Lars Jacob Stovner MD PhD, Norwegian National Headache Centre, St. Olav's University Hospital, 7006 Trondheim, Norway (tel.: +47 73 86 84 16; fax: +47 73 86 75 81; e-mail: firstname.lastname@example.org).
The present review of epidemiologic studies on migraine and headache in Europe is part of a larger initiative by the European Brain Council to estimate the costs incurred because of brain disorders. Summarizing the data on 1-year prevalence, the proportion of adults in Europe reporting headache was 51%, migraine 14%, and ‘chronic headache’ (i.e. ≥15 days/month or ‘daily’) 4%. Generally, migraine, and to a lesser degree headache, are most prevalent during the most productive years of adulthood, from age 20 to 50 years. Several European studies document the negative influence of headache disorders on the quality of life, and health-economic studies indicate that 15% of adults were absent from work during the last year because of headache. Very few studies have been performed in Eastern Europe, and there are also surprisingly little data on tension-type headache from any country. Although the methodology and the quality of the published studies vary considerably, making direct comparisons between different countries difficult, the present review clearly demonstrates that headache disorders are extremely prevalent and have a vast impact on public health. The data collected should be used as arguments to increase resources to headache research and care for headache patients all over the continent.
There is an increasing awareness in many countries that headache constitutes a major public health problem. In a World Health Organization (WHO) report of 2000, migraine is ranked 19 among disorders causing years lived with a disability for both sexes (rank 9 for women), and the disability during a severe migraine attack is considered on level with that of active psychosis, tetraplegia and dementia . Recognizing the scale of the global problem, WHO, in collaboration with three major international headache non-governmental organizations, has launched a global campaign to reduce the burden of headache [Lifting the Burden; 2].
In spite of the ubiquity of migraine and the enormity of its impact on public health, the resources allocated to research and patient care is probably not proportional. There may be a tendency to belittle the seriousness of the disorder for many reasons. One is that almost everyone suffers from headache from time to time, and it is therefore not considered a medical problem. Further reasons may be that headache disability is not visible in public, and that headache normally does not reduce life expectancy.
Quality of life (QoL) measurements show that migraine patients’ perception of their physical functioning and health is similar to that of patients with angina and diabetes. Migraine patients have, however, lower scores for role functioning, social functioning, pain and mental health, and their energy score is also remarkably low . Although migraine is among the more severe headache types, other forms may be as incapacitating as migraine. In addition to personal suffering, absenteeism from work and reduced effectiveness have huge socioeconomical implications, as headache and migraine are highly prevalent disorders in the general population. The prevalence of these disorders have been well documented in several epidemiological studies and an increased awareness among decision makers is needed in order to make the right priorities and allocate resources for treatment and research within this field. Epidemiological studies are also important for identifying risk factors, co-morbid conditions and monitor trends in disease prevalence over time.
The present review of headache epidemiology in Europe is part of a larger initiative to delineate the cost incurred by all disorders of the brain, called the European Brain Council (EBC). As resources to research programmes are now increasingly determined on an all-European basis, it is hoped that this initiative will increase funding to research on all brain disorders, and stimulate co-operative research across borders. As Europe is so diverse climatically, culturally and economically, a comparison of disease prevalence and incidence rates, with trends over time, could lead to useful hypotheses on risk factors. A first step in this direction is to review the current epidemiology on headache in Europe to get an overview of the situation.
At the outset, our main aim was to review studies on headache epidemiology in Europe in order to estimate the prevalence of the disorder, and then to identify epidemiologic studies dealing with the consequences of headache with regard to perceived QoL and in economic terms. In the course of this study, we encountered a plethora of methodological problems that made it difficult to compare the different studies, and it seemed important to discuss some of these for improving the quality and comparability of later studies on headache epidemiology.
In all epidemiologic studies, it is necessary to define who has a certain diagnosis (is a case) and who has not. It was therefore a great advantage for both headache research and clinical practice when the Headache Classification Committee of the International Headache Society (IHS)  provided the first relatively specific and unequivocal definitions of various headaches. This classification has later been incorporated into the International Classification of Diseases (ICD-10) . The IHS classification was revised in 2004 , but almost all headache epidemiologic studies since 1988 have been based on case definitions according to the first IHS classification.
