This issue of AJRCCM (pp. 773–780) features a randomized controlled trial (RCT) (1) to add to the growing database of continuous positive airway pressure (CPAP) treatment for obstructive sleep apnea/hypopnea syndrome (OSAHS). This blinded, multi-center, placebo-controlled Australian trial includes a comprehensive range of salient outcomes, including objective and subjective sleepiness, cognitive performance, and blood pressure and represents high-quality research into CPAP required for systematic reviews. It reports that patients with mild SAHS (apnea + hypopnea index [AHI] < 30 per hour) gained few benefits from CPAP over placebo treatment.

Before recalling the many CPAP units issued to patients with an AHI less than 30, consider that clinical practice seldom changes overnight. Except in new treatments, there is more commonly a gradual accumulation of evidence that must be periodically reviewed, taking into account the trials' quality, specific methodology, and sampling characteristics, as well as their results.

A systematic review would include at least 12 previous placebo-controlled RCTs of CPAP (2-13) of varying statistical power, study design, sampling methods, outcome measurements, and results. Of these, eight contain patient groups of more severe polysomnographic status (2-9), whereas the other four (10– 13) examine the role of CPAP in ‘mild' OSAHS. Among these 12, just two studies (8, 9) do not show benefits from CPAP.

Taking previous RCTs of patients with moderate to severe SAHS first (2-9), all but two (8, 9) of these provide robust evidence of better enhancements to function from CPAP than placebo. Within daytime function outcomes, six (2-7) of eight (2-9) studies showed improvements in subjective symptoms and/or sleepiness, three (2, 4, 5) of four (2, 4, 5, 8) studies showed improvements in objective sleepiness, two (2, 5) of four (2, 5, 8, 9) in cognitive function scores, and five (2-4, 6, 7) of six (2–4, 6– 8) in quality of life or well-being scores. Among studies of blood pressure, one (7) of the two (7, 8) showed reductions with CPAP.

Although statisticians expect ‘scatter' in results from similar RCTs, both the negative trials (8, 9) in moderate to severe OSAHS have specific qualities that might predispose them toward confirming the null hypothesis. The trial of Barbé and associates (8) recruited only less-sleepy (Epworth score less than 10) patients with significant AHI (greater than 30 per hour). In these patients, CPAP did not produce controlled improvements to daytime function or blood pressure. The negative U.S. trial (9) examined just subjective sleepiness and cognitive performance scores as outcomes. Both these trials used parallel and not crossover designs, providing less statistical power, but had sample sizes amongst the lowest in RCTs. These factors may have contributed to negative results.

The ‘mild' OSAHS of five trials' titles (1, 10-13) denotes low polysomnographic severity, with AHI values up to a ceiling of 15 or 30 per hour. Among previous trials, two (10, 11) of four (10-13) showed enhancements to subjective sleepiness and/or symptoms; two (10, 11) of three (10, 11, 13), improvements in cognitive performance; two (10, 11) of four (10-13), improvements in quality of life or well-being scores; but none in objective sleepiness (10-13) or blood pressure (13). Whereas no outcome was improved in the trial of Redline and associates (12), significantly more patients were classed as ‘responders' (improving in two or more of three daytime function domains) on CPAP than on conservative treatment.

CPAP appeared beneficial to at least some extent in four (10-13) of these trials, with the current Australian trial (1) being the one negative trial. An examination of sleepiness scores across these trials shows important distinctions. Both ‘mild' Edinburgh trials (10, 11) selectively recruited patients who were sleepy and whose Epworth scores averaged 14 and 13, respectively. Patients in the trial of Monasterio and associates (13) had a mean baseline Epworth score of 13, in line with other positive trials, whereas the somewhat less positive sample of Redline and colleagues averaged 10 on this scale. Patients in the negative trial of Barnes and associates (1) also had less-sleepy Epworth scores, with a mean of 11.

Perhaps the most important contributing factor to negative findings in the Australian trial (1) is discussed by the authors. This revolves around the daytime symptomatic status of the patient sample. Of the 13 RCTs (1-13), 8 required significant daytime symptoms or sleepiness as an entry criterion (2-7, 10, 11), and all of these demonstrated positive benefit from CPAP. Of the remaining five trials, two selected for low sleepiness (8, 12) and three (1, 9, 13) did not use sleepiness as an entry criterion. A review of subjective and objective sleepiness scores across trials (1-13) tends to confirm that where baseline Epworth is 12 or more, or mean multiple sleep onset latency less than 10 minutes, CPAP appears to enhance daytime function. The only exception to this rule is the positive U.S. trial in mild patients (11) with low subjective (Epworth 10) and objective (mean multiple sleep onset latency 10 minutes) sleepiness, where 49% of CPAP-treated subjects were classified as “responders.”

A further contributing factor to the negative findings of Barnes and associates might be the imbalance in treatment order. This seemingly trivial detail has impact when learning and placebo effects on retesting are large, as has been documented in the current (1) and previous (2, 4, 9) trials. CPAP scores, represented by 13 second testings and 15 first testings, would be disadvantaged compared with placebo scores, with 15 second testings and 13 first testings, thus biasing against a positive finding.

Barnes and colleagues (1) have quite properly interpreted the results of their trial conservatively, as showing overall unconvincing benefit from CPAP among the many outcomes. However, a closer reading suggests that at least some patients did experience improved daytime function with CPAP. Of 34 outcomes, 23 had mean values favoring CPAP over placebo. These were statistically significant for three of five symptom subscores and for verbal fluency score and showed trends for improvement in functional outcomes of sleep questionnaire score. The authors noted that better use of CPAP was more frequent among patients with higher baseline daytime sleepiness. For these reasons, this important trial may not be incongruous with the previous literature.

The message emerging from this personal interpretation of these RCTs is the importance of accompanying daytime symptoms, particularly sleepiness, to a diagnosis of SAHS. The syndrome definition makes clear that daytime symptoms must accompany sleep-disordered breathing. This appears to extend to a situation where daytime sleepiness severity in OSAHS is also linked to daytime benefit from CPAP treatment. This linkage may arise in part from the fact that CPAP is a relatively cumbersome and intrusive treatment, such that patients will be reluctant to accept the treatment without perceptible benefit to daytime symptoms. But it brings into question the use of CPAP to treat sleep-disordered breathing per se, without significant symptoms.

The blossoming of high-quality trials of CPAP over this last decade is a gift to clinicians, who can compare and contrast these trials to guide patient management. In my opinion, current evidence, including this issue's Australian RCT of CPAP, can be used to support CPAP therapy for SAHS of any polysomnographic severity, provided that the symptoms of sleepiness themselves are not ‘mild'.

References

Section:

ChooseTop of pageReferences <<CITING ARTICLES