Seizure description accuracy and seizure classification results from patient histories

To study the influence of inaccurate histories on the seizure classification results and explore the characteristics of the inaccurate descriptions from the relatives of people with epilepsy (PWE).

Fifty-four individuals from our multi-disciplinary meeting (MDM) were included. Their histories were collected. The relatives of the individuals answered a questionnaire regarding habitual seizures. Experienced epileptologists reviewed ictal videos, classified habitual seizures based on histories and videos separately with short term EEGs and brain MRIs, and answered the same questionnaire. With the answers from the epileptologists as references, the overall mistake rate for each relative was calculated. Mistake rate, kappa value, and other indicators were also calculated for each item of the questionnaire. Furthermore, seizure-related and socioeconomic factors were included in a multivariable logistic regression model to determine their roles in the overall mistake rates.

The inconsistency of classifications from the histories is 25.93 %, with 18.52 % missing focal onset descriptions. The average overall mistake rate of the questionnaire was 28.92 %. Answers to most items were in moderate agreement with the epileptologists. Relatives were not good at reporting eye, face and head movements; asymmetrical postures; and time estimations. Movements of the limbs were likely to be detected. Younger individuals as well as relatives with higher educational levels were related to higher overall mistake rates (P = 0.012 and P = 0.031, respectively).

The potential inaccuracy in histories didn't lead to a large deviation on seizure classifications. There is a certain space for improvement, the importance of detecting focal signs, subtle movements, and asymmetrical postures during seizures should be stressed.