Sleep and Breathing (2020) 24:783–790

https://doi.org/10.1007/s11325-019-01963-8

METHODS • ORIGINAL ARTICLE

Patient characteristics affecting accurate detection of sleep apnea

using a bed sheet-type portable monitor
Seiko Miyata 1 & Hironao Otake 2 & Motoo Ando 3 & Masato Okuda 4 & Hiroshige Fujishiro 1 & Kunihiro Iwamoto 1 &
Akiko Noda 4 & Michihiko Sone 2 & Norio Ozaki 1

Received: 24 March 2019 / Revised: 11 September 2019 / Accepted: 22 October 2019 / Published online: 22 November 2019
# Springer Nature Switzerland AG 2019

Abstract
Purpose Although performed inside a laboratory, attended polysomnography (PSG) has long been the gold standard for the
diagnosis of sleep apnea. However, high costs and long wait times have led to the development of home-based portable
monitoring devices. A bed sheet-shaped device called SD102 (Suzuken Co., Nagoya, Japan) has been developed, and its
accuracy in evaluating sleep apnea is becoming evident. The purpose of this study was to confirm the accuracy of SD102 in
evaluating sleep apnea and to investigate patient characteristics that may contribute to inaccurate test results in patients with
suspected obstructive sleep apnea (OSA).
Methods One hundred and eighty-nine patients simultaneously underwent PSG and portable monitoring by using a home sleep
apnea testing (HSAT) device. A blinded, experienced technologist using the American Academy of Sleep Medicine criteria
versions 2.1 and 2.3 scored the PSG data and HSAT device data, respectively.
Results The respiratory event index (REI) by HSAT significantly correlated with the apnea–hypopnea index (AHI) by PSG (r =
0.974, p < 0.001). HSAT sensitivity, specificity, and positive and negative predictive values of 0.99, 0.83, 0.95, and 0.97,
respectively. Body mass index and arousal index were significantly associated with the difference between REI from SD102
HSAT and AHI from PSG.
Conclusions This study demonstrates the good agreement between REI and AHI in patients with suspected OSA and suggests
that understanding the limitations of different testing methods may help in the accurate detection of OSA.

Keywords Home sleep apnea testing . Polysomnography . Body mass index . Arousal . Sheet-shaped device

Introduction OSA is associated with a range of adverse consequences, such

as decreased workplace productivity [2], automobile accidents
Obstructive sleep apnea (OSA) is both highly prevalent and [3], systemic hypertension [4], and cardiovascular disease [5].
consequential. The prevalence of OSA, which is defined as an Currently, the gold standard method for the diagnosis of OSA
apnea–hypopnea index (AHI) of ≥ 5 events per hour, is esti- is polysomnography (PSG), which is performed in a sleep
mated to be 33.9% in men and 17.4% in women [1]. Untreated laboratory. However, the limited number of sleep laboratories
has resulted in a search for alternative diagnostic methods.
Home sleep apnea testing (HSAT) devices have been wide-
* Seiko Miyata ly used to monitor sleep apnea. A bed sheet-like device called
[email protected]
SD102 contains 99 membrane-type pressure sensors that can
detect the gravitational alterations accompanying respiratory
1
Department of Psychiatry, Nagoya University Graduate School of movements in the body [6]. Unlike other HSAT devices, this
Medicine, 65 Tsurumai-cho, Showa-ku, Nagoya 466-8550, Japan
sensor avoids the discomfort associated with nasal, oral air-
2
Department of Otorhinolaryngology, Nagoya University Graduate flow, or snore sensors and chest/abdominal belts for sensing
School of Medicine, Nagoya, Japan
respiratory movements. Therefore, this method is advanta-
3
Department of Clinical Laboratory, Nagoya University Hospital, geous for home-based OSA screening because an examinee
Nagoya, Japan
simply sleeps on a bed sheet without physical restraint or
4
Department of Biomedicial Sciences, Chubu University Graduate invasiveness. A correlation between the results from this
School of Life and Health Sciences, Kasugai, Japan
784 Sleep Breath (2020) 24:783–790

