{"paper_id":"10894ce3-1d86-42b4-b672-72f2383a825a","body_text":"Migraine is one of the most common neurological disorders ( 1 , 2 ). Both constitutional and environmental factors have been shown to be important for its clinical manifestations ( 3 ). Several studies have identified stress as a major trigger of migraine attacks ( 4–10 ). The stress response involves activation of the sympathetic nervous system and the hypothalamic–pituitary–adrenal axis, and it is generally associated with a subjective feeling of external or internal threat or demand ( 11 ). The intensity of this reaction is dependent on the duration, frequency, and severity of the stressors as well as the current health condition of the affected individual. Some stressors are a genuine part of human physiology, but many are largely contextually defined and dependent on culture, life situation, previous experiences, and personality factors ( 12–15 ). However, it has been shown that psychological stress plays an important role not only before the onset of migraine ( 9 , 10 ), but also in the maintenance of the disorder, the frequency of attacks ( 16 , 17 ), as well as the change from episodic to chronic migraine ( 18 ). It has also been shown that migraine is associated with several other health conditions of both somatic and psychological character ( 14 , 19 , 20 ). Furthermore, it is well recognized that there is a strong influence of the female sex hormones in this disorder ( 21 ).\nThe importance of personality traits in migraine has been widely debated. Claims have been made that migraineurs display increased neuroticism and anxiety and are anti-aggressive, but complex interactions regarding personality traits need to be considered ( 22–24 ). It may be that some of these personality traits also could be an effect of serious and insufficiently processed life events, as it is shown that migraine patients may have a history of maltreatment, especially during childhood ( 19 , 25 , 26 ). Furthermore, migraineurs display increased psychiatric co-morbidity such as post-traumatic stress disorders, anxiety disorders, and depression ( 15 , 19 ), which may be the result of constitutional as well as of environmental factors. Patients with migraine also have an increased co-morbidity with somatic disorders, especially chronic pain disorders such as fibromyalgia, arthritis, and orofacial pain ( 20 , 27 ) as well as an increased frequency of menorrhagia and endometriosis ( 28 ).\nWomen and men display differences in the prevalence of many disorders that could be considered to be stress-related ( 12 ), including migraine ( 1 , 2 ). Gender differences in prevalence and intensity of migraine may also be effects of gender socialization ( 29 ), as it is well known that the gender roles of women and men, both professionally and in private life, are strong determinants of their life situation ( 30 , 31 ).\nIn view of the complexity of the roles of personality, stress, life events, and gender in individuals with migraine, the aim of the present study was to examine and describe these aspects in a cross-sectional study of 150 migraineurs.\n\nThe study population comprised 150 consecutively enrolled adults (18 years of age and above), 106 women and 44 men, with moderate to severe migraine, defined as two or more migraine attacks a month. The subjects were recruited through advertisements in the local daily newspaper. All participants were interviewed by K.H. and C.M. A medical history was recorded regarding headache characteristics, stress as a trigger for migraine, concomitant illnesses, and general well-being, followed by a neurological examination performed by one of the researchers, C.M., a specialist in neurology, thus ensuring a uniform mode of data collection. The migraine diagnosis was confirmed on the basis of the International Classification of Headache Disorders ( 32 ). The study was approved by the Regional Research Ethics Committee, and informed consent was obtained. Eighty-three migraneurs from the same population participated in a further study applying a multimodal behavioral treatment program (33).\nIn connection with the interview, all participants answered questionnaires, which are described below, on a computer. Complete responses were obtained from all participants. The time required to answer these questionnaires was approximately one hour. All answers were directly transferred into the Statistical Package for the Social Sciences 18.0 (SPSS) program for further analysis.