Classification of daily and near-daily headaches

Utilizing the expertise of groups of headache specialists, the International Headache Society (IHS) developed and published a new headache classification system in 1988. ^[1] This system included explicit inclusion and exclusion criteria for a broad range of headache disorders, but did not consider their natural history. Rather, classification was based on cross-sectional features of individual attacks, such as attack duration, pain location, and associated symptoms. Several studies have assessed the IHS headache classification system, variously focusing on reliability, validity, comprehensiveness, and generalizability. ^[2-9]

Some studies conducted in headache subspecialty centers showed that the IHS system was not comprehensive. ^[2-9] Many patients with frequent headaches--sometimes referred to as chronic daily headache (CDH)--could not be appropriately classified within the IHS system. ^[4-9] When they could be classified, these headaches were most often placed in the chronic tension-type headache (CTTH) group. Because these daily headaches often evolved from episodic migraine and had migrainous features, it seemed inappropriate to classify them as tension-type headaches (TTH). In addition, many of these patients had too many migrainous features to meet the criteria for CTTH, yet they could not be classified as migraine. ^[4]

Many headache specialists diagnosed these frequent headaches as transformed migraine (TM), a condition that, like CDH, is not defined within the IHS system. Originally described by Mathew et al., ^[10] TM is a clinical syndrome with a characteristic natural history. Patients with TM have a history of episodic migraine, often beginning in their teens or twenties. As the process of migraine transformation progresses, headaches increase in frequency but decrease in severity, and associated migrainous features (photophobia, phonophobia, and nausea) become less severe and less frequent than during typical migraine. Headaches that resemble TTH occur daily, or almost every day, and superimposed on these background headaches may be occasional full-blown migraine attacks. Other features of migraine may persist, including menstrual aggravation, identifiable trigger factors, and unilateral pain. ^[3,10-14] CDH evolving from migraine is to some a synonym for TM, but it encompasses other disorders. The older term, "mixed or combined headache," includes patients with this disorder. ^[3,4]

We recommended revising the IHS criteria for chronic, frequent primary headache disorders and proposed adding several headache types to the current IHS classification. ^[15] Our major intention was to make the IHS criteria more comprehensive by providing a biologically plausible niche for TM. Our initial recommendations are summarized in Table 1, with numbers that indicate the proposed place for each disorder within the IHS structure. We originally subdivided daily headache into TM, CTTH evolved from episodic tension-type headache (ETTH), new daily persistent headache (NDPH), and hemicrania continua (HC). The IHS does not currently include TM, NDPH, and HC in its classification.

We proposed operational criteria for TM 1994 Table 2, including it as a subset of migraine (1.8). ^[15] Its diagnosis depended on a history of IHS-defined migraine and the presence of head pain lasting more than 4 hours a day at least 15 days a month. We elected not to require particular characteristics for the daily or near-daily headaches, in part because these headaches are pleiomorphic; daily headaches may be unilateral or bilateral, mild to severe in intensity, with or without associated migrainous features. We originally required a history of transformation, that is, a period when migraine headaches increased in frequency while the prominence of associated migrainous features decreased. Patients with TM often continue to have episodic superimposed bouts of full-blown migraine. Some patients find that their migraine headaches disappear completely. ^[2,8] For this reason, we did not originally utilize the continuing occurrence of superimposed migraine attacks as part of our definition. We subdivided TM into two categories, one with and one without medication overuse, using a consensus of published reports to define medication overuse. ^[16-21] We recommended field-testing our proposed revisions to the IHS criteria to determine whether they served the intended purposes.

The objectives of the present study were to prospectively (1) assess (in a headache subspecialty center) the comprehensiveness of the IHS criteria with and without the addition of our new criteria for TM for the classification of frequent primary headaches; (2) determine whether subjects can provide the information necessary to assign diagnoses using our revised criteria; (3) identify subjects who cannot be classified; and (4) recommend modifications, if necessary, based on the results of the field tests.

