The Quality Standards Subcommittee of the American Academy of Neurology and the Practice Committee of the Child Neurology Society have developed practice parameters for evaluation of children and adolescents with recurrent headaches. Individual committee members reviewed titles and abstracts for content and relevance. Articles selected included studies with more than 25 patients, those with details of a neurologic examination, and those concerning etiology. Of those reviewed, only one study reported results of laboratory tests and none concerned the role of routine lumbar puncture. Eight studies assessed the use of EEG in 1,148 children with recurrent headache, but none compared the incidence of EEG abnormalities in migraine vs nonmigraine pediatric headache patients.
EEG was not recommended for routine evaluation, even in patients suspected of having seizure-related headaches. Neuroimaging should only be considered in patients with an abnormal neurologic examination or findings suggesting CNS disease. Predictors of a space-occupying lesion included: 1) recent-onset or change in type of severe headaches; 2) absent family history of migraine; 3) abnormal neurologic exam; 4) gait abnormalities; and 5) seizures.
Controlled prospective studies are needed to define risks for serious intracranial disease with recurrent headaches, to define the role of laboratory tests, and the value of neuroimaging in patients with normal neurologic exam. The AAN and CNS attach a disclaimer recognizing that each patient is an individual with different circumstances and specific diagnostic indications. 
COMMENT. The committee concluded that the diagnosis of recurrent headaches is made on a clinical basis and not by routine tests. Diagnostic studies such as neuroimaging are not recommended in the absence of associated risk factors and an abnormal neurologic examination.
The committee wisely attached a disclaimer regarding these recommendations which should not be accepted as all inclusive. For example, in deciding on the necessity for neuroimaging, the luxury of follow-up evaluation and observation over time may not be available to the neurologist who examines a patient in consultation. Deferral of neuroimaging may not always be practical or judicious . In particular, headaches in children younger than 4 years may pose a diagnostic problem. Straussberg R, Amir J  report 5 young children with headache as the first symptom of intracranial tumor, and a normal initial neurologic examination, including the fundi. These authors stress the need for neuroimaging studies in young children with recent-onset recurrent headaches, even when the neurologic exam is normal.
An EEG study in 100 consecutive children with recurrent headaches, migrainous and nonmigrainous, apparently not qualifying for inclusion in the Committee selection criteria, found epileptiform EEGs in 18% of the total group and with the same incidence in the migraine group. The EEG did not distinguish migraine from nonmigraine patients . A trial of phenytoin in 30 children with migraine in this study demonstrated a beneficial response to the anticonvulsant in 77%. In 13 with abnormal and 17 with normal EEGs, the beneficial response rates were 61% and 88%, respectively. Response to phenytoin was not significantly correlated with an abnormal EEG. Alternative anticonvulsants are now recommended in the prophylactic treatment of migraine, independent of EEG findings or asssociated seizures.