\nSSP was used to assess personality traits ( 34 ). This instrument is a revised version of the Karolinska Scales of Personality inventory, which was developed to identify stable personality traits of importance for psychological vulnerability ( 35 ).The SSP inventory comprises 91 items divided into 13 traits: 1)  Somatic trait anxiety  (autonomic disturbances, restlessness, tension); 2)  Psychic trait anxiety  (worry, anticipation, lack of self-confidence), 3)  Stress susceptibility  (easily fatigued, feeling of unease when urged to speed up); 4)  Lack of assertiveness  (lack of ability to speak up and to be self-assertive in social situations); 5)  Impulsiveness  (acting on the spur of the moment, non-planning, impulsive); 6)  Adventure seeking  (avoiding routine, need for change and action); 7)  Detachment  (avoiding involvement in others, withdrawn, schizoid); 8)  Social desirability  (socially conforming, friendly, helpful); 9)  Embitterment  (unsatisfied, blaming and envying others); 10)  Trait irritability  (irritability, lack of patience); 11)  Mistrust  (suspiciousness, distrust of people's motives); 12)  Verbal trait aggression  (getting into arguments, berating people when annoyed); and 13)  Physical trait aggression  (getting into fights, starting fights, hitting back). Each item is a statement, to which the participant responds by choosing an alternative on a four-point scale: ‘does not apply at all’, ‘applies to a certain extent’, ‘applies fairly well’, and ‘applies completely’. The SSP inventory results were standardized to normative data. The normative data were drawn from a random sample of the Swedish population, with the normative average for each trait defined as 50 and with a standard deviation of 10. Factor analysis for the instrument ( 34 ) showed one factor consisting mainly of traits 1–4 and 9, which was assessed as traits of neuroticism, another factor consisting of traits 8 and 10–13 was assessed as forms of aggressiveness, and a third factor consisting of traits 5–7 was assessed as an extroversion factor.\nFor evaluation of current stress, a questionnaire containing 75 questions, developed for clinical research at the Department of Environmental Stress Disorders (CEOS), Uppsala University, Sweden, was used ( 36–38 ). Forty-three items regarding stress-related psycho-social factors and stress coping, general health, and sleep were measured on a visual analog scale (VAS) graded from 0 to 100 (0 = low/no stress; 100 = maximal stress; Appendix); 38 of these 43 items refer to experiences of stress during the past month and 5 refer to the past or next year. The remaining 32 items concerned demographic data and concomitant illnesses. Items concerning the latter category were: Are you presently treated or have you previously been investigated or treated for: anxiety and depression, gastrointestinal diseases, allergies, cardiovascular diseases? Answering options were: never, yes previously, yes presently. In the further analyses the answers were dichotomized into: never or yes (previously and/or presently). Subjects were also asked if they suffered from tension-type headache.\nTo identify groups of the 43 stress-related items, a factor analysis with Varimax rotation was conducted on the current responses to the 43 items answered on a VAS scale. The eight factors with the highest eigenvalues were chosen to represent a measurement of current stress, all of which had eigenvalues above 1.0. These factors covered 29 of the 43 stress-related items (Appendix). Our designations of these factors were as follows and are listed in order of magnitude of their eigenvalues: Factor 1:  Feeling of meaningfulness regarding your home situation and your social situation ; Factor 2:  Mental symptoms ; Factor 3:  Subjective level of stress ; Factor 4:  Quality of sleep ; Factor 5:  General health ; Factor 6:  Difficulties with concentration and memory ; Factor 7:  Ability to prioritize recovery from stress ; Factor 8:  Ability to cope with stress.  The reliability of these eight components was tested using the Cronbach method, which yielded alpha values between 0.730 and 0.897 (Appendix).\nAs a measure of stress in the long-term, life events were evaluated. These were separated into events that had occurred during childhood/adolescence and during adulthood. The respective periods were defined by an age limit of 18 years. The instrument used ( 39 , 40 ) consisted of 11 childhood/adolescence life event categories and 21 adulthood categories, listed in  Table IV . The participants rated the impact of the events in the respective categories into three grades: ‘strongly negative’, ‘notably negative’, and ‘hardly negative at all’. For the last seven adulthood life event categories listed in  Table IV , a ‘positive’ response alternative was also provided. In the further analyses, the data were grouped into either event categories in which the participants had had experiences they rated as ‘strongly negative’ or into ‘any’ type of negative event, including all categories in which the rating was any of the following: ‘strongly negative’, ‘notably negative’, and ‘hardly negative at all’.\nDifferences in proportions were analyzed using the chi-square test. Mean differences were examined using either the  t -test or the Mann–Whitney test, depending on the distribution and sample size of the variables. Covariation was analyzed using correlation analyses, and the result was expressed as the Spearman's rho coefficient. Factor analysis was performed by means of principal components analysis with Varimax rotation. The number of latent variables was determined by using eigenvalues above 1.0 and scree plot analysis. Cronbach's alpha was used as a measure of the reliability of each component. All analyses were performed with SPSS 18.0 software. The significance level was uniformly set at 0.05, two-tailed test.