Abstract
Study Objectives:
People with obstructive sleep apnea (OSA) remain undiagnosed because of the lack of easy and comfortable screening tools. Through this study, we aimed to compare the diagnostic accuracy of chest wall motion and cyclic variation of heart rate (CVHR) in detecting OSA by using a single-lead electrocardiogram (ECG) patch with a 3-axis accelerometer.
Methods:
In total, 119 patients who snore simultaneously underwent polysomnography with a single-lead ECG patch. Signals of chest wall motion and CVHR from the single-lead ECG patch were collected. The chest effort index (CEI) was calculated using the chest wall motion recorded by a 3-axis accelerometer in the device. The ability of CEI and CVHR indices in diagnosing moderate-to-severe OSA (apnea-hypopnea index ≥ 15) was compared using the area under the curve (AUC) by using the DeLong test.
Results:
CVHR detected moderate-to-severe OSA with 52.9% sensitivity and 94.1% specificity (AUC: 0.76, 95% confidence interval: 0.67–0.84, optimal cutoff: 21.2 events/h). By contrast, CEI identified moderate-to-severe OSA with 80% sensitivity and 79.4% specificity (AUC: 0.87, 95% confidence interval: 0.80–0.94, optimal cutoff: 7.1 events/h). CEI significantly outperformed CVHR regarding the discrimination ability for moderate-to-severe OSA (ΔAUC: 0.11, 95% confidence interval: 0.009–0.21, P = .032). For determining severe OSA, the performance of discrimination ability was greater (AUC = 0.90, 95% confidence interval: 0.85–0.95) when combining these two signals.
Conclusions:
Both CEI and CVHR recorded from a patch-type device with ECG and a 3-axis accelerometer can be used to detect moderate-to-severe OSA. Thus, incorporation of CEI is helpful in the detection of sleep apnea by using a single-lead ECG with a 3-axis accelerometer.
Citation:
Hsu Y-S, Chen T-Y, Wu D, Lin C-M, Juang J-N, Liu W-T. Screening of obstructive sleep apnea in patients who snore using a patch-type device with electrocardiogram and 3-axis accelerometer. J Clin Sleep Med. 2020;16(7):1149–1160.
Keywords: apnea-hypopnea index, cyclic variation of heart rate, chest wall motion, electrocardiogram, sleep apnea, obstructive sleep apnea, sleep-disordered breathing
BRIEF SUMMARY
Current Knowledge/Study Rationale: To determine the accuracy and tolerability of a single electrocardiogram lead to screen for sleep apnea, we collected the data of 119 patients who snore who simultaneously underwent polysomnography and a single-lead electrocardiogram patch with 3-axis accelerometer. Chest effort index from the chest wall motion and cyclic variation of heart rate were calculated and compared with the apnea-hypopnea index recorded through polysomnography.
Study Impact: We found that chest effort index significantly outperformed the cyclic variation of heart rate on the discrimination ability in identifying moderate-to-severe obstructive sleep apnea. Our study validated that a combination of chest effort index and cyclic variation of heart rate is helpful in identifying obstructive sleep apnea in sleep tests conducted using a single-lead electrocardiogram with 3-axis accelerometer.
INTRODUCTION
Obstructive sleep apnea (OSA) occurs during sleep because of recurrent upper airway obstruction. Several complications associated with OSA have been detected, such as hypertension,1 sudden cardiac death,2 cerebrovascular diseases,3 diabetes,4 and excessive daytime sleepiness.5 Therefore, this disease considerably impacts the health and quality of life of affected individuals. Benjafield et al6 estimated that globally, 936 million adults 30–69 years of age (men and women) have mild-to-severe OSA, and 425 million adults 30–69 years of age have moderate-to-severe OSA. The prevalence is so high that there is a growing need for devices that are more accessible to people to more efficiently and conveniently diagnose new patients with sleep apnea; thus, a comprehensive assessment of novel diagnostic devices for sleep apnea is warranted.
In-laboratory polysomnography (PSG) is the first-line diagnostic study for suspected OSA.7 However, this test is expensive, time consuming, and uncomfortable. Therefore, several alternatives to PSG were proposed, and home sleep apnea testing is one of them.7 Oximetry is another less expensive and easier alternative that is widely available.8,9 However, attaching a device to the finger during sleep is uncomfortable and may limit natural position change during sleep; moreover, the device is removed on waking up, which limits its use in multiday, continuous recordings.
Cyclic variation of heart rate (CVHR) was first introduced by Guilleminault et al10 for screening of OSA. They found that at the onset of sleep apnea, patients showed progressive bradycardia, followed by abrupt tachycardia on resumption of breathing. As this screening tool has an advantage of comfort recording in sleep, especially in modern-type single-lead electrocardiogram (ECG) devices, it is potentially suitable for multiday evaluation. Several studies have been conducted to improve its algorithm and accuracy.11–16
Chest wall motion is affected in sleep apnea. During sleep apnea, the muscle tone of the upper airway is reduced. However, the diaphragm and intercostal muscle tone remain relatively the same.17 The difference in muscle function results in upper airway collapse and apnea.18 Staats et al19 suggested that respiratory efforts against the occlusion cause distortion of the chest wall and result in paradoxical motion of the rib cage, which can adequately characterize apnea in most patients. The authors also suggested that recognizing these paradoxical chest wall motions in sleep apnea could help avoid the use of invasive sleep monitoring techniques. We hypothesized that chest wall motion, as an indicator of a sleep apnea episode, can also be simultaneously recorded during recording using a single-lead ECG device.
Through this study, we aimed to evaluate the predictive performance of chest wall motion and CVHR in detecting OSA with wireless single-lead ECG monitoring patch with 3-axis accelerometer in subjects who snore. We hypothesized that using an overnight wireless single-lead ECG monitoring patch with a 3-axis accelerometer is an accurate method to identify OSA, especially when simultaneously considering CVHR and chest wall motion.
