Abstract
Background
Sleep-related hypoventilation, particularly during rapid eye movement (REM) sleep, has been linked to pulmonary hypertension and recurrent exacerbations in individuals with advanced chronic respiratory or neuromuscular diseases. Overnight pulse oximetry (OPO) serves as a valuable screening tool to depict episodic oxygen desaturation resulting from sleep-related hypoventilation. However, differentiating nocturnal desaturation caused by physical activity from that attributable to sleep-related hypoventilation remains clinically challenging. This study aimed to determine whether the integration of accelerometer data with OPO readings can assist in distinguishing exertional nocturnal desaturation from desaturation due to sleep-related hypoventilation.
Methods
Between July 2021 and December 2022, a prospective enrollment was conducted among consecutive individuals with stable chronic respiratory disorders who reported worsening exertional dyspnea. Participants underwent overnight monitoring involving transcutaneous carbon dioxide pressure (PtcCO₂) and pulse oximetry integrated with accelerometer sensors. The number of exertion-associated desaturation events was compared between participant self-reports and acceleration-derived data. Additionally, the diagnostic accuracy of accelerometer-integrated pulse oximetry for detecting episodic nocturnal hypercapnia was assessed using PtcCO₂ monitoring as the reference standard. The primary endpoint was the patient-level diagnostic accuracy of accelerometer-integrated pulse oximetry in detecting episodic nocturnal hypercapnia.
Results
Thirty-six individuals were enrolled, with a median age of 78.0 (IQR: 72.0–82.0) years and a mean daytime arterial carbon dioxide pressure (PaCO₂) of 42.4 ± 6.9 mmHg. Of the 89 desaturation events observed, 56 (62.9%) were identified as exertion-related using accelerometer data, including 19 events (21.3%) that were not self-reported. The device demonstrated a sensitivity of 100% (95% CI: 79.6–100%) and a specificity of 75.7% (95% CI: 64.8–84.0%) in identifying episodic nocturnal hypoxia associated with hypercapnia. At the patient level, sensitivity and specificity were 100% (95% CI: 100–100%) and 73.1% (95% CI: 53.9–86.3%), respectively. At the event level, sensitivity and specificity were 100% (95% CI: 79.6–100%) and 75.7% (95% CI: 64.8–84.2%), respectively.
Conclusion
Among individuals with suspected sleep-related breathing disorders, accelerometer-integrated pulse oximetry may serve as a valuable tool to distinguish nocturnal desaturation episodes caused by exertion from those due to sleep-related hypoventilation. These findings suggest that accelerometer-integrated pulse oximetry could offer a feasible screening method for detecting sleep-related hypoventilation in outpatient settings lacking access to PtcCO2 monitoring.
Supplementary Information
The online version contains supplementary material available at 10.1186/s12931-025-03477-2.
Keywords: Nocturnal exertional desaturation, Episodic nocturnal hypercapnia, Pulse oximetry, Sleep-related hypoventilation, COPD, Transcutaneous carbon dioxide monitoring, Diagnostic accuracy
Background
Sleep-related hypoventilation frequently occurs in individuals with chronic obstructive pulmonary disease (COPD), neuromuscular disease (NMD), and chest wall disorders, with a reported prevalence ranging from 9–43% [1–3]. This condition is attributed to reduced physiological ventilation during sleep and progressive deterioration of respiratory function due to the underlying disease [4]. Sleep-related hypoventilation initially manifests during rapid eye movement (REM) sleep and subsequently progresses into non-REM sleep [5]. During REM sleep, the activity of the intercostal and accessory respiratory muscles decreases, while hypoxic and hypercapnic ventilatory responses are blunted [6], resulting in episodic reductions in minute ventilation, oxygen desaturation, and transient elevations in arterial partial pressure of carbon dioxide (PaCO₂) [3]. Recurrent nocturnal and early morning elevations in PaCO₂ due to sleep-related hypoventilation may lead to sustained daytime hypercapnia [3], a condition associated with an increased risk of exacerbations and mortality [7, 8]. Although evaluation of sleep-related hypoventilation is crucial, its clinical daytime presentation is often nonspecific, and standard respiratory function tests inadequately predict nocturnal hypoventilation [5]. Therefore, overnight monitoring of respiratory parameters and carbon dioxide levels is recommended for assessing high-risk populations [5].
Recently, transcutaneous carbon dioxide pressure (PtcCO₂) monitoring has emerged as a noninvasive modality for evaluating hypoventilation during sleep [9]. Episodic nocturnal hypercapnia (eNH), primarily corresponding to REM sleep-related hypoventilation, has previously been defined based on PtcCO₂ monitoring [10, 11]. In individuals with COPD, eNH has been associated with pulmonary hypertension and a history of frequent exacerbations [10, 11]. Naito et al. also reported that eNH has been linked to future use of home noninvasive ventilation (NIV) and mortality among individuals with NMD [12]. Furthermore, the use of NIV specifically targeting eNH has demonstrated a reduction in exacerbation frequency in patients with COPD [10]. Consequently, PtcCO₂ monitoring serves as a valuable tool for the assessment of sleep-related hypoventilation and the initiation of NIV. However, overnight PtcCO₂ monitoring remains technically complex, cost-prohibitive, and is not widely accessible in North America [13, 14] and Japan. In Japan, overnight PtcCO₂ monitoring is typically performed during hospitalization, rather than in the home setting. Therefore, the use of PtcCO₂ monitoring as a screening modality for large populations suspected of sleep-related hypoventilation remains limited.
Overnight pulse oximetry (OPO) represents a cost-effective, safe, and reliable modality for assessing cardiorespiratory function in outpatient settings [15–17]. It is particularly known for its simplicity and accuracy in detecting sleep-related breathing disorders [18–20]. OPO may also facilitate the detection of sleep-related hypoventilation, manifested as episodic nocturnal desaturation, without requiring overnight PtcCO₂ monitoring. Nocturnal oxygen desaturation has been previously documented in individuals with COPD, interstitial lung disease (ILD), restrictive thoracic disease, and NMD [21–25]. In REM sleep-related hypoventilation, prolonged episodes of oxygen desaturation lasting 5–30 min and recurring every 90–120 min throughout the night are commonly observed [13, 26]. This episodic nocturnal desaturation has been associated with adverse clinical outcomes [21]. However, distinguishing whether episodic nocturnal desaturations arise from exertional activity (e.g., nocturnal ambulation) or from sleep-related hypoventilation remains challenging based solely on OPO data. This diagnostic challenge stems from the similarity in desaturation waveforms observed in REM sleep-related hypoventilation and exertional hypoxia.
