Interpretation Of The Adult

Otolaryngol Clin N Am 40 (2007) 713–743

Interpretation of the Adult Polysomnogram Rahul K. Kakkar, MD, FCCPa,b,*, Gilbert K. Hill, RPSGTa a

North Florida South Georgia VAHS Sleep Disorders Center, #111A, Malcom Randall Veterans Administration Medical Center, 1601 Archer Road, Gainesville, FL 32608, USA b Division of Pulmonary, Critical Care and Sleep Medicine, University of Florida, PO Box 100225, JHMHC, Gainesville, FL 32610-0225, USA

The term ‘‘polysomnography’’ (PSG) has Greek and Roman roots, and refers to the recording of multiple sleep-related signals. It employs various methods to simultaneously and continuously record neurophysiological, cardiopulmonary, and other physiological parameters over the course of several hours, usually during an entire night (overnight polysomnography). PSG provides information on the physiological changes occurring in many different organ systems in relation to sleep stages and wakefulness. It allows qualitative and quantitative documentation of abnormalities of sleep and wakefulness, sleep-wake transition, and of physiological function of other organ systems that are influenced by sleep. Many of these, such as sleep apnea, may not be present during wakefulness. Four types of sleep studies are available, depending upon the number of physiological variables recorded [1]: Level I. Standard PSG with a minimum of seven parameters measured, including electroencephalogram (EEG), electro-oculogram (EOG), chin electromyogram (EMG), and EKG, as well as monitors for airflow, respiratory effort, and oxygen saturation. A technician is in constant attendance. Level II. Comprehensive portable PSG studies are essentially the same, except that a heart rate monitor can replace the ECG and a technician is not in constant attendance. * Corresponding author. North Florida South Georgia VAHS Sleep Disorders Center, #111A, Malcom Randall Veterans Administration Medical Center, 1601 Archer Road, Gainesville, FL 32608. E-mail address: [email protected] (R.K. Kakkar). 0030-6665/07/$ - see front matter. Published by Elsevier Inc. doi:10.1016/j.otc.2007.04.003

oto.theclinics.com

714

KAKKAR & HILL

Level III. Modified portable sleep apnea testing is a cardiorespiratory study in which a minimum of four parameters must be measured, including ventilation (at least two channels of respiratory movement, or respiratory movement and airflow), heart rate or EKG, and oxygen saturation. Ventilation in this case is measured with at least two channels of respiratory movement or of airflow. Personnel are needed for preparation, but the ability to intervene is not required for all studies. Level IV. Continuous (single or dual) bioparameter recordings where devices that measure a minimum of one parameter, usually oxygen saturation are utilized. This article focuses on interpretation of Level I studies (attended PSG). According to the recently revised criteria published by the American Academy of Sleep Medicine (AASM), the following parameters should be recorded: Electroencephalogram (EEG) derivations (frontal, central and occipital) (Fig. 1), bilateral Electrooculogram (EOG) (Fig. 2), Chin electromyogram (EMG) (Fig. 3), leg EMG, airflow, respiratory effort, oxygen

Fig. 1. The EEG electrodes are placed using the conventional 10–20 system. For sleep recording and staging, the current standards require monitoring of at least central (C3, C4) and occipital (O1, O2) derivations. Additional EEG electrodes can be placed depending on the need. The new AASM standards to be published in spring 2007 will require frontal derivations to be recorded as well. (Courtesy of Compumedics USA, El Paso, Texas; with permission.)

INTERPRETATION OF THE ADULT POLYSOMNOGRAM

715

Fig. 2. Placement of EOG electrodes. See text for details. (Courtesy of Compumedics USA, El Paso, Texas; with permission.)

saturation and body position [2]. EKG is recorded routinely during all PSG studies. Video monitoring during a PSG, although not required, is extremely valuable, both from a diagnostic as well as medico-legal perspective. Additional variables can be recorded according to patient age, standards of the sleep laboratory, and indication for performing the study. Examples

Fig. 3. Placement of submental EMG electrodes. A third electrode is often used and placed under the chin. (Courtesy of Compumedics USA, El Paso, Texas; with permission.)

716

KAKKAR & HILL

of such additional variables include end-tidal carbon dioxide monitoring (EtCO2), 16-channel EEG recording for seizures, esophageal pressure monitoring (Pes), and pulse transit time (PTT). AASM [formerly The American Sleep Disorders Association (ASDA)] recommends performing at least 6 hours of overnight recording [1]. The neurophysiological, EKG, and sound channels are sampled at 100 to 1000 Hz with 12- to 20-bit resolution, respiratory mechanical channels at 50 Hz, and pulse oximetry and body position, at a significantly lower rate. The modern day computerized PSG systems provide powerful tools for technicians and physicians to customize data acquisition. One can change reference electrodes, resolution, and sensitivity, and add or remove digital filters to minimize recording artifacts. Many of these functions can be performed even after the data have been acquired, which allows interpretation of the studies despite malfunction of certain equipment during the recording. EEG is the recording of surface electrical activity of the brain. Only limited EEG data are obtained during a PSG recording to help identify stages of sleep and wakefulness. EOG is the recording of eye movements during sleep and wakefulness. The cornea and retina form a dipole, with the cornea being negative in relation to the retina. A movement in the eyes changes the electrical signal in the EOG electrodes, which is recorded as a deflection. The EOG electrodes are placed slightly outside the outer canthus of each eye, with one electrode being slightly lower than the other (see Fig. 2). This enables detection of horizontal as well as vertical movements of the eyes. EEG, EOG, and submental EMG are essential for sleep staging. Submental and leg (tibialis anterior) EMG recordings are performed routinely during PSG. The submental EMG recordings are essential for scoring sleep stages, especially rapid eye movement (REM) sleep. Three electrodes are placed for submental EMG recordings, in order to have a backup in case one of them malfunctions during the sleep study (see Fig. 3). Typically, the submental EMG tone is lowest during REM sleep. It is also helpful in detecting sleep bruxism (see below). Bilateral leg EMG recordings are used to diagnose periodic limb movements of sleep (PLMS). Additional EMG recordings can be used under special circumstances. For example, both upper and lower extremity EMG can be recorded for suspected REM behavior disorder (RBD), and gastrocnemius muscle EMG could be recorded for diagnosing nocturnal leg cramps. Snoring is recorded with a microphone, and airflow is measured with the help of a thermistor or with a nasal pressure monitor. The thermistor measures changes in the electrical conductance in response to temperature changes in the probe, which occur with inspiration and expiration. Continuous recording of body position by an accelerometer is important, because snoring and upper airway obstruction during sleep are influenced by gravity [3,4]. The most common indication for performing a PSG is diagnosis of sleep disordered breathing (SDB) and its treatment with positive airway pressure (PAP). Other indications for PSG include: evaluation for effectiveness of alternative treatments for SDB (eg, dental

INTERPRETATION OF THE ADULT POLYSOMNOGRAM

717

appliances or surgical procedures, diagnosis of sleep-related seizures, parasomnias, and evaluation of erectile dysfunction with nocturnal penile tumescence [NPT]).

