By Dara G. Jamieson, MD

Associate Professor of Clinical Neurology, Weill Cornell Medical College

Dr. Jamieson reports no financial relationships relevant to this field of study.

SYNOPSIS: The diagnosis of primary headache disorders by a computerized and clinical paradigm can predict a baseline prevalence of intracranial abnormalities on brain imaging. Some historical “red flags” in children with headaches, including morning headaches and occipital pain, are not associated with increased intracranial abnormalities.

SOURCES: Wang R, Liu R, Dong Z, et al. Unnecessary neuroimaging for patients with primary headaches. Headache 2018; Aug 23. doi: 10.1111/head.13397. [Epub ahead of print].

Irwin SL, Gelfand AA. Occipital headaches and neuroimaging in children. Curr Pain Headache Rep 2018;22:59.

Tsze DS, Ochs JB, Gonzalez AE, Dayan PS. Red flag findings in children with headaches: Prevalence and association with emergency department neuroimaging. Cephalalgia 2018; Jan 1. doi: 10.1177/0333102418781814. [Epub ahead of print].

Brain imaging often is used to differentiate between primary headaches without specific neuroimaging correlates and secondary headaches due to visible brain lesions. However, the ubiquity of headaches necessitates a parsimonious approach to brain imaging. The study from Wang et al at the Headache Center of the Chinese People’s Liberation Army General Hospital was designed to verify that primary headache patients do not need neuroimaging. Brain imaging was obtained in 1,070 healthy controls (345 male, aged 40.18 ± 12.46 years) and 1,070 primary headache patients (345 male, aged 40.05 ± 12.30 years) diagnosed by computerized clinical decision support systems (CDSS) and correlating clinical assessment. The CDSS is an interactive decision support system using computer software and International Classification of Headache Disorders criteria to diagnose based on clinical data.

The primary headache diagnosis, without “red flags” by history or abnormality on examination, was confirmed by a headache specialist. Primary headaches were migraine (62%), tension-type (32%), trigeminal autonomic cephalalgias (5%), and other (1.5%). In each group, 382 participants underwent CT scans and 688 underwent MRI scans based on individual non-neurological considerations. The neuroimaging findings were classified as significant abnormalities, nonsignificant abnormalities, or normal. Significant abnormalities were defined as neoplastic disease, hydrocephalus, vascular malformations (aneurysms, arteriovenous malformations, dural fistula, and/or cavernous angiomas), Chiari malformations, intracranial hemorrhages, and acute infarcts. White matter lesions, commonly seen in migraineurs, were not considered a significant finding. The rate of significant abnormalities, which were noted only on the MRI scans, was not significantly different in the primary headache group (four patients: two hydrocephalus, two nose/throat tumors) compared to the healthy controls (five controls: two cerebral infarction, one acoustic schwannoma, two cavernous angiomas). The authors concluded that with the possible exception of the very rare diagnosis of retinal migraine, neuroimaging is not necessary for patients with the presumed diagnosis of primary headaches.

Headaches in children are especially concerning to physicians and parents, so neuroimaging often is used for reassurance that the primary headache diagnosis is correct. The authors of two recently published articles evaluated the role of neuroimaging in children with headache. In a prospective cohort study, Tzse et al evaluated the prevalence of historical “red flags” and their association with intracranial abnormalities in 224 healthy children 2-17 years of age who were evaluated for headaches in the ED. The ED physicians completed standardized forms to document headache characteristics and associated symptoms, along with examination findings. At least one presumed historical “red flag” was found in 87.9% of the children, including headache upon awakening from sleep (34.8%), headache noted with or soon after awakening in the morning (39.7%), and headaches increasing in frequency, duration, and severity (40%, 33.1%, and 46.3%, respectively). Children who were not imaged in the ED received a four-month telephone follow-up.

In the 33% of children who received ED neuroimaging, the prevalence of emergent, serious, and incidental intracranial abnormalities was 1%, 1.5%, and 7%, respectively. Fifty-two of 55 children who underwent ED neuroimaging for a documented reason (as opposed to “no specified reason”) were noted to have historical red flags, such as headaches on awakening from sleep (46.1%), headaches with or soon after morning awakening (26.9%), or headaches of increasing frequency (19.2%). The authors concluded that while historical red flags are common in children presenting with headaches to the ED, their presence is associated with a low risk of emergent intracranial abnormalities.

Occipital head pain in children has been considered a red flag to indicate a secondary headache. Irwin’s review of the medical literature determined that 0-4.1% of children with occipital headaches and normal neurological examinations demonstrate significant findings on neuroimaging. Migraine was the most likely etiology for occipital headaches, which are noted in up to 20% of children with headaches. Occipital headaches in neurologically normal children are no more likely to be associated with intracranial pathology than headaches localized to other cranial locations.


The question “Is brain imaging needed to diagnose a headache?” plagues patients, parents, physicians, and payers. Clearly, an associated abnormality on neurological examination dictates brain imaging. The more common and more complicated scenario involves a seemingly characteristic history of a primary headache disorder, a normal neurological examination, and a patient skeptical of the proffered diagnosis. Wang et al used a “belt-and-suspenders” approach to diagnose primary headache disorders and then image both headache patients and headache-free controls, finding no difference in significant brain lesions. The value of clinical and computer diagnostic skills is reassuring, but the authors emphasized the importance of time and expertise in obtaining a detailed neurological history and in performing an appropriate neurological exam prior to deciding that imaging is not indicated. Historically, imaging advocated with some types of primary headaches, such as side-locked headaches including trigeminal autonomic cephalgias and cluster headaches.

This study and personal experience indicate that even with these side-locked headaches, imaging adds little to diagnostic accuracy if the history is characteristic and the examination is normal. Some pediatric red flags, such as morning and occipital headaches, appear to be of little significance and should no longer elicit a knee-jerk response to image. A detailed history, an appropriate examination, and a thoughtful approach should lead to a focused approach to ordering brain imaging studies in patients with headaches. However, if brain imaging is deemed appropriate, then an MRI scan, as opposed to a CT scan, generally should be obtained.