METHODS
Subjects and study protocol
In this prospective study, 119 patients who snore were included; they were subjected to in-laboratory PSG examinations at the sleep center. Among them, 97 were from Shin Kong hospital (SKH), and 22 were from Shuang Ho Hospital (SHH). The study was approved by the institutional review boards of both hospitals (SKH: 20171003R, SHH: N201709023), and written informed consent was provided by all participants. Adult patients aged between 18 and 70 years who complained of snoring or those with suspected sleep-disordered breathing were included. Patients with persistent atrial fibrillation, pacemaker implantation, ventricular tachycardia, or those who were pregnant were excluded. At the sleep center, a wireless single-lead ECG monitoring patch (Rooti Rx System, Rooti Labs Ltd., Taipei, Taiwan; Figure 1A) was simultaneously used; it was placed between the midsternal line and the left midclavicular line and around the third and fourth intercostal space. This device also provides 3-axis accelerometer signals, which could represent the motion of chest wall during respiration. The signals collected by PSG and Rooti Rx were synchronized, and the measurements derived from the two tests were compared.
Figure 1. Rooti Rx, a wireless single-lead ECG monitoring patch.
(A) Appearance of the device. (B) CVHR pattern recorded by Rooti Rx. Cyclic lengthening/shortening in the heart rate during apnea/postapneic hyperventilation is shown. The airflow and saturation data were derived from the PSG. The response of SpO2 and heart rate falls a little behind air flow in an apnea event. This phenomenon means that desaturation and heart rate variability are likely an aftermath of sleep apnea. CVHR = cyclic variation of heart rate, ECG = electrocardiogram, SpO2 = peripheral oxygen saturation.
PSG
The PSG recordings were obtained using Compumedics Grael (Compumedics Limited, Abbotsford, Victoria, Australia) (SKH) or Embla N7000 (Pleasanton, CA, USA) and Embletta MPR Sleep System (Pleasanton, CA, USA) (SHH). The sleep stages and respiratory events were scored using the updated standard diagnostic criteria of the American Academy of Sleep Medicine (AASM)20 and AASM 2017 scoring guidelines (https://aasm.org/resources/pdf/scoring-manual-update-april-2017.pdf) by an experienced registered polysomnographic technologist at the sleep center of SKH and SHH, and these scorings were rechecked by at least 2 technologists to ensure further accuracy of sleep apnea events and ECG staging. Obstructive apnea was defined as a drop in the peak thermal sensor excursion by more than 90% from the baseline for at least 10 seconds, caused by airway obstruction. If these events lacked respiratory effort, they were defined as central apnea. Hypopnea was defined as more than 30% reduction in airflow for at least 10 seconds with arousal or more than 3% oxygen desaturation. Apnea-hypopnea index (AHI) was determined as the mean number of apneas and hypopneas per hour of time in sleep for each PSG recording. Patients were considered to have moderate-to-severe OSA when their AHI was more than 15 events/h. All the PSG recordings were performed during the night, and split PSG studies were not included. The in-laboratory PSG data were carefully reviewed by a second sleep technologist. If there were differences in their assessments, these were carefully evaluated by both sleep technologists until a final consensus was met.
CVHR
CVHR is a characteristic heart rate pattern that often presents along with episodes of OSA. CVHR (Figure 1B) is an ECG pattern characterized by progressive bradycardia at the onset of sleep apnea, followed by abrupt tachycardia on resumption of breathing. This pattern is identifiable through computer analysis and can be used as a screening tool for sleep apnea.10 For single-lead ECG, Hayano et al11,16 developed an algorithm called autocorrelated wave detection with adaptive threshold for automated detection of CVHR and demonstrated that this pattern could be used to screen moderate-to-severe OSA. Hence, the analysis of single-lead ECG during sleep could be considered a screening tool.
In this study, single-lead ECGs were obtained using wireless single-lead ECG monitoring patches (Rooti Rx; Figure 1A). The ECG signals were then analyzed using RootiCare Sleep Monitoring, a certified cloud-based computing software. Using this, CVHR was detected through adaptive threshold and its time-domain dip-detection algorithm.11 Valid CVHR events were defined as at least 3 consecutive cycles of increase in heart rate (>6 beats/min) that last for 10–120 seconds. The CVHR index was then calculated as the mean number of CVHR events per hour of sleep.
Chest effort
To measure the chest wall motion during respiratory cycles and sleep apnea events, the recorded 3-axis accelerometer signals were analyzed using the Rooti Rx monitoring patches, and chest effort events were calculated (Figure 2).
Figure 2. The chest wall motion is translated from the signals of the 3-axis accelerometer.
(A) Placement of Rooti Rx monitoring patch and the axes of the 3-axis accelerometer. (B) Signals from PSG airflow and from 3-axis accelerometer of Rooti Rx in X, Y, and Z axes (labeled as G-X, G-Y, and G-Z, respectively). The Y-axis signal has the largest fluctuation amplitude during apnea and normal respiration. PSG = polysomnography.
The chest effort events were automatically detected through 3-axis G-sensor signal recording (Figure 2B). As the most significant chest wall motions were noted on the Y-axis and these signals were less affected by position change, the Y-axis signals contributed the most to the chest effort events in this study (Figure 2B).
The details of the calculation for chest effort events are shown in Figure 3. The Y-axis signals were obtained at a sampling frequency of 31.25 Hz. We initially applied the Butterworth bandpass filter with a pass band of 0.15–0.4 Hz and then performed downsampling to 1 Hz by calculating the summation of the signal in the moving window of 5 seconds. Then, the downsampled moving sum signal was used to determine the peaks and troughs. By searching forward and backward from the trough point, the points at which there was 80% peak–trough difference were also located at the start and the endpoint of each event. Based on the median of moving sum signal adjusted for each case, the drop threshold value was determined to detect sharp changes in chest wall motion, and the block threshold value was determined to detect blocked chest wall movement. Subsequently, chest effort events could be detected using these three criteria: (1) peak–trough differences should be greater than the drop threshold; (2) the trough value should be lower than the block threshold; and (3) the duration of each event (start point to endpoint) should be greater than 10 seconds.
Figure 3. Illustration of the calculated peaks and troughs of the chest effort event.
To determine the chest effort event by this diagram, 3 criteria should be achieved (marked in bold). First, peak–trough differences should be greater than the drop threshold. Second, the trough value should be lower than the block threshold. Third, the duration of each event (start point to end point) should be greater than 10 seconds.
Events that fulfilled these 3 criteria were considered chest effort events. The chest effort index (CEI) was calculated as the mean number of chest effort events per hour of sleep.
