Randomized Controlled Trial Comparing Flexible and Continuous Positive Airway Pressure Delivery: Effects on Compliance, Objective and Subjective Sleepiness and Vigilance

Abstract

Study Objectives:

Positive airway pressure (PAP) treatment for obstructive sleep apnea (OSA) can be limited by suboptimal compliance. C-Flex technology (Philips Respironics, PA, USA) reduces pressure during expiration, aiming to improve comfort and therefore compliance. This may be of particular relevance to patients requiring high pressures. Many studies thus far have suffered from design limitations and small sample sizes. This study aimed to compare compliance with C-Flex and CPAP, as well as analyzing objective and subjective sleepiness and vigilance.

Design:

Three-month, double-blinded, parallel-arm randomized controlled trial.

Setting:

A university-based sleep laboratory.

Patients:

76 consecutive patients with severe OSA (mean ± SD AHI 60.2 ± 32.9 events/hour, ESS 13.6 ± 4.5/24, BMI 35.6 ± 7.8 kg/m²), without significant cardiac, respiratory, psychiatric, or sleep comorbidities.

Interventions:

Patients were randomized to C-Flex (dip level 2) or CPAP.

Measurements and Results:

Patients underwent titration with C-Flex/CPAP (mean pressure 11.6 cm H₂O). Modified maintenance of wakefulness tests (mod-MWT), psychomotor vigilance tasks (PVT) and questionnaires were administered at baseline and after one and 3 months. Median compliance was 5.51 and 5.89 h/ night in the C-Flex and CPAP groups respectively (P = 0.82). There were no significant differences between groups in terms of PVT reaction time, subjective sleepiness, sleep quality, health-related quality of life, or treatment comfort. There was no significant difference between the groups regarding the change in mod-MWT sleep latency values.

Conclusions:

In patients with severe OSA both CPAP and C-Flex resulted in substantial improvements in sleepiness, vigilance, and quality of life. The use of C-Flex did not result in greater compliance, and neither treatment appeared superior.

Citation:

Bakker J; Campbell A; Neill A. Randomized controlled trial comparing flexible and continuous positive airway pressure delivery: effects on compliance, objective and subjective sleepiness and vigilance. SLEEP 2010;33(4):523-529.

OBSTRUCTIVE SLEEP APNEA (OSA) IS A SERIOUS CONDITION CHARACTERIZED BY LOUD SNORING AND FREQUENT PARTIAL OR COMPLETE BLOCKAGE of the upper airway during sleep. Affecting approximately 2% to 4% of the adult population,^1,2 OSA is associated with daytime hypersomnolence,^1,3 neurocognitive impairment,⁴ and development of cardiovascular disease.⁵ First-line therapy for moderate-to-severe OSA is positive airway pressure (PAP) therapy, which was developed with the principle of preventing upper airway collapse using positive pressure as a splint.⁶ PAP is able to successfully eliminate the symptoms of OSA,^7–12 and there is now a large amount of evidence that PAP is able to improve daytime consequences of OSA, as well as long-term cardiovascular outcomes.^10,13–18 However, the effectiveness of PAP is often limited by suboptimal compliance.^19,20 Definitions as to what level of PAP use is optimal change depending on the outcome measure in question,^21–23 but the literature most often uses the somewhat arbitrary value of 4 h/night.

