PMCCPMCCPMCC

Search tips
Search criteria 

Advanced

 
Logo of nihpaAbout Author manuscriptsSubmit a manuscriptHHS Public Access; Author Manuscript; Accepted for publication in peer reviewed journal;
 
Sleep. Author manuscript; available in PMC 2013 November 5.
Published in final edited form as:
Sleep. 2004 August 1; 27(5): 923–933.
PMCID: PMC3818284
NIHMSID: NIHMS43697

Using a Wrist-Worn Device Based on Peripheral Arterial Tonometry to Diagnose Obstructive Sleep Apnea: In-Laboratory and Ambulatory Validation

Stephen D. Pittman, MSBME, RPSGT,1 Najib T. Ayas, MD, MPH,1,2,4,3 Mary M. MacDonald, RPSGT,1 Atul Malhotra, MD,1,2 Robert B. Fogel, MD,1,2 and David P. White, MD1,2

Abstract

Study Objectives

To assess the accuracy of a wrist-worn device (Watch_PAT 100) to diagnose obstructive sleep apnea in the home.

Design

Participants completed 2 overnight diagnostic studies with the test device: 1 night in the laboratory with concurrent polysomnography and 1 night in the home with only the Watch_PAT. The order of the laboratory and home study nights was random. The frequency of respiratory events on the PSG was quantified using indexes based on 2 definitions of hypopnea: the respiratory disturbance index (RDI) using American Academy of Sleep Medicine Task Force criteria for clinical research, also referred to as the Chicago criteria (RDI.C), and the Medicare guidelines (RDI.M). The Watch_PAT RDI (PAT RDI) and oxygen desaturation index (PAT ODI) were then evaluated against the polysomnography RDI.C and RDI.M, respectively, for both Watch_PAT diagnostic nights, yielding IN-LAB and HOME-LAB comparisons.

Setting

Sleep laboratory affiliated with a tertiary-care academic medical center.

Patients

30 patients referred with suspected OSA.

Interventions

N/A.

Measurements and Results

The polysomnography and PAT measures were compared using the mean [2 SD] of the differences and the intra-class correlation coefficient (ICC). The receiver-operator characteristic curve was used to assess optimum sensitivity and specificity and calculate likelihood ratios. For the IN-LAB comparison, there was high concordance between RDI.C and PAT RDI (ICC = 0.88, mean difference 2.5 [18.9] events per hour); RDI.M and PAT ODI (ICC = 0.95, mean difference 1.4 [12.9] events per hour; and sleep time (ICC = 0.70, mean difference 7.0 [93.1] minutes) between the test device and PSG. For the HOME-LAB comparison, there was good concordance between RDI.C and PAT RDI (ICC = 0.72, mean difference 1.4 [30.1] events per hour) and RDI.M and PAT ODI (ICC = 0.80, mean difference 1.6 [26.4] events per hour) for the test device and PSG. Home studies were performed with no technical failures.

Conclusions

In a population of patients suspected of having obstructive sleep apnea, the Watch_PAT can quantify an ODI that compares very well with Medicare criteria for defining respiratory events and an RDI that compares favorably with Chicago criteria for defining respiratory events. The device can be used with a low failure rate for single use in the lab and home for self-administered testing.

INTRODUCTION

OBSTRUCTIVE SLEEP APNEA (OSA) IS A COMMON DISORDER WITH IMPORTANT CLINICAL CONSEQUENCES FOR AFFECTED INDIVIDUALS. Prevalence of the disorder is estimated to be 2% to 4% among a middle-aged population.1 OSA is characterized by repetitive collapse of the pharyngeal airway during sleep, yielding hypoxia, hypercapnia, and arousal to reestablish airway patency.2 The associated consequences include daytime sleepiness,3 decreased cognitive performance, decreased quality of life,4 increased risk of automobile and industrial accidents,5,6 and adverse cardiovascular sequelae.710 Treatment of OSA leads to improvements in many of these adverse outcomes and may reduce healthcare costs.1113 Thus, diagnosis of this disorder is important.

A commonly used indicator of OSA severity is the respiratory disturbance index (RDI). The RDI represents the total number of apneas and hypopneas per hour of sleep. Most commonly, the RDI is derived primarily from overnight in-laboratory polysomnography (PSG) which includes the continuous recording of many physiologic variables including airflow, chest/abdominal movements, electroencephalography, electrooculography, electromyography, electrocardiography, and oxygen saturation.14 However, full PSG in the laboratory is expensive, cumbersome, and not readily available in many geographic areas due to a growing demand for the procedure.15 Although unattended and telemonitored PSG are available, the reliability of these alternatives varies.1619 Gagnadoux and coworkers19 recently reported a failure rate of 23% for 98 unattended home PSGs even though subjects reported to the laboratory for equipment set up and then returned home for the study. Several investigations of unattended PSG have reported that the majority of subjects preferred in-laboratory PSG over ambulatory PSG.17,20

