In this large-scale population-based study a bidirectional relationship between anxiety and depression with migraine and TTH was found. We identified slightly higher risk ratio for definite migraine than for TTH for anxiety.
Comparison with previous studies
In the present study we found approximately 30% increased risk of depression among patients with migraine or TTH, and that depression at baseline more than doubled risk of migraine, whereas a 40% increased risk of TTH was found. Accordingly, a questionnaire-based 14-year follow-up study of 36,016 women without depression at baseline, reported that the risk of depression was 1.4 (1.3-1.6) for non-migrainous headache, 1.5 (1.4-1.8) for migraine with aura, 1.4 (1.3-1.6) for migraine without aura, and 1.6 (1.4-1.8) for past history of migraine [
14]. Most of the previous population-based follow-up studies from Canada and US have focused mainly on the temporal relationship between migraine and major depression. For example, a 12-year population-based follow-up of approximately 14,000 individuals aged >12 years from Canada demonstrated a bidirectional relationship between migraine and major depressive episodes [
13]. Thus, migraine gave a 60% increased risk (RR 1.6, 95% CI 1.3-1.9) of major depression, whereas major depression gave 40% increased risk (RR 1.4, 95% CI 1.0-1.9) migraine [
13]. A separate publication based on this Canadian population focused on 8-year follow-data of 9,288 participants aged 18-64 years [
35]. In sex and age-adjusted analyses they showed that depression was predictive of migraine (HR 1.62, 95% CI 1.03 - 2.53) and migraine was predictive of depression (HR 1.55, 95% CI 1.15 - 2.08) [
34]. Interestingly, however, supplementary adjustment for major life stressors e.g. childhood trauma and/or work stress decreased the association substantially [
35]. Another two-year follow-up study of adults aged 25-55 years in the US showed a bidirectional association between major depression and migraine [
36]. Major depression at baseline predicted the first-onset migraine (OR 3.4, 95% CI 1.4-8.7), and vice versa, migraine at baseline predicted the first-onset major depression (OR 5.8, 95% CI 2.7-12.3) [
36].
Some previous cross-sectional studies have diagnosed depression using HADS score. Depression defined as HADS-D score of ≥ 11 was more than two times more likely among individuals with migraine than non-migrainous headache based on cross-sectional data from HUNT2 [
5]. Similarly, merged data of six previous EU-based cross-sectional studies reported that migraine was twice as common (OR 2.1, 95% CI 1.3–3.4) among individual with depression measured by HADS [
37]. A recent meta-analysis based on data from cohort studies reported that migraineurs were almost twice (OR 1.81, 95% CI 1.20-2.72) as likely to develop depression as those without migraine [
38].
In the present study, migraine with aura was associated with higher RR (1.7) of co-existing depression and anxiety than RR (1.3) of depression alone. A previous cross-sectional study showed that individuals with migraine with aura were more likely to be depressed compared to migraine without aura in women but not in men [
8]. In the Michigan study, the risk for depression was almost twice as likely among migraine with aura compared to migraine without aura (ORs 4.0 vs. 2.2) [
39].
A bidirectional association between depression and TTH was also found in the present study. To our knowledge, data on such a relationship has not been provided for TTH in previous longitudinal studies. In cross-sectional population-based studies, divergent results have been reported for TTH. A recent population-based study performed in Korea reported that patients with TTH were almost two times more likely to have depression (4.2% vs. 1.8%) than those without [
40]. In contrast, a Nepalese population-based study found no association between depression and TTH [
41]. Similarly, no association was found between depression measured by HADS and TTH in analyses using the merged data of six previous cross-sectional EU studies [
37].
In the present study, we found that anxiety increased the risk of migraine, probable migraine, and TTH, and vice versa, that all these headache types increased the risk of anxiety.
Very few long-term prospective studies have evaluated the temporal bidirectional relationship between anxiety disorders and common primary headache disorders. Notably, prospective data of 591 young adults from Switzerland showed that the combination of major depression and anxiety disorder doubled the risk of migraine, and that anxiety disorders generally preceded the occurrence of migraine [
9]. Furthermore, in a 1.2-year follow-up study of 1,007 adults aged 21-30 years, migraine at baseline increase the odds ratio of panic disorders with 12.8 [
10]. Several cross-sectional population-based studies have reported evidence for positive associations between migraine and anxiety [
5,
41‐
44]. In studies from Norway, the US and Canada, anxiety disorders were respectively 3.2, 3.1 and 2.5 times higher among migraineurs compared to headache-free individuals [
5,
42,
43]. A recent systematic review evaluated the comorbidity of anxiety and migraine in the cohort studies, and estimated RR with an average of 1.63 [
45].