Modifications of conventional continuous PAP (CPAP) have been developed with the aim of increasing comfort and therefore compliance. A major complaint from CPAP patients is the difficulty of exhaling against incoming pressure, so flexible pressure (C-Flex, Philips Respironics, PA, USA) was developed, which reduces the delivered pressure at the start of exhalation and increases the pressure to therapeutic level for the latter part of exhalation and subsequent inhalation. The extent of pressure relief is proportional to expiratory flow, and the resulting delivered pressure pattern is scalloped rather than constant, with a lower overall mean pressure. C-Flex and the automatically titrating version, A-Flex, have been utilized in a number of trials reporting compliance,^24–29 only one of which has shown C-Flex to have significantly higher compliance than CPAP using intention-to-treat analysis.²⁴ This trial was limited by non-randomization and non-blinding, suggesting a placebo effect. Only one previously published trial included a range of secondary outcome measures including an objective measure of sleepiness: this was a short-term pilot randomized controlled trial (RCT) undertaken in our laboratory which indicated that patients with severe OSA may use C-Flex more than CPAP but advised that this be the focus of a larger trial.²⁸ We therefore aimed to evaluate the effectiveness of C-Flex versus CPAP primarily by analyzing mean objective compliance over one and 3 months in all patients, as well as in sub-sets of compliant patients and those requiring high pressure delivery. Additionally, we compared key therapeutic outcome measures to determine the effectiveness of therapy: objective sleepiness (modified Maintenance of Wakefulness test [mod-MWT])^30–32 and vigilance (Psychomotor Vigilance Task [PVT]),^33,34 subjective sleepiness (Epworth Sleepiness Scale [ESS],³⁵ Stanford Sleepiness Scale [SSS]),³⁶ sleep quality (Pittsburgh Sleep Quality Index [PSQI]),³⁷ Functional Outcomes of Sleep Questionnaire [FOSQ]),³⁸ and health-related quality of life (Short-Form 36 Health Survey [SF-36]).³⁹

MATERIALS AND METHODS

Ethical approval was granted by the Central Region Ethics Committee, and all patients gave written informed consent before commencement of the trial. The clinical trial was registered with the Australia and New Zealand Clinical Trials Registry (12607000359437).

The study was a double-blinded, parallel-arm RCT of C-Flex versus CPAP (Philips Respironics RemStar Pro with C-Flex, set to dip level 2 if randomized to C-Flex) with data collection occurring at baseline and after one and 3 months of treatment. Patients were not able to access the C-Flex menu option, and all references to “C-Flex” on the device were covered. The data analyst remained blinded by random 3-digit codes being assigned to all patients.

A pre-trial power calculation indicated that 76 patients would be required to give 80% likelihood of detecting a significant mean difference in compliance of 1.3 h (0.05 α), based on the variability in compliance found in our pilot study.²⁸ English-speaking, PAP-naöive, adult patients referred by respiratory physicians with suspected moderate-to-severe OSA (apnea hypopnea index [AHI] ≥ 30, or ≥ 20 with an ESS score ≥ 12) measured during a full-night diagnostic polysomnography (PSG; S-series or Somte, Compumedics, Victoria, Australia) at an accredited sleep laboratory were recruited from the Greater Wellington region, excluding those with significant cardiac, respiratory, psychiatric, or sleep comorbidities, and those with irregular sleep patterns. Randomization to treatment was performed prior to manual titration using a (1, 2) urn randomization procedure,^40,41 with all patients titrated using their allocated treatment (pressure was titrated to overcome obstruction and all flow limitation).

All studies were performed and scored by New Zealand registered polysomnographers according to Rechtschaffen and Kales sleep staging criteria,⁴² ASDA arousal criteria,⁴³ and the Chicago criteria for scoring respiratory events.⁴⁴ Studies were scored by a single polysomnographer and checked by the laboratory manager, with the referring physician also checking the final titrated pressure.

The primary outcome measure was objective compliance (reported as total h of use per 24-h period) measured using the internal pressure sensor in the C-Flex devices. Secondary outcome measures were mean sleep latency measured during the mod-MWTs (performed twice in the afternoon, 2 hours apart), simple reaction time and number of lapses (reaction time ≥ 500 ms) measured during the PVTs (also performed twice in the afternoon, 2 h apart), and 6 questionnaires—the ESS, SSS, PSQI, FOSQ, SF-36 (8 subscales), and an in-house treatment comfort questionnaire. The treatment comfort questionnaire is used routinely in the current laboratory, though it has not been validated, and addresses subjective sleep quality, feeling of being refreshed on waking, and daytime alertness compared to pre-PAP sleep, as well as an overall rating of comfort of the device/mask, and 5-point Likert scales relating to the level of annoyance of 11 side-effects (including blocked nose, irritation of the skin on face, dry mouth and throat, and air leakage). A copy of this questionnaire is available from the corresponding author.

