Effect of Obstructive Sleep Apnea on Frequency of Stroke in Patients With Atrial Fibrillation

Obstructive sleep apnea (OSA) is an independent risk factor for ischemic stroke that is not included in the usual cardioembolic risk assessments for patients with atrial fibrillation (AF). The aim of this study was to investigate the impact of OSA on stroke rate in patients with AF. Patients with AF and new diagnoses of OSA were identified from retrospective chart review. Those with histories of stroke at the time of the sleep study were excluded. The primary outcome was the incidence of stroke, determined by a physician investigator blinded to the results of polysomnography. Subgroup analysis was performed among different CHADS 2 and CHA 2 DS 2 -VASc scores. Of 5,138 patients screened for OSA, 402 (7.7%) had AF and 332 (6.4%) met the inclusion criteria. Among the study population, the occurrence of first-time stroke was 22.9%. Ischemic stroke was more common in patients with OSA compared with patients without (25.4% vs 8.2% respectively, p = 0.006). After controlling for age, male gender, and coronary artery disease, the association between OSA and stroke remained statistically significant, with an adjusted odds ratio of 3.65 (95% confidence interval 1.252 to 10.623). A positive dose effect of the apnea-hypopnea index on the rate of stroke was observed (p = 0.0045). Subgroup analysis showed significantly higher rates of stroke in patients with CHADS 2 scores of 0 and CHA 2 DS 2 -VASc scores of 0 and 1 and co-morbid OSA. In conclusion, OSA in patients with AF is an independent predictor of stroke. This association may have important clinical implications in ischemic stroke risk stratification.

Atrial fibrillation (AF) is an independent risk factor for stroke. Approximately 25% of patients with AF has documented obstructive sleep apnea (OSA). Multiple studies indicate a strong association between ischemic stroke and OSA. The role OSA plays with regard to stroke in patients with AF has not been well delineated. Risk assessment tools are used to estimate the risk for stroke and the need for antithrombotic therapy for patients with AF. This traditionally has been accomplished by the application of the CHADS 2 risk score. Modifications of this scoring system have resulted in the CHA 2 DS 2 -VASc score, which is currently the only assessment tool recommended in the 2014 American Heart Association, American College of Cardiology, and Heart Rhythm Society guideline for the management of patients with AF. Because these traditional cardioembolic risk stratification systems do not account for OSA as a risk factor, we sought to study stroke risk in patients with AF while accounting for the presence of OSA. The primary aim of this study was to evaluate the stroke risk attributed to OSA in patients with AF.


We conducted a retrospective cohort study of patients referred to the Western Connecticut Health Network Department of Pulmonary, Critical Care, and Sleep Medicine for polysomnography from January 1, 2008, to December 31, 2011. The study sample was limited to patients ≥18 years of age who underwent their first overnight polysomnographic studies. Of all patients who underwent sleep studies, those with previous diagnoses of AF were identified. The exposure group consisted of subjects with apnea-hypopnea index (AHI) values ≥5 and higher; patients with AHI values <5 constituted the comparison group. The follow-up period was defined as the time from index sleep study to the chart review and on average was 4.4 years.

Patients were excluded if they had histories of ischemic stroke or transient ischemic attack at the time of the sleep study. Those patients who underwent split studies without diagnostic polysomnography were eliminated from the study sample. Patients with diagnoses of central sleep apnea were also excluded from the study.

Data collected from the time of polysomnography included age, gender, body mass index, and smoking status. In addition, stroke risk-factor data including diabetes mellitus, congestive heart failure, hypertension, and hyperlipidemia were collected. CHADS 2 and CHA 2 DS 2 -VASc scores were calculated for each patient at the time of polysomnography. Data regarding prophylactic anticoagulation was also collected. For those patients who met the primary end point, anticoagulation status was recorded as the status before the event. The physician investigator blinded to the OSA status of the patient assessed prevalence of first-time ischemic stroke by reviewing electronic medical records. Diagnosis was determined according to criteria of the National Institute of Neurological Disorders and Stroke for the classification of cerebrovascular events. The diagnosis of stroke had to have been established by a neurologist. The primary outcome was the incidence of ischemic stroke or transient ischemic attack.

Continuous positive airway pressure (CPAP) use and compliance data were obtained from the outpatient chart review. Patients were deemed compliant if they used CPAP for >4 hours per night, ≥70% of the nights in a 30-day period (data obtained from data cards or wireless transmission). Patients were evaluated in the pulmonary and sleep department for 3 consecutive visits after CPAP setup.

The diagnosis of coronary artery disease was made if the medical record noted a history of myocardial infarction, angina, or coronary angiography revealing the presence of disease, percutaneous coronary intervention, or coronary artery bypass graft surgery.

Heart failure was assigned if the medical record documented the presence of the entity or if any cardiac imaging revealed a left ventricular ejection fraction <50%. Diabetes mellitus was recorded if the medical record had noted the presence of type I or II diabetes mellitus or the use of hypoglycemic medications. Hypertension and hyperlipidemia were recorded if diagnoses were noted in medical records.

The diagnosis of AF was determined through review of electronic medical records, including all hospital admissions, outpatient electrocardiograms, discharge and consult notes, Holter monitor reports, and hospital telemonitor strips.

