SHHS Polysomnography Database

Return to the SHHS PSG Database

Quality control procedures for data acquisition and scoring are described in [3]. All studies were reviewed at the Reading Center and assigned quality codes, grading overall study quality and quality of each channel. Signal and study quality codes, specific to each technician, each monitor, and each site, were summarized and reported on a monthly basis. These data were reviewed on a monthly basis by members of a Polysomnography Subcommittee. Sites and individual technicians were expected to produce at least 85% of studies with a grade of "Good" or better. Those who did not reach this standard were identified. Any downward trending of quality or deviation of specific technicians required a written response from the Principle Investigator from that respective site. These data were reviewed periodically with members of the Steering Committee, the NIH program office, and a Data and Safety Monitoring Board.

Scorer Training and Certification: Each scorer underwent extensive training with the other SHHS scorers, a Chief Polysomnologist and Reading Center Investigators. Scorer certification required the demonstration of a complete understanding of scoring rules and achievement of a 90% level of agreement with the Chief Polysomnologist for respiratory events and sleep stages and a 85% level of agreement for arousals for 10 or more independently scored practice records. Reliability and accuracy of scored data were evaluated on an ongoing basis, including formal reliability studies [4] and ongoing scoring exercises.

Record Grading: Each record is assigned an overall grade as follows:

• Outstanding: All channels are graded as mostly artifact free for > 5 hours and entire sleep time.
• Excellent: At least one EEG channel, one EOG channel, EMG, oximetry, all respiratory channels usable for > 5 hours and > 75% of sleep time.
• Very good: At least one EEG channel, oximetry, airflow and either chest or abdomen usable for > 5 hours and > 50% of sleep time.
• Good: At least one respiratory channel (airflow or either band), oximetry and one EEG usable for > 5 hours and > 50% of sleep time.
• Fair: At least one respiratory channel, oximetry and one EEG usable for > 4 hours or study scored sleep-wake only (because of the EEG artifact).
• Poor (Failed): < 4 hours of usable data on at least one respiratory channel, oximetry and one EEG.

(SHHS-1 records did not use the percentage of sleep time during which signals were artifact free, and only relied on the absolute duration of artifact free data.)

Manual Scoring: An Atlas detailing scoring, with pictorial examples, is found in the SHHS Reading Center Manual of Procedures.

Sleep stages were identified for each 30 second epoch using Rechtshafflen and Kales criteria [5].

Arousals were characterized by the American Academy of Sleep Medicine criteria [6]. An EEG arousal is an abrupt shift in EEG frequency, which may include alpha and/or theta waves and/or frequencies greater than 16 Hz, but not sleep spindles, lasting at least 3 s., and starting after at least 10 continuous seconds of sleep. In stage REM, an EEG frequency shift must be accompanied by a simultaneous increase in amplitude of the chin EMG (lasting over 0.5 s.). An arousal starts when a definite change in background EEG is visualized. The increase in the chin EMG can occur anytime during the arousal (can be at the end) and is not a marker for the beginning of the arousal. However, increased EMG activity without a change in background EEG does not constitute an arousal.

Apneas were identified if the amplitude (peak to trough) of the airflow signal was flat or nearly flat. This was noted when amplitude decreases below at least 25% of the amplitude of "baseline" breathing (identified during a period of regular breathing with stable oxygen levels), if this change lasted for > 10 s. Apneas cannot be designated in areas of the study where thermistry is missing or uninterpretable

Hypopneas were identified if the amplitude of any respiratory signal decreases below (approximately) 70% of the amplitude of "baseline" (identified during a period of regular breathing with stable oxygen levels), if this change lasted for > 10 s and for >2 breaths.

"Central" events were noted if displacement was observed on neither the chest nor the abdominal inductance channels. Otherwise, events were noted as "obstructive". Central events cannot be designated if either or both band data are missing or uninterpretable.

Summary data available for each record: Computer analysis of the records also linked data from varying channels to identify desaturation variables, sleep summary data, and various Respiratory Disturbance Indices (RDIs). The following describe how these data are used (See Data Dictionary for variable names associated with key summary parameters):

The desaturation associated with any respiratory event were based on the nadir desaturation reached within a user defined amount of the time (usually within 30 sec) of the end of the event. The magnitude of the desaturation for an event is the difference between the greatest saturation level observed during the event and this minimum. The scorer will manually check events to assure that the appropriate desaturation is identified and modify the time lag as needed.

Arousals were associated with events if they begin within 3 s. of the termination of the respiratory event.

RDI (Respiratory Disturbance Index) is defined as the number of respiratory events (apneas and hypopneas) per hour of the sleep. Events will be included in different indices according to level of associated desaturation and/or arousal: ? Summary RDI values, based on events associated with 0-2%, > 2%, >3%, >4%, and >5% desaturation levels. ? Summary RDI values based on events associated with 0-2%, > 2%, >3%, >4%, and >5% desaturation levels and associated arousal. ? Summary RDI values based on events associated with arousal regardless of desaturation.

Time in apnea or hypopnea:

• Percent of the sleep time in apnea (obstructive or central, with at least a 3% desaturation +/- arousal).
• Percent sleep time in hypopnea (with at least a 3% desaturation +/- arousal).

Oxygen desaturation Profile:

• Percent of the sleep time in desaturation (<95%, <90%, <85%, <80%, <75%).
• Number of desaturations/hour of sleep (unlinked with respiratory events) of 2%, 3%, 4%, 5%.

Sleep Architecture:

• Time and percent of sleep time in each sleep stage.
• Arousal index.
• Number/hour stage 1 shifts; upward shifts; wake shifts.
• Sleep efficiency.
• Sleep latency; REM latency (defined both as time from lights off to first REM and from onset of sleep to first REM).

Heart Rate data:

• Maximum, minimum and mean heart rate noted during sleep, associated with respiratory events and associated with the arousals (in REM and Non-REM separately).

Additional Data Available for each record (in QS form): Special events, generally recognized by pattern recognition, are also coded for each study. These include:

• Abnormal Awake EEG
• Alpha intrusion
• Abnormal Eye Movements
• Periodicity
• Cheynes Stokes Breathing (CSB)
• Periodic large breaths
• Sleep staging/arousal unreliable
• Apneas vs hypopneas unreliable

Return to the SHHS PSG Database