All patients received a phone call after one night of treatment at home, and attended a 2-week follow-up appointment with a trained physiologist in order to troubleshoot any problems encountered. No other advice or help likely to affect compliance was given to patients unless instigated by the patient.

All analyses were performed using SPSS (Version 15.0; IL, USA). Continuous data were analyzed for normality of distribution and homogeneity of variance where appropriate. Between-arm data were analyzed using independent-samples t-tests (parametric; means indicated below) or Mann-Whitney tests (nonparametric; medians indicated below); within-arm data were analyzed using dependent-samples t-tests (parametric) or Wilcoxon matched-pairs (signed-rank) tests (nonparametric); categorical between-arm data were analyzed using Pearson χ² tests. All statistical tests were considered significant when P ≤ 0.05. More details regarding the study procedure and analyses are available from the corresponding author.

RESULTS

Study Sample

As shown in Figure 1, 751 OSA patients were referred between February 2007 and September 2008, with 627 of these meeting exclusion criteria (of these, 32% had comorbidity; 19% were unsuitable because of previous experience with PAP, living outside the catchment area, or inability to read/write in English; 17% had mild OSA; and 16% were referred by non-respiratory physicians). Of those who met inclusion criteria and agreed to participate, 43 were randomized to C-Flex and 37 were randomized to CPAP. Two patients were withdrawn (one because of the extent of neurocognitive impairment not becoming apparent until the first day of data collection, and one because of Cheyne-Stokes breathing during the manual titration), and 2 patients dropped out of the CPAP group immediately following titration (no compliance data was available for these patients; they were replaced with additional recruitment, and therefore analysis was conducted on a per-protocol basis). Table 1 shows the clinical and descriptive characteristics of the patient sample. Patients were typical for a group with moderate-to-severe OSA, being predominantly male, middle-aged, and obese. The 2 groups were equivalent at baseline, with the only significant difference occurring comparing the periodic limb movement (PLM) index, though this was not considered clinically significant, as the figures for both groups were well below the threshold for PLM syndrome.

Figure 1.

Trial flow chart.

Table 1.

Clinical characteristics of the C-Flex and CPAP groups at baseline

	Total Sample (n = 76) Mean (SD)	C-Flex (n = 39) Mean (SD)	CPAP (n = 37) Mean (SD)	P-value between arms (2-tailed)
Number of females (%)	23 (30%)	11 (28%)	12 (32%)	0.80
Age at consent (years)	49.7 (range 30-77)	49 (range 32-73)	50 (range 30-77)	0.59
BMI (kg/m2)	35.6 (7.8)	35.6 (7.4)	35.6 (8.4)	1.0
Diagnostic sleep efficiency (%)	77.7 (12.8)	79.1 (10.7)	76.2 (14.7)	0.33
Diagnostic mean O2 desaturation (%)	6.5 (3.6)	6.7 (3.5)	6.4 (3.7)	0.45
Diagnostic AHI (events/h)	60.2 (32.9)	61.1 (32.1)	59.3 (34.2)	0.77
Diagnostic arousal index (arousals/h)	47.3 (29.3)	47.0 (26.6)	47.7 (32.2)	0.90
Diagnostic PLM index (events/h)	2.0 (4.4)	0.5 (1.4)	3.6 (5.8)	0.01
Titrated pressure (cm H2O)	11.6 (3.2) (range 6–20)	11.7 (3.7) (range 6-20)	11.5 (2.7) (range 7-18)	0.79

SD, standard deviation; bold indicates a significant result (P ≤ 0.05)

Effects of Treatment on PSG Variables

C-Flex and CPAP had similar effects on polysomnographic indices of OSA comparing the diagnostic PSG and manual titration studies, with both treatments significantly lowering the AHI, arousal index, and mean O₂ desaturation (all P ≤ 0.05). There were no significant differences between the 2 groups in terms of these variables as well as mean sleep efficiency measured during the manual titrations (all P > 0.05) (results not shown). The mean residual AHI and leak levels (median and 90th percentile) interpreted by the M-Series devices were downloaded at the conclusion of the trial. There was no difference between the C-Flex and CPAP groups in terms of residual AHI (3.2 ± 1.9 and 3.2 ± 2.8 events/h, respectively; P = 0.57), 90th percentile leak (54.8 ± 19.5 and 59.8 ± 20.8 liters/min, respectively; P = 0.30), or average leak (43.7 ± 14.8 and 47.6 ± 16.8 events/h, respectively; P = 0.31).

