Impact of headache disorders in Italy and the public-health and policy implications: a population-based study within the Eurolight Project

The World Health Organization identified headache disorders as a public health priority in 2000 [1, 2]. On a clinical level it has long been evident that primary headache disorders, migraine and tension-type headache (TTH) in particular, are often lifelong illnesses which, as well as directly causing pain and disability, hinder family and social relationships and impoverish quality of life. Epidemiologically, a large number of studies provide evidence that headache among populations is ubiquitous and common [3], although many published studies prior to the last decade focused on prevalence to the exclusion of burden. The Global Campaign against Headache, conducted by Lifting The Burden (LTB) in official relations with the World Health Organization [4–6], set out to fill this knowledge gap as its first priority, developing standardised methodology [7]. Population-based studies in China [8], India [9] and Russia [10] in time for inclusion in the Global Burden of Disease study 2010 (GBD2010) [11] saw migraine recognised as the seventh cause of disability worldwide, with a mean estimated global prevalence of 14.7 % among adults [11, 12]. More recently, the Global Burden of Disease Study 2013 (GBD2013), with better information still, reported migraine as the sixth highest cause of disability worldwide [13]. Importantly, GBD2013 included MOH for the first time, and it entered the top twenty causes of disability. TTH was considered relatively non-disabling but, collectively, headache disorders ranked third [14].

In Europe, meanwhile, although many studies had reported migraine prevalence, there were still knowledge gaps regarding TTH and MOH, and for all headache types there were few published reports of burden. The Eurolight Project commenced in May 2007, using modified cluster-sampling in 10 countries of Europe to assess the impact of headache [15]. Multiple scientific and lay organisations collaborated with headache experts in these countries, and not all were able to draw population-based samples. The survey used the same structured questionnaire in each country, translated into the local languages, with diagnostic questions based on ICHD-II [16] taken from the HARDSHIP questionnaire [17]. Eurolight gathered analysable data from 8271 adult participants, with participation-rates between 10.6 and 58.8 % and detectable moderate interest-bias. Gender-adjusted 1-year prevalence of any headache was 78.6 %, of migraine 35.3 %, of TTH 38.2 %, of headache on ≥15 days/month 7.2 %, of probable MOH 3.1 % [18]. Personal impact was high, with a gradient of probable MOH > migraine > TTH, and generally higher among females than males. Particularly notable were the proportions of lost useful time: 17.7 % of males and 28.0 % of females with migraine lost >10 % of days; 44.7 % of males and 53.7 % of females with probable MOH lost >20 %. The conclusions of Eurolight were: “The common headache disorders have very high personal impact in the EU, with important implications for health policy” [18].

The situation in Italy is that, Eurolight apart, few data are available from the scientific literature regarding the general population. Moreover, most published studies have limitations: they were restricted to segments of the general population (the elderly [19, 20] or adolescents [21]) or to selected types of headache (eg, cluster headache [22]); or they were carried out before the publication of ICHD-I [23]; or they did not assess the impact of headache. Recently, however, two surveys of the adult general population of Parma based on door-to-door interviews reported a 1-year prevalence of definite migraine of 24.7 % [24] and of TTH of 19.4 % [25]. The finding for migraine is in line with that of Eurolight (22.2 %) for Europe generally [18], and indicative of a major public-health concern with substantial implications for health-care policy, planning and resource-allocation. Here we present the findings of Eurolight for the Italian population in order to add to the knowledge on which that policy might be based in Italy.

We received 500 questionnaires (participation rate 14.3 %), of which 487 were complete and usable for analysis (males 203 [41.7 %], females 284 [58.3 %]; mean age 43.4 ± 12.6 years). All age groups were similarly represented except those aged 25–29 (underrepresented by about 45 %) and those aged 40–44 (overrepresented by about 50 %). Almost all participants (98.0 %) were native Italian speakers. A large majority (92.1 %) were living with a partner and two thirds (67.9 %) were employed.