Calculating RDI with a portable testing device based on a reduced channel set makes diagnosing OSA more feasible in an ambulatory setting such as the home. Ideally, the system is designed for self-administered home use. We and others have previously reported on 1 technology that may be useful in the ambulatory diagnosis of OSA, the peripheral arterial tonometer (PAT).2123 This technology uses a finger-mounted pneumo-optical sensor that eliminates venous pulsations and continuously measures the arterial pulse wave volume of the digit. Episodic vasoconstriction of digital vascular beds from sympathetic nervous system activation (mediated by α-adrenergic receptors) results in attenuation of the signal.21 Episodes of upper airway obstruction (eg, apneas, hypopneas) may cause arousal from sleep, sympathetic activation, and peripheral vasoconstriction and can therefore be detected on the PAT signal.24

In the present study, we assessed the accuracy of a wrist-worn device that combines this PAT technology with actigraphy and arterial oxygen saturation to diagnose OSA. Previous studies have assessed the accuracy of this device when compared to simultaneously obtained in-laboratory PSG and have demonstrated good results.22,23 This study utilized a similar approach but extended the work of Bar and coworkers by also studying patients in the home where the device is intended to be used.22

MATERIALS AND METHODS

Subjects

Adult patients referred to the clinical sleep laboratory of Brigham and Women’s Hospital with suspected OSA participated in this study. These participants were not consecutive patients but were a sample of patients who disclosed on a comprehensive questionnaire between June and December of 2002 that they were interested in being contacted about research studies conducted at the sleep laboratory. Exclusion criteria for the study were history of peripheral vascular disease, peripheral neuropathy, nonsinus cardiac rhythm, permanent pacemaker, severe lung disease, status-post bilateral cervical or thoracic sympathectomy, finger deformity that precluded adequate sensor application, and use of α-adrenergic receptor-blocking agents (24-hour washout period required). Informed consent was obtained from all participants after the protocol was approved by the Human Research Committee of Brigham and Women’s Hospital.

Protocol

All subjects completed a comprehensive sleep and health survey that included an Epworth Sleepiness Scale.25 Subjects underwent 2 separate evaluations of the home monitoring system (Watch_PAT 100, Itamar Medical Ltd., Caesarea, Israel): an in-laboratory comparison (IN-LAB) where subjects simultaneously wore the Watch_PAT during a full-night standard PSG and a home-laboratory (HOME-LAB) comparison during which Watch_PAT data acquired in the home were compared to PSG results obtained in the laboratory. The order of these 2 nights was random. The home and laboratory studies were scheduled to occur within 1 week of each other, thus avoiding the confounding influence of long time delays between studies.

Home Sleep Evaluation

For the home study, subjects reported to the sleep laboratory on the day of their study to receive instruction on the use of the Watch_PAT device. Instruction addressed proper application of the device and subject demonstration of correct use. This process required 5 to 10 minutes. The Watch_PAT device was then provided to the subject for transport to their home.

The Watch_PAT system has been described elsewhere22, 23 but consists of a battery-powered, wrist-mounted, recording device and software for postacquisition viewing and analysis of the recorded data. The wrist unit contains an actigraph (3-axis accelerometer for limb movement detection) to differentiate wake time from sleep time. It also includes 2 finger-mounted sensors: a PAT probe (Itamar Medical Ltd., Caesarea, Israel) and a pulse oximeter sensor (Nonin 8000J, Plymouth, Minn). The PAT probe applies a uniform pressure field over the distal two thirds of the finger, including the fingertip, which unloads arterial wall tension without causing distal venous pooling and distension, potential sources of venoarterial-mediated vasoconstriction. A transmission mode photoelectric plethysmograph is used to measure the optical density changes associated with pulsatile blood volume changes in the finger. The Watch_PAT device records 4 signals: PAT signal (arterial pulse wave volume), heart rate derived from the PAT signal, oxyhemoglobin saturation, and wrist activity (derived from the accelerometer). The device measures the oxyhemoglobin saturation at 1 sample per second from the internal pulse oximeter that uses 4-beat exponential averaging of the raw pulse-wave oxyhemoglobin-saturation measurements (8-beat exponential averaging was used for pulse rates between 112 and 225 beats per minute). The Watch_PAT device contains a rechargeable power supply, preliminary signal-conditioning hardware, 100-Hz data acquisition, and data storage on a removable COMPACTFLASH disk.

Watch_PAT studies were uploaded for automated analysis on a personal computer using the COMPACTFLASH reader provided with the PAT software (zzz_PAT version 2.0.39.13, Itamar Medical Ltd., Caesarea, Israel). The automated analysis used wrist activity to differentiate wake from sleep to determine a PAT valid sleep time (PAT VST) using a proprietary nonstandard algorithm. Respiratory events were detected during segments of PAT VST using a proprietary algorithm developed to match American Academy of Sleep Medicine guidelines for measurement in clinical research (also referred to as “Chicago criteria”).26 In particular, a respiratory event was automatically scored if 1 of 3 criteria were met: (1) PAT amplitude reduction occurred with an acceleration in the pulse rate or an increase in wrist activity, (2) PAT amplitude reduction occurred with a 3% or greater oxyhemoglobin desaturation, or (3) a 4% or greater oxyhemoglobin desaturation. The algorithm was developed using previous Watch_PAT data collected concurrently with PSG data to optimize event-by-event agreement. Oxyhemoglobin desaturations were quantified automatically in a similar fashion. No manual editing of the automated Watch_PAT scoring was performed. An RDI (PAT RDI) was then reported that represents the number of respiratory events per hour of PAT VST. In addition, an oxygen desaturation index (ODI) was reported that represented the number of oxyhemoglobin desaturations of at least 4% per hour of PAT VST.

