Chapter 18. Historical Features in Primary Headache Syndromes

Headache is a nearly universal symptom. As an example of the prevalence of headache, a study of 410 patients who had visited a primary care internal medicine practice found that headache was the fourth most common symptom and was exceeded only by fatigue, back pain, and dyspnea.1 In an early study of over 1 million unselected individuals from the general population, headache was the single most common current symptom and was reported by 39% of men and 56% of women.2 As physicians, nearly all of us have had personal experience with headache and can understand the headache descriptions that we hear from our patients. Chapter 17 elegantly details the prevalence of this common symptom and of common primary headache syndromes. Primary headaches are those without a pathologic basis.3–5 These are benign recurring headaches of unknown cause. The most common primary headache syndromes are migraine, tension-type, and cluster headache. Secondary headaches are the result of an underlying pathologic cause.

When faced with the large numbers of patients who seek medical evaluation for headache, clinicians seek to identify the rare patient with a serious headache from the rest whose headache is benign in nature. Two general approaches assist this effort. Firstly, one must learn the warning symptoms and signs that suggest a pathologic cause for headache. Many published reviews have offered such advice.4,6,7 A complementary approach is to learn to confidently diagnose benign primary headache syndromes through careful history taking and the systematic application of established diagnostic criteria. Primary headaches are clinical diagnoses that are based on history taking alone. With the exception of the occasional persistence of a partial Horner’s syndrome among asymptomatic patients with a history of cluster headaches, the physical examination of a patient with primary headaches is normal during headache-free intervals.

The most commonly used criteria are the International Headache Society (IHS) classification and diagnostic criteria for headache disorders, cranial neuralgias, and facial pain.8 The use of these criteria helps to identify uniform populations of patients for research and epidemiologic studies. The criteria themselves, however, are complicated, not easily committed to memory, and may be unnecessarily restrictive in the daily clinical care of patients.

When evaluating individual patients with headache, clinicians will benefit from understanding which historical features are most useful in establishing or excluding a particular primary headache diagnosis. In this chapter, I review and summarize published clinical series of patients with migraine, tension-type headache, and cluster headache to determine the sensitivity, specificity, and likelihood ratios of individual historical features.

Clinicians may initially classify all headaches as either old or new. Old headaches are similar to those that have occurred repeatedly over time. Primary headaches are old headaches. New headaches are either headaches of recent onset or those that represent a change in the character or pattern of an old headache. A new headache may ultimately prove to be the first instance of a primary headache syndrome; however, clinicians cannot reach this conclusion with certainty until a pattern of similar headaches emerges over time.

A change in the intensity or frequency of an old headache is still an old headache. In this case, the physician must establish the precipitant for the increase in the severity of the old headache. A change in the quality, character, or descriptors of the headache, however, is a new headache. As an example, if a patient’s headaches are typically unilateral and throbbing, a headache that is bilateral, constant, and progressive is a new headache. Most old headaches are benign, and the longer a headache syndrome has been present, the more likely it is to be benign.

In a cost-constrained environment, physicians must minimize the unnecessary use of diagnostic imaging in the evaluation of patients with headache. Neuroimaging is, by definition, normal or unhelpful in patients with primary headaches. A carefully obtained history of migraine, tension-type, or cluster headache minimizes the need for such testing. Several studies have confirmed the low yield of diagnostic imaging among patients with clinical diagnoses of primary headache syndromes.

The history and physical examination may identify patients who need no further diagnostic evaluation. A retrospective study evaluated 592 patients with headache and normal neurologic examinations who had been referred for cranial computed tomography (CT) studies.9 There were no cases of intracranial pathology sufficient to explain headache in this entire group. Mitchell and coworkers studied 350 patients with headache who were referred by their clinicians for head CT studies.10 Among the 320 patients with normal neurologic examinations, there were only three clinically significant CT abnormalities. Each of these three patients had a headache of recent onset with worrisome atypical features.

In 1994, the American Academy of Neurology reviewed the role of neuroimaging in the evaluation of patients with headache and normal neurologic examinations.11 The incidence of clinically important pathologic findings among patients with normal neurologic examinations and migraine or unspecified headache was 0.4% and 2.4%, respectively. They recommended no routine imaging for patients with typical migraine who had no recent change in pattern, seizures, or focal neurologic signs. They acknowledged insufficient evidence, but suggested that neuroimaging may be indicated for patients with atypical headache patterns, seizures, or focal neurologic signs or symptoms.

