Introduction

Obstructive sleep apnea (OSA) is a common sleep breathing disorder[1] associated with excessive daytime sleepiness[2] and the development of hypertension.[3] Meta analyses have shown that continuous positive airway pressure (CPAP) is an effective treatment for the daytime manifestations of severe OSA.[4, 5] CPAP therapy pneumatically splints open the collapsible sleeping airway and is nearly 100% effective in alleviating obstructions when properly applied.[6] Sub-optimal compliance thus appears to be a major factor limiting CPAP’s effectiveness.[7] Whilst there are many factors that have been shown to affect compliance levels,[7] technological changes in the delivery of positive airway pressure have been promoted as compliance enhancers. These newer devices have included bi-level PAP, auto-titrating CPAP devices and, most recently, flexible PAP (C-Flex, Respironics Inc., Murraysville, PA, USA). However, neither bi-level devices[8] nor auto-titrating devices[9] have been reliably demonstrated to increase compliance in randomised trials.

C-Flex is a technological development that reduces pressure during the first part of exhalation on a breath-by-breath basis in proportion to expiratory flow rate. Pressure is then increased towards the end of exhalation when upper airway narrowing and collapse is most likely. It is claimed that this feature increases patient comfort, aids compliance and improves clinical outcomes. However, whilst promising, only non-randomised trial data are currently published to support this notion.[10] The only randomised trial of the device was primarily designed to compare the effects on sleep of C-Flex to that of standard CPAP but subsequently found no compliance benefit after 7 weeks of treatment.[11] We aimed to test whether initial compliance levels with C-Flex are greater than those of CPAP in patients with severe OSA in a randomised controlled trial.

Materials and methods

We conducted a single-blind 4-week parallel arm randomised controlled trial comparing standard CPAP (Respironics REMStar Pro with C-Flex, Respironics Inc.) to C-Flex (we arbitrarily selected depth level II on the REMstar) in patients with severe obstructive sleep apnea (Apnea Hypopnea Index >30/h or 20–30/h with Epworth >12). Patients were randomised to treatment before manual CPAP titration with patients being titrated on the device they had been randomised to. Patients did not have significant other medical, sleep or psychiatric disorders or suffered from recent circadian shifts. The primary outcome was average hours of compliance with treatment per night. Compliance was objectively measured by an internal pressure sensor in the PAP device. Compliance reflects the amount of time that the device is switched on and being worn correctly so as to deliver effective therapeutic pressure. The secondary outcomes were improvements in daytime sleepiness (Epworth Sleepiness Scale[12] and a modified maintenance of wakefulness test[13]) and simple reaction time (Psychomotor Vigilance Task[14]). Measurements of sleepiness and reactions times were undertaken by an investigator blinded to treatment allocation; patients were blinded to treatment allocation. Both the Wellington Area Health Ethics Committee and the Massey University (Wellington) Human Ethics Committee approved the protocol and all patients gave written informed consent before entering the trial. More detailed descriptions of the patients and study protocols including randomisation and blinding are available from the corresponding author.

Statistical methods and data handling

All data were double entered and the data analyses were conducted in SPSS (v 12.0.1, SPSS Inc., Chicago IL, USA). Differences in the primary and secondary outcomes between and within treatments were analysed using the independent and paired t-tests, where appropriate, as these distributions were all approximately normal. Effect size calculations[15] were achieved by dividing the size (or net size of the effect) by the pooled standard deviation of both groups (primary outcome only) or the standard deviation of that measurement at baseline across both groups (secondary outcomes).

Results

Nine patients were randomised to C-Flex and ten to CPAP and patients’ characteristics and demographics were highly comparable between the treatment groups. Patients were typical for severely afflicted patients in our clinic with (mean ± SD) Apnea Hypopnea Index = 78 ± 33/h, Epworth Sleepiness Scale 14 ± 4, required PAP 8–17 cm H2O, BMI = 39 ± 10 kg/m2 and aged 20–63 years (see Table 1 for comparison of the randomised groups). CPAP and C-Flex had similar effects on indices of SDB during the titration phase. As expected, both treatments largely eliminated the AHI and improved oxygen desaturation indices.[6] This trial was not primarily designed to establish the equivalence between C-Flex and CPAP in controlling sleep disordered breathing as many of the patients received split-night combined diagnostic and titration studies. However, the performance of C-Flex and CPAP was not detectibly different on any index of SDB.