Methods.

We screened 150 consecutive outpatients with more than 15 headache days a month who consulted at a headache subspecialty center. Only patients who did not have an organic cause for their headaches and whose headaches lasted more than 4 hours on average were included. Herein we refer to headaches meeting these frequency and duration criteria as daily headaches. An experienced clinician conducted a semistructured interview designed to ascertain which headache features are essential for diagnosis.

The interview began by eliciting specific demographic information. It went on to assess the patients' lifetime headache history, including headache patterns before and during escalation (if it occurred) and current headache patterns. The interview focused on the frequency, duration, pain characteristics, and associated features of both the near-daily and the superimposed headache types (if present).

The analysis focused on individual symptoms and particular headache diagnoses. We reasoned that if a significant proportion of patients could not remember if they had a particular feature, then a case definition should not depend on the presence or absence of that feature. We also examined headache classification by applying diagnostic algorithms based on the published IHS criteria and our proposed 1994 criteria. We reviewed the symptom profiles of patients who could not be classified and proposed additional revisions intended to make the classification more comprehensive. For purposes of this assessment, we divided the IHS definition of migraine into three components. The first component reflects attack frequency and duration. The second considers features of pain. The third considers associated features. The final component of the IHS definition, exclusion of secondary headaches, was not applicable as patients with secondary headaches were excluded.

Results.

Our final sample consisted of 150 patients with a mean age of 40.7 +/- 15 years. The gender ratio of women-to-men was 2.7:1 (73% were women). Most of the subjects (91%) were Caucasian. Before developing a daily headache, 55% of the patients had migraine without aura, 15% had migraine with aura, and 63% had ETTH. These diagnoses were assigned at the time of interview based on the patients' recollection of their prior episodic headaches. The time from the onset of the daily headache to our evaluation averaged 36 months.

By definition, all patients had more than 15 headache days a month with the actual mean headache frequency being 28.9 +/- 3 days a month. The mean daily headache duration was 18.4 +/- 8 hours a day. The longest mean pain-free interval was 11.7 +/- 17 hours (last month). Table 3 summarizes the prevalence of selected migraine symptoms associated with daily headaches (for the entire sample) and with superimposed headaches for the patients who identified them. Although migraine symptoms were often associated with the daily headaches, they were far more common with the superimposed headaches. Additionally, 99 patients reported a history of headache escalation, and 83 did not meet our proposed criteria for TM (see Table 2) of escalation over at least a 3-month period, often because they could not recall details of the escalation process. Twelve patients did not know if they had escalated.

We attempted to classify these patients using the current IHS criteria and our proposed revisions Table 4. Using the IHS criteria, 43% of patients could not be classified. Subjects could not be classified as IHS migraine in 87% because the headaches were essentially continuous, in 65% because they did not have sufficient associated features, and in 22.1% because of a missing pain factor. Subjects could not be classified as CTTH in 45% because they were missing a pain feature and in 15.3% because they had too many associated features (see Table 4).

When we used our 1994 criteria for TM (see Table 4) we still could not classify 25% of patients; 61 (40.7%) were not able to provide a history of migraine, whereas 83 (55%) did not meet our criteria for escalation. Of these patients, 11 (13%) could not report how long ago escalation began, and 12 (14%) did not know if they had escalated. If the time requirement of 3 months of escalation was dropped, 10 more subjects met the criteria for TM (see Table 4). Based on these data, we developed our 1995 criteria for TM Table 5. We reasoned that the requirements of diagnosable migraine in the past and a history of escalation over 3 months imposed unreasonable burdens, given the limitations of patient recall. Accordingly, we made our definition less restrictive in hopes of increasing the proportion of patients who could be classified. We provided three alternative means for association with migraine: (1) history of IHS migraine or (2) a period of escalation (these were both required in the 1994 criteria) or (3) superimposed attacks of IHS migraine except for duration. The 1995 criteria are also hierarchical. The diagnosis of TM precludes the diagnosis of either migraine or CTTH. With these revisions, we were able to classify all our patients. The number of patients with TM increased dramatically. The patients not classified as TM or CTTH met the criteria for NDPH or HC and are included in the "Other" group in Table 4.

