Sleep Medicine Pearls 2nd

Answers: In Figure I, there is no arousal as the chin EMG does not increase. In Figure 2, alpha sleep is present. Discussion: According to the ASDA rules for scoring an arousal during REM sleep, an increase in the chin EMG must be present in order to score an arousal. The rationale for this rule is that bursts of alpha are common during REM sleep and do not necessarily reflect a change in sleep state. Of note, the frequency of alpha in REM sleep often is 1-2 Hz slower. In Figure I, notice that there is no change in heart rate (EKG). There is a change in airflow, but this is not uncommon during bursts of eye movements in REM sleep. In Figure 2, alpha waves are superimposed upon slow waves. This is an example of alpha sleep. As the alpha waves are present during delta sleep (slow wave sleep) this is also called alpha-delta sleep. Alpha sleep is defined as the diffuse presence of alpha activity during a stage of sleep in which it is normally not present. Alpha intrusion refers to a brief superimposition of alpha activity on sleep (although some use the term synonmously with alpha sleep). In this example, the alpha activity is very prominent. When alpha activity is less distinct, you C4-Al C3-A2 02-Al Ol-A2 ROC-Al LOC-A2 chin EMG can change the time base (if you are using digital monitoring) to show a lO-second page (30 mm/sec equivalent paper speed). Below is a section of the tracing in which you can see the alpha activity present on top of the slow wave activity. Alpha intrusion makes the scoring of sleep more difficult. You might assume that the presence of alpha activity in more than 50% of the epoch would make an epoch stage Wake. However, K complexes or slow waves are not present during wake. Hence you can stage sleep as stage 2, 3, or 4 dependent on the usual criteria using these waveforms. Scoring stage I is more difficult as it is usually defined by less than 50% alpha activity, rather than the presence of other EEG activity. If prominent theta activity is present or you can see definite vertex sharp waves, then you might be confident in scoring stage I. Alpha intrusion can be associated with any cause of discomfort during sleep. It has been associated with depression, fibromyalgia, chronic pain syndromes, and even discomfort from monitoring equipment. 9 cps Clinical Pearls l. Scoring an arousal during REM sleep requires an increase in the chin EMG. 2. Alpha intrusion can be present during any stage of sleep (alpha sleep) and can make sleep staging difficult. 3. Using a different time base (faster paper speed) can allow visualization of background alpha EEG activity. REFERENCE Butkov N: Atlas of Clinical Polysomnography. Ashland, OR, Synapse Media, 1996, pp 110-112. 39

PATIENT 14 A 40-year-old man being treated for depression A 40-year-old man being treated for depression underwent a sleep study to evaluate complaints of excessive daytime sleepiness. Sleep Study: The tracing below shows prominent slow eye movements and rapid eye movements (R) in conjunction with K complexes and slow wave activity. Question: What sleep stage is shown in the tracing? C4-Al 02-Al ROC-Al LOC-A2 EKG 40

Diagnosis: This is stage 2 sleep. The prominent eye movements are secondary to medication (serotonin reuptake inhibitor). Discussion: Slow eye movements (slow rolling eye movements) are characteristic of drowsy wakefulness and stage I sleep. They usually are absent during stages 2, 3. and 4 NREM sleep. They may folIowan arousal and typically are characteristic of transition to a lighter stage of sleep. Rapid eye movements are seen during Wake (usually eyes open) and during REM sleep. Eye movements both slow and rapid may be prominent during stages 2, 3, and 4 NREM sleep in patients taking serotonin reuptake inhibitors (e.g., fluoxetine, paroxetine) and, less commonly, tricyclic antidepressants. The fact that REM sleep is not present can usually be ascertained by noting the relatively high chin EMG activity and the presence ofK complexes and slow wave activity. Studies in depressed patients have shown that fluoxetine increased the REM density (number of REMs per minute of REM sleep) while increasing the REM latency and decreasing the amount of REM sleep. A medication history is essential when evaluating a sleep study. Most sleep laboratories have patients fill out both a pre-study and post-study questionnaire to carefully document drug and alcohol consumption. In the current case, the tracing shows both slow and rapid eye movements. However, K complexes are present and the EMG shows considerable activity. Therefore, the tracing is best classified as stage 2 sleep. The patient was taking paroxetine at the time the tracing was performed. Clinical Pearls I. Prominent slow and rapid eye movement activity may be seen in stages 2, 3, and 4 NREM sleep in patients taking serotonin reuptake inhibitor antidepressants. 2. The presence of an EEG typical for NREM sleep and a level of EMG activity that is higher than the REM level identify the affected sleep as NREM instead of REM sleep. 3. A drug history is essential when evaluating a sleep study. REFERENCES I. Schenck CH. Mahowlad MW. Kim SW. et al: Prominent eye movements during NREM sleep and REM sleep behavior disorder associated with ftuoxetine treatment of obsessive-compulsive disorder. Sleep 1992; 15:226-235. 2. Armitage R. Trivedi M. Rush AJ: Fluoxetine and oculomotor activity during sleep in depressed patients. Neuropsychopharrnacology 1995;12:159-165. 41

FUNDAMENTALS OF SLEEP MEDICINE 6 Additional Sleep Staging Rules The EEG, EOG and chin EMG characteristics of the different stages of sleep are discussed in Patients 1-12 and summarized in Appendix 2. There are additional scoring rules to handle special situations. These rules are necessary because K complexes, sleep spindles, and REMs are episodic. The three-minute rule concerns stage 2 sleep. This rule, as outlined by the classic sleep staging manual of Rechtschaffen and Kales, states that if the period of time between spindles or K complexes is shorter than 3 minutes and if the intervening sleep would otherwise meet criteria for stage I (less than 50% alpha activity) with no evidence of intervening arousal, then this period of sleep is scored as stage 2. If the period of time is 3 minutes or longer, then the intervening sleep is scored as stage I. Figure 1 (below) shows five epochs (30 seconds each) of sleep, with K complexes (K) in epochs 69 and 73. The central and occipital EEG tracings in epochs 70-72 are assumed not to contain sleep spindles, K complexes, or evidence of arousal. The time between the K complexes is less than 3 minutes; therefore, this intervening sleep is scored as stage 2. 70 71 72 I I K K ~---....;....---;..------;----:--t-.....: I °2- A1 'V I ROC - A 1 t1' =1- I LOC - A 2 :1~----.~~----r---..-----.--"~ I chin EMG :UI.UH.... il!I.1I Itl... I I'- .•,.....IllI,•.-MI" I d"'~ ",••• ..: Stage 2 Stage 2 Stage 2 Stage 2 Stage 2 Staging of REM sleep also requires special rules (REM rules) to define the beginning and end, because REMs are episodic, and the three indicators of stage REM (EEG, EOG, EMG) may not change to (or from) the REM-like pattern simultaneously. Rechtschaffen and Kales recommend that any section of the record that is contiguous with stage REM and displays a relatively low-voltage, mixed-frequency EEG be scored as stage REM regardless of whether REMs are present, providing the EMG is at the stage REM level. To be REM-like, the EEG must not contain spindles, K complexes, or slow waves. Figure 2 (next page) shows four epochs, with a K complex in epoch 69 and REMs in epoch 72. After epoch 69 there are no K complexes or sleep spindles, and the EMG falls to the REM level during the last part of epoch 70. Hence, epoch 71 meets criteria for REM sleep except that there are no eye movements. Epoch 71 is scored as stage REM because it is contiguous with an epoch of unequivocal REM sleep (epoch 72). Epoch 70 also does not contain K complexes or sleep spindles, but is scored as stage 2 by the three-minute rule. 42

® Epoch 69 70 71 72 I 1 K complex C4 - A1 1-1 O2 - A1 1 1---1\; REM ROC-A1 I~ A LOC-A2 V EMG ~1.Il. 11M ,••iblt"I.•••I'III ••• 1' Stage 2 Stage 2 Stage REM Stage REM These rules are more difficult to apply if arousals break the continuity of sleep. With respect to the three-minute rule, sleep following an arousal is scored according to its nature. In Figure 3 (below), a brief arousal occurs at the end of epoch 71 (alpha waves in EEG, increased EMG). The sleep before the arousal is scored by the three-minute rule as stage 2. Epoch 71 is scored as stage 2 because most of the epoch is stage 2. Epoch 72 is scored as stage I because there are no K complexes or spindles, and the slow rolling eye movements (SR) following the arousal are more characteristic of stage I than stage 2. Stage 2 *•• 1 tl~ ® Epoch 69 70 71 72 73 I I I K I K Cc A 1 :--{. aIP~: t-----:' I I °2- A1 Ht 'MIl: 'V I I ROC - A1 r-1t ~--,...1t---'T I 'SR I I ALOC - A 2 ,1 \..(V"--..,........'Y,--..--.; I I chin EMG ~11.UiL II1II••: 111M Itl'lI. ~.t ....II~IIllU.I1"'I...".iIIl-~-........... Stage 2 Stage 2 Stage 2 Stage 1 In REM sleep, bursts of alpha waves are common and do not signify an arousal unless the chin EMG amplitude also increases. Deciding how to score an epoch of sleep with a REM-like EEG/EMG but no REMs that is separated from contiguous unequivocal REM sleep by an intervening arousal is sometimes difficult. The decision in this case is between stage REM and stage I sleep (with an EMG at the REM level). The EEGs of stage I and REM sleep are similar, but subtle differences are present: REM Sleep EEG Stage 1 EEG Saw-tooth waves may occur Alpha waves 1-2 Hz slower than wakefulness No saw-tooth waves Vertex sharp waves may occur A very brief arousal and/or the presence of saw-tooth waves in the sleep following the arousal is evidence that the sleep after the arousal is still stage REM(until definite evidence for another sleep stage is noted). Conversely, a prolonged arousal (with persistent alpha waves) followed by slow rolling eye 43

movements is evidence that the arousal induced a change from stage REM to stage 1 sleep. Sharp waves or incipient spindles (shorter than O.S-second duration) on the EEG also are evidence for stage 1. If stage I is scored following the arousal, all subsequent epochs are scored as stage I until evidence of another sleep stage is noted (spindles or K complexes-stage 2). In Figure 4 (below), a brief arousal signified by EEG alpha waves and an EMG increase occurs at the end of epoch 70. Epoch 70 is scored as REM because the majority of the epoch is REM-like. and it is contiguous with an epoch of unequivocal REM sleep. Epoch 72 contains a K complex and is scored as stage 2. @ Epoch 69 70 71 72 I I K complex C 1 1 ~I 4 -A1 'YWv 1 alphal 1 02- A1 1 !#N- I iL I 1--.1\ REMs 1 -J\--- I ROC-A1 I I I LOC-A2 IV I ...JV I EMG I .. I ....M• • ,"'lInll , t••J I I I Stage REM Stage REM Stage REM Stage 2 Epoch 71 is scored as stage REM because the arousal was brief and the EEG and EMG are REM-like for most of the epoch. ® Epoch 69 70 71 72 I I I I alpha I waves K complex] C4 - A1 I f". - "V- I 02- A1 I lL- I I I ROC-A1 I SR I LOC-A2 :v --V I EMG J. I I ,.......ltll.'1 JUII I I I I Stage REM Stage REM Stage 1 Stage 2 An epoch containing slow rolling eye movements and following a prolonged arousal is illustrated in Figure S. This epoch is labeled stage 1. The reader is referred to reference 1 for special rules governing the unusual case in which spindles occur during REM sleep REFERENCES 1. Rechtschaffen A. Kales A (eds): A Manual of Standardized Terminology Techniques and Scoring System for Sleep Stages of Human Sleep. Los Angeles, Brain Information Service/Brain Research Institute, UCLA. J968. 2. Caraskadon MA. Rechschaffen A: Monitoring and staging human sleep. In Kryger MH, Roth T, Dement WC (eds): Principles and Practice of Sleep Medicine. 3rd ed. Philadelphia, WB Saunders Co .• 2000, pp 1197-1215. 44