All polysomnographic studies were conducted at the Western Connecticut Health Network sleep laboratory, and all patients were evaluated by a board-certified sleep physician. The diagnosis of OSA was defined as an AHI ≥5. OSA status was further classified as mild (AHI 5 to 15), moderate (AHI >15 to 30), or severe (AHI >30), in accordance with the American Academy of Sleep Medicine criteria. This study was approved by the Western Connecticut Health Network Institutional Review Board.

Baseline characteristics are presented as means and standard deviations. An analysis of variance was used to compare mean values. Categorical data were compared with the use of chi-square tests. Multiple logistic regression analysis was performed with covariates that were statistically significantly associated with the outcome on univariate analysis. Statistical significance was set at p <0.05. Subgroup analysis of patients on the basis of their CHADS 2 and CHA 2 DS 2 -VASc scores was performed. For every subgroup, univariate analysis was performed with stroke as the dependent variable and OSA as the covariate. Odds ratios were regarded as significant if the 95% confidence interval (CI) did not cross 1, corresponding to a p value <0.05.

All statistical analysis was performed using JMP version 9.0 (SAS Institute Inc., Cary, North Carolina).


Of 5,138 patients who underwent sleep studies for any reason at the Western Connecticut Health Network sleep laboratory during the study period, 402 (7.7%) had previous AF and were eligible for inclusion in the study ( Figure 1 ). After applying the exclusion criteria, 332 patients (6.4%) were included in the study.

Figure 1

Design of the study.

OSA was identified in 283 patients (85.2%). The percentage of men was higher among patients with OSA compared with those without (73.1% and 59.2%, respectively, p = 0.047), as was the percentage of patients with coronary artery disease (35.3% vs 20.4%, respectively, p = 0.048). No other recorded clinical characteristics differed between the 2 groups. Mean CHADS 2 score did not differ significantly between groups (1.45 ± 1.24 among non-OSA patients and 1.6 ± 1.06 among patients with OSA), nor did mean CHA 2 DS 2 -VASc score (2.65 ± 1.62 among non-OSA patients and 2.87 ± 1.91 among patients with OSA). All demographic and relevant clinical characteristics are listed in Table 1 .

Table 1

Demographic and clinical characteristics of the patients at baseline

Variable Obstructive Sleep Apnea P value
No (N=49) Yes (N=283)
Age (years) 67.10 (±12.1) 69.06 (±12.1) 0.29
Body-mass index (kg/m 2 ) 32.26 (±8.0) 33.2 (±7.2) 0.41
Men 29 (59.2%) 207 (73.1%) 0.04
Hypertension 31 (63.3%) 188 (66.4%) 0.74
Hyperlipidemia 37 (76%) 220(78%) 0.81
Diabetes mellitus 17 (34.7%) 89 (31.4%) 0.74
Congestive heart failure 10 (20.4%) 76 (26.9%) 0.38
Coronary artery disease (History of myocardial infarction, angina, percutaneous coronary intervention, or coronary artery bypass graft surgery) 10 (20.4%) 100 (35.3%) 0.04
Apnea hypopnea index 2.60 (±1.4) 36.18 (±29.3)
Percent of time spent with O2 saturation <90% 22.17 (±27.0)
Minimal oxygen saturation during polysomnography 79.27 (±9.6)
CHADS 2 score 1.45 ± 1.2 1.6 ±1.1 0.73
CHA 2 DS 2 -VASc score 2.65 ± 1.6 2.87 ±1.9 0.88

Plus – minus values are means ± SD.

The occurrence of first-time stroke in the study population was 22.9%. The prevalence of stroke was more common in patients with OSA compared with non-OSA patients (25.4% and 8.2%, respectively, p = 0.006). The odds ratio for first-time stroke in AF patients with OSA compared with those without OSA was 3.84 (95% CI 1.334 to 11.047). On multiple logistic regression analysis, the association between OSA and stroke remained statistically significant after controlling for parameters that were significantly associated with outcome on univariate analysis ( Table 2 ), with an adjusted odds ratio of 3.65 (95% CI 1.252 to 10.623).

Table 2

Univariate analysis

Variable Odds ratio Confidence interval P value
Obstructive Sleep Apnea 3.839 1.33-11.05 0.008
Body-mass index >30 1.01 1.93-3.47 0.93
Smoking 0.975 0.59-1.63 0.92
Hypertension 1.354 0.77-2.36 0.33
Hyperlipidemia 0.75 0.32-1.75 0.5
Coronary artery disease 1.33 1.79-2.27 0.05
Diabetes mellitus 1.03 1.33-2.28 0.06
Congestive heart failure 1.33 1.33-2.34 0.37
Age >75 2.45 2.45-4.16 0.001

A dose effect relation was observed between AHI and the incidence of stroke (p = 0.0045). Additionally, an incremental increase in the incidence of stroke in OSA patients was observed across all CHADS 2 and CHA 2 DS 2 -VASc groups. Subgroup analysis was performed after stratification of OSA and non-OSA groups on the basis of CHADS 2 and CHA 2 DS 2 -VASc scores. Among patients with CHADS 2 scores of 0 and 1, significantly increased incidence of stroke was detected in patients with OSA compared with those without ( Table 3 ). Among patients with CHA 2 DS 2 -VASc scores of 0, the odds ratio of having a stroke in patients with OSA was 1.62 (95% CI 1.155 to 2.259) ( Table 4 ).

Nov 30, 2016 | Posted by in CARDIOLOGY | Comments Off on Effect of Obstructive Sleep Apnea on Frequency of Stroke in Patients With Atrial Fibrillation

Full access? Get Clinical Tree

Get Clinical Tree app for offline access