INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Sleep-disordered breathing (SDB), a syndrome associated with repetitive apneas and hypopneas during sleep and associated with desaturation, sleep fragmentation, and daytime sleepiness, has been estimated to affect 2 to 4% of the adult middle-aged population (1). The overall health effects of this prevalent condition, and the potential benefits to be derived from specific therapies for it, are controversial (2). Among patients with severe SDB, usually defined on the basis of frequent apneas (> 20 apneas/h of sleep), beneficial effects of nasal continuous positive airway pressure (CPAP) or surgery have been reported (3). In these patients, elimination or reduction of apneas may result in improved sleep architecture and reduced daytime sleepiness (1, 2, 4, 5). Several studies also suggest that treatment of patients with moderate to severe sleep apnea may improve cognition and mood (5, 9).

Much less is known about the role of specific treatments in ameliorating symptoms in individuals with the milder degrees of SDB that are more commonly observed in the population (1, 13, 14). A paucity of data related to the role of specific therapies in reducing morbidity or improving the quality of life in individuals with milder degrees of SDB have led to markedly disparate approaches for screening and treating patients with snoring but without pathologic daytime sleepiness. Despite a lack of data on threshold effects related to SDB and potential benefits of treatment for it, many third-party payers have established minimal levels of SDB as criteria for reimbursing the costs of CPAP or surgery. However, all levels proposed have either been set arbitrarily or without consideration of data from patients representing a broad spectrum of disease. Additional data on treatment effectiveness across the spectrum of SDB and within patient subgroups are needed to develop a rational approach to screening and treating this prevalent condition.

In this paper we report the initial results of a randomized controlled study of treatment in a group of subjects aged 25 to 65 years with mild to moderate SDB (RDI < 30)a subgroup about which there is no consensus on the role of specific therapy. Subjects were randomized to receive nasal CPAP or conservative treatment (CT), and were reassessed after at least 8 wk of intervention. Outcome was assessed with a measure constructed to detect changes before and after therapy in areas thought to be clinically relevant to individuals with sleep disorders, consisting of mood, energy/fatigue, and functional status.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subjects

Recruitment was aimed at enrolling subjects with mild to moderate SDB, defined on the basis of an RDI between 5 and 30 and the absence of "pathologic" sleepiness. Eligibility criteria included: (1) age between 25 and 65 yr; (2) RDI between 5 and 30; (3) absence of (subjective) pathologic sleepiness (did not fall asleep driving or in other potentially dangerous situations); (4) absence of a sleep disorder other than SDB (narcolepsy; insomnia, defined as regularly sleeping less than 6 h per night; regular use of hypnotics; sleep insufficiency, defined as sleeping  2 h more on nonwork than on work days; or a history of periodic leg movements; and (5) absence of underlying conditions that could interfere with neuropsychologic test performance or with adherence to the study protocol. These included severe or unstable medical disease (myocardial infarction or congestive heart failure documented with the previous 3 mo, uncontrolled diabetes or thyroid disorder, cirrhosis, or recently diagnosed cancer); neurologic disease (a history of stroke, seizure disorder, or head trauma with loss of consciousness for > 6 h or associated with memory impairment); alcohol abuse (a history of  5 alcoholic drinks/d for > 6 yr) or drug abuse (current drug use or heavy past use leading to tolerance or dependency); regular use of medications that impair the sensorium (e.g., benzodiazepines); and < 8 yr of schooling.

Individuals were recruited by means of fliers distributed throughout the Cleveland Veterans Affairs Medical Center and Case Western Reserve University campus, soliciting participation of "snorers" in a research study, and letters sent to patients enrolled in general medical clinics. Letters were also sent to practitioners in the community, inviting referrals of patients with snoring. Participants in other ongoing clinical or epidemiologic studies (the Cleveland Family Study and P450 Cytochrome Blood Pressure Study) also were identified if a sleep study performed for these other studies showed an RDI between 5 and 30. Study eligibility was initially ascertained by orally administered questionnaires and subsequently by a screening in-home sleep study (Edentec II Monitoring Device; Eden Prairie, MN). After ascertaining study eligibility, subjects were randomized either to CT or nasal CPAP through the use of a computerized program based on random-number assignments.

