. It is the superior modality for detecting low levels of airflow. Thermistry takes advantage of the fact that exhaled air is warmer than environmental air as a result of being heated while in the lungs. By measuring fluctuations in air temperature at the nose and mouth, even modest levels of airflow can be detected.
Notably, the thermistor must not be allowed to touch the skin directly during monitoring; if contact occurs, the device will warm and be unable to detect fluctuations in temperature from exhaled air.
●Nasal pressure transduction – Nasal pressure transducers are the superior modality for measurement of subtle changes in airflow, as seen in hypopneas. They detect airflow by measuring variations in airway pressure: negative fluctuations in pressure represent inspiration, while positive changes signal expiration.
Compared with oronasal thermistry, nasal pressure transducers are less sensitive in detecting low levels of flow. This is in part due to an inability to detect purely oral flow, thereby mistakenly identifying minimal levels of nasal airflow as a total cessation of breathing [4].
The nasal pressure transduction signal may be altered with square root transformation, although this is not required. Square root transformation is helpful in reporting hypopneas, especially in low airflow situations.
Respiratory effort — Measurement of airflow alone is insufficient to distinguish between central and obstructive etiologies for variations in respiration during sleep. Concomitant assessment of respiratory effort during PSG can be performed using one of several recommended methods.
Most sleep laboratories use a combination of thoracic and abdominal respiratory inductance plethysmography (RIP) to provide an accurate representation of respiratory effort. RIP measures the change in magnetic field induced by an electric current running around the body as the thorax and abdomen change shape. Diaphragmatic and/or intercostal EMG can be used as an alternative or back-up sensor for respiratory effort [3].
An alternative technology, polyvinylidene fluoride (PVDF), does not require an external power source and is being increasingly adopted by sleep laboratories. Like RIP belts, PVDF generates an electrical signal with movement.
Esophageal manometry, while no longer routinely used in clinical studies, provides the only quantitative data on respiratory effort during PSG and remains a gold standard in the assessment of work of breathing [5]. It is performed by placing a pressure monitor in the esophagus during sleep testing to provide quantitative intrathoracic pressure measurements. In clinical studies, it has largely been replaced by nasal pressure monitoring, which has improved the ability to detect subtle sleep apnea that previously may have been missed without an esophageal probe.
Pulse oximetry — Pulse oximetry is a standard component of PSG, allowing for continuous monitoring of arterial oxyhemoglobin saturation (SpO2). The technology determines SpO2 using a dual-wavelength (red and infrared) light transmitter and receiver placed around a pulsatile arterial bed; the ratio of red to infrared transmission is used to derive SpO2.
Several locations are commonly used to sample SpO2, including an earlobe, a finger, or the forehead. The more peripheral the sampling location, the greater the delay between a respiratory event and the detected desaturation. This delay is more pronounced in patients with prolonged circulation time due to systolic heart failure.
Averaging time refers to the time window used as a signal filter for the oximeter to smooth out short-term fluctuations in SpO2, which are often artifactual. While longer averaging times are commonly used in the intensive care unit and other clinical environments, the use of shorter averaging times (less than three seconds) is critical for PSG oximetry in order to detect the transient desaturations commonly seen in sleep apnea.
Known sources of error in pulse oximetry have not been well studied in the sleep laboratory, but it seems reasonable to assume that factors identified in the general inpatient population also affect accuracy during PSG. Factors that may lead to a falsely normal or high oximetry reading and thereby underestimate hypoxemia include high glycohemoglobin A1c levels and dark skin pigmentation (see "Pulse oximetry"). No formal recommendations regarding changes in PSG protocols have yet been made for these or other factors, likely related to the absence of a more accurate alterative for noninvasive assessment of SpO2.
Ventilation — Assessment of ventilation is an optional component of PSG in adults and a recommended component of PSG in children [3]. Monitoring of ventilation in an adult should only be performed if there is significant clinical suspicion for hypoventilation, such as in patients with severe obesity and hypoxemia, or those with unexplained elevation of serum bicarbonate levels. (See "Clinical manifestations and diagnosis of obesity hypoventilation syndrome", section on 'Diagnostic approach'.)
Noninvasive techniques to monitor ventilation include transcutaneous and end-tidal carbon dioxide monitoring. These techniques have generally obviated the need for invasive monitoring with an arterial catheter during PSG, although both noninvasive methods have potential flaws.
