Case Definition

Migraine is a chronic disorder with episodic manifestations; it is characterized by attacks of head pain and various combinations of neurological, gastrointestinal, and autonomic changes. Although the IHS defines (p. 320 ) seven subtypes of migraine, by far, the two most important are migraine without aura and migraine with aura (see Table 13–1). Migraine is both a diagnosis of inclusion and exclusion. It is a diagnosis of inclusion because specific diagnostic features are required. It is a diagnosis of exclusion since secondary headache disorders have to be eliminated on the basis of the history, physical examination, or laboratory studies.

The gold standard for diagnosing migraine is assigned by a headache expert applying IHS criteria. In studies using videotaped patient encounters, the reliability of headache diagnosis is good to excellent (mean kappa = 0.74) but far from perfect (Granella et al. 1994). In epidemiology studies, a diagnosis of migraine is often assigned on the basis of a telephone interview or self-administered questionnaire. Using a clinical assessment as the gold standard, telephone interviews have sensitivities of approximately 85% and specificities of approximately 93% for migraine (Stewart et al. 1996a).

The migraine attack is often divided into four phases: the premonitory phase, the

(p. 321 ) aura, the headache phase, and the resolution phase. None of these phases is obligatory for diagnosis and most people with migraine do not have all four phases. The premonitory phase (prodrome) usually begins hours to days before headache onset. Symptoms are variable but may include changes in mood, hyperactivity, yawning, food cravings, and thirst (Silberstein et al. 1998). Premonitory features occur in about 60% of migraine sufferers.

In migraine with aura, focal neurological symptoms precede or accompany the attack. The aura develops over a period of 5–20 minutes and typically lasts less than 1 hour. Auras are usually visual and may include a mix of positive features (spots of light, zig-zag lines) and negative features (regions of visual loss). Auras may also involve other features, including motor, language, or brain stem disturbances. The aura is usually, but not always, followed by a headache. Approximately 20%–25% of migraine sufferers have migraine with aura. Most individuals who have migraine with aura also have attacks of migraine without aura (Johannes et al. 1995).

The third phase of the migraine attack is the headache phase. The headache of migraine is typically one-sided, throbbing, moderate to severe in intensity, and aggravated by routine physical activity. The headache phase may be accompanied by photophobia (sensitivity to light), phonophobia (sensitivity to sound), nausea, and vomiting, in variable combinations.

During the final phase, the resolution phase, the pain and accompanying symptoms subside. Many migraine sufferers report mood changes (euphoria, lethargy, fatigue) and scalp tenderness, even after spontaneous pain has subsided.

Incidence

Estimating the incidence of a chronic disorder with episodic manifestations is challenging. Since diagnostic criteria for migraine without aura require at least five lifetime attacks, should incidence be estimated using the time of the first or fifth attack? Further, the disease affects individuals of all ages and the incidence rate varies substantially by age. To accurately describe incidence requires a large cohort study of individuals across the life span. There have been few studies of migraine incidence and no studies that systematically ascertain new migraine cases across a broad range of ages. The ideal study design for estimating the incidence of migraine would be prospective in design, ascertaining prevalent migraine cases at baseline and newly incident migraine cases as the study proceeded. As yet, no study has achieved this ideal in an inception cohort across a broad age span, but one study of migraine incidence was conducted in an inception cohort of young adults.

Example 13–1 Breslau et al. (1996) conducted a prospective study in 1007 members of a health maintenance organization. The inception cohort ranged in age from 21 to 30 years. Most of the sample (972/1007) completed follow-up interviews 3.5 and 5.5 years after enrollment. The at-risk population was composed of 848 participants who did not meet the criteria for migraine at baseline. The 5.5-year cumulative incidence was 8.4% (71/848; 60 female cases; 11 male cases), for a rate of 17.0 per 1000 person-years (24.0/10² female, 6.0/10² male). Of interest is that the female/male ratio of incidence rates was 4:1 in this study of young adults.

