Kaplan-Meier estimates of survival for patients with obstructive sleep apnea (OSA) vs control subjects (P = .02) and patients with central sleep apnea (CSA) vs controls (P = .05).
Survival curves calculated from a Cox proportional hazards regression model after adjustments for age, sex, body mass index, current smoking, hypertension, diabetes mellitus, atrial fibrillation, Mini-Mental State Examination score, and Barthel index of activities of daily living. Mortality was higher among patients with obstructive sleep apnea (OSA) vs control subjects (adjusted hazard ratio, 1.76; 95% confidence interval, 1.05-2.95; P = .03); there was no difference in mortality between patients with central sleep apnea (CSA) and controls (P = .80).
Sahlin C, Sandberg O, Gustafson Y, Bucht G, Carlberg B, Stenlund H, Franklin KA. Obstructive Sleep Apnea Is a Risk Factor for Death in Patients With StrokeA 10-Year Follow-up. Arch Intern Med. 2008;168(3):297–301. doi:10.1001/archinternmed.2007.70
Sleep apnea occurs frequently among patients with stroke, but it is still unknown whether a diagnosis of sleep apnea is an independent risk factor for mortality. We aimed to investigate whether obstructive or central sleep apnea was related to reduced long-term survival among patients with stroke.
Of 151 patients admitted for in-hospital stroke rehabilitation in the catchment area of Umeå from April 1, 1995, to May 1, 1997, 132 underwent overnight sleep apnea recordings at a mean (SD) of 23 (8) days after the onset of stroke. All patients were followed up prospectively for a mean (SD) of 10.0 (0.6) years, with death as the primary outcome; no one was lost to follow-up. Obstructive sleep apnea was defined when the obstructive apnea-hypopnea index was 15 or greater, and central sleep apnea was defined when the central apnea-hypopnea index was 15 or greater. Patients with obstructive and central apnea-hypopnea indexes of less than 15 served as control subjects.
Of 132 enrolled patients, 116 had died at follow-up. The risk of death was higher among the 23 patients with obstructive sleep apnea than controls (adjusted hazard ratio, 1.76; 95% confidence interval, 1.05-2.95; P = .03), independent of age, sex, body mass index, smoking, hypertension, diabetes mellitus, atrial fibrillation, Mini-Mental State Examination score, and Barthel index of activities of daily living. There was no difference in mortality between the 28 patients with central sleep apnea and controls (adjusted hazard ratio, 1.07; 95% confidence interval, 0.65-1.76; P = .80).
Patients with stroke and obstructive sleep apnea have an increased risk of early death. Central sleep apnea was not related to early death among the present patients.
Stroke is a serious and common disorder and a major cause of death worldwide. Risk factors for stroke include arterial hypertension, atrial fibrillation, diabetes mellitus, smoking, and overweight. The secondary prevention of stroke includes smoking cessation, antihypertensive medication, lipid-lowering drugs, and antithrombotic medications.
Sleep apnea in the form of obstructive apnea and central apnea occurs frequently among patients with stroke.1- 11 Obstructive sleep apnea is a treatable disease characterized by recurrent episodes of upper airway obstruction during sleep and daytime sleepiness.12 Prospective studies13- 16 have shown that obstructive sleep apnea is associated with an increased risk for stroke and early death. Central apneas differ from obstructive apneas in that no efforts are made to breathe during the apnea. Most patients with central sleep apnea have Cheyne-Stokes respiration, characterized by a regular waxing and waning breathing pattern, followed by a central apnea.12
Three previous studies7,8,10 investigated the impact of sleep apnea in patients with stroke and suggested that it was a negative prognostic factor, but none of them reported that a diagnosis of sleep apnea was independently associated with mortality. These studies, however, did not differentiate between obstructive and central apneas. We sought to investigate whether obstructive or central sleep apnea was related to reduced long-term survival among patients with stroke.
