Abstract
In connection with the Vågå study of headache epidemiology, a search was made for caffeine-withdrawal headache in 1741 parishioners. Female to male ratio 1.05; mean age 44.2 years (range 18-65 years). A face-to-face interview technique was used. The mean consumption of coffee was 4.7 cups a day. Males on an average consumed more coffee (5.1 ± 3.3 cups/day) than females (4.4 ± 3.1 cups/day). Neither in those with a high consumption of coffee: ≥10 cups a day (n = 134), nor in those with a considerable variation in consumption: ≥10 cups/day (n = 31) did there seem to be a definite increase in headache resembling caffeine-withdrawal headache, for instance during weekends. In seven parishioners, however, there did seem to be such a headache, and in two of them, the evidence was rather convincing. This headache generally seemed to be mild and global and occurred mainly in the morning hours on weekends. There was no nausea, no throbbing quality of the pain, and no reported use of analgesics. Coffee seemed to abate the headache. This frequency (0.4%) should, clearly, be regarded as a minimum figure. Caffeine-withdrawal headache at the grassroots level may be a rather rare, generally vague, symptom-poor headache.
Bach: Kaffee-Kantate ‘Sweigt stille, plaudert nicht’ BWV 211
The father is concerned about the daughter's (Lieschen’s) consumption of coffee and reproaches her for her vice. Lieschen gives vent to her indignation:
‘Ei! wie schmeckt der Coffee süsse,
Lieblicher als tausend Küsse,
Milder als Muskatenwein.
Coffee, Coffee muss ich haben;
Und wenn jemand mich will laben,
Ach, so schenkt mir Coffee ein!′
(‘des Tages – dreimal mein Schälchen Coffee trinken darf …’)
Alas! the sweet taste of coffee
Is lovelier than a thousand kisses
And milder than muscatel wine.
I simply must have coffee,
and to put me in a good frame of mind,
just pour me out some coffee !
(… to drink my three cups of coffee daily …)
(Archiv; stereo 2533269)
In addition to the sacred cantatas, Bach composed around 50 secular (‘profane’) cantatas, one of the famous being the Kaffee-Kantate (BWV 211). This was first performed in Leipzig, probably in 1734. Coffee-houses were ‘allowed’ in Leipzig from the end of the 17th century.
Introduction
Coffee was introduced into Europe in the 17th century; thus the first coffee houses were opened in Oxford c. 1650. Scepticism towards this ‘brown poison’ was much more than marginal, as briefly illustrated in the ‘vignette’. Later, coffee became a widely used and socially acceptable, mild stimulant.
While the consumption may create various negative effects, its withdrawal has also been studied. Caffeine-withdrawal headache was probably first properly described in 1940/1943 (1, 2). It was readily accepted as a well defined headache, where the cause was established and also a sort of therapy was at hand. It appears in the textbooks (3–5), and it has a separate section in the IHS headache classification system (6). Everything therefore seems to be well beyond doubt. What is there then to discuss?
Two different circumstances have made us somewhat sceptical of this concept:
A small inquiry within the circle of headache specialists has given a fairly uniform picture: In spite of extensive practice over many years, the specialists had seen only a few cases where this diagnosis was suspected; one specialist even reported zero cases. Also the present first author (O.S.) has only seen 3 or 4 cases over the years with a putative diagnosis of caffeine-withdrawal headache.
In a session at the European Headache Federation meeting in Istanbul (2002), caffeine-withdrawal headache was discussed (7). Inquiry was made among the audience: Did they believe in this type of headache? What was the amount of caffeine necessary for its withdrawal to be harmful? etc. There was apparently no unanimity in the response to these questions. This is remarkable. In spite of an official, universal acceptance, some scepticism seems to linger among practitioners in the field of headache.
What can be the reason for this apparent, partial disbelief?