Important considerations Polysomnography is considered to be the ‘‘gold standard’’ for diagnosing SDB and other sleep disorders. A gold standard should have 100% sensitivity and specificity [5]; however, like most other diagnostic tests, PSG is not ideal, but rather the best available method to diagnose SDB. Furthermore, breathing disturbances may vary from night to night within certain limits [6,7]. Some sleep disorders such as nocturnal laryngospasm, sleep-related epilepsy, or parasomnias that occur episodically may be missed in a one-night recording [8]. A PSG should always be interpreted in light of a patient’s clinical history. The physician’s orders should be reviewed to determine the reason for obtaining the PSG and for the type of study requested. The interpreting physician should ascertain that the PSG was performed as requested. If a split-night study (diagnostic and PAP titration in the same night) was ordered but the patient has difficulty sleeping, demonstrates only mild obstructive sleep apnea (OSA), or does not tolerate the PAP trial, it is helpful to write a brief explanation as to why PSG was not performed as requested. This helps the referring physician to decide whether to order another study with PAP trial, or to refer the patient for alternative therapies. It is pertinent to review the referring physician’s notes and the prestudy questionnaire before interpreting PSG. The referring physician’s notes provide information on comorbid conditions that may increase the likelihood of prevalence of certain sleep disorders. For example, heart failure with ejection fraction of less than 40% increases the risk of having Cheyne-Stokes breathing, hypothyroidism increases the likelihood of having OSA, and Parkinson’s disease is associated with higher incidence of RBD. The physician’s notes are also helpful in explaining certain findings on the PSG, such as cardiac arrhythmias or abnormal oxygen saturation at baseline. Generally, each patient is given a pre-study questionnaire to fill out before performing the PSG. This questionnaire provides sleep history in greater detail than is usually available otherwise. The questionnaire is intended to obtain information on patient’s sleep-wake schedule and various sleep symptoms such as snoring, excessive daytime sleepiness, witnessed apneas, irresistible urge to move legs, dream enactment, teeth grinding, cataplexy, and so forth. The patient’s usual bedtime and rise time are taken into account when commenting on sleep latency and efficiency (see below). Some degree of insomnia can be expected in the unfamiliar environment of the sleep laboratory, especially on the first night (first-night effect) [9].

718

KAKKAR & HILL

Most questionnaires also incorporate the Epworth Sleepiness Scale (ESS) [10,11]. ESS has eight questions about subjective sleepiness under routine activities, and the patient answers them on a scale of 0 to 3, the highest possible score being 24 (Appendix, Table 1). A score of 12 or more is indicative of excessive daytime sleepiness, whereas a score of 10 to 12 is considered borderline. It should be noted that many patients who have SDB and low ESS score may show hypersomnia on objective measures such as the Multiple Sleep Latency Test (MSLT) [12]. The patient is also asked about any sedatives taken before the PSG. The authors do not administer, nor do we recommend administering, any medications (other than patient’s usual regimen) for the purpose of promoting sleep during PSG recording. Most patients are able to sleep for sufficient duration in the sleep laboratory to allow acquisition of interpretable data. Because most sleep centers do not have nurses at night, dispensing the medications and assessment of patients before discharge after the study may compromise patient safety if medications are administered for the sleep study. Sedatives have unpredictable side effects, and may impair the patient’s driving ability the next morning. They also skew the sleep architecture, can potentially worsen obstructive sleep apnea or attenuate central sleep apnea, RBD, and PLMS, thus interfering with the diagnosis. Other medications, which could affect sleep, should also be noted. For example, many antidepressants suppress REM sleep and increase EMG tone during REM sleep. The importance of technician comments cannot be overemphasized, because they provide an excellent source of additional information. The technician should fill out a checklist, which includes information about nocturnal oxygen usage, sedative, and alcohol usage among other things. The technician also makes notes of additional monitoring performed during the study (EtCO2, Pes) so that the reviewing physician can modify the montage when analyzing PSG data. Any technical difficulties encountered during the recording and any specific problems encountered by the patient, such as pain, panic, or breathing difficulty, that could be corroborated with the PSG findings are also mentioned. If a PAP titration is performed, the technician makes note of types of interfaces tried, interface with the best fit, patient preference for a specific interface, mouth breathing, use of humidifier, chinstrap, and use of bi-level PAP. Each time a change is made in PAP setting, the technician cites the reason for this. All this information should be taken into account for correct interpretation of a PSG. Last but not the least, a post-study questionnaire completed by the patient after the PSG provides an important insight into the patient’s overall experience. If PAP titration was performed, it is important to note the patient’s attitude toward PAP therapy. Specific problems reported by the patient during PAP titration (eg, nasal congestion, claustrophobia or pressure intolerance), can be specifically targeted and the treatment tailored to individual needs of the patient.

INTERPRETATION OF THE ADULT POLYSOMNOGRAM

719

Video recording The authors routinely perform video recording on all patients coming to our sleep laboratories, and a written consent for this is obtained from every patient. Video recording should begin from the time of starting the hookup, and should continue until the study is completed. It provides objective assessment of seizures, sleep-related movement disorders, and parasomnias. It can also be used to confirm the patient’s position in case of doubt. It may sometimes be possible to see mouth opening and paradoxical respiratory movements on the video. Video also provides added security to the patient and the laboratory personnel from inappropriate interaction or allegations of such an interaction. The video files typically consume a large amount of storage space. DVD and tape drives are the currently preferred methods of storing video data.