device and PSG data has been reported in adults, thus demon- trouble was observed. Finally, we analyzed data in 189 pa-
strating the usefulness of SD102 [6, 7]. The inclusion of oxy- tients (Fig. 1). Clinical diagnosis was determined based on
gen saturation (SpO2) in OSA screening improves the accura- international classification of sleep disorders third edition.
cy of the analysis [8]. However, the influence of patient char- All patients were informed of the objectives and conditions
acteristics on measurements using this HSAT remains unclear. of the study, and written informed consent was provided prior
The sleep, cardiovascular, oximetry, position, effort, and re- to participation. The study protocol was approved by the
spiratory (i.e., SCOPER) system emphasizes that clinicians ethics review committees at Nagoya University Graduate
should understand the technical specifications and limitations School of Medicine and Nagoya University Hospital.
of the devices at their disposal [9].
To address this important question, we confirmed the ac- Polysomnography
curacy of SD102 HSAT in our study population and investi-
gated the patient characteristics attributed to the inaccuracies PSG consisted of continuous recordings of electroencephalog-
of this device in patients with suspected OSA. raphy (F3-A2, F4-A1, C3-A2, C4-A1, O1-A2, O2-A1), electro-
oculography, electromyography, electrocardiography,
oronasal airflow with a thermistor and a pressure sensor, and
Patients and methods thoracoabdominal motion with piezo sensors. SpO2 was mon-
itored by a pulse oximeter. PSG data were analyzed using
Patients Polysmith software (Nihon Kohden Co., Tokyo, Japan).
Sleep staging and arousals were scored according to the
Consecutive patients, referred to the Department of AASM Manual for the Scoring of Sleep and Associated
Otolaryngology and Psychiatry Nagoya University Hospital, Events version 2.1 [10]. Apneas were defined as a decrease
were prospectively recruited between August 2012 and in peak thermal sensor excursion by > 90% from the pre-event
November 2016. Incursion criteria were patients of 20 years baseline, where ≥ 90% of apnea duration met the amplitude
and older with the symptoms such as snoring, excessive day- reduction criteria for apnea with ≥ 10-s duration. Hypopneas
time sleepiness, or sleep problems, for instance insomnia and were scored if the nasal pressure signal excursion decreased
sleep talking. Exclusion criteria were patients with body by > 30% of baseline for ≥ 10 s with either ≥ 3% desaturation
weight < 15 kg or > 200 kg, or had undergone implantation or arousal. The 3% oxygen desaturation per hour (oxygen
of medical devices such as a cardiac pace maker. Two hundred desaturation index 3%ODI), minimum SpO2, time SpO2 <
and fifteen of all patients underwent PSG during this period 90% (% total sleep time (TST)), arousal index, and periodic
consented to the collection of simultaneous measurements limb movement (PLM) index were evaluated.
using the SD102 during inlaboratory PSG. They underwent
PSG via PSG-1100 (Nihon Kohden Co., Tokyo, Japan) and SD102 home sleep apnea testing device
monitoring via the SD102 HSAT sleep recorder (Suzuken Co.,
Nagoya, Japan), simultaneously. Trained technologists who The SD102 sleep recorder was a bed sheet-shaped HSAT de-
were blinded to the data manually analyzed both PSG and vice with SpO2 monitoring. The device was 812 mm wide ×
HSAT data. Patients with insufficient data on backgrounds, 555 mm long and was spread between the mattress and a
PSG, and SD102 HSAT were excluded from this study. In sheet. This device has 99 thin membrane-type pressure sen-
detail, data in 26 patients excluded from the study because sors with high sensitivity arranged at 40-mm intervals.
of insufficient sleep duration and removing sensors/stopping Gravitational alterations in the body corresponding to respira-
recording of SD102 HSAT by patients. None of machine tory movements were detected by the multipoint sensors.

Fig. 1 Flow chart showing the

patients selection process leading
to the final analysis
Sleep Breath (2020) 24:783–790 785