Recent evidence suggests that wrist-worn accelerometers may serve as physiological markers of nocturnal exertional events [27]. It was hypothesized that accelerometer-integrated pulse oximetry could distinguish between exertion-induced nocturnal desaturation and that resulting from hypoventilation during sleep.
Accordingly, this study aimed to evaluate whether overnight accelerometer-integrated pulse oximetry could effectively distinguish between hypoxia secondary to exertion and that due to sleep-related hypoventilation. The diagnostic accuracy of accelerometer-integrated pulse oximetry for detecting eNH was prospectively validated against PtcCO₂ monitoring. This approach may facilitate broader implementation of outpatient screening for sleep-related hypoventilation, especially in resource-constrained settings.
Methods
Patients
A prospective observational study was conducted to evaluate the diagnostic accuracy of accelerometer-integrated pulse oximetry in detecting episodic nocturnal hypercapnia among individuals with suspected sleep-related breathing disorders. Consecutive, clinically stable patients admitted to the Medical Research Institute, KITANO HOSPITAL, for evaluation of suspected exertional dyspnea worsening were enrolled between July 2021 and December 2022. The inclusion criteria for this study were as follows: (1) age ≥ 20 years, (2) medical history of chronic pulmonary disease, and (3) absence of long-term oxygen therapy use. Exclusion criteria were: (1) a pre-existing diagnosis of obstructive sleep apnea, (2) monitoring duration for pulse oximetry and PtcCO₂ of less than four hours, and (3) sustained nocturnal SpO₂ levels below 85%. Sample size determination was guided by feasibility considerations and the availability of eligible participants during the study period.
Measurements and data collection
Sociodemographic, clinical, and laboratory data were extracted from electronic medical records. Body mass index (BMI) was calculated as weight in kilograms divided by height in meters squared (kg/m²). Pulmonary function tests were performed by trained personnel in accordance with the American Thoracic Society and European Respiratory Society guidelines [28]. Arterial blood gas analysis was performed during the day in the supine position using the RAPIDLAB 1200 System (Siemens Healthcare Diagnostics Incorporated, Tarrytown, NY, USA). All arterial blood gas samples were obtained while patients were breathing ambient air. Overnight evaluations without supplemental oxygen administration were performed using PtcCO2 monitoring with the SenTec Digital Monitor (SenTec, Therwil, Switzerland) and accelerometer-integrated pulse oximetry with the PULSOX-500i (KONICA MINOLTA, INC, Tokyo, Japan). During sleep, each patient wore the accelerometer-integrated pulse oximetry device on one wrist and a finger-mounted probe measuring SpO₂. Patients were instructed to press the event marker or manually record their behaviors during the night. Pulse oximetry data were analyzed using the DS-500 (KONICA MINOLTA, INC) (Fig. 1). Two board-certified respiratory specialists, each with over ten years of clinical experience, independently evaluated the PtcCO₂ and pulse oximetry data.
Fig. 1.
Episodic nocturnal hypercapnia and desaturation event due to exertion. Notes: Events were assessed using overnight accelerometer-integrated pulse oximetry and PtcCO₂ monitoring. Grey arrows indicate episodic desaturation events. Black arrowheads indicate episodic nocturnal hypercapnia, characterized by desaturation events and concurrent increases in PtcCO₂ without acceleration data. When desaturation events are accompanied by acceleration, they are attributed to exertion. Abbreviations: PtcCO2: transcutaneous carbon dioxide pressure; SpO2: oxygen saturation
Definition of oxygen desaturation event, possible eNH, and eNH
The most widely accepted definition of exercise-induced desaturation is a reduction in oxygen saturation greater than 4% from baseline and/or a SpO₂ below 90% during the 6-min walk test [29–31]. In accordance with prior research, “oxygen desaturation events” in this study were defined as a decrease in SpO₂ of ≥ 4% from baseline and a sustained SpO₂ <90% for at least 5 min, measured using accelerometer-integrated pulse oximetry. Among all desaturation events, “possible eNH” refers to a desaturation episode that occurs without concurrent exertional acceleration. This indicates that the desaturation may be related to sleep-related hypoventilation. Exertion was identified using acceleration data, as described in the supplementary materials. “eNH” was defined as a ≥ 5 mmHg increase from baseline PtcCO₂, accompanied by SpO₂ <90% for ≥ 5 min continuously, occurring at least once during the night, according to previously established criteria (Fig. 1) [10, 11]. “Subclinical eNH” was defined as a sustained increase in PtcCO₂ of ≥ 3 mmHg but < 5 mmHg from baseline for at least 5 min, accompanied by oxygen desaturation below 90% (Additional Fig. 1).
Event-level and patient-level analyses of oxygen desaturation were conducted in this study. In the event-level analysis, each oxygen desaturation event was assessed to determine whether it was due to sleep-related hypoventilation. Conversely, the patient-level analysis involved identifying the presence of sleep-related hypoventilation-induced desaturation on an individual basis. Patients were classified as having eNH if at least one eNH event was identified during overnight monitoring.
Overnight accelerometer-integrated pulse oximetry data were used to extract mean SpO₂, time spent with SpO₂ <90% (T90), 3% oxygen desaturation index (ODI), and mean pulse rate. Nocturnal mean and maximal PtcCO₂ values were also obtained from overnight PtcCO₂ monitoring. To facilitate analysis using modified SpO₂ waveforms, desaturation dip intervals were identified, and the waveform was reconstructed accordingly. Detailed methodologies for processing the modified SpO₂ waveforms are provided in the additional materials. Analysis using the modified SpO₂ waveforms demonstrated a flattened sawtooth pattern, typically observed in cases of sleep apnea or signal artifact.
Study endpoint
The primary endpoint was the patient-level diagnostic performance (sensitivity and specificity) of accelerometer-integrated pulse oximetry for detecting eNH, with PtcCO₂ monitoring serving as the reference. Secondary endpoints included event-level diagnostic accuracy, diagnostic performance using the modified SpO₂ waveform, and positive and negative predictive values.