A further explanation of a drop threshold and block threshold is described here. A drop threshold is used to check if a sharp chest movement change occurs (peak-trough difference > drop threshold), because a sharp chest movement change would be a signal for the start of an obstruction event. A block threshold is a fairly low chest movement level that is used to determine whether the chest movement is blocked, and an even lower chest movement is considered an obstruction event. Drop threshold and block threshold were adjusted for each case based on the median of the moving sum signal.
Statistical analysis
The baseline characteristics of the patients in the moderate-to-severe and none-to-mild OSA groups were compared using the independent-sample t test for age (continuous variable) or χ2 test for the categorical variables. The relationships among CVHR, CEI, and AHI index were studied using Spearman rank correlation. The outcome of primary interest was moderate-to-severe OSA (AHI ≥ 15 events/h, n = 85), whereas the outcomes of secondary interest were mild, moderate, or severe OSA (AHI ≥ 5 events/h, n = 110), as well as severe OSA (AHI ≥ 30 events/h, n = 58). The association between CVHR or CEI and the risk of OSA was assessed using bivariate and multivariable logistic regression analyses. Several known confounders of OSA, including age, sex, and body mass index (BMI), were adjusted in the multivariable analysis. The values of CVHR or CEI were naturally log-transformed in logistic analyses because of the lack of normality. The performance of CVHR or CEI to detect OSA was evaluated through receiver operating characteristic curve analysis. The optimal cutoffs were determined by Youden index. The positive likelihood ratio (+LR), negative likelihood ratio (−LR), positive predictive value (PPV), and negative predictive value (NPV) were obtained according to the prevalence of patients who snore, in which 40% had moderate-to-severe OSA.21 The difference in AUCs between CEI and CVHR was compared using the DeLong test. All tests were 2-tailed, and P < .05 was considered statistically significant. No adjustment of multiple testing (multiplicity) was made in this study. Data were analyzed using SPSS 25 (IBM SPSS Inc., Chicago, IL).
The scale of AHI and CVHR/CEI is different even if their unit is the same (ie, the number of events per hour). In other words, an AHI value of 1 is not equivalent to a CVHR/CEI value of 1; likewise, a CVHR value of 1 is also not equivalent to a CEI value of 1. Because not every apnea-hypopnea event results in heart rate variability or a chest effort change, a suitable screening threshold for CVHRI and CEI was required in this study. As such, the Bland-Altman plot was not appropriate in this scenario. Furthermore, neither ECG nor chest effort signals were synchronized in this study, so we only compared the final numbers for each patient.
RESULTS
Characteristics of the study population
A total of 119 patients who met the inclusion criteria were included in this study; of these, 97 patients were from SKH, and 22 were from SHH. Table 1 presents the baseline characteristics of the patients according to the severity of OSA. In this study population, 85 patients were classified as having moderate-to-severe OSA (AHI ≥ 15 events/h), and the other 34 were classified as having none-to-mild OSA (AHI < 15 events/h). No significant difference in age, sex, and height was observed between the 2 groups. However, patients in the moderate-to-severe OSA group had a larger BMI and greater neck length. Because the moderate-severe group was predominantly male and the none-to-mild group was predominantly female, sex could explain the differences observed between these 2 groups in terms of weight, BMI, and neck length. The mean heart rate, desaturation index, and snore index, as well as the CVHR, CEI, and AHI index, were also significantly higher in the moderate-to-severe OSA group than in the none-to-mild OSA group. Significantly lower mean saturation (92.4% in the moderate-to-severe OSA group vs 96.7% in the none-to-mild OSA group) and lowest saturation (76.8% in the moderate-to-severe OSA group vs 86.4% in the none-to-mild OSA group) were also noted in the moderate-to-severe OSA group. In addition, the moderate-to-severe OSA group had more apnea and hypopnea events.
Table 1.
Baseline characteristics of patients according to severity of obstructive sleep apnea.
| Variables | Total (n = 119) | Moderate-to-Severe OSA (n = 85) | None-to-Mild OSA (n = 34) | P |
|---|---|---|---|---|
| Age (yr) | 42.8 ± 11.6 | 43.6 ± 11.7 | 40.9 ± 11.2 | .245 |
| Male | 96 (80.7) | 72 (84.7) | 24 (70.6) | .069 |
| Height (cm) | 169.7 ± 7.5 | 170.5 ± 7.4 | 167.9 ± 7.4 | .086 |
| Weight (kg) | 78.5 ± 16.0 | 81.8 ± 16.4 | 70.5 ± 11.4 | <.001 |
| BMI (kg/m2) | 27.2 ± 4.7 | 28.0 ± 4.8 | 25.0 ± 3.6 | .001 |
| Neck length (cm) | 38.7 ± 3.4 | 39.4 ± 3.5 | 37.0 ± 2.7 | <.001 |
| Mean heart rate (beats/min) | 66.6 ± 10.0 | 68.1 ± 10.1 | 62.9 ± 9.0 | .011 |
| Mean SpO2 (%) | 93.6 ± 4.2 | 92.4 ± 4.4 | 96.7 ± 1.0 | <.001 |
| Lowest SpO2 (%) | 79.5 ± 13.3 | 76.8 ± 10.7 | 86.4 ± 16.4 | <.001 |
| Desaturation index (events/h) | 25.9 ± 28.3 | 35.1 ± 28.7 | 2.7 ± 3.1 | <.001 |
| Snore index (events/h) | 385 ± 238 | 427 ± 223 | 281 ± 244 | .002 |
| CVHR index (events/h) | 22.9 ± 20.6 | 27.9 ± 21.9 | 10.4 ± 8.4 | <.001 |
| CEI (events/h) | 13.5 ± 11.8 | 16.8 ± 12.3 | 5.4 ± 3.9 | <.001 |
| AHI (events/h) | 35.6 ± 26.9 | 46.6 ± 24.1 | 7.9 ± 3.8 | <.001 |
| Total sleep time (h) | 338.6 ± 55.5 | 342.0 ± 51.8 | 330.0 ± 63.9 | .291 |
| Sleep efficiency (%) | 85.1 ± 11.3 | 85.8 ± 9.7 | 83.5 ± 14.7 | .321 |
| Apnea counts | 94.3 ± 125.6 | 129.8 ± 132.9 | 5.4 ± 8.2 | <.001 |
| Hypopnea counts | 112.3 ± 84.5 | 141.0 ± 83.2 | 40.7 ± 22.1 | <.001 |
Values are presented as number (%) or mean ± SD. AHI = apnea-hypopnea index, BMI = body mass index, CEI = chest effort index, CVHR = cyclic variation of heart rate, OSA = obstructive sleep apnea, SpO2 = peripheral oxygen saturation.