Apnea or hypopnea events were automatically identified 35 patients (18.5%). Central sleep apnea was also recognized
when the decrease in the pressure alteration due to respiratory in 13 patients (6.8%).
movement was > 30% in mean respiratory waveforms for at SD102-REI showed good overall correlation with PSG-
least 10 s. In addition to automatic detection, well-trained AHI (r = 0.974, p < 0.001) (Fig. 2). Fig. 3 shows the Bland–
technologists reviewed the data. To prevent the erroneous de- Altman plot. The mean difference was 2.5/h, and the 95% CI
tection of pressure changes due to body movement, the SD102 was 8.5 to 13.6/h.
device promptly recognized the body position and automati- The sensitivity, specificity, and positive and negative pre-
cally selected the most suitable sensors to detect the pressure dictive values for the SD102 at a AHI cutoff on PSG of 5/h
alterations due to true respiratory movement. This sensor se- were 0.99, 0.83, 0.95, and 0.97, respectively. The diagnostic
lection system was described in more detail in previous re- accuracy of SD102 was 0.95, the false positive rate was 0.16,
ports [6–8]. Movement time, defined as the time of respiratory and the false negative rate was 0.02; however, SD102 incor-
waveforms, could not be read owing to the artifacts devel- rectly identified the severity of respiratory events in 24 pa-
oped, except for those after respiratory events and time that tients (12.6%). The positive and negative likelihood ratios
a patient left a bed, which was detected by disappearance of were 20.7 and 0.02 (Table 2). At a PSG-AHI cutoff value of
respiratory waveforms, and was excluded from the total re- 15/h, the sensitivity, specificity, and positive and negative pre-
cording time (TRT). The respiratory event index (REI) obtain- dictive values of RDI on the SD102 were 0.93, 0.96, 0.96, and
ed from this device is the total number of apnea and hypopnea 0.93, respectively. The accuracy, and positive and negative
events occurring per hour during the TRT [11]. likelihood ratios of SD102 were 0.95, 30.3, and 0.06, respec-
tively. The sensitivity, specificity, and positive and negative
predictive values at a AHI cutoff on PSG of 30/h were 0.87, 1,
Statistical analysis
1, and 0.94, respectively, on SD102. The accuracy and nega-
tive likelihood ratio of SD102 were 0.96 and 0.05,
The difference of the two methods was calculated as follows:
respectively.
PSG-AHI − SD102-REI (δPSG-AHI − SD102-REI). δAHI-
ROC curves is displayed in Fig. 4. The AUC for an AHI
3%ODI was indicated as apnea/hypopnea with associated
cutoff value of 5/h was 0.97 (95% CI 0.95–0.99). The AUC
arousal without desaturation. The normal distribution of re-
for AHI cutoff values of 15 and 30/h was 0.99 (95% CI 0.98–
sults was confirmed using the Kolmogorov–Smirnov test.
1.00) and 0.99 (95% CI 0.98–1.00), respectively. The REI cutoffs
Characteristics of men and women were compared with
from the SD102 at PSG-AHI cutoffs of 5, 15, and 30/h were 4.9
Mann-Whitney U tests. The correlation between PSG-AHI
Fig. 4a, 15.7 Fig. 4b, and 24.9 Fig. 4c, respectively.
and SD102-REI, δPSG-AHI-SD102-REI, patient characteris-
In all patients, δPSG-AHI-SD102-REI significantly corre-
tics, and sleep measurements was analyzed using Spearman’s
lated with age, BMI, TST, sleep efficiency, arousal index, and
rank correlation coefficient. Bland–Altman analysis was per-
δAHI-3%ODI (age: r = 0.167, p = 0.021; BMI: r = 0.245, p =
formed to determine and visualize the agreement between the
0.001; TST: r = − 0.151, p = 0.038; sleep efficiency: r = −
two methods, and bias (average difference between the two
0.283, p < 0.001; arousal index: r = 0.607, p < 0.001; and
methods), 95% confidence intervals, and 95% limits of agree-
δAHI - 3%ODI: r = 0.611, p < 0.001). In men, δPSG-AHI-
ment were calculated. The area under the receiver operating
SD102-REI significantly correlated with BMI, TST, sleep ef-
characteristic (ROC) curve (AUC) was calculated for the cut-
ficiency, arousal index, and δAHI-3%ODI (BMI: r = 0.243, p
off point based on PSG-AHI of 5, 15, and 30/h. Multiple linear
< 0.001; TST: r = −0.214, p = 0.014; sleep efficiency: r =
regression analysis was performed to examine the association
−0.377, p < 0.001; arousal index: r = 0.663, p < 0.001;
of δPSG-AHI-SD102-REI with age, gender, body mass index
δAHI - 3%ODI: r = 0.690, p < 0.001). In women, δPSG-
(BMI), and significant factors from the correlation analysis.
AHI-SD102-REI significantly correlated with arousal index,
All tests were two sided, and a p value of 0.05 was considered
δAHI-3%ODI, and PLMs index (arousal index: r = 0.472, p <
statistically significant. All statistical analyses were performed
0.001; δAHI - 3%ODI: r = 0.591, p < 0.001; PLMs: index r =
using SPSS software version 25 (SPSS, IBM).
−0.274, p = 0.041) (Table 3).
BMI, arousal index, and δAHI-3%ODI were independent-
ly associated with δPSG-AHI-SD102-REI (BMI β = 0.192, p
Results = 0.001; arousal index β = 0.324, p < 0.001; δAHI - 3%ODI, β
= 0.396, p < 0.001) according to multiple linear regression
A total of 189 patients were enrolled in this study (133 men analysis. Both men and women showed that BMI and δAHI-
and 56 women, mean age 56.1 ± 18.3 years). Table 1 shows 3%ODI were independently associated with δPSG-AHI-
the patient backgrounds, polysomnographic data collected at SD102-REI (BMI β = 0.162, p = 0.011; β = 0.360, p =
the study, and clinical diagnosis. Most patients were diag- 0.001; δAHI-3%ODI, β = 0.268, p < 0.001; β = 0.583, p <
nosed with OSA (146, 77.2%), and PLMs were observed in 0.001). In contrast, arousal index was the independent
786 Sleep Breath (2020) 24:783–790

