| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
Guest Access | Sign In via User Name/Password
|
|
|
|||||||
Guest Access | Sign In via User Name/Password |
|||||||||
* From the Sleep Disorders Unit, Repatriation General Hospital, Daw Park, South Australia, Australia.
 |
Abstract |
|---|
 TOP Abstract IntroductionMaterials and MethodsResultsDiscussionSummary and ConclusionsReferences |
|---|
Design and patients: Simultaneous, full polysomnographic recordings from the portable device (PSGP) and the laboratory-based system (PSGL) were obtained using separate sets of sensors on 20 patients referred for investigation of sleep apnea.
Setting: After initial optimization of signals, the portable device was left unattended in 10 of the patients (to simulate home studies), while in the other 10 the signals were reviewed on a laptop computer screen and adjustments to electrode or sensor placement made as needed during the studies. Recordings were manually scored by a technologist blinded to the origin of the data.
Measurements and results: The quality of signals was comparable between the PSGP and PSGL studies, apart from a slight decrease in respiratory signal quality during PSGP studies that led to reduced confidence in respiratory event scoring. SaO2 signal loss was also greater in unattended PSGP. There was good agreement between PSGP and PSGL for sleep variables and the apnea-hypopnea index (r = 0.99). The periodic limb movement index was slightly lower during unattended PSGP. Blinded physician assessment of the records led to a recommendation for repeat studies due to poor signal quality in one (10%) attended and one (10%) unattended portable recording. There was no significant discordance between PSGP and PSGL in the final diagnostic formulations.
Conclusion: Portable polysomnography is a viable alternative to laboratory-based polysomnography and may be improved further by better sensor attachment.
Key Words: home monitoring • portable polysomnography • sleep apnea
|
Introduction |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResultsDiscussionSummary and ConclusionsReferences |
|---|
treatment in sleep disorders. The high prevalence of sleep disorders and the increasing recognition of their importance in clinical practice has led to a remarkable increase in the demand for polysomnography services over recent years. PSGL studies are expensive, and in many instances demand for these services has outstripped supply. In response to this demand, a variety of portable devices have been developed that allow sleep studies to be performed in the home.1 ,2 ,3 ,4 ,5 ,6 Portable devices may be particularly valuable for epidemiologic studies and for subgroups of patients who have difficulty accessing main laboratory facilities, either because they live in remote communities or because of severe disability. The role of portable polysomnography (PSGP) in the evaluation of sleep apnea remains controversial and has been the subject of recent editorial comment7 and American Sleep Disorders Association (ASDA) statements.8 ,9
Portable EEG recording devices have been used successfully for many years by neurologists to investigate epilepsy.10 ,11 However, most of the ambulatory or screening polysomnographic devices used until recently recorded only a limited number of respiratory variables on their digital storage devices, and many of them were limited in recording of electrophysiologic variables that are necessary for the accurate staging of sleep (ie, EEG, electro-oculograms [EOGs], and submental electromyogram [EMG]). Sleep quantity and quality either were not measured or were assessed indirectly using body movement1 or a combination of body and eye movements.6 The latter approach appears to provide a reasonable estimate of total sleep time (TST), but it discriminates between sleep stages poorly.6 Actigraphy appears to provide reliable estimates of TST in normal subjects and patients with insomnia, but its reliability has not been systematically investigated in sleep apnea populations.9 A problem common to all such screening devices is that subtle alterations in sleep architecture (eg, brief arousals) that are important in disorders such as upper airways resistance syndrome12 cannot be discriminated.
It would be desirable if all physiologic variables could be recorded in the home with PSGP devices. Such a device that employs a magnetic-tape storage system has been available for a number of years and was used in a previous study by this group to record data from patients in the home setting.13. It provides accurate sleep stage data when the recordings are visually scored2 and a full range of respiratory recordings suitable for sleep apnea evaluation.9 However, the technology in this recorder has been superseded by less expensive high-density digital storage devices and further miniaturization of digital electronic components. A number of digital recording devices suitable for full PSGP have recently been released commercially. One of these devices (Compumedics PS1 Series portable system; Compumedics Pty Ltd; Melbourne, Australia) conforms to the ASDA standards-of-practice requirements for a Level II polysomnographic recorder. It allows full disclosure of electrophysiologic, respiratory, leg movement, and body position signals for manual and/or computer-assisted scoring. The present study was designed to compare this device with an established laboratory-based polysomnographic system (PSGL) in terms of (1) the technical quality of data recorded with respect to its suitability for sleep staging and respiratory event scoring, (2) derived indices such as sleep stages and apnea-hypopnea index (AHI), and (3) the final interpretive result. Half of the PSGP recordings were obtained in an unattended setting in the laboratory in an attempt to simulate the conditions of home monitoring and assess the quality of data thus recorded.