Transcutaneous carbon dioxide measurements are derived electrochemically, and some older models require frequent recalibration, often several times during a nocturnal study. However, the newest generation of transcutaneous monitors can provide more reliable data throughout a standard PSG.
End-tidal carbon dioxide monitors do not require frequent recalibration, but assessments may be inaccurate in patients with significant nasal obstruction or underlying pulmonary disease with areas of dead space, which can lead to an underestimation of the arterial carbon dioxide level. Also, they may not provide accurate data when supplemental oxygen is used.
Cardiac rhythm — Electrocardiographic monitoring is routinely performed as a component of PSG. A single lead is typically sufficient to identify cardiac dysrhythmias; lead II is usually used for simplicity of placement [6].
IDENTIFICATION AND SCORING OF EVENTS
General principles — The identification of respiratory events on polysomnography (PSG) requires measurement of airflow as well as confirmation that the event occurred during sleep, as assessed by electroencephalography (EEG). Additional information is obtained from the assessment of respiratory effort, which is necessary to distinguish between obstructive and central events. Measurement of oximetry improves the reliability of event identification and is necessary for scoring certain types of events.
Arousals — Arousals from sleep range from full awakenings to transient EEG shifts from a deeper to a lighter stage of sleep. The formal definition of an arousal on PSG includes the following criteria [3]:
●There is an abrupt shift of EEG frequency including alpha, theta, and/or frequencies greater than 16 Hz (but not spindles) that lasts at least three seconds, with at least 10 seconds of stable sleep preceding the change.
●During rapid eye movement (REM) sleep, there should also be an increase in submental electromyography (EMG) activity lasting at least one second.
Arousals are generally counted and then divided by the total sleep time to give the number of arousals per hour of sleep (ie, arousal index). Although EEG arousals can be subtle, leading to a high degree of interobserver variability, they are an important measure of sleep disturbance and provide a baseline from which to assess treatment response [7,8]. Many sleep-related breathing events end with an arousal, which may contribute to the poor sleep quality and daytime sleepiness associated with sleep-related breathing disorders.
Apneas — An apnea is the cessation, or near cessation, of airflow. Scoring an apnea on PSG requires documentation of a 90 percent or greater decrease in airflow, compared with preceding signals, for a minimum of 10 seconds.
Apneas are further categorized as obstructive, central, or mixed based on respiratory effort during the event.
●An obstructive apnea is associated with evidence of continued respiratory effort throughout the event (waveform 2).
●A central apnea is associated with an absence of respiratory effort throughout the event (waveform 3).
●Mixed apneas are associated with an absence of respiratory effort during the initial part of the event, followed by the appearance of respiratory effort during the latter part of the event (waveform 4).
Hypopneas — Hypopnea is a reduction of airflow to a degree that is insufficient to meet the criteria for an apnea. Precise definitions of hypopnea have varied significantly over the last decade. It is important to know which definition is being used by a given laboratory in order to interpret test results in the correct context, as the number of hypopneas directly influences summary measures like the apnea-hypopnea index [9,10]. (See 'Measures of sleep-disordered breathing severity' below.)
The current version of the American Academy of Sleep Medicine (AASM) scoring manual recommends that a hypopnea (waveform 5) be scored when all three of the following criteria are met [3,8]:
●Airflow decreases at least 30 percent compared with the pre-event baseline
●The diminished airflow lasts at least 10 seconds
●The event is associated with either a 3 percent oxygen desaturation from baseline or an EEG arousal
Because of high interobserver variability in scoring arousals [11], an alternative definition that does not include an arousal criterion is considered acceptable [3]. This definition requires a 30 percent decrement in airflow for at least 10 seconds associated with a 4 percent oxyhemoglobin desaturation. Based upon an increasing body of data demonstrating that respiratory event-related arousals are associated with morbidity, even in the absence of related desaturations, use of this alternative definition (which is sometimes mandated by insurers) should be performed alongside, not in place of, the recommended criteria [8].
Of note, use of supplemental oxygen may limit the frequency of desaturations during respiratory events that would otherwise be scored as hypopneas using either of these rules. However, there are currently no formal guidelines for identifying hypopneas when oxygen has been used during the study.
