Abstracts

Rationale: Mounting evidence suggests the need for early diagnosis of obstructive sleep apnea (OSA), affecting over one-third of adults with epilepsy and largely under-diagnosed. This implies unaddressed risk factors for a host of medical conditions including the seizure activating effect OSA has in some patients. Previously, the accuracy of OSA detection using the Sleep Apnea Scale of the Sleep Disorders Questionnaire (SA-SDQ) in adults with epilepsy was assessed (Weatherwax et al, 2003), proposing cut-offs of 29 and 26 for males and females, respectively, as optimal in the prediction of OSA based on an apnea-hypopnea index (AHI) >5 on polysomnography (PSG). This analysis classified OSA using different severity classes than those currently used in clinical practice. The purpose of this study was to assess the validity of the SA-SDQ as a screening tool for OSA in adults with epilepsy using standard OSA severity criteria.Methods: As part of a larger study exploring sleep disorders in adults with epilepsy, 90 consecutive subjects unselected for sleep disorder symptoms completed the SA-SDQ and underwent PSG with expanded EEG. The SA-SDQ is a validated self-administered 12-item questionnaire for OSA based on traditional predictors including symptoms, age, body mass index, tobacco use, and hypertension history. Scores range from 12-60 (higher more suggestive of OSA). OSA was defined as an AHI 5 with mild=5-<15, moderate=15-<30, and severe= 30. Descriptive analysis and T-tests were performed and receiver-operating characteristics curves (ROC) by gender were formed to provide sensitivity and specificity values for SA-SDQ cutoffs best predictive of OSA. SA-SDQ internal consistency was assessed with the Cronbach s procedure which detected the reliability coefficients of individual items.Results: 50 (56%) subjects had no OSA and 40 (44%) had an AHI 5. Cronbach s for the 12 items was 0.724. Seven of 12 scale items correlated significantly with AHI. The overall area under the curve for the diagnosis of OSA was 0.771 overall, 0.926 for males and 0.687 for females. For all subjects, a cut-off score of 25 provided high sensitivity (73%) and specificity (72%) for OSA. A score of 25 provided optimal sensitivity (94%) and specificity (83%) for males, but lower sensitivity (55%) and specificity (68%) for females. In females, a cut-off score of 24 improved sensitivity (68%) and specificity (58%). For moderate-to-severe OSA, the area under the curve was 0.766 overall, 0.861 for males and 0.722 for females, with an optimal cut-off of 28.Conclusions: The reliability of the SA-SDQ and its individual items as a screening tool for OSA in adults with epilepsy is high, consistent with previous findings. Our findings support the use of SA-SDQ cut-off scores of 25 for males and 24 for females for predicting any degree of OSA and 29 for males and 26 for females for predicting moderate-to-severe disease in epilepsy populations.