Published ahead of print on August 2, 2007, doi:10.1164/rccm.200611-1588OC
© 2007 American Thoracic Society doi: 10.1164/rccm.200611-1588OC
Continuous Positive Airway Pressure Treatment of Mild to Moderate Obstructive Sleep Apnea Reduces Cardiovascular Risk1 Department of Internal Medicine I, Marienhospital Herne, Ruhr University Bochum, Herne, Germany; 2 Department of Internal Medicine, Bethesda Hospital, Wuppertal, Germany; and 3 Department of Cardiology and Angiology, St. Josef-Hospital/Bergmannsheil, Ruhr University Bochum, Bochum, Germany Correspondence and requests for reprints should be addressed to Nikolaus J. Büchner, M.D., Department of Internal Medicine I, Marienhospital Herne, Ruhr University Bochum, Hoelkeskampring 40, D-44625 Herne, Germany. E-mail: nikolaus.buechner{at}rub.de
Rationale: Obstructive sleep apnea (OSA) is linked to increased cardiovascular risk, but the impact of mild forms of OSA and their treatment on cardiovascular outcomes remains controversial. Objectives: To prospectively investigate cardiovascular outcomes in treated versus untreated patients with OSA. Methods: Consecutive sleep laboratory patients with all degrees of OSA were included. Endpoints were nonfatal (myocardial infarction, stroke, and acute coronary syndrome requiring revascularization procedures) and fatal (death from myocardial infarction or stroke) cardiovascular events. Measurements and Main Results: Comparison of event-free survival rates in treated versus untreated patients (Kaplan-Meier estimates, log-rank test). Of 449 patients enrolled (age, 56.0 ± 10.5 years; body mass index, 30.8 ± 5.4 kg/m2), 364 patients received OSA treatment, and 85 patients remained untreated. Median follow-up was 72.0 months (range, 1–156). Mean apnea–hypopnea index before treatment was 30.9 ± 21.8/hour in treated and 15.3 ± 13.0/hour in untreated patients, but there were no differences in cardiovascular comorbidities or risk factors. In patients with mild–moderate OSA (n = 288), events were more frequent in untreated patients (estimated event-free survival at 10 yr, 51.8 vs. 80.3% [P < 0.001]; absolute risk reduction, 28.5%; number needed to treat to prevent one event/10 yr, 3.5). After adjustment for age, gender, cardiovascular risk factors, and comorbidities at baseline, OSA treatment was an independent predictor for events (hazard ratio, 0.36; 95% confidence interval, 0.21–0.62; P < 0.001). Conclusions: OSA treatment was associated with a cardiovascular risk reduction of 64% independent from age and preexisting cardiovascular comorbidities. OSA treatment should be considered for primary and secondary cardiovascular prevention, even in milder OSA.
Key Words: obstructive sleep apnea • continuous positive airway pressure therapy • outcome studies • cardiovascular mortality
It has been clearly established that patients with untreated obstructive sleep apnea (OSA) have high levels of cardiovascular morbidity and mortality (1–13). In addition, cross-sectional and case-control studies found high rates of OSA in unselected patients with cardiovascular disease (14–16), and high rates of cardiovascular morbidity have been reported in patients suffering from sleep apnea (1–3), even in patients with no preexisting cardiovascular disease (4–6). However, some of these epidemiological studies have been criticized for methodological limitations because they were retrospective (8, 10, 11) or did not sufficiently control for confounding factors (5, 8, 10, 11). Classical cardiovascular risk factors occur frequently in patients with OSA, which makes it difficult to separately assess the independent cardiovascular risk due to OSA itself. This may explain why some studies have failed to clearly demonstrate an independent association between OSA and cardiovascular risk after adjustment for age and obesity (3, 7, 9, 17). Another important issue is whether OSA treatment reduces cardiovascular morbidity and mortality. Early support for a causal relationship of OSA treatment and cardiovascular risk reduction was derived from intervention studies that showed beneficial effects of OSA treatment on blood pressure (18–20). Recent prospective studies have shown that continuous positive airway pressure therapy (CPAP) reduces cardiovascular risk in patients with OSA (5, 21–26). Thus, in a 10-year follow-up study, CPAP therapy reduced fatal and nonfatal events in men with severe OSA compared with a population-based sample of age- and body mass index–matched healthy men (23). Despite these results, the impact of OSA treatment on cardiovascular risk remains controversial (27–29). Other important questions remain. The proportion of elderly persons in the general population has dramatically increased over the past few decades. The prevalence of sleep apnea increases with age, but it is unclear whether elderly patients benefit from OSA treatment. For example, the mortality rate did not show a significant increase in patients with severe OSA aged 50 to 79 years (10) when compared with the general population. Moreover, prospective studies have failed to demonstrate any association between the apnea–hypopnea index (AHI) and mortality rates in elderly patients (7, 30–33). Another important question is whether CPAP treatment improves cardiovascular outcomes in patients with mild–moderate OSA. A trend toward higher cardiovascular risk has been repeatedly demonstrated in patients with mild–moderate OSA, but cardiovascular outcomes in these patients have not been sufficiently well evaluated in prospective treatment studies. In the future, it is likely that milder forms of OSA will be diagnosed because there is a growing public awareness of sleep apnea as an independent disease. Thus, it is essential to determine whether cardiovascular outcomes in such patients can be improved by CPAP therapy. The present prospective study evaluated the impact of CPAP therapy on cardiovascular outcomes in a large cohort of male and female patients suffering from all degrees of OSA.
