Apneas or hypopneas that occur within the hypnagogic state should be just as important to clinicians as those events that occur during 30-second epochs nominally scored as a particular stage of sleep.

photoMany sleep researchers and clinicians who have had considerable experience in scoring and interpreting polysomnograms have, at one time or another, scored or reviewed a record in which the objective measurements of sleep as defined by Rechtschaffen and Kales criteria1 indicate severe insomnia, sometimes with less than an hour or rarely even less than half an hour of total recorded sleep time. Sleep is defined in most laboratories as a 30-second epoch containing a majority of sleep-defining electroencephalographic (EEG) patterns (generally a low-voltage mixed-frequency EEG for stages 1 and 2 sleep and a relative absence [less than 50%] of alpha and/or beta activity).1 This is fine for scoring sleep according to the near universally accepted criteria developed by Rechtschaffen and Kales. It also gives a relatively manageable number of epochs (960 epochs for an 8-hour recording). However, if carried to a simplistic extreme, this process may interfere with good clinical judgment in terms of recognizing severe sleep-related breathing disorders in certain patients.

When sleep-related breathing events (either hypopneas or apneas) occur with such frequency during the night and/or the sleep EEG patterns are so short (less than 15 seconds in duration), then a problem develops in terms of what to do with those apneas or hypopneas that occur during epochs of wakefulness but within which there is obvious evidence of impending sleep. This hypnagogic phenomenon is sometimes referred to as microsleep, in which less than 15 seconds of low-voltage mixed-frequency EEG intrudes into a majority pattern of alpha and/or beta activity defining a nominal epoch of wakefulness.

The EEG changes during the transition from wakefulness to sleep have been well documented.2,3 The question arises as to whether or not to score sleep apneas or hypopneas within epochs of wakefulness in which an obvious respiratory disturbance has occurred. It has previously been recognized that EEG frequency changes, specifically a slowing of the EEG frequencies from alpha to theta, occur during sleep apnea episodes.4 However, it is often observed that the beginning of a sleep hypopnea (usually scored as the peak of the last full-sized or normal breath) often occurs several seconds before noticeable EEG slowing occurs.

Apneas and Hypopneas
An apnea is loosely defined as a cessation of airflow (or an airflow correlate as measured with either a nasal/oral thermocouple or a thermistor) lasting 10 seconds or longer. One of the earlier published, quantitative definitions of an apnea was stated as “More precisely, obstructive apnea is defined by a fall in amplitude of the thermistor tracing (with or without an associated change in the amplitude of the strain gauge tracings) to less than 50% of the basal value amplitude as judged from the period two minutes before or after the respiratory irregularity in question.”5 A more recent definition of apnea has been defined as an 80%-100% reduction in airflow at the nose and mouth for 10 seconds or longer.

Hypopneas have been much less clearly defined, and many papers have addressed the lack of a consensus definition for this common sleep-related breathing event.6,7 The American Sleep Disorders Association (ASDA) is currently developing a position paper to operationally define apneas and hypopneas.8

One thing that is clear from a reading of an early draft of the ASDA position paper is that most experts in the field recognize apneas and hypopneas based on some sort of consequence of the respiratory disturbance. The polysomnographically defined effects most often required in order for an event to be considered clinically significant are either an EEG arousal (which disturbs sleep) or an oxygen desaturation (which causes temporary hypoxia).

It is my experience that a patient is seen occasionally (probably less than one out of 100) who has severe insomnia and little polysomnographic evidence of sleep, both of which are secondary to extremely severe sleep apnea/hypopnea. Like most sleep apnea patients, they present in the clinic with one or more of the following signs or symptoms of obstructive sleep apnea (OSA): loud snoring, observed apneas, morning headaches, excessive daytime sleepiness, high blood pressure, frequent heartburn or gastroesophageal reflux, memory problems, concentration difficulties, tiredness, or depression. When performing a polysomnographic sleep recording on these rare sleep apnea patients, it is observed that the patients have extreme sleep fragmentation, hundreds of awakenings or EEG arousals, and severe insomnia as judged by sleep-

scoring the 30-second epochs of the polysomnogram. However, closer examination of the EEG patterns in relationship to the location of the sleep-related breathing events (hypopneas or apneas) reveals that the sleep-disordered breathing occurs during episodes of microsleep, within the epochs of wakefulness. The fact that sleep apnea can occur in awake patients has been previously documented, especially if they are lying in the supine position.9

Problematic RDIs
Some sleep laboratories ignore such events, adopting the approach that if the event did not occur during an epoch of sleep, it is not considered a sleep apnea or sleep hypopnea. Other laboratories have adopted the approach of scoring such apneas or hypopneas during epochs of wakefulness containing periods of microsleep, but only if there are also pronounced EEG arousals or obvious oxygen desaturations following the events.

