There is randomized controlled evidence that patients with sleep apnea–hypopnea syndrome (SAHS) treated with continuous positive airway pressure (CPAP) have improved daytime function and quality of life. “Before-and-after” data indicate that CPAP improves sleep quality, but there is no randomized controlled evidence of this. We tested the hypothesis that CPAP improved sleep quality in patients with SAHS. We also sought correlations between polysomnographic (PSG) indices and benefit from use of CPAP. Twenty-two patients with SAHS (median, [interquartile range] apnea–hypopnea index, 40 [25–65] events/h) spent 1 mo receiving CPAP and 1 mo receiving placebo capsules, in random order, and home sleep studies were performed at the end of each month. CPAP resulted in a lower arousal index, less Stage 1, and more Stage 3 + 4 sleep, but there was no increase in Stage REM (rapid eye movement) sleep. Improvement in Epworth score after 6–12 mo of treatment correlated significantly with PSG indices on CPAP: sleep efficiency index (r = 0.78, p < 0.001), Stage REM (r = 0.55, p = 0.001), Stage 3 + 4 (r = 0.5, p = 0.02), and arousal index (r = − 0.43, p = 0.05). This study provides randomized controlled trial evidence that patients with SAHS sleep better when receiving CPAP. Patients with good sleep quality on CPAP at 1 mo are likely to gain later benefit in subjective daytime sleepiness.
Keywords: sleep apnea syndromes; continuous positive airway pressure; sleep architecture
The current preferred treatment for the sleep apnea–hypopnea syndrome (SAHS) is continuous positive airway pressure (CPAP) (1). There is now good randomized placebo-controlled evidence showing that CPAP improves SAHS symptoms (especially sleepiness), mood, cognitive function, and quality of life (2-5). There is also evidence that CPAP may improve cardiovascular outcomes (6). Although several uncontrolled studies (7-10) have shown that CPAP can improve sleep quality, there is no randomized placebo-controlled evidence of this. A Medline and Embase literature search found only one placebo-controlled study; this study did not show any difference in sleep indices between the CPAP and placebo treatment limbs (11).
Furthermore, our understanding of the factors that determine benefit from CPAP and determine CPAP use is poor, and pretreatment predictors that have been identified are weakly associated with these outcomes (12-15). Polysomnographic (PSG) variables may be important in determining which patients benefit from and use CPAP.
We have examined the hypothesis that patients with SAHS treated with therapeutic CPAP would have improvement in key PSG variables and subjective sleep quality. The secondary aims of the study were to evaluate long-term nightly CPAP use and subjective sleepiness and to examine PSG variables and subjective sleep quality as predictors of long-term nightly CPAP use and subjective sleepiness.
Consecutive patients with SAHS booked for CPAP treatment were recruited, using the following criteria.
Inclusion. Patients with two or more symptoms of SAHS (16) and
Apnea–hypopnea index (AHI) > 15/h (electroencephalogram [EEG]-based study) or
Apneas and hypopneas > 30/h recorded (non-EEG-based study )
Apneas and hypopneas were defined by our usual criteria (18).
Medication or coexisting disorder likely to disturb sleep quality
Consumption of > 210 g of alcohol per week
Driving problems due to sleepiness
Randomization occurred the morning after CPAP titration, using a random number table and sealed envelopes. Patients received either 1 mo of placebo capsule (lactose; Nova Laboratories, Wigston, UK) or 1 mo of CPAP treatment. The treatment intervention was reversed during the subsequent month. Block randomization was used to balance the order of intervention. CPAP treatment used a Sullivan Elite machine (ResCare, Abingdon, UK). Placebo capsules were prescribed at one per evening and patients were told that the capsule treatment might improve upper airway muscle function in sleep.
At the end of each intervention patients had full PSG monitoring of their sleep in the home, using the Compumedics P-series portable system (Compumedics Sleep, Abbotsford, Australia). PSG monitoring was started at the patient's usual bedtime and stopped at the usual wake time. This total recording time (TRT) was kept the same, as was the time of the week on both study limbs. Patients were asked not to consume alcohol or caffeinated drinks for 6 h before overnight monitoring.
Outcomes included feeling refreshed in the morning (five-point Likert Scale), subjective sleepiness (Epworth Sleepiness Scale ), and median nightly CPAP use. Outcomes were assessed at the end of each study month and at clinic follow-up after 6–12 mo of CPAP.