Patients From 1993 until 1998, we prospectively recruited all patients with suspected obstructive sleep–related breathing disorders admitted to our sleep clinic through unselected referral from primary or secondary care physicians. Snorers without apnea and patients with central sleep apnea, Cheyne-Stokes' respiration, hypoventilation syndromes, or periodic limb movement in sleep as the predominant finding were excluded from the study (Figure 1).
Patients underwent overnight attended polysomnography (Somnostar 4100; SensorMedics Co., Yorba Linda, CA). Continuous recordings included electroencephalography, electrooculography, submental and leg electromyography, electrocardiography, noninvasive sensors for nasal and oral airflow, tracheal sounds, thoracic and abdominal respiratory movement, and oxyhemoglobin level (Sat-Trak finger-pulse oximeter; SensorMedics Co.). Polysomnography records were scored for sleep, breathing, and oxygenation at 30-s periods. Sleep stage was assessed manually using the methods of Rechtschaffen and Kales. Apnea was defined as a complete cessation of airflow lasting  10 seconds, and hypopnea was defined as a 50% reduction in respiratory airflow lasting longer than 10 seconds associated with an arousal or oxygen desaturation by 4%. Obstructive apnea was defined as absence of airflow in the presence of paradoxical chest-wall motion. The AHI was calculated according to the average number of episodes of apnea and hypopnea per hour of sleep. OSA was diagnosed if AHI was 5 events/hour (mild OSA was defined as AHI 5 to < 15/h, moderate OSA was defined as AHI 15 to < 30/h, and severe OSA was defined as AHI > 30/h).
Treatment
Data Collection
Definition of Risk Factors and Cardiovascular Disease
Follow-up
Endpoints
Statistical Analysis
A total of 449 patients (384 male, 65 female) were recruited. Treatment for OSA was initiated in 364 patients (CPAP in 296, bilevel positive airway pressure in 48, and intraoral protrusion devices in 20). Eighty-five patients refused treatment. Patient baseline characteristics are listed in Table 1. Briefly, BMI was lower in the untreated group (29.3 ± 5.4 vs. 31.2 ± 5.4 kg/m2; P = 0.003) without further significant differences regarding age, gender, medication, cardiovascular risk factors, and preexisting cardiovascular disease.
In the untreated group, AHI was lower, and the minimal oxygen saturation was higher without further significant differences in polysomnographic parameters (Table 2). In treated patients, AHI was reduced from 30.9 ± 21.8/hour to 5.6 ± 8.3/hour (P < 0.001), and values of oxygen saturation improved markedly (data not shown). In treated patients with mild–moderate OSA, AHI decreased from 15.5 ± 6.9/hour to 4.1 ± 6.7/hour (P < 0.001) (Table 3). No significant changes in these parameters were observed in untreated patients. In treated patients, 21.5% were noncompliant with CPAP or bilevel positive airway pressure.
The median follow-up period was 72.0 months (range, 1–156 mo). The mean duration of follow-up was significantly different between untreated and treated patients (50.0 ± 49.4 vs. 77.0 ± 55.0 mo; P = 0.001). One hundred sixty-two treated and 26 untreated patients visited the hospital during the last 6 months of the study. Thirty-four untreated and 151 treated patients for whom no data were available from the last 6 months of the study were reached by phone. In the remaining patients, who were lost to last follow-up (51 [14.0%] treated and 25 [29.4%] untreated patients), the latest available dataset was used for censoring of event-free survival. A total of 76 events occurred during the observation period in the whole study group. Among the treated patients, there were nine myocardial infarctions (2.4%), 25 revascularization procedures (6.8%), 10 strokes (2.7%), eight cardiovascular deaths (2.2%), and 24 deaths of all cause (6.6%). Among the untreated patients, there were five myocardial infarctions (5.8%), 11 revascularization procedures (12.9%), five strokes (5.8%), and three deaths due to cardiovascular causes (3.5%). Events were more frequent in untreated than in treated patients (24 [28.3%] in untreated vs. 52 [14.3%] in treated patients, P = 0.009; estimated event-free survival at 10 yr, 51.8% vs. 79.7%, log rank test: P < 0.001; absolute risk reduction, 27.9%; number needed to treat to prevent one cardiovascular event over 10 yr, 3.6) (Figure 2A).