Table 1 Characteristics of all

patients who underwent Total (n = 189) Men (n = 133) Women (n = 56) p value
polysomnography. (men vs women)

Age (years) 56.1 ± 18.3 56.9 ± 18.1 54.2 ± 18.7 0.277

BMI (kg/m2) 25.0 ± 4.3 25.3 ± 4.3 24.5 ± 4.2 0.174
ESS 8.9 ± 5.4 8.7 ± 5.2 9.5 ± 5.8 0.357
Sleep measurements by PSG
TRT (min) 499.4 ± 57.8 501.3 ± 59.1 494.6 ± 55.0 0.452
TST (min) 398.4 ± 79.4 399.7 ± 79.5 395.3 ± 79.8 0.861
Sleep efficiency (%TRT) 79.6 ± 12.2 79.6 ± 12.1 79.7 ± 12.5 0.976
Arousal index (/h) 26.9 ± 15.4 28.1 ± 16.2 23.9 ± 12.9 0.127
AHI (/h) 22.4 ± 21.2 25.1 ± 21.9 16.0 ± 18.1 0.001
3%ODI (/h) 18.1 ± 18.8 20.3 ± 19.8 12.7 ±15.1 0.005
Minimum SpO2 (%) 82.8 ± 9.3 81.7 ± 9.7 85.7 ± 7.3 0.007
Time < SpO2 90% (%TST) 4.7 ± 11.0 5.5 ± 11.5 2.6 ± 9.6 0.001
PLMs index (/h) 7.9 ± 17.1 8.8 ± 18.4 5.6 ± 13.8 0.032
Clinical diagnosis* (n, %)
Obstructive sleep apnea disorder 146 (77.2) 110 (82.7) 36 (64.2) 0.008
Central sleep apnea syndrome 13 (6.8) 12 (9.0) 1 (1.7) 0.113
Periodic limb movement disorder 35 (18.5) 26 (19.5) 9 (16.0) 0.684
Others** 33 (17.4) 17 (18.0) 16 (28.5) 0.012
WNL 14 (7.4) 7 (5.2) 7 (12.5) 0.124

Data are expressed as mean ± standard deviation

BMI body mass index, ESS Epworth sleepiness index, PSG polysomnogaraphy, TRT total recording time, TST
total sleep time, AHI apnea/hypopnea index, ODI oxygen desaturation index, SpO2 oxygen saturation, PLMs
periodic limb movements, WNL within normal limit on PSG
a
The number is overlapping
b
Insomnia, hypersomnia (included due to medication), narcolepsy, insufficient sleep syndrome, circadian rhythm
sleep-wake disorder, and REM sleep behavior disorder

significant factors related to δPSG-AHI-SD102-REI only in

men (β = 0.471, p < 0.001) (Table 3). 120
r=0.974
p < 0.001
100
Discussion

Our findings demonstrate that the SD102 HSAT device has 80

SD102-REI (/hr)