Distinguishing obstructive from central events is significantly more difficult for hypopneas than apneas, and many laboratories choose not to subclassify hypopneas. Because central hypopneas are associated with a decrease but not a total cessation in respiratory effort, thoracic and abdominal movements will still occur, just as they do during obstructive events. Use of an esophageal pressure monitor may allow the distinction to be made more readily, but such monitoring is not often performed during clinical studies. (See 'Respiratory effort' above.)
For laboratories that choose to subclassify hypopneas, the AASM recommends scoring a hypopnea as obstructive when there are other findings consistent with upper airway narrowing, namely snoring, flattening of inspiratory flow, or thoracoabdominal paradox [3]. In the absence of any of these phenomena, a hypopnea is scored as a central event.
Respiratory effort-related arousals — Respiratory effort-related arousals (RERAs) are arousals that are associated with a change in airflow that does not meet the criteria for apnea or hypopnea. They are formally defined as events lasting at least 10 seconds associated with flattening of the nasal pressure waveform (suggesting flow limitation) and/or evidence of increasing respiratory effort, terminating in an arousal but not otherwise meeting criteria for apnea or hypopnea (waveform 6) [3]. (See 'Apneas' above and 'Hypopneas' above.)
Scoring of RERAs is listed as an option rather than a requirement in the AASM scoring manual [3], and the necessity of scoring RERAs is not completely certain. In addition, there may be greater interobserver variation in the scoring RERAs compared with apneas and hypopneas. For these reasons, many laboratories report the severity of sleep-disordered breathing in two ways: both with and without inclusion of RERAs.
Cheyne-Stokes respiration — Cheyne-Stokes respiration is a cyclic pattern beginning with an apnea, followed by crescendo respiratory rate and tidal volume, and ending with a gradually shrinking respiratory effort leading to another apnea (waveform 7) [12]. It is most commonly seen in association with congestive heart failure, neurologic disease, or use of sedative medications. (See "Central sleep apnea: Risk factors, clinical presentation, and diagnosis".)
Cheyne-Stokes breathing during sleep is scored in the presence of three consecutive central respiratory events, separated by the characteristic crescendo-decrescendo respiratory pattern [3]. At least five such apneas or hypopneas must occur on average per hour of recorded sleep, and the length of each crescendo-decrescendo cycle must be at least 40 seconds; longer cycle lengths are associated with more severe cardiac disturbance.
Hypoventilation — Although sleep-related hypoventilation can be suggested by the presence of persistent hypoxemia during sleep that is not present during wakefulness, it cannot be reliably detected without monitoring of carbon dioxide levels during sleep. As discussed above, noninvasive carbon dioxide monitoring has generally replaced invasive arterial blood sampling for this purpose. (See 'Ventilation' above.)
Hypoventilation during sleep is defined by the presence of an elevation of arterial (or surrogate) carbon dioxide levels to greater than 55 mmHg for at least 10 minutes during sleep or, in the context of a known diurnal carbon dioxide level, an elevation of 10 mmHg above baseline values for the same length of time [3].
Due to the normal physiologic changes associated with REM sleep, including impaired responsiveness to carbon dioxide and skeletal muscle relaxation, hypoventilation is most commonly seen during REM sleep.
The finding of sleep-related hypoventilation can be attributable to obesity hypoventilation syndrome, an underlying pulmonary or neuromuscular disorder, or the use of medications or substances that impair respiratory drive, among other potential causes. (See "Clinical manifestations and diagnosis of obesity hypoventilation syndrome" and "Evaluation of sleep-disordered breathing in patients with neuromuscular and chest wall disease" and "Sleep-disordered breathing in patients chronically using opioids".)
INTERPRETATION AND SUMMARY MEASURES
Measures of sleep quality — Polysomnography (PSG) interpretation requires documentation of total sleep time and distribution of sleep stages. This is necessary in part for reliable interpretation of the severity of sleep-disordered breathing, as the total number of respiratory events must be normalized for the total duration of sleep.
The relative distribution of the different sleep stages (N1, N2, N3, and rapid eye movement [REM]) can also offer insight on how the PSG should be interpreted. High quantities of N1 sleep may suggest sleep fragmentation due to sleep-disordered breathing. The relative quantity of REM-phase sleep is also important, since apnea severity tends to vary significantly with sleep stage. In particular, REM sleep tends to exacerbate obstructive physiology but attenuate the severity of central sleep apnea (CSA) and Cheyne-Stokes respiration.