In patients with mild–moderate OSA, more events occurred in untreated (25.3% [n = 20]) than in treated patients (14.4% [n = 30]; P = 0.024). The estimated event-free survival after 10 years was 51.8% in untreated and 80.3% in treated patients (log-rank test, P < 0.001; absolute risk reduction, 28.5%; number needed to treat to prevent one event over 10 yr, 3.5) (Figure 2B). In treated patients who were compliant to CPAP, the estimated event-free survival after 10 years was 83.1% (log-rank test, P < 0.001 for comparison with untreated patients). The first event tended to occur earlier in untreated than in treated patients (46.3 vs. 56.5 mo), but this difference was statistically not significant. OSA treatment was assessed in a Cox proportional hazards model after adjustment for age, gender, BMI, cardiovascular risk factors, cardiovascular disease, COPD, and malignant disorders. The results are summarized in Table 4. In the final model, OSA treatment remained a significant predictor for events. The risk of events was significantly lower in treated patients when compared with that in untreated patients. The risk reduction resulting from OSA treatment was 64% for the whole study group and for patients with mild–moderate OSA.
In patients with mild–moderate OSA without preexisting cardiovascular disease, estimated event-free survival after 10 years was 90.7% in treated versus 68.5% in untreated patients (log rank test, P = 0.012) (Figure 2C). The adjusted HR for OSA treatment in these patients was 0.37 (95% CI, 0.17–0.78).
It is generally accepted that severe forms of OSA require treatment with CPAP to reduce cardiovascular risk, especially when associated with daytime sleepiness. However, it remains uncertain whether milder forms of OSA merit similar therapeutic attention. Reactive oxygen species (34), proinflammatory substances (35, 36), and soluble adhesion molecules (37) are increased in patients with OSA. In addition, one can expect that even in milder forms of OSA, repetitive apneas may increase sympathetic nerve activity and that this could persist during the day (38) and cause endothelial dysfunction and hypertension. We therefore performed a prospective observational study to evaluate the impact of OSA treatment on cardiovascular outcome in a large cohort of patients suffering from all degrees of OSA. The main new finding of our observational study is that CPAP therapy was associated with a significant decrease in cardiovascular events even in mild to moderate forms of the disease. Our study findings differ from other recent outcome studies (23) by showing a general benefit of OSA treatment on cardiovascular outcome in patients with all degrees of OSA. This is in contrast to the study by Marin and colleagues, who found a higher incidence of fatal and nonfatal cardiovascular events only in untreated patients with severe OSA in a large prospective controlled study, whereas simple snoring, mild–moderate OSA, or CPAP-treated severe OSA, was not associated with an increased risk for such events (23). This discrepancy is most likely explained by differences between the study cohorts in comorbidities, hypersomnolence, or compliance of the treated groups. Our study cohort was characterized by a higher proportion of cardiovascular risk factors. For instance, in the OSA-treated group using the same definitions as Marin and colleagues, hypertension (70.8 vs. 35.1%), diabetes mellitus (22.7 vs. 11.3%), and hyperlipidemia (59.2 vs. 7.9%) were much more frequent in our patients. In addition, there was a higher prevalence of preexisting cardiovascular disease in our patients (29.2 vs. 8.5%). However, hypertension, diabetes mellitus, and hyperlipidemia are not simple confounders independent from OSA. They are part of the causal pathway that may account for the adverse effects of OSA on cardiovascular outcome. Therefore, the higher proportion of cardiovascular risk factors and diseases could explain the benefit of OSA treatment in patients with mild to moderate OSA in the present study. In another large study that treated only patients with severe OSA (22), cardiovascular death rates but not new cases of hypertension, cardiac disorder, or stroke were reduced by CPAP therapy. This contrasts with our and earlier findings (4–6). A surprisingly low prevalence of hypertension (23.4% in CPAP responders, 19.7% in CPAP nonresponders) and diabetes (6.5% in CPAP responders, 3.3% in nonresponders) was reported by Doherty and colleagues (22). This differs from findings of previous large epidemiological studies (2, 4). OSA therapy with diet alone (defined as 10% weight loss in the first year after diagnosis) has shown a similar risk reduction as CPAP therapy (HR [95% CI], 0.23 [0.09–0.56] vs. 0.20 [0.06–0.62]; P < 0.05 for both) (24). However, changes of body weight were not a confounding factor in our study because BMI remained stable in treated or untreated patients with or without events. Our results have limitations that have to be taken into consideration. Limitations in causal inferences derived from observational studies such as ours lie in potentially uncontrolled extraneous factors, a selection bias, and reverse causality (39). The most convincing way to prove a cause-and-effect inference, in this case the benefit of CPAP on cardiovascular outcome, would be to conduct a prospective controlled, randomized trial (39). This study design is ethically not justifiable in patients with OSA, at least in a long-term setting. Consequently, there are no randomized controlled outcome studies that address this issue among previous reports. In this situation, one has to rely on observational studies to evaluate the effects of CPAP on