Benefit to patients on CPAP compared with placebo was assessed by subtracting scores on CPAP (at the end of each study month and at 6–12 mo) from placebo. To explore relationships between sleep architecture and outcomes four sleep indices were selected a priori: the Sleep Efficiency Index (SEI) (total sleep time [TST] per total recording time [TRT]), the arousal index (number of EEG arousals per hour slept), Stage 3+4, and Stage REM (rapid eye movement; minutes).
Sleep studies were analyzed by the researcher, after blinding of data by our usual criteria (20-22). The quality of signals was assessed on the basis of criteria from the Sleep Heart Health Study (SHHS) (5).
A power calculation with α = 0.05 showed we required 22 patients for a power of 90% (23). Parametric and nonparametric statistics were used. The general linear model was used with repeated measures to assess treatment changes for these variables; potential treatment × treatment order interactions were examined with order as a between-subject variable (24). Order effects were also assessed (25). Stepwise multiple linear regression was used to assess the independent role of predictor variables.
The study had the approval of the ethics subcommittee. All patients gave written informed consent. Subjects were debriefed after the study was completed.
Fifty-eight patients were invited to participate; 23 started and 22 completed the study (1 patient withdrew shortly after enrollment and refused to continue CPAP treatment). Thirty-five patients declined and they were similar to those agreeing to the study on common demographic and disease-related variables (Table 1). The median (IQR) closed airway CPAP titration AHI of participants was 3.4 (2.3–7.4)/h. On the basis of the criteria used in the Sleep Heart Health Study (5), 40 home sleep studies were of “outstanding” quality and 3 were “excellent.” In the remaining study, the battery ran out after 4 h and only the first 4 h of data were compared.
|Patient Characteristic||Study Patients (n = 23)||Study Refusers (n = 35)||p Value|
|Mean age, yr (SD)||53 (11)||51 (10)||0.4|
|Number of males||20||29||0.7|
|Mean BMI (SD)||31 (5)||31 (5)||0.8|
|Median Epworth score (IQR)||14 (10-17)||12 (8-16)||0.3|
|Median AHI (IQR), events/h||40 (25–65)||44 (32–56)||0.7|
|Mean CPAP pressure (SD), cm H2O||10 (3)||9 (2)||0.2|
On the CPAP night there was less Stage 1, more Stage 3+4, and a lower arousal index than on the placebo night (Table 2). Neither SEI nor Stage REM were significantly different between placebo and CPAP nights (Table 2). Analysis of normally distributed data (Stage REM) did not show any evidence of order (acclimatization) effects (CPAP − placebo; order 1 = 11 [± 38.3], order 2 = 12.4 [± 32.5] min, p = 0.9) or of treatment × treatment order interactions. There were also no order effects for nonparametric data. Patients were more refreshed after the CPAP night than the placebo night (median [IQR] refreshed: CPAP = 4 [3–5], placebo = 3 [2–3.25], p = 0.002). The median (IQR) nightly CPAP use during the study was 4.5 (2.6–6.2) h/night. The median (IQR) CPAP use on the study night of 7.2 (6.4–7.9) h/night was higher than the median nightly use for the month (p < 0.001).
|PSG Variable||Placebo||CPAP||p Value|
|Mean total recording time (± SD), min||489 (± 45)||489 (± 45)||1|
|Median sleep latency (IQR), min||19 (12-32)||15 (9-28)||0.6|
|Median awake time in sleep period (IQR), min||81 (45–113)||59 (34–95)||0.4|
|Median sleep efficiency index (IQR), %||77 (69–81)||80 (69–85)||0.4|
|Median latency to Stage 3/4 (IQR), min||22 (13-34)||18 (13-24)||0.4|
|Median Stage 2 (IQR), min||239 (173–258)||213 (186–251)||0.3|
|Median latency to stage REM (IQR), min||79 (57–129)||66 (55–95)||0.2|
|Mean Stage REM (± SD), min||74 (±29)||86 (±30)||0.1|
|Median Stage 1 (IQR), min||24 (14-34)||12 (6-19)||0.03|
|Median Stage 3/4 (IQR), min||23 (12–39)||41 (30–67)||0.007|
|Median arousal index (IQR), events/h||45 (32–77)||21 (17-31)||< 0.001|
Follow-up outcomes, at 6–12 mo (median [IQR] = 7.5 (6-9) mo), were available for all participants. The median (IQR) CPAP use at 6–12 mo of 3.8 (0.8–5.4) h/night was similar to that during the study month (p = 0.1). The median (IQR) Epworth score at the end of the study month on placebo was 12.5 (7.8–15.5), with lower values at the end of the CPAP study month of 6 (4.8–11) (p < 0.001) and at the 6 to 12 mo follow-up of 6.5 (4.8–10.8) (p = 0.002).
