Headache yesterday in Europe

The World Health Organization (WHO) has acknowledged headache disorders as of global public-health importance [1]. This is good for people with headache, who carry most of its burden but who find that it receives little priority in the queue for health care [1]. Headache disorders are under-diagnosed and mostly undertreated [1, 2], not because diagnosis is particularly difficult or because effective treatments do not exist but because of widespread failure of health services to recognize the need for health care for headache and to take steps to deliver it [1, 2]. In these circumstances, the recent Global Burden of Disease Survey 2010 (GBD2010) [3] found tension-type headache (TTH) and migraine to be the second and third most prevalent disorders in the world, and migraine the seventh highest specific cause of disability [4].

In fact there are large gaps in our knowledge of the prevalence and burden of headache worldwide [5, 6]. In Europe, knowledge is most complete in western countries, yet focuses strongly on migraine; rather few studies have addressed the more prevalent TTH or the more disabling headache disorders characterized by headache on ≥15 days/month. Countries to the east, particularly those of the former USSR, have until recently been badly neglected.

The Eurolight project was a collaboration of 25 partners from 15 countries in Europe [7] supported by the EC European Agency for Health and Consumers. Eurolight’s main objective was to bolster the political argument for better investment in health-care services for people with headache throughout Europe. For this purpose, it sought to adduce evidence of the impact of migraine, TTH and headache on ≥15 days/month across the EU, and achieved this by conducting surveys in ten countries. It added estimates of impact to those of prevalence [8], thereby filling an important part of the knowledge gap in Europe, and showed, among other large burdens, the enormous financial costs (over €100 billion annually) imposed by headache disorders [9].

Burden-of-headache studies generally estimate 1-year prevalence and enquire into quantifiable components of burden (for example, symptom burden, disability, time loss, financial loss, impact on quality of life) through questionnaires that depend upon recall over a preceding period (commonly 3 months). It is not known how reliable recall is for such matters, but at best it is likely to be variable and at worst highly misleading [10]. Recent studies initiated by the Global Campaign against Headache [5] in Russia [11], China [12, 13], India [14], Pakistan [15] and Zambia (unpublished) introduced enquiry into “headache yesterday”, and Eurolight followed this lead. In such an enquiry, numbers responding positively are obviously reduced: the prevalence of headache yesterday is substantially lower than the 1-year prevalence, given that headache-affected days vary in frequency between individuals with headache from one to 365 per year. On the other hand, responses to enquiries about headache yesterday are not subject to recall bias.

The full methodology [7, 16] and principal results [8] of Eurolight are published elsewhere. Here are presented the results on headache yesterday.

Eurolight was essentially an EU-wide study employing a modified and pragmatic form of cluster sampling to reach a large number of participants in many countries. Questionnaire-based surveys were conducted in 10 countries: Austria, France, Germany, Ireland, Italy, Lithuania, Luxembourg, Netherlands, Spain and United Kingdom (UK).

In nine countries, 8,271 participants not derived from specific headache-patient populations (ie, excluding the patient-organization samples from the Netherlands, Spain and Ireland), participated in the survey. A majority (58%) were female. For each country, the number of participants and their gender are shown in Table 2. The overall participation rate in the study (excepting the Netherlands, where the denominator was indeterminable, and Austria, where it was not recorded) was 27% (details are reported elsewhere [7]).Altogether, 1,422 participants (17.2%) reported headache yesterday. The prevalences of headache yesterday in the nine countries were rather constant in the main study, ranging from 11% in Lithuania to 19% in Spain (gender-adjusted mean: 17%; SD: 0.4%), with the UK, which had very low numbers, appearing as an outlier (25%) (Figure 1). In the non-responder study, with 1,007 participants, there were 91 (9.0%) with headache yesterday.

Most participants (47%) reported a duration of headache yesterday of 1-4 hours, which might reflect effect of treatment; 24% reported 5-12 hours and 15% reported >12 hours. The estimate of the overall mean (taking the mid-point of each range) was 6 hours (Table 3). In all samples, most headache yesterday was reported as “not bad” (overall mean 44%) or “bad” (overall mean 44%); even so, headache yesterday was “very bad” in 11%.