Electroencephalography and electro-olfactography A standard 10 to 20 EEG electrode placement system is used in placing the electrodes for recording of sleep (see Fig. 1). The recommended standard derivations for EEG recording are F4-M1, C4-M1, O2-M1. The EOG electrodes (ROC and LOC) are placed as shown in Fig. 2. Mastoid (auricular) electrodes are used as references. Conventionally right-sided electrodes are assigned even numbers (C4, O2) and left sided electrodes are assigned an odd number (C3, O1). The EEG electrodes could be referenced to mastoid electrodes (M1 and M2 [A1 and A2 in older terminology]) (see Fig. 1) or to a common reference electrode. Typically right-sided electrodes (O2, C4 and ROC) are referenced to left mastoid (M1) electrode and left sided electrodes (O1, C3 and LOC) are referenced to right sided electrodes (M2). The drawback of using mastoid electrodes is that EKG artifact commonly appears in these electrodes making it difficult to staging sleep in some old systems (Fig. 4). With the older systems, placing the auricular electrodes higher, jumping the two auricular electrode, or using a common reference electrode can minimize EKG artifact. The newer PSG systems use software to filter EKG artifact. Before reviewing digital PSG, it is important to ensure proper setting of the desktop display for resolution. It is recommended that the monitor for reviewing PSG should have a resolution of at least 1280  1024 and be 20 inches or larger for proper identification of waveforms and sleep staging and video reviewing [13]. The digital systems allow creation, storage, and individual configuration of several different montages for different users. One can adjust the sensitivity, trend color, and display range among many other variables. The display range is usually between 10 to 480 seconds. Sleep is usually staged according to the Rechtschaffen and Kales criteria in 30-second epochs [1]. In case of difficulty, reducing the display range to 10 seconds is helpful in recognition of EEG waveforms. On the other

720

KAKKAR & HILL

Fig. 4. EKG artifact (solid arrows). Note the presence of prominent EKG artifact in the EEG (C3, C4, O1, and O2) and EOG (ROC, LOC) channels. It persisted despite re-referencing the derivations to average (AVG) of all electrodes.

hand, detection of certain respiratory patterns is better appreciated by compressing the waveforms obtained over 60 to 480 seconds on the screen (Fig. 5). Another important consideration is checking the biocalibration data. Before the actual sleep recording is started, the technicians ask patients to perform certain voluntary maneuvers such as closing and opening of the eyes, moving the eyes vertically and horizontally, moving the legs, clenching the teeth, rapid breathing, and breath holding. This is useful in ensuring that all the electrodes are functioning and are recording the signal as intended. These waveforms serve an important reference for comparison later during the sleep study. The technicians also check and document impedance in the beginning and periodically during the study, to ensure accuracy of the recorded parameters. Sleep staging Until recently sleep staging was based on the original Rechtschaffen and Kales’ criteria devised more than 30 years ago. The AASM task force has recommended implementation of new criteria for sleep staging, which will replace the current staging system [2]. Sleep is now scored in the following stages; stage awake (W), NREM stages 1, 2, 3 ( N1, N2 and N3 respectively) and stage REM (R) sleep. EEG, EOG and chin EMG are required to score

INTERPRETATION OF THE ADULT POLYSOMNOGRAM

721

Fig. 5. Value of different display ranges. (A) Displaying the trends in 10-second range increases the resolution of EEG waveforms. Alpha waves are clearly recognized in the EEG (C3, C4, O1, and O2) tracings. Frequency is determined by counting the ‘‘peaks’’ in one-second interval. (B) Displaying the data in a compressed format (display range 480 seconds) accentuates the classical pattern of Cheyne-Stokes breathing.

the sleep stages. The identification of EEG waveforms is helpful in scoring sleep. The central leads (C4 and C3) are most helpful in the identification of EEG waveforms, except alpha waves, which are most prominent in the occipital leads. There are four different waveforms based upon the frequency of the rhythm, alpha waves are fast frequency (8–12 Hz) (Fig. 6). Beta waves (Fig. 7) consists of fast frequency of greater than 13 Hz. They are most

722

KAKKAR & HILL

Fig. 6. The arrow points towards alpha waves (30-second montage; vertical grid lines are 1 second apart). Note that they are most prominent in occipital leads.

Fig. 7. Beta waves (black arrows) recorded during a documented period of wakefulness during the PSG (see technician notations on top and bottom of (A). (A) 30-second and (B) 10-second display periods showing difference between beta and alpha (white arrows) rhythms.

INTERPRETATION OF THE ADULT POLYSOMNOGRAM

723

prominent during stage awake and eyes open. They are of low amplitude and disappear during sleep. The frequency for theta waves ranges from 4–8 Hz (Fig. 8) and for delta waves it is less than 4 Hz (Fig. 9). Theta frequency is the most prevalent frequency seen during a polysomnography. In addition to these four frequencies, four morphologic waveforms are important. Vertex waves are sharp negative waves (in EEG terminology, a negative wave causes an upward deflection) (Fig. 10). They first appear in stage N1 sleep and are most abundant during stage N2 sleep. K-complex is a biphasic wave with an initial upward (negative) deflection followed by a slower downward (positive) deflection (see Fig. 8). Sleep spindles are so-named because of their characteristic shape. They have a frequency of 12 to 14 Hz and duration of 0.5 to 1.5 seconds (Fig. 11). K-complexes and sleep spindles are characteristic of stage 2 sleep. Sawtooth waves also have a characteristic appearance (Fig. 12). They have low amplitude and are in the theta frequency. Staging of sleep is performed in 30-second epochs, and is more of a pattern recognition exercise. An epoch is assigned a sleep stage according to the predominant stage present in that epoch. Stage awake with open eyes is characterized by presence of low amplitude beta rhythm. Stage awake with eyes closed is characterized by prominent alpha waves (see Fig. 6). As the patient becomes drowsy, slow lateral eye movements (SEM) are often seen while alpha activity is still prominent (Fig. 13). Stage N1 sleep is the most difficult of all stages to recognize because it is not defined by the presence of any particular waveform, but rather by the disappearance of alpha

Fig. 8. Theta waves (black arrow) and K-complexes (white arrows) in a patient in stage 2 sleep (30-second montage; vertical grid lines are 1 second apart).

724

KAKKAR & HILL

Fig. 9. The solid arrows point towards the delta waves (30-second montage; vertical grid lines are 1 second apart). For sleep staging, the duration should be at least 0.5 sec (frequency 2 Hz or less) and amplitude should be more than 75 mV. The reference marker for voltage is displayed in lower right hand corner (broken arrow).

waves and the onset of a mixed frequency, low amplitude pattern. Other clues to the presence of stage N1 sleep are the appearance of SEM, a decrease in EMG tone, and the appearance of vertex waves (although they are not required to identify stage N1 sleep). Appearance of K-complexes and sleep

Fig. 10. Vertex waves, shown by the solid arrow, in a series (30-second montage; vertical grid lines are 1 second apart). Note that they do not have a component of positive (downward) deflection and are much narrower than K-complexes. They appear in stage 1 sleep, but are most prominent during stage 2 sleep.

INTERPRETATION OF THE ADULT POLYSOMNOGRAM

725

Fig. 11. Sleep spindles, shown by the solid arrow, (10-second montage; vertical grid lines are 1 second apart). Note the characteristic shape, frequency and duration.

Fig. 12. Sawtooth waves (solid arrow). Note characteristic appearance. They are in theta frequency and are most prominent during REM sleep.

