Improvement of sleep parameters by titration polysomnography could predict adherence to positive airway pressure therapy in obstructive sleep apnea

Abstract

Study Objectives:

Poor adherence to continuous positive airway pressure (CPAP) has been a critical issue in treating obstructive sleep apnea. Because long-term CPAP adherence may be established shortly after treatment begins, early intervention is essential. This study aimed to identify the potential factors affecting CPAP therapy adherence during diagnostic polysomnography and auto CPAP titration polysomnography.

Methods:

This retrospective observational study included 463 patients with obstructive sleep apnea who underwent consecutive diagnostic polysomnography and titration polysomnography. We recorded their demographic, anthropometric, and lifestyle factors and obtained self-reported comments regarding their sleep status following both polysomnography evaluations. CPAP adherence was evaluated following 3 months of treatment.

Results:

A total of 312 patients (67.4%) fulfilled the criteria for good adherence. Each patient’s CPAP adherence was categorized as “poor” (< 4 hours/night or <70% of nights), “good” (≥ 4 hours/night and ≥ 70% of nights), or “excellent” (≥ 6 hours/night and ≥ 80% of nights). There were no significant differences in arterial oxyhemoglobin saturation measured by pulse oximetry and apnea-hypopnea index during diagnostic polysomnography among 3 groups. The polysomnographic evaluations indicated that patients with better adherence displayed more significant improvements in sleep parameters, including apnea-hypopnea index, sleep efficacy, sleep latency, and sleep architecture, which were correlated with an improvement in self-reported sleep quality.

Conclusions:

Polysomnographic evaluations enabled CPAP adherence prediction and a comparison of self-reported sleep quality with and without CPAP; CPAP adherence led to improvements in polysomnographic parameters. Our findings suggest that titration polysomnography and self-reported sleep improvement with CPAP could be used for adherence prediction in clinical practice.

Citation:

Shirahata T, Uchida Y, Uchida T, et al. Improvement of sleep parameters by titration polysomnography could predict adherence to positive airway pressure therapy in obstructive sleep apnea. J Clin Sleep Med. 2023;19(8):1465–1473.

BRIEF SUMMARY

Current Knowledge/Study Rationale: Early intervention is essential for continuous positive airway pressure (CPAP) adherence, and there have been no large-scale studies associated with CPAP adherence using titration polysomnography parameters in patients with obstructive sleep apnea. This study identified CPAP adherence predictors using data from diagnostic polysomnography and titration polysomnography in addition to sleep questionnaires following polysomnography evaluations.

Study Impact: CPAP adherence was correlated with improvements in polysomnography parameters, and self-reported sleep improvement with CPAP could be a substitute for titration polysomnography in daily clinical practice.

INTRODUCTION

Obstructive sleep apnea (OSA) affects 32% of the Japanese population aged 30–69 years.¹ It is characterized by snoring associated with reductions or cessations in respiration owing to the partial or total collapse of the upper airways during sleep,² thereby leading to adverse health outcomes such as daytime somnolence, resistant hypertension, cardiovascular disease, neurological disease, impaired quality of life, increased motor vehicle accidents, and all-cause mortality.³

Continuous positive airway pressure (CPAP) has been recognized as the first-line treatment for OSA since the early 1980s for counteracting apneic episodes and consequent hypoxia. The appropriate use of CPAP is highly efficacious in controlled studies and normalizes sleep architectures.^4–9 However, poor adherence to CPAP therapy is a crucial problem and limits its overall effectiveness in patients of all ages with OSA.¹⁰