PATIENT 15 A 30-year-old man with severe snoring and occasional breathing lapses A 30-year-old man was evaluated for a history of loud snoring and possible sleep apnea. Below is a schematic representation of five epochs of the patient's sleep. Assume that epoch 69 is stage 2 sleep (a K complex is shown). No K complexes, sleep spindles, or arousals are noted in epochs 70-75. The EEG in these epochs otherwise meets criteria for stage 2. Question: What sleep stage is scored in epochs 70-75? Epoch 69 70 71 72 73 74 75 76 I , , KcomplexI K complexI I 1r-: I C4 - A1 I I I I I 02- A1 '-it I h , , , ROC-A1 I~ , I v: I LOC-A2 '-Jv I I , I , I I I EMG ,..••Url:+... lIMIII ..... l'i.I"",'''' t.1Il1''ff~''''.1*" .11 t·1It I I I , , , , , , Stage 2 Stage? Stage? Stage? Stage? Stage? Stage? Stage 2 45

Answer: Epochs 70-75 are scored as stage 1 because the interval between K complexes is longer than 3 minutes. Discussion: Stage 2 is characterized by the presence of either sleep spindles, which are bursts of 12-14 Hz activity, or K complexes, which are large-amplitude, biphasic EEG deflections. To qualify as stage 2, an epoch also must contain less than 20% of slow (delta) wave EEG activity. Slow waves are large-amplitude (> 75 microvolt) deflections with a frequency of < 2 Hz. K complexes and sleep spindles are episodic and may not occur in each epoch. According to the three-minute rule, if the period of time between spindles or K complexes is < 3 minutes and if the intervening sleep would otherwise meet criteria for stage I (less than 50% alpha activity) with no evidence of intervening arousal, then this period of sleep is scored as stage 2. If the period of time is z 3 minutes, then the intervening sleep is scored as stage I. In the current patient, the time between intervening K complexes in epochs 69 and 76 is longer than 3 minutes; therefore, the intervening sleep is scored as stage I. The 3-minute time frame is somewhat arbitrary. It was selected based on the spindle-tospindle and K complex-to-K complex intervals typically observed. Clinical Pearl Use the three-minute rule to stage sleep occurring between K complexes or spindles which would otherwise meet criteria for stage 2 sleep except that K complexes or sleep spindles are absent. REFERENCES I. Rechtschaffen A, Kales A (eds): A Manual of Standardized Terminology Techniques and Scoring System for Sleep Stages of Human Sleep. Los Angeles. Brain Information Service/Brain Research Institute, UCLA. 1968. 2. Caraskadon MA, Rechschaffen A: Monitoring and staging human sleep. In Kryger MH, Roth T, Dement WC (eds): Principles and Practice of Sleep Medicine, 3rd ed. Philadelphia, WB Saunders Co., 2000, pp 1197-1215. 46

PATIENT 16 A 35-year-old woman experiencing uncontrollable episodes of sleep A 35-year-old woman was evaluated for possible narcolepsy. A schematic representation of eight epochs of the patient's sleep is shown. A K complex occurs in epoch 69 (stage 2 sleep). The chin EMG amplitude decreases at the start of epoch 70 and briefly increases at the end of epoch 75. A REM occurs in epoch 76. Question: What sleep stage is scored in epochs 70-75? Epoch 69 70 71 72 73 74 75 76 j I I Kcomplex I I I C4 - A1 H I .. I alpha! I I I O2 - A1 I-'\J I I • I IREMs ROC-A1 :-A; I I ' I h LOC-A2 I-----Av I II'-{ I I I I EMG l'llitu1*' tid III MI I I I I I Stage 2 Stage? Stage? Stage? Stage? Stage? Stage? Stage REM 47

Answer: Epochs 70-75 are scored as stage REM. Discussion: Stage REM is characterized by a low-amplitude, mixed-frequency EEG and an absence of sleep spindles and K complexes. The eye movement channels show REMs, and the chin EMG is relatively reduced (equal to or lower than the lowest level of NREM sleep). These EEG, EOG, and EMG characteristics may not all start or end at the same time. REMs are episodic and may not occur in all epochs of REM sleep. Therefore, the REM rule is useful for scoring epochs that do not contain REMs: Any section of record contiguous with stage REM in which the EEG is relatively low-voltage and mixed-frequency (no spindles, no K complexes) is scored stage REM regardless of whether or not REMs are present, providing the EMG is at the stage REM level. The rule holds for both the beginning and end of a segment of REM sleep. During sleep scoring, once unequivocal REM sleep is noted, the examiner should work backward to determine if preceding epochs meet the above criteria for REM sleep (see Fundamentals 6). At the transition from NREM to REM sleep, the above REM rule takes precedence over the threeminute rule. However, when an arousal separates the end of unequivocal stages 2--4 NREM sleep and the beginning of unequivocal REM sleep, the scoring of the intervening sleep prior to the arousal is more problematic. The arousal mayor may not signify a change in sleep stage. The EEGs of stage I and REM sleep can look similar. If the EMG of the period in question is at the REM level, the decision is between stage I, stage 2 (based all the threeminute rule), or stage REM. The scoring manual recommends scoring sleep between the last spindle or K complex and an arousal as stage 2 if the time duration is < 3 minutes. If the segment is > 3 minutes, the sleep is scored as stage REM (rather than stage I-an exception to the three-minute rule). In the example below, note the arousal at the end of epoch 72. Sleep after the arousal is scored on the basis of the REM rule. Sleep before the arousal is scored according to a combination of the threeminute rule and the REM rule. On the other hand, if an arousal is prolonged or followed by an obvious change to stage I sleep (slow rolling eye movements), then the sleep following the arousal is scored as stage 1 until unequivocal evidence for another sleep stage is present. The presence of saw-tooth waves favors stage REM. In the present case, epoch 76 is unequivocal REM sleep (referto previous figure). Epochs 70-75 have a REM-like EEG and EMG except for a brief arousal at the end of epoch 75. The choices for epochs 70-75 include stages 1,2, or REM sleep. As the interval after the last K complex (epoch 69) exceeds 3 minutes, it is not stage 2 sleep. Thus, the choices are stage I or REM sleep. Epochs 70-75 look like REM sleep, except for the absence of REMs, and are contiguous with unequivocal REM sleep. Therefore, they are scored as stage REM. Epoch 69 70 71 72 73 74 75 76 I I I I I I I I I Kcomplex I I I C4 - A1 HI fIIJ I I I alpha' I I I I O2 - A1 :-t I .- I I I I I I I I REMs I ROC-A1 l---"\t I I I~I LOC-A2 :---At I I I~ I I I .., I EMG ~i 'Ill.'.'III' I I I I I I I I I Stage 2 Stage 2 Stage 2 Stage 2 Stage REM Stage REM Stage REM Stage REM 48

Clinical Pearls I. Sleep that is contiguous with an epoch of unequivocal REM sleep and meets criteria for stage REM. except that no REMs are present, is scored as stage REM (REM rule). 2. To identify the start of an episode of REM sleep, first identify an epoch of unequivocal REM (REM-like EEG, REMs present, EMG at REM level) and then work backward using the REM rule. To determine the end of an episode of REM. work forward from an epoch of unequivocal REM sleep. 3. When a brief arousal separates unequivocal stages 2-4 NREM sleep and the start of REM sleep, epochs with a REM-like EEG and EMG prior to the arousal are scored as stage 2 if the arousal occurred less than 3 minutes after the last sleep spindle or K complex. Otherwise, the segment is considered stage REM sleep (combination of three-minute rule and REM rule). REFERENCES I. Rechtschaffen A, Kales A (eds): A Manual of Standardized Terminology Techniques and Scoring System for Sleep Stages of Human Sleep. Los Angeles, Brain Information Service/Brain Research Institute, UCLA, 1968. 2. Caraskadon MA, Rechschaffen A: Monitoring and staging human sleep. In Kryger MH, Roth T, Dement WC (eds): Principles and Practice of Sleep Medicine, 3rd ed. Philadelphia, WB Saunders Co., 2000, pp 1197-1215. 49

FUNDAMENTALS OF SLEEP MEDICINE 7 Sleep Architecture Definitions A number of variables have been defined to help characterize the quantity, composition, and quality of sleep. Standard Sleep Variables Time in bed (TIB) Movement time Total sleep time (TST) Wake after sleep onset (WASO) Sleep period time (SPT) Sleep efficiency (0/0) Sleep latency (min) REM latency (min) Monitoring period-lights out to lights on Epochs in which stage is indeterminant due to artifact Total minutes of sleep (stages 1-4 and REM) Minutes of wake after initial sleep onset and before the final awakening TST + WASO (TST * 100) I TIB Time from lights out to the first epoch of sleep Time from sleep onset to the first epoch of REM sleep Any condition that results in frequent or prolonged awakenings (such as sleep-maintenance insomnia or sleep apnea) results in an increased WASO. Even if the WASO is small, it is possible to have a low sleep efficiency if the final awakening occurs early in the monitoring period (early-morning awakening). That is, TST/TIB is reduced, even though TST/SPT is normal. The sleep latency reflects how rapidly the patient fell asleep. Patients with sleep-onset insomnia (difficulty in initiating sleep) typically have a long sleep latency (more than 30 minutes). Some sleep laboratories also compute the latency to stage 2 sleep. The REM latency is the time from the first sleep (not lights out) to the first epoch of REM sleep. The REM latency, normally 70-120 minutes, can be reduced in narcolepsy, sleep apnea. and depression, and after withdrawal of REM-suppressing medications. REM latency is discussed in Patient 18 and Fundamentals 9. The division of total sleep time among the sleep stages often is called the sleep architecture. A common approach is to express the minutes spent in each stage of sleep as a percentage of either TST or SPT. There are few widely accepted normative values for sleep architecture. In this book, the normal ranges are the mean ::+::: one standard deviation, as presented by Williams et al., and the values are age- and sexdependent. In some laboratories movement time (MT) is considered part of the TST (TST = stage I + stage 2 + stage 3 + stage 4 + stage REM + MT). Recall that any epoch that would be scored as MT that is surrounded by wake is scored as Wake. For simplicity, in this book MT is assumed to be zero. The fraction of slow wave sleep (stages 3 and 4) decreases considerably with age, while the amount of sleep stages I and 2 and WASO increase. The reduction in stages 3-4 sleep is mainly because of a decrease in the amplitude ofslow waves that occurs with increasing age. The fraction of REM sleep changes little after young adulthood. The following table shows typical values for normal sleep at two different ages and in a patient with severe obstructive sleep apnea. Note that here and throughout this book, in the tables showing sleep stages as a percentage of sleep period time, Wake = WASO (does not include epochs of wake recorded before sleep onset or after final awakening). 50

