Introduction

Caffeine is one of the most popular food ingredients in modern society, with approximately 80% of the global population consuming caffeinated products almost daily1. Caffeine has beneficial effects on health, including decreased risk of cardiovascular diseases, diabetes, cancer, and Parkinson’s disease2. By contrast, caffeine may cause tremor, anxiety, insomnia, tachycardia, and increased blood pressure3. Considering both its beneficial and harmful effects on health, health organizations recommend up to 400–600 mg of daily caffeine consumption4,5.

Caffeine has various effects on migraine and non-migraine headaches. Caffeine has analgesic effects in the treatment of headache6,7,8,9. The combination of caffeine and analgesics is more effective than analgesics alone in the acute treatment of migraine6. Caffeine is also effective in the treatment of non-migraine headaches, including tension-type, hypnic, and post-dural puncture headache7,8,9. Moreover, acute caffeine withdrawal can induce headaches10. On the contrary, caffeine discontinuation improves acute migraine treatment11. Some individuals with migraine reported that caffeinated beverages triggered migraine attacks12. Two population studies reported that high caffeine intake increased the risk of chronic daily headache (CDH)13,14. Nevertheless, most studies did not adequately define migraine or evaluate the impact of caffeine on the clinical characteristics of migraine15,16,17. In addition, studies on the impact of caffeine intake on the efficacy of acute treatment of headache have not been conducted.

In Korea, 89% of daily caffeine intake is through coffee consumption18. Therefore, evaluating the relationship between coffee consumption and migraine can provide an opportunity to identify the impact of caffeine intake on the clinical characteristics and acute treatment response in individuals with migraine. We hypothesized that the clinical characteristics and acute treatment response of individuals with migraine would differ significantly according to coffee consumption. This study aimed to examine the relationship between coffee consumption and clinical characteristics of participants with migraine compared with those with non-migraine headache using data from a nation-wide population-based study in Korea.

Methods

Survey

We conducted a secondary analysis of previously collected data where the analysis process was preplanned. This cross-sectional study used baseline data from the Circannual Change in Headache and Sleep (CHASE) study, a nationwide web-based survey of adults aged 20–59 years investigating chronobiological aspects of headache and sleep. A detailed process of the survey was described previously19. In brief, two-stage stratified cluster sampling was used to select respondents representative of the Korean population. Data from the 2015 population and housing census by Statistics Korea were used for the population distribution of all Korean territories, except Jeju-do20. The target sample size was set at 3000 individuals with an estimated sampling error of ± 1.8%. The survey participants were followed up every 3 months for over a year after the baseline assessment. The baseline investigation consisted of validated questionnaires for the diagnosis of headache, anxiety, depression, insomnia, and impact of headache. Sociodemographic and lifestyle factors, including coffee consumption, were also evaluated. The baseline CHASE study was conducted in October 2020.

Assessment of migraine and non-migraine headache

Migraine was diagnosed using a series of questions based on diagnostic criteria from the International Classification of Headache Disorders, Third Edition (ICHD-3) for migraine without or with aura10. Additional telephone interviews were conducted by a headache specialist (M.K.C.) to diagnose migraine and assess the validity of the migraine diagnosis from the survey responses. The diagnostic sensitivity and specificity of our survey were 92.6% and 94.8%, respectively, with a kappa coefficient of 0.87521. Non-migraine headache was defined as any headache not fulfilling the migraine diagnostic criteria. Clinical characteristics including monthly headache days, monthly severe headache days, monthly days with acute medications, headache intensity, unilateral location, pulsating quality, and accompanying symptoms were determined as part of the survey. The Migraine Disability Assessment (MIDAS) was used to evaluate headache-related disability22.

Assessment of coffee consumption

Coffee consumption was assessed through the question, “On average, how much coffee did you drink in the past year?”. The participants were asked to select one of the following: (1) do not drink coffee, (2) < 1 cup per week, (3) 1–6 cups per week, (4) 1–2 cups per day, (5) 3–4 cups per day, and (6) ≥ 5 cups per day. The level of coffee consumption was further categorized into no-to-low (< 1 cup per day), moderate (1–2 cups per day), and high (≥ 3 cups per day).