Life is complex and the daily influence of possible internal and external headache-generating factors is legion. How regular are the meals (blood sugar level?); what is the level of the mental tone (tendency to a depressive mood (8))?; is there a slight tendency to ‘tension’ (‘tension-type headache’)?; how is the social adaptation (human interrelationship, social stress)?; there is also the anxiety combined with waiting and ‘compulsory quietude’, as emphasized by Lundberg (9); can the neck be absolutely ruled out (tendency to cervicogenic headache)? and in countries with a high, average alcohol intake, the influence of alcohol enters into the picture (can a certain ‘hangover’ component be excluded or alcohol-induced headache)?. In addition, the following can be added: influence of other food components (dietary headache?); are there disturbances of sleep (where coffee also may play a causal role)?; possible influence of allergy; possible influence of weather conditions (barogenic headache (9)). All of these components, and perhaps others, may, in an isolated case, make it most difficult to sort out the role of a single component as the cause of headache.
Perhaps the most important question of all in this context is: What about the grade of the complaint? The term contains the word ‘headache’. Is it really a ‘headache’, or is it a heaviness or a (minimal) unpleasantness? We are, of course, fully aware that a ‘severe’ brand of headache has been described after coffee withdrawal (1, 2, 8), but the crucial question remains: by which standard?
Another problem will also be studied. Norwegians are known as a coffee-drinking people. Even among Norwegians, Vågå has been considered a ‘parish of coffee-drinkers’ (10), possibly making the allusion in the ‘Kaffe-Kantate’ (‘three cups a day, see ‘vignette’) seem a trifle. If so, Vågå could be particularly suitable for this type of study. Are the inhabitants of Vågå in a particular position in this respect?
Aims of the study
The principal aim was to search for evidence for caffeine-withdrawal headache in Vågå. In doing so, particular attentions will be paid to those with a highly varying consumption of coffee, since theoretically the chances of finding such cases would probably be best within this group. The, admittedly scarce, IHS rules (6) can be interpreted to argue for this eventuality. Is a ‘tension-type-headache picture’ found more frequently than otherwise, when combined with heavy coffee consumption and in particular among those with a considerable variation in consumption? or: the other way around? A secondary aim was to try, if possible, to outline such headache at the grassroots level. A tertiary aim was to approximate the average coffee consumption in Vågå. This study was entirely retrospective. The conclusions from this study therefore have to be drawn with caution.
Vågå study of headache epidemiology; a brief outline
The Vågå study was carried out in a two-year period (1995–97) in the parish of Vågå in the mountainous part of southern Norway, approximately midway between Trondheim and Oslo. A detailed description of nature, population, and design has been given previously (11). Just prior to the initiation of the study, Vågå had 3907 inhabitants (11). The target group was the accessible 18–65-year-old parishioners (n = 2075), of whom 1838 (88.6%) could be personally interviewed and examined by the principal investigator (O.S) in a semistructured interview. The parishioner was not allowed to see the questionnaire. A direct, face-to-face interview technique was adhered to throughout the study.
Materials and methods
A total of 1741 parishioners (94.7% of those interviewed in connection with the Vågå study) were questioned regarding coffee consumption. As regards those who were not questioned (n = 97), this was mainly a matter of available time during the clinical interview. There was no other selection policy. There was a female/male ratio of 1.05 among those examined, a figure identical to that in the entire study population. The mean age among the coffee-drinkers was 44.2 years.
A thorough examination of head, face, and neck was invariably carried out, including the cranial nerves. A full scale neurological examination was carried out when deemed necessary.
The quantity coffee consumed
Only the spontaneous consumption and its consequences for headache have been studied. No interventions as regards coffee usage have, accordingly, been carried out. However, in the entirely exceptional case, ‘spontaneously arranged experiments’ came to our notice.
The quantity consumed has been estimated in a rough fashion by questioning individual parishioners. This methodology has some shortcomings. The ordinary ‘measure’ in this area seems to be a teaspoon, corresponding to approximately 5 ml (but with a convex surface) of instant coffee for each cup of coffee. This roughly corresponds to 2 g of coffee. A cup in this area may contain c. 140–200 ml, most frequently probably around 150 ml. Instant coffee was probably used as frequently as ground roast (‘brewed’) coffee. While most parishioners gave their consumption in cups, others gave it in litres. We have used 150 ml per cup as a conversion factor when changing litres into cups. The caffeine content has been stipulated at c. 75–85 mg/cup by James (12) and at 86–99 mg by Seltzer (13). The caffeine content in brewed coffee seems to be 40–50% higher than that in instant coffee (12). There may be considerable discrepancies between caffeine concentration in beverages prepared in the laboratory and under domestic conditions (12). Most parishioners seem to be steady consumers, but irregularity in consumption seemed to occur not infrequently, especially during weekends. Any irregularity in consumption was particularly noted, in order to uncover possible negative effects. The daily caffeine consumption can not be calculated exactly.
