Inter-rater reliability on rhythmic and periodic EEG patterns using Standardized Computer-based Organized Reporting...

This multicenter study aimed to evaluate inter-rater reliability (IRR) for rhythmic and periodic patterns (RPPs), background activity, and interictal findings using the Standardized Computer-based Organized Reporting of EEG (SCORE) across 10 European Reference Network (ERN) EpiCARE centers.

Fifteen neurologists and neurophysiologists, divided into two panels, independently analyzed a total of 277 adult electroencephalography (EEG) recordings, including RPP cases and controls without RPPs, using 5-min EEG epochs. IRR was measured among pairs of raters and quantified using the maximum-likelihood kappa (κML), an unbiased reliability estimator. EEG studies evaluated by at least six raters were included in the analyses (n = 263).

The overall identification of RPP's presence vs absence achieved substantial agreement (κML: .61-.66). Reliability for Main Terms 1 and 2 was substantial to almost perfect (κML: .73-.89), matching or exceeding previously published results. Notably, individual RPP types (κML up to .98) and complex modifiers, such as evolving dynamics (κML: .74-.87) and triphasic morphology (κML: .60-.88), achieved high reliability. Although background activity components showed almost perfect consensus (κML: .84-.98), agreement for posterior dominant rhythm was lower (κML: .06-.47). In contrast, the identification of sporadic interictal epileptiform discharges was high (κML: .89).

The SCORE framework enabled a detailed and comprehensive classification of RPPs, while also capturing background activities and other non-RPP findings in a setting that closely resembles clinical practice. It highlighted complex characteristics for which IRR remains low.