Ethics and statistics
The study was conducted in compliance with the ethical guidelines of the Japanese Ministry of Health, Labor, and Welfare and received approval from the Institutional Review Board of the Medical Research Institute, KITANO HOSPITAL Ethics Committee (Ethics Board approval number: P2102009). All participants provided written informed consent prior to study enrollment. In addition, the confidentiality of personal information was maintained in accordance with established ethical guidelines.
The Shapiro–Wilk test was used to assess the normality of data distribution. Parametric variables are presented as mean ± standard deviation, whereas nonparametric variables are reported as median (interquartile range [IQR]). Group comparisons were performed using Student’s t-test for normally distributed data with equal variances, Welch’s t-test for data with unequal variances, and the Mann–Whitney U test for non-normally distributed data. Diagnostic performance was assessed using sensitivity, specificity, and positive and negative predictive values, with 95% confidence intervals (CIs) at both the patient and event levels. Receiver operating characteristic (ROC) curves were constructed to calculate the area under the curve (AUC, 95% CI) for discriminating confirmed eNH. Agreement between accelerometer-integrated pulse oximetry and PtcCO₂-based classification at the patient level was assessed using Cohen’s κ coefficient. Primary diagnostic analyses were pre-specified, while subgroup and waveform-based analyses were exploratory. A P-value of less than 0.05 was considered statistically significant for all analyses. All statistical analyses were performed using IBM SPSS Statistics for Windows version 25 (IBM Corp., Armonk, NY, USA).
Results
This section summarizes the clinical characteristics of the study cohort and evaluates the diagnostic efficacy of accelerometer-integrated pulse oximetry for identifying episodic nocturnal hypercapnia. Figure 2 illustrates the patient selection flowchart used in this study. A total of 36 patients (30 men and 6 women) met the inclusion criteria. Table 1 presents patient characteristics, including a median age of 78.0 (72.0–82.0) years and a low BMI of 21.5 (16.5–24.2) kg/m². COPD was diagnosed in 66.7% of the patients. The mean forced expiratory volume in 1 s (FEV₁, % of predicted value) and FEV₁/forced vital capacity (FVC) ratio were 44.2 ± 17.8% and 58.6 (27.3–78.6), respectively, in the overall patient group. Daytime PaO₂, PaCO₂, and bicarbonate (HCO₃⁻) levels were 75.9 (60.3–85.5) mmHg, 42.4 ± 6.9 mmHg, and 26.5 ± 3.6 mmol/L, respectively.
Fig. 2.
Patient selection flow diagram. Abbreviations: PtcCO2: transcutaneous carbon dioxide pressure; SpO2: oxygen saturation; eNH: episodic nocturnal hypercapnia
Table 1.
Patient characteristics
| Variable | All (n = 36) |
|---|---|
| Age (y) [IQR] | 78.0 (72.0–82.0) |
| Male sex (%) | 79.1 |
| Body mass index (kg/m2) [IQR] | 21.5 (16.5–24.2) |
| Underlying disease | |
| COPD (%) | 66.7 |
| Interstitial lung disease (%) | 19.4 |
| Restrictive lung disease (%) | 5.6 |
| Others (%) | 8.3 |
| FEV1 (% of predicted value) | 44.2 ± 17.8 |
| FVC (% of predicted value) | 70.5 ± 22.9 |
| FEV1/FVC ratio [IQR] | 58.6 (27.3–78.6) |
| pH | 7.41 ± 0.03 |
| PaCO2 (mmHg) | 42.4 ± 6.9 |
| PaO2 (mmHg) [IQR] | 75.9 (60.3–85.5) |
| Bicarbonate (mmol/L) | 26.5 ± 3.6 |
Parametric data are presented as mean ± standard deviation and nonparametric data as median (interquartile [IQR]). Other data are presented as proportions (%)
Abbreviations: FEV1 forced expiratory volume in 1 s, FVC forced vital capacity, PaCO2, arterial carbon dioxide pressure, PaO2, arterial oxygen pressure, COPD chronic obstructive pulmonary disease
Overnight pulse oximetry and PtcCO2 monitoring analysis
Overnight accelerometer-integrated pulse oximetry and PtcCO₂ monitoring results are summarized in Table 2. The mean values of SpO₂, 3% ODI, and T90 were 92.4% (90.8–93.9%), 8.8/h (4.2–18.6/h), and 11.1% (1.4–31.1%), respectively. The mean and maximum values of PtcCO₂ were 41.2 (39.7–45.8) mmHg and 45.9 (43.5–55.9) mmHg, respectively.
Table 2.
Data on accelerometer-integrated pulse oximetry and transcutaneous carbon dioxide pressure monitoring
| Variable | All (n = 36) |
|---|---|
| Accelerometer-integrated pulse oximetry | |
| Recording time (minutes) [IQR] | 469 (432–494) |
| Mean SpO2 (%) [IQR] | 92.4 (90.8–93.9) |
| Lowest SpO2 (%) [IQR] | 81.3 (75.3–85.7) |
| 3% ODI (/ hour) [IQR] | 8.8 (4.2–18.6) |
| T88 (%) [IQR] | 3.5 (0.5–14.8) |
| T90 (%) [IQR] | 11.1 (1.4–31.1) |
| T90 > 30% (persons) | 9 |
| Mean pulse rate (times / mins) | 71.9 ± 8.8 |
| PtcCO2 monitoring | |
| Recording time (minutes) [IQR] | 7:43 (7:12–8:14) |
| Mean PtcCO2 (mmHg) [IQR] | 41.2 (39.7–45.8) |
| Maximum PtcCO2 (mmHg) [IQR] | 45.9 (43.5–55.9) |
| PtcCO2 > 50 mmHg (%) [IQR] | 0.0 (0.0–7.5) |
Parametric data are presented as mean ± standard deviation and nonparametric data as median (interquartile [IQR]). Other data are presented as proportions (%)
Abbreviations: SpO2 oxygen saturation, ODI oxygen desaturation index, T88 percent of time spent with oxygen saturation below 88%, T90 percent of time spent with oxygen saturation below 90%, PtcCO2 transcutaneous carbon dioxide pressure
Differentiation between episodic nocturnal desaturation events with and without exertion in event-level analysis
In the event-level analysis, the accelerometer-integrated pulse oximeter revealed a total of 89 oxygen desaturation events (2.4 ± 1.7 events per patient). When pulse oximeter data were analyzed in conjunction with accelerometer readings, 56 of the 89 events (62.9%) were classified as exertion-related. Among the 56 exertional events, 37 (66.1%) were attributed to patient-reported exertion. However, 19 events (32.9%) showed exertional patterns based on accelerometer data, although patient-reported exertion was not confirmed.