excellent sensitivity. The device also has excellent positive

and negative predictive values for the detection of AHI ≥ 5/h
60
on PSG, and the REI derived from SD102 shows good corre-
lation with AHI derived from PSG. Home-based testing is
widely used to evaluate sleep apnea, and some studies report- 40
ed that the effectiveness of HSAT is similar to that of PSG in
managing patients with suspected sleep apnea [12, 13].
Conversely, we observed that BMI, arousal index, and 20
hypopnea can affect the difference between PSG and
SD102. On the basis of these results, we conclude that
home-based testing may be a useful tool for managing patients 0
with OSA if the disadvantages can be mitigated. 0 20 40 60 80 100 120
Many kinds of HSAT devices have been developed, and PSG-AHI (/hr)
these devices correlate well with PSG [6–8, 14–16]. In the Fig. 2 Correlation between PSG-AHI and SD102-REI
Sleep Breath (2020) 24:783–790 787

Fig. 3 Bland–Altman plots 40

showing the mean difference
between PSG-AHI and SD102- 30

PSG-AHI - SD102-REI (/hr)

REI (solid line); the 95% confi-
dence intervals (broken lines) 20

10

-10

-20

-30

-40
0 20 40 60 80 100 120
Average of PSG-AHI and SD102-REI (/hr)

present study, SD102 showed high sensitivity, positive and discrepancy between PSG and SD102, and arousal index
negative predictive values, and excellent concordance with and hypopnea associated with arousal were significant factors
PSG. These data are consisted with those of previous studies for identifying the difference between the two methods. The
[6–8]. The correlation analyses performed by studies that analysis of autonomic signals from a wrist-worn ambulatory
monitored only respiratory effort show a high agreement be- sleep recorder based on peripheral arterial tone signal, pulse
tween PSG and SD series devices (r = 0.88 [6] and r = 0.83 rate, actigraphy, and pulse oximetry can detect sleep stages
[7]), and the inclusion of SpO2 in manual analysis improves that show moderate agreement with more standard techniques
the accuracy of sleep apnea screening [8, 17]. The advantage in normal subjects and patients with OSA [18]. Our previous
of the SD102 device is that the subject does not need to be study investigated the utility of actigraphy and showed that the
restrained by attached sensors, thus allowing the patient to TST and sleep efficiency obtained using actigraphy were sig-
experience natural sleep and less discomfort. This portable nificantly correlated with those obtained using PSG.
device may be favorable for evaluating sleep apnea in older Actigraphy provides a valuable sleep–wake rhythm assess-
individuals, patients with mental disabilities [8], and children ment in outpatients with OSA, for whom PSG is difficult to
[17]. Therefore, the manual analysis of SD102 data may pro- perform [19]. The disadvantage of home-based testing can be
vide important information for evaluating sleep apnea among partly overcome by making use of tools that can estimate sleep
individuals with suspected OSA. and wake cycles.
One of the limitations of HSAT is that it cannot differentiate The one important factor that affects the difference between
between wakeful and sleeping states. As a result, the severity PSG and SD102 in this study was BMI. There may be several
of a patient’s sleep apnea is sometimes underestimated. In our reasons for this result. First, it is well known that BMI is
study, sleep efficiency was negatively correlated with the significantly correlated with AHI severity. The Bland–
Altman findings revealed a trend of lower SD102-REI than
PSG-AHI in patients with severe OSA. This phenomenon is
Table 2 Concordance between PSG and SD102 for the severity of AHI quite consistent with the results of previous studies on type 3
and REI. HSAT devices [8, 14, 16]. As AHI increases, sleep becomes
Total SD102 REI (/h) more fragmented, thus leading to lower sleep efficiency and a
wider discrepancy between actual sleep time and recording
<5 < 15 < 30 ≥ 30 Total time. Second, overweight patients show a greater decrease in
SpO2 with shorter apnea/hypopnea and smaller respiratory
PSG-AHI (/h) <5 36 7 0 0 43
waveforms when using SD102. The sensors on SD102 detect
< 15 1 45 3 0 49
gravitational alterations in the body corresponding to respira-
< 30 0 6 33 0 39
tory movements. As the diaphragm shifts toward the abdo-
≥ 30 0 0 7 51 58
men, inspiration occurs, and the lungs are filled with air.
Total 37 58 43 51 189
Thus, the pressure distribution per unit area applied to the
PSG polysomnography, AHI apnea/hypopnea index, REI respiratory sensors in contact with the chest and abdomen paradoxically
event index changes. These changes in pressure distribution are reversed
788 Sleep Breath (2020) 24:783–790

