SleepCheck - Clinical Studies

The SleepCheck has been rigorously clinically tested in six clinical trials. The results of these studies have been presented and published. The presentations and publications are summarized in Tables One and Two below; the details of each study are also described.

Table 1: Professional Presentations

Table Two: Professional Publications

STUDY 1
Sleep lab evaluation of SleepCheck:
This study describes the airflow apnea detection component of a Multi-Recorder Project funded by the National Institute of Neurological Disorders and Stroke (NINDS) through an SBIR Phase I program # N43-NS-5-2328. The study was approved by OHRP at NIH and a formal IRB Committee and the results were reported to NINDS, and presented at the 15th Annual Meeting of the Associated Professional Sleep Societies, 2000, Las Vegas, NV. The report of this study was published in the peer-reviewed journal Sleep with the following reference:

Gorny, S.W., Allen, R.P., & Krausman, D.T. (2000). Evaluation of an unattended monitoring system for automated detection of sleep apnea. Sleep, 23 (supplement 2), A369.

The evaluation of the airflow apnea detection method was based on comparison of the SleepCheck technology using an oral/nasal thermocouple sensor with a complete PSG analyses for sleep times, sleep efficiency, and sleep disordered breathing events (apneas).

Subjects:
Seven patients (5 male, mean 51.7 years old, mean BMI of 34.2 Kg/m2) with symptoms of sleep disordered breathing were tested. (See Table 3.) The sample included one patient who snored with no significant apnea, three patients with minimal apnea, three patients with mild apnea, and one patient with severe apnea.


TABLE 3: Characteristics of patients used for sleep lab and home monitoring.

Methods:
Each subject had a gold standard all night polysomnogram which involved recording for EEG from C3-A2 and C3-O1, submental EMG, left and right EOG, one channel of ECG (modified lead 2), airflow at the mouth and nose, respiratory effort from standard piezo-electric belts at the thoracic and abdominal levels and finger oximetry for oxygen saturation. This represents a standard assessment for sleep apnea. Air flow data were taken from the same oral/nasal thermocouple sensors, which were connected to both the polysomnogram and the SleepCheck air flow monitor. The SleepCheck monitor recorded the airflow signal and provided download to a PC for analysis. The patient also wore activity monitors on the chest to give indications of body-movement activity associated with respiration and on the wrist to grossly determine the sleep-wake state.

A Diplomate of the American Board of Sleep Disorders Medicine scored the data from each subject. A sleep disordered breathing event (apnea) was defined as any reduction by 50% or more in airflow lasting at least 10 seconds, or any reduced airflow lasting 10 or more seconds and associated with either an oximeter decrease greater than 3% or a movement arousal at the end of the event. The epoch-by-epoch identification of apnea events provided a count of all events during the sleep time and measurement of event durations from a stratified random sample. The same apnea event data was determined from the SleepCheck events based on computer scoring of the events only without any technician adjustment of the scores. The parameters for the computer scoring was based on the first two subjects and then applied to all of the patients giving a limited validation of the system. The computer scoring provided a rigorous test of the accuracy of the SleepCheck detection algorithm since it did not allow for any adjustment of the computerized data by the technician or sleep specialist.

Data were analyzed first for the point-by-point agreement for reproduction of the PSG tracing in the computer format. The amplitude and distance between peaks were recorded for both the PSG and the computer output from each patient for 5 consecutive peaks at random intervals for air-flow.

In addition, the data were analyzed for actual hit and miss rates for detection of apnea events and for the agreement on duration of apneas based on measurements of random samples. For the obstructive apneas the events were sampled from each hour of the recording for the apnea patient with the highest disordered breathing rate (DBR), one patient with moderate DBR, and the one with the lowest DBR. The apneas were matched to determine the hit rate and false alarm rate for the monitor’s apnea detection compared to the PSG.

To analyze agreement between the SleepCheck monitors and the PSG for measurement of apnea duration, two time periods were taken from one subject who had the most pronounced obstructive hypopneas of all the subjects. For the central apneas all event duration and detections were scored.

Results:
The point-to-point correlation between PSG and the output from SleepCheck for the wave forms showed an excellent agreement with an overall average correlation for peak-to-peak times of 0.97. (See Figure 1 below.)