Assessment of anxiety, depression, insomnia, and stress

Participants were asked multiple questions on anxiety Generalized Anxiety Disorder-7 (GAD-7), depression Patient Health Questionnaire-9 (PHQ-9), insomnia Insomnia Severity Index (ISI), and stress Korean version of the Brief Encounter Psychosocial Instrument (BEPSI-K). Higher scores indicate more severe psychological distress and sleep disturbance. The Korean versions of the GAD-7, PHQ-9, ISI, and BEPSI-K have been validated23,24,25,26. Participants were classified as having anxiety, depression, insomnia, and stress if they had GAD-7, PHQ-9, ISI, and BEPSI-K scores greater than 7, 9, 15, and 2.4, respectively.

Assessment of acute treatment response

The use of medications during acute headache attacks was determined. The effectiveness and tolerability of the acute headache treatment were assessed using the six-item migraine Treatment Optimization Questionnaire (mTOQ-6), a validated measure developed to assess response to acute treatment in persons with migraine27. The mTOQ-6 was composed of six items regarding (1) quick return to function, (2) 2-h pain free, (3) sustained 24-h pain relief, (4) tolerability, (5) comfort in making plans, and (6) perceived control. Respondents were asked to rate the frequency of each item using the response options of (1) never, (2) rarely, (3) less than half the time, and (4) half the time or more. The mTOQ-6 total score ranges from 6 to 24. There is no cut-off value, and higher scores indicate better optimization of acute treatment.

Statistical analyses

Categorical variables were compared using the Pearson’s χ2 test or Fisher’s exact test. Linear by linear association was used to find trends for categorical variables. For continuous variables, the Kolmogorov–Smirnov test was performed to confirm the normality of the distribution. Nonparametric tests were used when a variable was not normally distributed in at least one group, and all continuous variables were subjected to nonparametric tests. The Kruskal–Wallis test followed by Bonferroni's post hoc test were conducted to compare age and body mass index (BMI). Monthly headache days, monthly severe headache days, monthly acute medication days, headache intensity (numerical rating scale [NRS], 0–10), headache-related disability (MIDAS), and acute headache treatment response (mTOQ-6) were compared using analysis of covariance (ANCOVA) followed by Bonferroni's post hoc test, as ANCOVA can be used even when the model does not meet parametric assumptions because it is robust to violation of assumptions28. Age, sex, and BMI were selected as covariates for ANCOVA for all variables. Anxiety, depression, stress, preventive medication use, and current smoking, which are known to affect acute migraine treatment response, were also adjusted for the mTOQ-6 total scores27. The significance level was set at two-sided p-value < 0.05 for all analyses. Results are shown as numbers with percentages for categorical variables and as median with interquartile range (IQR) for continuous variables. Statistical analyses were performed using IBM SPSS Statistics for Windows, version 24 (IBM Corp., Armonk, N.Y., USA). No statistical power calculation was conducted to guide the sample size.

Ethical approval

The institutional review board of Severance Hospital approved the CHASE study protocol (Approval Number 2020-0034-001). The study was conducted according to the principles of the Declaration of Helsinki. All participants volunteered and provided their written informed consent to participate in the research.

Results

Survey

A total of 91,153 individuals received an email invitation and 10,699 responded to participate. Of the 10,699 participants who consented, 1075 withdrew from participation, 6215 abandoned the survey, 379 were removed for exceeding the quota, and finally, 3030 (28.3% participation rate) completed the survey (Fig. 1). No significant differences were observed in the distribution of sex, age, education level, and size of residential area of the participants from the total Korean population (Table 1). This study had no missing data because the participants were required to answer all survey items to complete the survey. When asked, "Have you had headaches in the past year?", 1938 (64.0%) of the 3030 participants responded positively. Among the participants who reported headache in the past year, 170 (5.6%) were identified as having migraine. Accordingly, 1768 (58.3%) participants were identified as having non-migraine headache (Fig. 1).

Figure 1
figure 1

Flowchart of the participation process in the Circannual Change in Headache and Sleep study.

Full size image
Table 1 Sociodemographic characteristics of the survey participants, total Korean population, and cases with migraine and non-migraine headache.
Full size table

Coffee consumption in the survey participants

Among the 3030 participants, 1137 (37.5%), 1410 (46.5%), and 483 (15.9%) had no-to-low, moderate, and high coffee consumption, respectively. Among the 170 participants with migraine, 54 (31.8%), 83 (48.8%), and 33 (19.4%) had no-to-low, moderate, and high coffee consumption, respectively. Among the 1768 participants with non-migraine headache, 637 (36.0%), 851 (48.1%), and 80 (15.8%) had no-to-low, moderate, and high coffee consumption, respectively. Among the 1092 participants with non-headache, 446 (40.8%), 476 (43.6%), and 170 (15.6%) had no-to-low, moderate, and high coffee consumption, respectively. Coffee consumption was significantly different among participants with migraine, non-migraine headache and non-headache (p = 0.035). Coffee consumption tended to increase in the following order: participants with non-headache, non-migraine headache, and migraine (p for trend = 0.011) (Fig. 2).