726

KAKKAR & HILL

Fig. 13. Stage 1 sleep. Note the onset of low amplitude, mixed frequency waveforms in EEG channels and slow eye movements (SEM) in EOG channels.

spindles is diagnostic of stage N2 sleep. Previously stages 3 and 4 were scored separately. According to new terminology they are scored together as stage N3. This stage is also called delta sleep. These waves are characterized by amplitude of at least 75 mV, and must have a frequency of 2 Hz or less. Thus all delta waves do not qualify to be scored for the purpose of defining stage N3 sleep. If delta waves occupy more than 20% of the epoch, it is scored as stage N3; if they occupy more than 50% of the epoch, it is scored as stage 4 sleep. REM sleep is defined by the presence of rapid eye movements (Fig. 14) and decrease in submental EMG tone to its lowest level (REM atonia). It is important to differentiate REM from SEM activity, because the background EEG frequencies during REM sleep and stage 1 may look very similar to each other. Sawtooth waves are also seen predominantly during REM sleep, but are not required to score REM sleep. Abundance of sawtooth waves and absence of vertex waves helps differentiate REM sleep from stage 1 sleep. An arousal is defined as an abrupt change in the EEG frequency to alpha, theta, or faster frequency lasting at least 3 seconds [14]. If only delta waves or K-complexes occur, it is not scored as an arousal. An arousal in non-REM (NREM) sleep requires only a shift in the EEG frequency, whereas to score an arousal during REM sleep, both a shift in EEG frequency and an increase in submental EMG tone are required [14].

INTERPRETATION OF THE ADULT POLYSOMNOGRAM

727

Fig. 14. REM. Notice sharp deflections in EOG channels and compare them with SEM shown in Fig. 13. Normal ‘‘phasic’’ bursts in leg EMG during REM sleep are also present (arrows).

A sleep hypnogram is a summary of the entire night’s PSG data in a graphic form (Fig. 15). It gives a good snapshot of sleep architecture, distribution of respiratory events, and oxygen saturation trends in different sleep stages, sleep position, and at different times of the study night. It is

Fig. 15. Hypnogram. (Courtesy of Compumedics USA, El Paso, Texas; with permission.)

728

KAKKAR & HILL

helpful to open a window for the hypnogram simultaneously while reviewing the PSG. This allows one to easily select the part of the study one may wish to see. One can also switch the epochs rapidly for comparison of sleep stages and efficacy of different PAP settings. The most relevant parameters are: total recording time, total sleep time (TST), sleep period time (SPT), time in bed (TIB), sleep efficiency (SE), sleep latency and REM sleep latency, and sleep stage distribution. These are typically expressed in minutes (except sleep efficiency, which is a percentage). It is also important to note the percentage of time spent in the supine position. Total recording time is the interval from the beginning of the sleep recording to the end of the sleep recording. In addition to TST, it comprises the time taken up by wakeful periods. TST is the amount of actual sleep time in a sleep episode. TST is the total of all REM and NREM sleep in a sleep episode. SPT is the period of time measured from sleep onset to final awakening. In addition to the TST, it comprises the time taken up by arousals and movement time (MT) until wake-up. SE is the percentage of time actually spent sleeping during the period when the patient is given an opportunity to sleep. Thus, it is the percentage of TST during TIB. Sleep efficiency above 85% is considered to be normal. SE is reduced in the elderly and in patients who have insomnia. It is generally reduced in patients who have narcolepsy, as opposed to patients who have idiopathic hypersomnia, in whom very high sleep efficiency is seen (sometimes O99%). It could be high or low in patients who have SDB, depending upon the degree of sleepiness and the severity of sleep fragmentation with respiratory events. PAP and medications can variably affect SE, whereas psychiatric conditions, pain syndromes, breathing difficulties, and urinary symptoms usually decrease it. Sleep latency (SL) is the time elapsed between lights out to the first epoch of sleep (usually stage 1). Normal adult SL is considered to be 10 to 25 minutes. Generally, it is shortened in sleep deprivation and SDB, and in conditions causing hypersomnia. It may be falsely prolonged in an otherwise sleepy patient who has taken a nap in the afternoon before coming to the sleep center, in patients who have late bedtime (delayed sleep phase), or in shift workers. It may be prolonged in patients who have restless legs syndrome (RLS) and insomnia. RLS may be suspected in patients who have prolonged SL, and a notation of irresistible desire to move their legs at night in the sleep questionnaire, and characteristic movements may be evident on video recording. RLS is a clinical diagnosis, however, and should not be diagnosed by PSG. When RLS is suspected, one should alert the referring physician toward this possibility in the final report. Thus prestudy questionnaire, medical history, medications, and technician notes are important when commenting on abnormal SL. REM sleep latency (RL) is defined as the time elapsed from the onset of sleep to the first epoch of REM sleep. Thus RL may not necessarily be prolonged with the prolonged SL. A short REM sleep latency on nocturnal PSG (as opposed to MSLT) is most commonly seen in narcolepsy, depression, and

INTERPRETATION OF THE ADULT POLYSOMNOGRAM

729

with certain medications such as bupropion [15–17]. It hasbeen also been reported in normal individuals who have SDB [18]. Other potential causes include sleep deprivation, withdrawal from REM suppressants such as tricyclic, monoamine oxidase inhibitors (MAOI), withdrawal from selective serotonin reuptake inhibitors (SSRI), serotonin and noradrenaline reuptake inhibitors (SNRI) antidepressants, withdrawal from alcohol, and SDB. Sleep stage distribution varies with age [19]. NREM sleep constitutes about 75% to 80% of sleep, in which stage N2 is the most abundant (45%–55%), followed by stage N3 (13%–23%), and stage 3 (3%–8%). REM sleep constitutes about 20% to 25% of sleep time [19]. In general, slow wave sleep and REM sleep decrease with age, whereas sleep latency, stage N1 sleep, and wake after sleep onset (WASO) increase with age [20]. When interpreting a PSG for SDB, it is important to note if REM sleep was recorded, and whether any REM sleep was noted in the supine position. This is important both during the diagnostic and the PAP titration studies, because OSA is generally more severe during REM sleep and in the supine position [3,4]. One must note whether the therapeutic pressure chosen after the PAP trial was effective during REM sleep and in the supine position. Sometimes prolonged periods of REM sleep are observed during PAP titration, representing a rebound phenomenon from chronic REM sleep deprivation in sleep apneics. During these periods, oxygen desaturation may be significant, and occasionally worse than the degree of hypoxemia seen during the diagnostic study. Clinical judgment is required to make a decision on whether a pressure increment is warranted, and depends on severity of desaturations, frequency and length of respiratory events, snoring, pressure intolerance, mask leaks, and existence of concurrent cardiopulmonary conditions. Sometimes supplemental oxygen is required to alleviate hypoxemia. Before prescribing supplemental oxygen with PAP, one should bear in mind that the oxygen concentrators used at home deliver only low flow of oxygen (maximum 6 liters per minute [LPM]), which further gets significantly diluted by the high flows of PAP [21]. In addition, the concentrators are bulky, noisy, and cumbersome. Overnight pulse oximetry is an inexpensive way to determine if the patient has significant oxygen desaturation while using PAP at home.