There was no indication of any error in the data from the SleepCheck monitors. The SleepCheck monitors were able to capture, store and reproduce, for each of the parameters measured, essentially the same data as that recorded directly by the standard polysomnogram connected by wires to the patient.

The sleep disordered breathing rate (DBR/AHI) of apnea events determined by computer analysis of the SleepCheck monitor data also showed excellent correlations with those from the PSG. The total DBR rates (including central, mixed and obstructive events) were determined both for each full hour of sleep recording and for the total night’s recording. Correlation for DBRs from the PSG with that from monitors based on oral/nasal airflow was 0.94. (See Figure 2 below.)

Central apneas were rare in these patients, but when they occurred they were virtually always detected from the computerized analyses of the SleepCheck monitors. For the 19 central apneas detected by the computer from the monitor data the hit rate for the detection was 100% with a false alarm rate of 5.5 % when compared to the PSG data.

The measurement of the duration of the DBE for central apneas showed an excellent agreement between the computer measurements from the monitor and the visual measurements from the PSG. The average difference (PSG- Monitor durations) was 0.23 seconds with a standard error of 0.166.

Figure 1: SleepCheck compared to PSG for time between consecutive Airflow peaks.
The analyses of the DBR for the whole night for each patient showed similar results. There was an excellent agreement between the PSG and the SleepCheck data (r=0.97). However, SleepCheck’s airflow data showed a consistent bias for more DBR which becomes more pronounced for greater DBRs.

Analysis for actual hit and miss rates for detection of a SDB (apnea) for obstructive apneas was made on three selected subjects with high, moderate and low DBR. For the patient with highest DBR (severe sleep apnea) the hit rate = 91.7% with a false alarm rate of 4.9%, for the patient with the lowest DBR ( no significant sleep apnea) the hit rate = 81.3% with a false alarm rate of 0.0%, and for the patient with a mildly abnormal DBR (mild sleep apnea) the hit rate was 78.2% with a false alarm rate of 0.0%. This agreed with our impression that detection of events was easier and more accurate when events were pronounced. This result showed the advantages of using the range of apnea patients in this study, which emphasized the patients with milder disorder where detection appears to be the most difficult. The study was therefore focused on the important area of detecting the milder disorder, but included both severe and no disorder patients.

Figure 2: DBR (apnea) for each hour for PSG compared to airflow monitor.

Conclusions:
The data produced from the SleepCheck monitors not only produced essentially equivalent data for the appropriate measures as PSG, but the computer analyses from the monitors also provided an excellent agreement with the visually scored PSG. The data demonstrate the accuracy of capturing and recording apnea event data in self-contained monitors (SleepCheck) that maintained a separate data stream from the gold standard PSG. The data streams were successfully synchronized and were shown to produce reliable and equivalent apnea detection results for analyses of sleep-related breathing. This process worked well with the SleepCheck detection technology and clearly demonstrated the accuracy of single channel airflow apnea detection methods for use as an apnea-screening tool.

STUDY 2
Field test of SleepCheck:
This study is a continuation of Study 1 and describes the airflow apnea detection component of a Distributed Recorder in the at-home environment. This project was also funded by the National Institute of Neurological Disorders and Stroke (NINDS) through an SBIR Phase I program # N43-NS-5-2328. The study was approved by OHRP at NIH and a formal IRB Committee and the results were reported to NINDS.

Methods:
The same 7 patients evaluated in the lab study 1 above also wore the SleepCheck monitor at home during their normal sleep times. They were given the monitors in the morning after a nights sleep in the sleep lab and were given verbal instructions for self-attaching the unit. The patients applied the sensor and reset the monitor prior to going to bed and removed it the next morning after getting up for the day. They also kept a sleep-wake log and answered a questionnaire about their experience wearing the monitors at home. Patients on CPAP were asked to refrain from wearing their CPAP during this night at home. Data from the SleepCheck for SDB (apnea) measures were compared to the patient’s sleep lab data obtained above.

Results:
All seven subjects were successful in self-applying the monitors with only minimal verbal instructions. There were no significant problems with the data obtained from the at-home recordings. For about 2900 minutes of data, all but 120 minutes were captured by the monitors. This 96% success demonstrates ease of application and functional use for SleepCheck as a screening unit for sleep apnea.