Study Protocol

Testing schedule. Following enrollment, subjects underwent an in-laboratory polysomnogram preceding a baseline daytime test battery. When possible, this was scheduled to immediately precede the daytime test battery. In subjects in whom the laboratory polysomnogram did not immediately precede the testing day, an additional in-home sleep study was performed on the night before the test battery to "expose" the subject to a preceding monitoring night and to document sleep time.

Follow-up studies were scheduled after at least 8 wk of intervention, and after at least 2 wk following any intercurrent illnesses (upper-respiratory-tract illnesses, etc.). The initial protocol called for an upper outer window for follow-up of 12 wk. However, the occurrence of frequent personal obligations required extension of this window to 16 wk. Three subjects, however, were retested after the 16-wk outer window (two at 17 wk and one at 19 wk). Delays in scheduling were caused by hospitalization for treatment of nephrolithiasis (n = 1), relocation (n = 1), and personal obligations that prevented scheduling (n = 1).

Follow-up testing included a daytime test battery preceded by an in-home sleep study. The latter was done to provide a monitoring "exposure" that was comparable to the one prior to the baseline test battery, and to evaluate the effectiveness of treatment in reducing respiratory disturbances during sleep at home.

Assessment of sleep. Overnight, in-laboratory polysomnography consisted of measurement of airflow (by nasal/oral thermistry), chest and abdominal effort (by piezo sensors), finger-pulse oximetry, an electrocardiogram (ECG), C4/A1, C3/A2, and O1/A2 electroencephalograms (EEGs), submental and tibialis electromyograms (EMGs), and right and left electrooculograms (EOGs) with a Nihon Koden (Anaheim, CA) 4400 Polygraph or a SensorMedics Somno Star (Yorba Linda, CA) 4100 Polygraph.

Daytime test battery. The daytime test battery consisted of an extensive array of tests of mood, functional status, and neuropsychologic functions. The testing day included assessment of objective sleepiness level with the Multiple Sleep Latency Test (MSLT) (15). Subjects underwent an MSLT consisting of four naps at 2-h intervals, starting at approximately 9:00 to 9:30 A.M., with the subjects given 20 min to fall asleep per nap, according to a standard protocol (15). Most of the other tests were given between the first and second and between the third and fourth naps. There were two different testing sequences, to minimize time-of-day effects. Alternate forms of tests were used when feasible, to minimize practice effects. The order of alternate forms was randomized between baseline and follow-up assessments. Testing (including the MSLT, a lunch break, and rests between tests), was completed on average between 9:00 A.M. and 4:30 P.M.

Treatment Protocol

Subjects in both treatment arms of the study received counseling about sleep posture (i.e., discouraging sleep in the supine position with the use of props as needed) and sleep hygiene (maintaining regular sleep habits and discouraging the use of sedatives, stimulants, alcohol, tobacco, and food near bedtime). Subjects with a BMI > 29 kg/m² were referred to a dietitian for weight-reduction counseling, and subjects with symptoms of nasal congestion were provided with a nasal steroid spray (Becanase nasal spray). Additionally, subjects in the CT arm of the study were given a supply of mechanical nasal dilators (Breathe-Right, CNS Inc., Chanhassen, MN), and those in the CPAP arm were provided with a CPAP machine (Models 314, 318, or 320; Puritan Bennett, Lenexa, KS, or Tranquility Quest 7300H; HealthDyne, Marietta, GA). Mechanical nasal dilators were used as a component of treatment in the control arm of the study to provide the participant with a "tool" to be used nightly, following a treatment schedule similar to that for participants in the CPAP arm. Subjects were instructed to use the nasal dilators or CPAP each night. The prescribed CPAP pressure was based on the results of overnight polysomnography preformed after the test battery for titration purposes. Prescribed CPAP pressure was based on the level that obliterated snoring and the majority of respiratory events during the titration polysomnogram. The mean CPAP pressure prescribed was 7.4 ± 1.8 cm H₂O (range: 5 to 13 cm H₂O).

Phone calls were made approximately every 2 wk to subjects in both groups, reminding them to use the prescribed therapy and to troubleshoot any problems associated with the interventions. The technician involved with all aspects of treatment (i.e., assignments, counseling, follow-up phone calls) was different from the technician primarily assigned to do testing.