Various factors influence CPAP adherence, including demographics, psychological behavior, patient and disease characteristics, technological device factors, and side effects.¹¹ Several factors, such as demographics and lifestyle, may differ by country or region; therefore, predicting the degree of CPAP adherence requires evaluation after considering the backgrounds of each country/region. One investigation on the relationship between CPAP adherence and polysomnography (PSG) data was published in Japan; however, the researchers conducted only diagnostic polysomnography (dPSG) and CPAP adherence was only 25% (44 of 173 patients with OSA).¹² Furthermore, CPAP adherence is defined as the persistent use of CPAP for ≥ 70% per month and ≥ 4 hours per day¹³; however, the optimal duration for CPAP use is debatable. Recent reports have revealed that ≥ 6 hours of adherence per night was associated with greater improvements in clinical outcomes such as excessive daytime sleepiness, glucose metabolism, and memory performance.^14–16 For instance, Ioachimescu et al¹⁶ reported that they divided CPAP adherence into 4 categories as follows: poor, good (≥ 70% of nights and ≥ 4 hours/night), excellent (≥ 80% of nights and ≥ 6 hours/night), and outstanding (≥ 90% of nights and 8 hours/night); none of the patients in the excellent and outstanding groups developed type 2 diabetes mellitus. To address this hypothesis, we divided adherence into 3 categories and examined the factors influencing CPAP adherence.

Although it is essential to enhance CPAP adherence to maximize its effectiveness, improving CPAP adherence following its initiation has been challenging despite various efforts toward behavioral intervention and patient education.¹⁷ Moreover, several studies have reported that long-term CPAP adherence is correlated with short-term CPAP adherence.^18–20 Thus, an early intervention for promoting CPAP adherence may be most effective. Some patient characteristics, such as age, sex, marital status, excessive daytime sleepiness, and PSG parameters have been reportedly associated with CPAP adherence^21–24; however, there are no identical factors to predict CPAP adherence. In addition, few studies have examined the correlation of sleep parameters with CPAP adherence. CPAP resulted in a lower arousal index and an improvement of sleep efficacy and sleep architecture.^19,25–27 Further, CPAP adherence has been correlated with changes in PSG parameters with or without CPAP, despite assessment in a small sample size.^19,27 Thus, we aimed to investigate the factors that predict CPAP adherence in the large-scale population during dPSG and titration polysomnography (tPSG).

METHODS

Patients and study design

Existing clinical data from 755 patients with OSA who underwent both dPSG and tPSG at the sleep center of Saitama Medical University Hospital between January 2010 and June 2020 were analyzed. Because this was a retrospective study of previously collected data, the questionnaires were collected within 3 months of the patients’ first visit to our hospital. Of these 755 patients, 177 were excluded because they underwent tPSG following a few months of CPAP initiation. In addition, 7 patients received treatment other than CPAP, 35 patients refused or failed CPAP before the 3-month follow-up visit, and 73 patients were lost to follow-up (most transferred to other facilities and some lacked a smart card for collecting CPAP adherence data). Eventually, we included 463 patients who underwent consecutive dPSG and tPSG (Figure 1). CPAP adherence was evaluated 3 months following its initiation. Our hospital standard for new patients includes collection of height, weight, and neck and abdominal circumferences by well-trained nurses. Other information including age, sex, symptoms, comorbid diseases, and lifestyle habits are also collected using self-reported questionnaires. Body mass index (BMI) was calculated using these measurements (body weight in kilograms/height in meters squared). We included evaluation of this retrospective data for past patients in our present study.

Figure 1. Flowchart depicting participant selection.

CPAP = continuous positive airway pressure, dPSG = diagnostic polysomnography, OSA = obstructive sleep apnea, tPSG = titration polysomnography.

This study was approved by the Institutional Review Board of Saitama Medical University Hospital (Nos. 16030.01 and 20168.01). The need for patient approval or informed consent was waived because the study involved a retrospective review of the patient records.

Questionnaires

Because this was a retrospective study of previously collected data, the questionnaires were collected within 3 months of the patients’ first visit to our hospital.

New patients to our hospital all undergo a detailed clinical interview; we obtained medical history and lifestyle habit information such as smoking status, alcohol intake, and sleep habits. Further, they responded to the Epworth Sleepiness Scale.²⁸ In addition, the morning following the PSG examinations we requested they submit a self-reported questionnaire regarding sleep quality and quantity during the PSG nights. Sleep quality for the PSG night was compared with that on usual nights and calculated as follows: PSG night sleep was better than usual = +1, almost same = 0, and worse than usual = −1. The questionnaire submitted by our patients was originally created at our sleep center and is provided in the supplemental material. We obtained our retrospective data from the records of the patients who fit the criteria for our study.