Diagnostic accuracy of possible eNH on an accelerometer-integrated pulse oximetry in event-level analysis
In the event-level analysis, 33 of the 89 events (37.1%) were identified as positive for possible eNH based on accelerometer-integrated pulse oximetry. Among these 33 events, 15 were confirmed as true positive cases of eNH by PtcCO₂ monitoring, while 18 did not meet the established diagnostic criteria (Additional Table 1A). The accelerometer-integrated pulse oximetry demonstrated a sensitivity of 100% (95% CI: 79.6–100%) and a specificity of 75.7% (95% CI: 64.8–84.2%) in detecting suspected eNH. The device yielded a positive predictive value of 45.5% (95% CI: 29.8–62.0%) and a negative predictive value of 100% (95% CI: 93.6–100%) in identifying suspected eNH. Furthermore, diagnostic discrimination for confirmed eNH was strong, with an AUC of 0.867 (95% CI: 0.794–0.938, P < 0.001) (Additional Fig. 2).
Diagnostic accuracy of possible eNH using modified SpO2 waveform on an accelerometer-integrated pulse oximetry in event-level analysis
Upon analysis with modified SpO₂ waveforms, the number of episodic desaturation events decreased from 89 to 74, attributable to the flattening of the sawtooth pattern typically observed in transient sleep apnea or motion artifacts (Additional Fig. 3). When pulse oximetry data incorporating modified SpO₂ waveforms were analyzed, the number of exertion-related episodic desaturation events decreased from 56 to 48 events. In contrast, possible eNH increased from 26 events, as determined by the original SpO₂ waveform, to 33 events.
Among the 26 events, 15 were diagnosed as true positive cases of eNH, whereas the remaining 11 did not meet the diagnostic criteria for eNH. The sensitivity and specificity of accelerometer-integrated pulse oximetry for detecting eNH were 100% (95% CI: 72.2–100%) and 81.4% (95% CI: 69.9–89.3%), respectively.
Differentiation between episodic nocturnal desaturation events with and without exertion in patient-level analysis
In the patient-level analysis, 31 out of 36 patients experienced at least one episodic nocturnal desaturation event (Fig. 2). When pulse oximetry data were evaluated alongside acceleration data, 19 of these patients were identified as having exclusively exertion-induced episodic desaturation events. However, 54.8% (17 out of 31) were classified as positive for possible eNH (Fig. 2). Of these 17 patients, 10 were confirmed as true positives for eNH, while the remaining 7 did not meet the diagnostic criteria (Fig. 2, Additional Table 1B).
Diagnostic accuracy of possible eNH on an accelerometer-integrated pulse oximetry in patient-level analysis
Accelerometer-integrated pulse oximetry demonstrated a sensitivity of 100% (95% CI: 100–100%) and a specificity of 73.1% (95% CI: 53.9–86.3%) for detecting eNH. It also yielded a positive predictive value of 58.8% (95% CI: 36.0–78.4%) and a negative predictive value of 100% (95% CI: 83.2–100%) for detecting eNH. Furthermore, at the patient level, accelerometer-integrated pulse oximetry demonstrated good diagnostic discrimination for confirmed eNH (area under the receiver operating characteristic curve [AUC] = 0.865, 95% CI 0.749–0.981, P < 0.001; Additional Fig. 2). The agreement between accelerometer-integrated pulse oximetry and PtcCO₂-based classification for eNH detection was moderate (Cohen’s κ = 0.60, 95% CI 0.36–0.84).
Among the seven patients who fulfilled the criteria for possible eNH but did not meet the definitive diagnostic threshold, five were identified as having subclinical eNH (Additional Table 1C). Upon the inclusion of subclinical eNH in the diagnostic criteria, the sensitivity and specificity of accelerometer-integrated pulse oximetry increased to 100% (95% CI: 79.6–100%) and 90.5% (95% CI: 71.1–97.3%), respectively.
Comparison based on episodic nocturnal hypercapnia
Table 3 presents patient characteristics and data obtained from overnight monitoring using accelerometer-integrated pulse oximetry and PtcCO₂, comparing results between patients with and without eNH. Figures 3 and 4 present comparative analyses of the key parameters associated with eNH and possible eNH, as identified through various assessment methods. No statistically significant differences were observed in PaO₂, BMI, and mean SpO₂. However, the group of patients diagnosed with eNH exhibited significantly elevated levels of PaCO₂ (Fig. 3A), HCO₃⁻, and mean PtcCO₂ (Fig. 4A).
Table 3.
Characteristics of patients with and without episodic nocturnal hypercapnia
| Variable | Patients with eNH (n = 10) |
Patients without eNH (n = 26) |
P |
|---|---|---|---|
| PaO2 (mmHg) [IQR] | 74.9 (66.7–90.7) | 77.4 (60.3–85.5) | 0.804 |
| PaCO2 (mmHg) | 48.6 ± 5.6 | 38.5 ± 4.5 | < 0.001* |
| Bicarbonate (mmol/L) | 29.4 ± 3.1 | 24.8 ± 2.8 | < 0.001* |
| Body mass index (kg/m2) [IQR] | 16.6 (14.1–23.7) | 21.5 (18.0–24.1) | 0.198 |
| Mean SpO2 (%) [IQR] | 90.6 ± 3.0 | 92.6 ± 2.8 | 0.113 |
| 3% ODI (/ hour) [IQR] | 14.8 (5.7–26.2) | 7.5 (3.1–14.7) | 0.155 |
| T88 (%) [IQR] | 9.8 (1.4–32.7) | 3.9 (0.2–7.5) | 0.053 |
| T90 (%) [IQR] | 18.6 (3.8–67.6) | 6.5 (1.1–22.7) | 0.145 |
| Mean PtcCO2 (mmHg) [IQR] | 49.8 (41.6–56.0) | 40.3 (36.4–45.1) | 0.001* |
| Maximum PtcCO2 (mmHg) [IQR] | 56.4 (47.7–61.1) | 44.0 (40.4–50.2) | 0.001* |
Parametric data are presented as mean ± standard deviation and nonparametric data as median (interquartile [IQR]). Other data are presented as proportions (%). *P <0.05 was considered statistically significant
Abbreviations: SpO2 oxygen saturation, ODI oxygen desaturation index, T88 percent of time spent with oxygen saturation below 88%, T90 percent of time spent with oxygen saturation below 90%, PtcCO2 transcutaneous carbon dioxide pressure
Fig. 3.