Figure 2
figure 2

Distribution of coffee consumption according to headache diagnosis.

Full size image

Demographic and clinical characteristics of headache according to coffee consumption in participants with migraine and non-migraine headache

Demographic and clinical characteristics of headache were compared among groups divided by coffee consumption in participants with migraine (Table 2) and non-migraine headache (Table 3).

Table 2 Demographic and clinical characteristics of headache according to coffee consumption in participants with migraine.
Full size table
Table 3 Demographic and clinical characteristics of headache according to coffee consumption in participants with non-migraine headache.
Full size table

Among the 170 participants with migraine, participants with no-to-low coffee consumption were significantly younger (no-to-low, 36.0 [27.0–43.0]; p < 0.001 vs. moderate, 42.0 [36.0–50.0]; p = 0.006 vs. high, 42.0 [36.5–45.5]) and more likely to have depression (no-to-low, 31.5% [17/54]; p = 0.016 vs. moderate, 12.0% [10/83]; p = 0.048 vs. high, 9.1% [3/33]) and stress (no-to-low, 61.1% [33/54]; p = 0.005 vs. moderate, 33.7% [28/83]; p = 0.016 vs. high, 30.3% [10/33]). However, there were no significant differences in sex, headache days per month, severe headache days per month, acute medication days per month, pain characteristics, accompanying symptoms, headache-related disability (MIDAS), anxiety, and insomnia among the three groups (Table 2).

Among the 1,768 participants with non-migraine headache, the high coffee consumption group was older (high, 46.0 [39.0–51.0]; p < 0.001 vs. no-to-low, 34.0 [26.0–45.0]; p = 0.002 vs. moderate, 43.0 [35.0–51.0]), had a lower proportion of women (high, 41.1% [115/280]; p < 0.001 vs. no-to-low, 54.8% [349/637]; p < 0.001 vs. moderate, 58.0% [494/851]) and higher BMI (high, 24.2 [22.0–26.8]; p = 0.019 vs. no-to-low, 23.4 [21.2–26.2]; p = 0.017 vs. moderate, 23.4 [21.5–26.0]) than the other groups. Additionally, the moderate coffee consumption group was significantly older than the no-to-low group (p < 0.001). Participants with moderate coffee consumption were less likely to have photophobia (moderate, 17.9% [152/851]; p = 0.039 vs. no-to-low, 23.1% [147/637]; p = 0.019 vs. high, 25.4% [71/280]) and anxiety (moderate, 19.5% [166/851]; p = 0.006 vs. no-to-low, 26.2% [167/637]; p = 0.006 vs. high, 28.2% [79/280]) than those with no-to-low and high coffee consumption. The moderate group also had a lower proportion of participants with stress than the no-to-low group (moderate, 21.9% [186/851]; p = 0.002 vs. no-to-low, 29.5% [188/637]). Other headache-related variables including headache days per month, severe headache days per month, acute medication days per month, NRS for pain intensity, MIDAS, depression, and insomnia did not differ with coffee consumption (Table 3).

Use of acute treatment medications in participants with migraine and non-migraine headache

Among the 170 participants with migraine, 34 (62.7%), 70 (84.3%), and 23 (69.7%) participants in the no-to-low, moderate, and high groups, respectively, used acute treatment medications. The proportion of participants that used acute treatment medications was significantly higher in the moderate group than in the no-to-low group (p = 0.013). Nevertheless, there were no significant differences in the proportions of participants who used acute treatment medications between the no-to-low and high groups (p > 0.999) and between the moderate and high groups (p = 0.223). Medication classes used for acute treatment did not differ significantly between the three groups, except for tramadol, which lost significance in post hoc analysis (Table 4).

Table 4 Prevalence of acute medication use and classes in participants with migraine and non-migraine headache.
Full size table

Among the 1768 participants with non-migraine headache, 304 (47.7%), 453 (53.2%), and 153 (54.6%) participants with no-to-low, moderate, and high coffee consumption, respectively, used acute treatment medications. The proportion of participants using acute treatment medications was not significantly different among the three coffee consumption groups (p = 0.056). The use of acute treatment medication classes did not differ significantly among the three groups, except for the combination analgesic class (Table 4).