Two other population-based studies estimated the incidence of migraine using the reported age of migraine onset and reconstructed cohort methods. Using this method, a prevalence sample of migraine sufferers is identified and age of onset is determined, often by self-report. Self-reported age of onset is used to estimate age-specific incidence as described below.

Example 13–2 In Washington County, Maryland, telephone interviews were conducted (p. 322 ) among 10,169 residents between the ages of 12 and 29 to identify 392 males and 1018 females with migraine (Stewart et al. 1991). In both males and females, the incidence rate of migraine with aura peaked 3 to 5 years earlier than migraine without aura (Fig. 13–1). In addition, the incidence of migraine in females peaked at a later age than in males. The phenomenon of telescoping, or the tendency to report the occurrence of events in the past at times closer to the present (Brown et al. 1985), complicates incidence estimates derived

Figure 13–1 Age- and gender-specific incidence of migraine headache with and without aura among 10,131 survey respondents 12–29 years of age, Washington County, Maryland, 1987. Upper curve (○), migraine without aura; lower curve (■), migraine with aura. (Reprinted with permission from Stewart WF, Linet M, Celentano D, et al. Age- and sex-specific incidence rates of migraine with and without aura. Am J Epidemiol 1991;134:1111–1120.)

using reconstructed cohort methods. Studies that estimate the age-specific incidence of migraine based on recall would likely be biased toward older ages of headache onset (Brown et al. 1985; Cummings et al. 1990). In the Washington County study, age-specific incidence rates were adjusted for the time lag between the reported age of onset and the age at interview. This study is still limited by the narrow age range of the participants.

Another population-based study, conducted by Rasmussen (1995), reported that the age-adjusted annual incidence of migraine was 3.7 per 1000 person-years (females 5.8/10²; males 1.6/10²). Neither age-specific incidence nor incidence by migraine subtypes was reported.

A third approach to estimating migraine incidence used medical records to ascertain cases. In the linked medical records system in Olmstead County, Minnesota, a review of 6400 patient records yielded 629 individuals who fulfilled criteria for migraine (Stang et al. 1992). The age-adjusted incidence rates were 1.37 per 1000 person-years for males and 2.94 per 1000 person-years for females. Only individuals who consulted a health-care provider for headache were included in this study, which may explain why these incidence rates are lower than in the previously discussed studies.

Prevalence

Risk Factors for Developing Migraine

Risk factors for migraine should be distinguished from risk factors that trigger the episodic attacks of headache within people who suffer from migraine (Olesen et al. 1993; Rasmussen 1995). Positive family history, female gender, epilepsy, depression, and low socioeconomic status are associated with developing chronic migraine (Rasmussen 1995; Breslau and Rasmussen 2001).

Over the last decade, some progress has been made in defining risk factors for the disease. Numerous studies have examined the socioeconomic profiles, comorbidities, and genetics of migraine (Olesen 1993; Lipton and Stewart 1997; Breslau and Rasmussen 2001). Migraine is emerging as a clinically and biologically heterogenous disorder, however, further advances are still needed to identify and quantify the contributions of factors that cause the development of migraine.

Genetic risk factors may be involved in migraine. Studies have found that various headache types and migraine tend to aggregate within families (Russell and Olesen 1995, 1996; Stewart et al. 1997; Russell et al. 1999). This has raised the question of whether this is due to environmental or genetic origins (or both). While an active search is underway to identify genes that cause migraine, no definitive results have been published.

Triggering Factors among Migraine Sufferers

For an individual who suffers from migraine headaches, certain factors may trigger an attack (see Table 13–2), including endogenous (e.g., menses) and exogenous factors (e.g., diet, fatigue, stress, and seasonal changes) (Lipton 2000; Breslau and Rasmussen 2001). Migraine sufferers are eager to understand what initiates their attack. Identifying the precipitating factors can lead to preventive strategies such as trigger avoidance, stress management, and short-term prophylaxis of menstrual migraine, and can enhance feelings of self-control (Olesen 1993; Lipton and Stewart 1997).