\n\nThe distribution of participant age at inclusion is shown in 5-year intervals in  Figure 1 . Additional demographic data on age, body mass index, and education are presented in  Table I . The data were separated by gender.\nAge distribution at the time of inclusion of the studied women ( n  = 106) and men ( n  = 44) with migraine.\nDemographic profiles of women and men with migraine. Statistical comparisons were based on Mann–Whitney test and chi-square analysis.\nMigraine frequency, time since onset and age at onset of migraine, presence of aura and associated symptoms, pain degree, chronicity of symptoms, and menstrual correlation are presented in  Table II . All data were separated by gender, and sex differences were analyzed.\nClinical migraine characteristics and concomitant illnesses in women and men. Statistical comparisons were based on Mann–Whitney test and chi-square analysis.\na 15 attacks/month or more.\nMost notable gender differences, with up to double the frequency in women compared to men, were: associated nausea and osmophobia, tension-type headache, anxiety, and depression. Gender differences were seen also in age at migraine onset ( Table II ). The perceived role of stress as a trigger for migraine attacks was measured on a VAS scale 0–100 millimeters (0 = no impact; 100 = total impact). Women scored 75.6 on average (SD 24.1) versus men's average score of 58.5 (SD 29.7);  P  < 0.001 ( Z  = 3.7), Mann–Whitney test.\nPersonality traits were analyzed by gender and are shown in  Table III . Mean scores for personality traits that deviated significantly from normative means in both women and men were: high stress susceptibility, low verbal trait aggression, low physical trait aggression, and low adventure seeking. Women with migraine, but not men, also displayed significantly higher values than the normative reference for somatic trait anxiety and psychic trait anxiety. The most deviant personality trait for the entire study group was high stress susceptibility (data not shown).\nPersonality trait scores in women and men with migraine. Mean scores for the Swedish universities Scales of Personality inventory (SSP), compared to normative data (mean = 50; SD = 10), separated by gender. Statistical comparisons were based on one-sample  t -test.\na Controls were provided in the instrument (see Materials and methods section).\nA factor analysis of our data on personality traits yielded a neuroticism factor very similar to that of the normative data of the SSP instrument ( 34 ), with somatic trait anxiety (factor loading 0.56), psychic trait anxiety (factor loading 0.88), stress susceptibility (factor loading 0.79), lack of assertiveness (factor loading 0.81), and embitterment (factor loading 0.66) loaded on the neuroticism factor (eigenvalue 4.25). The reliability of these five scales was tested using the Cronbach method, which yielded an alpha value of 0.823.\nDetails of event categories rated as having had a ‘strongly negative’ impact are presented by gender in  Table IV , and Figure 2 summarizes data on life event categories according to the grouping into ‘any’ negative event and ‘strongly negative’ events, separated by gender and age period during which they occurred. Women reported life events in significantly more event categories and also significantly more event categories in which events had had a “strongly negative” impact, both during childhood/adolescence and during adulthood ( Table IV ;  Figure 2 ).\nLife events categories with a strongly negative impact in women and men with migraine. Statistical comparisons were based on chi-square analysis.\nLife events in women and men with migraine: the  y -axis shows the average numbers of life event categories reported, separated by ‘any’ negative event and events with a ‘strongly negative’ influence, as well as by gender and period in life in which they occurred. The Mann–Whitney test was used for statistical analysis of gender differences in average numbers.\nThe life events reported in the present study were also compared to those of a healthy control group (45 women and 30 men) presented in another study from the same time period and geographical region ( 40 ). When analyzing these data in relation to age-matched women ( n  = 57) and men ( n  = 24) from our study population, it appears that our migraineurs had experienced significantly more events—about twice as many. The respective ratios (present study versus controls in Marteinsdottir et al. ( 40 )) for women and men were: ‘any’ childhood/adolescence event, women: 1.7 ( P  = 0.006; Mann–Whitney test) and men: 1.9 ( P  = 0.015); ‘strongly negative’ childhood/adolescence events, women: 2.4 ( P  = 0.017) and men: 1.5 ( P  = 0.598); ‘any’ adulthood event, women: 2.0 ( P  < 0.001) and men: 2.6 ( P  < 0.001); ‘strongly negative’ adulthood events, women: 3.4 ( P  < 0.001) and men: 2.2 ( P  = 0.139).