Table 1 Comparison of patients randomised to C-Flex with those randomised to CPAP
Full size table

CPAP compliance averaged 3.0 h/night (SD = 2.1 h, range 0–6.6 h/night), whilst C-Flex compliance averaged 4.7 h per night (SD = 2.9 h, range 0.5 to 8.7 h/night). This difference was not statistically significant (t = 1.52, 95% CI −0.7 to 4.1 h/night, p = 0.15, see Fig. 1 for box and whisker plot of these data). The effect size of 0.68 was of moderate magnitude. A post hoc power calculation indicated that approximately 33 patients per group would be required to give an 80% likelihood of detecting a 1.7-h difference as being significant (alpha = 0.05) given the observed variability. Removing patients who were unable to tolerate PAP therapy (i.e. those averaging less than 1 h per night compliance) from the analysis resulted in the loss of two patients from each treatment group in a post hoc analysis. The remaining patients used CPAP for an average of 3.8 h per night (SD 1.5) and C-Flex for an average 5.9 h per night (SD 2.0). This 2.1-h difference was statistically significant (t = 2.4, 95% CI 0.2 to 4.1 h, p = 0.035). Using the same variability as the primary analysis (SD = 2.5) gives a large effect at 0.84. Whilst this was not an a priori comparison no other compliance cut-points were tested. Whilst all 19 patients were included in the primary analysis under the intention-to-treat principle due to missed appointments only 17 patients are included in the Epworth analysis and 16 each were included in the modMWT and PVT analyses. Secondary outcomes are listed in Table 2.

Fig. 1
figure 1

Box and whisker plot showing the apparent 1.7 h difference in compliance between C-Flex and CPAP (p=0.15)

Full size image
Table 2 Effects of treatments on secondary outcome measures
Full size table

Discussion

In this 4-week randomised controlled trial, in patients with severe OSA, C-Flex use resulted in a trend toward 1.7 h/night of additional compliance compared to CPAP (p = 0.15). In a post hoc analysis patients, who were able to tolerate PAP therapy, (mean use >1 h/night) used C-Flex for a significant 2.1 h/night more than CPAP. This represented a moderate to large compliance improvement (effect size 0.68 or 0.84) and, to our knowledge, is the largest compliance increase due to a PAP device modification reported from an RCT. However, the post hoc analysis should be treated with due caution as it was not a planned pre-trial investigation. Our data, in addition to providing powering estimates for larger trials, also provide close agreement with the magnitude of effect found by Aloia and colleagues[10] in a non-randomised study which also found a 1.7 h/night increase in compliance associated with C-Flex.

There is one other published randomised controlled trial that we are aware of that compared the efficacy and effectiveness of C-Flex (setting III) to CPAP in 52 patients found that the average compliance with therapy was 5.3 vs. 5.2 h/night, respectively (p = 0.99).[11] It is therefore possible that our results are due to chance alone. Conversely, the larger trial might not have observed an effect on compliance because of their combination of a cross-over study to compare efficacy as measured by PSG followed by a parallel study to measure compliance and patient outcomes. Thus all the patients in the compliance study had already been exposed to both treatment modalities before being sent home with the longer-term treatment modality. The other trial also used pressure relief setting level III which could potentially have different compliance effects than the pressure relief setting II which we used.

In our trial, increased compliance did not result in better outcomes for patients. Indeed the Epworth Sleepiness Scale was significantly more improved on CPAP treatment than on C-Flex despite the additional compliance with the novel modality. This finding is unexpected and difficult to explain. However, this does not mean that it is erroneous. It might reflect as yet undescribed differences between standard CPAP and C-Flex on sleep architecture that we were unable to investigate with our study design. In addition, before C-Flex can be widely recommended as a superior option to CPAP there needs to be a clear understanding of the marginal benefits of increasing compliance with PAP via this method which may not furnish the same benefits as increasing compliance with CPAP.[16] Ideally, these clear incremental benefits should include reductions in cardiovascular disease (CVD) or well-established markers of CVD.

This small randomised controlled trial has demonstrated some evidence of increased compliance with C-Flex compared to standard CPAP in patients with severe OSA. However, conflicting evidence from a recent randomised[11] and non-randomised studies,[10] that were not optimally designed to test compliance differences, means that a larger randomised trial designed specifically to test compliance and patient-orientated outcomes is now warranted.