More recently, the US Headache Consortium reviewed the incidence of significant intracranial abnormalities in patients with migraine or tension-type headache and normal neurologic examinations.12 They found only two abnormal imaging studies among 1,086 patients with migraine and no abnormal studies among 83 patients with tension headache. In contrast, among patients with unspecified types of headaches and normal neurologic examinations, the prevalence of significant intracranial abnormalities on imaging studies varied from 0% to 6.7%. These reports confirm the importance of careful history taking to identify patients for whom the yield of imaging is sufficiently low that clinicians can establish a clinical diagnosis without further study.

In the same review, focal neurologic symptoms, rapidly increasing frequency of headaches, headaches causing awakening from sleep, lack of coordination, subjective numbness or tingling, and headache with Valsalva maneuver each predicted a significantly higher likelihood of intracranial pathology. By definition, clinicians should not make a diagnosis of a primary headache syndrome if the neurologic examination is abnormal. An abnormal neurologic examination confers a positive likelihood ratio of 3.0 (CI 2.3-4.0) for a significant abnormality on neuroimaging.12

Migraine headaches are common. In a review by Stewart and colleagues of four prevalence studies that used the IHS criteria, overall migraine prevalence was 6% among men and 16% among women.13 The initial onset of migraine headaches most commonly occurs during between the ages of 10 and 25 years. Migraine headaches occur for the first time after age 40 uncommonly. In one incidence study, for example, 77% of patients with migraine first sought medical attention for migraine before age 40.14 The new appearance of a migraine-like headache in a person over age 40 should prompt consideration of other possible diagnoses. Hamelsky and colleagues characterize migraine prevalence in more detail in Chapter 17.

In practice, clinicians most commonly entertain the diagnoses of migraine and tension-type headache for patients with long-standing recurring headaches. Although less common than tension-type headache, migraine is more likely to cause sufferers to be disabled by their headaches and to seek medical attention for their symptoms.15

Clinical Features

Tables 18-1 and 18-2 list the IHS criteria for the diagnosis of migraine without aura and migraine with aura, respectively. The principal historical features are headaches that are unilateral, throbbing, moderate to severe in intensity, and worse with ordinary physical activity; last from 4 to 72 hours; and are associated with nausea, photophobia, and phonophobia. All of these features are, however, not equally useful to clinicians in establishing a diagnosis of migraine. Table 18-3 summarizes reported clinical series that detailed the frequency of particular clinical features in patients with migraine and tension-type headache. These data were pooled from multiple series published over the past four decades. All studies were classified as to their use of the IHS diagnostic criteria or other criteria. Sensitivity, specificity, and likelihood ratios were calculated for the diagnosis of migraine as compared with tension-type headache. A positive likelihood ratio indicates the increase in the odds of the diagnosis of migraine if the particular feature is present. A negative likelihood ratio indicates the decrease in the odds of the diagnosis of migraine if the feature is absent.

Nausea, exacerbation by physical activity, photophobia, and throbbing headache are the most sensitive features for the diagnosis of migraine. Sensitivities are 81%, 81%, 79%, and 73%, respectively. Despite the origin of the word migraine from “hemicrania,” only 65% of migraines are unilateral, and this is the least sensitive of the major clinical criteria. When compared with tension-type headache, the most specific features for migraine are nausea, phonophobia, photophobia, and unilateral headache, with specificities of 96%, 87%, 86%, and 82%, respectively.

The features with the best overall predictive value are nausea, photophobia, phonophobia, and exacerbation by physical activity. The particularly high positive predictive value of nausea results in part from the inclusion of large numbers of patients who were classified according the IHS criteria. The IHS criteria for the diagnosis of tension-type headache require the absence of nausea. However, questionnaire studies that have used less restrictive criteria for the diagnosis of tension-type headache have also found nausea to be highly specific.34,35 Headache duration is less useful to distinguish between the two diagnoses, with the exception that headaches lasting less than 4 hours are less likely to be migraine.

Authors used many different diagnostic criteria for migraine and tension-type headache in the pre-IHS era. Despite these varied definitions, the likelihood ratios for all pooled studies are not substantially different than those restricted to studies using the IHS criteria. The data suggest that the pre-IHS studies also included fairly uniform populations of patients.

Features of Migraine Aura

Among patients with migraine, one third experience migraine with aura. In a study of 4,000 randomly selected 40-year-olds in a Danish population, the lifetime prevalence of migraine without aura was 11.8%; that of migraine with aura was 5.5%.36 The migraine aura is sufficiently characteristic that a carefully obtained history of an aura substantially increases confidence in the diagnosis of migraine. Both the subjective aura elements and the duration of the aura are important features. Table 18-4 summarizes the sensitivity of various aura features among pooled series of patients having migraine with aura.