We need first to consider the limitations of this study, because there were several. The Eurolight Project itself “was a very large and organizationally complex study, involving multiple collaborating partners (academic and lay) in ten countries [making] pragmatic methodological compromises in order to complete it” [18]. It had considerable strengths: a questionnaire already used and validated by LTB in many different countries, cultures and translations (the HARDSHIP questionnaire [17]); neutral screening questions expected to lead to better ascertainment than questions incorporating degrees of frequency or severity [7]; diagnoses made according to a standard algorithm also developed and widely used by LTB [17]. Nevertheless, Eurolight could not undertake diagnostic validation in each of the translations used, so diagnostic accuracy was not directly assessed. This was one limitation here. The modified cluster-sampling approach of Eurolight, using different sampling methods in different countries, was seen as a strength of the main project [18], since it was not a primary purpose of Eurolight to generate representative data at national level. For this, its sampling methods were less suited, so this was a second limitation here. The third key limitation was the non-participation rate of 85.7 %.

With regard to diagnosis, we believe the instrument was reliable in view of its successful use elsewhere [17], and we draw attention as evidence of this to the expected gender differential in migraine but not TTH. Nonetheless, we are circumspect about making comparisons between migraine and TTH, which were not the main purpose of this report. Headache on ≥15 days/month was diagnosed (as a group of disorders) simply on the basis of reported frequency, and, among these, pMOH on the basis additionally of reported medication use. We see very clear differentials in impact between the episodic headaches and pMOH. The key messages of this report relate to headache disorders.

For the Eurolight project we sampled from one area of Italy: Pavia province in Lombardy region, Northern Italy, an area of industry, high population density and (relative to Italy as a whole) high gross domestic product (GDP). It is not representative of the country. Nevertheless, sampling, carefully stratified, was from the general population across the area, urban and rural; unless there are cultural, environmental or genetic factors varying across Italy that greatly influence the prevalence or expression of headache disorders, the findings from this area should at least be indicative for Italy as a whole. The sample, mostly of employed people living with partners, was good for investigating the impact of headache on the different aspects of life.

Much more troublesome is the high non-participation rate — not, of course, unusual in a postal survey. There was clear evidence of interest-bias, as might be expected; the non-participant survey recorded prevalences that were less than half those reported by the primary responders. A special reason for interest bias might be that, in Pavia, we have had a very active headache centre for over 30 years; many who participated may have been followed by the centre in the past, and responded because they were sensitive to the issue. However, it would not be correct to assume that this smaller sample, responding to a much shorter questionnaire, provided more reliable data.

With these limitations of our own study in mind, we review other available data. In Italy, as we noted, the epidemiology of headache disorders is not well established, and data on impact are very few. One study only has been performed on a sample representative of the general population [31], but this predated and therefore could not use accepted diagnostic criteria. It involved the small independent Republic of San Marino, and administered a questionnaire face-to-face to a random sample of 1500 subjects over 7 years of age summoned by mail. Of the 76.3 % who agreed to be interviewed, 528 (46.1 %) reported headache during the previous year. One-year prevalence of migraine was estimated at 9.3 % in males and 18 % in females. More recent but less population-based was the sample of 71,588 patients interviewed by 902 general practitioners (GPs) [32]. All patients who visited the GPs for any reason during five consecutive days of two different weeks were asked whether they suffered from headache; if they did, the GP completed a diagnostic questionnaire based on ICHD-I [23]. The estimated prevalence of migraine in this sample was 11.6 % (5.0 % in males and 15.8 % in females).

