In order to interpret polysomnograms accurately, clinicians must be aware of recording conditions such as the transducers used, filter and gain settings, and display parameters

Many sleep disorder specialists think that interpretation of polysomnograms begins when the technologist calls up a scored data file on the reading computer. In fact, what the specialist sees on the computer screen, and therefore the diagnosis and treatment plan for each patient, is the end result of a series of decisions made either consciously or by default. The character of the polysomnogram begins with the choice of physiological sensors, recording equipment, filter settings, and display montages. The size of the monitor and layout of the tracings also strongly influence the interpretation.

A comprehensive polysomnogram must include analysis of sleep stages, sleep-related breathing disorders, limb movements, and sleep fragmentation. During my tenure as chair of Part 2 of the American Board of Sleep Medicine (ABSM) examination, I attempted to reassure candidates just prior to the start of the examination by telling them that the procedure was similar to that which should occur daily in their offices—reviewing histories, developing a differential diagnosis, and evaluating polysomnograms. One candidate wrote me after failing the examination to explain that the test was not at all like a day in his office because he “saw only sleep apnea patients.” I countered that he saw only sleep apnea patients because he looked only for sleep apnea and that if he raised his gaze from the airflow and oxygen saturation channels, he would have found periodic limb movements at the least and possibly some less common sleep disorders.

In many cases, sleep centers have purchased turnkey systems that include recording equipment, electrodes, oximeters, and body position sensors, as well as automated scoring systems. The trade-off for this relative ease of operation is a lack of flexibility; the system may have fixed filter, gain, and display settings that overprocess signals to make a poor quality signal seem adequate.

The American Academy of Sleep Medicine (AASM) provides frequently updated Standards of Practice1 and, for centers willing to undergo a detailed application process and site visit, Standards for Accreditation.2 These documents provide evidenced-based and consensus-based recommendations for polysomnography. One area of consensus is the need for continuous oxygen saturation and cardiac rhythm monitoring during polysomnography. This is critical for deciding whether emergency intervention is necessary and provides a ready indicator of the severity of sleep apnea. There is also good agreement that the polysomnogram report must include measures of sleep stage pattern, breathing during sleep, limb movements, and sleep continuity.

Sleep Stage Scoring
Many automated polysomnography units fail to provide electroencephalogram (EEG), eye movement, and muscle activity monitoring necessary for sleep stage scoring. This is especially true of some portable or home study units. These systems may provide methods for determining the severity of sleep-related breathing disorders. They permit counts of the number of episodes of apnea, degree of oxygen desaturation, or cardiac rhythm abnormalities. These systems fail to address one of the most important effects of sleep-related breathing disorders: disruption of sleep. When evaluating the success of treatment, it may not be sufficient to eliminate episodes of apnea. A comprehensive polysomnogram determines the effects of treatment on sleep stages and sleep continuity as well as measures of airflow and oxygen saturation. Sleep stage disruption may be sufficient in itself to warrant therapeutic intervention. In addition, sleep stage percentages allow a comparison across sleep centers based on an accepted standard.

Sleep stage scoring is critical for diagnosing narcolepsy. The clinician should be aware of latency to rapid eye movement (REM) sleep on the nocturnal polysomnogram as well as the Multiple Sleep Latency Test (MSLT). Although narcoleptics may have normal latency to REM sleep at night, abnormally short REM latency on a polysomnogram must be explained. This finding may be normal on the first night of continuous positive airway pressure (CPAP) titration, and infrequently occurs with severe sleep apnea. In my own practice, I became suspicious of narcolepsy in a patient based on abnormally short REM latency on the polysomnogram. When re-interviewed, he described a competition among friends to tell him the most outrageous jokes. The friends called this “Flooring John” because when the joke was funny enough, John fell to the floor with cataplexy. A subsequent MSLT confirmed the diagnosis of narcolepsy. In addition, REM latency between 30 and 60 minutes should alert the clinician to the possibility of depression.