Electromyography and leg movements Submental and bilateral leg EMG recordings are standard during a PSG recording. The submental EMG is essential for scoring REM sleep, and is at its lowest amplitude during REM sleep. A useful way to discern this is to increase the sensitivity of EMG and compare the amplitude to that recorded during other sleep stages. Short bursts in leg or submental EMG associated with eye movements during REM sleep is a normal finding (see Fig. 14) and should not be confused with periodic limb movements (PLM) (Figs. 16).

730

KAKKAR & HILL

Fig. 16. Periodic limb movements causing arousals.

According to the old definition, periodic limb movements are characterized by occurrence of short bursts of EMG activity in limb EMG channels lasting 0.5–5 seconds and are associated with flexion of knee, dorsiflexion of ankle and extension of great toe. Scoring requires at least a sequence of four movements occurring at least 5 seconds apart and not more than 90 seconds apart [22]. This definition required that EMG tone must increase by at least 25% of baseline [22]. According to the World Association of Sleep Medicine, PLMS can be scored up to 10 seconds in duration and must have an absolute threshold value of 8 mV at the onset [23]. This new definition of PLMS has been adopted and recommended by AASM in their scoring manual [2]. The PLMS occur with a periodicity of 20–40 seconds, predominantly during stages N1 and N2 and during the first one third of the night. A PLM index (number/hour) of 5 is required for diagnosis. To meet the scoring criteria, PLM must not be preceded by a respiratory event. Many laboratories

INTERPRETATION OF THE ADULT POLYSOMNOGRAM

731

also report PLM arousal index (number/hour of PLM resulting in arousal); however, there is no agreement on whether PLM arousal index correlates with daytime symptoms. Most patients who have PLM do not require treatment unless they are symptomatic. Prolonged sleep latency in association with significant PLM on PSG should alert one to the possibility of RLS. Excessive EMG tone and abnormal limb movements during REM sleep are seen in REM behavior sleep disorder and in patients treated with certain antidepressants [24,25]. The movements in RBD are complex and semipurposeful, and are often associated with vocalization. The accompanying EEG does not show epileptiform activity. It should be remembered that RBD is a clinical entity, and should not be diagnosed based upon PSG findings alone in the absence of symptoms. RBD should be differentiated from seizures, which are uncommon during REM sleep and usually have a repetitive pattern. Because only four EEG derivations are used, characteristic EEG activity may not be apparent during a routine PSG recording, especially with frontal lobe epilepsy, and a note of this limitation of PSG should be made in the report. A whole night PSG with 16-channel EEG is recommended for diagnosing nocturnal seizures. RBD should be differentiated from confusional arousals, which are more common in younger subjects, whereas RBD is usually seen in elderly individuals. In confusional arousals the subjects usually arouse from stage 3 or 4 sleep, dream recall is not common, and significant autonomic activation (tachycardia, sweating) may be seen. Video recording, heart rate changes, and technician notes are extremely important for assessment of parasomnias. Another common finding on EMG is sleep bruxism (Fig. 17). During bruxism, repetitive submental EMG bursts are seen, occurring at a frequency of 1 to 2 Hz and causing artifact in all EEG and EOG channels. Similar findings can be seen in orofacial dyskinesia and rhythmic tongue movement disorder. A characteristic loud gritting sound is heard on the voice recording, and is diagnostic of bruxism. Therefore, confirmation by listening to the characteristic gritting sounds is important to make a diagnosis. Masseter muscle EMG recording is more sensitive than chin EMG for detecting and differentiating bruxism from other conditions. Airflow and respiration The most common reason for performing a PSG is to establish a diagnosis of SDB. Obstructive sleep apnea is the most common sleep diagnosis. Central sleep apnea (CSA) is much less common and is diagnosed in fewer than 10% of patients presenting for a PSG [26]. Most patients who have CSA have concomitant OSA. Airflow has been conventionally measured using an oral-nasal thermistor probe or thermocouple. The underlying concept is that respiratory air flow causes a change in temperature and the conductance of the thermistor, which is recorded as an electrical signal. It can detect airflow from the nose as well as the mouth, but it has low sensitivity

732

KAKKAR & HILL

Fig. 17. Bruxism. 29-year-old female with subvalvular stenosis was noted to have several episodes of prolonged 1–2 Hz disturbance in EEG and chin EMG. Prominent gritting sounds were heard on video replay.

to detect respiratory events, especially hypopneas. Nasal pressure can be monitored using cannulae similar to ones used for oxygen therapy (Fig. 18). They are much more sensitive than a thermistor probe, especially for detection of hypopneas (Fig. 19). For OSA, the events show not only a decrease in the pressure, but also a characteristic flattening. Thus both the amplitude and shape of nasal pressure tracing should be observed. The authors routinely use nasal pressure monitoring in addition to oralnasal thermistor. Nasal pressure may not be reliable in the event of mouth breathing, but the nasal pressure never drops to zero even with mouth breathing, and there is no flattening of the waveform (Fig. 20). Also, there may be a concomitant increase in thermistor signal. Another drawback of nasal pressure is that the cannulae can become occluded with moisture and nasal secretions. Respiratory excursions are measured by placing piezoelectric belts on the chest and abdomen. They do not quantitatively measure the effort, tend to dislodge overnight, and are not very sensitive in obese patients. Respiratory inductance plethysmography (RIP) belts are better for detection of respiratory movements. An apnea in adults is defined on PSG by cessation of airflow for 10 or more seconds. According to an AASM task force (Chicago criteria) it is not important to distinguish obstructive apneas from hypopneas, because they have similar pathophysiology [27]. Accordingly, a 50% reduction in

INTERPRETATION OF THE ADULT POLYSOMNOGRAM

733

Fig. 18. Nasal pressure cannulae. (Courtesy of Compumedics USA, El Paso, Texas; with permission.)

the amplitude of a valid measure of breathing from baseline is adequate to identify a ‘‘respiratory event,’’ and does not require oxygen desaturation or arousal to define it. Baseline is defined as mean amplitude for 2 minutes preceding the respiratory event, or the mean amplitude of the three largest breaths in the 2 minutes preceding onset of the event (in individuals who do not have a stable breathing pattern). If there is clear reduction (but

Fig. 19. Nasal pressure tracing (marked NPRES) shows significant decrease in amplitude, characteristic flattening, and reflection of snoring during a hypopnea while there is little change in the thermistor signal (marked AIRFLOW).