Compliance

Figure 2 shows that over the whole 3-month study period (average 88 days in both groups), there was no significant difference in median objective compliance between the 2 treatment arms (C-Flex 5.51 h/night [IQR 3.46] and CPAP 5.89 h/night [IQR 2.75], U = 699.5, P = 0.82; both C-Flex and CPAP compliance distributions were significantly non-normal; D₃₉ = 0.16, P = 0.02 and D₃₇ = 0.15, P = 0.04, respectively). The same result was found when analyzing data from the first month only (average 30 and 31 days in the C-Flex and CPAP groups, respectively; n = 74 due to 2 patients not having compliance downloaded at this point; C-Flex median 5.59 h/night [IQR 3.43] and CPAP median 5.92 h/night [IQR 1.81]; U = 659.0, P = 0.80). There was no significant difference between the 2 groups in terms of the change in mean compliance between the 1- and 3-month appointments (U = 689.5, P = 0.74).

Figure 2.

Compliance with C-Flex and CPAP in all patients after three months of treatment (outliers are those which lie > 1.5 interquartile ranges lower than the lower quartile).

Sub-analyses of compliant patients and those requiring high pressures were performed, the cut-off levels for which were decided post hoc. There was no significant difference in compliance at 3 months after eliminating non-adherent patients (those using treatment < 1 h/night; C-Flex, n = 38, median compliance 5.52 [IQR 3.42] h/night; CPAP, n = 33, median compliance 5.98 [IQR 2.29] h/night; U = 546.5, P = 0.35). The same results were found after eliminating low-compliers (those using treatment < 4 h/night; C-Flex, n = 26, mean compliance 6.35 ± 1.0 h/night; CPAP, n = 28, mean compliance 6.18 ± 1.1 h/night; t₅₂ = 0.58, P = 0.57).

Similarly, when analyzing compliance at three months in patients requiring high pressures (≥ 12 cm H₂O) there was no significant difference between the 2 groups (C-Flex [n = 15], mean compliance 4.85 ± 2.2 h/ night; CPAP [n = 16], mean compliance 4.99 ± 2.1 h/night; t₂₉ = −0.18, P = 0.86). The same results were found when analyzing patients requiring very high pressures (≥ 14 cm H₂O; C-Flex [n = 10], mean compliance 5.22 ± 2.3 h/night; CPAP n = 8, mean compliance 5.95 ± 1.6 h/night; t₁₆ = −0.76, P = 0.46).

Vigilance, Sleepiness, and Questionnaire Data

Table 2 details the effects of C-Flex and CPAP on objective sleepiness and vigilance. There was no significant difference in PVT reaction time or number of lapses per test obtained at 1 month or 3 months (all P > 0.05). However, as shown in Figure 3 there was a significant difference between the 2 groups at 3 months in mod-MWT sleep latency (U = 519.5, P = 0.05), though this was not evident after the first month (U = 601.5, P = 0.25). Although both groups showed a similar rate of improvement on the mod-MWT over the first month, the C-Flex group declined over the following 2 months while the CPAP group retained the maximum possible score (2400 seconds); there was no significant difference between the groups regarding the change in sleep latency values between baseline and 3 months (P = 0.72).

Table 2.