Normal Sleep (o/c SPT) Wake Stage I Stage 2 Stages 3 and 4 Stage REM AGE 20 I 5 45 21 28 AGE 60 8 10 57 2 23 Key Point Obstructive Sleep APNEA (%SPT) 10 25 55 o 10 The normal ranges for many parameters of sleep architecture. particularly the amount of slow wave sleep. are age-dependent. REFERENCES I. Williams RL. Karacan r. Hursch CJ: Electroencephalography (EEG) of Human Sleep: Clinical Applications. New York. John Wiley & Sons. 1974. pages 49-60. 2. Bonnet MH: Sleep deprivation. In Kryger MH. Roth T. Dement W (eds): Principles and Practice of Sleep Medicine. Philadelphia. WB Saunders. 1994. pp 50-67. 3. Caraskadon MA. Rechschaffen A: Monitoring and staging human sleep. In Kryger MH. Roth T. Dement WC (eds): Principles and Practice of Sleep Medicine. Philadelphia. WB Saunders Co .• 2000. pp 1197-1215. 51

PATIENT 17 A 23-year-old man with difficulty sleeping A 23-year-old man was monitored because he complained of poor sleep. He admitted to the sleep technician that he had taken his usual benzodiazepine sleeping pill before arriving in the sleep laboratory because he feared that otherwise he would be unable to sleep. Physical Examination: Normal. Sleep Study Time in bed (monitoring time) Total sleep time (TST) Sleep period time (SPT) WASO Sleep efficiency (%) Sleep latency REM latency 450 min (430-454) 428.5 min (405-434) 432.5 min (410-439) 4min 95 (91-99) 5 min (3-26) 120 min (78-99) Sleep Stages Stage Wake Stage I Stage 2 Stages 3 and 4 Stage REM %SPT 1 (0-1) 6 (3-6) 71 (40-51) 7.8 (16-26) 15.2 (22-34) ( ) = normal values for age; sleep efficiency =(TST * 100) I TIB Question: What is abnormal about the sleep architecture? 52

Answer: The percentages of stages 3, 4, and REM sleep are reduced, and the amount of stage 2 sleep is increased. The REM latency also is mildly increased. Discussion: Sleep architecture can be altered by several factors, including sleep disorders, coexistent medical disorders, prior sleep for the last week (sleep deprivation), medications (or withdrawal), beverages, and exposure to a novel sleep environment (the first-night effect). It is essential to know the patient's normal sleep patterns and recent sleep history (sleep diary). An earlier-than-normal bedtime during the sleep study can increase the sleep latency. A later-than-normal bedtime can decrease the REM latency. Prior sleep deprivation can cause a rebound in the amount of slow wave and REM sleep. While it is important to know the patient's medication intake, it is equally important to know if usual medications and/or beverages (caffeine or ethanolcontaining) were not taken (see table below). Abrupt withdrawal of certain medications can profoundly affect sleep architecture. Withdrawal of stimulants or tricyclic antidepressants (REM suppressors) can cause a rebound in the amount of REM sleep and/or shorten the REM latency. Abrupt withdrawal of medications decreasing a given sleep stage can cause a rebound in the amount ofthat sleep stage. Virtually all antidepressants-except for nefazodone, bupropion, and possibly trazodone and mirtazapine-decrease REM sleep. The former two may actually increase REM sleep in depressed patients. The older MAO inhibitors are said to be the most potent suppressors of REM sleep. In the current case, the total amount of sleep and sleep efficiency were normal. However, the amounts of stages 3, 4, and REM sleep were reduced. This change is not unusual with benzodiazepines, which tend to decrease slow wave sleep and, to a lesser degree, REM sleep. The amounts of stage 2 sleep and sleep spindle activity were increased tremendously. Common Medications Affecting Sleep Architecture • DECREASE REM SLEEP Ethanol Tricyclic antidepressants Selective serotonin reuptake inhibitors MAO inhibitors Lithium Amphetamines Methy Iphenidate Clonidine Benzodiazepines (mild) • INCREASE REM SLEEP Nefazodone Reserpine Bupropion (depressed patients) Withdrawal of REM-suppressing medications • No CHANGE IN REM SLEEP Mirtazapine* Trazodone* • DECREASE SWS Benzodiazepines SWS = slow wave sleep * = Some studies show decreased REM sleep. Clinical Pearls I. Analysis of sleep architecture can provide important insight into the causes of sleep disturbance. 2. A medication history (including over-the-counter medications) and a history of the pattern of sleep for several days prior to the sleep study are essential when analyzing sleep architecture. 3. The lights-out time should always mimic the patient's usual bedtime, if possible. 53

REFERENCES I. Bonnet MH: Sleep deprivation. In Kryger MH. Roth T. Dement W (eds): Principles and Practice of Sleep Medicine. Philadelphia. WB Saunders. 1994. pp 50-67. 2. Obermeyer WHo Benca RM: Effects of drugs on sleep. Neurol Clin (Sleep Disorders II) 1996; 14:l;27-840. 3. Walter TJ. Golish JA: Psychotropic and Neurologic Medications. In Lee-Chiong TL. Sateia MJ. Caraskadon MA (eds): Sleep Medicine. Philadelphia. Hanley and Belfus. 2002. pp 587-599. 4. Nofzinger EA. Reynolds CF. Thase ME. et al: REM sleep enhancement by bupropion in depressed men. Am J Psychiatry 1995;152:274-276. 5. Schweitzer PK: Drugs that disturb sleep and wakefulness. In Kryger MH. Roth T. Dement WC (eds): Principles and Practice of Sleep Medicine. 3rd ed. Philadelphia. WB Saunders. 2000. pp 441-461. S4

PATIENT 18 A 25-year-old man with daytime sleepiness and fatigue A 25-year-old man has experienced daytime sleepiness and fatigue during the last year. Prior to these symptoms, he broke up with his girlfriend and has been quite depressed. He has trouble waking up and getting out of bed in the morning. Recently, his primary care physician began him on 20 mg of f1uoxetine every morning. There is no history of cataplexy (muscle weakness triggered by emotion) or sleep paralysis. The patient has never been told that he snores. Physical Examination: Normal. Sleep Study Time in bed (monitoring time) Total sleep time (TST) Sleep period time (SPT) WASO Sleep efficiency (0/0) Sleep latency REM latency 457.5 min (430-454) 406.5 min (405-434) 432.5 min (410-439) 26 min 89 (91-99) 15 min (3-26) 140 min (78-99) Slee Sta es Stage Wake Stage I Stage 2 Stages 3 and 4 Stage REM O/OSPT 6 (0-1) 8 (3-6) 49 (40-51) 25 (16-26) 12(22-34) ( ) = normal values for age: sleep efficiency =(TST * 100) / TIB Questions: What is abnormal about the sleep architecture? What is the most likely cause? 55

Diagnosis: Prolonged REM latency and decreased REM sleep probably secondary to medication. Discussion: The REM latency is the time from the first epoch of sleep until the first epoch of stage REM - usually about 70-120 minutes, depending on the age of the patient. Alterations in REM latency can be due to the presence of disease processes, but many other factors increase or decrease this time period as well (see table below). Narcolepsy, a disorder causing excessive daytime sleepiness, often is associated with a very short REM latency (10-15 minutes or less) referred to as sleep-onset REM. However, this finding is neither specific for nor always present in narcolepsy. Obstructive sleep apnea or any other cause of REM deprivation also may be associated with a short REM latency. Depression typically causes only modest shortening (40-50 minutes); however, depression can have an effect as extreme as narcolepsy. The propensity for REM sleep is associated with the daily (circadian) change in body temperature. By delaying bedtime, the time of sleep onset may move closer to the time of initial REM sleep. Usually this is not a problem, unless the lights-out time is much later than the patient's normal bedtime. Abrupt withdrawal of REM-suppressing medications also can reduce REM latency. A medication history is essential for proper interpretation of the sleep study results. If a diagnosis of narcolepsy is suspected, patients usually are asked to stop medications altering REM latency for at least 2 weeks prior to the sleep study, because changes in REM latency are an essential part of diagnostic criteria. This is problematic for depressed patients with possible narcolepsy, as most antidepressants increase the REM latency. The exceptions are mirtazapine, nefazodone, and bupropion. The first two do not change the REM latency. Bupropion may actually decrease the REM latency in depressed patients. One case report found that bupropion decreased the propensity of a patient with narcolepsy to have early REM periods. The effect of bupropion on the multiple sleep latency test in narcolepsy has not been studied in a large group. A common mistake is to abruptly withhold REMsuppressing medications just before the sleep study. This may cause a rebound in the amount of REM sleep and possibly shorten the REM latency. In the present case, the sleep architecture is fairly normal except for the prolonged REM latency, decreased amount of REM sleep, and slightly decreased sleep efficiency (increased stage W as percentage of SPT). The changes in REM sleep are most likely secondary to the use of ftuoxetine (Prozac), a selective serotonin reuptake inhibitor (SSRI) antidepressant. Sleep efficiency often is decreased in patients with depression. In some of these patients, treatment with SSRIs improves sleep quality. In others, the medications themselves disturb sleep. However, when this patient was seen for discussion of the sleep study results, he claimed to be feeling better and denied any symptoms of daytime sleepiness. SHORT REM LATENCY Common Factors Altering REM Latency LONG REM LATENCY No CHANGE IN REM LATENCY Narcolepsy Prior REM deprivation Sleep apnea Depression, schizophrenia Withdrawal of REM suppressants Later than normal bedtime Bupropion* First-night effect Medical disease (COPD, chronic pain syndromes) Ethanol REM suppressant medications: Tricyclic antidepressants SSRIs (e.g., fluoxetine, sertaline, paroxetine, citalopram) Trazodone MAO inhibitors (most) Venlafaxine Stimulant medications (e.g., amphetamine, methlyphenidate) Clonidine Nefazodone Mirtazapine * in depressed patients COPD = chronic obstructive pulmonary disease, SSRls = selective serotonin reuptake inhibitors 56

Clinical Pearls I. The REM latency is the time from the first epoch ofsleep until the first epoch ofstage REM. 2. A short REM latency is associated with withdrawal of REM-suppressant medications as well as several sleep disorders, including narcolepsy, sleep apnea, and depression. REFERENCES I. Standards of Practice Committee, American Sleep Disorders Association: The clinical use of the multiple sleep latency test. Sleep 1992; 15:268-276. 2. Obermeyer WHo Benca RM: Effects of drugs on sleep. Neurol Clin (Sleep Disorders II) 1996; 14:827-840. 3. Rye DB. Dihenia B. Bliwise DL: Reversal of atypical depression. sleepiness. and REM-sleep propensity in narcolepsy with bupropion. Depress Anxiety 1998;7:92-95. 4. Walter TJ, Golish JA: Psychotropic and Neurologic Medications. In Lee-Chiong TL. Sateia MJ. Caraskadon MA (eds): Sleep Medicine. Philadelphia. Hanley and Belfus, 2002, pp 587-599. 5. Schweitzer PK: Drugs that disturb sleep and wakefulness. In Kryger MH. Roth T. Dement WC (eds): Principles and Practice of Sleep Medicine. Philadelphia, WB Saunders, 2000, pp 441--461. 57