734

KAKKAR & HILL

Fig. 20. An example of exaggerated mouth breathing. Increased signal at mouth transition from additional sensor at mouth. Note the nasal pressure signal does not disappear. (Courtesy of Compumedics USA, El Paso, Texas; with permission.)

!50%) in any of the valid measures of breathing lasting 10 or mores seconds, and it is either associated with oxygen desaturation of 3% or greater or results in an arousal, it is also a scorable event (see Fig. 20) [27]. An apnea is termed obstructive if it is associated with a discernible respiratory effort (Fig. 21). An apnea is termed central if there is no discernible respiratory effort associated with cessation of airflow [27]. AASM task force criteria require measurement of esophageal pressures

Fig. 21. Obstructive apnea with paradoxical breathing. Note the out-of-phase movements in chest and abdominal effort during the obstructive sleep apnea, which returns to in-phase movements at the termination of the event.

INTERPRETATION OF THE ADULT POLYSOMNOGRAM

735

to distinguish between central and obstructive sleep apnea events; however, most laboratories in the country do not measure esophageal pressures, and rely on other less accurate methods of determining respiratory effort. Paradoxical motion in chest and abdominal belts is highly suggestive that the apnea is obstructive (see Fig. 21). This refers to out-of-phase movements of chest and abdominal waveforms. One should be careful in such instances to make sure that the respiratory signal in preceding epochs and following the apnea returns to in-phase excursions. If out-of-phase waveforms are persistent, it may indicate reversed polarity in the belts. Paradoxical motion without OSA can be seen in patients who have severe chronic obstructive pulmonary disease (COPD), with morbid obesity, and in patients who have quadriplegia. Apnea hypopnea index (AHI) is the number of respiratory events scored per hour of sleep. AASM defines OSA as mild if AHI is 5 to 15, moderate if it is 16 to 30, and severe if it is greater than 30. Another measure of severity is the degree of hypoxemia; however, there is no universally accepted way of defining the severity, because symptoms and oxygen desaturation may not correlate well with the AHI. One of the biggest controversies in sleep medicine is the definition of hypopnea. The AASM Clinical Practice Committee and centers of Medicare and Medicaid services (CMS) define hypopnea in adult patients as an abnormal respiratory event lasting at least 10 seconds, with at least a 30% reduction in thorocoabdominal movement or airflow as compared with baseline, and with at least a 4% oxygen desaturation [28]. There are several other definitions and different sleep laboratories define hypopneas extremely variably. Such a wide variation in the criteria and the inter-observer variability in scoring PSG can have a huge impact on the ‘‘severity’’ of sleep apnea as defined by one laboratory versus another laboratory. It has been shown that this can result in a 10 fold variation in severity of SDB as defined by AHI, and a 16 fold variation in the prevalence of OSA [29,30]. Coexisting medical disorders such as hypothyroidism or cardiovascular disease should be taken into account while interpreting a borderline study. Many patients, especially young women, have severe daytime hypersomnolence and history of snoring. Despite the scarcity of scorable respiratory events, many of them will exhibit either crescendo snoring culminating in arousal (Fig. 22), or periods of prolonged snoring associated with concomitant flattening of nasal pressure tracing. If Pes is recorded in these patients, excessive swings in the negative Pes are recorded. These findings are suggestive of upper airway resistance syndrome (UARS) [31,32]. There is considerable debate whether UARS is a separate clinical entity or a mild form of OSA. UARS is more prevalent in younger women who have lower body mass index (BMI) than their counterparts who have OSA, who most often tend to be middle-aged obese men. Unlike OSA, CSA is a heterogeneous entity consisting of many different disorders. One of the common manifestations of CSA, especially in patients

736

KAKKAR & HILL

Fig. 22. Crescendo snoring culminating in arousal.

who have severe left ventricular dysfunction, is Cheyne-Stokes breathing (Fig. 23). It has a characteristic crescendo, decrescendo morphology. Pes monitoring is the best available method to distinguish CSA from OSA. Emergence of CSA during PAP titration usually disappears with time;

Fig. 23. Cheyne-Stokes breathing. Note characteristic crescendo decrescendo pattern in breathing effort, paralleling the airflow.

INTERPRETATION OF THE ADULT POLYSOMNOGRAM

737

however, some patients may show worsening CSA and oxygen desaturation with continuous positive airway pressure (CPAP) or bi-level PAP titration. This is called complex sleep apnea [33]. Several treatment options are available for the treatment of complex sleep apnea. Continuous pulse oximetry is used to assess baseline oxygen saturation and degree of hypoxemia during PSG recording. The authors prefer not to use supplemental oxygen during diagnostic or PAP titration studies as long as severe hypoxemia (!80%) is not present. Supplemental oxygen may make recognition of hypopneas difficult [34]. Many patients on nocturnal oxygen therapy for mild nocturnal hypoxemia often do not require supplemental oxygen with PAP therapy. This is because PAP increases the functional residual capacity (FRC), which is critical in determining oxygenation in an individual. True oxygen desaturation with respiratory events is associated with a smooth drop in the pulse oximetry tracing and recovery to baseline upon termination of the event (see Figs. 19 and 21). Respiratory events may cause an arousal resulting in abrupt movements of limbs. Such movements may make recognition of the pulse wave difficult, and result in an artifactual lowering of pulse oximetry. Abrupt drops and recovery in pulse oximetry signal are usually indicative of artifacts and should be ignored. A drop in oxygen saturation to zero is always artifactual (Fig. 24). Some of the manufacturers offer real time display of pulse plethysmography signal. This is particularly helpful in differentiating true oxygen

Fig. 24. Excessive leak during PAP titration causing ‘‘hypopnea.’’ Note a leak greater than 60 L/min (1 L/sec). Associated arousal caused artifactually low reading of oxygen saturation at 0%.