The most common method used to investigate trigger factors is patient surveys (Van den Bergh et al. 1987; Rasmussen 1993) in which migraine sufferers are asked to list their headache triggers and whether specific listed factors trigger their migraine attacks. In some cases, a clinician may conduct an N-of-1 study that helps suggest whether a factor triggers headaches or not. However, these studies are notorious for not providing clear evidence.

Example 13–4 In a study to determine the effect of Chinook weather conditions on the probability of developing a migraine (p. 326 )

Table 13–2 Risk Factors and Triggering Factors for Migraine

Risk factor or trigger	Examples
Food	Xanthine-containing foods (wine, chocolate, tea, coffee, colas) Tyramine-containing foods (cheese, bean pods) Preservatives (sulfites, nitrates) Pickled or fermented foods Flavor enhancers (aspartame, MSG) Alcohol
Environment	Weather Exposure to smoke Air pollution Noise Odors
Behaviors	Activity level Food consumption patterns (e.g., meal skipping)
Medications	Hormones Antihistamines Diurectics Vasodilators Bronchodilators Hypoglycemic agents Chemotherapy Sympathomimetic agents Overuse of pain or analgesic agents
Physiological and psychological factors	Stress Depression Sleep disturbances Hormone cycles

headache, 88% of the participants reported that the Chinook weather influenced their migraine (Cooke et al. 2000). However, using rigorous diary methods, Chinook weather precipitated migraine in only 29% of the participants. Thus, migraine sufferers may differentially recall only the attacks that support their belief that a particular putative factor triggered their migraine headache.

Randomized clinical trials are often conducted to examine exogenous triggers of migraine (Lipton 2000). Subjects are exposed to a placebo or to a trigger factor, such as aspartame, chocolate, or nitroglycerine (Olesen 1993; Van Den Eeden et al. 1994; Rasmusssen 1995). When only one trigger is tested, inferences about between-group differences in migraine occurrence with and without exposure to the factor may be possible. However, multiple exposures to the trigger factor for each patient is a better design to support inferences about individual patients. In any case, well-controlled trials that are blinded and adequately controlled are needed to assess these factors. Study designs such as crossover trials, if appropriate, or diary studies that have the participants blinded to the hypothesis can greatly enhance the believability of these putative associations.

Individual Impact-Frequency and Severity of Attacks

Migraine is a disabling disorder with an impact on both the individual and society. The individual impact of migraine is measured by quantifying the symptom profile and the frequency and severity of attacks, as well as through quality of life studies.

The individual impact of migraine has been evaluated in numerous studies (Stewart et al. 1992, 1994, 1996b; Lipton and (p. 327 ) Stewart 1993). In a population study of 1748 migraine sufferers identified by telephone interview, approximately 40% of subjects (female 39%; male 46%) reported severe pain, while an additional 40% reported very severe pain (female 43%; male 32%). The remainder of subjects reported pain that ranged from mild to moderate.

Most migraine sufferers reported a range of symptoms other than pain including photophobia (overall 82%; female 84%; male 77%) and phonophobia (overall 78%; female 79%; male 73%). More than half (59%) of migraine sufferers reported nausea that accompanies their migraine headache more than half of the time (female 62%; male 47%), while only a small proportion reported vomiting.

Most subjects reported one to two migraine attacks per month. The duration of an untreated attack varies considerably by gender. Among females, approximately 71% of attacks last longer than 24 hours. In contrast, 48% of males reported attacks that last longer than 24 hours. These results are comparable to the results of previous studies (Stewart et al. 1992, 1994; Lipton and Stewart 1993).

There have been at least three population-based studies of health related quality of life (HRQoL) in migraine cases versus controls (Terwindt et al. 1998; Lipton et al. 1999a; Steiner et al. 1999). In all of these studies, migraine sufferers had substantially lower HRQoL relative to population controls. In addition, HRQoL was inversely related to attack frequency and disability (Terwindt et al. 1998; Lipton et al. 1999a). One clinic-based study (Osterhaus et al. 1994) found that migraine sufferers have HRQoL scores similar to patients with osteoarthritis, hypertension, or diabetes. However, HRQoL measured in clinic-based samples may differ from that of migraine sufferers in the community. In addition, many studies conducted in clinic-based samples lack a contemporaneous control group derived from the same source population or sample (e.g., neurology clinic).

