LinkExchange Member | Free Home Pages at GeoCities |
The authors note that the effective positive-pressure level (Peff) of CPAP for a given individual varies in relation to a number of factors: ingestion of alcohol and other sedative drugs, body position while sleeping, sleep stages, and even the course of CPAP itself, which may result in decreases of Peff over time. This implies that the Peff could change from night to night, or even from hour to hour within a given night. The ability to continuously modify Peff to a patient's present needs could yield more effective settings, less side-effects, and better compliance. Such considerations led to the development of the auto-CPAP, which continuously adapts the positive pressure level during the night, allowing a decrease (of up to -4 cm H20) in pressure when apnea and hypopneas disappear and an increase (of up to +2 cm H20) in pressure level when they reappear. It has been found as effective as conventional CPAP in correcting sleep-related breathing disorders, nocturnal desaturations, and sleep fragmentation. The purpose of this study was to compare the influence of auto-CPAP and conventional CPAP on sleep and breathing, daytime alertness, and cognitive impairment in previously untreated patients with Sleep Apnea Hypopnea Syndrome (SAHS). Subjects were 16 men with untreated SAHS, average age 54 (SD=11), BMI=34 (SD=6), Apnea/Hypopnea Index or AHI averaging 44 (SD=20). All subjects underwent conventional laboratory polysomnography at baseline to confirm the presence of a sleep-related breathing disorder, then a second polysomnogram within a week to determine the level of effective positive pressure needed to abolish apneas, hypopneas, and snoring (Peff). They also underwent a Maintenance of Wakefulness Test (MWT) in which they attempted to remain awake as long as possible during four consecutive 40-minute trials every 2 hours after turning off the light. They completed a Trailmaking Test, used to evaluate brain damage, which required alertness and concentration to quickly connect circles in numerical and alphabetical order, the score being the time taken. After the second polysomnogram, subjects were randomly assigned to two groups matched for age, AHI, and sleep latency. One group used the CPAP machine in constant-pressure mode, the other in automatic mode that adjusted pressure within a range of 2 cm above and 4 cm below the Peff. Both groups used the same CPAP machines, the patients being blind to group assignment. Treatment on the study continued for three weeks. At the end, all subjects underwent another lab polysomnogram without adjustment of pressures, and the MWT, Trailmaking Test, and Epworth Sleep Questionnaire were repeated the next day. The two groups were the same in their sleep and respiratory characteristics at baseline and at the end of study treatment. Peff was the same for both (9.5 vs 9.1 SD=3.1). AHI and sleep architecture were similar at baseline and at end of study, with significant increases in REM and slow wave sleep and decrease of light sleep in both treatment groups, and decline of AHI in Auto-CPAP and constant-CPAP from 40 and 47 respectively at baseline to 2 and 3 at end of study. Minimum oxygen saturations increased from 93 and 91 respectively at baseline to 96 in both groups at end of study.Arousal index decreased from 36 in both groups at baselineto 10 and 9 at study end. Epworth Sleep Scores (ESS) were similar in the two groups before (15 and 14) and after (6 and 9) treatment, at which point they were significantly reduced. Likewise the Maintenance of Wakefulness Test showed similar improvement in the two groups from baseline to study end in mean sleep latency: 18 and 19 versus 27 and 26. Scores on one of the two Trail Making Tests were similarly improved, from 41 and 36 to 33 in both groups at study end. the second Trail Making test also showed improvement but it did not achieve statistical significance. With auto-CPAP, patients spent 49% of the time at pressure equal to or below the Peff. Pressure significantly decreased during slow wave sleep (Stages III & IV) compared to light sleep (Stages I & II) and REM sleep. Nightly duration of home CPAP use by time counter was 7.1 hrs with auto-CPAP vs. 5.7 hrs with constant-CPAP, a difference of statistically borderline significance which became more significant when comparing the number of hours per night patients experienced a positive pressure. The authors conclude their article with a brief discussion that emphasizes that seemingly equivalent efficacy of auto-CPAP versus conventional CPAP and its possible advantage in compliance. They adduce evidence from other studies they have done the patients find auto-CPAP more comfortable, and from this study that auto-CPAP should allow some overall lowering of pressures, for example during slow-wave sleep when pressure needs seem lower. |