IN-LABORATORY PSG

All subjects underwent a standard in-laboratory overnight PSG. Signals recorded included electroencephalogram (C4-A1, C3-A2, O2-A1 and O1-A2), electrooculogram, submental and bilateral tibial electromyogram, electrocardiogram, airflow (nasal-oral thermistor and nasal pressure [PTAF2, Pro-Tech Services, Woodinville, Wash]), chest and abdominal motion (piezo bands), oxyhemoglobin saturation (Model 930 Pulse-Oximeter, Respironics, Murrysville, Penn), body position, and snoring intensity. All physiologic data were collected and stored using the ALICE3 digital PSG system (Respironics, Murrysville, Penn). The oxyhemoglobin saturation was sampled once per second on the PSG, and the Model 930 Pulse Oximeter used the same 4-beat exponential averaging of the raw pulse wave oxyhemoglobin saturation measurements as the Watch_PAT device (8-beat exponential averaging was used for pulse rates between 112 and 225 beats per minute). The Watch_PAT signals described above were acquired concurrently with the PSG.

PSGs were all scored manually by 1 technologist (SDP) who was blinded to the Watch_PAT signals. A second technologist (MMM), who was also blinded to the Watch_PAT signals, manually scored the PSG records to help validate the manual PSG scoring of sleep and respiratory events. Sleep was staged according to standard criteria.27 Arousals were defined according to the American Sleep Disorders Association (ASDA) guidelines.28 Respiratory events were first scored according to the American Academy of Sleep Medicine guidelines for measurement in clinical research (Chicago criteria).26 In particular, an apnea was scored if airflow was absent for 10 seconds, and a hypopnea if airflow was reduced by 50% or a lesser extent (noticeable change) in association with a desaturation of at least 3% or an arousal. A respiratory disturbance index (RDI.C) was then calculated based on the number of apneas plus hypopneas per hour of sleep. Records were then rescored for respiratory events using an alternate definition of hypopnea (Medicare criteria).29 In this case, a hypopnea was defined as “an abnormal respiratory event lasting at least 10 seconds with at least a 30% reduction in thoracoabdominal movement or airflow as compared to baseline, and with at least a 4% oxygen desaturation.” A respiratory disturbance index (RDI.M) was then calculated based on the number of apneas plus desaturating hypopneas (per Medicare criteria) per hour of sleep.

Data Analysis

Sleep studies were considered acceptable for data analysis if none of the following rejection criteria occurred: (1) PSG total sleep time (TST) less than 1.5 hours, (2) Watch_PAT valid sleep time less than 1.5 hours, (3) poor-quality PSG recording (defined as a substantial portion of the PSG being not interpretable to score sleep and respiratory events), or (4) all Watch_PAT channels not available for review and automatic analysis. The PSG was considered the gold standard for identifying and quantifying OSA. The Watch_PAT was compared to respiratory events detected on the PSG using both Chicago and Medicare criteria. Detection of respiratory events using Chicago criteria was assessed by comparing the PAT RDI to the PSG RDI.C. Detection of respiratory events using Medicare criteria was assessed by comparing the PAT ODI to the PSG RDI.M. The manual PSG scoring of sleep stages and respiratory events (Medicare criteria) was validated by assessing agreement between the reference scorer and a second scorer. Epoch-by-epoch comparisons yielded the percentage of epochs with agreement, and Cohen κ statistic measured the chance-corrected level of agreement.30

The utility and accuracy of the Watch_PAT indexes in detecting OSA was based on summary data for both nights (laboratory and home) and was evaluated in a number of ways. These included evaluations for concordance using intraclass correlation (ICC: model 2, individual ratings),31 agreement using the method of Bland and Altman,32 and by constructing receiver operator characteristic (ROC) curves33 (Analyse-It Clinical Laboratory Software ver 1.67, Leeds, England) using RDI.M and RDI.C cutoffs of 15 events per hour on the PSG to differentiate normal cases from those with moderate to severe OSA.26,34 To assess the detection of varying degrees of OSA severity, we also constructed ROC curves using RDI.M and RDI.C cutoffs of 5, 10, 20, and 30 events per hour. Given that there is no well-defined RDI cutoff value for defining OSA, we also assessed diagnostic agreement using a clinical approach described previously to assess the diagnostic utility of another ambulatory device to detect OSA.35 According to this analysis, the Watch_PAT and PSG are considered in agreement if (1) both the RDI.C and PAT RDI were greater than 40 events per hour or (2) the RDI.C was less than 40 per hour on PSG and the PAT RDI was within 10 events per hour of the RDI.C.

All results are given as means ± 1 SD except for the means of the differences for the Bland-Altman analyses, which are given as means (± 2 SD for limits of agreement). Statistical significance was considered to be present when P < .05.

