Effect of Continuous Positive Airway Pressure vs. Placebo Continuous Positive Airway Pressure On Sleep Quality in Obstructive Sleep Apnea

Jose S. Loredo, MD, Sonia Ancoli-Israel, PhD, and Joel E. Dimsdale, MD
Departments of Medicine and Psychiatry, University of California at San Diego, and Department of Psychiatry, Veterans Affairs, San Diego Healthcare System
Published in Chest 1998; 244: 163-168

Summary

The authors make the point that the research literature on effectiveness of Continuous Positive Airway Pressure (CPAP), the standard treatment for Obstructive Sleep Apnea (OSA), suffers from a relative lack of studies well designed with an appropriate control treatment group to compare with those patients receiving CPAP. They have found only six studies, of which four used an inappropriate oral treatment (e.g., pills) as a control for CPAP, and one used as a control treatment, air given by nasal cannula. None of these resemble CPAP. Only one other study used the more appropriate control treatment given in this study--CPAP set at an ineffective pressure.

The study reported in this abstract used CPAP set at 2 cm H2O pressure as the placebo treatment and compared it with CPAP set at a pressure titrated to abolish (or minimally, reduce by 50%) all apneas, hypopneas, and snoring associated with arousals. In the average subject treated with active CPAP, this pressure came to 10 cm H2O.

The investigators limited their subjects to an age range from 30 to 65 years, weight no greater than 1.7 times their ideal, free of medication known to affect sleep, and free of several major medical illnesses. On two successive occasions of sleep polysomnography--the first at home, the second in the laboratory--subjects must have demonstrated a Respiratory Distress Index (RDI=average number of apneas and hypopneas per hours of sleep) equal to at least 20/hour.

Having met these criteria, a group of 48 subjects then entered the treatment phase of the study, in which they were randomly assigned to receive either the inadequate placebo CPAP or the adequate CPAP. However, only 41 subjects were included in the final analysis, one having had placebo treatment discontinued because of severe OSA with severe hypoxemia (low blood oxygen), another proving unable to sleep with the CPAP equipment, and four more turning out to have initial RDI's below 20. The remaining 41 subjects were mostly (75%) men and were moderately obese (weight averaging 136% of ideal).

The randomization failed to yield two groups of subjects equivalent in all respects. The active CPAP group proved significantly heavier (Body Mass Index or BMI=33.0 for active treatment subject versus 29.1 for placebo treated subjects) and also had significantly lower blood oxygen levels prior to treatment (89.1 versus 93.5). However, on all other sleep measures they proved similar prior to treatment.

The laboratory sleep study done at the done at the beginning was repeated a week later. Changes in sleep measures from the first study to the last represented the basis for comparing active and placebo treatment effects by statistical tests which detected and removed changes that occurred in both groups and therefore might have resulted from just the passage of time.

Active CPAP proved significantly more effective than placebo CPAP on three sleep measures:

• Reduction of RDI (from 56.4 to 3.3, vs placebo 44.2 to 28.3)
• Less frequent arousals (55.1/hr to 14.4 vs 48.6 to 33.55)
• Higher sleep blood oxygen (89.1 to 95.6 vs 93.5 to 94.1)

In contrast, active CPAP showed no significant advantage over placebo CPAP on seven sleep measures:

• Delay of sleep onset
• Percentage of Stage I (light) sleep
• Percentage of Slow Wave (deep) sleep
• Percentage of Rapid Eye Movement (so-called dreaming) sleep
• Time spent awake after first falling asleep
• Total Sleep Time
• Sleep Efficiency (percentage of total time spent asleep)

Note that all of the above outcome measures, with the exceptions of Total Sleep Time and Time Awake After Sleep Onset, improved with time in both the active and the placebo CPAP groups.

Note also that the CPAP machines all contained a hidden clock to measure the amount of time they were turned on each day. In both treatment groups, this proved to exceed 5 hours/night, and the amount of time the machine was on proved unrelated to the effects of treatment.

The authors were surprised that their results failed to confirm those of uncontrolled studies, which had reported that CPAP improved sleep quality according to percentages of time spent in different stages of sleep, sleep efficiency, total sleep time, or time spent awake after first falling asleep. They suspected one explanation for this failure might lie in the first-night effect, which means that subjects sleeping in the lab for the first time sleep worse than on later retests, when they become more familiar with the environment and procedures. This effect would result in improvement regardless of whether the subjective received placebo or active CPAP, reducing the difference in changes between the two groups. A second lab study prior to treatment, had it been done, might have reduced or eliminated this first-night effect from the measures of change during treatment.

However, the authors acknowledged the possibility that reports from uncontrolled studies, which cannot separate active treatment from placebo effects, may have led to an incorrect conclusion. Perhaps CPAP, while clearly effective in correcting OSA and its associated arousals and drops of blood oxygen, may have much less effectiveness in normalizing other measures of sleep.

Comments

The crucial importance of randomized, placebo-controlled studies in assessing the effectiveness of medication treatments is taken for granted throughout most of medicine. The reason it has much less importance for certain non-medication treatments is only sometimes obvious (for example, surgical procedures represent too much investment of time and risk to do in a "placebo" fashion expected to prove ineffective). The convention of dispensing with controlled studies as a prerequisite for gauging treatment effects may also occur when clinicians consider those effects too obvious, due to the stability of the disease measures and the magnitude of treatment effects, to require placebo controls.

On the other hand, this convention may tempt clinicians and clinical investigators with the convenience of avoiding the much greater effort of planning, carrying out, and obtaining subject consent, for the "gold standard" of placebo-controlled, randomized clinical trials. However, this may lead to substantial errors in judging the extent of treatment effects, by confusing them with the effects of time, the power of hope and suggestion (placebo effects) and still other factors.

Even in uncontrolled studies of OSA treated with CPAP, one gets the impression that extent of the symptomatic and functional improvement does not always correspond to the suppression of apneas and hypopneas.

This is not to raise doubts about the crucial importance of CPAP for people with OSA. It is only to question CPAP as a panacea (cure-all) for OSA.

One old saying reminds us: "If it looks too good to be true, it probably is!"

In face of extravagantly optimistic claims, as another even older saying warns, "Caveat emptor!" (Let the buyer--or consumer--beware!)

For more information on Sleep Apnea, please see our Sleep Apnea Section.