|
Materials and Methods |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResultsDiscussionSummary and ConclusionsReferences |
|---|
The procedures adopted for the attended and unattended portable PSG recordings (Att-PSGP and Unatt-PSGP) were as follows: during Att-PSGP, data were displayed in real time on a laptop computer regularly monitored by the attending sleep technologist. The technologist was instructed to attend the patient to rectify any problems with signal quality (eg, to reattach a sensor). During Unatt-PSGP, no laptop computer was used, and after initial optimization of signals, no technical intervention was permitted during the period of recording.
PSG Sensors and Montage
To keep PSGL and PSGP signals completely independent from one
another, two separate sets of sensors and electrodes (positioned
closely together) were used on each patient. The order and position of
sensor/electrode placement was randomized. Care was taken to ensure
that the PSGL and PSGP recordings began and ended at the same times.
The montage, electrodes, and transducers used for PSGL and PSGP recordings were identical, with the exception of the oximeters and the inductive plethysmography signal filtering/amplification systems (see below). Gold electrodes were used to record one channel of EEG (C3–A2), right and left EOGs, and submental EMG. Other parameters recorded included ECG, right and left leg movements (piezoelectric cell; Compumedics), arterial oxyhemoglobin saturation (SaO2 measured by finger probe pulse oximetry [For the Compumedics S-Series Sleep System: Criticare 504 oximeter; Criticare Systems Inc; Waukesha, WI. For the PS1-Series: Portable system oximeter; MSA; Pittsburgh, PA]). Body position (mercury switch; Compumedics), airflow (oronasal thermistor; Compumedics) and thoracic and abdominal efforts (inductive plethysmography bands; Vitalog Respironics; Redwood City, CA) were also recorded. The effort signals were amplified and filtered differently as follows: With the S-Series system, the input signal was supplied via a Vitalog interface box (Stand-Alone Medical Monitoring Interface, or SAMMI) that was also responsible for amplification and filtering of output signals. With the PS1-Series Portable, the Vitalog interface box was not used, and effort signal amplification and filtering were performed within the portable unit. Transducers were also mounted in the patient interface boxes of each system to monitor sound and to determine lights on/off.
Computerized Recording and Display Equipment
Hardware: The hardware used for data storage and
display were different for the PSGL and PSGP. PSGL data were recorded
on a 220-megabyte (MB) hard disk and displayed on a high-resolution
20-inch screen (resolution, 1,024 x 768 pixels; Multisync 6FG; NEC
Corp; Tokyo, Japan). PSGP data were recorded on a 20-MB PCMCIA card and
(during the Attended studies) displayed on a Toshiba T4700CS laptop
computer with a 9-inch VGA active matrix screen.
Software: Both PSGL and PSGP systems utilized specific recording software developed by Compumedics. The montages used were the same for each system and the majority of sampling rates were the same (EEG, 125 Hz; EOG, 50 Hz; EMG, 125 Hz; respiratory effort and flow, 25 Hz). The leg movement sampling rate was 25 Hz for PSGP and 50 Hz for PSGL. The SaO2 sampling rate was 1 Hz for PSGP and 5 Hz for PSGL. For PSGL, on-line gain and signal filtering adjustments were possible during the study and were performed as required; these adjustments were not possible during PSGP.
Data Analysis
Data recorded during PSGP were converted and downloaded onto the
Compumedics S-Series Sleep System to enable all studies to be analyzed
using the same analysis software (Compumedics Replay Version 4.0) and
high-resolution SVGA screen.