Test Instruments Used in the Study

Health status (history of sinus problems, hypertension, diabetes) and symptoms of SDB were assessed with a modified version of the Specialized Center of Research in Cardiopulmonary Disorders of Sleep (SCOR) Sleep and Health Questionnaire (16) that was completed independently by each subject. Habitual snoring and excessive daytime sleepiness were each considered to be present if the subject reported these symptoms as occurring "frequently" or "at least 3-4 times per week." Sleepiness was characterized with the MSLT and with the Epworth Sleepiness Scale (ESS) (17). The latter is a self-administered questionnaire that requires the subject to rate his or her chances of falling asleep in eight different situations, referring to his or her "usual way of life in recent times." Scores vary from 0 to 24, with increasing scores indicating greater sleepiness (17). Mood was assessed with the Profile of Mood States (POMS) (18) administered during the morning testing session, and the Positive and Negative Affect Scale (PANAS) (19), which provides separate assessments of positive and negative emotional states. According to recent psychological research, positive and negative emotional states are orthogonal dimensions rather than a single bipolar continuum (20). Well being and functional status were assessed with the Medical Outcomes Survey Short Form 36 (MOS) (21, 22). Neuropsychologic test performance is being presented separately, and therefore is not reported here.

Assessment of Adherence to the Study Protocol

Adherence to the use of CPAP was assessed with an internal compliance meter contained within the CPAP unit, which detected changes in airflow (Puritan Bennett Models 314 and 320, and Tranquility Quest 7300H) or hours of use (Puritan Bennett Model 318). The predicted number of hours of CPAP use was calculated from the self-reported average number of hours slept per night and the number of days of the intervention. Adherence in the use of nasal dilators was assessed through self-reporting.

Data Analysis

Scoring of sleep and respiratory data. A respiratory disturbance (apnea or hypopnea) was defined as a discernible change in airflow or chest-wall movement, lasting > 10 s and occurring in association with at least a 2.5% decrease in oxygen saturation (Sa_O2) or an arousal (arousals were used in the definition of respiratory events only for data obtained from laboratory polysomnograms; desaturation criteria alone were used in the identification of respiratory events for the home screening and follow-up studies). Sleep was staged according to the criteria of Rechtschaffen and Kales (23). The number of upward sleep-stage shifts was defined as the number of times that sleep shifted from a deeper to a lighter stage (e.g., Stage I to Wakefulness, Stage II to Stage I, etc.)

Data from the MSLT. The average sleep latency over four naps was computed. Any nap period during which the subject did not sleep was coded as having a latency of 20 min (15).

Assessment of treatment response. We identified three clinically relevant outcomes that have been suggested to be affected by SDB: energy/fatigue, mood, and general health/functional status (24). A treatment response score was derived from changes in tests in each of these three domains. Changes in mood were measured with the depression subscale of the POMS and the PANAS negative and positive affect scores. Energy/fatigue was assessed through changes in the energy/ fatigue subscore of the MOS and the fatigue subscore of the POMS. General health/functional status was assessed through changes in the following subscores of the MOS: general health perceptions, role limitations due to emotional problems, role limitations due to physical problems, and social functioning. (Sleepiness was not considered an outcome in these analyses because of our interest in using baseline sleepiness as a potential predictor of treatment response.) For each of these domains, a score of +1 was assigned if the score on any of the tests included in each respective area improved by > 0.5 standard deviation (SD) (a statistical change considered to represent a "moderate" effect) between baseline and follow-up assessments. A score of 1 was assigned to any domain in which test scores declined by > 0.5 SD. Scores of 0 were assigned to changes in tests of less than ± 0.5 SD. A total score was calculated on the basis of the sum (1 to +1) of scores across these three domains (providing a total possible aggregate score of 3 to +3). Subjects were considered to have an improved outcome on the basis of response scores of  2 ("responders"). This indicated improvement in at least two of the areas investigated (mood, fatigue, functional status/general health), with no deterioration in any domain.

Statistical analysis and sample size. Group differences were examined with unpaired t tests and contingency table analysis. Logistic regression analyses were used to assess the relationship between the outcome response (as previously defined) and group status and covariates. Within-group differences in conjunction with time were assessed with paired t tests and repeated measures analyses of variance (ANOVA). A sample size of 112 was identified as sufficient to detect a difference between groups of 0.5 SD with 75% power at an alpha level of 0.05.

