Abstracts

RATIONALE: Previous studies of risk factors for seizure-related injuries in epilepsy patients have been biased towards refractory epilepsy. Multivariate study of risk factors is lacking in the literature.
METHODS: Subjects consist of 247 Rochester, MN residents with epilepsy diagnosed between 1975 and 1984. Patients were followed from time of diagnosis until death, migration out of Rochester, or until July 1, 1999. Seizure-related injuries were defined as any injury occurring directly as a result of a seizure, other than oro-lingual trauma, sufficient for patient to seek medical attention or for injury occurrence to be determined during the course of medical care. Characteristics of patients with injuries were compared to those without injury to identify risk factors for seizure-related injury. Kaplan-Meier life table methods were used to determine cumulative probabilities of having a seizure-related injury after first seizure. Cox Proportional hazards regression models were utilized to assess the univariate associations between patient characteristics and occurrence of seizure-related injury. Stepwise multivariate Cox regression analysis was performed on factors that were significant on univariate analysis.
RESULTS: During a total follow-up of 2,714 patient-years, 62 seizure-related injuries were identified in 39 patients (16 %; one injury every 44 patient-years). Most injuries were minor extracranial soft tissue contusion or laceration of the head (79%). After the initial seizure, the cumulative probability of having a seizure-related injury increased steadily to 13% in the 5th year and then remained relatively stable through the 10th year. After year 10 it then increased steadily again until it reached 20% in the 15th year and then remained stable. On univariate analysis three factors were significant: seizure frequency score (p[lt]0.0001, Risk Ratio 1.33); history of generalized seizures (p[lt]0.0045, Risk Ratio 4.48); and number of anti-epileptic drugs used (p[lt]0.0191, Risk Ratio 1.94). Other factors of note not identified to be significant (i.e. p[gt]0.05) included gender, living environment, missing of anti-epileptic drug dose, Rankin score, epilepsy cause (cryptogenic/idiopathic versus symptomatic) and type of epilepsy. However, seizure frequency score was the only significant risk factor for seizure-related injury on multivariate analysis (p[lt]0.0001, Risk Ratio 1.33). For every increment in the seizure frequency score there was a 33% increase in the risk of seizure-related injuries.
CONCLUSIONS: In this population-based study, the only independent determinant of seizure-related injuries was degree of seizure control. Effective treatment with complete elimination of seizures is the only reliable means to reduce the risk of seizure-related injuries. Our findings have implications for counseling patients regarding therapeutic goals and treatment options in epilepsy.
[Supported by: Mayo Foundation for Research and Education.]