AUC 0.97
Cut off 4.9

a. AHI cutoff on PSG of 5 /hr

AUC 0.99
AUC 0.99
Cut off 15.7
Cut off 24.9

b. AHI cutoff on PSG of 15 /hr

c. AHI cutoff on PSG of 30 /hr
Fig. 4 Receiver operating characteristic curves with cutoff value AHI of 5, 15, and 30/h based on polysomnography

during expiration. Therefore, respiratory movements can be In men, obesity, arousals, and hypopneas associated with
detected as changes in pressure distributions [6]. Obesity arousal were significant factors. Meanwhile, in women, obe-
may create difficulty in detecting respiratory waveforms. sity and hypopneas associated with arousal independently re-
Previous studies performed in patients with suspected sleep lated to the deference of the number of respiratory events
apnea showed the low specificity [6–8]. This might be due to between the two methods. Previous studies showed gender
the low number of patients with < 5/h. Our study included difference in the polysomnographic features of patients with
wide variety of sleep disorder patients with common symp- OSA. The retrospective study of 830 patients with OSA re-
toms with sleep apnea. High specificity is desirable to exclude vealed that OSA was less severe in women owing to milder
false-positive findings in order to screen in general popula- OSA during non-REM sleep. Also, women had a higher prev-
tions. SD102 is possibly used to find patents with sleep apnea alence of OSA occurring almost exclusively during REM
although of eight (4%) patients were misdiagnosed using sleep than men [20]. Obese women with OSA showed in-
SD102. creased number of arousals and decreased sleep efficiency
To our knowledge, this is the first study on the effect of compared with age- and body weight-matched men [21].
gender difference on the difference between PSG and HAST. Furthermore, we observed gender difference in the prevalence
Sleep Breath (2020) 24:783–790 789

Table 3 Statistical analysis of the differences between PSG-AHI and SD102-REI

Total Men Women

Variables Simple correlation Multiple linear Simple correlation Multiple linear Simple correlation Multiple linear
analysis regression analysis analysis regression analysis analysis regression analysis

r p value β p value r p value β p value r p value β p value

Age 0.167 0.021 − 0.084 0.151 0.149 0.087 − 0.129 0.060 0.114 0.405 0.047 0.673
Gender – – − 0.008 0.867 – – – – – – – –
BMI 0.245 0.001 0.198 < 0.001 0.243 0.005 0.162 0.011 0.205 0.134 0.360 0.001
ESS − 0.049 0.503 – – 0.005 0.956 – – − 0.147 0.280 – –
TRT 0.077 0.294 – – 0.107 0.221 – – − 0.055 0.686 – –
TST − 0.151 0.038 − 0.132 0.128 − 0.214 0.014 − 0.100 0.305 − 0.042 0.757 – –
Sleep efficiency − 0.283 < 0.001 0.031 0.729 − 0.377 < 0.001 − 0.132 0.202 − 0.105 0.441 – –
Arousal index 0.607 < 0.001 0.367 < 0.001 0.663 < 0.001 0.471 < 0.001 0.472 < 0.001 0.071 0.581
δAHI-3%ODI 0.611 < 0.001 0.400 < 0.001 0.690 < 0.001 0.268 < 0.001 0.591 < 0.001 0.538 < 0.001
PLMs index − 0.077 0.293 – – − 0.039 0.661 – – − 0.274 0.041 − 0.190 0.065

PSG polysomnography, REI respiratory event index, BMI body mass index, ESS Epworth sleepiness index, TRT total recording time, TST total sleep
time, AHI apnea/hypopnea index, ODI oxygen desaturation index, SpO2 oxygen saturation, PLMs periodic limb movement