Arousal frequency is typically reported on PSG, but its clinical significance is not clear. Normal values range widely, between 10 and 25 events per hour depending upon patient age [13]. In addition, supernormal arousal frequencies are not specific to sleep-disordered breathing and can represent alternative sleep pathologies or simply discomfort with the monitoring environment or equipment.
Measures of sleep-disordered breathing severity — In addition to a report of the total number of apneas, hypopneas, and respiratory effort-related arousals (RERAs), PSG reports typically contain a number of other variables derived from these and other data.
Apnea-hypopnea index — The apnea-hypopnea index (AHI) is calculated by adding together the number of apneas and hypopneas and dividing the sum by the total sleep time (in hours). The index can be subdivided into an obstructive AHI and a central AHI. Apnea and hypopnea indices (AI and HI, respectively) may also be reported separately.
The total AHI is the measurement most commonly used to characterize the severity of sleep apnea. In adults, an obstructive AHI greater than 15 (or greater than five, in the presence of appropriate signs, symptoms, or comorbidities) meets diagnostic criteria for obstructive sleep apnea (OSA) [14]. A formal diagnosis of CSA requires a central AHI of at least five events per hour. (See "Clinical presentation and diagnosis of obstructive sleep apnea in adults", section on 'Diagnosis' and "Central sleep apnea: Risk factors, clinical presentation, and diagnosis", section on 'Diagnostic criteria'.)
Sleep apnea severity may vary significantly throughout the night based upon other physiologic parameters, and many laboratories report AHI (and respiratory disturbance index [RDI]) separately according to body position (supine and non-supine) and sleep stage (REM and non-REM). Reviewing each of these variables separately may suggest alternative therapies for a particular patient's disease, such as strict maintenance of the non-supine position using pillows or other devices during sleep. (See "Management of obstructive sleep apnea in adults", section on 'Sleep position'.)
Documentation of sleep-disordered breathing associated with cardiac arrhythmias may lower the threshold for implementing treatment, even in the presence of more mild forms of sleep apnea. (See "Obstructive sleep apnea and cardiovascular disease in adults".)
Respiratory disturbance index — Many sleep laboratories choose to report the RDI, which is derived by adding the number of apneas, hypopneas, and RERAs, then dividing the sum by the total sleep time in hours. By including RERAs, the RDI is necessarily greater than the AHI. However, it may have a greater degree of interobserver variability due to less reliable scoring of RERAs. (See 'Respiratory effort-related arousals' above.)
Despite these limitations, data exist to suggest that RERAs may impair neurocognitive function [15], even though they may not be as reliably associated with incident cardiovascular morbidity. These data suggest that the RDI may provide different clinical information than the AHI.
Degree of hypoxemia — Many respiratory events are associated with oxyhemoglobin desaturation, and PSG reports also include metrics related to the degree of hypoxemia. Several studies suggest that metrics of sleep-disordered breathing severity referencing hypoxemia may better predict adverse outcomes than strict use of AHI or RDI [16,17].
Minimum (nadir) and mean oxygen saturations during sleep are recommended components of PSG reports [3]. Many laboratories also assess "hypoxemic burden" as reflected in the percentage of sleep time during which a patient has an oxyhemoglobin saturation less than a given level, such as 90 percent.
The oxygen desaturation index (ODI), reported by some laboratories, represents the number of times per hour of sleep that a patient's oxyhemoglobin saturation drops by more than a certain amount from baseline, typically 4 percent (ODI 4 percent). Some laboratories use a 3 percent drop and report an ODI 3 percent as well.
Presentation of summary data — Formal sleep reports summarize the data above in both graphical and tabular form. A graphical summary (figure 1) usually displays a summary of sleep staging, body position, pulse, and oximetry and marks the occurrence of sleep-disordered breathing events. The data are usually also shown in tabular form (figure 2), which often includes documentation of the frequency of respiratory events and may be broken down by sleep stage and body position.
Interpretation — The most important portion of the sleep laboratory report is the interpretation, whereby the reviewing clinician draws attention to the most important conclusions of the study.