Within- and between-arms analyses of objective sleepiness and vigilance

	C-Flex			CPAP			P-value between arms (2-tailed)
	BL Mean (SD)	1 Month Mean (SD)	3 Month Mean (SD)	BL Mean (SD)	1 Month Mean (SD)	3 Month Mean (SD)	BL	1 Month	3 Month
mod-MWT (/2400) (s)	1259.3 (746.6)	1810.4 (722.6)*	1806.3 (665.5)*	1432.7 (752.1)	2037.5 (514.1)*	2073.3 (541.2)*	0.27	0.25	0.05
PVT Reaction Time (s)	294.9 (81.5)	275.2 (58.6)**	262.5 (35.1)**	318.9 (128.0)	266.5 (60.7)**	254.5 (37.8)**	0.75	0.30	0.21
PVT Lapses per test	4.5 (8.1)	2.1 (3.7)**	1.2 (1.6)**	9.6 (21.2)	2.4 (5.9)**	1.0 (1.9)**	0.89	0.28	0.31

BL, baseline; SD, standard deviation;*symbol indicates significant change from BL;

^*P ≤ 0.05;

^**P ≤ 0.01 (1-tailed);^†symbol indicates significant change between 1 and 3 months; ^†P ≤ 0.05; ^††P ≤ 0.01 (1-tailed); bold indicates a significant result (P ≤ 0.05)

Figure 3.

Effects of C-Flex and CPAP on modified MWT sleep latencies (*indicates P ≤ 0.05 from baseline).

Tables 3 and 4 detail the effects of C-Flex and CPAP on sleep/CPAP-related questionnaires and the SF-36, respectively. The 2 groups were comparable at baseline on all measures except the FOSQ (significantly higher in the C-Flex group, P = 0.04). There was no significant difference between the 2 groups concerning the scores of any questionnaires administered post-treatment (all P > 0.05).

Table 3.

Within- between-arms analyses of subjective sleep/CPAP-related questionnaires

	C-Flex			CPAP			P-value between arms (2-tailed)
	BL Mean (SD)	1 Month Mean (SD)	3 Month Mean (SD)	BL Mean (SD)	1 Month Mean (SD)	3 Month Mean (SD)	BL	1 Month	3 Month
ESS (/24)	14.5 (4.3)	7.5 (3.7)**	6.0 (3.9)**††	12.7 (4.6)	7.6 (4.4)**	5.5 (3.6)**††	0.10	0.89	0.41
SSS (/7)	3.5 (1.3)	3.1 (1.2)*	2.5 (1.1)**††	3.2 (1.0)	2.7 (1.3)*	2.4 (1.2)**	0.16	0.12	0.74
PSQI (/21)	10.1 (3.2)	5.8 (3.5)**	4.8 (3.0)**	9.3 (3.6)	6.3 (3.3)**	5.8 (3.0)**	0.23	0.44	0.19
FOSQ (/20)	15.3 (2.5)	17.1 (2.2)**	18.3 (1.9)**††	14.0 (2.6)	17.6 (2.3)**	18.7 (1.4)**††	0.04	0.20	0.78
Treatment Comfort (/81)	not measured	60.4 (7.5)	64.1 (8.1) ††	not measured	58.7 (10.0)	62.2 (11.5) †	-	0.66	0.56

BL, baseline; SD, standard deviation;

^*symbol indicates significant change from BL; *P ≤ 0.05;

^**P ≤ 0.01 (1-tailed);

^†symbol indicates significant change between 1 and 3 months; †P = 0.05;

^††P = 0.01 (1-tailed); bold indicates a significant result (P = 0.05)

Table 4.

Within- and between-arms analyses of subjective health-related quality of life (SF-36 sub-scales) data