RESULTS

We recruited a total of 30 subjects with suspected OSA who met prospective eligibility requirements for participation in this study. Of these participants, 1 subject was excluded because of significant signal artifact on the study night in the sleep laboratory that prevented analysis of the PAT signal. Thus, data from 29 subjects (21 men, 8 women) are included in the analysis. The mean age of these subjects was 43.2 ± 10.8 years and mean body mass index was 33.9 ± 7.1 kg/m2. The mean Epworth Sleepiness Scale score was 9.2 ± 4.7 (range 2–18). The mean interval between HOME and IN-LAB studies was 1.7 ± 1.0 days. The IN-LAB was first in 17 (58.6%) subjects.

Validation of the PSG manual scoring of sleep stages yielded epoch-by-epoch agreement of 93.4% (discriminating wake, non-rapid eye movement sleep, and rapid eye movement sleep) with a Cohen κ statistic of 0.86. The agreement was 95.1%, and Cohen κ statistic was 0.83 for the RDI.M calculation between the 2 manual PSG scorers using comparisons of 0, 1, or 2 respiratory events per epoch.

In-Lab Comparison

Chicago Criteria

For the night in the laboratory, mean TST per PSG was 347.4 ± 64.6 minutes, while the mean LAB-PAT VST was 340.3 ± 55.5 minutes (see Table 1). There was good agreement between LAB-PAT VST and PSG TST (ICC = 0.70 mean difference 7.0 [93.1] minutes). Mean RDI.C per PSG was 31.6 ± 20.6 events per hour, while the mean LAB-PAT RDI was 34.2 ± 19.2 events per hour. The LAB-PAT RDI and PSG RDI.C were concordant (ICC = 0.88, Figure 1A). A Bland Altman plot of LAB-PAT RDI and PSG RDI.C is shown in Figure 1B; there was no obvious systematic difference between the 2 variables.

Figure 1
In-laboratory (Watch_PAT data collected concurrently with polysomnography data) comparisons are shown. (A) Scatter plot of PSG RDI.C vs LAB-PAT RDI with a best-fit line of Y = −0.97 + 0.95X. (B) Bland-Altman plot of PSG RDI.C vs LAB-PAT RDI. Agreement ...
Table 1
Data for Each Subject Comparing Watch_PAT with Polysomnography

We constructed ROC curves to assess the sensitivity and specificity of the Watch_PAT system using a range of PSG RDI.C threshold values (5, 10, 15, 20, and 30 events per hour) to differentiate normal cases from those with OSA. The area under the ROC curves was undefined for a threshold of 5 events per hour, 0.96, 0.89, 0.93, and 0.95, respectively. Optimal combinations of sensitivity and specificity are shown in Table 2. When using the previously described clinical approach to assess agreement,35 concordance was found in 24 out of 29 subjects (83%). There was substantial disagreement in 1 case (subject #9: mild OSA by Chicago criteria). The PSG RDI.C was 10.8 events per hour in this individual, but the LAB-PAT RDI was 40.6 events per hour. In this individual, the Watch_PAT device seemed to be detecting sympathetic activation associated with events that did not meet the Chicago criteria for scoring hypopneas (see Figure 2). This individual’s score on the Epworth Sleepiness Scale was 16 out of a possible score of 24.

Figure 2
This represents a 3-minute sample from a polysomnogram (subject #9: mild sleep apnea by Chicago criteria). One hypopnea is shown (event indicated by the word Hypopnea in the box). Five respiratory events detected by the Watch_PAT software and included ...
Table 2
Area Under the ROC Curve; Optimum Sensitivity and Specificity, Likelihood Ratios (Chicago Criteria)

Medicare Criteria

The mean RDI.M per PSG was 18.3 ± 21.9 events per hour, while the mean LAB-PAT ODI was 16.9 ± 19.5 events per hour. There was excellent agreement between LAB-PAT ODI and RDI.M (ICC = 0.95) as is shown in Figure 1C. A Bland Altman plot of LAB-PAT ODI and RDI.M is shown in Figure 1D.

We constructed ROC curves to assess the sensitivity and specificity of the Watch_PAT system using PSG RDI.M threshold values (5, 10, 15, 20, and 30 events per hour) to differentiate normal cases from those with OSA. The area under the ROC curves was 0.99, 1.0, 0.99, 1.0, and 0.99, respectively. Optimal combinations of sensitivity and specificity are shown in Table 3. There was a substantial disagreement for 1 case (subject #3). The PSG RDI.M was 57.6 events per hour in this individual, but the Lab-PAT ODI was 28.3 events per hour. The source of the disagreement was obstructive apneas that met Medicare criteria but did not yield a 4% or greater oxyhemoglobin desaturation. Thus, respiratory events included in the PSG RDI.M were not included in the LAB-PAT ODI. A representative sample from the PSG is shown in Figure 3.