\nA gender-based correlation analysis was performed to discern interdependence between personality and stress as measured by life events or as current stress according to our CEOS instrument. For this purpose we selected the overall most deviant personality trait—stress susceptibility—which was compared to scores for negative life events and to factors of current stress in the present study, respectively: correlation analysis (Spearman) between stress susceptibility scores and number of life events revealed that these parameters were correlated in women, independent of the severity of the life events or the period in life in which they occurred ( Table V ). In men, a positive correlation was seen between stress susceptibility scores and the overall number of events during adulthood, which was not the case for events occurring during childhood/adolescence or for events rated as ‘strongly negative’ ( Table V ). With regard to the correlation between scores for stress susceptibility and scores for each of the eight factors of current stress, as defined by our factor analysis of the CEOS inventory results, women displayed significant positive correlations for seven and men for six out of eight of these factors ( Table V ). In both women and men, this correlation was strongest for the ‘difficulties with concentration and memory’ factor.\nCorrelations between stress susceptibility scores in women and men with migraine and A: scores for life events; B: scores for current stress. Correlations were analyzed using Spearman's rho coefficient.\n\nThe aim of the present study was to elucidate factors of personality and of past and on-going stress in migraineurs and the gender dependence of these factors. With regard to personality, we found that stress susceptibility was the overall most deviant trait. Women displayed significantly elevated scores also for the anxiety traits of the neuroticism cluster, and both sexes scored significantly lower for traits of aggression and adventure seeking than normative data. These profiles are in concordance with previous findings on personality traits in migraineurs ( 23 , 24 ). Regarding traits of neuroticism in women and men with migraine, Merikangas et al. ( 41 ) reported increased neuroticism in both sexes in a longitudinal study on migraine subjects between 19 and 29 years of age. Breslau et al. ( 42 ) showed that high neuroticism was associated with a higher incidence of migraine, which was more marked for women. High neuroticism levels were also found by Huber and Henrich ( 43 ), who studied women with migraine. The high values on the two neuroticism traits psychic and somatic anxiety for women in this study are in line with other reports where the researchers have found that migraine patients were more alexithymic, anxious, and depressed—especially those making frequent health care visits for their migraine (‘repeaters’) ( 44 ). Moreover, in another study the authors found that migraineurs displayed signs of low assertiveness, as evidenced by their avoidance of seeking social support to the same extent as healthy controls when they were experiencing stressful events ( 45 ).\nApart from traits of neuroticism, we found low average scores for traits of aggression and for adventure seeking in both sexes in the present study, which is in agreement with early data by Wolff, who suggested that migraineurs repress strong emotions, in particular aggression ( 24 ). It is possible that repressing feelings, especially those of anger, may increase the perception of stress, which in turn may affect the course of migraine. This personality-dependent mechanism could work in synergy with high stress susceptibility in causing a stronger impact of stressors in migraineurs.\nPrevious research speaks strongly in favor of the negative influence of stress on migraine ( 4 , 5 , 7 , 8 , 10 ). However, women and men display differences in their susceptibility to stressors ( 12 , 29 , 46 ). For example, claims have been made that women are more communally oriented and men have higher agentic, instrumental tendencies ( 46 ) and that women, therefore, are more reactive to social rejection challenges whereas men react more to achievement challenges ( 29 ). These differences, however, may be dependent also on social and cultural factors and may hence change with time ( 12 ).\nInterestingly, we found that our measurements of current stress were highly correlated with scores for the most deviant personality trait found, stress susceptibility. Intuitively, this appears logical, but we have not found any reference to this correlation in previous personality studies regarding migraine. It has been reported that migraineurs are not exposed to external stress to a greater degree than comparable controls ( 43 ), which would indicate that the personality-dependent differences in stress level observed are primarily due to how external demands are perceived. We find the correlation between stress susceptibility and different items of current stress worth discussing, because this correlation was strong in both women and men. In women a strong correlation to life events was also seen. The extent of these correlations further underscores the great importance of stress in migraine. A consequence of the fact that stress susceptibility is strongly correlated to parameters of stress is that either measurement—stress susceptibility or stress—can be used interchangeably in the individual characterization of migraine patients, which is similar to what Huber and Henrich have expressed ( 43 ).