Abstract
Some data indicate that migraine with aura (MA) is more strongly associated with anxiety disorder and depression than migraine without aura (MoA), but the evidence is not conclusive. In the Nord-Trøndelag Health study 1995-1997, a total of 49 205 (75% of the participants) subjects gave valid answers to both HADS (Hospital Anxiety and Depression Scale) and a validated headache questionnaire. Associations between anxiety disorder/depression and MA/MoA were evaluated by multiple logistic regression analysis. Depression (DEP) [odds ratio (OR) 1.7; 95% confidence interval (CI) 1.2, 2.6] and depression with comorbid anxiety disorder (COM) (OR 1.6; 95% CI 1.2, 2.1) were more likely in women having MA than in those with MoA. No stronger association was found for pure anxiety disorder (ANX) in MA vs. MoA (OR 0.9; 95% CI 0.7, 1.5). Among men, we found no difference in prevalence of depression and anxiety disorders between MA and MoA. This is a new finding that might have relevance for both research and clinical treatment.
Introduction
Anxiety disorders, depression and migraine affect many people. These disorders have a comparable prevalence of around 10–15%, and have in common that women are more often affected (1, 2). Population-based studies have provided consistent evidence for a positive association between migraine, anxiety disorder, and depression (3–6), but whether migraine with aura (MA) is more strongly associated than migraine without aura (MoA) is unclear. Although many of these studies have included detailed information concerning the diagnosis of anxiety disorders, depression and migraine, they have either not looked at MA separately, or had too small sample sizes to compare MA and MoA regarding their associations with anxiety disorders and depression. An epidemiological study of 1007 young persons from Detroit indicated that MA may have a stronger association with mental disorders than MoA (7), but the difference was not statistically significant. Genetic factors are important in both migraine (8, 9) and mood disorders (10, 11) and family occurrence and mode of inheritance indicate that MA is more determined by genetic factors than MoA (12). It would therefore be of interest to know whether MA is more strongly associated with anxiety or depression than MoA.
The HUNT study (13) included population data regarding symptoms of anxiety, depression, and migraine from 49 205 persons, and this sample size should have the statistical power to reveal a difference in these associations between MA and MoA. The aim of the present study was to evaluate the association between anxiety disorder/depression and MA/MoA based on data from the HUNT study.
Methods
The health study (HUNT-II) of Nord-Trøndelag County was conducted during 1995–1997 in a county that contained 3% of the Norwegian population. Everyone living in Norway has a unique personal identification number, and the registers of inhabitants in each municipality are continuously updated. The Central Population Register of Norway states that the validity and reliability of these local registers are excellent. According to these lists 92 566 individuals > 20 years old were eligible for HUNT-II. Of those eligible, 65 648 took part in the study (70.9%) after receiving a personal invitation with a specific appointment time at a central place in the municipality for physical examination. Two questionnaires including more than 200 health-related questions were administered to the participants. The first questionnaire (Q1) was enclosed with the invitation letter and brought to the examination, while the second questionnaire (Q2) was handed out at the examination to be filled in at home and returned by mail. Q1 included The Hospital Anxiety and Depression Scale (HADS) (14), and Q2 13 questions related to headache (15). These questions were designed to determine whether a person suffered from headache or not, the frequency of headache, and to diagnose MA and MoA according to a modified version of the migraine criteria of the International Headache Society (16). HADS showed 62 644 (95.4%) valid scores, and 51 383 (78.3%) participants responded to the headache questions. A total of 49 205 (75% of the respondents) gave valid answers to both HADS and the headache questions, and they constitute the current study population.
Weighting procedures
Weighting was performed to adjust for differences according to age and gender between attenders and non-attenders of HUNT-II, and for differences between the population of Nord- Trønderlag County and the population of Norway. The weighting procedure was based on the National population statistics of 1996, and was done according to the same procedure that was used in the National Comorbidity Survey (17). Crude numbers are based on unweighted data.