Manual analysis of all the studies was performed by a single trained scorer (IM) who was blinded to the origin of the data. Standard methods were used to score sleep stages.14 Sleep arousals were scored from the one EEG signal but otherwise conformed to ASDA criteria.15 Abnormal respiratory events were defined as follows: obstructive apnea = cessation of airflow for a minimum of 10 s with continued thoracoabdominal wall movement; central apnea = cessation of airflow and thoracoabdominal wall movements for a minimum of 10 s; mixed apnea = cessation of airflow and thoracoabdominal wall movements for at least the duration of one respiratory cycle, followed by a return of respiratory efforts but continued absence of airflow, the duration of the event being a minimum of 10 s; hypopnea = a minimum of 50% reduction in amplitude of airflow signal or both thoracic and abdominal efforts for at least 10 s. We did not include desaturation as a criterion for scoring apneas or hypopneas. Periodic limb movements (PLMs) were scored according to standard criteria.16
Outcome Measures
Signal Quality: Signal quality was assessed in two
ways. First, to determine the raw signal quality, the percentage of
study time that individual signals were either absent or
uninterpretable was calculated. Second, to determine the influence of
any such signal loss on the ability to score the PSGs, the technologist
noted the percentage of study time that sleep stage could not be
determined or abnormal respiratory events could not be detected or
properly characterized (eg, obstructive vs central apneas).
This second method of assessing signal quality was used because, in
practice, it is found that a skilled sleep technologist can often
continue to score the sleep record with confidence when one or more
signals is absent or severely distorted by artifact (eg, a
single EOG lead or one of the respiratory effort signals). We
considered that this additional method of data analysis would therefore
provide useful practical information regarding the reliability of the
PSGP device.
Derived Values: Derived values included sleep stages (total number of minutes spent in each individual sleep stage), TST (total number of minutes of sleep recorded after "lights out"), sleep efficiency [(time asleep/time in bed after "lights out") x 100%], AHI (number of apneas plus hypopneas per hour of sleep), arousal index (number of arousals per hour of sleep), and PLM index (number of PLMs per hour of sleep).
Clinical Interpretation: To assess whether signal quality affected the confidence or accuracy of the final interpretation of the polysomnographic recording, an experienced sleep physician who was blinded to the origin of the data assessed each study. The physician noted whether the technical quality of the data was excellent (no signal failure), good (signal failure occurred in the study but diagnostic confidence was not significantly impaired), poor (signal failure significantly reduced diagnostic confidence) or inadequate (serious signal failure necessitated repeat study). The physician also gave a final diagnosis, and sleep apnea was categorized as mild (AHI = 15–25), moderate (AHI = 26–40), or severe (AHI > 40). The moderate and severe categories could include patients from the adjacent less severe category of AHI if either severe desaturation (ie, nadir < 70%, or > 25% desaturations < 85%) or marked sleep fragmentation (arousal index > 15 per hour of sleep) was present in combination with daytime sleepiness (Epworth Sleepiness Score > 12).
Statistics
Since most parameters that were examined were not distributed
normally, comparisons between PSGL and PSGP data (Att and Unatt) were
made using the Wilcoxon signed rank test. Correlations were performed
using regression analysis. Bland-Altman plots17
were
constructed to assess the degree of agreement between PSGP and the gold
standard of PSGL. A p value of 0.05 or less was considered to be
statistically significant.
|
Results |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResultsDiscussionSummary and ConclusionsReferences |
|---|
Signal Quality
Individual Signal Failures: In Table 1
the percent of study time that individual signals (excluding
SaO2) were either absent or
uninterpretable due to poor quality, are given for Att-PSGP and for
Unatt-PSGP, and compared with data obtained during simultaneous PSGL.
In general, the percentage of time that individual signals were lost or
uninterpretable was very small (< 5%) and did not differ between the
portable and laboratory polysomnographies. The EMG signal during
Att-PSGP was unreliable due to an undetected fault in one of the two
pairs of EMG recording electrodes used with the portable device. This
set of electrodes was used more frequently in the attended than the
unattended portable studies, leading to a significant difference in EMG
quality in Att-PSGP vs PSGL. During Unatt-PSGP, there was a significant
increase in the percentage of study time that the airflow signal was
absent or uninterpretable compared to during PSGL.