Association among CVHR, CEI, and severity of OSA
Figure 4 illustrates the relationships among CVHR, CEI, and AHI index. The results demonstrated that both CVHR and CEI were significantly positively correlated to AHI values (Spearman rank correlation coefficient = 0.65 and 0.77, respectively; both P < .001). Table 2 shows the results of bivariate and multivariable logistic regression analyses for the association between CVHR or CEI and the risk of moderate-to-severe OSA. After adjustment for age, sex, and BMI, both CVHR and CEI were significantly associated with higher risks of moderate-to-severe OSA (odds ratio [OR] = 1.80, 95% confidence interval [CI]: 1.21–2.68 for CVHR; OR = 10.5, 95% CI: 3.81–28.92 for CEI). In addition, the CVHR was not associated with higher risks of mild, moderate, and severe OSA, even though the CEI was the following: OR = 3.89, 95% CI: 1.47–10.27. Nonetheless, both CVHR and CEI were significantly associated with higher risks of severe OSA (OR = 4.14, 95% CI: 2.16–7.92 for CVHR; OR = 12.72, 95% CI: 4.51–35.86 for CEI).
Figure 4. Relationship between CVHR index, CEI, and the severity of obstructive sleep apnea.
(A) Relationship between CVHR index and AHI. (B) Relationship between CEI and AHI. AHI = apnea-hypopnea index, CEI = chest effort index, CVHR = cyclic variation of heart rate.
Table 2.
Association between CVHR or CEI and risk of different types of OSA.
| Model | Log Value of CVHR | Log Value of CEI | ||
|---|---|---|---|---|
| OR (95% CI) | P | OR (95% CI) | P | |
| Mild, moderate, and severe OSA | ||||
| Model 1 | 1.48 (0.94–2.33) | .091 | 3.82 (1.57–9.27) | .003 |
| Model 2 | 1.53 (0.95–2.45) | .080 | 4.13 (1.59–10.74) | .004 |
| Model 3 | 1.41 (0.85–2.34) | .184 | 3.89 (1.47–10.27) | .006 |
| Moderate to severe OSA (primary outcome) | ||||
| Model 1 | 2.00 (1.37–2.93) | <.001 | 9.78 (3.99–24.01) | <.001 |
| Model 2 | 2.02 (1.38–2.94) | <.001 | 10.49 (4.09–26.89) | <.001 |
| Model 3 | 1.80 (1.21–2.68) | .004 | 10.50 (3.81–28.92) | <.001 |
| Severe OSA | ||||
| Model 1 | 5.06 (2.75–9.33) | <.001 | 11.50 (4.83–27.35) | <.001 |
| Model 2 | 4.86 (2.64–8.93) | <.001 | 13.26 (5.20–33.77) | <.001 |
| Model 3 | 4.14 (2.16–7.92) | <.001 | 12.72 (4.51–35.86) | <.001 |
Model 1 is without adjustment for any covariates. Model 2 is adjusted for age and sex. Model 3 is adjusted for age, sex, and body mass index. CEI = chest effort index, CI = confidence interval, CVHR = cyclic variation of heart rate, OR = odds ratio, OSA = obstructive sleep apnea.
Detection performance of CVHR or CEI
The performance of CVHR in detecting moderate-to-severe OSA was satisfactory (Table 3), with an AUC of 0.76 (95% CI: 0.67–0.84). The optimal cutoff according to Youden index was >21.2 events/h, with a PPV of 85.7% and an NPV of 75%. By contrast, the performance of CEI in diagnosing moderate-to-severe OSA was significantly better than CVHR (Table 3), with an AUC of 0.87 (95% CI: 0.80–0.94). The optimal cutoff was >7.1 events/h, with a PPV of 72.1% and an NPV of 85.6%. Figure 5 presents that the CEI significantly outperformed CVHR with regard to discrimination ability of moderate-to-severe OSA (ΔAUC = 0.11, 95% CI: 0.009–0.21, P = .032). However, the combination of the CVHR and CEI did not outperform CEI alone in discriminating any type of OSA (mild, moderate, or severe OSA) or moderate-to-severe OSA (Table 3). Noticeably, the performance of discrimination ability of severe OSA was excellent in terms of AUC (AUC = 0.90, 95% CI: 0.85–0.95) when combining these 2 signals. When using the optimal cutoffs of both signals, the PPVs of moderate-to-severe OSA (91.6%) and severe OSA (96%) were very high (Table 3).
Table 3.
Diagnostic properties of CVHR and CEI in detecting different severities of OSA.