Migraine
Persons were classified as suffering from MoA if they reported migraine or fulfilled the following three criteria: (i) headache attacks lasting 4–72 h; (ii) headache with at least one of the following three characteristics: (a) pulsating quality, (b) unilateral location, (c) aggravation by physical activity; (iii) during headache, at least one of the following: (a) nausea, (b) photophobia and/or phonophobia. A diagnosis of MA was accepted in subjects who reported frequent visual disturbances prior to headache, if they otherwise fulfilled the criteria for MoA, even if they had attacks that lasted < 4 h. This classification of the subjects with migraine has been described in detail previously, and has been validated by interview-based diagnoses (15). Based on these diagnoses, the positive and negative predictive values for migraine in general, based on the headache questions, were 84% and 78%, respectively. For the diagnosis of MA, the specificity was 100% whereas the sensitivity was 42%. The 1-year prevalence of migraine and non-migranous headache has been described previously (18).
Anxiety and depression
HADS is a well-established self-rating instrument for anxiety and depression in which 14 items are scored on a 4-point scale from 0 (not present) to 3 (maximally present). The seven depression items (HADS-D) are related to anhedonia, lowered mood, and psychomotor retardation, and assess some of the depressive symptoms that are part of the diagnostic criteria of depression in ICD 10 (19) and DSM IV (20). The seven anxiety questions (HADS-A) cover some of the criteria of generalized anxiety disorder as described in both these diagnostic manuals.
Several studies and reviews have concluded that HADS is a reliable and valid instrument for assessing anxiety and depression in somatic, psychiatric and primary care patients (21–24). A recent study of the psychometric properties of the Norwegian version of HADS based on analyses of the HUNT population (N = 52 265) concluded that HADS was a suitable instrument for assessment in the general population also (25). In this study the two-factor structure for the anxiety (HADS-A) and depression (HADS-D) subscales was confirmed, and explained 57% of the total variance. A high correlation between the two subscales was found (Pearson's r = 0.55). Internal consistency, as measured by Cronbach's coefficient α, was 0.76 for HADS-D and 0.80 for HADS-A.
In accordance with empirical evidence (21), clinical significant anxiety disorder was defined by a score of ≥ 8 on HADS-A, and for depression ≥8 on HADS-D, as these cut-off values represented an optimal balance between sensitivity and specificity. Based on the HADS-A and HADS-D scores, we classified persons into one of four groups: (i) no disorder: scores < 8 on both HADS-A and HADS-D; (ii) pure depressive disorder (DEP): ≥ 8 on HADS-D and < 8 on HADS-A; (iii) pure anxiety disorder (ANX): ≥ 8 on HADS-A and < 8 on HADS-D; and (iv) comorbid disorder (COM): scores ≥ 8 on both HADS-A and HADS-D.
Data analyses
The prevalence of MA and MoA, and the four disorder groups were estimated for both genders. Using multiple logistic regression we calculated the odds ratios (ORs) with 95% confidence intervals (CIs) for DEP, ANX, and COM in persons having MA and MoA for both males and females. The OR expressing the difference in association between MA and MoA in individuals with DEP, ANX, and COM was also estimated. Potential confounding was evaluated by adjusting for age (decades from 20 years), and by years of education (three categories: < 10, 10–12, > 12 years). We also investigated potential interaction between gender and DEP, ANX and COM groups, respectively, by including the product of the two variables in the logistic regression analyses. The interaction coefficients were tested using the Wald χ2 statistic.
Data analyses were performed with SPSS version 11.0 (SPSS Inc., Chicago, IL, USA). A significance level of P < 0.05 was chosen, and all tests were two-sided.
Results
Table 1 presents the demographic data for the migraine and HADS groups. The total prevalence of migraine in the study population was 12.3% (MA 2.1%, MoA 10.2%). The prevalence in women was more than two-fold higher in both migraine groups (MoA: women 13.6%, 95% CI 13.2, 14.0 vs. men 6.3%, 95% CI 6.0, 6.6, P < 0.001; and MA: women 2.8%, 95% CI 2.6, 3.0 vs. men 1.2% CI, 1.1, 1.3, P < 0.001). ANX was more frequent in women (11.4% vs. 7.0%, P < 0.001), whereas DEP was more common in men (6.1% vs. 3.9%, P < 0.001). The prevalence of COM was close to equal between the genders, but the difference was still highly significant (5.1% in men and 6.4% in women, P < 0.001).