The change in PSG variables on the study nights correlated moderately strongly with the change in feeling refreshed the morning after these nights (all variables; r = 0.4–0.5, p ⩽ 0.02). Change in feeling refreshed correlated with the change in Epworth score at follow-up (r = 0.63, p = 0.002) but did not correlate with CPAP use at 6–12 mo.
The SEI on the CPAP night was strongly correlated with change in Epworth score at follow-up (Figure 1) and was correlated with 6 to 12 mo CPAP use (Figure 2). Other CPAP study night PSG variables correlated with change in Epworth score at follow-up (Figures 3-5). Sleep architecture variables on CPAP were associated with each other; a Spearman rank correlation matrix shows that SEI was correlated with all of the other sleep architecture variables (for all: magnitude of r = 0.44–0.53, p < 0.05) and arousal index was correlated with Stage 3+4 (r = −0.46, p = 0.033). Multiple linear regression showed only SEI on CPAP had an independent predictive role.
The only correlations between changes in PSG variables and 6 to 12 mo outcomes were between change in arousal index and change in Stage 3+4 with change in Epworth at follow-up (Stage 3+4: r = 0.54, p = 0.01; arousal index: r = 0.45, p = 0.03). There were no significant correlations between PSG variables on placebo and 6 to 12 mo outcomes.
This is the first randomized placebo-controlled trial study to show that patients with SAHS have improved sleep architecture after 1 mo of CPAP treatment. CPAP treatment resulted in better sleep quality with fewer arousals, less light (Stage 1) sleep, and more deep (Stage 3+4) sleep. Although unsurprising, this adds important new randomized placebo-controlled evidence of efficacy of CPAP in SAHS. The study also showed that median nightly CPAP use at 6–12 mo had little correlation with PSG variables. Commonly used sleep indices on CPAP (assessed after 1 mo) were related to long-term benefit as assessed by change in Epworth score after 6–12 mo of CPAP from that on placebo. The sleep efficiency during CPAP was the best predictor of long-term benefit from CPAP and was also related to median nightly CPAP use at 6–12 mo.
Our data on the effect of CPAP on sleep quality confirm and extend a number of studies that have measured PSG sleep indices before and after patients with SAHS started CPAP treatment. These found an increased time in Stage 3+4 (7-9) and Stage REM (8-10), and reduced time in Stage 1 (7-10); while some found increased Stage 2 (8) and reduced Stage REM (7) and 3+4 sleep latency (7). However, such “before-and-after” studies do not allow for first-night effects, which can mean that sleep quality improves after the first study night because of acclimatization (26). Also, placebo effects are not taken into account in “before-and-after” studies. Our randomized placebo-controlled study has now confirmed some of these changes but found no significant increase in REM sleep. The changes in Stage REM reported in “before-and-after” studies were often small (9) and acclimatization effects may be the explanation for this earlier finding. A rebound increase in Stage REM immediately after CPAP treatment for SAHS has been described (27), which was not picked up in our assessment after 1 mo of CPAP treatment. Alternatively, REM sleep suppression in the untreated state could be a function of SAHS severity; our patients had a median AHI of 40 per hour, which was appreciably lower than the respiratory disturbance indices of study patients who have shown more Stage REM on CPAP (8-10). One study of first-night effects (25) found Stage REM latency decreased on second and third nights, possibly contributing to the earlier results for REM latency.