| Outcome/Statistics | CVHR | CEI | CVHR + CEI |
|---|---|---|---|
| Mild, moderate or severe OSA | |||
| AUC (95% CI) | 0.71 (0.56–0.85) | 0.80 (0.64–0.96) | 0.75 (0.61–0.89) |
| Cutoffa | >18.6 | >6.6 | CVHR>18.6 and CEI>6.6 |
| Sensitivity, % (95% CI) | 47.3 (37.7–57.0) | 70.9 (61.5–79.2) | 40.9 (31.6–50.7) |
| Specificity, % (95% CI) | 100 (66.4–100) | 88.9 (51.8–99.7) | 100 (66.4–100) |
| +LR (95% CI)b | NA | 6.4 (1.0–40.7) | NA |
| −LR (95% CI)b | 0.5 (0.4–0.6) | 0.3 (0.2–0.5) | 0.6 (0.5–0.7) |
| PPV (95% CI)b | NA | 81 (40.0–96.4) | NA |
| NPV (95% CI)b | 74 (70.4–77.2) | 82.1 (76.0–86.9) | 71.7 (68.5–74.8) |
| Moderate to severe OSA (primary outcome) | |||
| AUC (95% CI) | 0.76 (0.67–0.84) | 0.87 (0.80–0.94) | 0.82 (0.75–0.90) |
| Cutoffa | >21.2 | >7.1 | CVHR > 21.2 and CEI > 7.1 |
| Sensitivity, % (95% CI) | 52.9 (41.8–63.9) | 80.0 (69.9–87.9) | 48.2 (37.3–59.3) |
| Specificity, % (95% CI) | 94.1 (80.3–99.3) | 79.4 (62.1–91.3) | 97.1 (84.7–99.9) |
| +LR (95% CI)b | 9 (2.3–35.0) | 3.9 (2.0–7.6) | 16.4 (2.3–114.5) |
| −LR (95% CI)b | 0.5 (0.4–0.6) | 0.3 (0.2–0.4) | 0.5 (0.4–0.7) |
| PPV (95% CI)b | 85.7 (60.6–95.9) | 72.1 (57.0–83.5) | 91.6 (61.0–98.7) |
| NPV (95% CI)b | 75 (70.2–79.2) | 85.6 (79.0–90.4) | 73.8 (69.4–77.7) |
| Severe OSA | |||
| AUC (95% CI) | 0.86 (0.79–0.93) | 0.87 (0.81–0.94) | 0.91 (0.85–0.96) |
| Cutoffa | >21.9 | >11 | CVHR > 21.9 and CEI > 11 |
| Sensitivity, % (95% CI) | 70.7 (57.3–81.9) | 72.4 (59.1–83.3) | 58.6 (44.9–71.4) |
| Specificity, % (95% CI) | 91.8 (81.9–97.3) | 88.5 (77.8–95.3) | 98.4 (91.2–100) |
| +LR (95% CI)b | 8.6 (3.7–20.3) | 6.3 (3.1–12.9) | 35.8 (5.1–252.8) |
| −LR (95% CI)b | 0.3 (0.2–0.5) | 0.3 (0.2–0.5) | 0.4 (0.3–0.6) |
| PPV (95% CI)b | 85.2 (71.0–93.1) | 80.8 (67.3–89.6) | 96.0 (77.1–99.4) |
| NPV (95% CI)b | 82.5 (75.8–87.6) | 82.8 (75.9–88.1) | 78.1 (72.4–82.9) |
According to Youden index. bAccording to a prevalence of 40% of moderate-to-severe OSA in patients who snore. AUC = area under the curve, CEI = chest effort index, CI = confidence interval, CVHR = cyclic variation of heart rate, +LR = positive likelihood ratio, −LR = negative likelihood ratio, NA = not applicable, NPV = negative predicted value, OSA = obstructive sleep apnea, PPV = positive predicted value.
Figure 5. Receiver operating characteristic curves of CVHR and CEI in detecting moderate-to-severe obstructive sleep apnea.
The difference of AUCs between CEI and CVHR was 0.11 (95% CI: 0.009–0.21), with P= .032. AUC = area under the curve, CEI = chest effort index, CI = confidence interval, CVHR = cyclic variation of heart rate.
Clinical use
Figure 6A presents the clinical use of Rooti Rx in the screening of moderate-to-severe OSA. In our study, when CVHR was more than 21.2 and CEI was more than 7.1 (42 patients in total), the detection rate for moderate-to-severe OSA was 97.6% (41 of 42). When one of the indices (CVHR or CEI) was above the cutoff value (38 patients in total), the detection rate for moderate-to-severe OSA was 81.6% (31 of 38). When both indices were under the cutoff value (39 patients in total), only 33.3% (13 of 39) of the patients had moderate-to-severe OSA. Overall, when CVHR or CEI was found to be above the cutoff value under the Rooti Rx examination (80 patients in total), moderate-to-severe OSA was diagnosed in 90% (72 of 80) of the patients.
Figure 6. Clinical use for diagnosis of moderate-to-severe OSA and severe OSA.
(A) Moderate-to-severe OSA. (B) Severe OSA. CEI = chest effort index, CVHR = cyclic variation of heart rate, OSA = obstructive sleep apnea.
Figure 6B illustrates the clinical use of Rooti Rx in the screening of severe OSA. In our study, when CVHR was more than 21.9 and CEI was more than 11 (35 patients in total), the detection rate for severe OSA was 97.1% (34 of 35). When one of the indices (CVHR or CEI) was above the cutoff value (25 patients in total), the detection rate for severe OSA was 60% (15 of 25). When both indices were under the cutoff value (59 patients in total), only 15.2% (9 of 59) of the patients had severe OSA. Overall, when CVHR or CEI was found to be above the cutoff value in the Rooti Rx examination (60 patients in total), 81.6% (49 of 60) of the patients were diagnosed with severe OSA.
Figure 7 presents signal waveforms of the parallel trace of cyclical heart rate variations and the chest effort signals during apnea.
Figure 7. Signal waveforms during apnea.
Note the response of SpO2 and heart rate falls a little behind air flow and chest effort in an apnea event. CEI = chest effort index, CVHR = cyclic variation of heart rate, SpO2 = peripheral oxygen saturation.
DISCUSSION
This study compared the diagnostic accuracy of detecting chest wall motion and CVHR using a single-lead wireless ECG patch with a 3-axis accelerometer for the screening of OSA. Based on the survey on 119 patients who snore, a CVHR index of >21.2 as the criterion to detect moderate-to-severe OSA yielded only 52.9% sensitivity and 94.1% specificity. By contrast, when CEI > 7.1 was used as the criterion, our chest wall motion detection algorithm identified moderate-to-severe OSA with 80% sensitivity and 79.4% specificity, which outperformed the CVHR index. These observations indicate that CVHR index and chest wall motion detection using a single-lead wireless ECG patch with 3-axis accelerometer during sleep could be used as a screening tool for moderate-to-severe OSA.
In our study, the measurements of CVHR and CEI were complementary to each other. When abnormal results were revealed in this device (CEI > 7.1 or CVHR > 21.2), 90% of patients who snore were found to have moderate-to-severe OSA. Notably, when CVHR was less than the cutoff value, 55.5% (40 of 72) of patients had moderate-to-severe OSA. This indicates that CEI evaluation may be necessary for screening moderate-to-severe OSA in this wireless single-lead ECG patch with 3-axis accelerometer.
This study has several strengths.22 First, we conducted a survey of all patients who snore with both the index test (CEI and CVHR index) and the reference standard (AHI by a standard overnight attended PSG). Second, because the CEI, CVHR index, and AHI were simultaneously recorded, no time lag or treatment effect was observed between the index test and the reference standard. Third, the in-laboratory PSG data were carefully reviewed by a second sleep technologist, which improved the accuracy of the reference standard.