Characteristics of the HUNT population, N (%)
Crude numbers are based on unweighted data.
MoA, Migraine without aura; MA, migraine with aura; DEP, pure depressive disorder; ANX, pure anxiety disorder; COM, comorbid disorder.
The prevalence ORs for DEP, COM, and ANX in persons with MA and MoA are shown in Table 2, and a significant positive association was found for all three mental disorders, evident for MA and MoA and for both genders.
Prevalence odds ratios (OR) with 95% CI for anxiety disorder, depressive disorder, and comorbid disorder in individuals with migraine with (MA) and without aura (MoA) compared with migraine-free individuals (reference group)
Individuals with migraine with aura excluded from the reference group.
Individuals with migraine without aura excluded from the reference group.
DEP, Pure depression; ANX, pure anxiety disorder; COM, comorbid anxiety disorder and depression.
All numbers are adjusted for age and gender.
The prevalence ORs of DEP, ANX, and COM in persons with MA vs. those with MoA are shown in Table 3. Depression (DEP) (OR 1.7; 95% CI 1.2, 2.6) and depression with comorbid anxiety disorder (COM) (OR 1.6; 95% CI 1.2, 2.1), were more likely among women having MA than in those with MoA. No stronger association was found for pure anxiety disorder (ANX) in MA vs. MoA (OR 0.9; 95% CI 0.7, 1.5). Among men, no difference was found in prevalence of depression and anxiety disorders between MA and MoA.
Prevalence odds ratio (OR) with 95% CI for of anxiety disorder, depression and comorbid disorder in persons with migraine with aura compared with those having migraine without aura
DEP, Pure depression; ANX, pure anxiety disorder; COM, comorbid anxiety disorder and depression.
Gender difference is tested by an interaction term: DEP × gender, P = 0.159; ANX × gender, P = 0.553; COM × gender, P = 0.024.
All numbers are adjusted for age and gender.
Adjusting for educational level did not alter any results, and therefore results presented in Tables 2 and 3 are adjusted for age and gender only.
Discussion
In this cross-sectional study, we found a positive association between DEP, ANX, and COM and both MA and MoA, evident for both genders. Women with MA were more likely to have HADS-defined depression and depression with comorbid anxiety disorder compared with women with MoA.
The results of the present study should be evaluated with caution because it is questionnaire-based, and persons may be misclassified into the incorrect migraine group. The low sensitivity of the questionnaire considering MA makes this instrument unsuitable for determining the prevalence rates of MA in a population, since the true prevalence will be underestimated. The high specificity, however, indicates that those who fulfil the criteria of MA most probably have MA, and that the associations for anxiety and depression in MA are correct. Some persons who really have MA are probably misclassified in the MoA group, since the questionnaire is much better in detecting migraine headache than aura symptoms. In addition, some with MA may have somato-sensory types of aura symptoms that the questionnaire was not designed to detect. This probably means that any association between anxiety, depression and MA is underestimated in our study.
In accordance with previous studies (3–6), we found a positive association between migraine and depression and migraine and anxiety disorders. Merikangas (3), Breslau (5) and Swartz (6) found similar ORs expressing the association between migraine and depression (2.2, 3.6 and 2.25, respectively) and migraine and anxiety (2.7, 1.9 and 3.4). We have previously reported, from the HUNT study, ORs of depression (OR 2.7) and anxiety (OR 3.2) in migraine (MA and MoA combined) (4). The sample characteristics and methodology used in the HUNT study reveal the same associations between migraine, depression and anxiety as observed earlier, suggesting that this study sample is suitable for further subanalyses of the MA and MoA groups.