Doing so, however (on some occasions), leads to a problematic definition of the respiratory disturbance index (RDI). For instance, in one case seen at the Mountain Medical Sleep Disorders Center, Carson City, Nev, the patient slept only 12.5 minutes (as judged by Rechtschaffen and Kales [R & K] scoring criteria), yet had a total of 21 apneas, 73 hypopneas, and an RDI of 451.2 apneas or hypopneas per hour of sleep (94 apneas or hypopneas divided by 12.5 minutes times 60 minutes/hour). On average, this represents one apnea or hypopnea every 7.98 seconds. This is obviously impossible since apneas and hypopneas must have a duration of at least 10 seconds in order to be scored.

For such extreme cases, this may be solved by changing the denominator of the RDI from total sleep time (stages 1, 2, 3, 4, and rapid eye movement [REM] sleep combined) to sleep period time (the time interval from sleep onset to final awake onset). Adopting this proposed alternative definition of RDI for the above-mentioned patient changed the RDI from the nonsensical figure of 451.2 apneas or hypopneas per hour of sleep to 44.9 apneas or hypopneas per hour of sleep period time.

In another case, also from the Mountain Medical Sleep Disorders Center, a patient had a measured total sleep time (defined by R & K criteria) of only 60.5 minutes out of 430.5 minutes in bed, giving an extremely low sleep efficiency index of only 14%. Yet, in spite of such little R & K defined sleep, the patient had hundreds of episodes of microsleep accompanied by hundreds of respiratory disturbances. In this particular record, the periods of microsleep, hypopneas, EEG arousals, and oxygen desaturations began to occur at epoch number 58 long before (69.5 minutes before) even the first epoch of recorded sleep (which occurred on epoch number 197). In the case of this particular patient, the transition from wakefulness to drowsiness to sleep was repeatedly impeded by a sleep apnea or a sleep hypopnea, which was almost always followed by an EEG arousal long enough to cause the 30-second epoch to be scored as wakefulness (rather than sleep). During the entire sleep period time (first epoch of sleep to first epoch of the final awakening) for this particular study, the patient had a total of 282 apneas and 18 hypopneas, giving a respiratory disturbance index of 297.5 apneas or hypopneas per hour of sleep. On average, this represents one apnea or hypopnea every 12.1 seconds, which, although possible, is most certainly an overestimation of the severity of this patient’s sleep apnea/hypopnea condition. Even though this was an extremely high RDI (297.5), it did not include the dozens of hypopneas that occurred during repeated sleep onset attempts prior to the first 30-second epoch scored as sleep (which had to have at least 15 seconds of low-voltage, mixed-frequency EEG and less than 15 seconds of alpha or beta activity in it). Recalculating the RDI in terms of the sleep period time (which was 283 minutes) gave a more realistic, modified RDI estimate of apnea severity of 63.6 apneas or hypopneas per hour of sleep period time (rather than 297.5 events per hour of sleep).

In a third example studied at the Washoe Sleep Disorders Center, Reno, Nev, a patient had a total measured sleep time of only 29.5 minutes. This was out of a total time in bed (from lights out to lights on) of 461 minutes, giving an extraordinarily low sleep efficiency index of a mere 6.4%. However, during the entire sleep period time of 454 minutes, there was a total of 104 apneas and 439 hypopneas (including those events during episodes of microsleep intrusions within 30-second epochs of wakefulness). This gave an RDI of 1,104 apneas or hypopneas per hour of sleep and the unrealistic figure of one apnea or hypopnea every 3.2 seconds. Again, recomputing this in terms of sleep period time gave a more realistic RDI of 71.7 apneas or hypopneas per hour of sleep period time. This, of course, is a more sensible estimate of the severity of this patient’s sleep problem (one event every 50.2 seconds rather than an apnea or hypopnea every 3.2 seconds).

Some sleep experts could argue that scoring apneas and hypopneas within periods of microsleep overestimates the severity of a patient’s sleep apnea condition. In order to argue against this position, consider the severity of each patient’s insomnia (as judged objectively by sleep stage scoring of the polysomnography) and how much the insomnia improved after continuous positive airway pressure (CPAP) treatment.