The main finding of this study--that auto-CPAP yielded similar therapeutic results to constant-pressure CPAP--unfortunately falls short of what one would have hoped to find: an advantage for auto-CPAP over constant-pressure CPAP. For one thing, especially with studies involving small numbers of subjects, a finding of equivalent results is inherently less convincing than one of significant differences, sometimes referred to in statistical circles as "trying to prove the null hypothesis" (the null hypothesis being that there is no difference between two groups, which is theoretically impossible to prove because the differences may always be in variables not measured or in magnitudes below the level of detectability). The differences in compliance were in the expected direction but statistically marginal. The authors make the point that their data do support a reason for frequent adjustment of pressures in terms of the variation of pressure needs with sleep stages. They point out that their study was of brief duration and took place during a period of clinical stability, whereas a longer study could make the outcome of auto-CPAP better simply on the basis of continued better compliance, if not on the basis of greater opportunity for clinical events (such as medication use, concurrent illness, weight change, etc) which could produce greater alterations in Peff. Essentially, I agree with the authors' optimistic appraisal of the potential advantages for auto-CPAP over constant-pressure CPAP, and regret that their study was designed in such a way as to reduce their chances of finding such an advantage. For one thing, the comparison of two groups with only 8 subjects in each group places severe limits on the magnitude of differences in effects of treatment that are detectable--the so-called statistical "power" of the study design. In fact, despite the many different variables subject to statistical testing, with significant pre-post differences mainly because of the powerful effects of CPAP itself, whether conventional or automatic, there appear to have been no statistically significant differences in group outcome on any variable, suggesting to me that the power of the study design was simply inadequate to detect meaningful differences in treatment outcome between the groups. Such "errors" of design usually result from practical problems with subject acquisition and funding, which rarely occur when pharmaceutical companies decide to invest large amounts of money in testing drugs--they want to be sure that their investments pay off with statistically significant positive findings if such exist.However, this often occurs when funding is noncommericial and/or severely limited. I did not notice any acknowledgement of funding agencies in the article itself so I assume this was the case. The same kind of "practical" consideration that limits subject numbers below statistical adequacy also operates to limit study duration. The authors seem well aware that a longer period of treatment under the study would have allowed many more opportunities for the kind of factors they mention as potentially affecting Peff to come into play. It is unlikely that in 3 weeks of study subjects will gain or lose significant amounts of weight, fall ill with an upper respiratory infection or allergy, alter patterns of alcohol or sedative use, etc. There is also not enough time to demonstrate the long-term effects of the modest differences in compliance observed between the two groups, and the possible further divergence of the two groups on compliance (which one might expect) as months pass.I noted no indication of careful assessment for differences in side-effects but perhaps these were done and not reported because of lack of significant differences. One "failing" of the study design that especially disturbed me was an end-of-study comparison of the two groups on the polysomnographic effects of the two different treatments. The auto-CPAP group was, after all, being exposed to an essentially "new" treatment at the end of the study, whereas the conventional CPAP group was receiving the same treatment they had become used to for three weeks. Furthermore, any advantage for auto-CPAP in terms of differential AHI or frequency of arousals might have been more apparent after three weeks of treatment than on the initial pressure titration night. However, I fear that the greatest limitation on demonstrating an advantage for auto-CPAP over conventional CPAP is inherent in the design of the auto-CPAP itself. It is based on the assumption that the goal of treatment is to suppress apneas, hypopneas, and desaturations, whereas there is an accumulation of evidence that sleep fragmentation is a more crucial variable, and that this can occur in the absence of apneas, hypopneas, or desaturations, whether due to upper airway obstructive factors (as in the Upper Airway Resistance Syndrome), to non-respiratory factors (such as periodic movements of sleep), or to unknown factors. Seemingly, the best results would derive from a treatment that specifically suppressed arousals, but the problem is that technically it would be more difficult to design a machine for home use that could detect the wide variety of such arousals, and furthermore decide whether such arousals were of respiratory origin or not. Meanwhile, the usual mode of CPAP titration, using a technician who can observe and integrate a multitude of variables including arousals related to periodic leg movements and unknown events, seems preferable to automatic titration in the laboratory or at home by machine alone. One must have personal, direct experience with the enormous volume of information available to a technician not only through the polysomnogram but also through audio and video monitoring of the subject, to appreciate how complex a task it would be for a computer to attempt to integrate all this information. Ideally, the patient on auto-CPAP would have this nightly monitoring and adjustment of pressure levels supplemented by periodic, fairly frequent in-laboratory reassessments to check for those factors the machine cannot monitor. Unfortunately, limitation of resources makes this impractical at present. Nevertheless, it might be feasible in the context of a well-funded study, of adequate duration (such as one year) with adequate numbers of subjects (such as 30 in each group) to evaluate this kind of "optimally" adjusted CPAP with the conventional CPAP treatment. |