Economic Impact of Migraine Headache

The direct costs of migraine care may include outpatient visits, hospitalization, the use of emergency department services, the cost of prescriptions and diagnostic tests, as well as other treatments. For migraine, most of the direct costs are due to outpatient visits and the cost of prescription medications, rather than hospitalization (Stang and Osterhaus 1993; Clouse and Osterhaus 1994), or are due to the use of emergency department services (Celentano et al. 1992; Edmeads et al. 1993; Clouse and Osterhaus 1994).

Few studies have assessed medical care for migraine in economic terms. In a population sample, Hu et al. (1999) estimated the annual costs of migraine treatment. They found that the annual treatment costs for migraine were over $1 billion, about $100 per migraine sufferer per year. These are likely to be underestimates because the figures were derived from 1994 data, and consultation and medication use have increased substantially over recent years (Lipton et al. 1999b). Clouse and Osterhaus (1994) reported similar, though slightly higher, average claim costs per member per month ($145). Since migraine sufferers are likely to have comorbid conditions, the increased costs may reflect the costs due to the treatment of migraine as well as other medical conditions.

Lost productivity time is the major determinant of the indirect economic impact of migraine. When estimating indirect costs, it is important to account not only for lost workdays due to migraine (absenteeism) but also reduced productivity while at work. Some studies have asked patients to report absenteeism or reduced production over 1 year; however, the accuracy of recall over this period is uncertain (Rasmussen et al. 1992; Stewart et al. 1996b; Schwartz et al. 1997). For some purposes, it is desirable to combine lost workdays (absenteeism) and reduced effectiveness days into an overall measure, often termed lost workday equivalent (LWDE). The (p. 328 ) LWDE equals the total days of missed work plus the number of days at work with headache multiplied by the estimated decrease in percent effectiveness for working with headache (Stewart et al. 1996b).

Example 13–5 To circumvent limitations of prior studies, Von Korff et al. (1998) conducted a population-based diary study. Individuals recorded missed time from work and reduced productivity on a daily basis for 3 months. During this 3-month period, migraine sufferers missed an average of 1.1 days per month due to headache. Subjects who worked during a migraine attack reported work effectiveness that was reduced by an average of 41 %. The missed workday and LWDE estimates reported by Von Korff et al. (1998) are higher than those reported in other population-based studies (Osterhaus et al. 1992, Stang and Osterhaus 1993; van Roijen et al. 1995, Stewart et al. 1996b). The use of a daily diary may have improved the accuracy of reporting, which is inherently limited by asking people to recall events over the past 3 to 12 months. A relatively small percentage of migraine sufferers account for the majority of lost work time. In this study, the most disabled 20% of the participants accounted for 77% of the missed workdays; 40% of subjects accounted for 75% of the total LWDEs (Von Korff et al. 1998).

Recently, Hu et al. (1999) used the human capital method to estimate the indirect costs of migraine. These studies assume that the economic value of a missed day of work is equivalent to the lost wages for that day. According to the results of this population-based study, Hu et al. estimated that migraine costs American employers $13 billion per year because of absenteeism and reduced effectiveness at work. Approximately 62% of the cost was due to absenteeism. Importantly, the greatest indirect costs were found among middle-age (30–49) individuals. Although this study provided important information, the human capital method values the indirect costs of the disease according to the sufferers’ salary. A lost workday is valued by the individuals’ wages for that day, therefore the impact of the disease in a laborer or homemaker would be valued lower than that of a business executive with a high salary. A limitation of all studies reported to date is that reduced productivity while at work when suffering from a migraine is based on self-report. No studies to date have actually quantified reductions in productivity.