Visual auras are most common; 84% of patients having migraine with aura experience a visual aura. Positive visual phenomena occur slightly more frequently than negative visual phenomena. Positive phenomena include zigzags (fortification spectra), stars, or flashes. Many eloquent descriptions of visual auras exist in the medical literature. In an early review of migrainous visual aura, Alvarez described his personal experience of fortification spectra.42 “Another time I saw a fine zigzag line running up and down and a coarse one running below it, horizontally. Later, the two ran together, end to end, and bowed out to the right. The line resembles a snake fence, or an old-style fortification with projecting angles. In some spells, the line is so brilliant one can see it easily with the eyes open.”*

[* Reprinted from the American Journal of Ophthalmology, vol. 49, Alvarez WC. The migrainous scotoma as studied in 618 persons, pp. 489–504. Copyright 1960, with permission from Elsevier Science.]

Negative visual phenomena include scotoma and hemianopsia. The presence of hemianopsia is one of the features that establishes a diagnosis of migraine with typical aura (previously referred to as complicated migraine). Disturbances of visual perception are least common. In the study of Queiroz and colleagues,39 these included, in descending order of frequency, foggy vision, looking through heat waves or water, tunnel vision, mosaic vision, micropsia or macropsia, corona phenomena, and complex hallucinations. Lewis Carroll, the author of Alice in Wonderland, was known to suffer from migraine with aura; some authors have speculated that Alice’s visual distortions in his novel may have paralleled complex visual hallucinations that he experienced during migrainous auras.43

The duration of the aura is also characteristic. IHS criteria require that each aura feature last from 4 to 60 minutes. In practice, the most common aura duration is 20 minutes, and 70% of visual auras last less than 30 minutes (see Table 18-4). Auras that last a few seconds or minutes are distinctly uncommon in migraine and should raise the possibility of seizure phenomena.

Nonvisual auras nearly always occur in conjunction with visual auras rather than as isolated events. In one study, only 4% of auras were complex nonvisual auras that occurred in isolation without accompanying visual auras.37 Among nonvisual auras, sensory auras are most common, followed by aphasia and motor auras. The sensitivities are 20%, 11%, and 4%, respectively. Sensory auras are unilateral, usually begin in the hand, and then progress to the arm, face, and tongue.38 Aphasic aura symptoms include paraphasia, impaired production of language, and impaired comprehension of language. Motor auras usually occur in conjunction with sensory auras rather than in isolation.

Historical Features of Migraineurs

Individuals with migraine are more likely to have a family history of migraine, and a childhood history of vomiting attacks or motion sickness. Although these factors, by themselves, are insufficient to establish a diagnosis of migraine, they can be useful in the evaluation of an individual patient if the type of primary headache remains uncertain after taking a careful history.

Of these features, the familial tendency is the least controversial. In a recent review of over 2,500 patients with data on family history, 58% of migraineurs had a family history of migraine as compared with 12% of unselected individuals without headache.44 In a case-control study, Stewart and colleagues reported a relative risk of 1.50 among family members of probands with migraine.45 A positive family history was more often present among patients with severe migraine and disability. Russell and coworkers noted different family histories among patients having migraine without aura and those having migraine with aura.46 In their study of 183 patients, migraine without aura was associated with a 2.9 relative risk of family history of migraine without aura, but no increase in risk of migraine with aura. Patients having migraine with aura were twice as likely to have family histories of both migraine with and without aura than expected.

Neither childhood vomiting attacks nor motion sickness are criteria for the diagnosis of migraine in the IHS classification. However, each of these features is more common in patients destined to develop migraine that in those without migraine. Thirty-two percent of patients with migraine report a history of childhood vomiting attacks as compared with only 14% of individuals without headaches.44 Data for a history of motion sickness are similar. Children who develop disabling headaches by the age of 5 years are 2.8 times more likely to report motion sickness than those without disabling headaches.47 In addition, children with migraines score significantly higher on a motion sickness susceptibility questionnaire than children without migraines.48

Tension-type headache is considerably more common than migraine. Thirty-six percent of men and 46% of women suffer from episodic tension-type headache (see Chapter 17). Patients with tension-type headache are, however, less likely to seek medical attention than are those with migraine because their headaches are less frequently disabling.

