INTRODUCTION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Keywords: angina pectoris; coronary disease; prognosis; sleep apnea syndromes

The effect of sleep-disordered breathing in terms of cardiovascular morbidity and mortality is still largely unknown and is under debate (1). Studies from sleep clinics that have found an increase in mortality have met with major criticism, partly because of retrospective study designs (2, 3). Others have interpreted the increase in cardiovascular mortality in patients with disordered breathing as an effect of coexisting hypertension and other risk factors (4, 5).

Both men and women with verified coronary artery disease have a high occurrence of disordered breathing (6, 7). Repetitive apneas and desaturations cause sympathetic activation and hemodynamic stress that may be particularly hazardous in this patient group. Studies including patients with coronary disease have yielded conflicting data about the prognostic importance of disordered breathing (8). Differences in patient selection, study design, and small study groups could explain the inconsistent results.

The aim of the present study was prospectively to compare the mortality and cardiovascular morbidity in patients with and without disordered breathing. All the participants had verified coronary artery disease.

	METHODS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Patients

Men and women aged 70 yr and younger who had been referred to Umeå University Hospital for coronary angiography because of disabling angina pectoris were included in the present study between 1992 and 1995. Umeå University Hospital has a catchment area that covers the northern part of Sweden and includes about one million people. Coronary disease was verified by coronary angiography before inclusion. One group of patients (n = 299) participated in a prevalence study and was randomly selected, by lot, from all the eligible patients on that day for an overnight sleep study. The remaining patients (n = 109) were included consecutively according to recording capacity.

The study was approved by the local ethics committee and was conducted in accordance with the Second Declaration of Helsinki. All the patients gave their informed consent.

Sleep Studies

The following variables were recorded during one night's sleep in the hospital: oronasal air flow using a three-way thermistor (Nihon Khoden ZE-732A, Japan), blood oxygen saturation and heart rate by pulse oximetry with a finger transducer (Ohmeda Biox 3700, Louisville, CO), respiratory and body movements using a pressure-sensitive bed (polyvinylidenefluoride foil, Apnomat; Duorek Ltd, Raisio, Finland) (11), and sleep position with a body position indicator (Vitalog Monitoring Inc., Redwood City, CA). All the signals were continuously sampled and stored in a Macintosh II computer and displayed online.

Desaturations and apneas were scored manually without any knowledge of the clinical characteristics. A desaturation was defined as a decrease in oxygen saturation of 4% or more. An apnea was defined as a cessation in air flow lasting at least 10 s and a hypopnea as a reduction in air flow of at least 50% compared with baseline, in combination with a decrease in oxygen saturation or a pulse alteration. The duration of sleep was estimated from the pressure-sensitive bed recording. This technique compares favorably with electroencephalogram (EEG) in assessing sleep time (11). No sedative medication was given.

Sleep-disordered breathing was measured as the oxygen desaturation index (ODI) and apnea-hypopnea index (AHI), calculated as the average number of episodes of desaturation and apnea or hypopnea per hour of sleep, respectively. An AHI of  10 and an ODI of  5 were defined prospectively as the criteria for sleep-disordered breathing.

End Points

The primary end point was a composite of death from any cause, stroke or transitory ischemic attack (TIA), and myocardial infarction (MI). The individual components were analyzed as secondary end points. The WHO definition of stroke was used: rapidly developing clinical signs of focal (or global) disturbance of cerebral function lasting for more than 24 h with no apparent cause other than a vascular origin (12). A focal disturbance lasting less than 24 h was classified as TIA.

A diagnosis of MI was based on either autopsy findings, or typical chest pain, ECG findings, and a diagnostic elevation of cardiac enzymes. Two out of three of the clinical criteria were required. Only new Q waves were used for the diagnosis of myocardial infarction in association with coronary artery bypass grafting.

All patients, or relatives in the event of death, were contacted by telephone or written questionnaire for an assessment of outcomes. Hospital records were obtained for all patients for the source documentation of end points. Death certificates and autopsy reports, if available, were obtained for any patients who died during the follow-up period. End point classification was made without any knowledge of sleep study results.

Statistical Analysis

On the basis of previous experience and follow-up data, we estimated the 5-yr event rate for the composite end point at 20% in patients without disordered breathing. An absolute increase in event rate of 3% per year in patients with disordered breathing was considered to be of clinical importance. The recruitment of approximately 350 patients was therefore required to reach 80% power with a significance level of 0.05. To obtain a median follow-up time of approximately 5 yr, all the patients were followed until the last included patient had completed 3.5 yr of follow-up.