A comparison between ‘light users’ (1–2 cups/day); ‘moderate users’ (3–4 cups per day) and ‘heavy users’ (5 or more cups per day) (12) was originally made. Since the coffee consumption in the ‘heavy users’ group only seems to equate the average consumption within this county (14), groups with much higher daily consumption were later constructed. Parishioners in the latter groups were asked if they had heaviness or head-pain in the wake of irregular coffee consumption, such as during week-ends. These particular parishioners were also queried concerning putative, continual head complaints. In cases of a ‘light’ consumption, no further such questions would ordinarily be asked. This study solitarily focuses on caffeine-withdrawal headache, and not on phenomena like diaphoresis and muscular pain (15), that are unlikely to derive from the headache.
Most parishioners suffered from other headache; these could putatively interfere with the clinical manifestations of an associated caffeine-withdrawal headache. Headache grading can be difficult under these circumstances.
Results
Coffee consumption
Parishioners with a fixed daily consumption
The mean number of cups in these parishioners was 4.3 per day (Table 1). There was a certain female preponderance among those with ≤4 cups per day, as opposed to in those with ≥5 cups per day (P = 0.0048; χ2 = 7.96; d.f. = 1).
Coffee consumption among those with a ‘fixed’ and a clearly varying daily consumption.
∗137 parishioners drank no coffee.
Consumption among those with a varying intake of coffee and in the total material
The mean consumption in those with a variation was 5.0 cups per day (Table 1). The latter figure is probably not so exact as that in the ‘fixed amount’ group, since the intra-individual variation as to number of cups could be considerable. The consumption in the whole series (those with and without a ‘fixed’ no. of cups) was 4.7 cups daily (Table 1). The mean consumption among females (entire series of 1741) was 4.4 ± 3.1 cups per day (n = 891), as against 5.1 ± 3.3 among males (n = 850) (P < 0.00005; t= 4.56; d.f. = 1739).
A total of 137 individuals drank no coffee. Among the coffee-drinking parishioners therefore the average usage would be 5.1 cups per day (Table 1).
Heavy consumption of coffee
A number of parishioners (n = 134) seemed to have ingested ≥ 10 cups of coffee per day on a permanent basis (Table 2). The maximum consumption in this group was 20–30 cups per day. Although the number of cups was stipulated as a single figure for each parishioner in the one group with a ‘fixed’ consumption (Table 2a), even these figures should be regarded with caution. When the number is small, for example 5 cups a day, the number may be rather exact; when the number amounts to 20, some guess-work on the part of the parishioner is probably present. It is remarkable that the percentage of parishioners without headache complaints did not seem to vary appreciably between heavy coffee drinkers (i.e. ‘fixed’ consumption: ≥10 cups/day) and those that did not drink coffee (Table 2b).
Coffee drinkers with daily consumption ≥ 10 cups/day
No significant difference between (a) and (c): (P = 0.27); No significant difference between (a) and (e) (P = 0.18); χ2 tests.
Parishioners with a considerable variation as regards consumption of coffee
A considerable variation in daily consumption was observed in a number of parishioners (Table 2b). Even though these figures can not be taken at face value, they may give more than just a hint. The minimum usage usually took place during weekends and the maximum during regular workdays.
There did not seem to be any particularly increased tendency to hard-to-classify headache, tension-type headache, respectively, migraine among parishioners with major variation in consumption, when compared to those with a low consumption. It is remarkable, however, that the ‘no headache’ fraction was higher among those with a high ‘fixed’ consumption, i.e. 24%, than among those with high variation in consumption, i.e. 13% (Table 2b).
Hangover and coffee consumption
The percentage of hangover headache cases (in accordance with the IHS criteria: code 8.3.1 (6)) among those with a consumption of ≤4 cups of coffee per day (‘fixed’ amount) was 51%, as against 50% among those with ≥5cups per day. Among the heavy users, i.e. ≥10 cups per day, there was a hangover frequency of 55%. There does thus not seem to be any substantial evidence in favour of connection between caffeine-withdrawal headache and hangover headache.