Consider what happened to each of these three individuals. When patient A was placed on CPAP therapy, his total sleep time increased from 12.5 minutes to 437.5 minutes, the number of apneas decreased from 21 to 0, and the number of hypopneas decreased from 73 to 12. The RDI, as conventionally defined, decreased from 451 to 1.65 apneas or hypopneas per hour of sleep. Computing this per hour of sleep period time, the RDI decreased from 44.9 to 1.29 apneas per hour of sleep period time.

The outcome of CPAP treatment for patient B was similar. When he was placed on CPAP therapy, there were again significant changes in his sleep parameters (comparing before CPAP to with CPAP). Total sleep time increased from 60.5 minutes to 253.5 minutes (and sleep efficiency increased from 14% to 52.5%). The number of apneas decreased from 282 to zero; however, the number of hypopneas (including those during wakefulness epochs) increased from 18 to 42. In spite of this, the overall RDI computed the standard total sleep time way decreased from 297.5 to 9.9 apneas or hypopneas per hour of sleep. The modified RDI (using sleep period time decreased from 63.6 to a mere 5.4 apneas or hypopneas per hour.

When patient C was placed on CPAP therapy, his total sleep time increased from 29.5 minutes to 370.5 minutes and sleep efficiency increased from 6.4% to 81.1%. The total number of apneas decreased from 104 to 13 and the number of hypopneas decreased from 439 to 161. The RDI decreased from 1,104 to 28.2 apneas or hypopneas per hour of sleep. The alternate way of computing the RDI showed a decrease from 71.7 to 22.8 apneas or hypopneas per hour of sleep period time. Obviously, there were substantial clinical improvements in all three of these patients.

One further argument could be made to not score those apneas or hypopneas that occur during periods of microsleep within 30-second epochs of wakefulness because the epoch was not a sleep epoch. This argument can be negated with two considerations. First, it is important to note that sleep did in fact occur in most of these 30-second epochs scored as wakefulness; it is just that the duration of the sleep EEG patterns (as defined by the EEG changes of slowing and an increase in amplitude) was not long enough to affect the nominal score of the 30-second epoch. If the sleep stage scoring epoch in question had a duration of 10 seconds (rather than 30 seconds), most of these periods of microsleep would then get a 10-second epoch score of stage 1 (or occasionally stage 2) sleep. In other words, the choice of epoch duration (30 seconds versus 10 seconds) determines how many and what proportion of epochs will be scored as sleep.

Second, it is equally important to observe the clear, unambiguous presence of either EEG arousals or oxygen desaturations (usually both) following these microsleep apneas and microsleep hypopneas. Clearly, if desaturations drop by more than 3 or 4 Sao2 percentage points (and sometimes drop dramatically, down into the 70s or below), this is very clinically significant and a real sleep-related breathing event.

The fact that almost all of these events also caused a substantial EEG arousal is also evidence that the events were clinically significant. In fact, the EEG arousals caused by these events are the very reason why many of these events occurred during epochs of wakefulness. This is the sleep fragmentation commonly seen at the end of a sleep apnea or hypopnea episode.10 Had the apnea or hypopnea not occurred and the arousal not occurred during the hyperpneic phase of respiration following the respiratory disturbance, the epoch in question usually would have been scored as a sleep epoch (generally stage 1 or sometimes stage 2) rather than a wakefulness epoch. Furthermore, many of these events showed evidence of upper airway obstruction, such as either paradoxical breathing (out of phase or phase shifted abdominal and thoracic movements as detected with a pair of piezo-electric strain gauge belts) or snoring (as detected with a microphone output to the polygraph). Other clues that these events were real were the frequent observations of “airflow signal flattening” or alterations in the shape of the airflow correlate signal. The lack of a corresponding increase in the airflow signal with increases in respiratory effort indicates partial upper airway obstruction. This can be seen from a variety of surrogate airflow signals such as a nasal/oral thermocouple or thermistor, the volume estimate signals from a CPAP system (showing flow limitation by means of signal flattening),11 or the recently commercialized nasal pressure sensing devices using a nasal cannula inserted in the external nares.12,13

Some or all of these detectable polysomnographic events (EEG arousals, oxygen desaturations, paradoxical breathing, snoring, and/or airflow signal flattening) were always seen even in those apneas or hypopneas that occurred during periods of microsleep that happened to occur within 30-second epochs scored as wakefulness.