Figure 3
This represents a 2-minute sample from a polysomnogram (subject #3: severe sleep apnea). Four obstructive apneas are shown (events indicated by the words Apnea Obstructive in the boxes). Respiratory events detected by the Watch_PAT software and included ...
Table 3
Area Under the ROC Curve; Optimum Sensitivity and Specificity, Likelihood Ratios (Medicare Criteria)

HOME-LAB Comparison

Chicago Criteria

No technical failures occurred during the HOME studies, thus data were available for analysis in all 29 subjects. The mean HOME-PAT VST for the HOME studies was 344.1 ± 73.9 minutes. There was not a significant correlation between HOME-PAT VST and PSG TST (Pearson product-moment correlation coefficient = 0.28, P = .14). The mean HOME-PAT RDI was 30.2 ± 19.5 events per hour for the home study night. There was good agreement between HOME-PAT RDI and RDI.C (ICC = 0.72, Figure 4A) even though they were recorded on different nights. A Bland Altman plot of HOME-PAT RDI and RDI.C is shown in Figure 4B. With RDI.Cs in the severe range, the HOME-PAT RDI tended to underestimate OSA severity in some cases, although the mean difference between the Home-PAT RDI and RDI.C was only 1.5 events per hour. The night-to-night variability of the PAT RDI in this study (when PAT values in the HOME are compared to PAT values in the lab) is shown in Figure 5A. The ICC for HOME-PAT RDI and LAB-PAT RDI was 0.62.

Figure 4
Home-Laboratory (Watch_PAT and polysomnography [PSG] data collected on different nights) comparisons are shown. (A) Scatter plot of PSG RDI.C vs HOME-PAT RDI with a best-fit line of Y = 8.6 + 0.76X. (B) Bland-Altman plot of PSG RDI.C vs HOME-PAT RDI. ...
Figure 5
Home-Laboratory (Watch_PAT data collected on different nights) comparisons are shown. (A) Bland-Altman plot of HOME-PAT RDI vs LAB-PAT RDI. (B) Bland-Altman plot of HOME-PAT ODI vs LAB-PAT ODI. PAT refers to peripheral arterial tonometer; PSG, polysomnography; ...

ROC curves were constructed as stated for the laboratory night with various PSG RDI.C threshold values (5, 10, 15, 20, and 30 events per hour) to differentiate normal cases from those with OSA. The area under the ROC curves was undefined for a threshold of 5 events per hour, 0.82, 0.97, 0.92, and 0.89, respectively. Optimal combinations of sensitivity and specificity are shown in Table 2. When using the previously described clinical approach to assess agreement, concordance was found in 21 out of 29 subjects (72%).

Medicare Criteria

The mean PAT ODI was 16.7 ± 17.1 events per hour for the HOME study night. There was good agreement between home-PAT ODI and PSG RDI.M (ICC = 0.80) even though they were recorded on different nights, as shown if Figure 4C. A Bland Altman plot of HOME-PAT ODI and PSG RDI.M is shown in Figure 4D. The night-to-night variability of the PAT ODI in this study is shown in Figure 5B. The agreement between LAB-PAT ODI and HOME-PAT ODI was good (ICC = 0.83).

ROC curves were constructed as stated for the laboratory night with various PSG RDI.M threshold values (5, 10, 15, 20 and 30 events per hour) differentiating normal cases from those with OSA. The area under the ROC curves was 1.0, 0.99, 0.90, 0.86, and 0.87, respectively. Optimal combinations of sensitivity and specificity are shown in Table 3.

DISCUSSION

In this study, we assessed the diagnostic accuracy of a wrist-worn device to detect OSA with concurrent PSG in the laboratory and when subject-administered in the home. This device is unique in that detection of episodic vasoconstriction of the digital vascular beds contributes to the identification of episodes of upper airway obstruction, rather than conventional measures of airflow and chest movements. The results of the IN-LAB study suggest there is generally good agreement between Watch_PAT and PSG in quantifying apnea plus hypopnea frequency with acceptable sensitivity, specificity, and diagnostic agreement between systems. The LAB-PAT ODI was concordant with the PSG RDI.M (Medicare criteria). The HOME-LAB comparison results suggest that technical failures are rare with use of the Watch_PAT system in the home (0% in this study), but there was less agreement between respiratory events detected with the Watch_PAT in the home and the laboratory PSG.

Validating an ambulatory sleep diagnostic system is challenging due to the lack of a true gold standard in detecting sleep-disordered breathing events during sleep, absence of a well-accepted cutoff in apnea-hypopnea frequency to differentiate normal cases from those with OSA, and night-to-night variability in measures of sleep and respiration that makes home assessment versus laboratory evaluation difficult. These issues will be addressed separately. Detecting apneas and hypopneas during sleep studies is not exact because quantifying airflow with pneumotachometry is not practical for routine use and the transducers that are used to measure respiration are semiquantitative at best. Furthermore, the definition of hypopnea is not consistent from laboratory to laboratory, or even in published guidelines. One definition allows scoring a hypopnea with a noticeable change in airflow terminated by an arousal or a 3% desaturation (RDI.C),26 while another requires at least a 30% reduction in airflow or effort combined with an oxyhemoglobin desaturation of at least 4% to improve the reliability of scoring respiratory events (RDI.M).29 Published data suggest that manual PSG scoring of hypopneas based on arousal without desaturation is subject to more interscorer variability.36 We therefore chose to quantify sleep-disordered breathing using both Chicago and Medicare criteria and then compared these indexes to the Watch_PAT RDI and ODI, respectively.