In the present survey, we included epidemiologic studies that have appeared after 1988 on migraine (ICD-10 diagnosis G43) and tension-type headache (TTH) (G44.2), the two types that affect the great majority of headache patients. For migraine, we did not distinguish between migraine with (G43.1) and without (G43.0) aura, as the differentiation may be difficult in epidemiologic studies and the two types are probably not very different with regard to the patients’ suffering and subsequent economic consequences. In order to include all headache patients, we also considered epidemiologic studies that have investigated headache in general. The term ‘headache’ is not, however, defined in the IHS classification, and therefore we included studies on headache prevalence that appeared before 1988.
For TTH, the term ‘chronic’ has been applied to patients who have this type of headache for ≥15 days/month on average for ≥3 months . In many headache studies, a similar definition has been given to patients with headache, irrespective of whether it is of the tension type or not. We gathered data on ‘chronic headache’ (i.e. ≥15 days/month or ‘daily’ headache) to assess the prevalence of headache in these patients who are probably most incapacitated by their condition.
For our purpose, we included only studies performed on the whole population or a representative sample of the whole population within a certain age range in a community, town or country. Accordingly, we did not include studies based on selected populations (clinic-based, in workplaces, among university students, etc.). As primary schooling is obligatory in all European countries, studies on headache in children and youth based on schools were included.
For many patients, headache is troublesome only in certain phases of life. For this reason, most headache epidemiologic studies have questioned the subjects on headache within a limited time span, usually the previous year. The 1-year prevalence figures indicate the proportion of the population having an active disease, which is more relevant than lifetime prevalence for health economic calculations. Data on lifetime prevalence are also considered less reliable because of recall problems. Lifetime prevalence estimates were presented for those studies that also reported 1-year prevalence rates, and some additional studies with a 3-month or unspecified time frame were included.
Literature search and data extraction
A comprehensive literature search identifying population-based studies of headache and migraine according to the IHS criteria was conducted. Empirical studies in English were identified using Medline. The search was conducted using the expressions ‘migraine epidemiology’, ‘headache epidemiology’, and ‘migraine prevalence’ or ‘headache prevalence’ for each European country. References in relevant publications were also examined. From this, we also identified studies on QoL and functioning in relation to work and social life among headache patients.
The information extracted included the country where the study was conducted, year of publication, population characteristics and the prevalence estimates for headache, migraine, TTH and chronic headache, both overall and for each gender, and for various age categories. We also extracted data on headache frequency (headache days per month or year), absenteeism from work and effect on well-being and functioning, wherever available.
Figures 1–4 show maps of the European countries with prevalence rates for headache (Fig. 1), migraine (Fig. 2), TTH (Fig. 3) and chronic headache (Fig. 4), including both 1-year and lifetime prevalences for different age groups.
The one-year prevalence rates for headache and migraine displayed in Table 1 show that in adults, the overall (i.e. including males and females) headache prevalence rates range between 29 and 77%. For men, the rates vary between 19 and 69%, and for women between 40 and 83%. The mean value for all studies that include both sexes and all age groups from ≤25 to ≥60 years (six studies, Table 6) was 51% (men 41%, women 58%). For migraine, the overall values vary between 9.6 and 24.6% (Table 1), and the mean value of eight studies was 13.7% (Table 6). In men, the corresponding figures are from 2.7 to 13% with a mean of 7.5% and in women 6.9 to 25.0% with a mean of 16.6%.
Table 1. One-year prevalence (%) of headache and migraine
Table 6. Summary data of headache epidemiologic studies on adults (including both sexes and at least age groups 25–60 years)
Number of studies
Prevalence (% of population)
The lifetime prevalence rates for headache and summary data are shown in Tables 2 and 6 respectively. The figures vary between 35 and 96% (mean of two studies including only adults: 94.2%). For migraine, the corresponding figures vary from 12 to 27.5% (mean of six studies: 18.5%). In children and youth, and in the elderly, all these figures are generally somewhat lower both for 1-year and lifetime prevalences (Tables 1 and 2). Table 3 displays other prevalence rates that have been reported in studies with a 3-month or unknown time frame, the figures being similar to those of 1-year prevalences.