Possible caffeine-withdrawal headache
Caffeine-withdrawal headache can easily be confused with other headaches (cf Introduction). There are a number of hard-to-classify/nonclassifiable, unilateral headaches in the Vågå study population. If there exists such a headache as caffeine-withdrawal headache, it ought to be bilateral. What one should search for would, according to previous descriptions, probably be: Non-migrainous, bilateral headache cases that may be severe, and partly co-exist with nausea (25% of the cases) and occasional vomiting (2); such cases seem to be rare in our series.
In 7 parishioners, with a particularly high consumption of coffee, we had originally made speculations as regards a possible link between coffee and headache (Table 3). Some of them had suspected such a connection themselves. In group A, the consumption had been up to 20 cups a day. In some of them (group B), at the lowest, the consumption mostly seemed to exceed 20 cups per day. In these parishioners (n = 3(4?)), migraine without aura seemed to prevail. On scrutiny, two parishioners probably had ‘no headache’, or: stage 0, i.e. not even a discomfort in the head. In two parishioners, there seemed to be tension-type headache. There was generally no morning headache of a global nature – not even on Saturday and Sunday mornings – when the morning coffee might have been delayed; nor was there generally any analgesics use or other evidence of symptom-producing caffeine withdrawal, such as effect of coffee drinking. In only one of them (no. 1), such a headache may have been present, manifesting itself as augmentation of pre-existing tension-type headache.
Parishioner with extremely high coffee consumption. Presence of caffeine-withdrawal headache?
M ± A, migraine ± aura; T-TH, tension-type headache; CEH, cervicogenic headache.
∗For caffeine-withdrawal headache
∗∗Scale from 0 to 6+ pertaining to intensity (16).
Those parishioners in whom we really suspected caffeine-withdrawal headache are detailed in Table 4. In seven parishioners with a relatively high coffee consumption (>4–5 cups per day); there may have been a tendency to global heaviness occasionally, in particular on Saturdays/Sundays. and coffee seemed to help. The level of coffee ingestion in no. 5 seemed low; however, we suspect that it was considerably higher at times. We are naturally not able to rule out the possibility that a hangover headache component has been part of this picture. Can it for instance be that coffee may have a slight salutary effect even on hangover headache?
Parishioners with probable caffeine-withdrawal headache
Migraine-A, migraine without aura; T-TH, tension-type headache.
∗The upper level seemed uncertain
∗∗The usual, instant coffee brand had been replaced by decaffeinated coffee, a fact of which he was ignorant. Headache disappeared immediately when caffeinated coffee again was served (see case report).
If the seven cases (Table 4) can be regarded as acceptable, then the prevalence rate would be 0.4%. Not even all these cases find themselves at the same level of acceptability. In two of them, the evidence for caffeine-withdrawal headache seems extra strong – nos. 2 and 6, with the strongest evidence in the latter one – see casuistic report. On the other hand, this figure may not only be too low: it may be considerably too low, since there may be many cases that defy detection with this type of methodology.
Case history
A 48-year-old male office worker used to drink 1–2 l of coffee a day, much of it during the office hours. And he always had afternoon coffee, at approximately 1700–1800h, at home. He usually had no generalized headache.
One day, at 47, he started having discomfort in the head; it started in the late morning hours and continued until the late afternoon, when it completely disappeared. This pattern continued for approximately a week. It was then discovered that someone in the office had substituted de-caffeineated coffee for the regular caffeine-containing coffee. ‘Our’ office worker had been completely ignorant of this fact. When this irregularity was reversed, the office-hours headache disappeared immediately.
The ‘headache’ was mostly at the level 1 (‘unpleasantness’) on a 0–6 + scale (16), and could be up to stage 3 (‘mild’ headache, as defined by the IHS (6)). It probably interfered slightly with work in the later part of the working day, and he considered himself as a patient then. It was not associated with nausea, vomiting, phono-/photophobia or a sensation of throbbing. The interval between the last intake of coffee and onset of complaints was around 16–18 h.