Comparison of daytime arterial carbon dioxide pressure among three groups. Notes: Groups were classified based on the presence or absence of episodic nocturnal hypercapnia and identification via modified SpO₂ waveforms A: Daytime PaCO₂ compared between patients with and without episodic nocturnal hypercapnia B: Comparison of PaCO₂ based on possible episodic nocturnal hypercapnia C: Comparison of PaCO₂ among groups stratified by possible episodic nocturnal hypercapnia identified through modified SpO₂ waveforms. Abbreviations: PaCO2: arterial carbon dioxide pressure; eNH: episodic nocturnal hypercapnia
Fig. 4.
Comparison of transcutaneous carbon dioxide pressure among the three groups. Notes: A: Comparison of PtcCO₂ in patients with and without episodic nocturnal hypercapnia B: PtcCO₂ compared between those with and without possible episodic nocturnal hypercapnia C: PtcCO₂ compared among groups defined by possible episodic nocturnal hypercapnia identified using modified SpO₂ waveforms. Abbreviations: PtcCO2: transcutaneous carbon dioxide pressure; eNH: episodic nocturnal hypercapnia
Comparison based on possible episodic nocturnal hypercapnia
Table 4 presents patient characteristics and overnight monitoring using accelerometer-integrated pulse oximetry and PtcCO₂, comparing results between patients with possible eNH and those without. No statistically significant differences were observed in PaO₂, mean SpO₂, and mean PtcCO₂. However, patients classified in the possible eNH group demonstrated significantly elevated levels of PaCO₂ (Fig. 3B). In analyses incorporating the modified SpO₂ waveform, patients with possible eNH exhibited significantly elevated levels of daytime PaCO₂, HCO₃⁻, mean PtcCO₂, and maximum PtcCO₂ compared to those without possible eNH (Figs. 3C and 4C, Additional Table 1).
Table 4.
Characteristics of patients with and without possible episodic nocturnal hypercapnia
| Variable | Patients with possible eNH (n = 17) |
Patients without possible eNH (n = 19) |
p |
|---|---|---|---|
| PaO2 (mmHg) [IQR] | 77.1 (60.4–87.8) | 74.7 (62.4–82.3) | 0.682 |
| PaCO2 (mmHg) | 44.3 ± 7.7 | 39.2 ± 4.5 | 0.041* |
| Bicarbonate (mmol/L) | 27.4 ± 3.9 | 25.1 ± 2.8 | 0.077 |
| Body mass index (kg/m2) [IQR] | 17.3 (15.1–24.2) | 21.7 (19.7–24.2) | 0.184 |
| Mean SpO2 (%) [IQR] | 92.8 (89.8–93.5) | 92.3 (91.1–94.6) | 0.165 |
| 3% ODI (/ hour) [IQR] | 10.2 (3.3–13.9) | 7.5 (3.4–13.9) | 0.616 |
| T88 (%) [IQR] | 3.2 (0.9–21.8) | 4.6 (0.1–7.0) | 0.138 |
| T90 (%) [IQR] | 9.3 (2.4–52.2) | 11.2 (0.6–21.0) | 0.330 |
| Mean PtcCO2 (mmHg) [IQR] | 43.4 (40.5–54.7) | 40.3 (38.0–45.5) | 0.093 |
| Maximum PtcCO2 (mmHg) [IQR] | 49.3 (44.5–60.4) | 44.4 (41.9–45.5) | 0.257 |
Parametric data are presented as mean ± standard deviation and nonparametric data as median (interquartile [IQR]). Other data are presented as proportions (%). *P <0.05 was considered statistically significant
Abbreviations: SpO2 oxygen saturation, ODI oxygen desaturation index, T88 percent of time spent with oxygen saturation below 88%, T90 percent of time spent with oxygen saturation below 90%, PtcCO2 transcutaneous carbon dioxide pressure
Discussion
This study evaluated the utility of nocturnal accelerometer-integrated pulse oximetry for episodic nocturnal hypercapnia in individuals with chronic respiratory diseases. Among 89 identified episodic desaturation events, accelerometer-integrated pulse oximetry excluded 56 (62.9%) that were attributable to exertion. This substantially improved the ability to distinguish exertional desaturation from hypoventilation-related events. In event-specific analysis, the sensitivity and specificity of accelerometer-integrated pulse oximetry in detecting eNH were 100% and 75.7%, respectively. Accelerometer-integrated pulse oximetry is effective in distinguishing episodic nocturnal desaturation events with hypercapnia due to hypoventilation from exertional desaturation events and in screening for sleep-related hypoventilation.
In OPO, distinguishing between desaturation waveforms resulting from sleep-related hypoventilation and those due to physical exertion is challenging, as their patterns are often similar. Typically, self-reported behavior logs are employed to identify exertion-induced desaturation events. However, the findings of the present study indicated that self-reported behavior logs may lack reliability. In 19 of 56 exertional desaturation events (33.9%), patient-reported exertion was not corroborated by objective data. Previous research similarly identified discrepancies between self-reported physical activity and accelerometer data, although the assessment methods differed from those used in the current study [32]. As a result, patients may not be able to complete the self-report accurately. These inaccuracies in self-documented behavior record sheets may affect the interpretation of previous clinical trials on nocturnal oxygen therapy [33]. The clinical utility of oxygen therapy for nocturnal hypoxemia should be re-assessed following an accurate differentiation between exertion-related and hypoventilation-related desaturation.