Table 2. Lifetime prevalence (%) of headache and migraine
In Table 4, the different prevalence studies on TTH are summarized. There is only one study showing the 1-year prevalence in a wide age range of adults (overall prevalence 74%), but two studies among the youth (mean overall prevalence 20.5%,) and two studies of the elderly (overall mean: 30.0%). As to lifetime prevalence, there are three studies covering a wide age range, the overall mean being 50.3%.
Table 5 shows the data on chronic headache. In the studies covering a wide age range and both sexes, the overall figures for 1-year prevalences vary between 2.4 and 4.5% (mean of two studies: 3.5%; Table 6). In men, the figures are between 1.7 and 2.1% (mean 1.9%), and in women between 2.8 and 6.8% (mean 4.8%). One study of the elderly showed similar figures (men 2.5%, women 6%, overall 4.4%), but another of youths showed markedly lower figures (boys 0.2%, girls 0.8%, overall 0.5%). Studies from Norway and Spain relating chronic headache to chronic analgesic overuse (data not shown) indicate that this combination may occur in as much as 1% of adults [7–9].
In order to get an overview of the prevalence of headache and migraine at different age groups, the mean prevalence for each age category was calculated, including all studies that provided age-specific prevalences. In order to fit the data into standardized 10-year age categories, the data for some studies had to be recalculated by calculating the average of the lower and higher age category. In Figs 5 and 6, the age distributions of migraine and headache prevalence are shown for each 10-year category from the first to the eighth decade for each gender. For migraine, the prevalence is somewhat higher among boys under the age of 10 years, but after this age higher rates are seen among women in all age groups. The prevalence increases gradually and peaks during middle age for both men and women and declines thereafter. The decline in headache prevalence with advancing age was similar, but there was little increase from youth to adulthood, and there was a smaller difference between the genders compared with migraine.
As to the frequency of headache among headache sufferers, there are several studies, but the way the frequencies have been reported differs widely, making comparisons between studies almost impossible. Only a few studies give exact figures; in a German study , it was found that migraineurs had an arithmetic average of 34 days/year of headache, and TTH patients 35 days/year. The figure for migraineurs is probably somewhat higher than that found in a Swedish study reporting attack frequency (16 attacks per year, mean duration of attack: 19 h) . Somewhat higher figures for mean attack frequency in migraineurs were found in England (26.3/year in men and 23.6/year in women) .
Regarding the consequences of headache, some studies give data on the proportion of headache sufferers in whom headache causes absence from work or school. In two studies, one from Finland in 1979  and one from San Marino in 1986 , 7% of working individuals had been absent from work in the previous year because of headache. In the Copenhagen study in 1992 , 43% of migraineurs (5% of the population) and 12% of TTH patients (9% of the population) had been absent from work during the previous year because of headache, i.e. a total of 14% of the population. In this study, the TTH patients who had been absent seem to be as much or more absent than the migraine patients, and the number of workdays lost per year because of migraine and TTH was estimated to be 270 per 1000 persons and 820 per 1000 persons, respectively, a sum of 1090 days per 1000 persons. In a study from England in 2003 , 15% had been absent from work or had reduced ability to work because of headaches in the last 3 months. Expressed in yearly terms, headache accounted for 1327 missed and 5213 reduced-ability days per 1000 workers per year, representing 0.5 and 2.0%, respectively, of all working days in the adult population, irrespective of the headache status. This study does not relate absenteeism to different headache diagnoses. In a Swedish study in 2004 , 65% of migraineurs reported some degree of absence from either school or work during the previous year; 54% of them had 1–2 days of absence per year, but as many as 11% reported absence from work or school once or more per month. Among migraineurs in England, an estimated 5.7 workdays/year was missed by those working or attending to school . This seems to be higher than in France where a diary-based registration of absenteeism showed that migraineurs were away from work 2.18 days/year because of headache .