Another feature adds credibility to this story: By pure coincidence, this parishioner was among those that were re-examined in a totally blinded fashion (n = 41), the interval between the two examinations being 17 months. The two histories given at different times proved to be congruent to the minute details. This story thus probably is an example of genuine caffeine-withdrawal headache.
This headache also fulfilled the IHS criteria for acute tension-type headache (6).
Discussion
Our study shows that caffeine-withdrawal headache is probably less common than previously considered.
Coffee consumption in Vågå
The Scandinavians, apparently, are at the top of the list of coffee consumption (12). The first demography from Norway concerns sobriety and is written by E. Sundt in 1857. Vågå is specifically mentioned, but not the consumption of coffee (17). Coffee is, however, mentioned in connection with neighbouring parishes, where liqueur was claimed to be replaced by coffee at the time.
There were already 12–14 coffee houses in the parish centre, Vågåmo, between the two world wars and as many as 10–12 could be operative at the same time (10). At that time, there were <400 inhabitants in the parish centre. The coffee drinking tradition has been kept up in modern time; however, the parish centre now has upwards of 1600 inhabitants.
The average consumption in the same county (Oppland, Norway) has previously been established as 4.5 cups/day among 35–49-year-old individuals (14). That study comprised the whole of the county (a sample of 24240 individuals studied, by questionnaire, from a total population of >180000 (14)). In Vågå, the average consumption among 18–65-year-olds who drank coffee (average age: 44.2 years) was around 5 cups and in the whole population 4.7 cups a day (Table 1). The latter figure is most apt for a direct comparison with 4.5 cups/day in the mentioned study (14). In a recent questionnaire study from the south-eastern part of Norway (n = 191, aged 24–69 years), an average consumption of 4.9 cups/day was found (18). The strong belief that Vågå is a ‘coffee-drinking parish’ was thus not sustained by our research.
Does caffeine-withdrawal headache exist?
There have been strong statements about this in the literature: ‘No doubt that caffeine-withdrawal headache is a real phenomenon’ (19); ‘… does exist as a clinical psychiatric syndrome’ (8); ‘… should be considered a common and important medical and psychiatric entity’ (15).
We are, nevertheless, in doubt about the existence of a full-fledged and frequently occurring form of caffeine-withdrawal headache. There seem to be various counterarguments against such a concept:
The headache was first described prior to the IHS criteria, for example prior to the present definition of tension-type headache (1988; 6). The lack of such criteria could have been the cause of considerable diagnostic confusion.
Intensity description. Headaches ‘extreme in severity’ (1, 2), a ‘splitting headache’ (8), as well as headache of maximum intensity (‘5’, on a 0–5 scale (15)) have been described. It would be somewhat strange if caffeine-withdrawal per se generally should lead to headache in the top pain category. There may be various explanations why such strong headaches were encountered. If the grading of headache has been made by the test subject only, and no assessment according to objective criteria has been made, an over-statement of the pain situation may have occurred. A headache free individual rather definitely will have only vague ideas about what a ‘splitting headache’ (8) is. One is aware that later and in other contexts, the severity of this headache has been softened (e.g. ‘mild’ (4)).
In such large groups that have been studied (n = 205 (8), n= 22 (2)) there might have been several migraine patients, in whom migraine attacks may have confused the picture.
Vomiting during attacks, in 9% of the cases (2). This seems strange, if it were to be due to caffeine-withdrawal solely. It should be emphasized that the gastrointestinal symptoms, originally reported by Dreisbach & Pfeiffer (2), have been toned somewhat down later (8).
The studies have partly been done by questionnaire filled in by the patients (8). This type of methodology may not be entirely suited for this extremely complex clinical situation.
The percentage of individuals/patients who responded with headache upon caffeine-withdrawal varied greatly between various studies: from 84% (55%) (2) to 20% (28%) (8). This may suggest methodological differences.
The early studies have been carried out by non-neurologists. It would have been reassuring if the results in this field had been corroborated by neurologists with headache expertise, with proper grading of the complaints.
There still probably is a caffeine-withdrawal headache (Table 4). We do, however, not feel that convincing evidence has been presented for its existence in the full-fledged form, with severe headache and with nausea deriving from this source, and where even vomiting could occur.
How frequent is caffeine-withdrawal headache in the population?
