Background
The reported prevalence of chronic migraine in the population is 1.4-2.2.% [
1]. Patients with episodic migraine have an annual risk of 2,5% to develop chronic migraine [
2]. An intermediate headache frequency of 6 to 9 days per month and even more a critical frequency of 10 to 14 headache days per month increase the risk for chronicity [
3]. Further risk factors are obesity, stressful life events, snoring, and overuse of certain classes of medication. Up to 90% of migraine patients are able to name trigger factors like emotional stress, sleep deprivation, visual triggers (e.g. flickering lights) or hormonal changes that precipitate their attacks [
4,
5]. Triggers are equally named by patients with episodic and chronic migraine [
6]. The mode of action of these migraine triggers is as yet unknown. Several authors proposed trigeminal signaling mechanisms that sensitize certain brain areas, resulting in general sensitization of meningeal nociceptors and subsequently migraine pain [
7,
8]. It has also been suggested that the interaction of several triggers leads to migraine attacks [
9,
10].
An attentional bias towards pain-related cues, i.e. selective attention to pain-related information, is one of the psychological variables that are thought to be involved in turning an episodic into a chronic pain disorder [
11,
12]. The emotional Stroop paradigm is a means to assess this bias. In this task subjects have to name the color of emotionally relevant stimuli as fast as possible while ignoring the affective content of the stimuli. The underlying theory assumes that threatening (pain-related) stimuli draw the subjects' attention, and thus increase the latencies for colour naming compared to neutral stimuli or positive stimuli [
13,
14]. The opposite effect can occur when the affective content of the stimuli is avoided and color naming of negative stimuli therefore is speeded up [
15]. A recent metanalysis identified five studies using an emotional Stroop task with pain-related words that found an attentional bias towards sensory and affective pain words in chronic pain patients [
16]. Similar results were found in a study using personalized pain words [
17]. However, no such attentional bias was found in chronic headache patients [
18]. Thus, until now, there is no evidence of attentional biases in the Stroop task in patients with migraine. Other studies investigated chronic pain patients using the so called visual probe task and failed to find the expected attentional bias [
19‐
21]. In contrast, Liossi and collegues [
22] found selective attention towards pain related words in a group of headache patients.
We hypothesized that for migraine patients trigger related stimuli rather than pain related stimuli are threatening and thus cause an attentional bias. As most of the patients associate their attacks to trigger factors as reported above, it is likely that these trigger factors over time develop a threatening character for the patients. Therefore, we conducted two emotional Stroop tasks, one with general affective words and the second with social affective cues. In the latter, pictures of affective faces are used due to their higher ecological validity [
23]. An attentional bias towards negative affective stimuli was expected in both tasks for a group of subjects with very frequent migraine (FM, thus at higher risk for or already transformed to chronic migraine) compared to subjects with less frequent, episodic migraine (EM) and healthy controls. We hypothesized that FM patients should be faster in colour naming of neutral or positive stimuli than of negative stimuli, which would be reflected in higher stroop indices.
Discussion and conclusions
According to the results from the self reports, the migraine patients in this study, especially the patients with very frequent and chronic migraine, resulted to be highly disabled by their headaches. Moreover, they reported more anxiety, depressive symptoms, and worries than the healthy control group.
In task 1, no group effects for the ESI-N were found despite a significant difference in color naming latencies of neutral and negative words in the FM group. All groups reacted the same way, with slower reactions to negative than to neutral words. No group effect was found for the index comparing the reaction times to negative words with the reaction times to positive words (ESI-P), either. Because of the missing group effects, the results cannot be interpreted in favor of the hypothesis of an attentional bias for negative words in the FM group.
One reason for the weak results in task one might lie in the small sample size. With low numbers like in the present study it is only possible to detect strong effects. Smaller but possibly meaningful differences do not become statistically significant. Moreover, already several former studies doubted the use of words as stimuli, because words might not be strong enough to elicit significant attentional biases, especially in pain patients [
32‐
34]. Also, the words we used in the present study were general affective words. The use of words that reflect the common migraine triggers more specifically might lead to more pronounced effects.
In task 2, the controls showed a significant attentional bias to negative faces, whereas the FM group and the EM group did not. There was almost no difference between the reactions to neutral faces and the reactions to negative faces in the FM group, which was reflected by an index near zero. Comparing the color naming latencies for negative faces to those for positive faces, the effect became even stronger. The FM group here showed faster reactions to negative faces than to positive faces, with an index below zero, in contrast to the control group, which showed a high positive index.
The small indices of the FM patients, which in case of ESI-P is even negative, and the differences between FM patients and controls, could be an indicator of avoidance. Instead of resulting in the expected attentional bias towards negative social stimuli as reflected by angry faces, the subjects with FM showed unexpected avoidance behavior, i.e. they were faster in color naming for negative pictures than for neutral or positive pictures. Furthermore, an association of this avoidance with disease duration and migraine frequency, as reflected by significant correlations, could be found.
Our behavioral data support the self report data from a recent study. There, patients with chronic migraines reported significantly more avoidant coping strategies than patients with episodic migraines [
35]. These findings are in line with the assumptions of Martin [
36] and Martin and MacLeod [
8], who stated that general avoidance of headache triggers might be a part of migraine worsening. They claimed that the role of avoidance in migraine exacerbation might be twofold: first, trigger avoidance does not decrease but rather increases the sensitivity towards the triggers, and therefore leads to an increased probability of migraine attacks in response to triggers [
37]. Second, constantly avoiding situations that might possibly lead to migraine attacks may further result in loss of functional coping abilities in these situations and therefore may cause exactly the stress to the subjects they actually try to avoid [
38]. Some findings of Martin and colleagues [
38,
39] examining the relationship between sensitivity and exposure to certain triggers like stress, noise or visual disturbances, support the idea of heightened trigger sensitivity with avoidance.
Our results indicate that the advice to avoid personal trigger factors in subjects with frequent migraines might lead to automated and maybe dysfunctional avoidance behavior in the long term. From the anxiety literature it is known that behavioral and cognitive avoidance play a role in the maintenance of disorders [
40]. Clinicians should be made aware of the role dysfunctional avoidance behaviors could play in the process of migraine chronification.
As our study did not primarily aim to assess avoidance behavior, studies with psychological paradigms like the visual probe task, where both directions of attention can be investigated, would be needed to further strengthen this point. The relationship between trigger avoidance, dysfunctional coping strategies and disability in migraine patients merits further investigation.
Acknowledgements
We thank P. Pauli, P. Weyers, G. Alpers and M. Andreatta, Deparment of Psychology I of the University of Wuerzburg for support during the planning phase of the study.
A.P. was supported by a stipend from the Research Training Group „Emotions‟, RTG 1253/1 (German Research Fundation, DFG) and by research funds of the University of Wuerzburg. This publication was funded by the DFG and the University of Wuerzburg in the funding programme Open Access Publishing.
Competing interests
The authors declare that they have no competing interests.
Authors' contributions
AP conceived of the study, conducted the experiments and performed the statistical analyses. CS participated in the design of the study and its coordination and helped to draft the manuscript. All authors read and approved the final manuscript.