In the present study, we could not address the question of migraine comorbidity in specific mental disorders beyond HADS-defined anxiety disorder and depression, but we were able to make a distinction between migraine subtypes and to study its impact in both genders. Of particular interest is that we found a stronger relationship with DEP and COM for MA than for MoA, evident for women only. This is a new finding. However, in the study from Detroit MA was also more strongly associated with mental disorders than MoA, although this tendency was non-significant (26). The prevalence of major depression was 21.7% in MoA and 32.2 in MA and the prevalence of any anxiety disorder was 50.7% in MoA and 57.6% in MA. This study also included information concerning subtypes of affective and anxiety disorders in both MA and MoA. Only the risk of attempting suicide was analysed by gender. This OR was increased two-fold in both (26). Information from the Zurich study may also indicate that MA is more strongly associated with major depression than MoA, since 45.5% of those with MA had major depression over a period of 10 years compared with 31.9% with MoA, but the MA group included only 11 individuals (7).
It is unclear why only women had a stronger relationship with DEP and COM for MA than for MoA. Although both migraine types predominate in women, this gender-specific finding is unlikely to be an effect of a difference in statistical power, as there is no tendency of a stronger association with DEP and COM for MA in men. It is known that there is a relation between migraine and female sex hormones (27), and alterations in monoamine systems or channelopathy (28) are thought to be involved in the pathogenesis of both migraine (29) and mood disorders (30). In a recent population-based magnetic resonance imaging study from the Netherlands, it was found that patients with migraine had a higher prevalence of infarcts in the cerebellar region than controls, and this relation seemed to be most evident in patients suffering from MA (31). Recent studies have provided new information on the underlying mechanisms of MA with a sequence of events that lead from the cortex (aura phase) to the activation of trigeminal fibres (headache phase) (32). Looking for differences in the pathophysiology of migraine, anxiety and affective disorders in neurophysiolological, genetic or neuroimaging studies, related to the influence of female sex hormones, could be a way to explore our findings further.
In the present study the largest group of persons with migraine (women with MoA, 13.6% in HUNT) had the slightest association with both DEP (OR 1.3) and COM (OR 1.72). Men with MoA, conversely, were more affected by anxiety and depressive symptoms, independent of presence of aura symptoms. This impact of migraine on depression in men is in line with a recent study from Turkey (33), reporting that the presence of migraine increased the association with depression 2.5 times more in men than in women, since the prevalence of depression was 18.6% in women and 6.2% in men without migraine, and 32.8% in women vs. 31.7% in men with comorbid migraine (no significances given).
Several studies have shown that patients with migraine and major depression most often have a comorbid anxiety disorder (7, 34), and the ‘anxiety link’ has been described as a hypothetically important connection as to pathophysiology. The comorbidity of anxiety and depressive disorder is generally high (35), and anxiety is a prominent symptom in depressive episodes. HADS is a dimensional instrument, and permits presentation of patients with varying levels of anxiety and depression.
In our study, HADS-defined anxiety disorder alone (ANX) does not show any significant difference between MA and MoA. Our result indicate that the well-documented relation between anxiety and migraine headaches may be less important with regard to identifying causes separating MA from MoA.
The studies from Detroit and Zurich both included rates of migraine among persons with bipolar disorders. Our study did not include data on any axis I disorders as defined by ICD-10 or DSM-IV. Bipolar disorders have a strong genetic basis and we have earlier reported from a clinical study that there seems to be a preferential association between migraine and bipolar II disorder (34). It is known that these patients also have an increased prevalence of anxiety disorders and suicidal behaviour. It would therefore be particularly interesting to include questionnaires revealing mania and hypomania in further epidemiological studies of MA and MoA. Specific mental disorder comorbidity in migraine subtypes might have both research relevance and clinical implications that should be taken into consideration in the treatment of these patients.
Footnotes
Acknowledgements
The Nord-Trøndelag Health Study (the HUNT Study) is a collaboration between the HUNT Research Centre, Faculty of Medicine, Norwegian University of Science and Technology (NTNU), Verdal, The Norwegian Institute of Public Health, and Nord-Trøndelag County Council. The Norwegian Research Council, GlaxoSmithKline (Norway), Ltd, H. Lundbeck Ltd (Norway), the Bjarne Willmann's Foundation for Brain Research and the legacy of Gerda Meyer Nyquist Gulbrandson & Gerdt Meyer Nyquist, gave grants to this study.