Risk Of Atrial Fibrillation And Stroke After Bariatric Surgery In Patients With Morbid Obesity With Or Without Obstructive Sleep Apnea

Background and Purpose: Weight loss in morbidly obese patients reduces atrial fibrillation (AF); however, it is unknown whether similar benefits are maintained in patients with obstructive sleep apnea (OSA). We sought to determine whether incident AF and stroke rates are affected by OSA after weight loss and to identify predictors of AF and stroke. Methods: Differences in laparoscopic adjustable gastric banding-induced weight loss on incident AF and stroke events in those with and without OSA in the entire and in propensity-matched cohorts were determined longitudinally, and independent predictors of AF and stroke were identified. Results: Of 827 morbidly obese patients who underwent laparoscopic adjustable gastric banding (mean age, 44 +/- 11 years; mean body mass index, 49 +/- 8 kg/m(2)), incident AF was documented in 4.96% and stroke in 5.44% of patients during a mean 6.0 +/- 3.2-year follow-up. Despite a similar reduction in body weight (19.6% and 21% in 3 years), new-onset AF was significantly higher in patients with OSA than without OSA in the entire (1.7% versus 0.5% per year; P<0.001) and propensity-matched cohorts. Incident stroke was higher in the OSA than in the non-OSA group (2.10% versus 0.47% per year; P<0.001), but only 20% of patients with stroke had documented AF. On multivariate analysis, OSA (hazard ratio, 2.88 [95% CI, 1.45-5.73]), age, and hypertension were independent predictors of new-onset AF, and OSA (hazard ratio, 5.84 [95% CI, 3.02-11.30]), depression, and body mass index were for stroke events. Conclusions: In morbidly obese patients who underwent laparoscopic adjustable gastric banding, despite similar weight loss, patients with OSA had a higher incidence of AF and stroke than patients without OSA. Both non-AF and AF-related factors were involved in increasing stroke risk. Further investigation is warranted into whether OSA treatment helps reduce AF or stroke events in this population.