Response to acute treatment of migraine and non-migraine headache

ANCOVA adjusted for age, sex, BMI, anxiety, depression, stress, preventive medication use, and current smoking showed no significant differences in acute treatment response as measured by the mTOQ-6 between the three coffee consumption groups in participants with migraine (no-to-low, 18.5 [16.8–21.3]; moderate, 22.0 [17.8–24.0]; high, 21.0 [17.0–23.0]; p = 0.437) and non-migraine headaches (no-to-low, 22.0 [19.0–24.0]; moderate, 23.0 [20.0–24.0]; high, 22.0 [20.0–24.0]; p = 0.440) (Table 5).

Table 5 Effect of acute medications according to coffee consumption in participants with migraine and non-migraine headache.
Full size table

Discussion

The main findings of this study were as follows: (1) coffee consumption showed an increasing trend in the order of non-headache, non-migraine headache, and migraine; (2) depression and stress decreased with increasing coffee consumption in participants with migraine; (3) most headache-related variables and response to acute headache treatment did not differ significantly according to coffee consumption in either migraine or non-migraine headache. Based on our findings, we could partially reject our hypothesis that clinical characteristics were significantly different according to coffee consumption in participants with migraine.

Several studies have evaluated caffeine consumption in migraine and headache. A large study in the United States (US) involving 25,755 women reported that caffeinated coffee consumption did not differ significantly between individuals with migraine and non-migraine headache17, but significantly more coffee was consumed by individuals with non-migraine headache than those without headache. A Norwegian study showed that high caffeine intake (> 540 mg/day) was associated with a modest increase in headache prevalence, but there was no significant association between migraine prevalence and caffeine intake16. In contrast, an epidemiologic study in Japan found that individuals with migraine consumed significantly more coffee and tea than individuals without headache in the same community15. Our study showed a trend toward higher coffee consumption in the order of participants with non-headache, non-migraine headache, and migraine. One possible explanation for the discrepancy between the present study and studies from the US and Norway is the difference in ethnicity. Although the average daily caffeine consumption of 68 mg in Koreans is lower than that of Americans (186 mg) and Norwegians (426 mg), the average daily caffeine consumption of 262 mg in the Japanese population is similar to that of Americans, suggesting that the higher caffeine consumption in migraine compared to non-migraine headache or non-headache is likely due to ethnic differences16,18,29,30. The present study may provide evidence of increased coffee consumption in participants with migraine than those with non-migraine in the Asian population, which was different from the US and Norwegian populations. Differences in socioeconomic status and the method to assess caffeine consumption may be other possible explanations for the discrepancy.

Although caffeine has various effects on migraine, detailed information on the impact of caffeine consumption on the clinical characteristics of migraine is scarce. Two studies from the US reported the association between daily caffeine consumption and CDH and identified that high daily caffeine consumption was significantly associated with an increased risk of CDH13,14. One study analyzed the outcome by dividing the groups into daily coffee drinkers and non-drinkers, and the other study divided the groups into high and non-high caffeine consumers. Nevertheless, these studies did not evaluate the relationship between caffeine consumption and clinical characteristics of migraine in detail. The present study evaluated detailed clinical characteristics of migraine including headache days per month, headache characteristics, accompanying symptoms, disability by migraine, comorbidities, and response to acute treatment. In contrast to the two previous studies, we found that there were no significant differences in headache days per month, severe headache days per month, and days with acute medications. The discrepancy may be due to differences in the categorization of coffee consumption. We categorized high coffee consumption as ≥ 3 cups/day and assumed that the average caffeine content in a cup of coffee in Korea is 75 mg. This estimation was based on a previous study that calculated the caffeine content of the most popular types of coffee in Korea, using the caffeine levels specified in the Korea Food Additives Code31. Therefore, consuming ≥ 3 cups of coffee per day would result in an intake of ≥ 225 mg of caffeine per day. Among the 33 migraineurs in our study who were classified as high coffee consumers, only one consumed ≥ 5 cups/day, while the remaining 32 consumed 3–4 cups/day (equivalent to 225–300 mg/day of caffeine). A US study categorized high caffeine consumption as > 287 mg/day, which was the top quartile of caffeine consumption13. Therefore, a considerable number of participants with migraine in the high coffee consumption group in our study may be classified as non-high caffeine consumers in the American study, which may have contributed to the difference in results.