The only other placebo-controlled trial of the effect of CPAP on sleep quality that we can find showed no differences in sleep after 7 d of therapeutic compared with subtherapeutic CPAP (11). Significant changes with time, but not with treatment, were found in that study (11), and these are likely to reflect first-night effects. In the current study all patients had had previous (in-laboratory) sleep studies and assessments were done by home studies, which have small first-night effects (28, 29). Furthermore, we reduced potential bias from possible first-night effects by using balanced blocks in the randomization of the order of intervention and we found no evidence of first-night (acclimatization) effects on our data. Loredo and colleagues (11) used repeated measures in two different groups of patients whereas in the current study each patient acted as his or her own control, reducing variability from between-group differences. Although the number of patients assessed in this earlier study was similar to the current study the increased statistical power of the cross-over design may have been important. Loredo and colleagues (11) assessed changes after 7 d, which might possibly be before sleep benefits are maximal.
Assessments of feeling refreshed the morning after CPAP and placebo nights were used as a measure of the quality of the previous night's sleep. This measure was used to try to examine the effect of interventions on study nights, unaffected by the previous CPAP use pattern. In this study nearly all patients used their CPAP treatment on the monitoring night as prescribed (i.e., “all night”) whereas use over the whole study month was more variable. When patients used CPAP as suggested, the feeling of being refreshed in the morning was better on CPAP than placebo. Changes in feeling refreshed from the morning after CPAP compared with placebo also predicted long-term benefit from CPAP (i.e., in subjective daytime sleepiness). Further, feeling refreshed in the morning related to all four commonly used sleep indices to a similar degree, suggesting that feeling refreshed in the morning is a reasonable gauge of the sleep architecture during the previous night. Improvement in SAHS symptoms on starting CPAP can be dramatic (27) and, although not investigated in this study, feeling refreshed in the morning after initial CPAP titration may provide an early clue to future benefit.
Two clinical outcomes were measured after 6–12 mo of CPAP treatment. One of these, median nightly CPAP use, is a widely used measure of CPAP compliance, but there may be considerable individual variability in the nightly use needed to gain benefit (30, 31). Median nightly CPAP use was similar to previous values from our center (2) and to values in the literature (14, 15). We also assessed the change in Epworth score between placebo and the score at 6–12 mo as a more direct measure of long-term benefit from CPAP treatment. There were large improvements in daytime sleepiness seen after 1 mo of CPAP treatment and sustained at 6–12 mo. Sleep quality indices on CPAP nights were the best predictors of this long-term benefit from CPAP, with the SEI showing a strong relationship. Surprisingly, changes in sleep quality between placebo and CPAP were less well correlated with change in Epworth score at follow-up, indicating that actual sleep quality on CPAP has a more important effect on long-term benefit than the change in sleep quality with treatment. Some studies have shown that automatic CPAP results in better clinical outcomes than fixed pressure CPAP, and in some of these (32-34) the automatic CPAP treatment was also associated with improved sleep quality. This is in keeping with our findings that good sleep quality on CPAP treatment is associated with subsequent benefit from treatment; however, we have not shown causality. If there is a robust causal link between sleep quality on CPAP and short- and long-term benefit, then developing CPAP machines that minimize sleep disturbance should improve outcomes for patients with SAHS. Sleep indices on placebo did not predict any subsequent treatment outcomes and this may be because sleep quality on placebo is influenced by many factors other than SAHS. Apart from SEI on the CPAP study night, no other sleep indices correlated with 6–12-mo median nightly use of CPAP, emphasizing that different factors seem to determine nightly use compared with those determining long-term benefit in daytime sleepiness.
Sleep architecture relates to sleep quality in normal subjects (35) and to objective sleepiness in those with sleep disorders (36). Further, among patients with SAHS the arousal index has an inverse correlation with the amount of Stage 3+4 sleep and stage REM sleep (expressed as a percentage of total sleep time, TST) (9). Experimental sleep fragmentation (22) induces sleep architecture changes of decreased Stage 3+4, Stage REM and a trend to increased wakefulness. We found that the sleep architecture variables on CPAP were frequently correlated with each other. The interrelationship of sleep architecture variables makes it difficult to determine which of these variables are most likely to be of pathogenic significance. Although arousal frequency is often considered to have an important role in the pathogenesis of daytime sleepiness we found that other sleep architecture variables had stronger associations with long-term benefit in daytime sleepiness from CPAP treatment and only SEI had an independent effect.