The study protocol was approved by the Institutional Review Board of the Cleveland Veterans Affairs Medical Center, and written informed consent was obtained for all subjects.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Subjects

Screening of volunteers with questionnaires and sleep screening studies identified 114 subjects eligible for study participation and able to commit themselves to 2 full days of testing. Figure 1 shows the flow of participants in the study. During randomization, one subject refused to be randomized to the CT arm because of a desire to start CPAP, and two subjects refused randomization to CPAP. Three subjects who were randomized to CPAP declined further study participation because of problems in using CPAP during the intervention period (associated with complaints about discomfort related to breathing against a positive pressure). Of the remaining 108 subjects, six subjects in the CT group and five subjects in the CPAP group declined participation in follow-up assessments. The predominant reason these subjects provided for failure to schedule follow-up assessments were the time requirements of the all-day test battery. The characteristics of the subjects who failed randomization or were lost to follow-up are shown in Table 1 (column one). As compared with subjects randomized and assessed in the follow-up battery, these subjects tended to have a higher RDI and a lower social functioning subscore: these differences, however, were not statistically significant.

View larger version (36K): [in this window] [in a new window]   Figure 1.   Flow of participants, including number randomized (R) and retained for follow-up.

The 97 subjects who were randomized (46 to CT and 51 to CPAP) and followed for the duration of the study period had an average age of 48 ± 9.8 yr (mean ± SD), and consisted of 48% females and 62% European-Americans (the majority of the remaining subjects were African-Americans) (Table ). Their RDI, on average, was mildly increased (13.3 ± 9.8), and their MSLT, on average, indicated minimal sleepiness (10.1 ± 4.8 min). Values for energy and fatigue and for general health perceptions obtained from the MOS were lower than what has been reported for generally healthy patients (19). The demographic characteristics and physiologic data were comparable for subjects in each of the two treatment arms. However, the MOS subscore for "role limitations due to physical problems" was lower for members of the CPAP than for those of the CT group (p < 0.05). Subjects randomized to CPAP were followed up at 10.8 ± 2.7 wk, and those randomized to CT at 10.1 ± 1.8 wk (p > 0.05).

Adherence to Therapy

Subjects in the CT group reported use of the mechanical nasal dilators during 82 ± 26% of intervention nights. In the CPAP intervention group, the machine's internal compliance monitors demonstrated CPAP use for 44 ± 34% of the time subjects were estimated to be asleep (3.1 h). Fifty percent of subjects randomized to CPAP demonstrated CPAP use for at least 40% of estimated sleep time.

Impact of Treatment on SDB and Sleepiness

Significant decreases in RDI with treatment were observed in each intervention group (Table 2). However, the posttreatment RDI (assessed while the subject was sleeping at home, prior to the follow-up battery and while using CPAP or mechanical nasal dilators) was significantly lower with CPAP than with CT. The oxygen-saturation nadir improved significantly with CPAP but not with CT (Table ).

Sleepiness, as measured with the MSLT, did not change with treatment for subjects treated with CT or with CPAP ( = 0.07 ± 5.30 min and 0.46 ± 5.4 min, for differences between the post- and pretreatment levels for the CT and CPAP groups, respectively) (Table ). Subjective sleepiness, as measured with the ESS, changed significantly in the CPAP group ( = 1.48 ± 4.13; p < 0.05) but not in the CT group (0.39 ± 2.95). With treatment, no significant changes were observed for BMI for either the CT or the CPAP group (overall change in BMI:  = 0.11 ± 0.20 kg/m²).

Impact of Treatment on Outcome

Of the 51 subjects randomized to CPAP therapy, 25 (49%) experienced improvement in at least two of the three domains described previously. This contrasted with improvement in only 12 of the 46 subjects randomized to CT (26%) (p < 0.05) (Figure 2). The odds of observing a positive treatment response in the CPAP as compared with the CT group was 2.72 (OR: 1.18 to 6.58, 95% CI). Logistic regression analyses demonstrated that neither RDI nor sleepiness (measured with either the MSLT or ESS) was significantly associated with treatment response.

