Comparison of Screening Instruments for Obstructive Sleep Apnea in Adults with Epilepsy
Maeve Pascoe1, Noah Andrews1, James Bena2, Irene Katzan3, Nancy Foldvary-Schaefer1
1Sleep Disorders Center, 2Quantitative Health Sciences, 3Neurology, Cleveland Clinic
We aimed to compare performance characteristics of obstructive sleep apnea (OSA) screens to predict disease severity in adults with epilepsy (AWE).
OSA is highly prevalent in AWE, yet screening instruments have not been adequately studied.
Scores for STOP, STOPBANG, STOPBAG, and STOPBAG2, a modified instrument by the Cleveland Clinic, were calculated for a sample of AWE who underwent PSG. STOP was calculated from positive responses to the following items (+1 point each): presence of Snoring, Tiredness/sleepiness/fatigue, Observed apneas, high blood Pressure. STOPBAG and STOPBANG were calculated using those items with addition of: Body mass index ≥30kg/m2, Age ≥50yrs, Neck circumference ≥40cm (15.75in) and Male gender. STOPBAG2 was calculated using STOP and male gender and continuous variables for age and BMI. For all instruments, greater scores indicate higher OSA risk.  Apnea-hypopnea index (AHI) defined disease severity (mild 5≤15; moderate 15≤30; severe >30). Logistic regression and receiver operating characteristic (ROC) analyses evaluated optimal cutoff values and overall discriminatory power for each instrument.
In 133 patients (age 41.9±13.7, 53.6% female, AHI 9.2[2.4,21.7], (32% <5, 32% 5≤15, 19% 15≤30, 17% ≥30), all screening instruments were highly correlated with OSA severity (p<0.001 for all). STOPBAG2 provided best sensitivity for predicting each OSA severity level (mild: 0.86, moderate: 0.79, severe: 0.82) and best specificity for mild OSA (0.86), with comparable specificity for moderate and severe OSA to other instruments. STOPBAG2 had greatest ROC discriminatory power of all instruments for each level of OSA severity (mild: 0.90, moderate: 0.78, severe: 0.80).
While all screening instruments were adequate for predicting OSA severity in AWE, the STOPBAG2 has the greatest discriminatory power for predicting all levels of severity, supporting its implementation in epilepsy clinics.