Daytime hypercapnia is associated with frequent exacerbations and poor prognosis in patients with COPD [7, 8]. Hypercapnic respiratory failure manifests earlier during sleep than during wakefulness [5, 34]. This is because sleep has adverse effects on breathing, including disturbances in respiratory control, respiratory muscle function, and lung mechanics [5]. Sleep-related alterations in respiratory control involve decreased chemoreceptor sensitivity, diminished ventilatory responses, and reduced function of accessory respiratory muscles, particularly during REM sleep [5, 35]. eNH has been reported to be associated with elevated daytime PaCO₂, pulmonary hypertension, and frequent exacerbations in advanced COPD. In contrast, in neuromuscular disorders, eNH is linked to greater use of home NIV and an increased risk of mortality [12]. Thus, eNH is considered an important surrogate marker of poor outcomes that can be assessed noninvasively through PtcCO₂ monitoring.
However, in Japan, overnight PtcCO₂ monitoring is complex, costly, and primarily conducted during hospitalization because it requires specialized equipment and trained personnel [13, 14]. Compared with outpatient OPO, inpatient PtcCO₂ monitoring in Japan incurs more than six times the cost. Sleep-related hypoventilation affects a substantial proportion of patients with advanced chronic respiratory disease, with prior studies reporting a prevalence of 43–51% in hypercapnic COPD and neuromuscular disorders [3, 24]. According to national statistics, approximately 8.3 per 100,000 people use home NIV in Japan, a prevalence similar to that of Western countries, where NIV use continues to rise, and a similar trend is expected in Japan [36, 37]. These considerations highlight the necessity for simple, cost-effective outpatient screening tools, particularly given that home initiation of NIV has proven to be more cost-efficient than hospital-based initiation [38].
To address this, the diagnostic accuracy of accelerometer-integrated pulse oximetry was evaluated as an alternative to PtcCO₂ monitoring for identifying eNH. The method demonstrated sufficient sensitivity and specificity for detecting eNH. In this present study, patients with possible eNH indicated by accelerometer-integrated pulse oximetry exhibited higher daytime PaCO₂ levels than those without possible eNH. These findings are consistent with previous studies that compared patients with and without eNH [10, 11]. These results suggest that accelerometer-integrated pulse oximetry could serve as a practical home screening tool for eNH prior to conducting PtcCO₂ monitoring.
In the patient-level analysis conducted in this study, the specificity of accelerometer-integrated pulse oximetry for detecting eNH was 73.1%, which was lower than the anticipated value. A review of the false-positive cases revealed that five out of seven patients (71.4%) exhibited a marked decrease in SpO₂ without a corresponding increase in PtcCO₂, defined as a rise of ≥ 3 mmHg but < 5 mmHg from baseline for more than 5 min. Including subclinical eNH within the diagnostic criteria increased the specificity of accelerometer-integrated pulse oximetry for detecting eNH to 90.5%. It is plausible that certain instances of sleep-related hypoventilation induced a more substantial reduction in SpO₂ accompanied by only a minimal elevation in PtcCO₂ (≥ 3 mmHg to < 5 mmHg). However, notably, PtcCO₂ may also increase during REM sleep in healthy individuals. Further investigation is needed to determine the clinical relevance of subclinical eNH.
After removing exertional events, most of the remaining misclassifications consisted of short, apnea-like desaturation dips likely related to coexisting sleep apnea or transient artifacts, consistent with our review of false-positive cases (Additional Fig. 3). To mitigate the confounding influence of coexisting sleep apnea and transient signal artifacts, we generated a modified SpO₂ waveform using linear interpolation between the onset and offset of each desaturation dip. This approach flattens brief sawtooth fluctuations while preserving longer desaturation patterns that may reflect underlying physiological changes and remains computationally efficient for practical clinical use. The accuracy of detecting eNH using accelerometer-integrated pulse oximetry improved from 75.7% to 81.4% in event-level analyses. Additionally, in the analysis using the modified SpO₂ waveform, daytime PaCO₂, HCO₃⁻, mean PtcCO₂, and maximum PtcCO₂ were significantly higher in patients with possible eNH compared to others (Figs. 3C and 4C, Additional Table 1). These findings were consistent with the overall assessment outcomes for eNH. Therefore, the application of the modified SpO₂ waveform may facilitate the identification of episodic oxygen desaturation events attributable to hypoventilation. Although the modified SpO₂ waveform reduces motion and apnea-like artifacts, it may also mask brief sleep apnea-related desaturations. Therefore, the choice of SpO₂ analysis should depend on the clinical objective, with original waveforms used for detecting sleep apnea and modified waveforms used for hypoventilation screening. Further research is warranted to validate the effectiveness of the waveform modification technique utilized in this study.
Limitations
The present study has several limitations. First, this was a single-center pilot study with a small, hospitalized cohort, and no formal sample size calculation was performed. The limited number of patients with confirmed eNH resulted in wide confidence intervals, reduced statistical precision, and may have contributed to the high sensitivity observed in this cohort. This hospitalized cohort may have overrepresented more severe phenotypes. These factors restrict the generalizability of our findings, particularly to home settings. Second, PSG was not performed; therefore, residual confounding by coexisting sleep apnea, arousals, or sleep stage distribution cannot be excluded. Third, the modified SpO₂ waveform was designed to suppress brief motion-related or apnea-like dips, but it may also attenuate genuine sleep apnea-related desaturations. External validation using PSG-based datasets is necessary to determine the optimal use of this approach. Fourth, although our drift analysis indicated that increases in PtcCO2 of at least 3 mmHg exceeded expected measurement variability, the clinical significance of such subclinical elevations remains uncertain. Overall, our findings should be viewed as exploratory and supportive of a screening role rather than a diagnostic role. Larger multicenter studies with multi-night and home-based monitoring are needed to confirm reproducibility and clinical utility.
Conclusion
This pilot observational study suggests that accelerometer-integrated pulse oximetry is a feasible approach for distinguishing nocturnal desaturation related to physical activity from that possibly associated with sleep-related hypoventilation. The findings suggest that accelerometer-integrated pulse oximetry may serve as an effective screening tool for episodic nocturnal hypercapnia, particularly in situations where PSG or transcutaneous CO₂ monitoring is unavailable. These preliminary results require validation in home settings with PSG-anchored sleep staging and in larger, multicenter cohorts.