Dreisbach & Pfeiffer (2) in an experimental situation found this headache in 21 (55%) of 38 trials. In 11 other trials, the headache was ‘definite, but not severe enough to necessitate treatment’– a total of 84%. Greden et al. (8) in an experimental situation (n = 205) found this headache in 20% or 28%, depending upon how the calculations were made. Roller (15) in nine experiments in one individual found this headache in each of the experiments.
The prevalence figures in the present study are at a much lower order of magnitude (0.4%). This latter figure should be taken as a minimum figure. Lack of prospective design and lack of interventions necessitate this cautious conclusion. It goes without saying that the figures from the various studies are significantly different. It should, however, immediately be emphasized that the first mentioned studies are experimental ones. Our study is a cross-sectional prevalence study in an unselected, rural Norwegian population. For that reason, these studies are not directly comparable. The interventional studies (2, 15) cannot tell much about what will happen at the grassroots level in real life.
For what it is worth, the first author (O.S.) has repeatedly tried to reduce coffee consumption from the usual level of 5–7 cups to 0–1 cup for a day or two. In one, week-long experiment, the coffee consumption was gradually increased to 2–2.25 litres per day for the last three days before coffee was abruptly discontinued. No appreciable headache manifested itself after that. Several colleagues have tried a similar procedure, and with similar results. Obviously, there are individual differences; there may be individual predisposition to such headache.
The clinical appearance
Although based on a far-from-solid foundation, a picture somewhat different from the previous one may seem to emanate with respect to caffeine-withdrawal headache: a more soft variety with vague, mild to extremely mild, hard-to-discern complaints, with no nausea and vomiting, no photo- and phono-phobia, and no throbbing. To the best of our knowledge, analgesics were not used. Coffee seemed to abate this headache. An outline of our present impressions is given in Table 5, where a comparison with previous works is made.
Assumed symptomatology of caffeine-withdrawal headache
∗In the group with caffeine-withdrawal headache more than 50% more caffeine was used (mean 616 mg/day) than in the nonheadache group (395 mg/day)
∗∗Grades 1–3 on a 0–6 + intensity scale (16).
Although the present study, due to limitations of the methodology, cannot furnish even close to correct figures as regards prevalence of caffeine-withdrawal headache, the following can probably be stated:
The idea that caffeine-withdrawal headache is a ubiquitous, severe headache of considerable social and health importance can, with reasonable certainty, be toned down. That caffeine-withdrawal headache may appear seems highly likely, however, it seems to be rather rare at the grassroots level. Possibly, a combination of sensibilization and inconsiderate/untoward usage of coffee underlie scattered instances of low-degree caffeine-withdrawal headache. Those with a marked variation in caffeine consumption may possibly be somewhat more headache-prone than those without such variation in coffee consumption. Such a tendency could, however, not definitely be demonstrated in the present study.
In spite of the general shortage of evidence for implicating caffeine-withdrawal as being a causal factor in a particular and frequently occurring headache in the population at large, the possibility remains that it may aggravate already existing brands of headache. For that reason, and in view of the negative influence of caffeine on blood cholesterol (14, 18), it may be a piece of good advice to all headache patients to limit their coffee intake. A daily consumption of 250–300 mg caffeine is considered to be ‘considerable’ (12, 13). Even the average caffeine consumption in the Vågå study seems to be far beyond that.
In a relatively recent double-blind study by Silverman et al. (20), ‘psychological distress and impairment of performance’ have been assessed in a careful way. ‘Headache’ has also been included, and its severity assessed according to a 0–3 scale, the latter assessment apparently being carried out by the participants themselves. ‘Moderate’ to ‘severe’ headache was found in 52%; there were no figures for ‘severe’ headache, specifically.
Their conclusion is very carefully worded: ‘Physicians should be aware of this syndrome when they encounter patients who have symptoms after ceasing to consume caffeine or when they advise patients to stop consuming caffeine’(e.g. in connection with ‘cardiac catheterization’, ‘anxiety, arrhythmia, … insomnia’). We quite agree. They have made no extrapolation of their results to the situation at the grassroots level, the topic in the present study.
In our opinion, further and well-planned studies are needed to solve some of the extremely intricate problems touched upon in this presentation.