Functional impairment yesterday because of headache showed that most people (72%) were adversely affected to some extent, while about a quarter (overall mean 27%) were severely affected (could do less than half of what they expected) and a minority (overall mean 7%) could do nothing (Table 3). Variation between countries was not enormous (Figure 2). Yesterday was a workday for 51% of those with headache yesterday. Absence rates from work were more variable but, again overall, 6% of those with headache yesterday (or 0.5% of all participants) lost the entire day. Of those who worked yesterday despite headache, 20% did less than half or nothing of what they expected, and only a small majority (55%, somewhat more than the 28% reporting normal function) responded that they actually did everything. Of those who did less, the majority (71%) claimed they would be able to make up for it later. The pattern was somewhat different for household chores, with fewer people (36%) achieving everything expected and more (11%) doing nothing, and for social activities (33% achieving everything, 24% doing nothing).

The prevalence of headache yesterday was much higher in women than in men (21% vs 12%). However, with regard to headache duration and intensity and the measures of functional impairments, men and women were quite similar (Table 3).

The importance of measuring the impact of headache yesterday, and the major strength of this study, lies in the freedom from recall bias. This form of bias potentially affects all studies retrospectively recording impact over a period of time, to an unknowable degree. We emphasise the distinction between recall error and recall bias. Recall of what happened yesterday may not be error-free, but the error is likely to be neither systematic nor large enough that it might lead to bias. There may be exaggeration, or mis-description in other ways, but these are not forms of recall bias, and they occur in any enquiry.

Although diagnostic questions were included in the methods, the responses to these are not analysed here. This is mostly because these questions were applied to the subjectively most bothersome headache reported in the last year, not to headache yesterday. Although we asked whether headache yesterday was of the same type as the most bothersome, and in over two thirds (69%) of cases it was, there was no diagnosis made in the remaining 31%. This is not important to our message, which is a public-health one: headache disorders, without focus on migraine, are a major cause of population ill-health, representing a need for which health services must make due provision.

A potential source of bias was headache today: participants with headache on the day they encountered the questionnaire might have deferred responding until they became headache-free, and then been highly likely to report headache yesterday. Headache today is as probable as headache yesterday; therefore, it might seem that this factor could double the reported prevalence of headache yesterday. In reality this was not so, because about half of the 1-day prevalence of headache is attributable to daily headache (the probability of episodic headache yesterday is <7% [ie, 2/30] of its long-term prevalence, assuming an average frequency of 2 days/month, whereas the probability of daily or near-daily headache being present yesterday is close to 100% of its prevalence). When headache occurs daily, it is of no consequence whether the enquiry is conducted today or tomorrow. The problem would not have occurred in Lithuania, where participants were visited without prior notice, and did not have the option of deferring their responses until tomorrow. The reported prevalence of headache yesterday was lowest in this country, but not outlying (Figure 1).

Low participation rates did inject bias. The purpose of the non-responder study was to detect participation bias, which, according to this study, led to an over-estimation of the prevalence of headache yesterday. However, the non-responder study had relatively few numbers, and a much less-detailed enquiry; the data from it were of lower quality than those from the main study. It would be quite inappropriate to substitute the former for the latter; rather, the former might be considered to demarcate minima. Responders in the main study, of whom 79% reported headache last year, were 27% of the surveyed sample (this being the participation rate); the non-responder study therefore represented the other 73%, of whom 64% (gender-adjusted) had headache last year [8]. The weighted average of 68% ([79*27 + 64*73]/100) establishes the minimum 1-year prevalence. We assumed that headache yesterday could occur only in those reporting headache last year, and adjusted on this basis. In the continuing discussion below, ranges are presented that reflect this.

