Prevalence and Assessment of Sleep-Disordered Breathing in Patients With Atrial Fibrillation: A Systematic Review and Meta-analysis

In this study, we sought to estimate the prevalence of concomitant sleep-disordered breathing (SDB) in patients with atrial fibrillation (AF) and to systematically evaluate how SDB is assessed in this population. We searched Medline, Embase and Cinahl databases through August 2020 for studies reporting on SDB in a minimum 100 patients with AF. For quantitative analysis, studies were required to have systematically assessed for SDB in consecutive AF patients. Pooled prevalence estimates were calculated with the use of the random effects model. Weighted mean differences and odds ratios were calculated when possible to assess the strength of association between baseline characteristics and SDB. The search yielded 2758 records, of which 33 studies (n = 23,894 patients) met the inclusion criteria for qualitative synthesis and 13 studies (n = 2660 patients) met the meta-analysis criteria. The pooled SDB prevalence based on an SDB diagnosis cutoff of apnea-hypopnea index (AHI) ≥ 5/h was 78% (95% confidence interval [CI] 70%-86%; P < 0.001). For moderate-to-severe SDB (AHI ≥ 15/h), the pooled SDB prevalence was 40% (95% CI 32%-48%; P < 0.001). High degrees of heterogeneity were observed (I2 = 96% and 94%, respectively; P < 0.001). Sleep testing with the use of poly(somno)graphy or oximetry was the most common assessment tool used (in 22 studies, 66%) but inconsistent diagnostic thresholds were used. SDB is highly prevalent in patients with AF. Wide variation exists in the diagnostic tools and thresholds used to detect concomitant SDB in AF. Prospective systematic testing for SDB in unselected cohorts of AF patients may be required to define the true prevalence of SDB in this population. Estimer la prévalence des troubles respiratoires du sommeil (TRS) concomitants à la fibrillation auriculaire (FA) et examiner systématiquement le mode d’évaluation des TRS dans ce contexte. Nous avons effectué des recherches dans les bases de données MEDLINE, EMBASE et CINAHL en ciblant une période allant jusqu’à août 2020, afin de trouver des études signalant des TRS survenus chez des patients atteints de FA, en fonction d’un nombre minimum de patients fixé à 100. Aux fins d’analyse quantitative, les études devaient comporter une évaluation systématique des TRS chez des patients consécutifs atteints de FA. L’estimation groupée de la prévalence a été effectuée à l’aide du modèle à effets aléatoires. Les différences moyennes pondérées et le rapport de cotes ont été calculés, lorsque