Serious consideration should be given to redefining the operational definition of a sleep apnea or sleep hypopnea. More attention should be given to events that occur in the hypnagogic time intervals (the periods of time in which a sustained alpha and/or beta EEG pattern begins to disintegrate and break up into either less intense alpha patterns or dissolve completely into a low-voltage mixed-frequency EEG pattern defining stage 1 sleep).

As clinically observant sleep specialists, we must not let a blind adherence to a 30-second scoring epoch that was developed for scoring sleep stages interfere with our clinical judgment for assessing sleep-related breathing disorders. Sleep apnea was not widely recognized at the time the R & K committee was deliberating, as the first two published observations of sleep apnea were published only in 1965.14,15 It is apparent that the members of the R & K sleep scoring committee did not deal with the sleep apnea/hypopnea issue and the resulting sleep fragmentation.10 Therefore, we must occasionally remind ourselves that we are looking at the process of sleep. Apneas or hypopneas that occur within the hypnagogic state (the impending behavioral and neurophysiological transition from wakefulness to sleep) should be as important to us as sleep disorder specialists as those events that occur during 30-second epochs nominally scored as a particular stage of sleep (stage 1, 2, 3, 4, or REM).

John T. Zimmerman, PhD, is laboratory director of Mountain Medical Sleep Disorders Center, Carson City, Nev, and adjunct professor of psychology at the University of Nevada, Reno.

1. Rechtschaffen A, Kales A (ed), Berger RJ, et al. A Manual of Standardized Terminology, Techniques and Scoring System for Sleep Stages of Human Subjects. Washington, DC: Public Health Service, US Government Printing Office; 1968.
2. Tanaka H, Hayashi M, Hori T. Topographical characteristics and principal component structure of the hypnagogic EEG. Sleep. 1997;20:523-533.
3. Wright KP Jr, Badia P, Wauquier A. EEG analyses and transition from wake: topographical and temporal patterns of brain activity during the transition from wakefulness to sleep. Sleep. 1995;18:880-888.
4. Svanborg E, Guilleminault C. EEG frequency changes during sleep apneas. Sleep. 1996;19:248-253.
5. Keenan-Bornstein S. Respiratory monitoring during sleep: polysomnography. In: Guilleminault C, ed. Sleep and Waking Disorders: Indications and Techniques. [AQ: NEED CITY], Calif: Addison-Wesley Publishing Co; 1982:183-212.
6. Moser NJ, Phillips BA, Berry DT, Harbison L. What is hypopnea, anyway? Chest. 1994;105:426-428.
7. Redline S, Sanders M. Hypopnea, a floating metric: implications for prevalence, morbidity estimates, and case finding. Sleep. 1997;20:1209-1217.
8. Report of the American Sleep Disorders Task Force. Criteria for the measurement, definition and severity rating of sleep related breathing disorders in adults. 1998 draft report. [AQ: NEED MORE INFO].
9. Miura C, Hida W, Miki H, Kikuchi Y, Chonan T, Takishima T. Effects of posture on flow-volume curves during normocapnia and hypercapnia in patients with obstructive sleep apnea. Thorax. 1992;47:524-528.
10. Kimoff RJ. Sleep fragmentation in obstructive sleep apnea. Sleep. 1996;19:S61-S65.
11. Condos R, Norman RG, Krishnasamy I, Peduzzi N, Goldring RM, Rapoport DM. Flow limitation as a noninvasive assessment of residual upper-airway resistance during continuous positive airway pressure therapy of obstructive sleep apnea. Am J Respir Crit Care Med. 1994;150:475-480.
12. Berg S, Haight JSJ, Yap V, Hoffstein V, Cole P. Sleep-disordered breathing: comparison of direct and indirect measurements of respiratory airflow: implications for hypopneas. Sleep. 1997;20:60-64.
13. Norman RG, Ahmed MM, Walsleben JA, Rapoport DM. Detection of respiratory events during NPSG: nasal cannula/pressure sensor versus thermistor. Sleep. 1997;20:1175-1184.
14. Gastaut H, Tassinari C, Duron B. Etude polygraphique des manifestations episodiques (bypniques et respiratoires) du syndrome de Pickwick. Rev Neuro. 1965;112:568-579.
15. Jung R, Kuhlo W. Neurophysiological studies of abnormal night sleep and the pickwickian syndrome. Prog Brain Res. 1965;18:140-159.