Potential limitations of this study include the use of a capsule as a placebo and the increased CPAP use on the night of study. It has been argued that a capsule placebo may not be the most appropriate placebo for CPAP treatment (37). We believe that a placebo capsule is a valid placebo provided it is actively sold to patients as a possible therapeutic agent, as in this study. Subtherapeutic CPAP is another usable placebo, but we remain concerned that it could have worsened sleep quality (due to sleeping with an ineffective encumbrance on one's face), particularly in patients with less severe SAHS, and this could have led, by itself, to a spurious finding of improved sleep architecture from therapeutic CPAP. Although we studied patients between their reported usual bedtime and usual wake time on both study limbs, these sleep times were not confirmed to be typical for each individual by other means, such as by a sleep diary. In this study we allowed 1 mo for patients to acclimatize to CPAP, and all patients had at least 25 h of use before assessment. The improved sleep quality found applies to nights of generally good CPAP use; the benefit on other nights of lesser use may be less. However, the study clearly shows that improved sleep quality is attainable with CPAP.
We have shown, in a randomized placebo-controlled trial, that patients with SAHS have improved sleep quality on CPAP, extending the growing randomized controlled trial evidence of CPAP benefit. Further, those that have good sleep architecture on CPAP will report subsequent benefit in daytime sleepiness.
The authors thank the administrative, technical, nursing, and research staff at the Edinburgh Sleep Center for assistance they provided during this study.
|1.||Douglas NJSystematic review of the efficacy of nasal CPAP. Thorax531998415|
|2.||Engleman HM, Martin SE, Kingshott R, Mackay T, Deary IJ, Douglas NJRandomised placebo controlled trial of daytime function after continuous positive airway pressure (CPAP) therapy for the sleep apnea/hypopnea syndrome. Thorax531998341345|
|3.||Engleman HM, Kingshott R, Wraith PK, Mackay R, Deary IJ, Douglas NJRandomized placebo-controlled crossover trial of continuous positive airway pressure for mild sleep apnea/hypopnea syndrome. Am J Respir Crit Care Med1591999461467|
|4.||Jenkinson C, Davies ROJ, Mullins R, Stradling JRComparison of therapeutic and subtherapeutic nasal continuous positive airway pressure for obstructive sleep apnoea: a randomised prospective parallel trial. Lancet353199921002105|
|5.||Redline S, Sanders MH, Lind BK, Quan SF, Iber C, Gottlieb DJ, Bonekat WH, Rapoport DM, Smith PL, Kiley JPMethods for obtaining and analysing unattended polysomnography data for a multicentre study. Sleep211998759767|
|6.||Dimsdale JE, Loredo JS, Profant JEffect of continuous positive airway pressure on blood pressure: a placebo trial. Hypertension352000144147|
|7.||Fietze I, Quispe-Bravo S, Hansch T, Rottig J, Baumann G, Witt CHArousals and sleep stages in patients with obstructive sleep apnoea syndrome: changes under nCPAP treatment. J Sleep Res61997128133|
|8.||Lamphere J, Roehrs T, Wittig R, Zorick F, Conway WA, Roth TRecovery of alertness after CPAP in apnea. Chest96198913641367|
|9.||Collard P, Dury M, Delguste P, Aubert G, Rodenstein DOMovement arousals and sleep-related disordered breathing in adults. Am J Respir Crit Care Med1541996454459|
|10.||Bonsignore G, Marrone O, Bellia V, Giannone G, Ferrara G, Milone FContinuous positive airway pressure improves the quality of sleep and oxygenation in obstructive sleep apnea syndrome. Ital J Neurol Sci81987129134|
|11.||Loredo JS, Ancoli-Israel S, Dimsdale JEEffect of continuous positive airway pressure vs placebo continuous positive airway pressure on sleep quality in obstructive sleep apnea. Chest116199915451549|
|12.||McArdle N, Devereux G, Heidarnejad H, Engleman HM, Mackay T, Douglas NJLong-term use of CPAP therapy for sleep apnea/hypopnea syndrome. Am J Respir Crit Care Med15919991114|