Data were analyzed with the STATISTICA 5.5 software modules (StatSoft Inc., Tulsa, OK). Group data were expressed as the mean and standard deviation (SD) for continuous variables and as rates for variables on a nominal scale. Differences between two means were assessed with a t test for unpaired data. Differences between proportions were analyzed with the chi-squared test. In all the statistical tests, the null hypothesis was rejected at the 5% level (p  0.05).

Graphs of the Kaplan-Meier estimate of the survival function were calculated for patients with and without disordered breathing and were compared between groups with the log-rank test. End point results are expressed as risk ratios and 95% confidence intervals (CI). Cox's proportional hazards model was used to identify predictors of the study end points. Variables of disordered breathing (ODI  5 and AHI  10) were analyzed in separate models, which also included variables associated statistically with the study end point in univariate analysis or considered to be of potential clinical interest.

	RESULTS

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Two hundred and eighty-nine men and 142 women were included. Sixteen patients (5 men, 11 women) refused to participate. After exclusions for technical failures, complete analyses of oxygen desaturation were obtained in 277 men and 130 women. Complete apnea analyses were obtained in 264 men and 128 women. All of them had stable angina pectoris in Canadian Cardiovascular Society class II (15%, n = 61) or class III (85%, n = 347) (13). No differences in age, body mass index, ODI, or AHI were found between randomly and consecutively included patients.

One-vessel disease was found in 19%, two-vessel disease in 28%, and three-vessel disease in 39%. Left main stem stenosis was found in 14%. Left ventricular function was assessed visually from left ventriculograms and scored as good (72%), fair (24%), and poor (4%), corresponding to ejection fractions of approximately > 0.5, 0.35-0.5, and < 0.35, respectively.

The median follow-up period was 5.1 yr. A coronary intervention was performed in 77.5% (316 of 408) of the patients during follow-up. Eight patients were on treatment (continuous positive airway pressure [CPAP] in two, dental devices in three) or had been operated on (uvulopalatopharyngoplasty in three patients) for disordered breathing at the end of follow-up and another six had stopped treatment for various reasons. All patients and their referring physicians (in cases with disordered breathing) were informed about the sleep study result. The most obvious explanation for the infrequent use of treatment is that the patients were included on the basis of coronary artery disease and not because of sleep disturbances or daytime sleepiness.

Clinical characteristics and medical treatment are presented in Tables E1 and E2 in the online data supplement. Patients with disordered breathing were slightly older, had a higher body mass index (BMI) and waist/hip ratio, and more frequently had an impaired left ventricular function. They also used diuretics more frequently. Other medical treatment was similar in the two groups.

The distribution of ODI and AHI in the study population is presented in Table 1. The proportion of patients with disordered breathing was 38% when an ODI of  5 was used as the diagnostic limit and 34% when an AHI of  10 was used.

At the end of the study, there was a significant 70% relative increase and 10.7% absolute increase in the primary composite end point of death, cerebrovascular events, and myocardial infarction in patients with disordered breathing defined as an ODI of  5. Similarly, patients with an AHI of  10 had a significant 62% relative increase and a 10.1% absolute increase in the composite end point (Table 2). Of the individual end point components, the occurrence of cerebrovascular events was most markedly increased. The 33 cerebrovascular events consisted of 26 cerebral infarctions, 5 TIAs, and 2 cerebral hemorrhages. All these patients were examined with computed tomography of the brain.

In the disordered breathing group, the cause of death was myocardial infarction or sudden death in eight patients, congestive heart failure in five, stroke in one, and aortic disease in one. In the nondisordered breathing group, the cause of death was myocardial infarction or sudden death in four patients, congestive heart failure in one, aortic disease in one, cancer in six, and traumatic death in two. Fifteen cardiovascular deaths occurred in patients with disordered breathing. Four of them occurred between midnight and 6 A.M. and four between 6 A.M. and noon.

The cumulative event-free survival in relation to ODI is shown in Figure 1. The curves diverge throughout the follow-up period.

View larger version (7K): [in this window] [in a new window]   Figure 1.   Cumulative event-free survival in patients with and without sleep-disordered breathing (ODI  5). The primary end point included death of any cause, cerebrovascular events, and myocardial infarction.

The primary end point occurred in 28.3% of men with disordered breathing and in 15.9% of men without disordered breathing measured as an ODI  5. The corresponding percentages for women were 20% and 14.4%.