Case Definition

Tension-type headache (TTH) is diagnosed on the basis of a characteristic symptom profile and the exclusion of secondary headache (Table 13–3). In TTH, there is no prodrome or aura. The headache is usually bilateral, with pressing, tightening, or squeezing pain of mild to moderate intensity. Tension-type headaches are of two major forms, episodic tension-type headache (ETTH) and chronic tension-type headache (CTTH) (see Table 13–3). The major difference between these disorders is the frequency of headache attacks. If headaches occur less than 15 days per month, they are classified as ETTH, while those that occur 15 or more days per month are classified as CTTH. For ETTH, nausea excludes the diagnosis but one of either photophobia or phonophobia is permitted. For CTTH, any one of nausea, photophobia, or phonophobia may occur; if more than one of these features is present, the diagnosis is excluded.

Most subjects with ETTH (90%) report mild to moderate headache pain; attacks typically occur 3 times per month (Rasmussen et al. 1991; Gobel et al. 1994; Lavados and Tenhamm 1998; Schwartz et al. 1998). Chronic tension-type headache, by contrast, is typically associated with higher pain intensity and more frequent attacks. By definition, CTTH headache frequency ranges from 15 to 30 headaches per (p. 329 )

month (Gobel et al. 1994; Schwartz et al. 1998). In one study, 86% of subjects with CTTH reported moderate or severe pain (moderate, 44%; severe, 42%) (Gobel et al. 1994). Using a 10-point scale, Schwartz et al. (1998) found significantly higher pain intensity scores among CTTH subjects than among ETTH subjects (CTTH, 5.55; ETTH, 4.98; p < 0.001).

The clinical profile of TTH varies by gender. Bilateral pain occurs in most TTH sufferers, but occurs with even greater frequency among women than among men (Lavados and Tenhamm 1998). Throbbing pain occurs approximately equally in 66% of men and 57% of women; this feature, often viewed as a hallmark of migraine, poorly discriminates the two disorders. Pressing pain is also frequently reported. Many subjects with TTH report that pain is exacerbated with movement (men, 70%; women, 76%), which is surprising since pain that is exacerbated by movement is normally associated with migraine rather than TTH. Photophobia or phonophobia is common, especially in women. If both features are present simultaneously, the diagnosis of TTH is excluded; however, each feature occurs in isolation with surprising frequency.

(p. 330 ) Prevalence

Economic Impact of Tension-type Headache

The first population-based study to examine work loss data in ETTH was reported by Rasmussen et al. (1992) in Denmark. (p. 332 ) Among employed participants, 12% were absent from work at least once during the previous year because of ETTH. Of those who missed work, the majority (68%) were absent from 1 to 7 days during the previous year. Twenty-five percent (25%) were absent between 8 and 14 days during the year, and only 16% were absent more than 14 days during the previous year.

Schwartz et al. also measured the impact of headache in the workplace in a study conducted in Baltimore County, Maryland (Schwartz et al. 1997, 1998). Reduced ability to function and inability to function (actual missed work) were measured separately. Of the lost work time associated with headache, 19% of the missed workdays and 22% of the reduced-effectiveness days were specifically due to ETTH (Schwartz et al. 1997). Among subjects with ETTH, 8% reported missed workdays (absenteeism), while 44% reported reduced-effectiveness days at work due to headache. Among those with missed workdays, an average of 8.9 missed workdays were reported, while subjects with reduced-effectiveness days reported approximately 5.0 reduced-effectiveness days per person. Lavados and Tenhamm (1998) found higher levels of missed work among their sample of TTH sufferers: 25% of males and 39% of females reported missed work due to their headaches.

The proportions of subjects with CTTH and ETTH who report lost and reduced-effectiveness days were similar: 12% of CTTH sufferers reported lost workdays and 47% reported reduced-effectiveness days. Sufferers of CTTH reported more frequently lost workdays and reduced-effectiveness days in comparison with ETTH sufferers. Subjects with lost workdays reported an average of 27.4 lost workdays per person; subjects with reduced-effectiveness days reported approximately 20.4 reduced-effectiveness days per person.