The absence of a clear cutoff in the RDI (Chicago or Medicare criteria) by which sleep apnea can be diagnosed presents significant challenges for calculating the prevalence of OSA37 and ROC curves that require discrimination of normal from abnormal cases using a gold standard. Thus, we used a variety of cutoffs (5, 10, 15, 20, and 30 events per hour of sleep) to assess discrimination of subjects with OSA from normal cases with the Watch_PAT. A cutoff of 5 events could not be used for the RDI.C analysis because OSA prevalence was 100% in our study population.

A recent investigation of short-term variability in respiration and sleep during unattended nonlaboratory nights reported no significant bias in RDI between study nights.38 However, both study nights were in the same environment. Previous investigations report considerable night-to-night variability in measures of RDI.39,40 In our study, subjects were studied on different nights with different environments, and we suspect this could be a source of variability in the HOME-LAB comparison. As illustrated in Figure 5A, there was a mean difference of 4.0 (33.5) events per hour when comparing the HOME-PAT RDI and LAB-PAT RDI with the bias toward larger RDIs in the lab. Therefore, false positives and false negatives encountered in the home may be attributed to night-to-night or home-to-lab variability rather than diagnostic inaccuracy of the Watch_PAT system. In spite of the problems described above (no gold standard for respiration assessment, no clear RDI cutoff for apnea diagnosis, and night-to-night variability), we conclude that the Watch_PAT system is producing accurate, clinically interpretable data.

The ASDA has classified sleep diagnostic systems into 4 categories based on the testing environment, technician attendance, and number of parameters recorded.41 Level I was reserved for in-laboratory PSG. A number of portable sleep diagnostic systems are available, but relatively few have been validated in an unattended home setting. These vary from simple oximetry (ASDA level IV) to complex systems that fully monitor sleep and respiration in the home (ASDA level II). A comprehensive evidence review of the literature on home monitoring for sleep apnea and practice parameters for the use of portable monitoring in the investigation of suspected OSA are available.34,42

Unattended, portable, full PSG performed in the home is considered a level II study by ASDA criteria. Home PSGs have been used for large epidemiology studies16 but may not be practical for clinical practice due to the requirement of technician set up. Furthermore, there is insufficient evidence to determine the validity of level II studies.34

The Watch_PAT system offers several potential advantages over other portable monitoring systems. First, the Watch_PAT detects respiratory events during epochs of sleep estimated by automated analysis of actigraphy and calculates an RDI based on this sleep time. Most ambulatory devices calculate an RDI based on the recording time, since they cannot differentiate wake from sleep. This method may lead to reduced sensitivity and artificially increased specificity. Second, the Watch_PAT can detect respiratory events that meet the Chicago criteria for scoring a hypopnea but do not cause a substantial desaturation. Since the Watch_PAT can detect sympathetic activation associated with arousal from sleep, these events are included in the PAT RDI. In addition, the Watch_PAT calculates an ODI based on desaturations of 4% or greater, a measure that is less inclusive in the detection of obstructive respiratory events but compares very well to the Medicare criteria for scoring events. Third, the Watch_PAT was simple to use for participants in our study, with a low failure rate for single use (no technical failures for the home studies). Fourth, performing Watch_PAT studies requires minimal effort from staff. Subject instruction for operating the device only required 5 to 10 minutes in our study, and the analysis of the studies was fully automated. Uploading a study from the COMPACTFLASH disk to the computer combined with automated analysis required less than 3 minutes (Dell Optiplex 1.8 GHz GX240 Pentium 4).

There are several potential disadvantages to the Watch_PAT system as well. The principal one relates to the novelty of the PAT signal and the lack of conventional measures of airflow, chest/abdominal movement, and body position on the studies. Thus, apneas that meet Medicare criteria but do not yield at least a 4% desaturation will not be included in the PAT ODI. This was the source of the disagreement between the LAB-PAT ODI and PSG RDI.M in subject #3 in our study (see Figure 3). Also, the Watch_PAT system reports a PAT RDI that does not differentiate the type of respiratory event (e.g., obstructive apnea, mixed apnea, central apnea, hypopnea) and cannot calculate a supine-only RDI, since body position is not detected by the device. These limitations may prove unsatisfactory in determining a differential diagnosis in some patients. Furthermore, using our clinical approach to assess agreement, concordance was not achieved in 28% of the cases when the HOME-PAT RDI was compared to the PSG RDI.C. Finally, the PAT signal is also susceptible to artifact due to certain arrhythmias such as atrial fibrillation and premature atrial contractions, and its response may be limited by certain medications (e.g., α-adrenergic receptor-blocking agents and nitrates).