|13.||Krieger J, Kurtz D, Petiau C, Sforza E, Trautmann DLongterm compliance with CPAP therapy in obstructive sleep apnea patients and in snorers. Sleep191996S136S143|
|14.||Reeves-Hoche MK, Meek R, Zwillich CWNasal CPAP: an objective evaluation of patient compliance. Am J Respir Crit Care Med1491994149154|
|15.||Pieters T, Collard P, Aubert G, Dury M, Delguste P, Rodenstein DOAcceptance and long-term compliance with nCPAP in patients with obstructive sleep apnoea syndrome. Eur Respir J91996939944|
|16.||Whyte KF, Allen MB, Jeffrey AA, Gould GA, Douglas NJClinical features of the sleep apnea/hypopnea syndrome. Q J Med721989666|
|17.||Rees K, Wraith PK, Berthon-Jones M, Douglas NJDetection of apneas, hypopneas and arousals by the AutoSet in the sleep apnea/hypopnea syndrome. Eur Respir J121998764769|
|18.||Douglas NJ, Thomas S, Jan MAClinical value of polysomnography. Lancet3391992350|
|19.||Johns MWA new method for measuring daytime sleepiness: the Epworth Sleepiness Scale. Sleep141991540545|
|20.||Rechtschaffen A., Kales A. A manual of standardized terminology and scoring system for sleep stages of human subjects. NIH Publication No. 204. Washington, DC: U.S. Government Printing Office; 1968.|
|21.||Cheshire K, Engleman HM, Deary IJ, Shapiro CM, Douglas NJFactors impairing daytime performance in patients with the sleep apnea/ hypopnea syndrome. Arch Intern Med1521992538541|
|22.||Martin SE, Engleman HM, Deary IJ, Douglas NJThe effect of sleep fragmentation on daytime function. Am J Respir Crit Care Med15319961332|
|23.||Florey Cdu V. Sample size for beginners. Br Med J306199311811184|
|24.||Hills M, Armitage PThe two period crossover trial. Pharmacology81979720|
|25.||Altman DG. Clinical trials. Practical statistics for medical research. London: Chapman & Hall; 1991. p. 467–471.|
|26.||Toussaint M, Luthringer R, Schaltenbrand N, Carelli G, Lainey E, Jacqmin A, Muzet A, Macher J-PFirst-night effect in normal subjects and psychiatric inpatients. Sleep181995463469|
|27.||Sullivan CEObstructive sleep apnoea. Clin Chest Med61985633650|
|28.||Coates TJ, George JM, Killen JD, Marchini E, Hamilton S, Thorensen CEFirst night effects in good sleepers and sleep-maintenance insomniacs when recorded at home. Sleep41981293298|
|29.||Sharpley AL, Solomon RA, Cowen PJEvaluation of first night effect using ambulatory monitoring and automatic sleep stage analysis. Sleep111988273276|
|30.||Engleman HM, Martin SE, Douglas NJCompliance with CPAP therapy in patients with the sleep apnoea/hypnoea syndrome. Thorax491994263266|
|31.||Engleman HM, Martin SE, Deary IJ, Douglas NJEffect of continuous positive airway pressure treatment on daytime function in sleep apnoea/hypopnoea syndrome. Lancet3431994572575|
|32.||Konerman M, Sanner BM, Vyleta M, Laschewski F, Groetz J, Sturm A, Zidek WUse of conventional and self-adjusting nasal continuous positive airway pressure for treatment of severe obstructive sleep apnea syndrome. Chest1131998714718|
|33.||Randerath WJ, Parys K, Feldmeyer F, Sanner B, Ruhle KHSelf- adjusting nasal continuous positive airway pressure therapy based on measurement of impedance: a comparison of two different maximum pressure levels. Chest1161999991999|
|34.||Scharf MB, Brannen DE, McDannold MD, Berkowitz DVComputerized adjustable versus fixed NCPAP treatment of obstructive sleep apnea. Sleep191996491496|
|35.||Akerstedt T, Hume K, Minors D, Waterhouse JGood sleep—its timing and physiological characteristics. J Sleep Res61997221229|
|36.||Stepanski E, Lamphere J, Badia P, Zorick F, Roth TSleep fragmentation and daytime sleepiness. Sleep719841826|
|37.||Wright J, Johns R, Watt I, Melville A, Sheldon THealth effects of obstructive sleep apnoea and the effectiveness of continuous positive airways pressure: a systematic review of the research evidence. Br Med J3141997851860|
N. McArdle was supported by funding from the San Diego Foundation.
N. J. Douglas is a consultant for ResMed, Ltd., a commercial CPAP company, where he acts in an advisory capacity only.