Purpose: To examine different approaches to classifying seizure outcomes. M
ethods: In a prospective cohort study of children (N = 613) with newly diag
nosed epilepsy, seizure outcomes at 2 years were classified as 'good' (grea
ter than or equal to1 year remission),'bad' or 'intractable' (greater than
or equal to2 AED failures, greater than or equal to1 seizure/month over gre
ater than or equal to 18 months), and 'indeterminate' (neither 'good' nor '
bad'). Outcomes at 2 years were compared to outcomes in those followed 4 or
more years. The associations of three commonly studied prognostic factors,
etiology, age at onset, and syndromic grouping with the three-level outcom
e were assessed. Results: 595 (97.1%) children were followed greater than o
r equal to2 years. A 'good', indeterminate, and 'bad' outcome was present i
n 314 (52.8%), 235 (38.3%), and 46 (7.7%) children. Problems with treatment
were recorded in 64.7% of the indeterminate group. In 390 children followe
d greater than or equal to4 years, early 'good' and 'bad' outcomes persiste
d in similar to 80%. About half of those with indeterminate 2-year outcomes
later achieved remission, 8% met criteria for intractability, and 37% rema
ined indeterminate. Most of the associations with etiology, age, and syndro
me were due to variation in the proportion that met criteria for intractabi
lity and not remission. Conclusions: Many children have indeterminate outco
mes, often in association with treatment issues. Clearly 'good' and 'bad' e
arly outcomes can be identified and persist greater than or equal to2 years
later. In the absence of pharmaco-resistance, lack of early remission (ind
eterminate outcome) is usually not associated with a bad outcome, at least
over the next few years. (C) 2001 Elsevier Science B.V. All rights reserved
.