Several epidemiologic studies have investigated the effects on well-being and functioning. Significant limitations have also been found in those suffering from headache, when compared with the general population [8,12,15,19,20,21] and with other chronic diseases . In a recent study, the QoL was found to be lowest among those with chronic headache, intermediate among migraineurs and highest among subjects with other forms of episodic headache , and in a Dutch study, there was an inverse relationship between headache frequency and headache-related QoL (HRQoL) . Headache frequency may have a greater impact than headache intensity on HRQoL . Chronic headache with medication overuse was associated with a decrease in all QoL aspects studied with SF-36, most marked for role physical and bodily pain . In addition to the disability experienced during a migraine attack, many migraine patients also report an impairment between attacks .
The total burden of headache patients may not only be related to headache per se, but also to comorbid conditions. European population-based studies have demonstrated that depression and/or anxiety occur two to three times more often among migraineurs than in the general population [24,25]. This comorbidity may be important for non-migrainous headache as well . In addition, it has been found that headache is also comorbid with other bodily pain, both in children  and adults .
From the maps (Fig. 1–4) it is evident that data on headache epidemiology from large parts of Europe, most notably from Russia and all countries within the former Soviet Union, and also from most of the other countries in Eastern Europe (except Hungary and Croatia) are lacking.
In the countries where epidemiologic studies on headache have been performed, there are a number of studies on the prevalence of headache and migraine (Tables 1, 2 and 3), but disappointingly scant data on TTH (Table 4) which is the most frequent type. Some studies have also focused on chronic headache (Table 5). Although this is not a proper diagnosis, this category is important as it designates the patient group carrying the heaviest burden of suffering and the highest societal costs per patient.
The summary data given in Table 6 confirm that headache is a highly prevalent disorder, affecting approximately 50% of the adult population within a year. Around 14% have migraine, and chronic headache is present in as much as 4% of the population. The accuracy of these figures for any particular country can, however, be questioned. From the tables and the maps, it can be seen that there are large variations in the prevalence of the various headache types among different countries. It would be of considerable interest if these variations in prevalence reflected real differences, but as there are also large differences within one country, they may to a large extent be due to variations in methodology.
Although it lies outside the scope of the present study, it could be of great scientific interest to compare prevalence data from Europe with those of other continents. This has been done in one meta-analysis, indicating that the prevalence of headache and migraine in Europe is somewhat lower than that in North America but higher than in Asia and Africa .
Burden of headache
The main burden of headache is carried by a minority of sufferers, with approximately 4% of the population having headache half of the days or more per month (Table 6, Fig. 4). On average, patients with migraine or TTH are affected with headaches approximately for a month every year ; 7 to 15% of the population will have reduced ability to work or be absent from work because of headache, and the number of days missed per year in migraineurs seems to vary between 2 and 6 [12,18]. Two relatively recent and population-based studies indicate 14–15% absenteeism because of headache in Denmark and England [15,16], and the number of days (1090 and 1327 days per 1000 persons/year) missed from work was not very different in these two studies. The number of days with reduced efficacy at work have not been entered in these calculations, and this number may be five times higher , and result in even more loss of work efficacy than the days missed.
As to the patients’ suffering, there is ample documentation of a considerable reduction in QoL experienced by headache patients, migraineurs in particular. The data indicate that those with most frequent headache and most work-related disability also have the highest reduction in HRQoL. The fact that many migraineurs report an impairment in QoL also between attacks may be an intrinsic feature of the disease, or it may be due to comorbid conditions such as anxiety and depression or other bodily pain.
Among the many methodological differences, variations in the specified time frame for the headache (3 months, 1 year, lifetime, etc.) is obviously very important, and in general, the figures on lifetime prevalences are higher than those on 1-year prevalences (Table 6). The differences do not seem to be so large between the 1-year and 3-month prevalence rates, as demonstrated by the fact that figures in Table 3 (3 months or unspecified) are similar to those in Table 1 (1 year). As headache prevalence data are dependent on patients’ recall, it is probable that headache during the last few months are most readily remembered, which would tend to make the 3-month and 1-year prevalences rather similar.
