Improving the inter-rater agreement of hypsarrhythmia using a simplified EEG grading scale for children with infantile spasms

Mytinger, J. R.; Hussain, S. A.; Islam, M. P.; Millichap, J. J.; Patel, A. D.; Ryan, N. R.; Twanow, J. D.; Heyer, G. L.

Epilepsy Res. 2015 Aug 19; 116:93-8

Abstract

BACKGROUND: There is poor inter-rater agreement in determining the presence or absence of hypsarrhythmia among patients with infantile spasms. Yet, remission of hypsarrhythmia has been used as a clinical and research outcome measure. Two important features of hypsarrhythmia are the burden of epileptiform discharges and the amplitudes of background slow waves. We hypothesized that an electroencephalogram (EEG) grading scale emphasizing epileptiform discharge burden and the amplitudes of background slow waves would improve inter-rater agreement in interpreting hypsarrhythmia. Our aim was to assess inter-rater agreement of hypsarrhythmia using a novel and simplified EEG grading scale called the 'BASED' (Burden of Amplitudes and Epileptiform Discharges) score and compare this to the traditional method of EEG analysis. METHODS: Twenty patients with infantile spasms were prospectively evaluated and electroclinical outcomes were determined. Forty EEG clips (20 pre-treatment and 20 post-treatment), representing the most severely abnormal five minute sleep epoch of each study, were assessed by three reviewers blinded to treatment and clinical outcome. Fleiss' kappa (capital KA, Cyrillic) was used to assess the inter-rater agreement in the interpretation of hypsarrhythmia when using the BASED score compared to the traditional method of EEG analysis. RESULTS: Reviewers had favorable inter-rater agreement using the BASED score in interpreting hypsarrhythmia (capital KA, Cyrillic: 0.87) compared to when using the traditional method of EEG analysis to interpret hypsarrhythmia (capital KA, Cyrillic: 0.09). The three reviewers all agreed on the presence or absence of hypsarrhythmia in 37/40 (93%) epochs using the BASED score but in only 15/40 (38%) epochs using the traditional method of EEG analysis, p=<0.001. CONCLUSION: When compared to the traditional method of EEG analysis, the BASED score allowed for better inter-rater agreement in the interpretation of hypsarrhythmia. Future infantile spasms clinical trials must better define criteria for hypsarrhythmia.

Read More on PubMed