All patients admitted to the geriatric stroke rehabilitation unit at Umeå University Hospital from April 1, 1995, to May 1, 1997, were eligible for inclusion. This unit is the only facility for the long-term in-hospital rehabilitation of patients with stroke in the catchment area of Umeå, with 140 000 inhabitants, and all patients in need of such rehabilitation in the area were asked to participate. Subjects who died shortly after stroke onset were not eligible because they had to survive for about 3 weeks to be in the rehabilitation facility as a criterion for inclusion. The patients who were included underwent overnight cardiorespiratory sleep apnea recordings, electrocardiography, blood pressure measurements, serum glucose measurement, height and weight measurements, and assessments using the Mini-Mental State Examination and the Barthel index of activities of daily living. All patient assessments were completed before the cardiorespiratory recordings, and the scorer (C.S.) was blinded to the patient's condition. The vital status until August 1, 2006, and the dates of death were obtained from the Causes of Death Register at the Swedish National Board of Health and Welfare. Approval for the study was obtained from the Medical Ethics Committee at Umeå University, and all patients gave their informed consent.
Stroke was defined according to World Health Organization recommendations as an acute neurological dysfunction of vascular origin with the rapid occurrence of symptoms and signs corresponding to the involvement of focal areas in the brain for more than 24 hours, with no apparent cause of other vascular origin according to a computed tomographic scan of the brain.
Overnight cardiorespiratory sleep apnea recordings were made in the stroke rehabilitation unit at a mean (SD) of 23 (8) days after the onset of stroke and sampled using computer software (Micro Digitrapper SAS; Synectics AB, Stockholm, Sweden). The nurses at the ward checked the patients and the recordings regularly during the night. They included nasal and oral air flow (3-port thermistor), thoracic and abdominal respiratory movements (resp-EZ; EPM Systems, Midlothian, Virginia), respiratory and body movements using a pressure-sensitive bed (PVDF Motion sensor; Duorec Ltd, Turku, Finland), heart rate and oxygen saturation with a sampling rate of 2 Hz (Nonin Medical Inc, Plymouth, Minnesota), body position, and a microphone placed on the throat. All recordings were scored manually, and the duration of sleep was estimated from the recordings by an experienced specialist in polysomnographic scoring (C.S.). Obstructive apnea was defined as a cessation of air flow for at least 10 seconds and hypopnea as a 50% reduction in the thermistor tracing compared with baseline, in combination with an oxygen desaturation of 3% or more and persistent respiratory movements.12 A central apnea was scored at the cessation of thoracoabdominal movements, while central hypopneas were scored when respiratory movements decreased in parallel and chronological to air flow. The apnea-hypopnea index was defined as the mean number of apneas and hypopneas per hour of estimated sleep. Obstructive sleep apnea was defined when the obstructive apnea-hypopnea index was 15 or greater, and central sleep apnea was defined when the central apnea-hypopnea index was 15 or greater.12,17 Patients with an obstructive sleep apnea index of less than 15 and a central apnea-hypopnea index of less than 15 served as control subjects.
Mini-Mental State Examination scores range from 0 to 30 points, and a low score indicates cognitive impairment. The Barthel index of activities of daily living ranges from 0 to 20, and a low score corresponds to a dependency on different activities of daily living.
Hypertension was defined as a resting blood pressure of 140/90 mm Hg or higher or a prior diagnosis of hypertension. Body mass index was defined as weight in kilograms divided by height in meters squared.
Baseline data are presented as means (SDs) or as proportions. The Wilcoxon signed rank test was used to compare the apnea-hypopnea index in different body positions. Cox proportional hazards regression was used to analyze the impact of central and obstructive sleep apnea on the time to death from the onset of stroke, with adjustments for age, sex, body mass index, current smoking, hypertension, diabetes mellitus, atrial fibrillation, Mini-Mental State Examination score, and Barthel index of activities of daily living. Survival curves were calculated from the Kaplan-Meier method and the Cox proportional hazards regression model. Hazard ratios for continuous values were based on a value close to the standard deviation. Hazard ratios were considered as significant when the 95% confidence interval did not include 1, corresponding to P < .05. All statistical calculations were made using a commercially available software program (SPSS, version 14.0; SPSS Inc, Chicago, Illinois).