The IHS criteria define tension-type headache largely as a chronic recurring headache with few or no migrainous features (Table 18-5). Tension-type headache is diagnosed based as much on the absence of particular features as the presence of specific historical features. The principal features are headaches that are pressing or tightening, mild or moderate (not severe) in intensity, and bilateral, and are not associated with nausea, photophobia, phonophobia, or aggravation by ordinary physical activity. Table 18-3 summarizes the prevalence of these features (1-specificity) based on pooled data from clinical series. The IHS criteria are restrictive and do not permit the diagnosis of tension-type headache if nausea is present or if both phonophobia and photophobia are present. As such, published series that report the frequency of particular clinical features run the risk of circular reasoning. Few series precede the publication of the IHS criteria so it is difficult to determine if specific clinical features would be equally frequent using less restrictive criteria.

Although the IHS criteria allow tension-type headaches to last from 30 minutes to 7 days, Ulrich and colleagues studied 499 patients with tension-type headache and found that 82% of patients reported headache duration of less than 24 hours.26 Chronic recurring headaches that usually last more than 72 hours are more likely to be tension-type than migraine, but this occurrence is rare. In general, the duration of headache episodes is not useful as a distinguishing feature between the two diagnoses.

The differential diagnosis of tension-type headache includes several secondary causes for headache. Cervicogenic headache, temporomandibular joint (TMJ) dysfunction, and temporal arteritis may occasionally be confused with tension-type headache. Cervicogenic headache is suggested by a predominantly occipital and occasionally frontal location in addition to neck pain. This headache is more commonly daily as opposed to episodic and may occur for the first time in an older individual more than 50 years of age. In contrast, the peak prevalence of tension-type headache is in individuals aged 30 to 39 years,49 and the first occurrence is most commonly in the teens or twenties.

TMJ dysfunction is suggested by a temporal location of pain, morning headaches, bruxism with excessive dental wear, and jaw pain. These features help to distinguish it from tension-type headache. The headache of temporal arteritis is variable but, similar to that of tension-type headache, is commonly pressing, bilateral, and without migrainous features.50 The most important factor that distinguishes the headache of temporal arteritis from that of tension-type headache is the later age at onset. Physicians should consider this possibility in any patient with a new-onset headache who is older than 50 years.

Cluster headache is a distinct syndrome that is easily recognized by clinicians familiar with its historical features. It is the least common primary headache syndrome. The prevalence has not been well studied but was found to be 0.1% in a single report of nearly 10,000 18-year-old men.51 Given the mean age of onset of this disorder at age 30,52–54 the actual prevalence is probably higher and may be as high as 0.4% to 1.0% among men.55Table 18-6 lists the IHS criteria for the diagnosis of cluster headache. The principal features are headaches that are severe, unilateral, and supraorbital or temporal; last from 15 to 180 minutes; and are associated with autonomic symptoms on the involved side.

Table 18-7 outlines the sensitivity of particular clinical features of cluster headaches and was derived by pooling data from published clinical series. With one exception, all of the large clinical series of patients with cluster headache precede the development of the IHS criteria. The most commonly used criteria for these series were the Ad hoc criteria,68 which were published in 1962.

Cluster headache is the only primary headache syndrome that is more common in men than in women; the male:female ratio is 6:1. In contrast to patients with migraine, a family history of migraine or other primary headache syndrome is no more common among patients with cluster headache than among individuals without headache. The pain is ocular in 80% of patients; the next most common locations are temporal, frontal, and maxillary. The pain is, by definition, strictly unilateral. Interestingly, in individual patients the headache commonly occurs on the same side.54,64 In the pooled data, 48% of patients experience their headaches exclusively on the right side, 38% exclusively on the left side, and only 14% of patients on both sides over their lifetimes. The onset of the headaches is rapid, and the most common duration of each headache is 30 to 60 minutes. This is an important distinguishing characteristic from migraine headaches, which nearly always last for at least 4 hours.

These headaches are called cluster headaches because of their tendency to cluster over the lifetime of each patient. In between each cluster, patients are completely asymptomatic. In 57% of patients, each cluster lasts from 1 to 2 months, and the most common cluster frequency is 1 to 2 per year. A patient with cluster headache often reports that during a cluster, each headache occurs at the same time of the day. Nocturnal headaches are most common and are reported by 54% of patients.

The character of the pain is variable and is less helpful to clinicians who seek to make a diagnosis of cluster headache. Tradition holds that cluster headaches are piercing or neuralgic. However the pooled data indicate that only 30% of patients describe their headaches in such a fashion; this is identical to the percentage of patients that report their cluster headaches as throbbing. The pain is excruciating and sufficiently distracting that most patients cannot continue their daily routine during an episode. Patients commonly pace the room and appear restless and agitated to observers.69 This behavior is in contrast to that of patients with migraine, who usually prefer to rest in a dark, quiet room.

