Headaches are the most common neurologic condition, affecting more than 90% of the population at some point in their lives. Even though headaches are one of the most common indications for MRI, the optimal MRI protocol for this prevalent condition is unknown. Furthermore, patients and physicians are often worried about brain tumors, but the yield of neuroimaging in those with chronic headaches is 1-3%, which is comparable to a healthy population. In response, multiple guidelines recommend against routine neuroimaging in certain headache populations, but which factors should prompt neuroimaging is not clear. Furthermore, false positives are likely common on headache MRI, yet downstream harms of false positive findings have not been studied. We propose three potential solutions to improve the care of veterans with headaches.
In Aim 1, we plan to determine the optimal set of MRI sequences for patients presenting with headaches. Our hypothesis is that FLAIR only MRI studies are non-inferior to conventional MRI for identifying brain lesions in patients with headache. Two neuroradiologists will read each MRI scan obtained in veterans with a headache diagnosis that are receiving an MRI as part of their routine care. Our primary analysis will determine the sensitivity of FLAIR-only MRI compared to conventional MRI for identifying headache-causing lesions. If our hypothesis is correct, fewer sequences may lower the incidence of false positive findings without substantially reducing the yield of true positives. Fewer false positives may also decrease the downstream harms associated with current neuroimaging practices.
In Aim 2, we plan to define the incidence of downstream harms after false positive findings, which our preliminary data indicates may be quite high. False positive results will be defined as all abnormal MRI findings in the clinical radiologist's report that did not lead to a change in clinical management as determined by two neurologists with adjudication via consensus committee. Downstream harms of diagnostic imaging will be defined as additional diagnostic testing, procedures and consultations. We will also identify the patient phenotypes that are predictive of headache-causing lesions including headache characteristics, headache classification, and high risk clinical features (?red flags?). Previous small studies from selected populations have attempted to address this question, but we will be able to overcome many of the limitations of prior work. By characterizing the harms and high risk patient phenotypes, we will be able to inform rational clinical neuroimaging decisions to improve the health of veterans. Third, in Aim 3, we will identify the core reasons behind MRI overuse in the headache population but surveying providers and patients using the Theoretical Domains Framework. All three of these approaches will directly lead to future implementation studies. Our project is innovative is many ways. If we are able to determine an optimal protocol that is different than conventional imaging, our project may lead the way in changing the paradigm of current neuroimaging from an anatomical based approach (brain) to a symptoms based approach (headaches). Furthermore, we will also be the first group to comprehensively study the harms associated with current neuroimaging practices while also refining our understanding of high risk patient phenotypes. Understanding both of these factors will allow clinicians to have the necessary evidence of the risks and benefits of neuroimaging in different clinical scenarios.

Public Health Relevance

Headache imaging is common, costly, and increasing over time despite guidelines that recommend against these studies. New approaches are needed to improve the care of the many veterans with headaches. We currently do not know the correct MRI images needed for patients with headaches. We also do not know the benefits or harms of headache scans. Furthermore, we do not know the main reasons that physicians and patients want to order these studies despite the evidence in the guidelines. Our project attempts to address these problems. We will compare a one sequence MRI test to the typical MRI that veterans with headaches receive as part of their routine care. We will also determine how common abnormal findings are seen that do not help patients and the harms that result from these findings. We will also determine which patient factors predict finding an important finding on MRI. Finally, we will discover the reasons that physician and patient want these tests despite the guidelines that suggest that we should order them only rarely.

National Institute of Health (NIH)
Veterans Affairs (VA)
Non-HHS Research Projects (I01)
Project #
Application #
Study Section
Epidemiology (EPID)
Project Start
Project End
Budget Start
Budget End
Support Year
Fiscal Year
Total Cost
Indirect Cost
Veterans Health Administration
Ann Arbor
United States
Zip Code