Home Recording And Video Selection: Their use In A Low-Resource Setting For Epilepsy Diagnosis

Objective: In management of epilepsy, identification of an epileptic seizure, classification, epilepsy syndromes, and management decisions relies heavily on seizure semiology. However, since most seizures are not witnessed, obtained semiology has its limitations. This study aimed to determine how many patients could successfully submit a home recording of a seizure event and if adapted video compilations would improve epilepsy diagnosis and classification in a low resource setting.

Methods: A prospective cross-sectional study carried out at a neurology clinic in a teaching hospital in a low-resource setting. Sixty-seven randomly selected patients with recurrent unprovoked seizures and an informant who had regular observed the seizures and had access to video recording facilities were enrolled. Participants were required to fill an interviewer-administered questionnaire, select from a pre-designed video compilation what best described seizure witnessed, then encouraged on acquisition of video recordings at home. In the absence of video electroencephalography, information obtained was compared with a pre-defined algorithm which combined clinical history, physical examination, EEG results and neuro-images. Accuracy and reliability was calculated for different semiological signs and seizure classification.

Results: Sixty seven patients were recruited comprising of 30 females and 37 male patients. Only eight (12%) participants returned with an adequate home recording of seizure episode. Incorporating video selection with questionnaire obtained description improved accuracy for generalized seizure (0.85 vs 0.79) and focal onset seizure (0.84 vs 0.73). Test-retest reliability on video selections by informants showed kappa coefficients ranging from 0.88 – 1.000.

Significance: Home video recording may not be as practical in our environment depending on the setting as adjustments may be required to make it routine. However, selecting videos from pre-selected video compilation may be a viable alternative to improve accuracy.