Abstracts

Rationale: EEG data are collected on EPGP study subjects to confirm epilepsy diagnosis and classification, and facilitate the investigation of genetic contributions to specific EEG abnormalities and phenotypic characteristics. The challenges faced in creating a platform to collect these data include the selection of tools to view and de-identify multi-vendor format EEGs, efficient electronic transfer of EEG files, and development of web-based tools to collect accurate phenotypic data and to facilitate the EEG Core Review process. Methods: Clinical site study coordinators upload each subject s EEG to a secure FTP server, and each EEG is then examined and archived by the EPGP Data Manager and converted to Persyst (Persyst Development Corp, Prescott, AZ) format in the process. Newly developed web-based tools allow neurologists at each site to collect phenotypic data and enable the EEG Core members to review and score the EEG data collected. Following clinical site submission and review of EEG data, Initial Inclusion Review is performed by one EEG Core member, Final Review by two independent EEG Core members, and Final Consensus Review by three or four EEG Core members. Results: 755 EEGs have been uploaded to date, of which 680 are initial inclusion EEGs and 75 are supplemental EEGs. A total of 189 EEGs are awaiting evaluation by the clinical site neurologist and 563 EEGs have been, or are in the process of being, reviewed by the EEG Core members. 35 EEGs did not meet EEG inclusion criteria or were deemed technically inadequate on EEG Core Initial Inclusion Review and 14 EEGs did not meet EEG inclusion criteria based on Final Consensus Review. To date, 289 eligible EEGs have been phenotyped via Final Consensus Review (215 digital), of which 210 (73%) had modifications made to EEG phenotypic details during Core Review. The majority of these modifications were minor phenotyping disagreements between site reviewers and Core reviewers, which were adjudicated on Final Consensus Review. Uploading digital EEGs to the FTP server was very efficient. A one hour EEG can be archived (i.e. converted to Persyst format) and de-identified in less than 45 seconds using Persyst Insight II . Some digital EEGs were not compatible with Persyst Insight II and had to be viewed using their native browsers. Conclusions: The rigorous review process likely improves the accuracy of EEG phenotyping. The customized EPGP EEG web tools provide a feasible model for use in other epilepsy studies. We anticipate that the collection of phenotypic EEG data using these semi-automated processes will help to identify the genetic contributions to specific EEG abnormalities in EPGP and future epilepsy studies.