|
|
|
Derived Values
Sleep:
TST, sleep efficiency, and the frequency of arousals did not
differ between portable and laboratory recordings (Fig 3
). Slightly more rapid eye movement (REM) sleep was scored in Att-PSGP
studies compared with PSGL (Fig 4
). This difference was attributable to poor EMG signal quality (faulty
electrode used in some of the portable studies) that affected the
accuracy of REM sleep scoring. The duration of non-REM sleep stages did
not differ between PSGP (attended or unattended) and PSGL recordings
(Fig 4 ).
|
|
|
|
|
Comparisons of the diagnostic formulations made from PSGP and PSGL recordings were possible in nine of 10 study pairs in both the attended and unattended components of the study (one attended PSGP and one unattended PSGP were considered of insufficient technical quality to enable confident interpretation). There was diagnostic concordance between Att-PSGP and PSGL in eight of nine patients, and between Unatt-PSGP and PSGL in eight of nine. The discordance in final diagnostic formulation was slight and of little clinical relevance: mild OSA vs probable upper airways resistance (Unatt-PSGP vs PSGL) and moderate OSA vs mild OSA (Att-PSGP vs PSGL).
|
Discussion |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResultsDiscussionSummary and ConclusionsReferences |
|---|
Technical Adequacy of PSGP vs PSGL
A feature of the present study was the generally high level
of reliability of electrophysiologic recordings using the portable
device, even in the unattended setting. The single exception to this
was the presence of poor EMG signals in some of the PSGP recordings due
to a faulty electrode that was undetected until late in the study. We
consider that the problem we encountered with the EMG electrode during
Att-PSGP recordings was serendipitous and does not signify any design
weakness of the PSGP device. It could equally have occurred during the
PSGL recordings. However, the laptop screen used during Att-PSGP
studies was not ideal for assessing the quality of the signals and
probably contributed to the late detection of this faulty component.
The problem with poor EMG signals during Att-PSGP studies did not
significantly affect the confidence of the technologist in scoring
sleep, but it did lead to slight inaccuracy in REM staging (see below).
It is generally acknowledged that an accurate description and quantification of sleep would add significantly to the diagnostic accuracy and utility of PSGP devices. Several devices have made attempts to quantify sleep using simplified sleep recording techniques (eg, body movement supplemented by eye movement)1 ,6 or software to perform on-line sleep analysis from standard electrophysiologic signals.5 It appears that the reluctance to employ full electrophysiologic monitoring in portable devices to this time has been due, at least in part, to concern than there would be a high signal failure rate. Our study suggests that this fear may not be justified. The mean time that signals were considered inadequate for sleep scoring using the portable device was 4% or less. This finding concurs with the experience of neurologists who have successfully used portable EEG for more than 20 years to monitor patients with epilepsy.10 The other impediment to collecting electrophysiologic data in portable sleep studies had been, until recently, the unavailability of miniature storage devices of sufficient storage capacity (at least 20 MB) for overnight, high-frequency (125 to 250 Hz), multichannel digital recordings. This problem has now been solved with the advent of newer portable storage devices such as that used in the Compumedics P-Series Portable Sleep System.
There was a small increase in the percentage of study time that respiratory effort and/or flow signals in PSGP recordings were poor and judged by the technologist to be suboptimal for respiratory event scoring. This was generally caused by falls in the amplitude of signals (eg, due to body position change) that could not be rectified by on-line gain changes. The software on the PSGP device does not allow on-line changes to signal gain or filtering. There were significant problems with the SaO2 sensor during portable studies. Based on the higher frequency of signal drop-outs during attended portable studies, it appears that the sensor used with the portable device may be more prone to signal failure or drop-out than the one used in the main laboratory system. This problem was compounded during unattended studies when signal absence was undetected in some studies for relatively long periods. These problems with respiratory signals are well recognized, particularly in unattended recording settings. The most common signal problems encountered by White et al,6 who used a portable device in the home, were respiratory effort, SaO2, and airflow signals, in decreasing order of importance. In one study, 10% of home oximetry records were rejected because of signal failure;18 in another home study using a Level III device, data loss was 15%.1 It should be noted that the definition of sleep-disordered breathing events used in this study did not include desaturation. Had it done so, the respiratory event scoring would likely have been more problematic. Finally, it is acknowledged that the Unatt-PSGP recordings were performed in the laboratory and might underestimate the frequency of signal failures likely to occur in true home studies.