738

KAKKAR & HILL

desaturation from an artifact, because the latter does not have the characteristic waveform of a peripheral arterial pulse. Snoring is recorded from a snore microphone, and the signal is displayed as a continuous waveform. Technician notes are the most useful way of assessing the severity of snoring. Snoring signal is also prominently reflected in chin EMG, in nasal pressure tracing (see Fig. 22), and occasionally in the thermistor tracing. Crescendo snores associated with progressively increasing effort are indicative of upper airway obstruction. It should be noted that crescendo snoring could also be seen in CheyneStokes breathing; however, an associated decrescendo component is characteristic. During a PAP titration study, the airflow signal is detected by the device itself and is displayed as ‘‘C-flow.’’ One should look for the effective pressure to eliminate apnea, hypopneas, hypoxemia, and snoring. Some laboratories attempt to eliminate flow limitation also during PAP titration. Persistent arousals with seemingly adequate titration are either caused by underlying mild airway obstruction or by excessive pressure. Under such circumstances, the technicians generally increase or decrease the pressure according to their judgment, and compare the results on different pressures. Occurrence of excessive air leak during PAP titration is a challenging situation. Most modern PSG and PAP systems continuously display breath-to-breath measurement of leak. Some leak (intentional leak) is needed for the effective operation of PAP machines. Excessive leak is defined as the degree of leak that compromises effectiveness of PAP therapy (see Fig. 24). Different manufacturers have different guidelines for their equipment, but a leak in excess of 0.4 liters per second is usually considered excessive. Leak can also be suspected by looking at the PAP waveform, which may show a sudden dip at the beginning of expiration. Presence of snoring-like artifact during the expiration is also indicative of air leak rather than upper airway obstruction. Excessive air leak can be caused either by problems with the mask seal or by mouth opening; however, true differentiation between the two may be difficult while looking at the PSG data. The technician’s notes are quite valuable in making this determination. Mask leaks are more common with full-face mask than with nasal interfaces. When choosing a therapeutic pressure, the clinician should ensure that it is effective in both supine position and during REM sleep. Other factors in choosing a therapeutic pressure are the degree of consolidated sleep obtained at that pressure, patient’s tolerance, degree of air leak, and the length of time it was tested. The authors prefer to choose the lowest effective pressure, provided it was tested for an adequate length of time. If bilevel PAP was used, the technician should document the reason for using it. If no therapeutic pressure could be determined, a repeat PAP titration should be requested. If supplemental oxygen was used, one should ensure that it was indeed effective and necessary for treatment of hypoxemia.

INTERPRETATION OF THE ADULT POLYSOMNOGRAM

739

Electrocardiography Patients who have SDB often exhibit cardiac arrhythmias at night (Figs. 25, 26). Most laboratories monitor one EKG lead to determine heart rate and to detect cardiac arrhythmias. The authors routinely record two EKG leads. The extra EKG electrode provides a backup in case one electrode becomes loose during the course of the study. The most common arrhythmias noticed in patients who have OSA are sinus bradycardia, sinus arrest, and sinus tachycardia. Periods of bradycardia intervening with normal heart rate or tachycardia are also sometimes noted in patients who have OSA. The bradycardia is a result of massive vagal discharge that occurs with breathing against a closed upper airway [35]. Termination of sleep apnea invariably results in a surge in catecholamine levels, which results in tachycardia and a rise in blood pressure [36]. It should be noted that atrial fibrillation is especially common in OSA. Presence of atrial fibrillation is one of the strongest independent predictors of the presence of OSA [37]. The medical director of the sleep laboratory should be alerted about any dangerous arrhythmias. Reporting a sleep study The PSG report should be brief and succinct. The first part of the report is generated automatically, and contains the patient’s demographic and anthropometric information, indication for performing the PSG, the

Fig. 25. Atrial flutter appearing with an obstructive sleep apnea. The patient did not have any previous history of cardiac arrhythmias and was noted to have three prolonged episodes of atrial flutter during this PSG, which returned to normal sinus rhythm spontaneously.

740

KAKKAR & HILL

Fig. 26. Nonsustained ventricular tachycardia during a PSG recording.

physiological parameters that were recorded, the equipment used for recording, and pertinent numerical data in tabular form. Any additional monitoring, such as extra EEG or EMG derivations recorded during the study, should also be mentioned. The tabular data should contain information on recording time, TIB, TST, SPT, SE, SL, RL, sleep stage distribution, percentage of time spent in supine position, total AHI, supine AHI, AHI during REM sleep, the nature of respiratory events (obstructive or central), whether baseline hypoxemia was present, degree of hypoxemia, heart rate, and the PLM index. Some laboratories also report arousal index. There are many other indices reported on the automatically generated PSG reports. Most of the data in these reports are redundant and confusing, and the clinical utility of many of these indices is unclear. The descriptive part of the report should be brief, with emphasis on important findings while avoiding unnecessary repetition of numerical data. One should develop an overall impression of the study as opposed to just the numerical interpretation, especially when reporting severity of sleep apnea and prescribing an effective PAP setting. Long apneas may artificially lower the AHI, whereas many patients have severely disrupted sleep or significant oxygen desaturation with respiratory events of short duration. Lack of REM sleep or lack of sleep in the supine position may underestimate the severity of sleep apnea, and a note should be made to reflect this. The AHI reflected on the PAP table may be misleading because it does not characterize sleep consolidation, patient discomfort, or air leak. Further, if only a short time is spent on a given pressure, it may artifactually lower or elevate the AHI because of sampling error. On should always

INTERPRETATION OF THE ADULT POLYSOMNOGRAM

741

mention whether or not the recommended pressure was adequately tested in the supine position and during REM sleep. The occurrence of cardiac arrhythmias and parasomnias should be documented, with the epoch numbers in which they occurred. Some sleep centers do not routinely archive the video data because of the size of the files. If parasomnias or seizures are documented, the laboratory should be instructed to store the video files of that PSG. Summary PSG interpretation is a time-consuming process that requires close attention to the patient’s history, medications, technician notes, pre- and poststudy sleep questionnaires, and overall impression of the raw data. It is a unique opportunity to correlate clinical and electrophysiological data, however, and is a good investment of time toward improving patient outcomes and avoidance of unnecessary testing. The adage, ‘‘Your eyes see what your brain knows’’ holds particularly true for interpreting a PSG. Appendix How likely are you to doze off or fall asleep in the following situations, in contrast to feeling just tired? This refers to your usual way of life in recent times. Even if you have not done some of these things recently, try to work out how they would have affected you. Use the following scale to choose the most appropriate number for each situation:    

0 1 2 3

¼ ¼ ¼ ¼

no chance of dozing slight chance of dozing moderate chance of dozing high chance of dozing

Table 1 Epworth sleepiness scale Situation Sitting and reading Watching TV Sitting inactive in a public place (eg, a theater or a meeting) As a passenger in a car for an hour without a break Lying down to rest in the afternoon when circumstances permit Sitting and talking to someone Sitting quietly after a lunch without alcohol In a car, while stopped for a few minutes in traffic Score: 0–10, normal range; 10–12, borderline; 12–24, abnormal.