FUNDAMENTALS OF SLEEP MEDICINE 8 Polysomnography "Polysornnography" is the term used to denote the continuous and simultaneous recording of multiple variables during sleep. Routinely Monitored Variables VARIABLES EEG (central, occipital), right and left EOG, chin EMG EKG Airflow (nasal and oral) Respiratory effort Anterior tibialis EMG Arterial oxygen saturation Upper airway sound/vibration PURPOSE Detect the presence and stage of sleep Measures cardiac rate and rhythm Detects apnea and hypopnea Detects respiratory effort Detects periodic leg movements Measures Sa02, detects desaturation Detects snoring METHODS Scalp/face surface electrodes Chest electrodes Thermistors, thermocouples Pneumotachograph in mask, nasal pressure Respiratory movement-chest and abdominal bands (piezoelectric, impedance) Intercostal EMG Esophageal pressure Separate channel for each leg or one common channel Alternate: leg movement Transducers Pulse oximetry Microphone, vibration transducer EEG = electroencephalogram, EOG= electro-oculogram, EMG = electromyogram, EKG = electrocardiogram A typical twelve-channel recording montage (the set of variables being recorded): Channel Variable Channel Variable I Central EEG 7 Right leg EMG 2 Occipital EEG 8 Left leg EMG 3 Right EOG 9 Airflow 4 Left EOG 10 Chest movement 5 Chin EMG II Abdominal movement 6 EKG 12 Sa02 Additional channels are commonly used to record pulse rate (from the oximeter), snoring, body position (from position sensor), and end-tidal CO? (pediatric cases). In special circumstances, esophageal pressure and transcutaneous CO2 are also recorded. During positive-pressure titration, the flow signal from the CPAP machine is usually substituted for airflow. Many laboratories also record a nasal-oral thermister signal and a nasal pressure signal on separate channels. The variables shown here are recorded on a polygraph, using a standard paper speed of 10 mmfsec, and/or digitally acquired on a computer system. 58

Sleep monitoring also requires continuous visual and auditory monitoring of the patient. This is especially important to detect changes in the sleeping posture (supine-lateral decubitus) and allows the patient to signal the technician if assistance is needed. Visual monitoring generally is accomplished via a low-light camera system with video monitors in the recording room. Video recording is needed for evaluation of parasomnias and is now used in most laboratories. Systems are available to allow synchronization of the polysomnographic recording and the video record. Today may systems offer digital video recording, and this makes synchronization of video and polysomnographic recordings very convenient. Where evaluation for possible nocturnal seizures is a consideration, a full EEG montage also should be recorded. If sufficient EEG channels are not available, a limited montage may be helpful (see Fundamentals 20). The signal from each variable recorded enters the system via an amplifier that must be correctly adjusted (sensitivity, low filter, high filter, polarity). EEG, EOG, and EMG channel AC amplifiers are calibrated with a standard voltage signal (usually 50 micovolts) that is available by pushing a button on the polygraph. The sensitivity is adjusted so that the desired output signal (pen deflection) for a given voltage input is obtained (see table below). Low- and high-frequency filters attenuate signals with frequencies outside (below and above, respectively) the desired range of recorded frequency. While filters reduce artifact from unwanted signals, they also can impair recording of desired variables. For example, the low filter must be set low enough so that slow waves and eye-movement signals are not attenuated. In many digital systems, signals are recorded without filtering. Digital filters are than added to change the display of the signals. This is helpful if you prefer to look at the tracings using an alternative filter setting. Polysomnograph Settings SENSITIVITY Low FILTER (Hz) HIGH FILTER (Hz) EEG (central, occipital) EOG EMG (chin, legs) EKG Airflow (therrn) Chest/abdomen Snoring Oximetry Nasal pressure CPAP flow 5 uv/mm 5 uv/rnm 5-10 uv/mrn 50 uv/mm variable variable variable I volt = 100% variable variable 0.3 0.3 10.0 0.3 0.1 0.1 30.0 DC DC or .01 0.01-0.03 30 30 90 30 15 15 90-100 15 15 (100 if snoring to be seen) 15 Filter settings are given as 1/2 amplitude frequency (the amplitude of a signal at this frequency is attenuated by 50%). DC = direct current (no low filter) 30 30 30 30 30 sens LF HF ~(J.l.v/mm) V - 5 - 0.3 - 5- 0.3- =i-C 55 -_ 0.3 - ,1 0.3- ---+--t- 10 - 10 EEG EEG EOG EOG In addition to high and low filter settings, most recording systems have 50-60 Hz "notch" filters to remove 60-cycle interference. These filters do not add that much when a high filter setting of 30 Hz is used as this already reduces 60-Hz activity considerably. However, for EMG channels in which the high filter is set at 90-100, the notch filter may be used. Sixty-Hz filters should never be used as an alternative to adequate electrode application, which is essential to reducing 60-cycle artifact (see Pa- chin EMG tient 20). Deflections from a calibration signal of 50 microvolts are shown in the sample tracing above. The smaller signal in the chin EMG channel is secondary to the higher low-frequency (LF) setting. 59

Biocalibration A biocalibration procedure is performed while signals are acquired with the patient connected to the monitoring equipment. This procedure permits checking of amplifier settings, integrity of monitoring leads/transducers, and recording abilities of airflow and respiratory effort transducers as well as leg EMG. It also provides a record of the patient's EEG and eye movements during wakefulness with eyes closed and open. Biocalibration Eyes closed Eyes open Look right, look left; look up, look down; blink eyes Grit teeth Breathe in, breathe out Hold breath Wiggle right toe, left toe What To Check (Technician and Scorer) Alpha EEG activity, slow rolling eye movements Attenuation of alpha EEG activity, pattern of eyesopen wakefulness Integrity, amplitude, polarity of eye channels, pattern ofREMs Chin EMG Adjust chin EMG gain so that some activity is present during relaxed wakefulness Airflow channel working Airflow, chest, abdomen tracings should have same polarity and be of reasonable amplitude Direction of inspiration (upward deflection) Apnea occurrence Leg movements The next figure is a sample tracing during eye movement biocalibration. The patient has been asked to hold the head still and look in different directions (Left, Right, Down, Up) and then blink (B) the eyes. chin EMG Next, respiratory channels (recording airflow and chest/abdominal movement) are adjusted so that tidal breathing induces a reasonable deflection in all three channels (see figure, right) with the same polarity (inspiration up, arrows). Amplifier gain (sensitivity) and chest/abdominal band positions may need adjustment. The patient is asked to breathe in and hold the breath to simulate apnea, then resume normal breathing. 60 Airflow Chest Abdomen In

The patient is then asked to wiggle the right and left toes to check the ability of the anterior tibialis EMG to detect leg movements (see tracing below). chin EMG EKG R Leg EMG L Leg EMG REFERENCES I. Harris CD: Recording montage. In Shepard JW (ed): Atlas of Sleep Medicine. Mount Kisco. NY, Futura Publishing Co., 1991, pp 1-5. 2. Introduction to the Polysomnograph (video/manual). Available from: Synapse Media, 4702 Cloudcrest Drive, Medford, Oregon; Tel. 800/949-8195. 3. Butkov N: Polysomnography. In Lee-Chiong TL, Sateia MJ, Carskadon MA (eds): Sleep Medicine. Philadelphia, Hanley and Belfus, 2002, pp 605-637. 61

PATIENT 19 A 30-year-old man having difficulty staying awake during the day A 30-year-old man was studied to evaluate complaints of excessive daytime sleepiness. An artifact was noted in his recording. The EEG leads C4-A) and 02-A" both eye leads (ROC-A) and LOC-A2), a chin EMG, and an EKG lead were monitored. After the initial artifact was noted, the EEG leads were "double referenced" (C4-A I2 and 02-Al)' meaning the combined leads AI and A2 were used as the reference for the central and occipital EEG derivations (see figure). Question: What is the artifact? C4-A , 2 .ifV~ O2 - A ' 2 ROC -At LOC - A2 \J Chin EMG '" T T T T T , t T , t T , , T T T T T t , 1 t t , , EKG 62

Answer: It is an EKG artifact, minimized by double referencing. Discussion: The EKG artifact is one of the most common and easily recognizable recording artifacts. Itcan be identified by sharp deflections in the signals of affected channels corresponding exactly in time to the QRS of the EKG. Fortunately, this artifact does not interfere a great deal with visual sleep staging, as the artifact does not mimic usual EEG patterns. The artifact can be minimized by placing the mastoid electrodes sufficiently high (behind the ear) so that they are over bone instead of neck tissue (fat). Double referencing to both mastoid electrodes also can minimize EKG artifact. This works because if the EKG voltage vector is toward one mastoid, it is away from the other. Hence, the EKG component of the two signals (C4-A1 and C4-A2) tend to cancel each other out. If the recording system does not allow double referencing, then you can link the mastoid electrodes A 1 and A2 with a jumper cable at the electrode box. In the present case, EKG artifact is prominent in the eye leads and chin EMG. It is less prominent in the EEG leads (refer to figure, arrows) because of double referencing. The artifact is larger than desirable, but the record still can be scored. Clinical Pearls 1. EKG artifact can be easily recognized as sharp deflections in the affected leads corresponding to the QRS complex in the EKG lead. 2. Proper application of the mastoid electrodes and double referencing can prevent or minimize this artifact. REFERENCES I. Harris CD, Dexter 0: Recording artifacts. In Shepard JW (ed): Atlas of Sleep Medicine. Mount Kisco, New York, Futura Publishing, 1991. pp 50-51. 2. Butkov N: Clinical Polysomnography. Ashland, OR, Synapse Media, 1996. pp 344-346. 63

PATIENT 20 A 25-year-old man complaining of excessive daytime sleepiness A 25-year-old man was being monitored for complaints of excessive daytime sleepiness. After several minutes of recording, the patient was noted to scratch his chin, and a humming noise was heard from the polygraph pens. A portion of the tracing is shown below. Question: What artifact is responsible for the humming? C 4 - A 1 O 2 -A 1 ROC-A 1 LOC - A 2 chin EMG A 64

