Clinical Question

What is the best way to diagnose migraine in patients presenting with headache?

Evidence Summary

Headache is a common reason for primary care office visits and is a complaint in more than 1 percent of patient visits.¹ The prevalence of migraine headache in the general population is approximately 6 percent in men and 15 percent in women, peaking at 40 years of age and then declining.² The pretest probability of migraine in patients presenting to a primary care physician with a chief complaint of headache probably is higher than 6 to 15 percent because migraine headaches often are more severe and are accompanied by more diverse nonpain symptoms than other headaches. On the other hand, the pretest probability most likely is lower than the 75 percent prevalence seen in diagnostic studies of patients with possible migraine.^3,4 For the purpose of this article, a reasonable estimate of migraine prevalence in patients with headache is 33 percent.

A meta-analysis identified nausea, photophobia, phonophobia, and exacerbation of headache with physical activity as the best individual symptoms for ruling in or out migraine.⁵ The best predictor is nausea (positive likelihood ratio [LR+] = 19.2). Given a 33 percent overall likelihood of migraine among patients presenting with headache, the presence of nausea alone increases the likelihood of migraine to 90 percent.⁵ The meta-analysis also confirmed that chocolate (LR+ = 7.1) and cheese (LR+ = 4.9) are the triggers most strongly associated with migraine.⁵

Several researchers have attempted to develop and validate clinical decision rules to diagnose migraine. Two researchers have developed rules for use in the primary care setting.^3,4 In the first study, 1,049 French National Railway employees were asked screening questions.³ The 166 employees who had at least one headache every three months were included in the study.³ The participants were interviewed by a neurologist specializing in headache disorders, who made the final diagnosis using International Headache Society (IHS) criteria.³ Logistic regression was used to identify the best independent predictors of migraine.³ The five-item clinical decision rule from this study is shown in Table 1.^2,3

In a more recent U.S. study, researchers developed the three-item Identification of Migraine (ID Migraine) clinical decision rule (Table 2^2,4) for the diagnosis of migraine.² The same researchers validated the rule in a separate group of patients.⁴ The validation study included adults 18 to 55 years of age presenting at 27 primary care and 12 headache clinics.⁴ The mean age of patients was 39 years, and 75 percent were women. As in the first study, the reference standard was application of IHS criteria by a neurologist.⁴

Although the methodology for both rules is strong, the ID Migraine rule has the advantage of being prospectively validated and can be used confidently as a self-administered tool. An additional advantage for American physicians is that it was validated in a domestic population.

It is important to remember that some of the features that make migraine more likely (e.g., photophobia, aura) also make it more likely that the patient will have a significant abnormality on an imaging study. Imaging is recommended for patients with high-risk features such as acute “thunderclap” headache (these patients also require lumbar puncture); abnormal neurologic examination findings; aura; vomiting, if not part of the patient’s usual migraine pattern; headache aggravated by exertion or Valsalva maneuver; or a nonclassic pattern of migraine, cluster, or tension headache.⁶ Imaging also should be considered in patients with new-onset headache, change in character of headache, or adult-onset migraine.⁶

Applying the Evidence

A 35-year-old woman has a recent history of headache episodes. Typically, her headaches are accompanied by nausea without vomiting or photophobia and are unilateral. The headaches last six to 12 hours, are usually gone when she awakens, and are fairly disabling. She denies having fever, weight loss, or other red flags for serious disease and has a normal neurologic examination. What is the probability that the patient has migraine headaches?

Answer: Based on the five-item clinical decision rule (Table 1^2,3), the patient has four out of five clinical features of migraine and, therefore, has a 92 percent probability of migraine. She also has a 62 percent probability (the highest possible) using the ID Migraine rule (Table 2^2,4). Because she has no red flags and a normal neurologic examination, you determine that imaging is not necessary, and you discuss migraine treatment options with the patient.