PSG, CPAP initiation, and follow-up

Each participant underwent overnight PSG at our sleep laboratory. PSG was performed using a SomnoStar Pro System (SensorMedics, Milano, Italy) and Somnoscreen Plus (SOMNOmedics, Randersacker, Germany), and tPSG with an auto CPAP machine (System One REMstar series or DreamStation, Philips Respironics, Murrysville, Pennsylvania and S9 or S10 AutoSet, ResMed, San Diego, California) was conducted. The following parameters were recorded continuously: electroencephalography, electrocardiography, chin muscle electromyography, electrooculography, and oxygen saturation. We measured the respiratory effort using thoracoabdominal strain gauges and oronasal airflow (thermistor signals and a pressure cannula). Trained technicians performed all procedures. All computerized sleep data were further analyzed manually by experienced staff. Sleep staging was assessed according to the American Academy of Sleep Medicine.²⁹ The apnea-hypopnea index (AHI) score refers to the total number of apnea and hypopnea events per hour of sleep duration.

CPAP therapy was initiated to patients with an AHI (OSA) greater than 20 events/h in the dPSG, according to the Japanese health insurance system. In addition, 5 patients with a total AHI less than 20 events/h received CPAP therapy owing to severe rapid eye movement sleep–related OSA. In this study, all patients received treatment with auto CPAP and were educated about mask fittings and CPAP use by a dedicated technologist at the sleep center prior to CPAP initiation. We reviewed the effects of treatment on disease-related symptoms and problems of CPAP device and masks at a 1-month follow-up visit. We modified the airflow pressure or let the patients consult about mask problems with specialized clinical engineers, if required. Patients who reported difficulty in using CPAP owing to nasal closure symptoms were treated with antiallergy drugs or by an otorhinolaryngologist. Those who could not continue CPAP therapy switched to other treatments, such as oral appliances. CPAP adherence (+) was defined as using CPAP for ≥ 70% of the days each month (including days when it was used for < 4 hours) and for > 4 hours per day on the days it was used,¹³ and we divided adherence into 3 groups as follows: good (≥ 4 hours/night and ≥ 70% of nights), excellent (≥ 6 hours/night and ≥ 80% of nights), and poor (< 4 hours/night or < 70% of nights).¹⁶ CPAP data, such as the average CPAP daily use (hours) and the rate of CPAP use, were downloaded using software (Fukuda Multi Viewer, Fukuda-Denshi Co., Ltd., Tokyo, Japan) from a smart card attached to the device at each follow-up visit.

Statistical analyses

EZR 4.1.2 (R, open-source, Saitama Medical Centre, Jichi Medical University, Saitama, Japan) was used for database management and all statistical analyses. Data pertaining to the continuous and categorical variables were expressed as the mean ± standard deviation and interquartile range (IQR) and as numbers or percentages, respectively. We analyzed the relationship between PSG parameters and CPAP adherence using Pearson’s correlation. The Jonckheere-Terpstra and Cochran-Armitage trend tests were performed for a trend analysis of the continuous and categorical variables, respectively. Pearson’s correlation was performed to evaluate changes in different sleep parameters with CPAP. We conducted an unpaired Student’s t test to compare the PSG parameters between the good and excellent adherence groups. P values <.05 were considered statistically significant.

RESULTS

A total of 463 patients (aged 22–86 years, 367 men, 96 women) were included in this study. Table 1 summarizes their baseline characteristics stratified by CPAP adherence; their mean age was 56.1 ± 13.3 years, BMI was 28.7 ± 5.3 kg/m², and Epworth Sleepiness Scale score was 8.6 ± 4.6. The OSA severity was as follows: mild (AHI ≥ 5 and < 15 events/h, n = 2), moderate (AHI ≥ 15 and < 30 events/h, n = 95), or severe (AHI ≥ 30 events/h, n = 366). At baseline, 35.4%, 42.5%, and 22.1% of the patients were obese (BMI ≥ 30), overweight (25 ≤ BMI < 30), and normal weight (BMI < 25), respectively. The BMI was positively correlated with disease severity (r = .436, P < .0001) and negatively correlated with age (r = −.25, P < .0001).

Table 1.

Patients’ characteristics stratified by CPAP adherence.