Excellent quality recordings were obtained from all seven subjects. The DBR showed a good agreement (r=0.91) between the sleep-lab PSG and the at-home SleepCheck data. As shown in Figure 3 below, the at-home monitoring tended to show slightly more DBR at home than in the sleep lab. It should be noted that it was expected that there would be some variation in these data since the patients reported better sleep at home than in the sleep lab. The improved sleep status would be expected to alter the DBR rate depending upon the type of apnea. In particular the apneas associated with stage 1- wake transitions would be decreased with better sleep while the more severe obstructive apneas would be less changed. Even the more severe apneas might decrease somewhat if there were more slow wave sleep on night two.

Figure 3: Apneas per hour from night 'one' sleep lab PSG compared to night 'two' home recording from the SleepCheck monitors.

Conclusions:
The SleepCheck monitors were successfully worn for a night’s recording at-home with minimal data loss and excellent patient acceptance. The DBR (apnea) rates from the at-home monitoring were similar to that from the prior night in the sleep lab and showed no systematic bias compared to sleep lab data. The results indicated that not only could these monitors be easily used for recording at-home, but also that they were well- accepted by the patients and reliably recorded the data. The at-home monitoring with these self-contained, easy-to-use monitors provided substantially the same information regarding apnea breathing rates as the full gold standard sleep lab PSG. The final analysis of these studies indicate that the SleepCheck monitor concept is an valuable tool for initial assessment of sleep apnea and is ideally suited for screening and possibly for treatment evaluation.

STUDY 3
Sleep lab evaluation of SleepCheck system:
This sleep disorders study was a large multi-night sleep laboratory testing program for validating apnea detection and funded by the National Institute of Neurological Disorders and Stroke (NINDS) through an SBIR Phase 2 program # N43-NS-8-2328. The study was approved by OHRP at NIH and a formal IRB Committee and the results were reported to NINDS, and presented at the 16th Annual Meeting of the Associated Professional Sleep Societies, 2001, Chicago, IL. The report of this study was published in the peer-reviewed journal Sleep with the following reference:

Gorny, S.W., Spiro, J.R., Phillips, B., Allen, R.P. & Krausman, D.T. (2001). Initial findings from a multi-site evaluation of an unattended monitoring system for automatic detection of sleep disordered breathing events.

A total of 45 subjects with sleep disorders (56% female, 46% male; mean age = 43.84 years) were studied at two sites: The IM Systems sleep laboratory and the Sleep Disorders Center at Samaritan Hospital in Lexington, Kentucky.

Patients were brought into the sleep lab for a single night evaluation during which they slept from approximately 11:00 PM until 7:00 AM the following morning; this schedule was modified as required to match each patient's usual sleep schedule. Subjects were prepared for a full-night sleep evaluation during which they wore the SleepCheck recorders and were prepared for a full PSG evaluation, which included monitoring of sleep EEG (C3-A2, O1-A2), EMG - submental, bilateral and anterior tibialis, oral and nasal airflow, chest and abdominal respiratory effort, EOG (left and right), ECG and pulse oximetry. PSG data was independently scored by trained technologist under the supervision of a board certified sleep specialist for number of Sleep Disordered Breathing Events (SDBE's); the PSG data were then compared to the SleepCheck count.

Results:
The comparison between the SleepCheck and the PSG scoring of apnea events per night yielded an overall correlation of 0.97 (see Figure 5). The mean, absolute difference between the two measures was observed to be 6.79 events (standard deviation = 15.78 events) and was not found to be statistically significant (t(38) = 0.75, p > 0.05).

Conclusions:
The comparison between the SleepCheck and the PSG scoring of apnea events were highly correlated; the SleepCheck was able to yield highly similar apnea counts, and could be used as a screening processor for disordered breathing events.

Figure 5: Comparison of the SleepCheck data using airflow and the PSG data for detection of total number of apnea events (SDBE's) per night.

STUDY 4
At-Home evaluation of SleepCheck system:
This sleep disorders study was a large multi-night at-home testing program for validating apnea detection and funded by the National Institute of Neurological Disorders and Stroke (NINDS) through an SBIR Phase 2 program # N43-NS-8-2328. The study was approved by OHRP at NIH and a formal IRB Committee and the results were reported to NINDS, and presented at the16th Annual Meeting of the Associated Professional Sleep Societies, 2001, Chicago, IL. The report of this study was published in the peer-reviewed journal Sleep with the following reference:

Gorny, S.W., Spiro, J.R., Phillips, B., Allen, R.P. & Krausman, D.T. (2001). Initial findings from a multi-site evaluation of an unattended monitoring system for automatic detection of sleep disordered breathing events. Sleep, 24(supplement), A387.