Answer: There is a sixty-cycle artifact in the chin EMG channel. Discussion: Sixty-cycle artifact is a common problem in sleep-study recording. It is caused by 60-Hz electrical activity from power lines and can be minimized by correct application of electrodes and proper design of the sleep laboratory. When prominent, the artifact causes a characteristic humming of the pens as they oscillate at 60 cycles per second. The artifact usually is easy to spot in the EEG and EGG leads, but may be more subtle in the EMG leads. Most EEG amplifiers have a 60-cycle notch amplifier to minimize the recording of this signal. If a 60-cycle filter is out (disengaged), the amplitude of the artifact increases tremendously (see figure below). If the paper speed is increased to 60 millimeters per second, then there will be 10 cycles in I centimeter (Yr: sec at that paper speed) or a frequency of 60 Hz. In a digital system, changing the time base to a 5- or IO-second page is equivalent to increasing the paper speed and will also reveal the problem. The problem can also be recognized on either type of system on the usual 3D-second page by a very dense, uniform, "squared off" tracing that does not vary. EEG amplifiers are alternating current (AC)-eoupled, which allows them to record low-voltage EEG activity (50 microvolts) while rejecting high-voltage Paper speed 10 mm/sec 60 mm/sec direct current (DC) activity. Differential amplifiers can record low-voltage physiologic signals by amplifying the difference in voltage between two electrodes while rejecting the common-mode signal consisting of higher-voltage, 60-Hz, background activity. When recording the voltage difference between two electrodes, the background AC activity is rejected only if the electrode impedances are low and fairly equal. If one electrode is faulty (disconnected or high impedance), then the 60-Hz AC activity will be more prominent. Although most AC amplifiers have notch filters to minimize AC activity, these filters may not prevent 60-Hz activity from being prominent when electrode impedances are very different. The ideal impedance of electrodes is below 5000 ohms. Electrode impedance should be checked by the sleep technician after electrode application. In the present case, when the patient scratched his chin he moved one of the EMG electrodes, altering its impedance (refer to previous figure, point A). The problem was fixed (note portion of tracing after A) by switching to a spare chin EMG electrode that had been placed in the submental area. Note the considerable EKG artifact in the left eye channel in this tracing. ( 60 cycle filter on ~\~~~\\\\~ I I 1/6 sec 65

Clinical Pearls I. Sixty-cycle artifact causes a humming in the recording pens. 2. Causes of sixty-cycle artifact include high and unequal electrode impedances (faulty attachment), lead failure, and interference from nearby power lines. 3. Sixty-cycle filters can minimize 60-cycle interference, but the problem often requires switching to a different electrode. 4. The ideal electrode impedance is 5000 ohms or less, but 5000-10,000 is adequate. The impedance of all electrodes should be checked before lights out. Low and equal impedances will minimize 60-cycle electrical interference. REFERENCES I. Harris CD. Dexter DO: Recording artifacts. In Shepard JW (ed): Atlas of Sleep Medicine. Mount Kisco. NY. Futura Publishing Co, 1991,pp 19-23. 2. Butkov N: Atlas of Clinical Polysomnography. Volume II. Ashland, OR, Synapse Media. 1996, pp 331-333. 66

PATIENT 21 A 30-year-old man with loud snoring A 30-year-old man seeking assistance for loud snoring was studied using a digital monitoring system. The EEG, EOG, EMG, and chin leads were acquired in a referential manner (each electrode recorded against a common reference), while the EKG, airflow, chest movement, and abdominal movement were acquired as bipolar tracings. The arterial oxygen saturation (Sa02) was acquired as a DC signal. Only some of the signals showed 60-cycle artifact, which confused the technicians. Questions: What is the problem? How would you fix the bad tracings? _"'1"'1&",.,. ""I'~~I1I... tl ,...... • ,'ttI .. 'y.. \lJWu, .. ..,. lU"."f,'~" ......"...'M••uIH"'hll ~----~---~------ .~ ... ~'~'~~~~~I~~ __.I~ ~~-,~~ ~_»_lM~~U.~¥~~ljll"UIUI. LOC-A2 .,.. --~_._.~ ....... ... ,,--_......... _ "'\Il!'~" ROC-Al C3-A2 C4-Al 01-A2 02-Al chin1-chin3 EKG R,L Legs airflow chest abdomen 5a02 67

Answer: The common referential ground electrode(s) is bad and should be replaced immediately. reference Discussion: Most digital sleep systems now permit acquisition of a mixture of referential, bipolar, and DC signals. Usually C4, C3, 02' 01' AI' A2, chin I, chin2, chin3, ROC, and LOC are each recorded against a common reference electrode (or several linked electrodes; see chart below). Of note, some labs also record the EKG or leg EMGs referentially as well. Referential recording has many advantages. If all referential electrode signals are displayed, this display view can quickly identify if any of the individual electrodes are bad. However, the usual bipolar display views are still possible, as during review the computer than performs subtractions to show typical views. For example (C4-reference)-(AI- reference) = C4-Al. The referential method of acquiring signals allows re-referencing after the data has already been acquired. The original data is not changed, but many display views are possible. You can see any signal against any other signal recorded referentially. If one of the mastoid electrodes should fail during the night, you can also change the display view (for example C4-A2 to C4- AI)' You could look at any combination of chin electrodes (e.g., chin l-chin2, chin l-chin3, chin2-chin3). Referential recording C4 C3 02-- 01 A1 A2 However, a potential disadvantage to referential monitoring is that if the reference electrode (s) should fail, all the referential electrode signals acquired will be bad (usually showing popping or 60- cycle artifact). As discussed in the previous case, the ability of a differential amplifier channel to reject 60-cycle interference depends on the integrity of both signals. If the reference electrode is bad, each of the channels (X-reference) will be bad where X is the electrode acquired referentially (for example, C4 or AI)' True bipolar recording is typically used for airflow, chest and abdominal movement, snoring, EKG, and often leg EMGs. Each sensor has two outputs, and the voltage between them is recorded. None of these signals is affected by the common reference electrode. In the present case, all of the referential electrode tracings showed artifact (refer to previous figure). No bipolar or DC channels showed the problem. This meant that the common reference electrode(s) (or one of the linked electrodes used for the common reference) was bad. When the common ground was fixed, the tracing on the next page was noted. Bipolar recording C4 - A1 C3 -A2 02 - A1 01- A2 Display bipolar views C4-A1 = (C4 - ref) - (A1-ref) C4-A2 = (C4 - ref) - (A2-ref) 68

LOC-A2 ROC-Al C3-A2 C4-Al 01 -A2 w-.~~~~"""""""loo\r'4Jio""V\MN."""""~~~~rw.N 02-Al chin l-chin3 EKG R,L Legs _ airflow chest abdomen Sa02 Clinical Pearls I. Many digital systems allow referential recording of EEG, EGG, and EMG leads. This allows re-referencing and multiple display views during study review. However, failure of the common reference in this system will impair all leads acquired referentially. 2. If all referential leads show artifact, then the problem is the common reference (or one of the linked electrodes used as the common reference) 3. Bipolar and DC channels that do not depend on the common reference will not be affected by failure of the common reference electrode(s). REFERENCES I. Geyer JD, Payne TA, Carny PR, Aldrich MS: Atlas of Digital Polysomnography. Philadelphia, Lippincott Williams and Wilkins, 2000. 2. Butkov N: Polysomnography. In Lee-Chiong TL, Sateia MJ, Carskadon MA (eds): Sleep Medicine. Philadelphia, Hanley and Belfus, 2002, pp 605-637. 69

PATIENT 22 A 40-year-old man with complaints of snoring An obese 40-year-old man was monitored in the sleep laboratory. The EEG, EGG, and EMG tracings showed a definite artifact. Below is a sample 15-second tracing. Questions: What is this type of artifact? How can it be minimized? LOC-A 2 Chin EMG . 5 SEC 70

Answers: This is sweat or slow frequency artifact. It can be minimized by cooling the patient. Discussion: Sweat artifact (or slow frequency artifact) is characterized by a slowly undulating movement of the baseline of affected channels. The movement mayor may not be synchronous with the patient's respiration. When in-phase with the patient's respiration, the artifact also is called respiratory artifact. Sweat artifact is believed to be secondary to the effects of perspiration. Sweat alters the electrode potential, thereby producing an artifact that mimics delta waves and results in overscoring of stages 3 and 4 NREM sleep. When the artifact is not present in all channels, it may be secondary to pressure on an electrode (or pulling on the electrode). In any case, the artifact is usually coming from one or more electrodes on the side the patient is lying on. For example, if the patient is sleeping with the left side down and C4-A1, 02-A" and ROC-AI are affected, but LOC-A2 shows no artifact, then lead AI requires attention. Switching to CrAz or C4-Az may be tried, but if switching electrodes does not solve the problem, then other actions are necessary. Options include reducing the room temperature, uncovering the patient, and/or using a fan. As a lastditch alternative, the setting of the low-frequency filter may be increased (e.g., from 0.3 to I). Unfortunately, this maneuver decreases the amount of delta activity recorded, but still may be preferable to a totally unscorable record. Sweat artifact can be prevented by maintaining a low room temperature, especially when very obese or heavily perspiring patients are studied. In the present case, the sweat artifact is present in leads referenced to both AI and Az. The room temperature was lowered, and the patient was uncovered. Over the next 15 minutes, the artifact resolved. Clinical Pearls I. Sweat or respiratory artifact is characterized by a slowly undulating baseline. 2. Maintaining a sufficiently cool room temperature is essential when studying obese patients. 3. Changing the electrodes to those opposite the side the patient is lying on can eliminate the artifact in some cases. 4. If all electrodes are involved, use a fan or lower the room temperature. REFERENCES 1. Harris CD, Dexter 0: Recording artifacts. In Shepard JW (ed): Atlas of Sleep Medicine. Mount Kisco. New York. Futura Publishing, 1991, pp 50-51. 2. Butkov N: Atlas of Clinical Polysomnography, Vol II. Ashland, OR, Synapse Media, 1996, pp 348-349. 71

PATIENT 23 Two patients with recording artifacts Patient A: A 40-year-old man with frequent awakenings at night was monitored in the sleep laboratory. The EEG, EGG, and EMG tracings showed a definite artifact. Below is a sample IS-second tracing. Questions: What is the artifact? Which lead is responsible for the problem? LOC· Al Chin EMG .,. , ..... 4 ••• ,., FI Or I. I' .. i. 1 • .,....' Patient B: A 50-year-old man was undergoing sleep monitoring as an evaluation for excessive daytime sleepiness. The tracing (below) looked much like REM sleep except for an unusual pattern in the eye leads showing deflections only in the right eye (points A). Questions: What sleep stage is shown? How do you know the eye channels are working? ROC· A1 72 chin EMG A A A A A