	All (463)	Poor (151)	Good (180)	Excellent (132)	P	r*	P*
Age, years	56.6 ± 12.9	52.4 ± 13.6	56.5 ± 12.4	61.5 ± 12.7	<.0001	.29	<.0001
Sex, male	79.3%	79.5%	80.0%	78.0%	.78	.00	.90
BMI, kg/m2	28.7 ± 5.3	30.1 ± 6.1	28.4 ± 4.9	27.7 ± 4.4	<.001	−.21	<.0001
AHI severity, % mild / moderate / severe	0.4 / 20.5 / 79.1	0.6 / 17.9 / 81.5	0.7 / 21.7 / 77.8	0 / 22.0 / 78.0	.44	.05	.51
NC, cm	39.8 ± 3.8	40.6 ± 3.9	39.7 ± 3.9	39.1 ± 3.6	<.01	−.15	<.01
AC, cm	97.0 ± 12.7	99.6 ± 14.2	96.3 ± 12.1	95.1 ± 11.4	<.05	−.16	<.001
SBP, mmHg	133.4 ± 17.4	133.5 ± 17.3	133.2 ± 18.1	133.4 ± 16.8	.94	.00	1.00
DBP, mmHg	83.1 ± 13.3	83.8 ± 13.7	83.4 ± 14.2	81.8 ± 11.6	.17	−.08	.09
Hypertension	52.9%	49.0%	53.3%	56.8%	.19	.04	.36
Hyperlipidemia	43.2%	35.8%	46.7%	47.0%	.052	.03	.46
Diabetes mellitus	16.2%	18.5%	13.9%	16.7%	.64	.03	.60
Sleep medicine intake	21.8%	24.7%	18.3%	23.4%	.75	.01	.76
Smoking status, % never / past / current	35.5 / 43.3 / 21.2	33.8 / 41.1 / 25.2	36.2 / 41.2 / 22.6	36.5 / 48.8 / 14.7	.23	.06	.22
ESS score	8.6 ± 4.6	9.6 ± 4.8	8.7 ± 4.4	7.4 ± 4.6	<.0001	−.16	<.001

Data reported as the number of patients (%) or mean ± SD. *Association with 3-month CPAP use (hours/night). AC = abdominal circumference, AHI = apnea-hypopnea index, BMI = body mass index, CPAP = continuous positive airway pressure, DBP = diastolic blood pressure, ESS = Epworth Sleepiness Scale, NC = neck circumference, SBP = systolic blood pressure.

Patients with better adherence had lower BMIs (P < .001), smaller neck circumference (P < .01), smaller abdominal circumference (P < .05), and older age (P < .0001). There were no differences in smoking habits, sleep medicine intake, and comorbidities such as hypertension, hyperlipidemia, and diabetes mellitus among the groups.

CPAP adherence after 3 months of CPAP therapy

The overall CPAP adherence was 83.2 ± 22.4% (IQR 76.5–100) per month and average usage was 5.0 ± 1.7 (IQR 3.9–6.2) hours/night. The average residual AHI was 3.7 ± 3.8 (IQR 1.9–4.5), with only 2.2% of the patients having values > 10 events/h. Overall adherence to CPAP was observed in 312 (67.4%), and 151 (32.6%) patients displayed poor adherence. “Good adherence” was observed in 180 (38.9%) patients, excluding those with excellent adherence (132 patients, 28.5%). Patients with moderate OSA used CPAP on 86.1 ± 19.4% (IQR 80.0–100.0) of the nights for an average usage of 5.00 ± 1.7 hours (IQR 4.0–6.2); those with severe OSA used it on 82.5 ± 23.1% (IQR 75.0–100.0) of the nights for an average usage of 5.0 ± 1.7 hours (IQR 3.9–6.3). There were no significant differences in the frequency or length between the moderate and severe OSA groups. Older patients and those with Epworth Sleepiness Scale scores < 10 displayed better CPAP adherence (P < .0001 and P < .01, respectively).