of OSA and respiratory event measurement. These features approximately 60% patients with OSA having supine predom-
might influence the factors associated with SD102 inant OSA [23]. We did not evaluate the respiratory event
uncertainty. regarding sleeping positions in this study. SD102 can detect
The results of our study showed that SD102 was useful for body positions by monitoring the pressure distribution.
confirming or excluding the diagnosis of sleep apnea in pa- Hence, additional study concerning the sleeping positions
tients with sleep problems such as snoring, excessive daytime may provide useful information for evaluating OSA.
sleepiness, or insomnia. Specifically, on the basis of a PSG In summary, the portable HSAT device SD102 shows high
AHI cutoff of 5/h, an AHI of 5/h from SD102 would effec- sensitivity, positive predictive value, and negative predictive
tively exclude and confirm an sleep apnea diagnosis. The use value in evaluating respiratory events and can accurately iden-
of HSAT provides numerous advantages, including increased tify AHI severity in patients suspected of sleep apnea. These
health care accessibility, shorter waiting times, earlier treat- data indicate that SD102 could be used to facilitate the simple
ment initiation, and a potential cost reduction [22]; on the and rapid screening of patients suspected of sleep apnea, re-
other hand, a limitation in the evaluation of OSA severity duce waiting time for diagnosis, and enable treatment to be
using SD102 is that BMI and arousals affected the difference initiated quickly. BMI and microarousals influence the differ-
between PSG-AHI and SD102-REI. Although SD102 cannot ences between PSG and SD102. Our study also suggests that
detect arousals during recording time, excluding movement the use of HSAT should be performed with the knowledge of
time, unreadable respiratory waveforms due to the artifacts its limitations.
developed, and time that a patient left a bed could be improve
the accuracy of OSA severity. We suggest that understanding Funding information This study was partly supported by JSPS
KAKENHI Grant-in-Aid for Young Scientists B (Grant No. 16K16600).
the technical specifications and limitations of SD102, more-
over, carefully eliminating artifacts and scoring respiratory
events by visually may contribute to the accurate evaluation Compliance with ethical standards
of OSA using SD102.
Conflict of interest The authors declare that they have no conflict of
This study has several limitations. First, the study popula- interest.
tion involved patients with snoring, excessive daytime sleep-
iness, and/or apnea from one hospital; this approach may Ethical approval All procedures performed in studies involving human
cause sampling bias. Second, HSAT devices are usually used participants were in accordance with the ethical standards of the institu-
at home; however, the installation and operation of SD102 tional and/or national research committee and with the 1964 Helsinki
declaration and its later amendments or comparable ethical standards.
were performed by a trained technologist. The utility of this
device in clinical settings should be examined by further eval- Informed consent Informed consent was obtained from all individual
uations. OSA related to the supine position is more severe than participants included in the study.
that related to the lateral position and is very common, with
790 Sleep Breath (2020) 24:783–790