Methods:
Following the sleep lab participation in Study 3, the 12 subjects at the Kentucky site and 21 of the 33 subjects at the IM Systems site wore the SleepCheck at home for two consecutive sleep periods. An additional 7 subjects were recruited at the IM Systems site for a total of 40 subjects (67.5% female, 32.5% male; mean age = 47.69 years).

For the 33 subjects who had participated in the sleep lab study, their data was compared to the data collected with the SleepCheck in the lab in order to determine the degree of agreement from night to night.

Results:
Correlations between the number of apnea events obtained during the sleep lab observation and the two nights of home recording for these subjects, using the SleepCheck data, demonstrated a high degree of night-to-night consistency as seen in Table 6 below.

Table 6: Correlations (r) between apnea rates obtained from the SleepCheck data in the one-night sleep lab observation and two nights of home recordings.

Conclusions:
The SleepCheck yielded similar results at home as in the lab; in addition, data loss was minimal and participant compliance was high. The results from this study indicate that the SleepCheck could be used as an at-home screening test; reliable results and high patient compliance could be expected.

STUDY 5
Sleep Lab evaluation of SleepCheck apnea detection methodology:
This sleep disorder apnea study was a Phase I study for validation of the oral/nasal airflow SleepCheck method and funded by the National Institute of Heart Lung and Blood (NIHLB) #1R43HL65166-01A1 through an SBIR grant award. The study was approved by OHRP at NIH and a formal IRB Committee and the results were reported to NIHLB, and presented at: the 16th Annual Meeting of the Northeastern Sleep Society, 2002, Baltimore, MD; the 24th Annual Meeting of the Southern Sleep Society, 2002, Bethesda, MD; and the 16th Annual meeting of the Association of Professional Sleep Societies, 2002, Seattle, WA. The report of this study was published in the peer-reviewed journal, Sleep:

Spiro, J.R., Gorny, S.W., Allen, R.A., & Krausman, D.T. (2002). Pilot evaluation of an ambulatory airflow pressure monitor for immediate identification of sleep disordered breathing events. Sleep, 25(supplement), A275.

Methods:
Six subjects were studied in the sleep research facility at IM Systems (mean age= 38.7 years, sd = 10.0 years). The sample had 3 males and 3 females, and was 33% African American and 67% Caucasian. For the six subjects, two were controls with no history of sleep disordered breathing; two had a history of snoring, but no diagnosis of sleep apnea; two had previously been diagnosed with obstructive sleep apnea, but had not been treated for the disorder within the year prior to this study.

All subjects came into the sleep lab for one night of observation. Normal bed and wake times were maintained for this study. Each subject was outfitted for a standard PSG evaluation: EEG (O1-A2, C3-A2), EOG (left and right referenced to a common mastoid), submental EMG, EKG (one channel), thoracic respiratory effort, and nasal/oral air pressure. Each subject also wore the SleepCheck system, which independently recorded nasal/oral airflow pressure. Subjects were instructed on how to put on the SleepCheck system and then placed the system on themselves for the nights recording. The SleepCheck system was also directly connected to one channel of the PSG.

Following the overnight evaluation, the monitor’s data was both displayed on the LCD readout and also downloaded to a PC. A trained technician scored the PSG recordings for apnea events and the scoring was checked by a Diplomate of the American Board of Sleep Medicine. This scoring was done from the PSG with the SleepCheck channel showing the airflow pressure removed from the display provided by the digital PSG system. Thus the PSGs were visually scored blind to any information from the SleepCheck system.

These tracings were essentially identical with no differences found in form or amount of deflection. The SleepCheck data downloaded to a PC was independently visually scored for apnea events by the technician blinded to the results from the PSG and following the same scoring procedures as for scoring the PSG except that the only data available were those from the SleepCheck. The apnea event rates per hour (DBR/AHI) form the SleepCheck were compared to those from the PSG. The correlations for apnea events (DBR/AHI) between the PSG and the SleepCheck were determined for the all night recordings for all subjects and for each of the first 3 hours for all subjects. The latter analyses provided repeated evaluations in a reasonably wide range of DBRs.