One hundred fifty-one consecutive patients were admitted to the stroke rehabilitation unit during the inclusion period, and they were all asked to participate in the present study. Thirteen patients refused to participate, and 138 were investigated. Six patients were excluded because of failure in the sleep apnea recordings. A total of 132 patients were included in the analysis, and they were followed up prospectively for a mean (SD) of 10.0 (0.6) years after stroke onset; no one was lost to follow-up.
A hemorrhagic stroke was recorded in 15 patients (11.4%) at study inclusion, and 49 patients (37.1%) had experienced 2 or more strokes. Twenty-three patients (17.4%) had obstructive sleep apnea, and 28 (21.2%) had central sleep apnea during Cheyne-Stokes respiration. Two patients had both central and obstructive sleep apnea, and they were excluded from further analysis. Seventy-nine patients had central and obstructive apnea-hypopnea indexes of less than 15, and they served as controls. Baseline data for the included patients are given in Table 1. Patients with obstructive sleep apnea had a mean obstructive apnea-hypopnea index of 28 (14) and a mean central apnea-hypopnea index of 2 (4). The mean supine apnea-hypopnea index was 35 (17), and the mean nonsupine apnea-hypopnea index was 9 (18) (P < .001 for the difference). Patients with central sleep apnea had a mean central apnea-hypopnea index of 33 (12) and a mean obstructive apnea-hypopnea index of 3 (3). The mean supine apnea-hypopnea index was 40 (13), and the mean nonsupine apnea-hypopnea index was 18 (20) (P < .001 for the difference).
At follow-up, 116 of 132 patients (87.9%) had died. The causes of death were regarded as cardiovascular in 74.0%, cancer in 9.9%, other causes in 14.3%, and unknown in 1.8%. All the patients with obstructive sleep apnea had died at follow-up, as did 96.4% of patients with central apnea and 81.0% of control subjects. Obstructive sleep apnea, age, atrial fibrillation, a low Mini-Mental State Examination score, and a low Barthel index of activities of daily living were related to an increased risk of death from any cause in the unadjusted analysis (Table 2 and Figure 1). Obstructive sleep apnea remained as a significant risk of death from any cause in the multivariate analysis (adjusted hazard ratio, 1.76; 95% confidence interval, 1.05-2.95; P = .03), independent of age, sex, body mass index, current smoking, hypertension, diabetes mellitus, atrial fibrillation, Mini-Mental State Examination score, and Barthel index of activities of daily living (Table 2 and Figure 2). Moreover, age and a low Mini-Mental State Examination score remained as significant risks of death after adjustment for confounders. Central sleep apnea was not associated with any increased risk of death (P = .80).
Adjusted hazard ratios were also tested using cutoffs at 5 and 10 instead of 15 for the apnea-hypopnea index. An obstructive apnea-hypopnea index of greater than 10 was independently related to death (adjusted hazard ratio, 1.66; 95% confidence interval, 1.00-2.73; P = .048), but not an obstructive apnea-hypopnea index of greater than 5 (adjusted hazard ratio, 1.47; 95% confidence interval, 0.88-2.47; P = .15). Central sleep apnea was neither related to death at a central apnea-hypopnea index of greater than 10 (adjusted hazard ratio, 1.31; 95% confidence interval, 0.80-2.16; P = .29) nor at a central apnea-hypopnea index of greater than 5 (adjusted hazard ratio, 1.19; 95% confidence interval, 0.70-2.02; P = .53).