View larger version (22K): [in this window] [in a new window]   Figure 2.   Percentage of subjects who demonstrated improvement in a least two of three domains (mood, energy/fatigue, functional status/general health) and in each treatment arm (CPAP: dark bars; conservative medical therapy [CT]: light bars.) Results are shown for the total sample, and for subsamples classified according to severity of SDB (RDI), sleepiness (MSLT), sinus problems, and underlying comorbidity (hypertension or diabetes).

A comparison of the characteristics of subjects within each treatment group according to the response to treatment is shown in Table 3. Within each treatment group, response was unrelated to gender, race, age, symptoms of SDB, level of RDI, level of sleepiness, or sleep architecture. However, in each group, a positive treatment response was significantly associated with higher BMI. Within the CT group, improved outcome also was significantly related to a history of sinus problems (i.e., 11 of the 12 CT responders reported a history of sinus problems). Compliance with CPAP therapy was significantly greater among CPAP responders than among nonresponders (i.e., those that had a response to intervention used the CPAP machine for 58 ± 33% of predicted sleep hours as compared with 31 ± 29% of predicted sleep hours for the nonresponder group, p < 0.05).

Additional exploratory analyses were undertaken to identify which subsets of individuals might best respond to CPAP as compared with CT. Figure 2 shows the percentage of responders in both the CPAP and CT groups within subsets defined by sleepiness (MSLT < 8 min), SDB severity (RDI > 15), and underlying comorbidity (history of hypertension or diabetes). As suggested by the logistic regression analyses, differential effects of CPAP as compared with CT were not more evident in subgroups of participants with greater sleepiness or higher RDIs. However, a significant beneficial effect of CPAP over CT was demonstrated among the 44 subjects with diabetes or hypertension. Group differences in treatment response were also demonstrated among subjects who did not report sinus problems, a comorbidity that could be anticipated to complicate treatment of SDB with CPAP.

A post hoc analysis was conducted to identify the specific measures that changed most with specific therapy. Of all measures assessed in the treatment-response score, the only individual measurement that significantly changed more with CPAP than with CT was the energy/fatigue subscore of the MOS (by 2.3 ± 16.8 and 10.3 ± 17.8 in the CT and CPAP groups, respectively, p < 0.05 by repeated-measures ANOVA for group differences).

Complications of Therapy

No serious complications of therapy were noted among subjects in either treatment arm of the study. Side effects related to therapy were skin irritation associated with CPAP mask use (n = 2) and minor nosebleeds related to use of the nasal steroid spray (in two subjects in the CT and one subject in the CPAP arm). These resolved with adjustments to the mask and discontinuation of the nasal spray. Two subjects in the CT arm and seven subjects in the CPAP arm reported use of antibiotics during the intervention period, for treatment of "colds" (n = 2), sinusitis (n = 2), nephrolithiasis (n = 1), and a skin infection (n = 1).

Interest in Continuing Therapy

Data from an exit interview, added to the protocol after study inception, were available for 57 subjects. In response to a query regarding interest in continuing current therapy (CPAP or nasal dilators), 13 of 18 (72%) of the subjects in the CT arm answered affirmatively, as compared with 36 of 39 (92%) subjects in the CPAP arm (p < 0.05).

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

The present study was the largest randomized study of the effectiveness of CPAP treatment for SDB, and the first study to focus on subjects with milder disorders (i.e., a group for whom there exists no consensus about optimal approaches to treatment). The study showed that even in subjects with mild disease (those with an RDI < 30 and without pathologic degrees of sleepiness) CPAP treatment resulted in a greater treatment response, as assessed by measures of changes in mood, functional status/general health, and energy/fatigue, than did a regimen that did not include nasal CPAP.

Previous studies of CPAP effectiveness have focused on patient groups with moderate to severe disease (apnea indices > 10 or RDI > 15). Small case series of patients studied before and after treatment have described decreased sleepiness (1, 4, 10), better mood (6, 7), and improved cognitive function (9, 10) with CPAP therapy. Engleman and associates compared sleepiness and cognition in 37 subjects who received either a CT regimen or CPAP (22). The findings in their study, in which treatments were not randomized, showed improved mood and MSLT scores, but no changes in cognition in the group that received CPAP as compared with the group that received CT. Engelman and associates later reported the results of a placebo-CPAP crossover study involving 32 patients with a wide range of apneic activity (mean RDI: 28; range: 7 to 128) (8). While receiving CPAP therapy, subjects were less sleepy and demonstrated improved mood, vigilance, executive functions, and quality of life. These previous studies, considered in aggregate, suggest beneficial effects related to the use of CPAP in subjects with moderate to severe SDB. Our study extends data to subjects with milder SDB (RDI: 5 to 30), demonstrating an approximately twofold improvement in an index of well-being in subjects treated with CPAP as compared with subjects treated with CT.