Furthermore, it has been suggested that migraine and psychiatric comorbidities are closely intertwined. A review article demonstrated that individuals with migraine were more likely to suffer from psychiatric disorders than the general population, and individuals with migraine who suffer from mood disorders were more likely to be refractory to migraine treatments32. Additionally, depression, anxiety, and stress were risk factors for treatment refractoriness and migraine chronification in patients with episodic migraine. On the other hand, several studies have shown a protective role for coffee in depression and anxiety33,34,35. Results from a meta-analysis showed that coffee consumption was significantly associated with a lower risk of depression33. We also found an association between increased coffee consumption and lower rates of depression and stress in participants with migraine. As it has been suggested that mood disorders are associated with increased migraine frequency and disability36, and that coffee consumption is inversely associated with mood disorders, it was expected that there would be differences in migraine symptoms based on coffee consumption. However, in this study we observed that most clinical features of migraine did not differ by coffee consumption. This finding might be explained by the balance of beneficial and detrimental effects of caffeine on migraine, including the triggering and relief of migraine.

The mechanism of action of caffeine in migraine is not fully understood. It has been reported that caffeine, competes with adenosine, with which it shares a similar structure, for the A1 and A2A adenosine receptors; these two receptors have opposing effects with each other3,37. Caffeine inhibits the activation of the trigeminal nerve pain pathway and vasodilation by blocking the A2A receptor. Conversely, by antagonizing the A1 receptor, caffeine promotes nitric oxide production, causing vasodilation and triggering migraine. The beneficial and triggering effects of caffeine on migraine may be related to this dual mechanism of action.

The present study has several limitations. First, we assessed coffee consumption based on participant self-reported number of coffee cups consumed. However, we did not differentiate between types of the coffee consumed; instead, we assessed consumption solely based on the total number of cups. The amount of caffeine in a cup of coffee varies depending on coffee bean variety and roasting methods, serving size, and coffee type38. Therefore, the caffeine intake of two individuals reporting the same number of coffee cups consumed may differ. In addition, we did not assess caffeine ingestion through non-coffee beverages, foods, and other sources. Tea, chocolate, cola, carbonated beverages, and caffeine-containing medications are common sources of caffeine other than coffee. However, in Korea, only about 11% of the daily intake of caffeine is consumed from sources other than coffee18. The cup-based coffee consumption in our study was similar to a previous study in Korea (< 1 cup/day: 33.4%, 1–2 cups/day, 43.0%, and > 2 cups/day: 23.5%). Second, although we investigated the relationship between coffee consumption and migraine using data from a large sample size, some subgroup analyses had smaller sample size and did not have sufficient sample power. Third, we categorized coffee consumption as no-to-low, moderate, and high coffee consumption based on the coffee consumption of the Korean general population. However, this classification may not be able to properly identify the impact of high caffeine use in Western countries. Studies from European and American countries reported significant effects of high caffeine consumption in migraine or headache13,14,16. However, the average caffeine consumption in Korea was lower than that in American and European countries, making it difficult to see the effect of high-dose caffeine intake in the present study. Only 3.0% of the participants in our study consumed ≥ 5 cups in a day, which was estimated to be ≥ 375 mg of caffeine. Fourth, we did examine additives that are often added to coffee. Common additives in coffee such as milk, sugar, creamer, and artificial sweeteners, may influence the clinical presentation of migraine39,40,41. Although many epidemiological studies have not examined the use additives, this is a limitation of the present study. Fifth, this study used a self-reported, web-based questionnaire to assess coffee intake, headache diagnosis, insomnia, psychological status, and other relevant factors. The questionnaire used to diagnose migraine had high sensitivity and specificity21. Insomnia and psychological status were assessed using questionnaires with high validity and reliability. However, these self-report measures rely on personal recall, which may be prone to error. Lastly, our study was cross-sectional and could not identify causality.

Conclusions

The present study investigated the relationship between migraine and coffee consumption using data from a nationwide population-based study. Coffee consumption showed an increasing trend in the order of participants with migraine, participants with non-migraine headaches, and those with non-headache. A significant correlation was found between coffee consumption and psychiatric comorbidities in participants with migraine, with higher coffee consumption associated with lower levels of depression and stress. However, most clinical characteristics and response to acute treatment were not significantly different according to coffee consumption in participants with migraine.