Chance of dozing

742

KAKKAR & HILL

References [1] ASDA Standards of Practice. Portable recording in the assessment of sleep. Sleep 1994;17: 378–92. [2] Iber C, Ancoli-Israel S, Chesson A and Quan SF for the American Academy of Sleep Medicine. The AASM Manual for the Scoring of Sleep and Associated Events- Rules, Terminology and Technical specifications. 1st edition. Westchester Illinois: American Academy of Sleep Medicine; 2007. [3] McEvoy RD, Sharp DJ, Thornton AT. The effects of posture on obstructive sleep apnea. Am Rev Respir Dis 1986;133:662–6. [4] Oksenberg A, Khamaysi I, Silverberg DS, et al. Association of body position with severity of apneic events in patients with severe nonpositional obstructive sleep apnea. Chest 2000;118: 1018–24. [5] Sackett DL, Haynes RB, Tugwell P, et al. Clinical epidemiology: a basic science for clinical medicine. 2nd edition. Boston: Lippincott Williams & Wilkins; 1991. [6] Chediak AD, Acevedo-Crespo J, Seiden DJ, et al. Nightly variability in the indices of sleepdisordered breathing in men being evaluated for impotence with consecutive night polysomnograms. Sleep 1996;19:589–92. [7] Witting RM, Romaker A, Zorick FJ, et al. Night-to-night consistency of apneas during sleep. Am Rev Respir Dis 1984;129:244–6. [8] Hirshkowitz M, Moore CA, Hamilton CR III, et al. Polysomnography of adults and elderly: sleep architecture, respiration, and leg movement. J Clin Neurophysiol 1992;9:52–62. [9] Agnew HW Jr, Webb WB, Williams RL. The first night effect: an EEG study of sleep. Psychophysiology 1966;2:263–6. [10] Johns MW. A new method for measuring daytime sleepiness: the Epworth Sleepiness Scale. Sleep 1991;14:540–5. [11] Johns MW. Daytime sleepiness, snoring, and obstructive sleep apnea. The Epworth Sleepiness Scale. Chest 1993;103:30–6. [12] Chervin RD, Aldrich MS, Pickett R, et al. Comparison of the results of the Epworth Sleepiness Scale and the Multiple Sleep Latency Test. J Psychosom Res 1997;42:145–55. [13] Penzel T, Conradt R. Computer based sleep recording and analysis. Sleep Med Rev 2000;4: 131–48. [14] American Sleep Disorders Association. EEG arousal: scoring rules and examples. Sleep 1992;15:173–84. [15] Reynolds CF III, Hoch CC, Buysse DJ, et al. Sleep in late-life recurrent depression: changes during early continuation therapy. Neuropsychopharmacology 1991;5:85–96. [16] Benca RM, Obermeyer WH, Thisted RA, et al. Sleep and psychiatric disorders: a meta-analysis. Arch Gen Psychiatry 1992;49:651–68. [17] Nofzinger EA, Reynolds CF III, Thase ME, et al. REM sleep enhancement by bupropion in depressed men. Am J Psychiatry 1995;152:274–6. [18] Olivier LB, Luc S, Guy H, et al. Shorter REM latency associated with more sleep cycles of a shorter duration in healthy humans. Psychiatry Res 2001;104:75–83. [19] Kryger MH, Roth T, Dement WC, editors. Principles and practice of sleep medicine. 3rd edition; 2000. p. 20. [20] Ohayon MM, Carskadon MA, Guilleminault C, et al. Metaanalysis of quantitative sleep parameters from childhood to old age in healthy individuals: developing normative sleep values across the human lifespan. Sleep 2004;27:1255–73. [21] Yoder EA, Klann K, Strohl KP. Inspired oxygen concentrations during positive pressure therapy. Sleep Breath 2004;8:1–5. [22] American Sleep Disorders Association. Recording and scoring leg movements. The Atlas Task Force. Sleep 1993;16:748–59. [23] Zucconi M, Ferri R, Allen Baier PC, et al. The official World Association of Sleep Medicine (WASM) standards for recording and scoring periodic leg movements in sleep (PLMS) and

INTERPRETATION OF THE ADULT POLYSOMNOGRAM

[24] [25] [26] [27]

[28] [29]

[30] [31]

[32]

[33] [34]

[35] [36] [37]

743

wakefulness (PLMW) developed in collaboration with a task force from the International Restless Legs Syndrome Study Group (IRLSSG). Sleep Med 2006;7:175–83. Jobert M, Jahnig P, Schulz H. Effect of two antidepressant drugs on REM sleep and EMG activity during sleep. Neuropsychobiology 1999;39:101–9. Winkelman JW, James L. Serotonergic antidepressants are associated with REM sleep without atonia. Sleep 2004;15(27):317–21. De Backer WA, Verbraecken J, Willemen M, et al. Central apnea index decreases after prolonged treatment with acetazolamide. Am J Respir Crit Care Med 1995;151:87–91. Sleep-related breathing disorders in adults: recommendations for syndrome definition and measurement techniques in clinical research. The report of an American Academy of Sleep Medicine task force. Sleep 1999;22(5):667–89. Morgenthaler TI, Kapen S, Lee-Chiong T, et al. Practice parameters for the medical therapy of obstructive sleep apnea. Sleep 2006;29:1031–5. Redline S, Kapur VK, Sanders MH, et al. Effects of varying approaches for identifying respiratory disturbances on sleep apnea assessment. Am J Respir Crit Care Med 2000;161: 369–74. Redline S, Bonekat W, Gottlieb D, et al. Variation in the apnea hypopnea index (AHI) according to hypopnea definition. Am J Respir Crit Care Med 1997;155:A128. Guilleminault C, Black JE, Palombini L, et al. A clinical investigation of obstructive sleep apnea syndrome (OSAS) and upper airway resistance syndrome (UARS) patients. Sleep Med 2000;1:51–6. Loube DI, Andrada TF. Comparison of respiratory polysomnographic parameters in matched cohorts of upper airway resistance and obstructive sleep apnea syndrome patients. Chest 1999;115:1519–24. Gilmartin GS, Dalyb RW, Thomas RJ. Recognition and management of complex sleepdisordered breathing. Curr Opin Pulm Med 2005;11:485–93. Mayos M, Hernandez Plaza L, Farre A, et al. [The effect of nocturnal oxygen therapy in patients with sleep apnea syndrome and chronic airflow limitation]. Arch Bronconeumol 2001;37:65–8 [Spanish]. Zwillich C, Devlin T, White D, et al. Bradycardia during sleep apnea: characteristics and mechanism. J Clin Invest 1982;69:1286–92. Launois SH, Averill N, Abraham JH, et al. Cardiovascular responses to nonrespiratory and respiratory arousals in a porcine model. J Appl Physiol 2001;90:114–20. Gami AS, Pressman G, Caples SM, et al. Association of atrial fibrillation and obstructive sleep apnea. Circulation 2004;110:364–7.