cardiovascular outcome. Because randomization of patients was not appropriate for the present study, patients who refused CPAP therapy served as control subjects. This design was chosen by the majority of previous outcome studies (22–26). Patients who declined medical treatment could have been different in terms of lifestyle, comorbidities, and compliance with cardiovascular medications. Nevertheless, at baseline our untreated patients did not significantly differ in cardiovascular comorbidities, medication, or coexisting malignant or pulmonary disorders. BMI was lower in the untreated group (see Table 1), which should have favored these patients with regard to cardiovascular outcomes (40). In addition, BMI, which is a poor surrogate parameter of lifestyle, did not change in treated or untreated patients during follow-up. As in most cohort studies, other potential cardiovascular risk factors have been documented at baseline, but changes were not sufficiently controlled during follow-up. However, changes in cardiovascular medication at last follow-up did not differ between the two groups. Therefore, it is unlikely that our results are influenced by differences of incident cardiovascular risk factors. Most of the classical cardiovascular risk factors are common in patients with OSA. Their role as confounding factors in the interpretation of the present results is unclear because some of them (in particular hypertension, diabetes, and hyperlipidemia) are thought to be involved in the pathological mechanisms that could explain the benefit of CPAP regarding cardiovascular protection. The impact of OSA treatment on cardiovascular events was found to be independent from the defined cardiovascular risk factors. A selection bias could be assumed among the cardiovascular risk factors. However, the distribution in these risk factors was not different between the OSA-treated and nontreated patients. Furthermore, there is no evidence of an adverse effect of cardiovascular risk factors on the acceptance or nonacceptance of the OSA treatment and thus no evidence of a reverse causality. Because a randomized trial was not appropriate to this area, we feel that the observational approach allows a useful causal inference to be made about the beneficial effect of OSA treatment on cardiovascular events. Another important point is that the duration of observation period for the two groups was different. It was shorter in untreated patients because they are less likely to show up for follow-up investigations. No data could be obtained during the last 6 months of the observation period in 14.0% of treated and in 29.4% of untreated patients. In these patients, we used the latest available data set from annual visitations for censoring of the event-free survival. As the main finding of our study, events were more frequent and tended to occur earlier in untreated patients. We cannot exclude that event rates in untreated patients would have declined if the follow-up period had been longer, resulting in an attenuation of the beneficial effects of OSA treatment on cardiovascular outcomes. Campos-Rodriguez and colleagues found CPAP compliance to be a strong predictor of mortality mostly from cardiovascular disease, with an adjusted OR of 0.10 (95% CI, 0.04–0.29) for high compliance (>6 h/d) and an OR of 0.28 (95% CI, 0.11–0.69) for moderate compliance (1–6 h/d) (21). At the beginning of the present study, we were not able to record compliance data reliably because most of the available CPAP devices were not equipped with a built-in time counter. Subsequently, some of the devices were exchanged to allow accurate recording of compliance data. In the remaining cases, we used information derived from the patients. Even if patients had tended to overestimate the hours of CPAP usage, the presented compliance data agree with previously published results (41–44). We therefore cannot exclude that our results for the treated group were affected by undetected CPAP refusers. Thus, we assume that the potential cardiovascular benefit of consequent CPAP treatment could be underestimated in our study. In the general population, the ratio of male to female patients with OSA is estimated to be about 2:1 to 3:1. In our study, the percentages of women in treated and untreated patients were 14.1 and 16.5%, respectively. This is a typical phenomenon in a sleep laboratory cohort (21, 22, 24). Nevertheless, our results may not be generalizable to the female population. Another limitation of our study is that we did not compare our outcome data with healthy matched control subjects. However, event-free survival rates of treated patients were within the range seen in previous observations. Taken together, the inherent limitations of this observational study might impair the strength of cause-and-effect inference of our results, but we believe that the improvement in cardiovascular outcome of treated versus untreated patients OSA is too large to be explained by the limitations discussed here.
Conclusions
* These authors contributed equally to this article.  Originally Published in Press as DOI: 10.1164/rccm.200611-1588OC on August 2, 2007 Conflict of Interest Statement: None of the authors has a financial relationship with a commercial entity that has an interest in the subject of this manuscript. Received in original form November 6, 2006; accepted in final form August 1, 2007
This article has been cited by other articles:
|
||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
TOP
ABSTRACT
AT A GLANCE COMMENTARY
0.25 were included in a Cox proportional hazards model. Results of Cox regression are presented as hazard ratio (HR) ± 95% confidence interval (CI). 