There are a number of limitations to our study. The study included a small number of consenting patients with suspected OSA at only 1 sleep center. Thus, the prevalence of OSA and mean RDI in our study population were high. We therefore cannot assess whether these results would extend to populations with a lower probability of OSA, for example patients at primary care facilities. One important use of a home diagnostic device might be to exclude OSA in a population where the pretest probability is relatively low, thus saving the cost of in-laboratory PSGs. A study investigating the accuracy of home studies with the Watch_PAT device in such a population would be worthwhile. Another limitation of our study is that we evaluated the accuracy of the Watch_PAT device in the home by comparing measurements in that environment on one night to the gold standard PSG acquired in the laboratory on a different night. There are inherent flaws with this method as mentioned above and in previous publications.35 We also did not perform event-by-event analysis of the Watch_PAT and PSG data to determine agreement due to technical limitations of synchronizing PSG and Watch_PAT data while maintaining blinded manual scoring of the PSG.

In conclusion, this study indicates that the Watch_PAT device is easy to use for home sleep studies with a low failure rate for single use and minimal technician time when compared with PSG. Based on the likelihood ratios for our home study results, the system yielded a very large increase in the probability of having moderate to severe OSA (Chicago criteria) for a PAT RDI ≥ 12.5 events per hour (abnormal cases: PSG RDI.C ≥ 15 events per hour) and a large reduction in the probability of having moderate to severe OSA for a PAT RDI < 12.5 events per hour.43 Thus, the Watch_PAT could become a useful diagnostic tool in diagnosing moderate to severe OSA in high-risk populations where the prevalence of sleep-disordered breathing is high. Studies in other sleep disorder centers are needed to confirm our experience. Nonetheless, the Watch_PAT system could become an important clinical tool and may play an import role in reducing per patient cost in diagnosing and managing OSA.

Footnotes

Disclosure Statement

Mr. Pittman and Dr. White served as consultants for Itamar Medical Ltd., the developer of the Watch_PAT device. Itamar Medical Ltd. provided the funding necessary to perform the study. Subsequent to all data collection and initial data analysis, Mr. Pittman became an employee of Respironics Inc. Dr. White is also a consultant for Respironics Inc., WideMED Ltd., the Alfred Mann Foundation, and Aspire Medical.