Among ipsilateral autonomic symptoms, lacrimation is experienced by 76% of patients. Next most common are rhinorrhea and a partial Horner’s syndrome (miosis and ptosis). The partial Horner’s syndrome may persist after resolution of the headache. The differential diagnosis of cluster headache includes trigeminal neuralgia. Clinicians may distinguish trigeminal neuralgia from cluster headache by its briefer duration of seconds to several minutes, the absence of autonomic features, and the lack of periodicity.

Many patients with recurring primary headaches recognize that certain triggers often precipitate their headaches. Clinicians may be tempted to use these triggers as clues to particular headache diagnoses. Table 18-8 summarizes the published experience on the sensitivity of particular headache precipitants among patients with migraine and tension-type headaches. The only triggers that are significantly more frequent among patients with migraine are chocolate, cheese, and any food. Using pooled data from all series, the positive likelihood ratios for these findings are 7.1, 4.9, and 3.6, respectively. Other precipitants that are present in at least 30% of patients with migraine include stress, alcohol, weather change, menses, missing a meal, lack of sleep, and perfume or odors. None of these triggers is significantly more common among patients with migraine than those with tension-type headache. These data, however, are limited by the relatively small numbers of patients with tension-type headache for whom there are reports of trigger frequency.

The finding that menstrual variation of headache is no more common among patients with migraine than with tension-type headache is particularly surprising but is based on observations of only 221 women with tension-type headache. Most clinicians find this clinical feature to be useful in suggesting a diagnosis of migraine. Migraines occur most commonly in the several days before the onset of menses.24,79

Caffeine withdrawal may precipitate headaches both in patients with and in those without a history of primary headaches, but they occur more commonly in patients with a background of primary headaches.80–82 No data suggest that the development of caffeine-withdrawal headaches distinguishes patients with migraine from those with tension-type headache.

A recent study compared the frequency of precipitating factors for 366 patients with IHS-diagnosed migraine with that of 169 nonmigraineurs.77 Presumably the nonmigraineurs suffered from tension-type headaches, but they were not required to meet the IHS definition for tension-type headache. Fatigue, stress, certain food or drinks, menstruation, and weather were all significantly more common triggers for patients with migraine than for those without migraine. The stronger association of these factors with migraine than in the pooled data of Table 18-8 may reflect a more heterogeneous population of patients in the nonmigraine category that included patients who would not meet the IHS definition for tension-type headache.

The most common triggers for cluster headache are alcohol and stress. These occur in 32% and 30% of patients, respectively (see Table 18-7). These values are similar to those reported for tension-type and migraine headache and are not helpful in differential diagnosis. From a review of these data, one can conclude that the decision not to include headache triggers in the IHS criteria was appropriate. Obtaining a careful history of headache triggers is most useful to guide subsequent advice to patients related to lifestyle changes that may decrease headache frequency. With the exception of chocolate, cheese, and any food, this information does not make a particular headache diagnosis more or less likely.

Nausea, photophobia, phonophobia, and exacerbation by physical activity confer the highest positive likelihood ratio for the diagnosis of migraine when compared with tension-type headache. Nausea, exacerbation by physical activity, photophobia, and throbbing headache confer the highest negative likelihood ratio, and their absence makes the diagnosis of migraine significantly less likely. Headache duration is not a useful distinguishing feature.

Among the one third of patients with migraine who experience an aura, visual symptoms are most common, occurring in 84% of patients. Positive visual phenomena, including zigzags and stars or flashes, are more common than negative visual phenomena. Sensory, aphasic, and motor auras are each less common and nearly always occur in conjunction with a visual aura, rather than in isolation.

Headache precipitants are helpful when counseling individual patients about nonpharmacologic approaches to headache management. Only chocolate, cheese, and any food significantly increase the odds that a particular headache is migraine. Other commonly reported triggers, including menses, stress, alcohol, missing a meal, fatigue, and weather change, are equally common among patients with tension-type and migraine headache. Alcohol and stress are the most common triggers of cluster headache but are not helpful in establishing this diagnosis.

Cluster headache is a distinct headache syndrome that is easy to distinguish from the other two primary headache syndromes. It is strictly unilateral, periorbital, lasts most commonly less than 2 hours, occurs primarily in men, and is associated with ipsilateral lacrimation, rhinorrhea, ptosis, and miosis. The most common cluster duration is 1 to 2 months, and affected patients have symptom-free intervals that average 6 to 9 months.

The use of the IHS criteria for diagnosing patients with suspected primary headache syndromes ensures homogeneous populations of patients for inclusion in studies. Practicing clinicians, however, may find these criteria too complicated and restrictive. This chapter identifies those clinical features that are most useful to suggest or exclude particular primary headache diagnoses.