The early identification of children at risk for intractable epilepsy (IE) would facilitate a choice of more aggressive and effective medical or surgical methods of management. Children with newly diagnosed epilepsy were prospectively identified by participating pediatric neurologists in the State of Connecticut, Jan 1993 through Dec 1997. Parental interviews and follow-up medical record reviews and analyses were conducted by researchers at Montefiore Medical Center, Bronx, NY; and Yale and Northern Illinois Universities. Median follow up was 4.8 years, and 98% were followed for more than 18 months. Intractability was defined as failure of response to two or more drugs and at least one seizure per month over 18 months. Drugs were given in maximal tolerated doses and noncompliance was ruled out. Of a total of 613 newly diagnosed cases, 60 (10%) met criteria for intractability. Of these, 35% were classified (according to International League Against Epilepsy guidelines) as cryptogenic/symptomatic generalized epilepsies, 2.7% as idiopathic, 11% as other localization-related, and 8% were unclassified. Risk of intractability was highly correlated with etiology and epilepsy syndrome (p<0.001). The highest risk of IE was in the symptomatic group and the lowest in idiopathic cases. IE was also correlated with a high initial seizure frequency (an interval of <6 months vs. 4,7 months between 2 seizures) (p<0.0001); focal EEG slowing (p=0.02); and acute symptomatic or neonatal status epilepticus (p=0.001). Age at onset between 5 and 9 years was associated with a lowered risk of IE (p=0.03). Factors not significantly correlated with early intractability included age of onset less than 1 year, absolute number of seizures and duration of epilepsy, and unprovoked or febrile status epilepticus. 
COMMENT. The authors have provided some valuable statistics on the risk factors involved in children with intractable epilepsy (IE) and a means of early predictability. The prognosis of IE may be improved by a more timely selection of cases for trials of newer antiepileptic drugs, and surgical or dietary methods of management. According to the International Classification of Epilepsy, epilepsies are grouped as localization-related (focal, local, partial), or generalized. Localization-related epilepsies are idiopathic/genetic (eg, benign childhood epilepsy with centrotemporal spikes (BECTS)) or symptomatic/cryptogenic (of known or unknown etiology). Generalized epilepsies are grouped as idiopathic, with age-related onset (eg. benign neonatal, benign myoclonic in infancy, childhood and juvenile absence), symptomatic, and idiopathic and/or symptomatic (eg. West’s and Lennox-Gastaut syndromes). Idiopathic or primary forms of epilepsy usually carry a better prognosis for response to treatment than symptomatic/secondary epilepsies. As may be predicted, symptomatic generalized epilepsies, a high initial seizure frequency, and focal EEG findings correlate with a likely intractable course.
The certainty of predicting intractability based on these criteria is questioned by some authorities . Is failure to respond after only 2 drugs sufficient for inclusion as IE, when 14% eventually were controlled within 1.5 and 3 years? A rush to surgical intervention based on these criteria may be inappropriate. May some patients initially responsive later develop intractability? Children with mesial temporal lobe epilepsy, most commonly treated surgically, are particularly susceptible to late onset intractability (Engel J Jr et al, 1997), a condition not addressed in this study. Despite these criticisms and cautionary comments, the criteria for IE identified by Berg and associates will be valuable in counseling parents on the increased necessity for adequate treatment and careful and frequent monitoring of seizures.