PDF《Epilepsia, 45(6):667–672, 2004》

5 years or older, with epilepsy diag- nosed in the recruiting center, and living in the proximity of the center. To prevent the inclusion of acute or progress- ing epileptogenic conditions, which might be an indirect cause of illnesses or accidents, only patients with idio- pathic, cryptogenic, or remote symptomatic epilepsy (4) were enrolled. For each case, a control was sought among persons without epilepsy, matched for age (±5 years) and sex. Controls were recruited among relatives or, if not avail- able, among friends, schoolmates, or workmates. Patients and controls were invited to keep daily diaries for record- ing details about any accident occurring during the study period. An accident was considered any event resulting from a sudden unexpected cause, not a disease, leading to physical damage requiring medical attention or result- ing in financial obligation. For every accident, the type, circumstances (i.e., home, street, work, school, sport), causes, possible correlation to a seizure, complications

667 668 M. VAN DEN BROEK AND E. BEGHI (i.e., any interference with daily living activities), and medical actions were recorded. Monthly telephone contacts and quarterly visits were scheduled to check compliance and improve the quality and completeness of the recordings. The study started in June 1993. Recruitment was completed on June 30, 1997. The follow-up was between

1 and

2 years for each individual. Patients with epilepsy and controls were assessed as a group and in subgroups according to their work and schooling. The working population included current skilled and unskilled workers. Current students repre- sented the school population. Patients with epilepsy also were assessed in subgroups defined by whether they had had seizures in the preceding

2 years (active or inactive epilepsy), and by seizure timing, disease duration, and epilepsy syndrome (5). The data were processed by using the SPSS statistical package and analyzed with actuarial methods. To test the statistical significance of the results, Pearson'

s and the CochranCMantelCHaenszel χ2 tests were used, as appro- priate. The relative risk (RR) with 95% confidence interval (95% CI) was used to indicate the ratio between the inci- dence of each measured event in the cases and controls. Actuarial methods were used to assess the time-dependent risk of accidents (in general and by type). The first acci- dent after inclusion in the study was considered an event. Data were censored at the date of the last follow-up visit. A multivariable analysis also was done, by using the Cox proportional hazard function (6), to adjust for the most important confounders. RESULTS A total of

951 patients with epilepsy and

909 con- trols were enrolled, with

42 unpaired cases for which no matched controls were available. Five controls were later excluded from the calculations because of inconsistent or missing data. The general characteristics of the sample and the disease characteristics in patients are shown in the companion article (7). An associated disability (defined as a physical or psychological problem receiving help from public or private agencies) was present in

123 (13%) pa- tients and

15 (2%) controls (p <

0.0001). Patients and controls were followed up for 17,484 and 17,206 person-months. During the study period,

199 (21%) patients and

123 (14%) controls reported an acci- dent (p <

0.0001). There were

127 patients reporting one accident and

72 patients with two or more accidents (con- trols,

92 and 31). The total number of accidents was