Several previous studies have demonstrated a high performance of CVHR index in screening OSA both in the PhysioNet sleep apnea-ECG database (https://www.physionet.org/physiobank/database/apnea-ecg/)11–14,23 and in clinical settings.15,16,24–26 Hayano et al16 reported 92% sensitivity and 96% specificity of automated ECG detection algorithms in identifying moderate-to-severe OSA among 165 adult male workers. Although the algorithms were the same as those used in this study, their performance was much higher than that observed in our study. This difference may be attributed to the difference in the scoring systems used in the 2 studies. The scoring system for AHI has changed extensively, especially in hypopneas.20,27 Many studies have shown a higher AHI if updated 2012 AASM respiratory event criteria were used.28,29 Hayano et al used AASM 2007 scoring guidelines, and updated AASM 2017 scoring guidelines were used in our study. This difference may have had a substantial impact on our result regarding sensitivity to CVHR. Although the difference in sensitivity and specificity between CVHR and CEI for different respiratory event criteria should be investigated further, it is beyond the scope of this study.
Magnusdottir et al26 used cardiopulmonary coupling to improve CVHR performance and reported 89% sensitivity and 79% specificity in identifying moderate-to-severe OSA by cardiopulmonary coupling–CVHR algorithms among 47 patients. There are some differences between their study and ours. First, cardiopulmonary coupling–CVHR was calculated using heart rate variability and ECG-derived respiration. ECG-derived respiration referred to amplitude variations in the QRS complex because of shifts in the cardiac electrical axis during respiration and changes in thoracic impedance. It was not the direct measurement of chest wall motion, as used in this study. Second, their sample size was also smaller than ours. Third, their study cohort may not be fully representative of the general population with sleep apnea, because their cohort comprised 71.4% women and only 28.6% men. In our cohort, there were only 19.3% women and 80.7% men, which was closer to the general epidemiology of individuals with sleep-disordered breathing.30,31 Moreover, the BMI of the patients in their study was higher than the BMI of patients in our study (average BMI for both sexes was 33.9 kg/m2 in their study compared with 27.2 kg/m2 in our group). The BMI of patients in our cohort was closer to that of the general population with sleep-disordered breathing.32 These differences in study design and patient demographic data may account for the different results of the 2 studies. Notably, Magnusdottir et al also found that the sensitivity of cardiopulmonary coupling–CVHR dropped from 100% to 89% when they changed scoring guidelines from 2007 and 2017, which proved that different scoring systems strongly affect the accuracy and screening ability of a single-lead ECG test. In our study, detection of chest wall motion may have been better than CVHR for identifying moderate-to-severe OSA with updated AASM respiratory event criteria.
As this single-lead wireless ECG patch is very comfortable and could be easily hidden under clothes during the daytime, it is suitable for multiday examination. Analysis of multiday data could lead to higher sensitivity and enable a more accurate diagnosis. Sleep hygiene and circadian rhythm analysis could also be reviewed in this manner, and more information other than sleep apnea numbers could be provided to the patients.
In primary care settings, screening and assessment for OSA is a priority. Primary care providers need accurate screening tools to predict the presence of OSA.33 Although the STOP-Bang and Berlin questionnaires have the highest sensitivity (97.7% for STOP-Bang and 95.5% for Berlin questionnaires), not only is their specificity very low (3.7% for STOP-Bang and 7.4% for Berlin questionnaires) but so is their NPV (20% for STOP-Bang and 20% for Berlin questionnaires).34 Moreover, the STOP-Bang and Berlin questionnaires do not perform very well when excluding low-risk OSA.33 Our patch-type device with an ECG and a 3-axis accelerometer could serve as a supplemental screening tool to exclude low-risk OSA with higher specificity and NPV in the diagnosis of moderate-severe OSA (specificity for CVHR and CEI is 94.1% and 79.4%, respectively, and NPV for CVHR and CEI is 74% and 82.1%, respectively). When using the optimal cutoffs of both signals, the specificity and NPV for moderate-severe OSA were also very satisfactory (specificity of 97.1% and NPV of 73.8%; Table 3).
This study has several limitations. First, patients with atrial fibrillation, pacemaker implantation, ventricular tachycardia, or those who were pregnant were excluded because we could not determine CVHR in these patients. Pregnant women were excluded because arrhythmias in pregnancy are common, and analysis of CVHR is often difficult under excessive arrhythmias.35 Furthermore, CEI is difficult to interpret in women who are pregnant, particularly in the second and third trimesters, given the effect of the gravid uterus on chest wall motion. Further studies for these subgroups should be performed to evaluate the use of chest wall motion in detecting OSA. Second, the influence of medical treatment for our patients should be evaluated, although they were mostly healthy other than the snoring habit. Third, several studies reported that automated ECG detection algorithms tend to overestimate AHI in patients with central apnea and those with periodic leg movements (PLMs).11–14,36 PLMs may overlap CVHR and affect its accuracy. During PLM episodes, autonomic activations and heart rate changes could occur, resulting in CVHR.37,38 PLM-related changes in CVHR are reported to have a briefer duration and shorter cycle length than those of CVHR changes related with OSA. In our study, the proportion of abnormal PLM (PLM > 5) was 34.5% (41 of 119 patients). This may be one of the reasons for the higher performance of CEI than that of CVHR in our study. The impact of PLM on the accuracy of CVHR and CEI warrants further investigation. Moreover, because most patients (85 of 119) actually had moderate or severe sleep apnea, the results of this study may well be artificially biased toward high performance. For instance, the lack of a more substantial control group with either healthy subjects or additional patients without sleep-disordered breathing prevents an effective assessment of the proposed method in real clinical settings. On a further note, paradoxical motions of the rib cage and abdomen generally require 2 sensors, and only using chest motion may not be sufficient to differentiate an obstructed breath. Because this patch-type device is only a screening tool, false-negative results are still possible. Therefore, full PSG is still suggested for symptomatic patients who do not meet the cutoffs of this screening tool.
CONCLUSIONS
Our study validated the effectiveness of screening for OSA using a patch-type device with ECG and 3-axis accelerometer. We also proposed a new method for the quantification of chest wall motion, namely chest effort index, to identify moderate-to-severe OSA, which outperformed CVHR index in our study. A CVHR index ≥21.2 or CEI ≥ 7.1 after screening suggests moderate-to-severe OSA, and such patients should be referred for further studies. Furthermore, compared with questionnaires, the device has higher specificity and NPV; therefore, it could serve as a supplemental screening tool to exclude low-risk OSA.