CPAP may have led to greater treatment responses by improving sleep quality and/or breathing during sleep, and/or reducing daytime sleepiness. CPAP therapy improved oxygen saturation and RDI level more so than did CT. Subjective sleepiness, measured with the ESS, also improved by a small amount in the CPAP group. Treatment responses of the MSLT, however, occurred in the absence of marked baseline sleepiness or significant changes with therapy. The finding of nonsignificant changes in the MSLT with treatment is not inconsistent with the findings in previous studies, in which changes in mood and cognition with CPAP were reported despite very small improvements in sleepiness (8, 10, 22) (e.g., MSLT times were 7.2 versus 6.1 min in the CPAP and CT groups, respectively, in the study by Engleman and colleagues [8]). In these studies, the MSLT, although increased by a small degree with treatment, remained within the abnormal range (8, 10, 22). Larger changes in sleepiness would be expected in more severely affected groups, because of statistical regression to the population mean. Persons with more extreme initial scores will show more change with repeated measures.

We explored the extent to which the impact of CPAP over CT varied in several subgroups. No relationship was observed between severity of SDB and likelihood of a CPAP treatment response within the RDI range examined. The subgroups that appeared to benefit the most from CPAP as compared with CT were subjects with underlying comorbidity (hypertension or diabetes) and those without a history of sinus problems. The first observation might be explained by greater perceived benefits in subjects concerned about their underlying health status. It is also possible that the subjects with hypertension or diabetes were more susceptible to adverse and reversible effects of SDB than were subjects without these medical problems. However, subjects with hypertension/diabetes were of comparable age and had BMI values similar to those of subjects without this comordibity, and also had similar MSLT values, ESS scores, and RDI values (data not shown).

Subjects most likely to respond to CT had a history of sinus problems. Conversely, a beneficial effect of CPAP was most evident among subjects without a history of sinus problems. These observations probably relate to potential difficulties in implementing CPAP therapy in patients with nasal or sinus problems. The observations also support the role of CT regimens that include treatment of underlying nasal inflammation and congestion as potentially beneficial interventions for mild SDB associated with nasal congestion or sinusitis.

Subjects with higher BMIs were more likely to improve with either CT or CPAP than were subjects with less obesity. Weight, however, did not significantly change with either intervention. Thus, it is possible that weight was a surrogate for disease severity not captured by RDI. Alternatively, it is possible that subjects with greater obesity benefited to a greater extent from interventions used in both the CPAP and CT arms (e.g., counseling on sleeping position, sleep hygiene, etc.) than did subjects with less obesity.

The object of this study was to evaluate the effects of therapy in subjects who could be successfully randomized to therapy and followed for 2 mo regardless of their specific adherence to therapy (i.e., an intention-to-treat analysis). However, patient acceptance of and adherence to CPAP therapy are known limitations of this therapy (26, 27). Adherence with CPAP in the present study was comparable with levels reported for patients referred to a sleep clinic because of symptoms, including studies of patients with more severe SDB (26, 27). Greater adherence was associated with a greater likelihood of treatment response. If analyses were restricted to subjects who used CPAP or nasal dilators (in the CT group) for at least 40% of sleep time, a stronger effect of CPAP over CT was observed (i.e., 17 of the 26 compliant CPAP participants experienced a treatment response [65%] as compared with eight of the 32 compliant CT participants [25%]; p < 0.01). In the entire sample, adjustment for hours of CPAP use increased the odds of a treatment response in the CPAP as compared with the CT group to 6.1 (CI: 1.8 to 20.5). Alternatively, if all subjects who refused to be randomized to a specific treatment arm or who were lost to follow-up are considered to be treatment failures, then the proportion of subjects who benefited from CPAP declined from 48% to 40%, and the proportion of responders in the CT group declined from 27% to 25% (the p value for CPAP-versus-CT difference increased to 0.056.) Thus, although our data suggest that treatment responses were greater in subjects randomized to CPAP than to CT, they also emphasize the importance of compliance in influencing treatment effectiveness, and the need to develop approaches for using CPAP that improve patient adherence with this therapy.