\nRegarding life events, obvious gender differences were seen in the present study. Overall, women reported approximately three times more experiences of strongly negative events. When separated by individual event category, higher female rates for such experiences were reported in all 32 categories, reaching statistical significance in 12 of them. These clear gender differences may reflect both differences in life situations for women and men and gender-dependent differences in how the events were perceived. For example, it is possible that life events related to the support of close relatives or friends are more negative for women because they face greater expectations to provide care. On the other hand, it seems reasonable that gender differences in perception of stress could explain why the women more often rated life events concerning their own disease, or the death of a close relative or friend, as having a strong negative influence. It is therefore possible that women are more sensitive to all kinds of stress. This is indicated by several studies showing that women experience more anxiety and depressive disorders, particularly in relation to different kinds of long-lasting perceived stress e.g. in fibromyalgia—another pain disorder that also is found to be related to stress and negative life events ( 39 , 47 , 48 ). Compared to healthy controls from the study by Marteinsdottir et al. ( 40 ), the participants of this study displayed approximately twice as many life events independently of time period in life and of gender. It has also been found previously that migraineurs experience much more negative life events during childhood than healthy persons ( 26 ). Furthermore, an independent support for a correlation between life events and migraine was provided by Peterlin et al. ( 15 ) in a study on post-traumatic stress disorders in migraine. However, these events were generally of greater magnitude than those reported in the present study.\nIt is well documented that migraine is concomitant with psychiatric disorders such as depression, anxiety, and post-traumatic stress disorder ( 15 , 19 , 49 ). It is notable that in the present study ongoing or a history of previous anxiety and/or depression were reported twice as frequently by the women as by the men. Again, it is possible that the high incidence of concomitant psychiatric disorders in migraineurs may worsen the migraine when these disorders have not been addressed and treated adequately. Tension-type headache was also considerably more prevalent in women than in men in the current study, which is also known from earlier studies ( 50 , 51 ). These gender differences in concomitant psychiatric/psychosomatic disorders are in line with the other observed gender differences in personality traits and in life events of this study and emphasize the importance of taking gender into account when evaluating migraine patients.\nThe number of male participants in our study was more limited compared to the number of females, despite special efforts to recruit men. Hence, low statistical power may have led to lack of detection of correlations in the male group as well as to detection of gender differences. Due to our mode of recruitment via advertisements in the local press, selection factors may also have influenced our results. Given its cross-sectional design, the study is descriptive in nature, and thus drawing conclusions about causality is precluded. Another weakness is the lack of a control group. Regarding life events, a healthy, age-matched control group from another study provided a possibility to compare our data, as described. With respect to personality we used normative data provided in the SSP instrument. Our own factor analysis of the SSP results yielded a profile similar to that described in the instrument, strengthening the validity of the instrument in our setting.\nThe present study confirms previous research showing that stress is an important factor in migraine. It indicates that high stress susceptibility is characteristic for migraine sufferers, and that stress susceptibility correlates with the individual's level of stress. Clear gender differences were found in personality traits and in the number of stressful life events, in particular those rated as strongly negative. Furthermore, anxiety, depression, and tension-type headache were twice as frequent in women. We conclude that stress susceptibility, life events, and concomitant disorders, especially of psychiatric nature, ought to be considered when investigating and treating individuals with migraine.","source_license":"CC-BY-4.0","license_restricted":false}