References N, Monasterio C, Terán J, Alonso-Alvarez ML, Muñoz-Méndez J,

Arias EM, Cabello M, Montserrat JM, de la Peña M, Serrano JC,
Barbe F, Masa JF, Spanish Sleep Network (2017) Conventional
1. Peppard PE, Young T, Barnet JH, Palta M, Hagen EW, Hla KM
polysomnography is not necessary for the management of most
(2013) Increased prevalence of sleep-disordered breathing in adults.
patients with suspected obstructive sleep apnea. Noninferiority, ran-
Am J Epidemiol 177:1006–1014
domized controlled trial. Am J Respir Crit Care Med 196:1181–
2. Mulgrew AT, Ryan CF, Fleetham JA, Cheema R, Fox N, Koehoorn
1190
M, Fitzgerald JM, Marra C, Ayas NT (2007) The impact of obstruc-
tive sleep apnea and daytime sleepiness on work limitation. Sleep 14. Ward KL, McArdle N, James A, Bremner AP, Simpson L, Cooper
Med 9:42–53 MN, Palmer LJ, Fedson AC, Mukherjee S, Hillman DR (2015) A
3. Noda A, Yagi T, Yokota M, Kayukawa Y, Ohta T, Okada T (1998) comprehensive evaluation of a two-channel portable monitor to
Daytime sleepiness and automobile accidents in patients with ob- "rule in" obstructive sleep apnea. J Clin Sleep Med 11:433–444
structive sleep apnea syndrome. Psychiatry Clin Neurosci 52:221– 15. Alshaer H, Fernie GR, Tseng W-H, Bradley TD (2016) Comparison
222 of in-laboratory and home diagnosis of sleep apnea using a cordless
4. Peppard PE, Young T, Palta M, Skatrud J (2000) Prospective study portable acoustic device. Sleep Med 22:91–96
of the association between sleep-disordered breathing and hyper- 16. Körkuyu E, Düzlü M, Karamert R, Tutar H, Yılmaz M, Çiftçi B,
tension. New Eng J Med 342:1378–1384 Güven SF (2015) The efficacy of Watch PAT in obstructive sleep
5. Marin JM, Carrizo SJ, Vicente E, Agusti AGN (2005) Long-term apnea syndrome diagnosis. Eur Arch Otorhinolaryngol 272:111–
cardiovascular outcomes in men with obstructive sleep apnoea- 116
hypopnoea with or without treatment with continuous positive air- 17. Arimoto M, Shiomi T, Sasanabe R, Inagawa S, Ueda H, Inafuku S
way pressure: an observational study. Lancet 365:1046–1053 (2011) A sheet-type device for home-monitoring sleep apneas in
6. Agatsuma T, Fujimoto K, Komatsu Y, Urushihata K, Honda T, children. Sleep Biol Rhythm 9:103–111
Tsukahara T, Nomiyama T (2009) A novel device (SD-101) with 18. Hedner J, White DP, Malhotra A et al (2011) Herscovici S, Pittman
high accuracy for screening sleep apnoea-hypopnoea syndrome. SD, Zou D, Grote L, Pillar G: Sleep staging based on autonomic
Respirology 14:1143–1150 signals: a multi-center validation study. J Clinical Sleep Med 7:
7. Tsukahara M, Sakao S, Jujo T (2014) The accuracy and uncertainty 301–306
of a sheet-type portable monitor as a screening device to identify 19. Otake M, Miyata S, Noda A, Koike Y, Hara Y, Sugiura M,
obstructive sleep apnea-hypopnea syndrome. Intern Med 53:1307– Minoshima M, Kojima J, Nakata S, Nakashima T (2011)
1313 Monitoring sleep-wake rhythm with actigraphy in patients on con-
8. Kobayashi M, Namba K, Tsuiki S (2013) Validity of sheet-type tinuous positive airway pressure therapy. Respiration 82:136–141
portable monitoring device for screening obstructive sleep apnea 20. O'Connor C, Thornley KS, Hanly PJ (2000) Gender differences in
syndrome. Sleep Breath 17:589–595 the polysomnographic features of obstructive sleep apnea. Am J
9. Collop NA, Tracy SL, Kapur V, Mehra R, Kuhlmann D, Fleishman Respir Crit Care Med 161:1465–1472
SA, Ojile JM (2011) Obstructive sleep apnea devices for out-of- 21. Resta O, Carpanano GE, Lacedonia D, di Gioia G, Giliberti T,
center (OOC) testing: technology evaluation. J Clin Sleep Med 7: Stefano A, Bonfitto P (2005) Gender difference in sleep profile of
531–548 severely obese patients with obstructive sleep apnea (OSA). Respir
10. Berry RB, Brooks R, Gamaldo CE et al (2014) The AASM manual Med 99:91–96
for the scoring of sleep and associated events: rules, terminology 22. Boulos MI, Elias S, Wan A, Im J, Frankul F, Atalla M, Black SE,
and technical specifications, version 2.1. American Academy of Basile VS, Sundaram A, Hopyan JJ, Boyle K, Gladstone DJ,
Sleep Medicine, Illinois. Swartz RH, Murray BJ (2017) Unattended hospital and home sleep
11. Berry RB, Brooks R, Gamaldo CE et al (2016) The AASM manual apnea testing following cerebrovascular events. J Stroke
for the scoring of sleep and associated events: rules, terminology Cerebrovasc Dis 26:143–149
and technical specifications, version 2.3. American Academy of
23. Barnes H, Edwards BA, Joosten SA, Naughton MT, Hamilton GS,
Sleep Medicine, Illinois.
Dabscheck E (2017) Positional modification techniques for supine
12. Hui DS, Ng SS, To KW et al (2017) A randomized controlled trial
obstructive sleep apnea: A systematic review and meta-analysis.
of an ambulatory approach versus the hospital-based approach in
Sleep Med Rev 36:107–115
managing suspected obstructive sleep apnea syndrome. Sci Rep 8:
45901
13. Corral J, Sánchez-Quiroga MÁ, Carmona-Bernal C, Sánchez- Publisher’s note Springer Nature remains neutral with regard to jurisdic-
Armengol Á, de la Torre AS, Durán-Cantolla J, Egea CJ, Salord tional claims in published maps and institutional affiliations.