Thirty-three patients tried continuous positive airway pressure (CPAP) therapy in the hospital. However, only 10 patients continued with CPAP therapy at home for 5 to 57 months. Four of them had central sleep apnea, and 6 had obstructive sleep apnea. Four patients continued with CPAP therapy until death, while 6 had quit therapy before they died.
We prospectively followed up 132 of 151 consecutive patients in need of in-hospital stroke rehabilitation for a mean of 10.0 (0.6) years. Patients with stroke and obstructive sleep apnea with an obstructive apnea-hypopnea index of 15 or greater had a 75% increase in the risk of early death compared with patients with stroke without sleep apnea, independent of age, sex, smoking, body mass index, hypertension, diabetes mellitus, atrial fibrillation, cognition, and dependency during daily living. Central sleep apnea was not related to increased mortality. To our knowledge, these are novel findings.
Three previous studies have investigated the relationship between sleep apnea in persons who experience stroke and mortality after a mean of 0.5 to 5 years. Turkington and colleagues8 investigated 120 patients within 24 hours from the onset of stroke and reported that 37% of them were dead 6 months later. Parra and colleagues7 investigated 161 patients within 2 to 3 days from the onset of stroke or transient ischemic attack and observed that 14% had died after a mean follow-up of 23 months. These 2 studies report a lower apnea-hypopnea index in survivors and an increased mortality in patients with sleep apnea in a univariate analysis. Bassetti and colleagues10 investigated 152 patients 3 days after stroke onset and observed 132 of them during a mean of 60 months. Of these patients, 14% were dead at follow-up. The apnea-hypopnea index was lower in survivors, but not independent of confounders. The studies described support our results, but none of them found that a diagnosis of sleep apnea was independently related to mortality and none of them differentiated between obstructive apneas and central apneas in their analysis. Rowat and colleagues,11 in a recent study, investigated the impact of central periodic breathing during daytime in the awake state on mortality in the immediate phase after a stroke. They investigated 138 patients with stroke at a median of 4 hours after stroke and found that 24% of the patients had central periodic breathing. Patients with central periodic breathing were more likely to have severe stroke, and more of them died within the first week. These patients were also more likely to be dead or dependent at 3 months. As different from the previous studies, we investigated our patients 3 weeks after the onset of stroke. Some of the immediate effects of stroke on mortality and possibly also on apneas secondary to acute stroke were, therefore, avoided. We followed up our patients for a much longer time, until most patients had died, and divided our patients into those with obstructive sleep apnea and those with central sleep apnea. These differences in design could explain the different results and that we observed that a diagnosis of obstructive sleep apnea in patients with stroke was independently related to increased mortality.
Different cutoff points in the apnea-hypopnea index have been used to define sleep apnea, usually 5, 10, and 15. A cutoff of 5 is suggested to define mild sleep apnea and a cutoff of 15 is suggested to define moderate sleep apnea in subjects seeking medical attention for snoring and daytime sleepiness.12 The frequency of subjects in population-based studies with an apnea-hypopnea index of greater than 5 is, however, high and many of them do not have daytime sleepiness. The apnea-hypopnea index increases with age, but there is no consensus on the cutoff in older subjects. We predefined sleep apnea at a cutoff of 15 because this was used in a previous CPAP trial on patients with stroke.17 Our results do, however, support an increased mortality among patients with stroke, even at an obstructive apnea-hypopnea index of greater than 10.
Possible mechanisms causing increased mortality among patients who experience stroke and who have sleep apnea include nocturnal hypoxemia and an increase in the risk of sudden cardiovascular death during sleep.18 Cerebral blood flow velocity increases concomitant with arterial pressure during obstructive apnea, with a maximum 5 seconds after apnea termination.19 Thereafter, the cerebral blood flow velocity and arterial pressure decrease rapidly to a minimum 20 seconds after apnea termination when oxygen saturation is low. The changes in cerebral blood flow velocity may be an immediate result of changes in blood pressure and cerebral autoregulation may be insufficient to protect the brain from the rapid potentially harmful arterial pressure changes that occur during obstructive apnea.19 The opposite is seen during central apnea, with a decrease in cerebral blood flow during apnea and an increase after apnea termination.20 These researchers19,20 suggest that obstructive apneas are followed by cerebral ischemia, which is not the case after central apneas. The present results support that obstructive and central apneas affect patients differently and that obstructive apneas are harmful for the brain.
