The effectiveness of cerebral imaging in the diagnosis of chronic headache

The increasing availability of high-resolution cerebral imaging scanners has fueled enthusiasm for their use to "rule out" brain tumor and other serious neurologic conditions in patients with headache. The effectiveness of this practice, however, has not been tested since the advent of newer scanning equipment. Our objective was to measure the usefulness of cerebral imaging in patients with chronic isolated headache.

A retrospective study with a 15- to 27-month follow-up period.

A group-model health maintenance organization.

Adult patients, 100,800, in a health maintenance organization and an enriched sample of 63 patients with neurosurgical conditions from other health maintenance organization hospitals.

During 1990, 1083 cerebral computed tomographic scans were performed on 863 adults (0.9% of health maintenance organization adults). Eighty-nine patients were scanned for chronic isolated headache; none of the scans provided important new information (95% confidence interval, 0%, 3%). Long-term patient follow-up confirmed that this low yield could not be attributed to diagnostic work-up bias. Further attempts to support a policy of imaging patients with isolated headache were also unsuccessful. Review of an enriched sample of patients with malignant brain tumor and patients requiring craniotomy for other reasons (n = 40) demonstrated that no patient had headache alone at the time of diagnosis (95% confidence interval, 0%, 8%) and that only 5% (95% confidence interval, 0%, 12%) of these patients sought medical attention for headache alone. Sampling a second enriched sample of patients who were referred from other hospitals (n = 63) because of conditions requiring neurosurgical procedures demonstrated that only 6% of patients presented with chronic isolated headache alone (95% confidence interval, 0%, 12%). Uncertainty regarding the appropriateness of imaging patients with headache was illustrated by the extreme interphysician variability of this practice.

Our study demonstrates the large potential cost and low (although not zero) yield associated with nonselectively imaging patients with chronic isolated headache.