Supplementary Information
Acknowledgements
We thank Toshiro Katayama for providing valuable advice on statistical analysis. We also acknowledge Ryo Yamanaka, Atsushi Funauchi, Shinya Tsukamoto, Yasumitsu Ueki, Hirotaka Tamesada, Takamitsu Imoto, and Yoko Hamakawa for their contributions to data collection.
Abbreviations
- OPO
Overnight pulse oximetry
- SpO2
oxygen saturation
- eNH
episodic nocturnal hypercapnia
- REM
rapid eye movement
- PtcCO2
transcutaneous carbon dioxide pressure
- NIV
non-invasive ventilation
- COPD
chronic obstructive pulmonary disease
- ILD
interstitial lung disease
- NMD
neuromuscular disease
- T90
SpO2 below 90% during sleep
- ODI
oxygen desaturation index
- Cis
confidence intervals
- PaO2
arterial oxygen pressure
- FEV1
forced expiratory volume in 1 s
- HCO3−
bicarbonate
- BMI
body mass index
- PaCO2
arterial partial carbon dioxide pressure
- FEV1
forced expiratory volume in 1 s
- FVC
forced vital capacity
Authors’ contributions
TK is the guarantor of the manuscript and assumes full responsibility for its content. HT contributed to study design, data collection, data analysis, and manuscript editing. EN contributed to data collection and manuscript editing. SJ contributed to data collection, data analysis, and manuscript editing. CM, DI, and SM contributed to data collection and manuscript editing. AM contributed to study design, data analysis, and manuscript editing. MF, as senior author, contributed to study design, data analysis, manuscript drafting, and manuscript editing. All authors read and approved the final manuscript.
Funding
This study was financially supported by KONICA MINOLTA, INC.
Data availability
The datasets generated and/or analyzed during this study are available from the corresponding author upon reasonable request.
Declarations
Ethics approval and consent to participate
The study was conducted in accordance with the ethical guidelines of the Japanese Ministry of Health, Labor, and Welfare and was approved by the Institutional Review Board of the Medical Research Institute, KITANO HOSPITAL Ethics Committee (Ethics Board approval number: P2102009). Written informed consent was obtained from all patients. Additionally, we prioritized the protection of personal information in compliance with ethical guidelines.
Consent for publication
Written informed consent for the publication of anonymized clinical details and/or accompanying data was obtained from all participants included in this study. All identifying information has been removed to protect participant privacy.
Competing interests
The authors declare no competing interests.
Footnotes
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
References
- 1.Hilbert J. Sleep-disordered breathing in neuromuscular and chest wall diseases. Clin Chest Med. 2018;39(2):309–24. [DOI] [PubMed] [Google Scholar]
- 2.Aboussouan LS, Mireles-Cabodevila E. Sleep-disordered breathing in neuromuscular disease: diagnostic and therapeutic challenges. Chest. 2017;152(4):880–92. [DOI] [PubMed] [Google Scholar]
- 3.O’Donoghue FJ, Catcheside PG, Ellis EE, et al. Sleep hypoventilation in hypercapnic chronic obstructive pulmonary disease: prevalence and associated factors. Eur Respir J. 2003;21(6):977–84. [DOI] [PubMed] [Google Scholar]
- 4.Ward S, Chatwin M, Heather S, Simonds AK. Randomised controlled trial of non-invasive ventilation for nocturnal hypoventilation in neuromuscular and chest wall disease patients with daytime normocapnia. Thorax. 2005;60(12):1019–s1024. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 5.McNicholas WT, Hansson D, Schiza S, Grote L. Sleep in chronic respiratory disease: COPD and hypoventilation disorders. Eur Respir Rev. 2019;28(153):190058. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 6.Berthon-Jones M, Sullivan CE. Ventilatory and arousal responses to hypoxia in sleeping humans. Am Rev Respir Dis. 1982;125(6):632–9. [DOI] [PubMed] [Google Scholar]
- 7.Yang H, Xiang P, Zhang E, et al. Is hypercapnia associated with poor prognosis in chronic obstructive pulmonary disease? A long-term follow-up cohort study. BMJ Open. 2015;5(12):e008909. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 8.Kessler R, Faller M, Fourgaut G, Mennecier B, Weitzenblum E. Predictive factors of hospitalization for acute exacerbation in patients with chronic obstructive pulmonary disease. Am J Respir Crit Care Med. 1999;159(1):158–64. [DOI] [PubMed] [Google Scholar]
- 9.Parker SM, Gibson GJ. Evaluation of a transcutaneous carbon dioxide monitor (TOSCA) in adult patients in routine respiratory practice. Respir Med. 2007;101(2):261–4. [DOI] [PubMed] [Google Scholar]
- 10.Kitajima T, Marumo S, Shima H, et al. Clinical impact of episodic nocturnal hypercapnia and its treatment with noninvasive positive pressure ventilation in patients with stable advanced COPD. Int J Chron Obstruct Pulmon Dis. 2018;13:843–53. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 11.Kitajima T, Marumo S, Amimoto H, et al. Relationship between episodic nocturnal hypercapnia and history of exacerbations in patients with advanced chronic obstructive pulmonary disease. Int J Chron Obstruct Pulmon Dis. 2022;17:1553–63. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 12.Naito H, Sugimoto T, Kimoto K, et al. Detection of episodic nocturnal hypercapnia inpatients with neurodegenerative disorders. Sleep Breath. 2024;28(1):393–399. 10.1007/s11325-023-02876-3. [DOI] [PubMed]
- 13.Janssens JP, Borel JC, Pepin JL, Somno NIVG. Nocturnal monitoring of home non-invasive ventilation: the contribution of simple tools such as pulse oximetry, capnography, built-in ventilator software and autonomic markers of sleep fragmentation. Thorax. 2011;66(5):438–45. [DOI] [PubMed] [Google Scholar]