Relationship between PSG data and CPAP adherence

Table 2 summarizes the data for dPSG and ΔPSG (Δ = changes in the parameters between dPSG and tPSG) in the poor, good, and excellent adherence groups. Trend analyses revealed that patients with better adherence displayed longer sleep latency (P < .01), lower sleep efficiency (P < .05), higher total sleep time (TST) stage N1% (P < .05), and lower TST stage N3% (P < .001) in the dPSG (Table 2A), besides significantly improved AHI (P < .01), TST (P < .05), sleep latency (P < .05), sleep efficiency (P < .01), and sleep architecture, such as stage N1% and N3% (P < .05), with CPAP (Table 2B). Further, we examined these associations with the average numbers of hours per night instead of the 3 adherence categories. Similar correlations were observed except for sleep latency and efficacy in the dPSG and except stage N3% and arousal index in the ΔPSG. In addition, patients with better adherence slept longer and better according to the sleep questionnaires following the PSG nights; the Δsubjective sleep quality was correlated with an improvement in various sleep parameters as follows: ΔAHI (P < .01), ΔTST (P < .0001), ΔTST stage N1% (P < .001), ΔTST stage N3% (P < .01), Δsleep efficacy (P < .0001), and Δsleep latency (P < .0001) (Figure 2). Furthermore, we investigated the associations at the 6-month follow-up visit as midterm CPAP adherence and provide the findings in Table S1 ^{(44.2KB, pdf)} ). The associations of 6-month CPAP adherence with ΔPSG parameters, such as AHI, stage N1%, and stage N3%, were persistent, although approximately 100 patients were lost to follow-up.

Table 2.

PSG studies stratified by CPAP adherence.

	All (463)	Poor (151)	Good (180)	Excellent (132)	P	r *	P *
A. Diagnostic PSG study
AHI, events/h	49.2 ± 21.5	49.6 ± 21.0	49.0 ± 22.5	49.0 ± 20.9	.78	.01	.76
TST, min	365.3 ± 76.9	368.8 ± 80.8	370.7 ± 74.5	353.9 ± 75.1|	.07	−.06	.21
TST stage N1, %	39.8 ± 18.4	38.4 ± 17.0	37.8 ± 18.0	44.1 ± 19.8|	<.05	.14	<.01
TST stage N2, %	39.9 ± 14.4	39.9 ± 13.9	40.7 ± 14.2	38.8 ± 15.1	.55	−.05	.31
TST stage N3, %	11.1 ± 10.0	12.2 ± 9.8	12.2 ± 10.1	8.2 ± 9.6	<.001	−.17	<.001
TST stage R, %	9.23 ± 5.8	9.5 ± 5.6	9.2 ± 5.5	9.0 ± 6.5	.31	−.04	.44
Arousal index, per hour	27.9 ± 19.6	25.7 ± 16.6	29.3 ± 20.6	28.4 ± 21.2	.49	.08	.08
Sleep latency, minutes	14.5 ± 21.3	13.1 ± 23.9	14.3 ± 20.2	16.5 ± 19.5	<.01	.08	.10
Sleep efficiency, %	75.9 ± 13.9	77.0 ± 14.3	76.5 ± 13.5	73.9 ± 14.1	<.05	−.07	.12
SpO2 > 90% TST, %	67.2 ± 28.7	67.6 ± 28.1	65.6 ± 29.6	68.9 ± 28.2	.65	.00	.98
Average SpO2, %	90.6 ± 4.5	90.5 ± 4.6	90.4 ± 4.8	90.9 ± 4.1	.62	.01	.80
Minimum SpO2, %	73.1 ± 97	73.2 ± 9.6	72.4 ± 10.0	73.9 ± 9.5	.48	.04	.46
Heart rate, beats per minute	72.3 ± 12.1	71.8 ± 11.4	73.2 ± 11.9	71.5 ± 13.2	.57	−.01	.85
Questionnaire
Sleep quality	−0.45 ± 0.6	−0.44 ± 0.6	−0.40 ± 0.6	−0.51 ± 0.5	.41	.07	.16
Sleep duration, h	6.0 ± 1.7	6.2 ± 1.6	6.2 ± 1.6	5.6 ± 2.0	<.05	−.12	<.05
B. Changes between dPSG and tPSG studies
Δ AHI, events/h	−35.5 ± 21.7	−31.8 ± 22.1	−36.5 ± 20.2	−38.5 ± 22.8	<.01	−.13	<.01
Δ TST, minutes	5.6 ± 89.4	−0.5 ± 98.2	−1.5 ± 82.0	22.3 ± 86.9	<.05	.10	<.05
Δ TST stage N1, %	−11.9 ± 20.5	−9.9 ± 20.1	−10.3 ± 18.9	−16.5 ± 22.4	<.05	−.13	<.01
Δ TST stage N2, %	3.4 ± 16.1	2.7 ± 15.9	2.4 ± 15.9	5.6 ± 16.5	.21	.08	.10
Δ TST stage N3, %	5.8 ± 11.1	4.7 ± 10.5	5.3 ± 11.8	7.5 ± 10.8	<.05	.08	.07
Δ TST stage R, %	2.7 ± 7.2	2.4 ± 7.2	2.6 ± 7.2	3.3 ± 7.4	.33	.07	.12
Δ Arousal index, /h	−13.0 ± 19.9	−10.7 ± 16.9	−12.9 ± 20.8	−15.8 ± 21.7	.08	−.11	<.05
Δ Sleep latency, minutes	1.9 ± 23.4	4.8 ± 26.9	0.59 ± 22.2	0.37 ± 20.4	<.05	−.11	<.05
Δ Sleep efficiency, %	1.4 ± 16.2	−0.8 ± 17.0	0.93 ± 14.9	4.5 ± 16.6	<.01	.13	<.01
Δ SpO2 > 90% TST, %	25.8 ± 26.9	23.2 ± 24.6	26.8 ± 27.9	27.3 ± 27.8	.30	.08	.10
Δ Average SpO2, %	4.0 ± 4.0	3.7 ± 3.7	4.4 ± 4.2	4.0 ± 3.8	.62	.05	.33
Δ Minimum SpO2, %	11.1 ± 9.1	10.1 ± 8.8	12.1 ± 9.1	11.0 ± 9.3	.68	.03	.49
Δ Heart rate, beats per minute	−4.1 ± 9.1	−3.9 ± 7.6	−3.4 ± 10.7	−5.4 ± 8.1	.12	−.03	.56
Questionnaire
Δ Subjective sleep quality	0.17 ± 0.7	0.09 ± 0.7	0.14 ± 0.7	0.30 ± 0.8	<.05	.11	<.05
Δ Subjective sleep duration, hours	0.01 ± 2.1	−0.21 ± 2.0	−0.20 ± 2.1	0.56 ± 2.0	<.01	.15	<.01

