Abstracts

Rationale: Encephalopathy with electrical status epilepticus in slow wave sleep (ESES) is a severe epileptic encephalopathy with significant activation of epileptiform activity on EEG in sleep. EEG is used to guide diagnosis, course and management. In a paper publishing guidelines on recording EEG in ESES- an activation of at least 50% was considered adequate for the sake of uniformity for comparison in and between centres.1 This would also be an important step in moving forward with multicenter research on ESES. Many centers have included a Spike wave index (SWI) of > 25% in sleep to be considered as a defining criterion for ESES. We report our experience with Persyst 13(P13) and compare the results of clinician expert(CE) to P13 in measuring spike counts (SC) and SWI. Methods: All routine and overnight EEGs between 2009 and 2015 on patients with a diagnosis ESES were identified. Initial 30 minute samples were chosen by their proximity to each other- the last 30 minutes of awake time prior to sleep onset and the subsequent first 30 minutes of uninterrupted sleep." Other samples were labeled "Random".. For patients who did not have 30 minute samples, two 15 minute samples were identified and then averaged. Of these 5 minutes samples were clipped and used for visual and P13 analysis of SC and SWI. A total of 40 such clips were analyzed. Two epileptologists (CJ and MC) independently obtained SC and SWI. These were averaged and compared to the same obtained by P13. SC were counted in P13 from peak to peak and all spikes > 200 msecond apart were marked. SWI was calculated as number of 1 second bins containing at least 1 spike expressed as a percentage over time. The SWI tool in P13 was configured to calculate the percentage of 1-second epochs that had one or more spike detections per second over the length of each 5 minute recording. A Spearman correlate was used to ascertain agreement for SC and SWI between CE. Shrout reliability index, a measure of intra-class correlation, and Bland Altman plots were used to compare SC and SWI as measured by CE and P13. Results: A Spearman correlate of 0.931 was obtained for SC and 0.948 for SWI respectively between the two readers. SC readings correlated closely with CE and P13 with a Shrout Fleiss reliability index for SC of 0. 79 (95% CI 0.63-0.88) and 0.91 for SWI with 95% CI (0.84-0.95). Variability was found between CE and P13 as expected on the Bland Altman plot but suggesting that the difference in the means of SC and SWI by CE and P13 may be clinically acceptable: 28.93 for SC (95% CI (-3.3362, 61.2862)- p=0.07) and 0.5965 for SWI ( 95%CI -2.4183, 3.6113;p=0.7) Conclusions: There is good agreement between CE and P13 for SC and clinically acceptable variability for SWI- if one considers the average SWI of > 30% to be clinically significant ( as the lower limit of the P13 mean is -2.42). These results need to be reproduced in larger numbers and many more centers to assess for validity especially to be able to use P13 for ease of multicenter research Funding: none