R & K Scoring
The accepted standard for sleep stage scoring is the manual of Rechtschaffen and Kales.3 This manual was originally intended for sleep stage scoring of normal, healthy young adult volunteers for research studies, but has become the gold standard for evaluating all adult polysomnograms. For more than 3 decades, “R & K” scoring has been used for the evaluation of the efficacy of hypnotics, sleep onset latency in insomniacs, sleep deprivation experiments, and the development of sleep-related breathing disorder indices. Using the manual requires recording a single central EEG, eye movement from electrodes placed on the outer canthi, and submental electromyography (EMG). Typically, these are recorded using standard gold cup electrodes attached to the scalp or skin with collodion, paste, or tape. The manual requires that a score be given for each 30-second epoch of recording. In the predigital era, this consisted of a single page of printout, usually 12-inch square, with 12 to 21 channels of pen-and-ink tracings. Most digital systems can provide a reasonable approximation of this standard. Rechtschaffen4 viewed sleep stages as systematic changes of a constellation of variables and favored recording of additional information as an aid in situations where the three primary measures yielded ambiguous results. These extra measures may include occipital EEGs, vertical eye movements, regularity of respiration and cardiac rhythm, and additional EMG channels.

The Standards for Accreditation of the AASM require that any automated scoring system be validated against R & K scoring, and reviewed in detail whenever used. In addition, a physician and a diplomate of the ABSM must review each record in sufficient detail to ensure the accuracy of the interpretation.2 In most centers this consists of a “fast forward” scan through the raw data. Relying on hypnograms, tachygraphs, and technician scores is not sufficient. Another anecdote from my tenure on the ABSM examination committee: A group of 20 sleep experts meet before the examination to develop the key for scoring. I cannot remember a single epoch used as a test question for which all 20 experts agreed on the sleep stage. Some epochs yielded as many as four different sleep stages. Sleep scoring is an art that requires experience. The AASM requires that the ABSM diplomate, usually the most experienced member of the sleep center team, review the raw data.

Breathing During Sleep
Technological advances have improved the ability of polysomnographers to measure breathing during sleep. Ear oximeters that were state of the art in 1970 seem like lumbering dinosaurs compared to current finger models. Piezo-crystal belts provide accurate measures of respiratory effort and miniature microphones can be taped to the neck to measure snoring.

Nasal pressure transducers provide a significantly more sensitive measure of airflow than temperature-based transducers; many believe that the pressure transducers can provide a measure of upper airway resistance. This is important in cases where periodic arousals and leg movements occur without changes in temperature-based airflow. A flattening or decrease in amplitude of the pressure transducer signal may tip the balance in favor of a diagnosis of upper airway resistance syndrome and lead to titration of nasal CPAP. The new transducers provide additional information for scoring hypopneas. Most centers require a decrease of airflow, oxygen desaturation of 2% or 4%, and an arousal to score hypopnea. Pressure transducers are more sensitive to hypopnea and the signal usually flattens completely.

If used for evaluation of sleep-related breathing disorders, the new level of sensitivity may lead to scoring of many more events than are typically scored with the older technology. These events may be as significant as conventionally scored apneas, but at present virtually all of the clinical literature is based on temperature-based airflow transduction. An apnea-hypopnea index greater than five has been correlated with clinical consequences; a similar number of upper airway resistance events may or may not have similar consequences.

At the time of this writing, experts recommend using both measures of airflow simultaneously. Conventionally scored apneas and hypopneas should be counted and used to calculate an index. Nasal pressure events may tip the balance toward more events, but events based on nasal pressure changes alone should be reported separately.

Limb movements
The standard measure of limb movements is surface EMG recording of anterior tibilias muscles. Events typically last between 0.5 and 5 seconds and recur with a minimum interval of 5 seconds. These are counted and, like sleep-related breathing events, used to calculate an index of number of events per hour of sleep. Many centers count events with or without arousals separately.