It may also be of importance to learn how the information was obtained (personal interview, telephone, questionnaire), but in the present review, there seems to be no substantial differences in reported prevalence rates between studies using questionnaires or interviews. Previous reports have stated that questionnaire-based studies are less suitable and that interviews should be considered the gold standard . If other methods are used, it is essential to perform a validation study of the method by comparing the diagnoses made in the study with the diagnoses made in a subsample using the gold standard method (interview and examination by a neurologist). Only one study has been performed with the gold standard method , and relatively few studies carried out using other methods have been validated by this method [32–36].
In many studies, some sort of screening question has been used before the subjects were asked about features of the various headache subtypes. That this can introduce some misclassification is demonstrated in a Dutch study  in which a diagnostic telephone interview of both screen positive and screen negative sub-samples of the general population was conducted. They found that the prevalence of migraine among screen-positive individuals was 6.5% among men and 18% among women, with an overall prevalence of about 12%. The prevalence increased after including screen negative individuals to 7.5 and 25% among men and women, with an overall prevalence of about 16%.
Of utmost importance is the way the IHS criteria are interpreted or applied . This is illustrated by two French studies [38,39] performed 10 years apart. In both studies only those reporting that they ‘suffered from headache’ without specifying the time frame were included, and those that reported suffering from headache every day were excluded. The overall prevalence of migraine (IHS categories 1.1 and 1.2) in the two studies were very similar (8.1%  and 7.9% ). The difference between the two studies was with respect to the prevalence of migrainous disorder. In their first study, using a relatively restrictive definition of borderline migraine, there was a prevalence of 4% , whereas it was 9.1% in the second study, where, according to the IHS they required at least three of four criteria for migrainous disorder (IHS category 1.7) . This illustrates that the estimated prevalence of migraine is sensitive to small changes in the diagnostic algorithm used for classification. The overall prevalence of migraine (IHS categories 1.1, 1.2 and 1.7) was 12% in their first study  and 17% in their last study . In the German study , the lifetime prevalence of those fulfilling migraine criteria was 11.3%, while the prevalence of those fulfilling all criteria but one (1.7) was 16.2%, resulting in an overall prevalence of 28%.
Moreover, in studies reporting only the prevalence of headache, it is important to consider the exact phrasing of the screening question. Higher prevalences are generally found in answer to a neutral question (e.g. ‘do you have headache?’) than to questions involving some specification of the degree of headache in the sense of suffering or frequency (e.g. ‘do you suffer from headache?’‘do you have intense headache?’, ‘do you have repeated headaches?’, etc). In studies reporting the 1-year prevalence of headache with a screening question specifying some headache degree, the prevalence rates vary between 29 and 49% among adults [40–43]. In those with higher figures [11,44], the exact way of questioning has not been given. This may also explain the great variations in overall lifetime prevalence rates of headache in adults, among which only the German study , reporting 71% prevalence, had a screening question including some degree of suffering, whereas the rest with more than 90% prevalence rates had neutral questions [16,31,45].
Two studies from Italy reported the prevalence of headache among elderly subjects >64 years of age. In one of these studies, using the screening question ‘have you had a headache in the previous year?’, a headache prevalence of 51% was found , while it was only 22% in the other study which included only those reporting at least three headache attacks . The latter is in accordance with the prevalence rates among elderly >60 years of age in the Norwegian study of 25%, who were asked whether they ‘suffered from headache’ . A problem with some studies, particularly the older ones, is that the exact phrasing, and the way criteria have been applied, are often not clearly described.
As evidenced by Figs 5 and 6 relating headache and migraine prevalence to age, it is obvious that differences in age composition between population samples may explain differences in headache prevalences. Low participation rates among elderly would therefore tend to make population prevalences higher, particularly for migraine.