^*Association with the 3-month CPAP use (hours/night). AHI = apnea-hypopnea index, CPAP = continuous positive airway pressure, dPSG = diagnostic polysomnography, SpO₂ = blood oxygen saturation, TST = total sleep time.

Figure 2. Correlation between Δsubjective sleep quality and changes in sleep parameters between dPSG and tPSG.

(A) AHI, (B) TST, (C) Stage N1, (D) Stage N3, (E) Sleep efficacy, (F) Sleep latency. ΔSubjective sleep quality is indicated by the score and the corresponding number of patients in parentheses. AHI = apnea-hypopnea index, dPSG = diagnostic polysomnography, tPSG = titration polysomnography, TST = total sleep time.

Comparison of changes in sleep parameters between the dPSG and tPSG

Figure 3 depicts the bivariate analysis of changes in PSG parameters during the tPSG, compared with the dPSG. There were significant associations among changes in different sleep parameters.

Figure 3. Relationship among the changes in PSG parameters between dPSG and tPSG.

AHI = apnea-hypopnea index, dPSG = diagnostic polysomnography, PSG = polysomnography, tPSG = titration polysomnography, TST = total sleep time.

Exploring the PSG variables associated with better CPAP adherence

We further investigated the factors that distinguished good and excellent adherence among the PSG parameters related to adherence. Despite a moderately higher ΔTST stage N3% (P = .09), the excellent adherence group displayed significant improvements in the ΔTST stage N1% (P < .01), Δsleep efficiency (P < .05), ΔTST (P < .05), and Δsubjective sleep quality and duration (P < .05 and P < .01, respectively) compared with the good adherence group.