In the present study, higher risk of anxiety was found for migraine with aura than for migraine without aura (1.8 vs. 1. 39). In accordance with the present study, depression with comorbid anxiety was more likely among women with migraine with aura than among those without aura in a cross-sectional study based on HUNT2, whereas an association between anxiety alone and migraine with aura was not found [
8]. In the Michigan study, the association between anxiety and migraine was slightly higher for migraine with aura than for migraine without aura (ORs 3.1 vs. 2.3) [
39].
Very few population-based cross-sectional studies have evaluated the impact of anxiety on TTH. A Korean population-based study reported that the prevalence of anxiety was almost twice as high (9.5% vs. 5.3%) among patients with TTH compared to headache-free controls [
40]. Furthermore, TTH was associated with anxiety measured by HADS score both for males (ORs 2.5, 95% CI 1.7–3.7) and females (OR 1.5, 95% CI 1.1–2.1) based on data from six EU studies [
37]. In the present study, a bidirectional relationship between co-existence of anxiety and depression (defined as total HADS score of ≥15 or ≥22) was more evident for migraine than for TTH. For example, total HADS score ≥22 at baseline was associated with more than four times increased risk of migraine at follow-up. In accordance, comorbidity of anxiety and depression was more strongly associated with increased risk of migraine than pure anxiety or depression in the Zurich study [
44]. Furthermore, migraine was associated with increased OR of major depression combined with panic disorders 25.2 (95% CI 2.5-251) in the 1.2-year US follow-up study, whereas OR of 2.5 (95% CI 1.0-6.5) was found for major depression alone [
10].
Interpretation
The bidirectional relationship between affective disorders and migraine and TTH could be the results of shared pathophysiological mechanisms related to neurotransmitters, genetic basis and/or environmental factors. The follow-up studies from Canada demonstrated that life stressors such as childhood trauma may be an important common underlying factor, supporting the environmental aspect [
13,
35]. Most previous studies focusing on the relationship between migraine and depression, suggest a common dysfunction related to serotonin, dopaminergic and GABAergic systems [
46,
47]. Furthermore, twin and family studies have indicated that the bilateral relationship between migraine and depression at least in part could be explained by genetic factors [
46,
47,
49]. However, it is still unclear how specific genetic variants are associated with both the primary headaches and affective disorders. In the present study, we identified a bidirectional relationship between anxiety and depression with migraine and TTH. Thus, in the diagnostic headache interview, clinicians should take into account the existence of psychiatric comorbidities and consider to include questions regarding traumatic life events. Potential beneficial or synergistic effects as well as treatment complications should be considered when choosing treatment [
16]. Vice versa, headache questions should be included when assessing patients admitted for psychiatric evaluation, and comorbid headache should be taken into account during treatment.
Strengths and limitations of the study
The major strengths of this study are the prospective design, an 11-year follow-up period, and the large cohort from the adult population of an entire population of Trøndelag County. The present study design is more likely to identify causal relationships than cross-sectional studies. Furthermore, in the longitudinal analysis of risk factors at baseline, we have adjusted for the same confounding factors in all analyses, making the estimated RRs comparable. Finally, the study included use of validated diagnoses of headache [
25,
26] and use of well-established cutoff score of HADS [
21‐
23]. However, since no previous longitudinal studies have evaluated depression or anxiety using HADS score, direct comparison with these prospective studies should be done with caution.
Several study limitations should also be addressed. Firstly, generalization of the results to the entire population must be made with caution, since the participation rate was 54% and 58% respectively in the two surveys, and only 36% of the invited population in HUNT3 participated in HUNT4. On the other hand, a bias in relative risk estimates will only arise if participation rates differ regarding headache status or affective disorders. The fact that neither headache nor anxiety and depression were primary objectives of these surveys makes selective participation and loss follow-up unlikely. Secondly, the diagnostic accuracy of questionnaire-based diagnoses of headache and affective disorders was not optimal, and the possibility of misclassification cannot be ruled out. However, most likely, such misclassification goes in both directions, decreasing the difference in RR between the diagnostic subgroups. The gold standard for the diagnosis of common primary headache disorders and affective disorders is the face-to-face interview by experts, but such strategy is not possible in large-scale population-based studies. Thirdly, even prospective studies are more likely to identify causal relationships than cross-sectional studies, the present study lacked information on the time of onset of both depression/anxiety and migraine/TTH, which would allow identification of causal relationship between both disorders. Finally, the number of individuals with HADS score ≥11 were relatively low. However, the associations studied were sufficiently powered to detect significant differences using the cut-off score ≥8 and for the 8-10 HADS score groups.