Derived Indices: PSGP vs PSGL
Apart from a slight overestimation of REM sleep in Att-PSGP
recordings that can be attributed to a faulty EMG electrode connection,
there was very good agreement between portable and laboratory-based
devices with respect to indices of sleep. It is important to note that
TST used for calculations of the frequency of disordered breathing
events and PLMs did not differ between PSGP and PSGL recordings. The
small increase in the rate of failed or inadequate respiratory effort
or flow signals during portable studies did not lead to any significant
disagreement in AHI between the two methods. It is possible that, had
we incorporated a reduction in SaO2
into the definition of hypopnea, as have some other investigators, we
would have observed a discrepancy in AHI between Unatt-PSGP and PSGL
because of the higher SaO2 signal
failure time in Unatt-PSGP studies.
The PLM index was significantly lower in Unatt-PSGP compared with PSGL studies. We do not have any satisfactory explanation for this finding. It did not appear to be due to errors in assessing TST or to displacement of leg movement sensors during Unatt-PSGP studies. All patients found during PSGL studies to have moderate to severe PLMs (PLM index > 20/h) were correctly diagnosed by PSGP, and there were no false-positive diagnoses of moderate to severe PLMs from PSGP recordings.
|
Summary and Conclusions |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResultsDiscussionSummary and ConclusionsReferences |
|---|
|
Footnotes |
|---|
Correspondence to: R. Douglas McEvoy, MBBS, MD, Sleep Disorders Unit, Repatriation General Hospital, Daw Park, SA 5041, Australia; e-mail: smcevrd@rgh.sa.gov.au
Abbreviations: AHI = apnea-hypopnea index; ASDA = American Sleep Disorders Association; Att = attended; EMG = electromyogram; EOG = electro-oculogram; MB = megabyte; NS = not significant; OSA = obstructive sleep apnea; PLM = periodic limb movement; PSGL = laboratory-based polysomnograph; PSGP = portable polysomnography; REM = rapid eye movement; SaO2 = arterial oxygen saturation; TST = total sleep time; Unatt = unattended
Received for publication May 4, 1998. Accepted for publication July 9, 1999.
|
References |
|---|
|
TOPAbstractIntroductionMaterials and MethodsResultsDiscussionSummary and ConclusionsReferences |
|---|
This article has been cited by other articles:
|
|
 |
|
  W. W. Flemons, M. R. Littner, J. A. Rowley, P. Gay, W. M. Anderson, D. W. Hudgel, R. D. McEvoy, and D. I. Loube Home Diagnosis of Sleep Apnea: A Systematic Review of the Literature: An Evidence Review Cosponsored by the American Academy of Sleep Medicine, the American College of Chest Physicians, and the American Thoracic Society Chest, October 1, 2003; 124(4): 1543 - 1579. [Full Text] [PDF]
|
|
|
|
 |
|
 M. Arzt, M. Harth, A. Luchner, F. Muders, S. R. Holmer, F. C. Blumberg, G. A.J. Riegger, and M. Pfeifer Enhanced Ventilatory Response to Exercise in Patients With Chronic Heart Failure and Central Sleep Apnea Circulation, April 22, 2003; 107(15): 1998 - 2003. [Abstract] [Full Text] [PDF]
|
|
|
|
|
|
F. Gagnadoux, N. Pelletier-Fleury, C. Philippe, D. Rakotonanahary, and B. Fleury Home Unattended vs Hospital Telemonitored Polysomnography in Suspected Obstructive Sleep Apnea Syndrome : A Randomized Crossover Trial Chest, March 1, 2002; 121(3): 753 - 758. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 F. PORTIER, A. PORTMANN, P. CZERNICHOW, L. VASCAUT, E. DEVIN, D. BENHAMOU, A. CUVELIER, and J. F. MUIR Evaluation of Home versus Laboratory Polysomnography in the Diagnosis of Sleep Apnea Syndrome Am. J. Respir. Crit. Care Med., September 1, 2000; 162(3): 814 - 818. [Abstract] [Full Text]
|
|
|
|
 |
|
 E A. Balas and I. Iakovidis Distance technologies for patient monitoring BMJ, November 13, 1999; 319(7220): 1309 - 1309. [Full Text]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