Answers: Patient A- This artifact is due to electrode popping in lead A I. Patient B - Stage REM in a patient with an artificial left eye. Check the biocalibrations. Discussion: Electrode popping is a common and severe artifact that makes the staging of sleep very difficult. It is characterized by sudden, highamplitude deflection (channel blocking) secondary to an electrode pulling away from the skin (sudden loss of signal). The popping tends to be regular and corresponds to body movement during breathing. Electrode popping often is caused by the patient lying on one mastoid electrode or pulling on an electrode during respiration. Popping also can occur if the electrode gel dries out during the night. This artifact frequently can be handled by switching to an alternate lead. For example, if 0, is the problem, the exploring occipital electrode is switched to 0 I' This is one reason that redundant electrodes are routinely placed. Alternatively, the offending electrode is repaired by adding electrode gel, or replaced. In Patient A, the regular high-voltage deflections are noted in all EEG and EOG channels except 0,- A2. The common electrode to all the affected channels is AI; therefore, the problem is most likely in electrode Ar- The patient was sleeping on his left side. After changing the reference electrode to A, (C3-A2, ROC-A2, LOC-A2) the problem was eliminated. The change could have been to C4-A2, but using an exploring electrode opposite the reference is preferable (if possible) as this produces a larger voltage signal. Biocalibrations are an important initial part of all sleep studies (see Fundamentals 8). Patients are asked to look left, right, up, and down to test the effectiveness of both the eye electrodes and the amplifier settings at detecting eye movements. In Patient B, no movement was seen in LOC-A, during biocalibration. The technician questioned-the patient, who reported that he had an artificial (glass) left eye. At the usual amplifier settings, movements of the right eye (relatively far away) did not result in deflections in LOC-A,. The absence of any deflection in the EEG lead-s coincident with the deflections in ROC-AI indicated that these deflections were due to eye movements and not to transmitted EEG activity. The eye movements, low-voltage EEG, and low-amplitude EMG are consistent with stage REM sleep. Clinical Pearls Patient A: I. Electrode popping artifact is a sudden, high-voltage deflection occurring at regular intervals usually coincident with respiration. 2. Electrode popping artifact is due to an electrode pulling away from the skin. 3. The offending electrode sometimes can be identified by noting if the affected channels have a common electrode. 4. Recording from alternative electrodes may eliminate the problem of electrode popping. Patient B: I. Always check biocalibrations prior to sleep study interpretation. 2. Pay special attention to the deflections resulting from voluntary eye movements. In the usual two-channel setup for detecting eye movements, proper calibration should result in reasonably sized, out-of-phase deflections. REFERENCES I. Harris CD, Dexter D: Recording artifacts. In Shepard JW (ed): Atlas of Sleep Medicine. Mount Kisco, New York, Futura Publishing, 1991, pp 30-31. 2. Butkov N: Polysomnography. In Lee-Chiong TL. Sateia MJ, Carskadon MA (eds): Sleep Medicine. Philadelphia, Hanley and Belfus, 2002, pp 605-637. 3. Caraskadon MA, Rechschaffen A : Monitoring and staging human sleep. In Kryger MH. Roth T, Dement WC (eds): Principles and Practice of Sleep Medicine, 3rd ed. Philadelphia, WB Saunders Co., 2000, pp 1197-1215. 73

PATIENT 24 A 29-year-old man struggling with daytime sleepiness A 29-year-old, generally healthy man was studied to evaluate complaints of daytime sleepiness. He was taking no medications, and his wife reported only occasional snoring. During the initial part of the test, the technician noted considerable sweat artifact (the air-conditioner thermostat was malfunctioning). The low-frequency filter (j-j amp) setting on the EEG channels was increased from 0.3 to I to control the artifact after only 30 minutes of recording. Physical Examination: Unremarkable. Sleep Study Time in bed (monitoring time) Total sleep time (TST) Sleep period time (SPT) WASO Sleep efficiency Sleep latency 435 min (430-454) 406.5 min (405-434) 432.5 min (410-439) 26 min 93% ( 91-99) 2.5 min (3-26) Sleep Stages Stage Wake Stage I Stage 2 Stages 3 and 4 Stage REM % SPT 1 (0-1) 6 (3-6) 60 (40-51) 8 (16-26) 25 (22-34) ( ) = normal values for age, sleep efficiency = (TST * 100) / TIB Question: Can you explain the abnormality in sleep architecture? 74

Answer: Reduced recording of slow wave activity secondary to the low-frequency filter setting. 1/2 amp Lo .3 Hz 1Hz 37~LIn the present case, the change in low filter settings was the most likely cause for the small amount of slow wave sleep recorded. When the study was scored without using voltage criteria, a higher but still subnormal amount of slow wave sleep was scored. The low filter setting should be increased only as a last resort (see treatment of sweat artifact, Patient 22). Fe = 1/(21T Tc). The Tc = RC where R is the resistance and C the capacitance. Since amplifiers are routinely calibrated by step (square wave) voltage changes rather than sine wave signals, the time constant can be noted from the time it takes for the deflection to return to baseline. In RC circuits, an increase in step voltage produces an abrupt increase in voltage across the resistor, then an exponential fall in voltage to lie (0.37) of the maximum voltage in one time constant. When the amp frequency is higher, the time constant is smaller (more rapid fall). Discussion: The high- and low-frequency filter settings can significantly alter the EEG amplifier response to a signal. The low filter settings can dramatically reduce the amplitude of slow wave activity if set too high. A amp low-filter setting of 0.3 Hz means that a sine wave input of 0.3 Hz is attenuated by 50%. On some amplifiers, a low filter setting of 0.3 means that a signal of 0.3 Hz is attenuated by 70.7 or 80% of the original signal. Regardless of the exact meaning for a given amplifier, a low filter setting of 0.3 does not significantly attenuate the majority ofslow wave activity « 2 Hz). However, a low filter setting of I Hz does produce considerable attenuation of slow wave activity. Thus, less activity meets the minimum voltage criterion of 75 microvolts. At the bottom of the page is a tracing showing the effect of changing the low filter from a amp setting of 0.3 Hz (usual setting) to I Hz. Note the abrupt reduction in the slow wave activity. Setting the low-frequency filter of the eye movement channels too high also markedly reduces eye movement amplitude. The low filter setting usually recommended for EEG and EOG leads is 0.3 Hz. For EMG and EKG channels, a low filter setting (~ amp) of 10 is used, as the relevant activity is of a much higher frequency. Filter settings are sometimes given as time constants. Most altematingcurrent amplifiers employ resistance-capacitance (RC) circuits as input filters. In a simple, low filter RC circuit, the frequency (fc) at which the output voltage across the resistor is attenuated to 70.7% of the input voltage is related to the time constant (Tc) by the formula t---4 tc I I 1 sec 1/2 amp LO .3 hz 1 hz chin EMG 5 sec 75

Clinical Pearls I. The filter settings should be recorded during calibration and any changes during sleep monitoring noted. The sleep scorer should be informed of any changes. 2. A higher-than-recommended low filter setting in the EEG leads (especially the central EEG) decreases the amplitude of slow wave activity and hence the amount of slow wave sleep that is scored. REFERENCES I. Tyner FA, Knott 1R, Mayer WB, 1r: Fundamentals of EEG technology. New York, Raven Press, 1983, pp 89-119. 2. Fisch B1: Spehlman's EEG Primer. New York, Elsevier, 1991, pp 51-60. 76

FUNDAMENTALS OF SLEEP MEDICINE 9 Multiple Sleep Latency Test and Maintenance of Wakefulness Test The multiple sleep latency test (MSLT) consists of sleep monitoring during five naps spread over the day, usually at two-hour intervals (10 AM, 12 noon, 2 PM, 4 PM, 6 PM). Generally, the test is preceded by nocturnal polysomnography. In the morning, the patient changes into comfortable street clothes, and nap monitoring begins 1.5 to 3 hours after nocturnal recording has ceased. Monitoring includes central EEG, occipital EEG (optional but recommended), chin EMG, EKG, and airflow (if sleep apnea is a possibility). The patient is instructed to fall asleep at lights out and is given 20 minutes to do so. Once sleep is attained, the patient is given another 15 minutes to reach stage REM sleep. The nap test is stopped if the patient fails to fall asleep within 20 minutes, or fails to reach REM sleep within 15 minutes of sleep onset. After nap termination, the patient is instructed to get out of bed and remain awake until the next nap opportunity. The sleep latency (time from lights out until the beginning of the first epoch of any stage of sleep) and the REM latency (time from the first sleep until the beginning of the first epoch of REM sleep) are determined for each nap. Results ofthe MSLT in normal subjects show a mean sleep latency longer than 15 minutes and zero to one REM periods in five naps (Table I). However, some otherwise normal subjects have a mean sleep latency in the 10- to 15-minute range. This range represents mild sleepiness, and a mean sleep latency < 10 minutes is considered abnormal. A mean sleep latency of 5-10 minutes represents moderate sleepiness; shorter than 5 minutes is severe (pathological) sleepiness. It is important not to confuse mean MSLT sleep latency with nocturnal sleep latency. A short sleep latency at the regular bedtime is not abnormal. However, a majority of patients with narcolepsy have a nap sleep latency shorter than 5 minutes, and patients with moderate-to-severe sleep apnea usually have a mean nap sleep latency shorter than 10 minutes. Table 1. Multiple Sleep Latency Test Results MEAN SLEEP LATENCY REM ONSETS < 5 min 5-10 min 10-15 min Severe sleepiness Moderate sleepiness Mild sleepiness oor I in 5 naps 2 or more in 5 naps Normal Abnormal The presence of two or more REM periods in five naps is characteristic of narcolepsy. However, only 70-80% of patients with narcolepsy and cataplexy will have this finding on a given day. Furthermore, two or more REM periods is not specific to narcolepsy and can occur with any cause of REM sleep deprivation or disturbance. Sleep apnea and, occasionally, psychiatric disorders can be associated with a short REM latency. Acute withdrawal of REM-suppressing medications (tricyclic antidepressants, lithium, serotonin reuptake inhibitors, stimulants) can be associated with REM sleep during naps. Ideally, any medication affecting either the sleep latency or the REM latency should be withdrawn for 2 weeks before the sleep study and MSLT. When this is not practical, the medications should not be abruptly discontinued just prior to testing. Most sleep centers have the patient keep a sleep log (diary) of the amount and pattern ofsleep during the 2 weeks preceding the MSLT, because sleep loss or disturbance during this period can affect the sleep latency. Proper interpretation of the MSLT requires analysis of the nocturnal polysomnogram. Note if the 77

amount and quality of sleep were adequate and if sleep disorders were present that could affect the sleep latency. Specific conditions such as sleep apnea or periodic leg movements in sleep should be documented. Decreased amounts of REM sleep (or REM-sleep fragmentation) during the night from any cause can increase the number of REM periods recorded on the MSLT. For example, sleep apnea is a common cause of two or more REM periods on the MSLT. In general, if the nocturnal sleep is significantly disturbed it is best to cancel the MSLT and repeat the test after the identified sleep disorder is adequately treated. The standard MSLT criteria for narcolepsy are: mean sleep latency shorter than 5 minutes and two or more REM episodes in five naps (Table 2). Narcoleptic patients with a longer sleep latency (5-10 minutes) and two more REM onsets often show a shorter sleep latency on retesting. When sleep apnea is present, interpretation of the MSLT is problematic. The sleep apnea should be treated first. If narcolepsy is suspected, a repeat sleep study showing adequate sleep (adequate treatment of sleep apnea) should be performed, followed by an MSLT. When treatment is with nasal continuous positive airway pressure (CPAP), both the repeat sleep study and the MSLT are performed on the prescribed level of CPAP. Table 2. MSLT Findings in Patients Evaluatedfor Daytime Sleepiness NARCOLEPSY WITH SLEEP-RELATED CATAPLEXY BREATHING DISORDERS ;::: 2 sleep onset REM period Mean sleep latency < 5 minutes Both 74% 87% 67% 7% 39% 4% From Aldrich MS, Chervin RD, Malow BA. Value ofthe multiple sleep latency test (MSLT) forthe diagnosis of narcolepsy. Sleep 1997; 20:620-629; with permission. The maintenance of wakefulness test (MWT) was designed to test the patient's ability to stay awake. The patient is seated upright in bed in a dimly lighted room and asked to remain awake for either 20 or 40 minutes. The usual EEG, EOG, and EMG monitoring is performed to detect sleep. The test is terminated if sleep is noted or after 20/40 minutes if the patient maintains wakefulness. The test is repeated four to five times across the day, and the mean sleep latency is determined (20 or 40 minutes if no sleep is recorded). When both the MSLT and MWT were administered to a group of patients with excessive daytime sleepiness, the correlation was significant but low. Several individuals did not fall asleep during the MWT, but had some degree of daytime sleepiness as assessed by the MSLT. The MWT is more likely than the MSLT to show improvement after treatment of daytime sleepiness. In one study, the MWT sleep latency increased from 18 to 31 minutes in a group of patients with obstructive sleep apnea (OSA) after adequate treatment. One normative study (Table 3) has suggested that a normal MWT latency should be > 19 minutes on a 40-minute test (sleep defined as three consecutive epochs of stage I or anyone epoch of another stage of sleep) or > II minutes on an abbreviated 20-minute MWT (sleep defined as any epoch of sleep). Currently, there is no consensus on what constitutes a normal MWT. Table 3. Proposed "Normal" for the Maintenance of Wakefulness Test LENGTH OFMWT MEAN SLEEP LATENCY 20-minute naps 40-minute naps Normal> II min Normal> 19 min The MWT latency required for a person to safely pursue an occupation critically dependent on alertness has not been standardized. Furthermore, the ability to stay awake is not the same as maintaining alertness. Studies using driving simulators have attempted to provide a performance-based test of alertness. Test results showed decreased alertness in patients with OSA and in patients with narcolepsy, as compared to a control group. However, these results did not correlate with MSLT results, and half of each group performed as well as controls. While these studies are important first steps, they have not been validated by performance tests of the real thing. 78