Other epidemiological studies in Italy have focused on selected groups. Prencipe et al [19], surveyed 1147 people aged ≥65 years in three rural villages in central Italy, 72.6 % of whom completed the protocol, including clinical evaluation by a neurologist and diagnosis according to ICHD-I [23] (with minor modifications in order to classify patients with headache on ≥15 days/month, then referred to as “chronic daily headache” [CDH]). Estimated 1-year prevalence of migraine was 11.0 %, of TTH 44.5 % and of CDH 4.4 % (with one third of these being pMOH). The similar study of 1031 residents aged ≥65 years of a rural village in southern Italy found that 225 (21.8 %) suffered from recurrent headaches (defined as ≥3 attacks within the past 12 months) [20]. Estimated 1-year prevalence of migraine was 4.6 %, of TTH 16 % and of pMOH 1.3 %. Both migraine and TTH were significantly more prevalent among females than males; only migraine prevalence significantly decreased with increasing age. Raieli et al surveyed 1445 students aged 11–14 years, diagnosing migraine by questionnaire and neurological examination and using the criteria of ICHD-I [23]. Prevalence of migraine without aura was 2.35 %, of migraine with aura 0.62 %.

More recently, Ferrante and colleagues [24] collaborated with GPs to survey the prevalence and clinical features of primary headaches in an adult population of the Emilia-Romagna region in Northern Italy. They evaluated 904 subjects using face-to-face interviews by a specialist of the Headache Centre of Parma. The 1-year adjusted prevalence of definite migraine (including chronic migraine) was 24.7 % and of probable migraine 5.1 %.

The emerging picture is far from clear. Ferrante’s study indicates a prevalence of all-migraine of 29.8 %, about one third lower than our gender-adjusted 42.9 % but considerably higher than the 19.8 % of our non-participant sample. While the mean global prevalence of migraine was estimated at 14.7 % in GBD2010 [11], several other LTB studies estimating the 1-year prevalence of migraine have found values in the range 20–26 %: in Russia 20.8 % [10]; in India 25.6 % [9]; in Zambia 22.9 % [33]; in Pakistan 22.9 % [34]. Ferrante’s finding is not unreasonable in the light of these, and we would say it is closer to the truth than ours, since we must recognise in ours the effect of interest bias. Eurolight was more focused on impact than prevalence, and it is insights into impact that this report is able to add.

As regards TTH, Ferrante et al [25] reported a crude prevalence of 19.4 %, similar to the 21.2 % reported by Vukovic in nearby Croatia [35], while our reported prevalence of 28.6 % is closer to the 30.9 % observed by Ayzenberg et al in Russia [10]. Our finding (as, indeed, that of Ayzenberg et al, who used a similar questionnaire) was subject to systematic underestimation since we diagnosed only one headache type (the most bothersome) in any participant with more than one. The ample variability in TTH prevalence that can be found in the literature (from 5.1 to 74.0 % [36–38], is probably dependent on a multiplicity of reasons, but more on variable methodological approaches and cultural differences in reporting the often mild and infrequent headache of TTH than true differences between populations due to geographical, demographic, racial, economic and psychosocial factors. This consideration underscores the need of national data for the correct implementation of adequate health care policies.

We saw that more than half (56.4 %) of those reporting headache in the last year were experiencing one or more headache episodes per week. One participant in eight (12.3 %) reported headache on ≥10 days in the last month. The frequencies reported generated an expectation that 60 of the sample of 487 would have headache on any one day. In fact, 63 reported headache yesterday, strongly supporting the responses based on recall (and validating the enquiry).

More than two thirds of our sample were employed. Each person with headache lost, on average, 2.3 days from paid work in the preceding 3 months, extrapolating to 9.2 days/year or 4.0 % of total working time. Spread between the entire sample of 487, this would represent a huge 3.1 % loss to GDP. Much of this was attributable to the highly-disabled small minority – the 20 people (4.1 % of the sample) losing >10 days and, particularly, the eight of these (1.6 %) losing >30 days. Even so, we suspect this finding was also influenced by interest bias; since losses at individual level were greater from migraine than from TTH, we suggest a conservative downrating of the estimate by one third, in line with the prevalence estimate of Ferrante et al [24] rather than our own. The revised loss to GDP of 2.0 % still exceeds estimates made by LTB for Russia (1.75 % [10]), China (1.9 % [8]) and Zambia (1.9 %) [33]). It should be noted that losses from household work were equally great, and these are also economically important.