DISCLOSURE STATEMENT
All authors have read and approved the manuscript. Work for this study was performed at the Shin Kong Wu Ho-Su Memorial Hospital and Shuang Ho Hospital. This study was funded by grants from Shin Kong Wu-Ho-Su Memorial Hospital (2018SKHADR017) and Shuang Ho Hospital (106TMU-SHH-17). The funders had no role in the study design, data collection and analysis, decision to publish, or preparation of the manuscript. The authors report no conflicts of interest.
ACKNOWLEDGMENTS
The authors thank the anonymous reviewers and editors for their comments.
ABBREVIATIONS
- AASM
American Academy of Sleep Medicine
- AHI
apnea-hypopnea index
- AUC
area under the curve
- BMI
body mass index
- CEI
chest effort index
- CI
confidence interval
- CVHR
cyclic variation of heart rate
- ECG
electrocardiogram
- +LR
positive likelihood ratio
- −LR
negative likelihood ratio
- NPV
negative predicted value
- OSA
obstructive sleep apnea
- PLM
periodic leg movement
- PPV
positive predicted value
- PSG
polysomnography
- SHH
Shuang Ho Hospital
- SKH
Shin Kong Wu Ho-Su Memorial Hospital
- SpO2
peripheral oxygen saturation
REFERENCES
- 1.Kapur VK, Weaver EM. Filling in the pieces of the sleep apnea-hypertension puzzle. JAMA. 2012;307(20):2197–2198. 10.1001/jama.2012.5039 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 2.Gami AS, Olson EJ, Shen WK, et al. Obstructive sleep apnea and the risk of sudden cardiac death: a longitudinal study of 10,701 adults. J Am Coll Cardiol. 2013;62(7):610–616. 10.1016/j.jacc.2013.04.080 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 3.Durgan DJ, Bryan RM. Cerebrovascular consequences of obstructive sleep apnea. J Am Heart Assoc. 2012;1(4):e000091. 10.1161/JAHA.111.000091 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 4.Reichmuth KJ, Austin D, Skatrud JB, Young T. Association of sleep apnea and type II diabetes: a population-based study. Am J Respir Crit Care Med. 2005;172(12):1590–1595. 10.1164/rccm.200504-637OC [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.Engleman HM, Douglas NJ. Sleep. 4: Sleepiness, cognitive function, and quality of life in obstructive sleep apnoea/hypopnoea syndrome. Thorax. 2004;59(7):618–622. 10.1136/thx.2003.015867 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Benjafield AV, Ayas NT, Eastwood PR, et al. Estimation of the global prevalence and burden of obstructive sleep apnoea: a literature-based analysis. Lancet Respir. Med. 2019;7(8):687–698. 10.1016/S2213-2600(19)30198-5 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 7.Kapur VK, Auckley DH, Chowdhuri S, et al. Clinical practice guideline for diagnostic testing for adult obstructive sleep apnea: an American Academy of Sleep Medicine clinical practice guideline. J Clin Sleep Med. 2017;13(3):479–504. 10.5664/jcsm.6506 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.del Campo F, Crespo A, Cerezo-Hernández A, Gutiérrez-Tobal GC, Hornero R, Álvarez D. Oximetry use in obstructive sleep apnea. Expert Rev Respir Med. 2018;12(8):665–681. 10.1080/17476348.2018.1495563 [DOI] [PubMed] [Google Scholar]
- 9.Lin SH, Branson C, Leung J, Park L, Doshi N, Auerbach SH. Oximetry as an accurate tool for identifying moderate to severe sleep apnea in patients with acute stroke. J Clin Sleep Med. 2018;14(12):2065–2073. 10.5664/jcsm.7538 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 10.Guilleminault C, Connolly S, Winkle R, Melvin K, Tilkian A. Cyclical variation of the heart rate in sleep apnoea syndrome. Mechanisms, and usefulness of 24 h electrocardiography as a screening technique. Lancet. 1984;1(8369):126–131. 10.1016/S0140-6736(84)90062-X [DOI] [PubMed] [Google Scholar]
- 11.Hayano J, Watanabe E, Saito Y, et al. Screening for obstructive sleep apnea by cyclic variation of heart rate. Circ Arrhythm Electrophysiol. 2011;4(1):64–72. 10.1161/CIRCEP.110.958009 [DOI] [PubMed] [Google Scholar]
- 12.Penzel T, McNames J, Murray A, de Chazal P, Moody G, Raymond B. Systematic comparison of different algorithms for apnoea detection based on electrocardiogram recordings. Med Biol Eng Comput. 2002;40(4):402–407. 10.1007/BF02345072 [DOI] [PubMed] [Google Scholar]
- 13.Khandoker AH, Palaniswami M, Karmakar CK. Support vector machines for automated recognition of obstructive sleep apnea syndrome from ECG recordings. IEEE Trans Inf Technol Biomed. 2009;13(1):37–48. 10.1109/TITB.2008.2004495 [DOI] [PubMed] [Google Scholar]
- 14.Mendez MO, Corthout J, Van Huffel S, et al. Automatic screening of obstructive sleep apnea from the ECG based on empirical mode decomposition and wavelet analysis. Physiol Meas. 2010;31(3):273–289. 10.1088/0967-3334/31/3/001 [DOI] [PubMed] [Google Scholar]
- 15.Roche F, Celle S, Pichot V, Barthélémy J-C, Sforza E. Analysis of the interbeat interval increment to detect obstructive sleep apnoea/hypopnoea. Eur Respir J. 2007;29(6):1206–1211. 10.1183/09031936.00042606 [DOI] [PubMed] [Google Scholar]
- 16.Hayano J, Tsukahara T, Watanabe E, et al. Accuracy of ECG-based screening for sleep-disordered breathing: a survey of all male workers in a transport company. Sleep Breath. 2013;17(1):243–251. 10.1007/s11325-012-0681-7 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Guilleminault C, Hill MW, Simmons FB, Dement WC. Obstructive sleep apnea: electromyographic and fiberoptic studies. Exp Neurol. 1978;62(1):48–67. 10.1016/0014-4886(78)90040-7 [DOI] [PubMed] [Google Scholar]