Problems with electrode connectivity and painful hair removal in the morning may be eliminated by the use of piezo-electric movement detectors. When wrapped around the ankle, these devices are sensitive enough to respond to pulse artifact and can easily detect leg movement. It should be noted, however, that the EMG recordings might detect simultaneous activation of opponent muscle groups without overt movement. Movement detectors might miss these events, but the clinical significance of such events is not known.

Some patients may complain of periodic arm movements and sensors may be placed on the arms to quantify this activity. The alert technician visually monitors patient movements during the night, and may record additional information as indicated.

Sleep fragmentation
Scoring arousals during sleep is controversial and the ongoing debate is reflected in the ambiguity of guidelines prepared by the AASM5; the preliminary guidelines written in 1992 have not been updated and are not widely accepted. These guidelines define arousals as “an abrupt shift in EEG frequency” and rely more on a series of examples rather than a precise definition. Despite this, evidence suggests that brief arousals may be extremely significant clinically. Experimental sleep fragmentation studies have shown that daytime sleepiness is a function of “the frequency of the arousals… the type of arousal… and the age of the subject.”6 There is general agreement that the 30-second epoch fails to provide sufficient detail and additional measures are required.

Most polysomnogram analysis programs allow for classification of arousals as associated with apnea, limb movements, or “spontaneous.” In many cases, the spontaneous arousals may result from movement of unmonitored limbs or other muscles. If this assumption is correct, treatment of these arousals with medications used to treat periodic limb movements should be efficacious. Environmental noises may also result in arousals, reinforcing the need for quiet in the sleep disorders center.

Interpretation of polysomnograms begins long before the data are presented on the computer screen. The clinician must be aware of the conditions of the recording, including the transducers used, the filter and gain settings, and the display parameters. Advances in technology may increase the sensitivity of the polysomnogram, but the clinical significance of finer and finer events must be determined experimentally.

My advice is to begin with traditional measures and R&K scoring. The vast majority of the literature is based on temperature-based transduction of airflow and this should remain the focus for evaluating breathing during sleep. Based on preliminary reports, the addition of a nasal pressure transducer seems warranted, but not at the expense of the temperature-based evaluation. At minimum, the polysomnographic report should include sleep stages, number and type of apneas and hypopneas, counts of periodic limb movements, and a measure of sleep fragmentation.

There is no substitute for examining each second of the raw data. Interpretations based on hypnograms or processed data are clearly inadequate. Most computer systems can run through an entire polysomnogram in less than 10 minutes. An experienced polysomnographer can scan the recording to confirm the scoring and events within this time frame and this is time well spent.

Richard S. Rosenberg, PhD, is director of the Evanston Hospital Sleep Disorders Center, and is associate professor of neurology, Northwestern University Medical School, Evanston Northwestern Healthcare, Evanston, Ill.

1. Standards of Practice Committee, American Academy of Sleep Medicine. Clinical Practice Parameters. Available at: Accessed December 15, 2000.
2. Accreditation Committee, American Academy of Sleep Medicine. Standards for Accreditation of Sleep Disorders Centers. Available at: Accessed December 15, 2000.
3. Rechtschaffen A, Kales A. A Manual of Standardized Terminology, Techniques and Scoring System for Sleep Stages of Human Subjects. Los Angeles: Brain Information Service; 1968.
4. Rechtschaffen A. The psychophysiology of mental activity during sleep. In: McGuigan F, Schoonover R, eds. The Psychophysiology of Thinking. New York: Academic Press Inc; 1973:153-205.
5. Atlas Task Force, American Sleep Disorders Association. EEG arousals: scoring rules and examples. Sleep. 1992;15:174-184.
6. Bonnet MH. Sleep deprivation. In: Kryger MH, Roth T, Dement WC, eds. Principles and Practice of Sleep Medicine. 3rd ed. Philadelphia: WB Saunders; 2000:53-71.