- 14.Farmer MJS, Callahan CD, Hughes AM, Riska KL, Hill NS. Applying noninvasive ventilation in treatment of acute exacerbation of COPD using evidence-based interprofessional clinical practice. Chest. 2024;165(6):1469–1480. 10.1016/j.chest.2024.02.040. [DOI] [PMC free article] [PubMed]
- 15.Netzer N, Eliasson AH, Netzer C, Kristo DA. Overnight pulse oximetry for sleep-disordered breathing in adults: a review. Chest. 2001;120(2):625–33. [DOI] [PubMed] [Google Scholar]
- 16.Singh S, Khan SZ, Singh D, Verma S, Talwar A. The uses of overnight pulse oximetry. Lung India. 2020;37(2):151–7. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 17.Ward NR, Cowie MR, Rosen SD, et al. Utility of overnight pulse oximetry and heart rate variability analysis to screen for sleep-disordered breathing in chronic heart failure. Thorax. 2012;67(11):1000–5. [DOI] [PubMed] [Google Scholar]
- 18.Yoshizaki A, Nagano T, Izumi S, et al. Prospective study of nocturnal desaturation in patients receiving home oxygen therapy. Intern Med. 2021;60(19):3071–9. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 19.Vitacca M, Bianchi L, Guerra A, et al. Tele-assistance in chronic respiratory failure patients: a randomised clinical trial. Eur Respir J. 2009;33(2):411–8. [DOI] [PubMed] [Google Scholar]
- 20.Hang LW, Wang HL, Chen JH, et al. Validation of overnight oximetry to diagnose patients with moderate to severe obstructive sleep apnea. BMC Pulm Med. 2015;15:24. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 21.Fletcher EC, Donner CF, Midgren B, et al. Survival in COPD patients with a daytime PaO₂ greater than 60 mm hg with and without nocturnal oxyhemoglobin desaturation. Chest. 1992;101(3):649–55. [DOI] [PubMed] [Google Scholar]
- 22.Chaouat A, Weitzenblum E, Kessler R, et al. Sleep-related O₂ desaturation and daytime pulmonary haemodynamics in COPD patients with mild hypoxaemia. Eur Respir J. 1997;10(8):1730–5. [DOI] [PubMed] [Google Scholar]
- 23.Lacasse Y, Sériès F, Vujovic-Zotovic N, et al. Evaluating nocturnal oxygen desaturation in COPD: revised. Respir Med. 2011;105(9):1331–7. [DOI] [PubMed] [Google Scholar]
- 24.Bauman KA, Kurili A, Schmidt SL, Rodriguez GM, Chiodo AE, Sitrin RG. Home-based overnight transcutaneous capnography/pulse oximetry for diagnosing nocturnal hypoventilation associated with neuromuscular disorders. Arch Phys Med Rehabil. 2013;94(1):46–52. [DOI] [PubMed] [Google Scholar]
- 25.Petitjean T, Philit F, Germain-Pastenne M, Langevin B, Guérin C. Sleep and respiratory function after withdrawal of noninvasive ventilation in patients with chronic respiratory failure. Respir Care. 2008;53(10):1316–23. [PubMed] [Google Scholar]
- 26.Shah NM, Steier J, Hart N, Kaltsakas G. Effects of non-invasive ventilation on sleep in chronic hypercapnic respiratory failure. Thorax. 2024;79(3):281–288. 10.1136/thorax-2023-220035. [DOI] [PubMed]
- 27.Chan LLY, Choi TCM, Lord SR, Brodie MA. Development and large-scale validation of the watch walk wrist-worn digital gait biomarkers. Sci Rep. 2022;12(1):16211. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 28.Miller MR, Hankinson J, Brusasco V, et al. Standardisation of spirometry. Eur Respir J. 2005;26(2):319–38. [DOI] [PubMed] [Google Scholar]
- 29.Waatevik M, Johannessen A, Gomez Real F, et al. Oxygen desaturation in 6-min walk test is a risk factor for adverse outcomes in COPD. Eur Respir J. 2016;48(1):82–91. [DOI] [PubMed] [Google Scholar]
- 30.Crisafulli E, Iattoni A, Venturelli E, et al. Predicting walking-induced oxygen desaturations in COPD patients: a statistical model. Respir Care. 2013;58(9):1495–503. [DOI] [PubMed] [Google Scholar]
- 31.Perez T, Deslée G, Burgel PR, et al. Predictors in routine practice of 6-min walking distance and oxygen desaturation in patients with COPD: impact of comorbidities. Int J Chron Obstruct Pulmon Dis. 2019;14:1399–410. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 32.Ferrari GLM, Kovalskys I, Fisberg M, et al. Comparison of self-report versus accelerometer-measured physical activity and sedentary behaviors and their association with body composition in Latin American countries. PLoS ONE. 2020;15(4):e0232420. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 33.Lacasse Y, Sériès F, Corbeil F, et al. Randomized trial of nocturnal oxygen in chronic obstructive pulmonary disease. N Engl J Med. 2020;383(12):1129–38. [DOI] [PubMed] [Google Scholar]
- 34.Boing S, Randerath WJ. Chronic hypoventilation syndromes and sleep-related hypoventilation. J Thorac Dis. 2015;7(8):1273–85. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 35.Johnson MW, Remmers JE. Accessory muscle activity during sleep in chronic obstructive pulmonary disease. J Appl Physiol Respir Environ Exerc Physiol. 1984;57(4):1011–7. [DOI] [PubMed] [Google Scholar]
- 36.Kotanen P, Brander P, Kreivi HR. The prevalence of non-invasive ventilation and long-term oxygen treatment in Helsinki university hospital area, Finland. BMC Pulm Med. 2022;22(1):248. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 37.Dretzke J, Wang J, Yao M, et al. Home Non-Invasive ventilation in COPD: A global systematic review. Chronic Obstr Pulm Dis. 2022;9(2):237–51. [DOI] [PMC free article] [PubMed] [Google Scholar]
- 38.Duiverman ML, Vonk JM, Bladder G, et al. Home initiation of chronic non-invasive ventilation in COPD patients with chronic hypercapnic respiratory failure: a randomised controlled trial. Thorax. 2020;75(3):244–52. [DOI] [PMC free article] [PubMed] [Google Scholar]
Associated Data
This section collects any data citations, data availability statements, or supplementary materials included in this article.
Supplementary Materials
Data Availability Statement
The datasets generated and/or analyzed during this study are available from the corresponding author upon reasonable request.