DISCUSSION

This study obtained sleep parameters and self-reported sleep status during both dPSG and tPSG. CPAP adherence in patients with OSA could be predicted by examining the changes in various indices with CPAP use. PSG evaluations demonstrated that CPAP adherence can be expected in patients who demonstrated improvements in AHI, sleep architecture, such as stage N1% and N3%, TST, sleep efficiency, and sleep latency with CPAP. These findings were correlated with self-reported sleep improvement. In addition, more significant improvements in sleep architecture, sleep efficacy, and self-reported sleep status with CPAP could result in better adherence. These findings suggest the importance of evaluating both self-reported and objective sleep indices during tPSG for predicting CPAP adherence.

Several previous reports have investigated the relationship between CPAP adherence and sleep parameters during tPSG. Somiah et al¹⁹ reported that 93 patients with short-term (1 week) good adherence displayed lower stage N2% and higher stage R% on tPSG; Drake et al²⁷ demonstrated that an improvement in sleep efficiency on tPSG was correlated with CPAP adherence (71 patients, 47 days). In the present study, we demonstrated that short to midterm CPAP adherence was correlated with an improvement in sleep parameters during the first night of CPAP use in a larger population. Although there were some different findings in our study compared to other studies, such as the relatively high baseline stage N1%, they may be due to different patients’ backgrounds such as age, sex, and severity of disease.³⁰ Furthermore, in our study, patients with better adherence to CPAP had higher stage N1%, lower stage N3%, longer sleep latency, and lower sleep efficacy in the dPSG. Although the full impact of these parameters on CPAP adherence is unclear, it is reasonable to assume that the degree of improvement patients using CPAP experience in these parameters contributes to enhanced CPAP adherence. In addition, a randomized control study revealed that CPAP resulted in a lower arousal index, less stage N1, and higher stage N3 but no increase in stage R; these parameters were correlated with decrease in sleepiness.²⁶ These findings suggested the importance of a change in sleep parameters and the initial experience with CPAP. This necessitates an active approach to enhance CPAP adherence in patients who may display poor adherence from tPSG data. For example, researchers should make active efforts to improve sleep parameters during the tPSG study. In lieu of considering conducting another tPSG study, we recommend patients visit our clinic earlier than the usual 1-month visit to investigate the cause of poor PSG results. We recommend patients visit our clinic earlier than the usual 1-month visit to investigate the cause of poor PSG results, in addition to considering conducting another tPSG study.

By contrast, Younes et al²⁴ recently reported that CPAP adherence was correlated with AHI and the odds ratio product, which is a continuous index of sleep depth, in the dPSG; prediction using an improvement in sleep parameters with CPAP was not considerably different from the dPSG measures of sleep, which were not consistent with our findings. One explanation for this difference is the varied patient characteristics; the researchers recruited more women and patients with obesity, severe disease, and lower average arterial oxyhemoglobin saturation measured by pulse oximetry (SpO₂) (16% of the population used bilevel positive airway pressure), which could have affected the sleep indices, particularly during dPSG. For example, the average SpO₂ and AHI in our study displayed no significant difference among the 3 groups, thus highlighting the importance of tPSG (changes in sleep parameters). Second, they used the split-night PSG and total recording time (TRT) for evaluations, whereas we used full-night PSG and TST. Total recording time is not similar to TST; total recording time percentage–based sleep architecture was likely to be inaccurate. These differences may have contributed to the inconsistency of findings.

Several issues about split-night PSG studies that might affect these studies should be mentioned. The individual variability in sleep patterns and an accurate assessment of the sleep architecture are major concerns about split-night PSG. Some patients may have rapid eye movement–related OSA; Ciftci et al³¹ found that TST stage N3% and TST stage R% during the first half (3 hours) of TST were significantly higher and lower than that during the second half. Brillante et al³² found a 40% rebound in N3% but only a 20% rebound in R% in titration studies. Given these findings, although split-night PSG has some advantages over full-night PSG, particularly in terms of time and cost efficiency, it has limitations and may not provide as much information or accuracy as full-night PSG.