	C-Flex			CPAP			P-value between arms (2-tailed)
	BL Mean (SD)	1 Month Mean (SD)	3 Month Mean (SD)	BL Mean (SD)	1 Month Mean (SD)	3 Month Mean (SD)	BL	1 Mo	3 Mo
SF-36 Sub-scales:
Physical Function (/100)	72.2 (23.8)	80.6 (17.9)**	80.9 (18.3)**	70.1 (24.9)	73.1 (22.4)*	74.7 (22.5)*	0.73	0.15	0.21
Role-Physical (/100)	49.4 (39.5)	65.8 (35.1)**	71.7 (39.1)**	46.6 (37.4)	64.8 (36.1)*	81.2 (31.3)**††	0.84	0.93	0.34
Bodily Pain (/100)	73.0 (25.8)	73.2 (23.8)	77.2 (22.9)	71.1 (24.2)	68.5 (25.1)	70.0 (25.4)	0.67	0.40	0.26
General Health (/100)	55.8 (21.0)	64.1 (19.1)**	67.6 (21.8)**	54.3 (19.7)	58.9 (19.1)*	66.6 (18.4)**†	0.76	0.25	0.83
Vitality (/100)	36.0 (20.5)	59.0 (20.6)**	61.7 (22.1)**	31.5 (18.0)	55.3 (21.1)**	64.2 (17.2)**††	0.44	0.51	0.74
Social Function (/100)	63.1 (25.6)	77.2 (22.4)**	75.7 (25.5)**	67.2 (25.4)	75.0 (21.7)	84.0 (16.3)**†	0.49	0.60	0.23
Role-Emotion (/100)	72.6 (38.9)	81.6 (29.7)	86.0 (28.6)	67.6 (37.8)	65.8 (42.7)	78.7 (35.8) *†	0.49	0.14	0.44
Mental Health (/100)	70.7 (19.0)	77.6 (15.9)**	78.9 (17.3)**	64.3 (18.8)	73.5 (13.2)**	76.9 (13.0)**	0.15	0.10	0.20

BL, baseline; SD, standard deviation;

^*symbol indicates significant change from BL; *P ≤ 0.05;

^**P ≤ 0.01 (1-tailed);

^†symbol indicates significant change between 1 and 3 months; ^†P ≤ 0.05;

^††P ≤ 0.01 (1-tailed)

DISCUSSION

In patients with moderate-to-severe OSA with minimal comorbidity, this double-blinded, parallel-arm RCT found that the use of the flexible pressure mode, C-Flex, did not significantly improve compliance over CPAP after either one or three months of treatment (median differences of 22:48 and 19:48 min, respectively). Both C-Flex and CPAP modes of therapy resulted in improvements in objective sleepiness and vigilance, and subjective sleepiness, sleep quality, and health-related quality of life. This pattern persisted when patients using treatment at a suboptimal level and those who required low pressures were eliminated from analyses.

Although there was no significant difference between the two treatment modes in terms of vigilance (PVT), the CPAP group was significantly less sleepy than the C-Flex group when measured using the mod-MWT after 3 months of treatment (median difference of 6.4 min). However, this was due to the CPAP group having (nonsignificantly) higher sleep latency at baseline, evidenced by the fact that there was no significant difference between the groups concerning the change in sleep latency values over time. The modified-MWT protocol (used in the current study as well as our pilot²⁸) included 2 (rather than 4) 40-min tests, at 13:30 and 15:30.³² We chose to administer 2 MWTs rather than 4 for the following reasons: Firstly, there has been no reported evidence that 4 tests give any additional benefits over 2 tests (the normative MWT study on which clinical recommendations were based administered 4 tests and found no major differences between the sleep latencies of these tests⁴⁵). Secondly, 4 adequately spaced MWTs take an entire day to administer, and most patients find this tedious and time-consuming, possibly contributing to increased drop-out and/or a lower number of patient volunteers. Thirdly, timing the 2 tests to take place during the postprandial period is likely to give more meaningful information as to alertness/sleepiness than the information gained from the 2 earlier tests.

The results concerning compliance in the current study are in agreement with the 5 published articles which also found no significant difference between flexible and continuous pressure compliance using intention-to-treat analysis.^25–29 The patients in the current study had a greater mean body mass index (BMI) than in many other studies which is most likely the reason for the higher average pressure requirement— the mean pressures of 11.7 cm H₂O and 11.5 cm H₂O in the C-Flex and CPAP groups, respectively, were markedly higher than all other flexible pressure trials except our pilot study.²⁸ It is feasible that the difficulty of exhaling against the incoming pressure of CPAP would be more pronounced in those receiving high pressures, so that C-Flex would lead to greater compliance in these patients. However, the current study found no significant differences between C-Flex and CPAP when analyzing the compliance of patients requiring pressures ≥ 12 cm H₂O and ≥ 14 cm H₂O. Only one other study has analyzed the compliance of patients requiring high pressures separately and also found no significant difference between the two treatment modes, but the relatively low cut-off point of ≥ 9 cm H₂O was chosen.²⁵ Alternatively, it is possible that patients with severe OSA will comply with PAP regardless of the mode of delivery, and that C-Flex may be more beneficial to those with mild OSA. The article mentioned above as being the only study which has found significantly higher compliance with C-Flex reports a relatively low baseline AHI (39.4 events/h in the CPAP group and 43.2 events/h in the C-Flex group),²⁴ but a more methodologically sound study published recently also recruited patients with a lower baseline AHI (35.4 events/h) and found no significant difference in compliance between C-Flex and CPAP.²⁹