Limitations of the study include the use of oronasal thermistors to detect airflow instead of nasal pressure cannulas and the use of piezoelectric belts instead of esophageal pressure monitors to differentiate central from obstructive apneas. Nasal pressure cannulas were, however, not available when the investigations were done, and esophageal pressure monitors are still seldom used. Only patients in need of in-hospital stroke rehabilitation were eligible, which excludes patients with minor stroke and transient ischemic attack. The study is limited by the sample size to rule out that central apnea does not affect survival in patients with stroke.
Nasal CPAP during sleep is the primary treatment for reducing obstructive apneas and excessive daytime sleepiness. We identified 6 studies10,17,21- 24 that tested CPAP in patients with stroke, and they all report problems tolerating CPAP. One randomized controlled trial17 reported a reduction in depressive symptoms after 1 month of CPAP, but compliance was low among patients with delirium and those with severe cognitive impairment. Another study22 reported that CPAP-treated patients had a smaller risk of new vascular events. A recent randomized controlled trial24 reports that, despite intensive efforts, objective CPAP use was poor. Tolerance of CPAP was also low among the present patients. From the present study and the previously described studies, it seems reasonable to suggest that sleep apnea recordings should be performed in patients with stroke and that CPAP should be offered to those with obstructive sleep apnea, because there is no harm in trying this treatment; however, most patients with stroke will not comply with this treatment. An individually fabricated mandibular advancement appliance is another treatment for obstructive sleep apnea that has not yet been tested in patients with stroke. It is possible that the avoidance of the supine position is another option in patients with stroke. The frequency of obstructive and central sleep apneas is more prevalent in the supine position compared with other positions.25 The supine dependency was also observed among the present patients with stroke.
In conclusion, patients with stroke and obstructive sleep apnea run an increased risk of early death. Central sleep apnea was not related to early death among the present patients.
Correspondence: Karl A. Franklin, MD, PhD, Department of Respiratory Medicine, University Hospital, SE-901 85 Umeå, Sweden (Karl.Franklin@Lung.Umu.Se).
Accepted for Publication: September 2, 2007.
Author Contributions: Dr Franklin had full access to all the data in the study and had final responsibility for the decision to submit for publication. Study concept and design: Sahlin, Sandberg, Gustafson, Bucht, and Franklin. Acquisition of data: Sahlin, Sandberg, Gustafson, and Franklin. Analysis and interpretation of data: Sahlin, Bucht, Carlberg, Stenlund, and Franklin. Drafting of the manuscript: Sahlin and Franklin. Critical revision of the manuscript for important intellectual content: Sandberg, Gustafson, Bucht, Carlberg, and Stenlund. Statistical analysis: Stenlund. Obtained funding: Sandberg, Gustafson, Bucht, and Franklin. Administrative, technical, and material support: Sahlin, Sandberg, Gustafson, Bucht, and Franklin. Study supervision: Gustafson and Franklin.
Financial Disclosure: None reported.
Funding/Support: This study was supported by grants from the Swedish Heart and Lung Foundation; the Swedish Society of Medicine; the Swedish Stroke Foundation; the Foundation for Medical Research at Umeå University; the Gun and Bertil Stohne Foundation; the Thureús Foundation; and the Loo and Hans Osterman Foundation.
Role of the Sponsor: The funding bodies had no role in the design and conduct of the study; in the collection, analysis, and interpretation of the data; or in the preparation, review, or approval of the manuscript.