Abstracts

To identify characteristics predictive of obstructive sleep apnea (OSA) on polysomnography (PSG) in medically refractory epilepsy patients who have already completed a screening questionnaire and clinical interview suggesting OSA. These patients were participating in a pilot clinical trial of the effects of treating sleep-disordered breathing on seizure frequency. Minimizing false positives who appear to have symptoms of OSA by survey with confirmatory clinical history, but not by PSG, will contribute to the success of a phase 3 trial. Thirty adults with four or more seizures a month and a score of 26 for women or 29 for men on the Sleep Apnea Scale of the Sleep Disorders Questionnaire (SA-SDQ) were invited to a screening interview with one of the authors. The SA-SDQ is a validated instrument for predicting OSA which has been used in the general population and in those with epilepsy. If the suspicion of OSA was confirmed by clinical interview and other criteria were met, patients were enrolled in a prospective study in which they underwent two nights of PSG to confirm OSA. OSA was defined by an apnea-hypopnea index (AHI) of 10 events/hour on either PSG night. Eighteen of 30 patients (60%) had OSA on PSG. The only two variables that predicted OSA on PSG were age at PSG and gender. Subjects with OSA were older (45.5 [plusmn] 11.8 vs. 35.8 [plusmn] 11.8, mean [plusmn] standard deviation; t= 2.3; p = 0.03;two-tailed student t-test) and more likely to be male (pearson chi-square 3.8, p = 0.05). Epworth sleepiness scale, body mass index (BMI), number of seizures/month, habitual loud snoring or witnessed apneas, number or type of antiepileptic drugs, hypertension, SA-SDQ score, or nocturnal seizures were not helpful in predicting which patients suspected of having OSA actually had PSG-documented OSA. Age and gender were useful variables in predicting which patients with suspected OSA had PSG-documented OSA. Variables useful for predicting OSA in the general population (daytime sleepiness, hypertension, BMI) and those unique to the epilepsy patient (seizure frequency, antiepileptic drugs, and nocturnal seizures), were not informative, although the small sample size may bias our results. Identification of OSA in epilepsy patients is challenging. Questionnaires, while useful, are not foolproof in detection of OSA. Clinical trials may need to plan for a proportion of patients testing as false positives for OSA. Using PSG in epilepsy monitoring units to screen patients for OSA may also be an effective strategy to explore. (Supported by NINDS R01 NS 042698 (BAM), NINDS K02 NS2099 (BAM), and GCRC grants RR000095 (Vanderbilt), RR00042 (Michigan), and RR00046 (University of North Carolina).)