The real time SleepCheck detection as recorded on the full PSG was compared with the scored PSG for each 30 second epoch and for each apnea event. For this analyses the full set of channels on the digital PSG system was displayed on the screen. Once the SleepCheck and PSG visually scored events were recorded, an analyses of 50 consecutive breaths was made starting at a randomly selected time in each tracing. This permitted estimating the sensitivity and specificity of the apnea detection.

Results:
The SleepCheck pressure tracing on the PSG was compared to the PSG pressure tracing for changes in the waveform. The SleepCheck’s air pressure signal was excellent throughout all the recordings, showing fidelity to the PSG air pressure in signal amplitude contour and time. The correlation with deflection from a fixed line was 0.998 with no detectable differences. This was the expectation from the methods used but confirms that the data recorded in the SleepCheck matches that on the PSG within the accuracy of the measurement of the waveforms.

The visually scored PSG apnea rates varied from 0.0 to 74 per hour. The correlation of the visually scored DBR for the SleepCheck compared to the PSG was 0.999 with an average absolute error in SleepCheck compared to the PSG of 0.46 (sd = 0.62) and a range of errors from 0.0 to 1.45. (See table 7.) The SleepCheck data tended to be slightly less than the PSG but there was no other consistent pattern of bias related to PSG values. The correlation for the rates for each of the first 3 hours was 0.99 and the analyses of the difference between the SleepCheck and the PSG showed no overall significant bias related to the PSG DBR (r2 = 0.093, p>0.25).

The breath-by-breath analyses showed the visual scored SleepCheck compared to the PSG had 100% sensitivity and 88% specificity for detection of an apnea.

Table 7: DBR (Disordered breathing rate) or AHI (Apnea/Hypopnea index) for visual scoring of the PSG vs. visual scoring and automatic scoring of SleepCheck.

Conclusions:
The SleepCheck produced data that were identical to the PSG data in both signal form and function for detecting apnea events. The small differences between waveforms and also apnea represent the expected level of error inherent in the measurements. The results documented the basic data integrity for the SleepCheck monitor. The accuracy was in the range expected for this monitor. The amount of error is remarkably small and validates the use of SleepCheck’s single channel airflow pressure detection method to accurately record apnea/hypopnea events and AHI per hour.

STUDY 6
Field evaluation of SleepCheck real-time detection methodology:
This sleep disorder apnea study was the final component of the Phase I study for validation of the oral/nasal airflow SleepCheck detection and display method and funded by the National Institute of Heart Lung and Blood (NIHLB) #1R43HL65166-01A1 through an SBIR grant award. The study was approved by OHRP at NIH and a formal IRB Committee and the results are to be reported to NIHLB, and presented at: the 16th Annual Meeting of the Northeastern Sleep Society, 2002, Baltimore, MD; the 24th Annual Meeting of the Southern Sleep Society, 2002, Bethesda, MD; and the 16th Annual meeting of the Association of Professional Sleep Societies, 2002, Seattle, WA. The report of this study was published in the peer-reviewed journal, Sleep:

Spiro, J.R., Gorny, S.W., Allen, R.A., & Krausman, D.T. (2002). Pilot evaluation of an ambulatory airflow pressure monitor for immediate identification of sleep disordered breathing events. Sleep, 25(supplement), A275.

Methods:
The same six subjects as described in Study 5 above were given an SleepCheck monitor to wear at home for one night. The number of DBE counted by the SleepCheck per hour in the home were correlated with the DBEs per hour of sleep that were detected on the PSG in the lab in the above study.

Results:
The correlations between the DBR at home and in the lab for each patient were high. The number of apnea events per hour (DBR) for each participant, as determined from the SleepCheck worn at home and the PSG report from the in-lab study, were strongly correlated, (r =.99). See Table 8 for a breakdown by DBR per patient in each environment.

Table 8: Lab & Home DBR Comparison

Conclusion:
The correlations between the PSG and the SleepCheck results are strong, and sensitivity and specificity of the SleepCheck are high. No systematic bias was evident in either the sleep lab or at-home SleepCheck data. The results indicate that the SleepCheck may be a useful tool for the identification of disordered breathing in sleep.

Return to Top