Interpretation of numbers such as the AHI and minimum oxygen saturation, in the context of the available patient clinical history, cannot occur by algorithm or strict reference to insurer-mandated thresholds for treatment coverage. This is because many features of the raw data, apparent on visual inspection, cannot be summarized numerically and are not captured in report tables or graphs. As one of many possible examples, hypopneas that occur in association with phasic eye movements during REM sleep often meet scoring criteria, and contribute to the AHI. However, in the absence of associated arousals or substantial oxygen desaturation, they are thought to be normal physiologic events that must be discounted in assessment of the overall AHI.
The interpretation should indicate whether study results suggest clinically significant sleep-disordered breathing, and if so, what type (ie, obstructive or central). Many reports also indicate overall severity. Discrepancies between the interpreter's assessment of clinical significance and common thresholds for concern, with reference to the AHI or other reported numbers, are explained.
Shortcomings or limitations of the study are noted, as are ancillary findings that could sometimes, depending on the patient's clinical history, affect diagnosis and treatment for conditions other than sleep-disordered breathing. Such findings can become important when treatment for sleep-disordered breathing does not result in amelioration of a patient's presenting complaints. PSG findings related to nonrespiratory parameters such as abnormal movements and behaviors during sleep are reviewed in detail separately. (See "Polysomnography in the evaluation of abnormal movements during sleep" and "Polysomnography in the evaluation of parasomnias and epilepsy".)
SOCIETY GUIDELINE LINKS — Links to society and government-sponsored guidelines from selected countries and regions around the world are provided separately. (See "Society guideline links: Sleep-related breathing disorders in adults".)
SUMMARY
●Polysomnography (PSG) requires monitoring of a number of physiologic signals that are relevant to the assessment of sleep-disordered breathing, including sleep stage, respiratory airflow, respiratory effort, pulse oximetry, and ventilation. Technical specifications for monitoring and scoring of these signals are published and routinely updated by the American Academy of Sleep Medicine (AASM). (See 'Sleep staging' above and 'Respiratory airflow' above and 'Respiratory effort' above and 'Pulse oximetry' above and 'Ventilation' above.)
●Apnea is the cessation, or near cessation, of airflow. An apnea on PSG is defined as a 90 percent or greater decrease in airflow compared with preceding signals for a minimum of 10 seconds. Apneas are further subclassified as obstructive, central, or mixed based on the pattern of respiratory effort throughout the event. (See 'Apneas' above.)
●Compared with apneas, hypopneas are less complete reductions in airflow. Although definitions have varied over time, hypopneas are currently scored in the presence of a 30 percent decrease in airflow compared with the pre-event baseline, lasting at least 10 seconds, and associated with either a 3 percent oxygen desaturation from baseline or an electroencephalography (EEG) arousal. (See 'Hypopneas' above.)
●Respiratory effort-related arousals (RERAs) are events lasting at least 10 seconds associated with flattening of the nasal pressure waveform and/or evidence of increasing respiratory effort, terminating in an arousal but not otherwise meeting criteria for apnea or hypopnea. Interobserver variability in scoring RERAs may be higher than that for apneas and hypopneas. (See 'Respiratory effort-related arousals' above.)
●Cheyne-Stokes respiration is a regular, cyclic pattern of apneas separated by hyperpneas that show gradual, relatively symmetric increases and then decrease in tidal volume. This respiratory pattern is most commonly seen in association with congestive heart failure and neurologic disease. (See 'Cheyne-Stokes respiration' above.)
●Sleep-related hypoventilation is recognized and scored on PSG based on the presence of elevated carbon dioxide levels during sleep. (See 'Hypoventilation' above.)
●In addition to a report of the total number of apneas, hypopneas, and RERAs (and a breakdown into obstructive and central events as appropriate), PSG reports contain a number of other derived variables to provide an indication of the nature and severity of sleep-disordered breathing. These include the apnea-hypopnea index (AHI), respiratory disturbance index (RDI), and various measures of the degree of hypoxemia during sleep. (See 'Measures of sleep quality' above and 'Measures of sleep-disordered breathing severity' above.)
●The most important part of a PSG report is the interpretation, in which the reviewing clinician summarizes clinically relevant conclusions based on review of the recorded signals, results of scoring, and the clinical history. (See 'Interpretation' above.)