Key Points I. Proper interpretation of a multiple sleep latency test (MSLT) requires analysis of: the preceding nocturnal sleep study, a careful medication history, and sleep habits (diary) for at least I week preceding the MSLT. 2. The mean nap sleep latency is a measure of the tendency to fall asleep during normal waking hours. 3. The number of naps with REM sleep can provide evidence for narcolepsy. However, this finding is neither highly sensitive nor specific for the disorder. 4. Obstructive sleep apnea (OSA) is also a common cause of two or more REM periods on an MSLT. Therefore, if OSA is present it should be adequately treated before the MSLT can be used to support a diagnosis of narcolepsy. 5. The maintenance of wakefulness test may be more likely to show a significant improvement after treatment of daytime sleepiness. REFERENCES Multiple Sleep Latency Test I. Richardson GS. Carskadon MA. Flagg W: Excessive daytime sleepiness in man: Multiple sleep latency measurements in narcoleptic and control subjects. Electroencephalogr Clin Neurophysiol 1978; 45:621-627. 2. Carskadon MA: Guidelines for the multiple sleep latency test. Sleep 1986; 9:519-524. 3. Standards of Practice Committee. American Sleep Disorders Association: The clinical use of the multiple sleep latency test. Sleep 1992; 15:268-276. 4. Chervin RD, Aldrich MS. Sleep onset REM periods during multiple sleep latency tests in patients evaluated for sleep apnea. Am J Respir Crit Care Med 2000; 161:426--431. 5. Aldrich MS. Chervin RD. Malow BA. Value of the multiple sleep latency test (MSLT) for the diagnosis of narcolepsy. Sleep 1997; 20:620-629. Maintenance of Wakefulness Test 1. Poceta, JS. Timms RM. Jeong D. et al: Maintenance of wakefulness test in obstructive sleep apnea syndrome. Chest 1992; 10I:893-902. 2. Sangal RB, Thomas L. Mitler MM: Maintenance of wakefulness test and multiple sleep latency test: Measurements of different abilities in patients with sleep disorders. Chest 1992; 101:898-902. 3. George CFP. Boudreau AC. Smiley A: Comparison of simulated driving performance in narcolepsy and sleep apnea patients. Sleep 1996; 19:711-717. 4. Doghramji K, Mitler MM, Sangal RB, et al: A normative study of the maintenance of wakefulness test (MWT). Electroencephalogr Clin Neurophysiol 1997; 103:554-562. 79

PATIENT 25 A 25-year-old man with daytime sleepiness A 25-year-old man complained that for 2 years he'd had problems with sleepiness during the day. A nocturnal sleep study failed to show any abnormality. A multiple sleep latency test (MSLT) was performed the next day. Figure: The results of the five naps are shown below in tabular format. The sleep stage of each 30- second epoch is listed below the epoch number. Questions: What is the sleep latency and REM latency of each nap? What is the mean sleep latency for the MSLT? Nap 1 R = REM. LO = lights out, W =stage Wake epoch 70 71 72 73 74 75 76 77 78 79 80 stage LO W W W I 1 I I 2 2 R Nap 2 epoch 90 91 92 93 94 95 96 97 98 99 100 stage LO W W I I I I I 2 2 2 epoch 101 102 103 104 105 106 107 108 109 110 III stage 2 3 3 3 4 4 4 4 4 4 4 epoch 112 113 114 115 116 117 118 119 120 121 122 stage 2 3 3 3 4 4 4 4 4 4 4 Nap3 epoch 130 131 132 133 134 135 136 137 138 139 140 stage LO W W W I I I I I I I epoch 141 142 143 144 145 146 147 148 149 150 151 stage 2 2 2 2 3 3 3 3 R R Nap4 epoch 160 161 162 163 164 165 166 167 168 169 170 stage LO W W W W W I 1 W W I epoch 171 172 173 174 175 176 177 178 179 180 181 stage 2 3 3 3 4 4 4 R R Nap 5 epoch 190 191 192 193 194 195 196 197 198 199 200 stage LO W W W W W W W W W W epoch 201 202 203 204 205 206 207 208 209 210 211 stage W W W W W W W W W W W epoch 112 113 114 115 116 117 118 119 120 121 122 stage W W W W W W W W W W W epoch 123 124 125 126 127 128 129 130 131 132 133 stage W W W W W W W W 80

Answers: Nap I Nap 2 Nap 3 Nap4 Nap 5 Mean Sleep Latency 1.5 min (3 epochs) 1.0 min (2 epochs) 1.5 min (3 epochs) 2.5 min (5 epochs) 20 min 5.3 min REM Latency 3 min (6 epochs) None 7.5 min (15 epochs) 6 min (12 epochs) None Discussion: The patient is given 20 minutes after lights out to fall asleep for each MSLT nap. If sleep does not occur, then the sleep latency is set at 20 minutes by convention. Once sleep is attained, the patient is given another 15 minutes to attain REM sleep. The sleep latency is the time from lights out until the first epoch of any stage of sleep. The REM latency is the time from the first epoch of sleep until the first epoch of REM sleep. There is a normal variation in sleep latency over the day in most subjects, with the minimum usually occurring near noon or early afternoon (naps 3 or 4) and the maximum in the late afternoon. The propensity of REM also varies, with REM periods most likely to occur in the morning naps. In the present patient, the mean sleep latency was consistent with moderate-to-severe sleepiness. Note that in nap 5 no sleep was attained, and the sleep latency was set at 20 minutes. The patient had difficulty falling asleep in this nap because he was "nervous and ready to go home." If nap 5 were excluded, the mean sleep latency would be much lower. REM sleep was present in three of five naps. In nap 4, epochs of wakefulness were noted between sleep onset and the first epoch of REM. The intervening wakefulness was included in the computation of the REM latency. The findings of this study were interpreted in light of the nocturnal polysomnographic findings and clinical history, and a diagnosis of narcolepsy was supported. (See Patients 77 and 78 for detailed discussions of narcolepsy). Clinical Pearls I. If no sleep is attained after 20 minutes of monitoring during an MSLT, then the nap is terminated, and the sleep latency is considered to be 20 minutes. 2. Sleep latency can vary considerably between naps in the MSLT. This is the reason for five naps spread out over the normal waking period. 3. The normal propensity for REM sleep is highest during the first naps. 4. The sleep latency tends to be shortest at noon and early afternoon naps. REFERENCES 1. Richardson GS, Carskadon MA, Flagg W: Excessive daytime sleepiness in man: Multiple sleep latency measurements in narcoleptic and control subjects. Electroencephalogr Clin Neurophysiol 1978; 45:621-627. 2. Carskadon MA: Guidelines for the multiple sleep latency test. Sleep 1986; 9:519-524. 3. Standards of Practice Committee, American Sleep Disorders Association: The clinical use of the multiple sleep latency test. Sleep 1992; 15:268-276. 81

FUNDAMENTALS OF SLEEP MEDICINE 10 Monitoring Respiration During Sleep Respiratory Definitions. The apnea + hyponea index (AHl), sometimes called the respiratory disturbance index (RDl), is the number of apneas + hyponeas per hours of sleep. Apnea is defined as an absence of airflow at the nose and mouth for IOseconds or longer. This time duration is arbitrary, but widely applied. The presence or absence of respiratory (inspiratory effort) determines the type of apnea. Central apnea is defined by an absence of inspiratory effort. Obstructive apnea occurs despite persistent respiratory effort. Mixed apnea has an initial central part (no inspiratory effort) and a terminal obstructive portion. In the figure below, respiratory effort (signaled by movement) is detected by bands around the chest and abdomen as well as esophageal pressure deflections. In central apnea, no movement of the chest and abdomen or esophageal pressure deflections are detected during the apnea. In obstructive apnea, respiratory effort persists during the apnea. Note that during obstructive apnea, the chest and abdomen often move in a paradoxical manner (one inward and the other outward, see arrows). Esophageal pressure deflections increase during the terminal portion of apnea. In mixed apnea, an initial central portion of the apnea (point C) is followed by an obstructive portion (Fig. I). APNEATYPES CENTRAL Airflow NV"v------/\A Chest Abdomen /\IV'v-----I\/\ Esoph pressure MIXED Airflow Chest Abdomen EsOph pressure FiGURE I OBSTRUCTIVE 1 Paradoxical Movement Hypopnea definitions vary widely and are controversial. Basically, hypopnea is a reduction in airflow. However, how airflow is measured and whether an associated desaturation or arousal is required can significantly affect the number of hypopneas detected in a given patient. This prompted some to characterize hypopnea as" a floating metric." Techniques of airflow detection/measurement are discussed below. Desaturation means a drop in the arterial oxygen saturation from baseline. A consensus conference sponsored by the American Academy of Sleep Medicine (AASM) published one set of guidelines for hypopnea indentification 82