Some studies are limited to younger age groups. Among adolescents the 1-year prevalence of ‘had any headache’ was found to be 77% in a study from Norway, while ‘recurrent headaches’ were only reported by 29% . The latter figure is similar to that found in a study from Italy reporting a 1-year prevalence of recurrent headaches of 23.9% . In a study from UK, 66% reported having ‘had headache over the past year’, but this study also included children below 10 years of age . The study from Sweden also included children and adolescents and reported an overall 1-year prevalence of ‘experienced headache during or prior to the last year’ of 44.8% . A German study used a different approach, asking the participants ‘Have you had any pain during the last three months?’. They were subsequently asked to indicate the body region, and 66% reported headache during the last 3 months .
Suggestions for future studies
If future headache epidemiologic studies are to be comparable, it seems from the considerations above, that they should be performed according to a relatively uniform standard with regard to the time frame (e.g. 1-year prevalence), age groups (e.g. decades for studies of persons above 20 years of age, 5-year groups below), and reporting headache frequency (number of days may be easier to interpret than number of attacks per month or year). As to the intensity of headache, a scale similar to that used in drug trials on headache would seem appropriate for many purposes . It may be particularly difficult to decide which headache screening question to use. With a neutral question most of the population will be headache sufferers, and additional data on frequency, intensity and disability will then be needed to define groups with a clinically and economically important headache problem. If some degree of suffering (e.g. intensity or frequency) is mentioned in the screening question, lower prevalence figures will be found. The degree of headache mentioned will probably be somewhat arbitrary, however, and it may also be a disadvantage that no further headache data are obtained on the screening-negative patients. Theoretically, a neutral question with additional data to define patient groups of interest would seem to be optimal. Data obtained from questionnaire-based studies should be validated on a randomly chosen sub-sample interviewed and examined by a neurologist, a method which is considered the gold standard for making headache diagnoses. If accurate headache diagnoses are the main goal, all data should be obtained by the gold standard method.
In spite of the discrepancies among different studies, the overall picture resulting from these evaluations on the prevalence and burden of headache clearly confirms that headache is a major health issue. On the basis of the studies already performed, a tentative estimate of the costs of the headache disorders in Europe has already been published by the present initiative . Such studies evaluate both the direct (use of healthcare system and medication) and indirect costs (work absenteeism and reduction in work efficacy), in addition to the intangible costs (the personal sufferings).
As can be seen from the figures, a large part of Europe (particularly in the eastern part) is still a ‘terra incognita’ as to headache epidemiology. Hence, there is a huge challenge for researchers in the headache field to fill in gaps of knowledge. Rational use of limited public resources is highly dependent on valid information on the prevalence and impact on health and well-being and the costs of various disorders. For headache, it is believed that correct treatment can mitigate the burdens, whereas incorrect treatment both adds to these burdens and wastes resources . For example, chronic headache related to overuse of analgesics and other acute medications is probably a big problem in most countries [7–9]. Making epidemiological surveys of headache prevalence and drug use followed by information campaigns to obtain more appropriate medication for headache may be first steps to improve headache care.
In addition to procure data on prevalence and incidence, which are of value for those who allocate resources, epidemiological studies may be used to identify causes and risk factors for disorders which may lead to better treatment and prevention. A powerful means to obtain this goal would be to perform studies with an identical methodology in various parts of Europe, where possible risk factors are assessed along with headache prevalence. Such studies would require an organization for co-ordinated planning and implementation, and should be a high-priority goal for the appropriate institutions.
This paper was prepared in the framework of the European Brain Council (EBC; http://www.ebc-eurobrain.net) initiative ‘Cost of Disorders of the Brain in Europe’ (CDBE steering committee: Jes Olesen, Bengt Jönsson, Hans-Ulrich Wittchen, Patrik Andlin-Sobocki), supported by an unrestricted educational grant from H. Lundbeck A/S.
Authorship to main document: P. Andlin-Sobocki, B. Jönsson, J. Olesen, H.-U. Wittchen and the EBC task force.1