Abstract
To investigate the possible association between migraine and left-handedness, we enrolled 100 patients with a diagnosis of migraine according to the International Headache Society diagnostic criteria and 100 age- and sex-matched control subjects into a case—control study. Handedness was determined by the Edinburgh Handedness Inventory. There was no significant difference in the frequency or grade of left-handedness between the two groups. Additionally, we pooled our data with those from five similar studies, which did not alter the result. Thus, neither our study nor the meta-analysis support Geschwind and Behan's hypothesis of an association between migraine and left-handedness.
Introduction
Geschwind and Behan (1) hypothesized that there is an association between left-handedness and several diseases, including migraine. They reported two separate investigations to evaluate this hypothesis. In both investigations, handedness was measured with a version of the Edinburgh Handedness Inventory (2), in which a laterality quotient (LQ) is determined that ranges between −100 for complete left-handedness and +100 for complete right-handedness. In Geschwind and Behan's first study, complete left-handers (LQ =−100) were compared with complete right-handers (LQ = +100) with respect to the presence of certain diseases. No details about the frequency of migraine could be given, because a proper distinction between migraine and other types of headache could not be made with the questionnaire used. In their second study, the LQs of patients with migraine or with immune disorders were measured and compared with the LQs of control subjects from the general population. Results showed that there was a higher percentage of left-handers among migraine patients than among control subjects. However, regarding the complete range of left-handedness (LQ from −1 to −100), this difference was not statistically significant (P < 0.1). The association became significant (P < 0.02) when different LQ values (<−30 and <−50) were used as cut-off points.
Geschwind and Behan's publication was followed by several studies using different approaches in order to investigate the association between migraine and left-handedness. In one case–control study, it was shown that the number of left-handers was the same in a group of 50 migraine patients as in a group of 50 control subjects (3). Messinger et al. (4) investigated the frequency of left-handedness among patients with migraine, tension-type-headache and cluster-headache. They did not find significant differences, either, by comparison of the different headache types with each other, or by comparison with the assumed prevalence of left-handedness in the general population of 10%. Additionally, they pooled their data with those from another survey, but again found no significant differences (5).
Chavance et al. (6) also performed a case–control study with patients suffering from different diseases, including migraine, and control subjects from the general population. No significant differences concerning handedness emerged from the investigation. Another case–control study was performed by Hering (7), who investigated patients with migraine and tension-type headache as well as healthy control subjects, but did not find a significant difference regarding their handedness either. However, Aygül et al. (8) found a significantly higher rate of left-handedness in male, but not in female migraine patients or in the overall sample of their case–control study. Guidetti et al. (9) investigated handedness in children with and without migraine, also using the Edinburgh Inventory, and observed random handedness, defined by a value between −70 and +70 on the Edinburgh scale, significantly more often in children with migraine than in the control group. However, no information is given on the prevalence of left-handedness.
Further studies have been published, in which various subpopulations were investigated regarding handedness and, mostly with questionnaires, regarding the presence of certain disorders in order to find out if the prevalence of the disorders varied depending on handedness. Most of these studies showed no significant differences with respect to handedness and migraine (10–13). However, Bryden et al. (14) showed that right-handed persons, who appeared to suffer from migraine, were less right-handed on a handedness scale than those who did not report having migraine headaches. When right-handers were compared with left-handers, though, there was no difference in migraine incidence. Hassler and Gupta (15) noticed in their study of musicians and non-musicians that male left-handers had a nearly significant (P = 0.07) higher incidence of asthma/allergies, migraine and myopia in a combined score than right-handers. Unfortunately, no details about migraine incidence alone are given.
The purpose of our study was to reopen the question of whether there is an association between migraine and left-handedness. Handedness is linked to hemisphere dominance. Hemisphere dominance, in turn, is influenced by hormonal factors. Prenatal hormonal status, especially the testosterone level, is considered to influence cortical lateralization (16). In addition, hormonal changes during the menstrual cycle have been shown to influence functional cerebral asymmetries (17). Hormonal factors are also known to play a role in the pathophysiology of migraine (18), which could be a link between the two conditions.
Methods
We included 100 consecutive patients with a diagnosis of migraine with and/or without aura from our supraregional headache out-patient clinic. The diagnosis of migraine was made based on a personal interview according to the International Headache Society (IHS) diagnostic criteria (19). We also enrolled 100 age- and sex-matched subjects who did not suffer from migraine as a control group. Control subjects were staff members of our department and waiting relatives of patients from the general out-patient clinic. Patients and control subjects were asked to complete a questionnaire that included the Edinburgh Handedness Inventory (2).
The mean LQ of the migraine patient group according to the Edinburgh Inventory was compared with that of the control group using the Mann–Whitney U-test. The number of left-handed subjects in each group was compared with Fisher's exact test. Values of P < 0.05 were considered to be significant. A sample of 100 cases results in a power of about 85% to detect differences of at least 10% in the prevalence of left-handedness. We considered this power as sufficient.