In addition, we demonstrated that the improvement in self-reported sleep status was correlated with CPAP adherence and progress in various sleep parameters. Verma et al³³ previously reported on a subjective improvement in sleep quality correlated with the sleep architecture in 44 patients with OSA. Lewis et al³⁴ demonstrated that patients experiencing something unpleasant during the first night of CPAP therapy displayed worse adherence, despite not evaluating the sleep parameters. Taken together, our results demonstrating a correlation between self-reported sleep improvement and CPAP adherence appear reasonable. Moreover, the method of evaluating changes in the sleep questionnaires may also help predict CPAP adherence in patients diagnosed using a home-based monitoring device and treated with auto PAP, despite not undergoing PSG studies at a sleep center. Further, improvements in TST, sleep efficacy, and sleep architecture such as stage N1% led to better adherence by dividing the patients into 2 groups, namely “good” and “excellent” adherence. In addition, the sleep questionnaires reflected the differences in the sleep parameters between the groups. Inquiring about self-reported sleep status with and without CPAP could be a substitute for changes in the sleep parameters between dPSG and tPSG. Because these findings also indicated the importance of patients’ initial impression, optimal settings are required on the first night of CPAP.

BMI was negatively correlated with CPAP adherence in our study; nonetheless, higher BMI has been reported to be correlated with better CPAP adherence in previous studies.^35,36 One of the possible reasons for the inconsistency is that the Epworth Sleepiness Scale score was low at 8.6 ± 4.6 in our study, which suggests our patients were mainly nonsleepy patients. Obese patients with OSA display more drowsiness and a higher arousal threshold for airway stenosis, and the benefits of CPAP treatment may motivate them to continue CPAP therapy. Thus, our nonsleepy patients were unlikely to receive the benefits of CPAP treatment by improving sleepiness. In addition, age was positively correlated with CPAP adherence in this study. Factors such as insomnia, cognitive impairment, and various comorbidities are related to poor adherence in older adults; however, our older adult patients were mostly in their mid-50s and did not have these factors and were more likely to accept CPAP therapy.

This study had several limitations. First, we did not investigate the side effects of CPAP devices or social factors, such as socioeconomic status and partner involvement, that may have affected CPAP adherence. Second, there may be some biases that resulted in relatively higher CPAP adherence in this study than that in previous similar studies. This may be attributed to several reasons. First, people are recognizing the importance of treatment for OSA, and Japanese patients are likely to display better adherence owing to their diligence. For example, a recent prospective multicenter randomized study has shown an adherence of approximately 70%. Second, only patients who underwent both dPSG and tPSG were enrolled in this study. We did not perform tPSG in 297 patients for several reasons, such as limited finances, busy schedule, and distant hospital location, which may have affected adherence. Third, we performed a retrospective study vulnerable to selection bias caused by unmeasured confounders, which could have caused unexpected differences among the groups. Because existing clinical data were used in this retrospective study, our study is not as accurate as a prospective, laboratory-based study. Finally, because our tPSG data were pre-CPAP initiation data, the present study has limitations in determining the effect of CPAP on sleep architecture. To better understand how changes in sleep architecture affect CPAP adherence, tPSG study after 3 months of CPAP initiation should be performed.

In conclusion, our findings suggest the difficulty of predicting adherence from dPSG data alone and that tPSG should be performed for evaluating an improvement in sleep parameters with CPAP to predict its adherence. Furthermore, merely asking patients about their sleep quality with and without CPAP could be a substitute for changes in the PSG parameters and a convenient predictor of CPAP adherence even if they do not undergo tPSG. Thus, our results demonstrate that tPSG and the questionnaire can facilitate identifying the patients at risk for poor CPAP adherence.

DISCLOSURE STATEMENT

All authors have seen and approved the manuscript. Work for this study was performed at the sleep center at Saitama Medical University Hospital. This was not an industry-supported study. The authors report no conflicts of interest.

ABBREVIATIONS

AHI

apnea-hypopnea index

BMI

body mass index

CPAP

continuous positive airway pressure

dPSG

diagnostic polysomnography

IQR

interquartile range

OSA

obstructive sleep apnea

PSG

polysomnography

SpO₂

arterial oxyhemoglobin saturation measured by pulse oximetry

tPSG

titration polysomnography

TST

total sleep time