Current issues of ACP Journal Club are published in Annals of Internal Medicine


Review: Routine neuroimaging is not recommended for migraine

ACP J Club. 1994 Nov-Dec;121:79. doi:10.7326/ACPJC-1994-121-3-079

Source Citations

Frishberg BM. The utility of neuroimaging in the evaluation of headache in patients with normal neurologic examinations. Neurology. 1994 Jul;44:1191-7.

Quality Standards Subcommittee of the American Academy of Neurology. Practice parameter: the utility of neuroimaging in the evaluation of headache in patients with normal neurologic examinations (summary statement). Neurology. 1994 Jul;44:1353-4.



To make recommendations about the utility of neuroimaging in patients with headache and normal neurologic examinations.

Data sources

MEDLINE was searched for the years 1974 through 1991 using the terms magnetic resonance imaging (MRI), computed tomography (CT), headache, migraine, and diagnosis; pertinent journals were manually searched to July 1993; bibliographies of retrieved articles were reviewed; and content experts were contacted.

Study selection

Selected studies had ≥ 18 patients with normal neurologic examinations. 17 case series (8 of unspecified headache comprising 1825 patients and 9 of migraine comprising 897 patients) met inclusion criteria.

Data extraction

Data were categorized as pertaining to unspecified headaches or migraine; MRI or CT scan; type of study (prospective or retrospective); number of scans; use or nonuse of intravenous iodine contrast media; number of tumors, arteriovenous malformations (AVM), and other neurologic diseases (hydrocephalus, aneurysm, cerebral infarction, subdural hematoma, and venous angioma) identified; and indications for doing the scan.

Main results

5 retrospective and 9 prospective studies examined CT, and 5 prospective studies examined MRI. 1 study of CT was both prospective and retrospective, and 1 prospective study evaluated both CT and MRI. The pooled results showed that 24 (0.9%) tumors (21 [1.2%] in patients with unspecified headache and 3 [0.3%] in patients with migraine), 4 (0.1%) AVM (3 [0.2%] in unspecified headache and 1 [0.1%] in migraine), and 37 (1.4%) in other diagnoses (32 [1.8%] in unspecified headache and 5 [0.6%] in migraine) were identified. CT and MRI were similar for rate of correct diagnoses of treatable lesions. All studies were compromised by differing definitions for migraine, inconsistent indications, variable techniques, and insufficient reporting of results. 5 studies used first-generation CT scanners that had a high potential for false-negative results.


The routine use of neuroimaging is not warranted in patients with recurrent headaches defined as migraine, including those with visual aura with no recent change in pattern. It may be indicated in patients with atypical headache patterns, history of seizures, and focal neurologic signs or symptoms. Insufficient evidence exists to define the role of CT and MRI in evaluation of headaches that are not consistent with migraine.

Source of funding: Not stated.

For article reprint: Dr. B.M. Frishberg, North County Neurology Associates, 3907 Waring Rd, Oceanside, CA 92056, USA. Fax 760-732-0358.


This meticulous, well-done review convincingly establishes that routine neuroimaging in patients with headaches who have normal neurologic examinations is unlikely to show clinically significant abnormalities. Based on this evidence, the American Academy of Neurology has issued a practice parameter discouraging the routine use of these studies in patients whose recurrent headaches are defined as migraine. Despite the strength of the underlying evidence, I suspect that this parameter will be largely ignored in clinical practice. Among the factors not addressed by this study and the Academy that influence the use of neuroimaging in patients with headaches, 2 deserve special attention.

First, although the likelihood of finding an abnormality on neuroimaging is strikingly low, it is not zero; many patients and physicians are unwilling to accept even this low risk. Physicians perceive the legal and medical risks of any missed diagnosis as unacceptable. They are socialized throughout medical school, residency, and practice to be thorough in the search for disease and, therefore, have difficulty tolerating even low levels of diagnostic uncertainty (1). Patients, who have substantial anxiety and disability from this chronic condition and who rarely bear the direct financial burden of testing, likewise view any diagnostic uncertainty as unacceptable.

Second, diagnosis of benign headache conditions is entirely clinical, based on changing, incompletely validated criteria (2). Treatment is not curative and is only partially effective for many patients, frequently leading to concern that something serious has been overlooked. This combination of diagnostic confusion and modest treatment efficacy in a chronic, disabling disease encourages the search for conditions whose diagnosis and treatment are more clearly defined.

Improved methods of diagnosis, better treatment for benign headache disorders, and efforts to increase tolerance of diagnostic uncertainty are likely to be more effective than statistics in reducing the demand for neuroimaging studies.

Elizabeth Loder, MD
The Spaulding HospitalBoston, Massachusetts, USA


1. Woolf SH, Kamerow DB. Testing for uncommon conditions. The heroic search for positive test results. Arch Intern Med. 1990;150:2451-8.

2. Headache Classification Committee of the International Headache Society. Classification and diagnostic criteria for headache disorders, cranial neuralgias and facial pain. Cephalalgia. 1988;8(Suppl 7):1-96.