There were a few limitations relating to the design of this study. The limited number of data collection time points meant that early patterns of PAP use could not be monitored. It is possible that, along with other modifications such as humidification,^46–48 C-Flex is most useful during the early establishment period, causing little difference after this time. Compliance was calculated during this study as the total hours of use divided by the number of nights available for treatment; it is acknowledged that this is a simplification of available data and does not take into account the number of nights the patient did not use treatment at all, or the number of nights a patient was compliant or not (≥ 4 / < 4 h/ night use). It is possible that C-Flex might result in incomplete control of upper airway obstruction, and unfortunately this study did not include a PSG with treatment at the end of the 3-month trial to address this issue. Although the machine-interpreted residual AHI and leak levels were not significantly different between modes, it cannot be said with absolute certainty that C-Flex and CPAP both resulted in improved indices of sleep disordered breathing overnight. All devices given to the C-Flex group were locked to a dip setting of 2 (the middle level) as this was the level used in our pilot study. However, a better comparison may have been to compare CPAP and C-Flex at a dip setting of 3—the two most contrasting pressure delivery modes in the RemStar devices. Although the amount of follow-up support was kept the same across patients as far as possible, it would have been useful to keep a tally as to the number of additional interventions required to ensure that this was the same between groups. Patients were not asked at the conclusion of the trial whether any additional medications which may have affected sleepiness/alertness were supplied to them during the 3-month study period, or whether any changes to existing medications took place. Finally, as mentioned above, this study was powered to detect a difference in mean compliance between C-Flex and CPAP of 1.3 h (based on the results of our pilot study which found a difference of 1.7 h²⁸), so could not have detected as statistically significant a more modest, but still clinically significant, difference in compliance.

The major strength of this study is that it addresses a gap in the literature, in that it provides a well-designed, adequately powered RCT over three months comparing C-Flex and CPAP in patients with severe OSA including a comprehensive battery of secondary outcome measures. The quality design of this study extends to the adequacy of blinding—although not formally assessed, it was observed by non-blinded staff members that few patients were able to correctly guess their treatment allocation, and the data analyst remained blinded for the duration of the trial. This therefore eliminated a placebo effect which was the major flaw of the only study published to date which has found significantly higher compliance with C-Flex over CPAP using intention to treat analysis.²⁴ Improvements in all secondary outcome measures as a result of both treatments were of a large magnitude, so clinicians can be confident that both C-Flex and CPAP are highly effective treatments of severe OSA, particularly given the high usage rates of each mode. Additionally, the fact that both objective and subjective outcome measures changed in the same direction over the course of the study indicates high internal consistency.

Future research could focus on whether the real appeal behind C-Flex lies with the fact that it provides the patient with some control over their treatment, which is denied them with CPAP. Allowing the patient to vary the C-Flex dip setting according to comfort and preference may lead to drastic improvements in compliance (if only via a placebo effect), particularly in those patients who have indicated at baseline that they are not comfortable with the idea of PAP use.

This RCT has demonstrated that C-Flex does not lead to significantly greater compliance than CPAP measured after either one or three months. Both treatments resulted in substantial improvements in daytime sleepiness, vigilance, subjective sleepiness, sleep quality. and health-related quality of life. There was no indication that either treatment was superior, and patients reported both treatments to be equally comfortable. Now that a number of peer-reviewed articles have been published comparing C-Flex and CPAP, a meta-analysis is warranted.

DISCLOSURE STATEMENT

This study was funded by Philips-Respironics. All authors received research support from Philips-Respironics. Philips-Respironics manufactures and markets CPAP and C-Flex devices.