Using multivariate analysis with Cox's proportional hazards model, including an ODI of  5 as a measure of disordered breathing, four variables were found to be independently associated with the primary end point (Table 3). An ODI of  5, diabetes, and left ventricular dysfunction all increased the risk of having an end point whereas coronary intervention decreased the risk. When an ODI of  5 was replaced by an AHI of  10 in the same model, disordered breathing did not reach significance (p = 0.15). The models also included the variables of age, sex, BMI, and hypertension. An ODI of  5 and an AHI of  10 were both independently associated with cerebrovascular events (hazard ratio 2.62, 95%CI 1.26-5.46, p = 0.01 and hazard ratio 2.98, 95%CI 1.43-6.20, p = 0.004, respectively) in separate models, which also included the variables of age, sex, BMI, hypertension, diabetes, left ventricular dysfunction, and coronary intervention. Diabetes was the only additional independent predictor of cerebrovascular events in these models (Table 3). Neither an ODI of  5 nor an AHI of  10 was independently predictive of the single end points of death or myocardial infarction.

	DISCUSSION

TOP ABSTRACT INTRODUCTION METHODS RESULTS DISCUSSION REFERENCES

Our results show that sleep-disordered breathing in patients with coronary artery disease is associated with a 60% to 70% increase in the risk of death and cardiovascular morbidity during long-term follow-up. Moreover, disordered breathing is associated with an almost 3-fold higher risk of cerebrovascular events after adjustment for other risk factors.

The present study is the first prospective evaluation designed and powered reliably to assess the consequences of sleep-disordered breathing in terms of mortality and cardiovascular morbidity in patients with coronary disease. The increase in risk is applicable to approximately one-third of all patients with coronary artery disease who have concomitant disordered breathing (6, 7).

The association between disordered breathing and mortality has been previously examined in different general populations with inconclusive results (1). The acquisition of reliable estimates of a potential association cannot be expected, however, when inadequate sample sizes or retrospective study designs are used (2, 3, 14, 15). Even in studies with adequate sample sizes and a prospective design, only a weak or insignificant independent association with mortality has been reported (4, 16).

It is reasonable to believe that the effect of disordered breathing in terms of hypoxemia, changes in blood pressure, and sympathetic activation is particularly critical in patients with established vascular disease. The evidence of a clinically important association between disordered breathing and cardiac events in this patient group is growing but still limited (8, 10, 17). Most investigators have examined the impact of disordered breathing on cardiac arrhythmia and myocardial ischemia, whereas the long-term effect on hard end points has been unknown. In the present study, an independent association with the composite primary end point was found for an ODI of  5. Diabetes and left ventricular dysfunction were both associated with an approximate 2-fold increase in risk, whereas coronary revascularization reduced the risk of having an end point by 50%. These findings correspond well with previous knowledge.

The relationship between sleep-disordered breathing and cerebrovascular events has not been previously examined in a prospective study. In a case-control study, a relationship was found between a self-reported history of snoring and the risk of stroke (21). Cross-sectional studies have reported a higher prevalence of sleep apnea in patients suffering a recent stroke than in control subjects (22, 23). Stroke can cause sleep apnea, but preexisting disordered breathing was presumed on clinical grounds in a large proportion of the patients (23). In a recent cross-sectional study, a significant association between disordered breathing and self-reported stroke was found (24). The present results indicate that disordered breathing is an independent risk factor for stroke in patients with coronary artery disease.

Several mechanisms may contribute to an increased risk of vascular events in patients with disordered breathing. Prothrombotic effects as a result of a decrease in fibrinolytic function have been reported (25, 26). Effects on cerebral blood flow and other hemodynamic effects may also be of importance (27, 28). An atherogenic effect by hypoxemia has been suggested and would increase the risk of vascular morbidity, particularly in patients with established vascular disease (29, 30). Our results suggest that the cerebral circulation is the most vulnerable.

We used a prospective cohort study design with adjustment for the effects of confounding factors, as it is not possible to use a randomized design in the present kind of investigation. One limitation of the methodology used here was that EEG with sleep stage recording was not performed. Sleep time was instead estimated from the pressure-sensitive bed recording. The pressure-sensitive bed is a movement sensor, which effectively changes mechanical displacement into electrical signals. The duration of sleep measured by EEG and the pressure-sensitive bed compares favorably (11, 31). Poor left ventricular function with congestive heart failure may be associated with apneas of the central type (without concomitant breathing efforts) (32). We did not distinguish the different apnea types, but the proportion of patients with poor left ventricular function was low and none of them had symptoms of congestive heart failure at the time of the sleep study, indicating a low frequency of central apneas among the present patients.

Poor left ventricular function was more common in patients with disordered breathing, adding strength to the hypothesis that sleep-disordered breathing may be a risk factor for heart failure (24, 33).

We conclude that sleep-disordered breathing is associated with a worse prognosis in patients with coronary disease. Furthermore, there is an independent association with cerebrovascular events.