- 18.Remmers JE, deGroot WJ, Sauerland EK, Anch AM. Pathogenesis of upper airway occlusion during sleep. J Appl Physiol. 1978;44(6):931–938. 10.1152/jappl.1978.44.6.931 [DOI] [PubMed] [Google Scholar]
- 19.Staats BA, Bonekat HW, Harris CD, Offord KP. Chest wall motion in sleep apnea. Am Rev Respir Dis. 1984;130(1):59–63. 10.1164/arrd.1984.130.1.59 [DOI] [PubMed] [Google Scholar]
- 20.Berry RB, Budhiraja R, Gottlieb DJ, et al. Rules for scoring respiratory events in sleep: update of the 2007 AASM Manual for the Scoring of Sleep and Associated Events. Deliberations of the Sleep Apnea Definitions Task Force of the American Academy of Sleep Medicine. J Clin Sleep Med. 2012;8(5):597–619. 10.5664/jcsm.2172 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Jung Y, Junna MR, Mandrekar JN, Morgenthaler TI. The National Healthy Sleep Awareness Project Sleep Health Surveillance Questionnaire as an obstructive sleep apnea surveillance tool. J Clin Sleep Med. 2017;13(9):1067–1074. 10.5664/jcsm.6724 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 22.Bossuyt PM, Reitsma JB, Bruns DE, et al. Towards complete and accurate reporting of studies of diagnostic accuracy: the STARD initiative. BMJ. 2003;326(7379):41–44. 10.1136/bmj.326.7379.41 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 23.Hilmisson H, Lange N, Duntley SP. Sleep apnea detection: accuracy of using automated ECG analysis compared to manually scored polysomnography (apnea hypopnea index). Sleep Breath. 2019;23(1):125–133. 10.1007/s11325-018-1672-0 [DOI] [PubMed] [Google Scholar]
- 24.Poupard L, Court-Fortune I, Pichot V, Chouchou F, Barthélémy J-C, Roche F. Use of high-frequency peak in spectral analysis of heart rate increment to improve screening of obstructive sleep apnoea. Sleep Breath. 2011;15(4):837–843. 10.1007/s11325-010-0446-0 [DOI] [PubMed] [Google Scholar]
- 25.Zamarrón C, Hornero R, del Campo F, Abásolo D, Alvarez D. Heart rate regularity analysis obtained from pulse oximetric recordings in the diagnosis of obstructive sleep apnea. Sleep Breath. 2006;10(2):83–89. 10.1007/s11325-005-0049-3 [DOI] [PubMed] [Google Scholar]
- 26.Magnusdottir S, Hilmisson H. Ambulatory screening tool for sleep apnea: analyzing a single-lead electrocardiogram signal (ECG). Sleep Breath. 2018;22(2):421–429. 10.1007/s11325-017-1566-6 [DOI] [PubMed] [Google Scholar]
- 27. Iber C, Ancoli-Israel S, Chesson AL Jr, Quan SF; for the American Academy of Sleep Medicine. The AASM Manual for the Scoring of Sleep and Associated Events: Rules, Terminology and Technical Specifications. 1st ed. Westchester, IL: American Academy of Sleep Medicine, 2007. [Google Scholar]
- 28.BaHammam AS, Obeidat A, Barataman K, Bahammam SA, Olaish AH, Sharif MM. A comparison between the AASM 2012 and 2007 definitions for detecting hypopnea. Sleep Breath. 2014;18(4):767–773. 10.1007/s11325-014-0939-3 [DOI] [PubMed] [Google Scholar]
- 29.Duce B, Milosavljevic J, Hukins C. The 2012 AASM Respiratory Event criteria increase the incidence of hypopneas in an adult sleep center population. J Clin Sleep Med. 2015;11(12):1425–1431. 10.5664/jcsm.5280 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 30.Hsu YS, Hsu WC, Ko JY, Yeh TH, Lee CH, Kang KT. Readmission after adult uvulopalatopharyngoplasty: a population-based inpatient cohort study in Taiwan. Otolaryngol Head Neck Surg. 2019;160(3):559–566. 10.1177/0194599818813337 [DOI] [PubMed] [Google Scholar]
- 31.Hsu YS, Hsu WC, Ko JY, Yeh TH, Lee CH, Kang KT. Incidence of multilevel surgical procedures and readmissions in uvulopalatopharyngoplasty in Taiwan. JAMA Otolaryngol Head Neck Surg. 2019;145(9):803. 10.1001/jamaoto.2019.1656 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Liu WT, Wu HT, Juang JN, et al. Prediction of the severity of obstructive sleep apnea by anthropometric features via support vector machine. PLoS One. 2017;12(5):e0176991. 10.1371/journal.pone.0176991 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Miller JN, Berger AM. Screening and assessment for obstructive sleep apnea in primary care. Sleep Med Rev. 2016;29:41–51. 10.1016/j.smrv.2015.09.005 [DOI] [PubMed] [Google Scholar]
- 34.El-Sayed IH. Comparison of four sleep questionnaires for screening obstructive sleep apnea. Egypt J Chest Dis Tuberc. 2012;61(4):433–441. 10.1016/j.ejcdt.2012.07.003 [DOI] [Google Scholar]
- 35.Adamson DL, Nelson-Piercy C. Managing palpitations and arrhythmias during pregnancy. Heart. 2007;93(12):1630–1636. 10.1136/hrt.2006.098822 [DOI] [PMC free article] [PubMed] [Google Scholar]
- 36.Varon C, Caicedo A, Testelmans D, Buyse B, Van Huffel S. A novel algorithm for the automatic detection of sleep apnea from single-lead ECG. IEEE Trans Biomed Eng. 2015;62(9):2269–2278. 10.1109/TBME.2015.2422378 [DOI] [PubMed] [Google Scholar]
- 37.Sforza E, Pichot V, Barthelemy JC, Haba-Rubio J, Roche F. Cardiovascular variability during periodic leg movements: a spectral analysis approach. Clin Neurophysiol. 2005;116(5):1096–1104. 10.1016/j.clinph.2004.12.018 [DOI] [PubMed] [Google Scholar]
- 38.Guggisberg AG, Hess CW, Mathis J. The significance of the sympathetic nervous system in the pathophysiology of periodic leg movements in sleep. Sleep. 2007;30(6):755–766. 10.1093/sleep/30.6.755 [DOI] [PMC free article] [PubMed] [Google Scholar]