Discussion.

Our results confirm that the IHS criteria are not comprehensive; we were unable to classify 43% of daily headache sufferers. This result is concordant with several prior studies conducted at subspecialty centers. ^[1-8] For example, Messinger et al. ^[6] were unable to classify 35.9% of patients. Overall only 9.1% had CTTH, but about 86% of these CTTH patients had two or more migrainous features. Solomon et al. ^[4] could not classify 33% of 100 consecutive CDH patients as CTTH because they had too many migrainous features. Many of these headaches would be classifiable as migraine except for their prolonged duration (IHS migraine headache attacks last 4 to 72 hours).

Using our proposed 1994 revisions to the IHS criteria, the proportion of patients whose headaches could be classified increased. We were able to classify an additional 27 subjects. In applying these criteria, several difficulties became apparent. We were still not able to classify 25% of patients. Some patients had difficulty in remembering the characteristics of their prior headaches, whether their headaches had escalated, when they had escalated, or how long the process of escalation took. We therefore explored several modifications of our proposed 1994 criteria. We modified the definition of TM to include subjects with a history of IHS migraine, a history of escalation over 3 months, or a current headache that met the IHS criteria for migraine except for duration. This allowed us to use both historical and current features of the headache, which we believe are crucial to the diagnosis. Using our new 1995 TM criteria, 15.3% had CTTH, 78% had TM, none had NDPH, and 6.7% had none of these disorders. To avoid more than one diagnosis for a single headache type, we imposed hierarchical diagnostic rules; patients could not be diagnosed with CTTH if they met the criteria for TM.

Providing alternative links to migraine (requiring one of three features) makes our 1995 definition less restrictive. We no longer absolutely require a period of transformation, not because we believe that it is unnecessary but because it was too difficult to ascertain. This change would be expected to increase the sensitivity of the criteria at the price of specificity. In the absence of an external gold standard (a diagnostic test like a gene or blood level), we cannot empirically assess the sensitivity/specificity trade-off.

Others have proposed revisions of the IHS criteria. Manzoni et al. ^[22] define "chronic migraine" as migraine headache that occurs for at least 6 days a week for at least 1 year. This definition excludes patients whose headache occurs >15 days a month but less than 6 days a week on average. Their second category, "migraine with interparoxysmal headache," requires a headache >15 days a month for 1 year. In addition, it requires that both recurrent attacks of migraine and headache at onset were migraine without aura. Our concept of TM encompasses both "chronic migraine" and "migraine with interparoxysmal headache."

We arbitrarily defined all types of CDH as daily or near-daily headache lasting more than 4 hours a day for more than 15 days a month. We used the 15-day-a-month figure to be consistent with the IHS definition of CTTH. We picked 4 hours a day to differentiate CDH from cluster. Most of our patients' headaches were constant. Whether patients with shorter duration and less frequent headaches are biologically different from patients with constant headache is uncertain at this time. It may represent a transition from episodic to daily headache. This is open to empiric verification.

The focus of our current effort has been to make the IHS criteria more comprehensive by providing a place for patients with TM. At the same time, we believe the revision improves validity because of the close relationship between migraine and TM. Ultimately, we need to actually study the reliability and validity of our nosologic entities. Additionally, we need to explore the biological connection between TM and migraine based on natural history, biological markers, genetics, and treatment response.

We hope that the new criteria for TM, and the modification of the criteria for CTTH, will make the IHS criteria more useful for physicians. The 1995 criteria require additional testing in an independent sample. We recognize that we are engaging in an iterative process and expect additional modifications to our proposal as the field advances.