Sleep apnea screening instrument evaluation and novel model development and validation in the paroxysmal atrial fibrillation population

Standard sleep apnea (SA) screening instruments perform suboptimally in the atrial fibrillation (AF) population. We evaluated and optimized common OSA screening tools in the AF population. Participants of the Sleep Apnea and Atrial Fibrillation Biomarkers and Electrophysiologic Atrial Triggers (SAFEBEAT, NCT02576587) age (±5 years)-, sex-, body mass index (BMI ± 5 kg/m2)-matched case control study (n = 150 each group) completed concurrent questionnaires and overnight polysomnography. Models based on STOP, STOP-BANG, Berlin, NoSAS and Epworth Sleepiness Scale and also models with STOP-BANG predictors with resting heart rate or left atrial volume were constructed. “Best subset” analysis was used to select a predictor subset for evaluation. We assessed test performance for two outcome thresholds: apnea-hypopnea index (AHI) ≥ 5 and AHI ≥ 15. Paroxysmal AF participants were: 61.3 ± 12.1 years, BMI = 31.2 ± 6.6 kg/m2 with median AHI = 11.8(IQR: 3.8, 24.5); 65 (43.3%) with AHI ≥ 15. Only STOP and STOP-BANG did not perform worse in AF relative to controls. For AHI ≥ 15, STOP-BANG (AUC 0.71, 95%CI:0.55–0.85) did not perform as well as NABS – a composite of neck circumference, age, and BMI as continuous variables and snoring (AUC 0.88, 95%CI:0.76–0.96). Optimal model for AHI ≥ 15 was NABS (sensitivity = 45%, specificity = 97%). For AHI ≥ 5, NABS was also the best performing (AUC 0.82, 95%CI:0.68–0.92, sensitivity = 78%, specificity = 67%). We identify a novel, short-item SA screening instrument for use in paroxysmal AF, i.e. NABS, with improved discriminative ability compared to commonly-used instruments. Further validation studies are needed to assess utility in other AF subtypes. Trial registration: clinicaltrials.gov NCT02576587.