An estimated 15-17% (the lower limit = 17*68/79) of adults aged 18-65 years in Europe had headache yesterday, which implies a similar percentage every day and signals a huge level of population ill-health. This range should be internally consistent with the approximate estimate derived from the 1-year prevalence (78.6% [8]) and average frequency (5.8 headache days/month [unpublished]) in the same population. These data generate a mean of 459 headache days per month per 100 of the population, from which it can easily be calculated that 15% of the population have headache on any day (13% when potential participation bias is considered). These figures are reassuringly similar, despite being subject to recall bias. Few other data exist on headache on any single day; there is one recent study from mainland China specifically on headache yesterday, reporting an age-adjusted prevalence of 4.8% in the general population aged 18-65 years [13]. This is a little less than one third of our finding, but the estimated 1-year prevalence of headache in China (23.8% [18]) is also about a third of that in Europe [14]. Our finding of 15-17% for headache yesterday in Europe is consistent with the point prevalences (16 and 19%) found in two carefully conducted Danish studies [19, 20].

The purpose of reporting headache yesterday was, however, to focus not on prevalence but on impact. Duration and intensity are dimensions of symptom burden; whilst they may be misleading because they are reported subject to the effects of any treatments taken, they give rise to the following estimation. If 15-17% of a population had headache yesterday of average duration 6 hours, then one quarter of these (say 4.0%) had headache at any particular moment (which compares with 1.8% reported in China [13]). Headache was bad or very bad in 56% of those with headache, or 2.2% of the population aged 18-65 years (1.4% in China [13]). Taking the population of the EU in the age range 18-65 years as 300 million (http://​europa.​eu/​about-eu/​facts-figures/​living/​), we calculate that, at this and any moment, between 6.3 (300*0.56*0.15/4) and 7.1 (300*0.56*0.17/4) million people are suffering in Europe with bad or very bad headache.

As telling are the secondary effects – in particular, functional impairment and its consequences. Thus, an estimated 0.4 (0.5*68/79) to 0.5% of working-age adults in Europe lost the whole of yesterday at work because of headache, again implying a similar percentage every (working) day. A further 1.2-1.6%, working despite headache yesterday, reported a significant level of inefficiency (less than half or nothing achieved). In line with the theory behind the MIDAS instrument [21] on which our lost-productivity questions were based, we counted these also as wholly lost days: the overestimate arising from this was offset by discounting days when >50% but <100% was achieved. Adding these generates a range-estimate of 1.8-2.1% as the total lost productivity at work due to headache yesterday (and every working day). This is not so very different from the 1.3% reported in China [13], despite the much lower prevalence there, reflecting a higher proportion in China reporting disablingly bad headache. Furthermore, in our study, 72% of participants working with impaired efficiency believed they could make up for it later. If so, a reduction by this factor should be applied to this element of lost time (now 0.3-0.4%), yielding a revised total of 0.7-0.9%. It is actually questionable whether “making up later” can achieve complete restitution, but clearly there must be some degree of offsetting. In other words, at least 0.7% of the available working capacity in Europe, and possibly 2-3 times this, is surrendered to headache, which opens a window upon a vista that employers and governments both should find alarming.

The evidently greater negative impact of functional impairment upon household chores and social activities (Table 3) reflects, almost certainly, the more optional nature of these in comparison with duties at work. They are more easily abandoned or reduced, which should not be to underrate their importance as a part of everyday life that headache takes away. Given the number of people with headache yesterday, losses of this sort must be acknowledged because their cumulative impact is very high. And there is more: for the 72% of affected participants who could make up later for impaired efficiency at work, something else would presumably be given up in order to do this. There would remain a possibly immeasurable but not necessarily intangible loss expressed elsewhere, and in all probability this would be of leisure time, already a serious casualty of headache. While lost working time has, in theory at least, a calculable financial consequence (either wages are lost or the employer pays for work not done), how leisure time is valued in monetary terms is a complex subject which we will not attempt to explore. It should nonetheless be noted that any value attached to lost leisure time must be added to that of lost work time when the cost of headache is calculated.

Our data show that the burden of headache is much larger in women than in men, mostly because women have headache more often. During headache, men and women appear to have similar symptom burdens and functional impairment.

Headache is common, and its impact is high. Assessment in nine countries in Europe based on headache yesterday, and therefore free from recall bias, shows that headache affects 15-17% of adults aged 18-65 years on any day. It removes at least 0.7% (and up to three times this) from workforce capacity. Leisure time and social activities are also serious casualties.