This randomized study was not conducted in a sleep-clinic population. The majority of participants were volunteers (interested or concerned with their snoring) who had had no previous contact with a sleep referral center or medical evaluation for a sleep problem. The use of a largely volunteer population was required to recruit subjects who met strict study eligibility criteria (to address issues related to the neuropsychologic consequences of SDB), and also to allow random assignment to interventions regardless of the preferences of the participant's health-care provider. Additionally, the "control" group did not receive a placebo, but rather was treated with a medical regimen that some clinicians consider "first-line" treatment for mild SDB. Both of these design features could be expected to minimize any group differences between the CT and CPAP groups. First, subjects referred to a sleep clinic because of concerns about sleepiness or impaired function might be expected to experience greater effects of intervention as a result of the initial severity of their condition. Second, differences between the CPAP and the comparison group would likely have been larger had the comparison group been a no-intervention control group.

The findings in this study may be less biased than those in studies of outcomes performed with patients identified exclusively through a sleep laboratory. Such subjects are more likely to suffer from symptoms (for any given RDI) than are randomly identified subjects from the community. Subjects with subjective pathologic sleepiness were excluded from the present study; however, the use of a volunteer sample raises the possibility that treatment responses may have been influenced by unmeasured comorbidities that influenced study participation. Nevertheless, the greater treatment response to CPAP than to CT in subjects with an RDI < 30 suggests that beneficial effects may be derived from CPAP even in subjects with a level of apneic activity usually considered "mild" or "mild-moderate" in severity.

All efforts in this study were made to interact as intensively with subjects in the CT as in the CPAP arm. For example, subjects in both arms were called every 2 wk to remind them to use a "device" (a nasal dilator or CPAP) on a nightly basis. Subjects in both arms were treated similary regarding weight reduction and sleep-hygiene counseling. We cannot, however, exclude the possibility that subjects in the CPAP arm may have perceived a greater treatment response because of the perception of being in a more "active" arm of the protocol. However, subjects in the CT arm appeared as "interested" in the use of nasal dilators as subjects in the CPAP arm appeared interested in the use of CPAP.

The outcome measure used in the study was constructed with data collected from well-accepted outcome tools. A priori, we identified mood, energy/fatigue, and functional status/ general health to be domains of interest in SDB. Because relatively little is known about the sensitivity of the study instruments to interventions in patients with sleep disorders, we considered improvement in each domain on the basis of changes of moderate size (0.5 SD) in any of several overlapping measurements. Although this approach may increase the likelihood of observing improvements by chance alone, such a phenomenon would influence responses in both the control and CPAP groups. Observing a differential effect of treatment in the CPAP arm, and requiring observed improvements in at least two of the three domains of interest, make it unlikely that the more than two-fold difference in treatment responses between subjects randomized to CPAP as compared with CT was spurious.

There clearly is a paucity of data that might clarify which patients with SDB benefit from specific therapy aimed at reversing breathing abnormalities during sleep. Characterization of threshold effects (levels of RDI, sleepiness) that identify those individuals who might benefit most from relatively expensive diagnostic (i.e., polysomnography) and therapeutic (i.e., CPAP) interventions are important for improving the rational utilization of health-care resources. This study does not address the long-term impact on morbidity and mortality related to SDB, and the extent to which specific therapies may affect this. This study does provide new data, however, that suggest that the addition of CPAP therapy to standard medical approaches to treatment of snoring (i.e., weight reduction, position changes, nasal decongestants, and sleep hygiene) positively influences domains relevant to patients with sleep disorders. Our data suggest that if improved well-being and functional status are goals of medical therapy for SDB, it may be inappropriate to strictly use levels of RDI or MSLT to make treatment decisions. Confirmation of our findings in other populations and with other outcome measures would further support screening and treatment strategies aimed at the large proportion of the population with snoring and mild to moderate levels of SDB.