("Chicago criteria"), and subsequently the Clinical Practice Review Committee (CPRC) of the same organization advocated still another definition of hypopnea. Of note, the Centers for Medicare and Medicaid Services (CMS) has adopted the CPRC definition. In the table below, the first two definitions are samples of what has been used in some laboratories, and the second two summarize the "Chicago" and CMS criteria. Four Definitions of Hypopnea Hypopnea I Hypopnea 2 AASM Consensus Conference "Chicago criteria" AASM-CPRC and CMS (Medicare) A B AIRFLOW REDUCTION 50% reduction in airflow for 10 seconds or longer 30% reduction in flow for 10 seconds or longer • 50% reduction in airflow from baseline for IO seconds or longer • Airflow measured using: pneumotachograph, nasal pressure, or RIP sum reduction in both channels of dual-channel RIP (RC, AB) • Discernable reduction in airflow signal from baseline for 2: IO seconds (but less than a 50% reduction) • Airflow measured using pneumotachograph, nasal pressure, or RIP sum with reduction in either channel of dual-channel RIP (RC,AB) Reduction in airflow by 30% from baseline for 2: 10 seconds DESATURATION OR AROUSAL Not required 2: 2% desaturation or arousal Not required 2: 3% desaturation or arousal 2: 4% drop in Sa02 (arousal not considered) RIP = respiratory inductance plethysmography. RC = rib cage, AS = abdomen. 4% desaturation means a 4% drop in the oxygen saturation (SaO z) The CPRC provided several reasons for their choice of hypopnea definition. First, the scoring of desaturation has good intra- and inter-scoring reliability while the scoring of arousals does not. Second, the Sleep Heart Health Study using this definition of hypopnea was able to show that even mild elevations of the AHI (2: 5/hr) are associated with an increased risk of cardiovascular disease. However, this definition does not recognize the sleep-disturbing effects of hypopneas associated with arousal but less than a 4% desaturation. Another problem in any definition of hypopnea is that the "baseline" flow may be hard to define in patients with repetitive apnea/hypopnea. Hypopneas can be further classified as central, obstructive, or mixed. However, classification is presumptive unless upper airway resistance is measured (supraglottic pressure or esophageal pressure deflections). Central hypopneas are characterized by a reduction in flow that is proportional to the reduction in inspiratory effort (see decreased esophageal pressure deflections, Fig. 2, arrows). Obstructive hypopneas are produced by upper airway narrowing. There is an increase in esophageal pressure deflections Central Hypopnea airflow (nasal pressure) Esophageal pressure airflow (nasal pressure) Esophageal pressure Obstructive Hypopnea FIGURE 2 83

which may have a crescendo pattern prior to event termination (Fig. 2, dots) compared to baseline deflections (b). Obstructive hypopneas are often associated with reduction in chest and abdominal movements, which tend to be paradoxical (Fig. 3, double arrow). If airflow is monitored with a pneumotachograph or nasal pressure, airflow has a flattened shape in obstructive hypopneas in contrast to the round shape during central hypopneas. The detection of snoring during a hypopnea also suggests an obstructive nature. A mixed hypopnea is characterized by both a decrease in respiratory effort and an increase in upper airway resistance. However, in routine clinical practice, hypopneas are rarely classified. RIPsum RC AB t inspiration FIGURE 3. Obstructive hypopnea with respiratory impedance plethysmography. The apnea index (AI), hypopnea index (HI), and the apnea + hypopnea index (AHI) are all defined as the total number of events divided by the total sleep time in hours. The AHI is commonly used to quantify the severity of OSA. An AHI < 5/hr is considered normal in adults. Guidelines for severity are: AHI 5 to < 15 /hr = mild OSA, AHI l5-30/hr = moderate OSA, and AHI > 30/hr = severe OSA. However, these are only rough guidelines. The validity of these cutoffs is questionable unless you use the same measurement techniques and hypopnea definitions that were used to establish these values. It is also often helpful to compute the AHI for NREM, REM, and the supine and non-supine sleeping positions separately. Doing so can identify REM-associated or positional OSA (AHI more than twice as high in a given situation). Techniques to Measure Airflow or Tidal Volume. Until recently, airflow was detected in the clinical setting using thermistors or thermocouples, which measure changes in temperature induced by airflow. They are generally adequate to detect an absence of airflow (apnea), but their signal does not vary proportionately to airflow. Hence, they are not accurate means of detecting hypopnea (changes in airflow). The gold standard for measuring airlow is a pneumotachograph. This device works by allowing measurement of the pressure drop across a linear resistance (wire screen): Pressure = Flow x Resistance. The resistance is fixed over a range of flows. The pneumotachograph is worn in a mask covering the nose and mouth. Recently, nasal cannula connected to pressure transducers have been used to measure nasal pressure (pressure change across the nasal inlet). Unlike the pneumotach, the resistance is not constant, and: Pressure = constant (flow)? or Flow = constant x vnasal pressure. However, in clincal practice nasal pressure rather than the square root is used as a flow signal. While this is much more accurate than using a thermistor, it tends to underestimate airflow at low flow rates and overestimate flow at high flows (Fig. 4). If the nasal pressure signal is "linearized by taking the square root," it very closely approximates the flow from a pneumotachograph. In both pneumotachographs and nasal pressure the signal vs time profile (shape) gives additional important information. The profile is flattened during airflow limitation (constant or decreased flow despite increased driving pressure). Airflow limitation is characteristically present during obstructive hyopopneas or snoring. In central hypopnea, the nasal pressure signal magnitude is reduced, but the profile is round. The most important limitation of the nasal pressure technique is that about 10% of patients are "mouth breathers," and the nasal pressure signal will be misleading. Another problem is that events that are actually hypopneas may be classified as apneas (nasal pressure underestimates flow at low flow rates). The former problem is handled by the simulataneous use of a nasal-oral or oral thermistor. The second issue is not really a problem in clinical practice unless it is essential that apnea and hypopnea 84

Pneumotachograph Flow Linearized Nasal Prongs o 10 20 30 40 time (sec) FIGURE 4. Nasal pressure compared to pneumotachograph during an obstructive hypopnea. be differentiated. Many sleep laboratories consider a drop in flow to below 10-25% of baseline as an apnea. Respiratory inductance plethysmography is another method that can be used to detect apnea and hypopnea (refer to Fig. 3). The ribcage and abdominal sensorsignals can be summed in an uncalibrated manner (RIPsum = RC + AB) or as a calibrated signal (RIPsum = [a X RC] + [b X AB]) as an estimate of tidal volume (not flow). Here a and b are calibration factors determined during a calibration procedure. During an apnea the sum of rib cage and abdomen signals may be near zero. In the case of hypopnea there is a reduction in the RIPsum signal (low tidal volume) as well as both the RC and AB signals. In the case of obstructive hypopnea, there may also beparadox, with chest and abdomen moving in opposite directions (Fig. 3, double arrow). Exhaled CO2 has also been used to detect airflow. A small nasal cannula connected to a COz monitor samples the exhaled air and gives a value as the PCO z (partial pressure ofCOz)' During normal tidal breathing the end tidal PCOz (peak value from each breath) is an estimate of arterial PCO z (see Patient 26). This type of monitoring is most frequently used in children, in whom 'periods of hypoventilation (increased end tidal PCOz) can be detected (see Patient 64). In adults, absence of fluctuation in the exhaled CO2 signal can be used to detect apnea. However, the signal can also be absent if the patient switches to mouth breathing. Hypopnea Detection OBSTRUCTIVE HYPOPNEA HYPOPNEA CENTRAL HYPOPNEA Nasal pressure RIP Esophageal pressure Reduced deflection Reduced deflection in RIPsum, RC, AB Depends on type Flattened shape Paradox in RC and AB Increased deflections Round shape No paradox in RC and AB Decreased deflections RIP = respiratory inductance plethysmography, RC = rib cage, AB = abdomen Measuring Respiratory Effort. Inspiratory effort is usually detected with a piezo-electric sensor connected to a band around the chest and abdomen. These are inexpensive, but can be misleading if the belts are not applied properly. The signal from these devices depends on the degree of tension on the transducer. They are adequate in detecting effort in most patients, but do not really quantify the changes in chest or abdominal volume during breathing. Respiratory inductance plethysmography is a more accurate 85

method of detecting changes in chest and abdominal volume. The inductance of coils in bands around the chest (ribcage) or abdomen changes during movement (dependent on the cross-sectional area encircled by the band) and is converted to a voltage. The ribcage and abdomen signals can be added as explained above. If tightly calibrated, during obstructive apnea RIPsum :::: zero (see Patient 26). The most sensitive method of detecting respiratory effort is by measurement of esophageal pressure. Changes in esophageal pressure are estimates of changes in pleural pressure. Such monitoring is routinely performed in only a few centers. Esophageal pressure is measured using air-filled balloons, fluid-filled catheters, or catheters with pressure transducers on their tips. Measuring Arterial Oxygen Saturation. Arterial oxygen saturation (SaO z) is measured during sleep studies using pulse oximetry (finger or ear probes). A desaturation is defined as a decrease in SaOz of 4% or more from baseline. Note that the nadir in SaO? commonly follows apnea (hypopnea) termination by approximately 6-8 seconds (longer in severe desaturations; Fig. 5). This delay is secondary to circulation time and instrumental delay (the oximeter averages over several cycles before producing a reading). In Figure 5, the apneas and the corresponding nadirs in saturation are identified. Various measures have been applied to assess the severity of desaturation, including computing the number of desaturations, the average minimum SaOz of desturations, the time below 80%, 85%, and 90%, as well as the mean SaOz and the minimum saturation during NREM and REM sleep. Oximeters may vary considerably in the number of desaturations they detect and their ability to discard movement artifact. Using long averaging times can dramatically affect the detection of desaturations (Fig. 6). Here, the results of identical oximeters monitoring the same patient.are shown. The averaging time of one of the oximeters is increased, and there are obvious large differences in the results. The ability of oximeters to detect desturations is especially important in light of the recent CMS definition of hypopnea that depends critically on detection of desaturation. Airflow 35sec 48 sec 82% 26sec 75% FIGURE 5 180 240 averaging time 21 sec ( ) 120 nme (s) o 60 averaging time 3 sec ~'f ) FIGURE 6 86

REFERENCES I. Block AJ. Boysen PG. Wynne JW, et al: Sleep apnea, hypopnea, and oxygen desaturation in normal subjects: A strong male predominance. N Engl J Med 1979: 330:513-517. 2. Redline S, Sander M: Hypopnea, a floating metric: Implications for prevalence, morbidity estimates. and case finding. Sleep 1997:20: 1209-1217. 3. American Academy of Sleep Medicine Task Force: Sleep-Related breathing disorders in adults: Recommendation for syndrome definition and measurement techniques in Clinical Research. Sleep 1999:22:667-689. 4. Redline S, Kapur VK, Sanders MH, et al: Effects of varying approaches for identifying respiratory disturbances on sleep apnea assesment. Am J Respir Crit Care Med 2000; 161:369-374. 5. Farre R. Rigau J. Montserrat JM. Balleste E. Navaja D: Relevance of linearizing nasal prongs for assessing hypopneas and flow limitation during sleep. Am J Respir Crit Care Med 2001: 163:494-497. 6. Meoli AL, Casey KR, Clark RW: Clinical Practice Review Committee-AASM. Hypopnea in sleep-disordered breathing in adults. Sleep 200 I;24:469--470. 7. Berry RB: Nasal and esophageal pressure monitoring. In Lee-Chiiong TL, Sateia MJ. Caraskadon MA (eds): Sleep Medicine. Philadelphia, Hanley and Belfus, 2002. pp 661-671. 87

PATIENT 26 A 45-year-old man with possible sleep apnea A 45-year-old man underwent polysomnography for evaluation of suspected sleep apnea. Apnea detection was performed by monitoring airflow with a thermocouple and measuring exhaled COo. Figure: Although the thermocouple revealed apnea (no deflections), the CO2 tracing revealed persistent deflections that differed from those observed during the pre-apnea period. Questions: Is apnea present? If so, why is the exhaled CO2 tracing fluctuating? rincreasing Exhaled CO 2 Airflow (Thermocouple1 88