People with difficulties at work were not helped by employers or colleagues who offered no understanding (in 50 reported cases), perhaps unsurprisingly when they bore the burden of the headache sufferer’s absenteeism. At home, 31 felt a lack of understanding. Nonetheless, most people with headache were not aware of negative consequences on education, career and professional prospects, family life or quality of life, although in each of these domains there was a disaffected minority. The consequences for the 27 (most with migraine) who believed they had done less well at school, or given up early, were potentially cumulative through a lifetime. So were they for the 30 who believed their headaches had hindered their careers, and the 12 who believed their headaches had limited their incomes. Again, most had migraine. Still the great majority of people with headache – at least those with episodic headache – seemed to get on with their lives. The two quality-of-life questions showed no differences between those with no headache, those with migraine and those with TTH, although those with pMOH scored well below all others on both questions.

Despite that 76.6 % of our sample reported headache within the last year, only 16.6 % of these had a doctor’s diagnosis and, in the last year, only 9.7 % had consulted a primary-care physician and 4.8 % a headache specialist. Here, interest bias is likely to mean that these numbers are higher than the norm (ie, a better picture than reality). There are many reasons for consultation-failure, including patient inaction. Torelli et al [39] assessed disease perception among a sample of 904 subjects; of 387 (123 males, 264 females) who reported headache in the past year, only one third perceived themselves as “headache sufferers”, this perception being dependent on headache frequency in females only. But, whatever its cause, the end-result of health-care failure, so evident in our study with only 2.4 % taking preventative medication despite that more than half the sample reported ≥4 headache days/month, is unrelieved burden, including symptom burden, disability, productive time losses and the inordinate losses to GDP.

Eurolight is not, generally, a source of reliable epidemiological data [18], a point we emphasise. The data from Italy were population-based, but from a selected area of the industrial north; furthermore, there was low participation. Eurolight’s strength was its detailed enquiry into impact. Taking other prevalence data into account, a fair picture can be painted of headache in Italy. As in other countries, headache disorders are highly prevalent and burdensome. In Italy, the high impact on individuals is matched by high economic burden; both are unmitigated by health-care failure. There is a highly-disabled small minority. Health policy in Italy must take note of these findings, of suggestions that Italy fares worse than other countries, and of the World Health Organization’s advice that effective treatment of headache is desirable not only for its health benefits but also because it is likely to be cost-saving [40].

The Eurolight Project was completed under the guidance of the Eurolight Steering Committee: C Andrée (project leader), Centre of Public Health Research, CRP-Santé, Strassen, Luxembourg; JM Laínez, Department of Neurology, Hospital Clinico Universitario, University of Valencia, Valencia, Spain; Z Katsarava, Department of Neurology, University of Essen, Essen, Germany; C Lampl, Department of Neurology and Pain Medicine, Konvent hospital BarmherzigeBrüder, Linz, Austria; E Ruiz de la Torre, Asociacion Española de Pacientes con Cefalea, Valencia, Spain; TJ Steiner, Department of Neuroscience, Norwegian University of Science and Technology, Trondheim, Norway, and Division of Brain Sciences, Imperial College London, London, UK; LJ Stovner, Department of Neuroscience, Norwegian University of Science and Technology, Trondheim, Norway, and Norwegian National Headache Centre, St Olavs University Hospital, Trondheim, Norway; P Vriezen, Nederlandse Vereniging van Hoofdpijnpatiënten, Amersfoort, The Netherlands. The Eurolight Project was funded by the European Agency for Health and Consumers (EAHC), Fonds National Luxembourg (FNR), with additional support from Lifting The Burden. Financial support for the collection of Italian data was provided by the C. Mondino National Neurological Institute of Pavia, Italy.

The data reported here were collected in collaboration with the University Consortium for Adaptive Disorders and Head pain (UCADH) and with Alleanza Cefalalgici - CIRNA Foundation Onlus.

The Authors are grateful to Jessica Barré for help in the statistical analysis of data.