Additionally, an online search of databases was conducted with the aim of identifying other studies on the association between migraine and handedness whose data could later be pooled with the data of our study. Thus, the search aimed at case–control studies that were performed with migraine patients and control subjects without migraine from the general population, which compared the prevalence of left-handedness between the two groups.
The search strategy included an independent search by two of the authors (K.B. and S.E.) Keywords used were ‘migraine’ and ‘handedness’, as well as ‘migraine disorders’ and ‘functional laterality’. Databases searched were MEDLINE, EMBASE and Web of Science including publications from 1960 to 2007. All publications that resulted from this search were reviewed to, first, identify any studies that investigated an epidemiological association between migraine and handedness. Reference lists of these studies were examined in order to identify more suitable studies. Second, the studies were reviewed to identify those with a case–control design.
The quality of these studies was analysed regarding study outcome (association of migraine and left-handedness), assessment tools (both for left-handedness and migraine), power calculation (prestudy) and enrolment and matching of control subjects.
As the underlying statistical model for the meta-analysis, a fixed effects model was used. χ2 test and odds ratio (OR) including 95% confidence intervals (CI) were used to determine statistical significance.
Results
In our study, the migraine and control groups consisted of 87 female and 13 male subjects. The mean age was 40.1 years in the migraine group (
Comparison of migraine and control group with respect to left-handedness
Data presented as arithmetic mean with standard deviation.
CI, confidence interval; LQ, laterality quotient; OR, odds ratio.
Concerning the meta-analysis, 171 publications were displayed altogether by search of the databases. Among them, 16 studies were identified that investigated the association between migraine and handedness from an epidemiological aspect. Of these, eight studies were carried out using case–control design, all of them published in English. Review of the reference lists did not reveal any more suitable studies. Three of the eight studies were excluded from the meta-analysis, because they did not fulfil the inclusion criteria: two used patients with other primary headache disorders as control subjects, and one study compared the prevalence of random handedness, but not left-handedness in groups of children.
All five remaining studies were included in the meta-analysis (Table 2). The analysis of these studies revealed moderate quality of two studies, the Chavance study (no power calculation) and the Hering study (no power calculation), and poor quality of the Geschwind study (no validated migraine diagnosis tool, no power calculation), the Blau study (no validated assessment tools, no power calculation, no exact description of control subjects) and the Aygül study (no power calculation, insufficient matching).
Meta-analysis of case–control studies on the association of migraine and left-handedness
CI, confidence interval; IHS, International Headache Society; ND, no date; NS, not significant; OR, odds ratio.
The pooled data of the five identified case–control studies together with our own data did not show a significant association between migraine and left-handedness. The overall OR was 0.93 (95% CI 0.69, 1.25).
Discussion
None of the several subsequent studies, including the case–control study and meta-analysis introduced here, was clearly able to corroborate Geschwind and Behan's hypothesis. Thus, the majority of the data indicate that there does not seem to be a link between migraine and left-handedness.
Concerning Geschwind and Behan's publication, some methodological problems should be mentioned. First, the patient group is described as ‘carefully diagnosed patients with severe migraine’. Unfortunately, it is not mentioned which criteria were used to establish the diagnosis of migraine, or what the description ‘severe’ means. Was the frequency of attacks particularly high, or was it the pain intensity that was severe? Most of the subsequent studies that did not show an association used the IHS criteria for the diagnosis of migraine.
Concerning the results, it should be mentioned that in Geschwind and Behan's publication a clearly significant association has been shown only for sections of stronger left-handedness on the handedness scale. In all other studies, which determined an LQ by means of the Edinburgh Inventory (4, 5, 7, 8), data are only published for the whole range of left-handedness. We suppose that, if one of the other investigations had shown an association between migraine and stronger left-handedness, this would have been published. Therefore, either this aspect has not been investigated in the other studies, or the result was not significant.
One limitation of our study is the comparatively small number of subjects. However, we assume the power of the study to be sufficient to detect relevant differences in the rate of left-handedness. A further limitation of our meta-analysis is that there is certainly some heterogeneity between the studies that were included, and that these studies are only of moderate to poor quality. Only in part of the studies were the applied diagnostic criteria for migraine mentioned, and the assessment of handedness differed between studies. None of the studies was based on a power calculation. However, we found it acceptable to include all five studies in view of the otherwise very small number of studies and the fact that this is not a meta-analysis resulting in therapeutical suggestions, where, of course, the value of the included studies is of much higher importance. All authors agreed on this selection of studies for the meta-analysis.
Additionally, possible publication bias has to be considered, as negative results are less likely to be published than positive results. Since all of the studies that were included in the meta-analysis showed negative results, we do not think that a possible publication bias has a high influence on our data.
Our study and meta-analysis do not support Geschwind and Behan's hypothesis. However, they corroborate the majority of previously published studies that came to the conclusion that migraine and left-handedness are not associated.