References

1. Young T, Palta M, Dempsey J, Skatrud J, Weber S, Badr S. The occurrence of sleep-disordered breathing among middle-aged adults. N Engl J Med. 1993;328:1230–5. [PubMed]
2. Remmers Je, Degroot Wj, Sauerland Ek, Anch Am. Pathogenesis of upper airway occlusion during sleep. J Appl Physiol. 1978;44:931–8. [PubMed]
3. Roehrs T, Conway W, Wittig R, Zorick F, Sicklesteel J, Roth T. Sleep-wake complaints in patients with sleep-related respiratory disturbances. Am Rev Respir Dis. 1985;132:520–3. [PubMed]
4. Flemons Ww, Tsai W. Quality of life consequences of sleep-disordered breathing. J Allergy Clin Immunol. 1997;99:S750–6. [PubMed]
5. Teran-Santos J, Jimenez-Gomez A, Cordero-Guevara J. The association between sleep apnea and the risk of traffic accidents. Cooperative Group Burgos-Santander. N Engl J Med. 1999;340:847–51. [PubMed]
6. Findley Lj, Unverzagt Me, Suratt Pm. Automobile accidents involving patients with obstructive sleep apnea. Am Rev Respir Dis. 1988;138:337–40. [PubMed]
7. Lavie P, Herer P, Hoffstein V. Obstructive sleep apnoea syndrome as a risk Factor for hypertension: population study. BMJ. 2000;320:479–82. [PMC free article] [PubMed]
8. Peppard Pe, Young T, Palta M, Skatrud J. Prospective study of the association between sleep-disordered breathing and hypertension. N Engl J Med. 2000;342:1378–84. [PubMed]
9. Hung J, Whitford EG, Parsons RW, Hillman DR. Association of sleep apnoea with myocardial infarction in men. Lancet. 1990;336:261–4. [PubMed]
10. Dyken ME, Somers VK, Yamada T, Ren ZY, Zimmerman MB. Investigating the relationship between stroke and obstructive sleep apnea. Stroke. 1996;27:401–7. [PubMed]
11. Kribbs NB, Pack AI, Kline LR, et al. Effects of one night without nasal CPAP treatment on sleep and sleepiness in patients with obstructive sleep apnea. Am Rev Respir Dis. 1993;147:1162–8. [PubMed]
12. Kryger MH, Roos L, Delaive K, Walld R, Horrocks J. Utilization of health care services in patients with severe obstructive sleep apnea. Sleep. 1996;19:S111–6. [PubMed]
13. Rodenstein DO. Sleep Apnoea Syndrome: the Health economics point of view. Monaldi Arch Chest Dis. 2000;55:404–10. [PubMed]
14. Practice parameters for the indications for polysomnography and related procedures. Polysomnography Task Force, American Sleep Disorders Association Standards Of Practice Committee. Sleep. 1997;20:406–22. [PubMed]
15. Pack AI, Gurubhagavatula I. Economic implications of the diagnosis of obstructive sleep apnea. Ann Intern Med. 1999;130:533–4. [PubMed]
16. Redline S, Sanders MH, Lind BK, et al. Methods for obtaining and analyzing unattended polysomnography data for a multicenter study. Sleep Heart Health Research Group. Sleep. 1998;21:759–67. [PubMed]
17. Fry JM, Diphillipo MA, Curran K, Goldberg R, Baran AS. Full polysomnography in the home. Sleep. 1998;21:635–42. [PubMed]
18. Mykytyn IJ, Sajkov D, Neill AM, Mcevoy RD. Portable computerized polysomnography in attended and unattended settings. Chest. 1999;115:114–22. [PubMed]
19. Gagnadoux F, Pelletier-Fleury N, Philippe C, Rakotonanahary D, Fleury B. Home unattended vs hospital telemonitored polysomnography in suspected obstructive sleep apnea syndrome: a randomized crossover trial. Chest. 2002;121:753–8. [PubMed]
20. Portier F, Portmann A, Czernichow P, et al. Evaluation of home versus laboratory polysomnography in the diagnosis of sleep apnea syndrome. Am J Respir Crit Care Med. 2000;162:814–8. [PubMed]
21. Schnall RP, Shlitner A, Sheffy J, Kedar R, Lavie P. Periodic, Profound peripheral vasoconstriction—a new marker of obstructive sleep apnea. Sleep. 1999;22:939–46. [PubMed]
22. Bar A, Pillar G, Dvir I, Sheffy J, Schnall RP, Lavie P. Evaluation of a portable device based on peripheral arterial tone for unattended home sleep studies. Chest. 2003;123:695–703. [PubMed]
23. Ayas NT, Pittman S, Macdonald M, White DP. Assessment of a wrist-worn device in the detection of obstructive sleep apnea. Sleep Med. 2003;4:435–42. [PubMed]
24. O’Donnell CP, Allan L, Atkinson P, Schwartz AR. The effect of upper airway obstruction and arousal on peripheral arterial tonometry in obstructive sleep apnea. Am J Respir Crit Care Med. 2002;166:965–71. [PubMed]
25. Johns MW. A new method for measuring daytime sleepiness: The Epworth Sleepiness Scale. Sleep. 1991;14:540–5. [PubMed]
26. Sleep-related breathing disorders in adults: recommendations for syndrome definition and measurement techniques in clinical research. The Report of an American Academy of Sleep Medicine Task Force. Sleep. 1999;22:667–89. [PubMed]
27. 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. [PubMed]
28. EEG arousals: scoring rules and examples: a preliminary report from the Sleep Disorders Atlas Task Force Of The American Sleep Disorders Association. Sleep. 1992;15:173–84. [PubMed]
29. Meoli AL, Casey KR, Clark RW, et al. Hypopnea in sleep-disordered breathing in adults. Sleep. 2001;24:469–70. [PubMed]
30. Cohen J. A Coefficient of Agreement for Nominal Scales. Educational and Psychological Measurement. 1960;20:37–46.
31. Portney LG, Watkins MP. Foundations of Clinical Research: Applications to Practice. 1. Norwalk: Appleton & Lange; 1993.
32. Bland JM, Altman DG. Statistical methods for assessing agreement between two methods of clinical measurement. Lancet. 1986;1:307–10. [PubMed]
33. Hanley JA, Mcneil BJ. The meaning and use of the area under a receiver operating characteristic (ROC) curve. Radiology. 1982;143:29–36. [PubMed]
34. Flemons WW, Littner MR, Rowley JA, et al. 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. 2003;124:1543–79. [PubMed]
35. White DP, Gibb TJ, Wall JM, Westbrook PR. Assessment of accuracy and analysis time of a novel device to monitor sleep and breathing in the home. Sleep. 1995;18:115–26. [PubMed]
36. Whitney CW, Gottlieb DJ, Redline S, et al. Reliability of scoring respiratory disturbance indices and sleep staging. Sleep. 1998;21:749–57. [PubMed]
37. Redline S, Sanders M. Hypopnea, a floating metric: implications for prevalence, morbidity estimates, and case finding. Sleep. 1997;20:1209–17. [PubMed]
38. Quan SF, Griswold ME, Iber C, et al. Short-term variability of respiration and sleep during unattended nonlaboratory polysomnography—The Sleep Heart Health Study. Sleep. 2002;25:843–9. [PubMed]
39. Redline S, Tosteson T, Boucher MA, Millman RP. Measurement of sleep-related breathing disturbances in epidemiologic studies. Assessment of the validity and reproducibility of a portable monitoring device. Chest. 1991;100:1281–6. [PubMed]
40. Wittig RM, Romaker A, Zorick FJ, Roehrs TA, Conway WA, Roth T. Night-to-night consistency of apneas during sleep. Am Rev Respir Dis. 1984;129:244–6. [PubMed]
41. Ferber R, Millman R, Coppola M, et al. Portable recording in the assessment of obstructive sleep apnea. ASDA Standards Of Practice. Sleep. 1994;17:378–92. [PubMed]
42. Chesson AL, Jr, Berry RB, Pack A. Practice parameters for the use of portable monitoring devices in the investigation of suspected obstructive sleep apnea in adults. Sleep. 2003;26:907–13. [PubMed]
43. Flemons WW, Littner MR. Measuring agreement between diagnostic devices. Chest. 2003;124:1535–42. [PubMed]