Abstract
Aim
To describe the frequency and number of premonitory symptoms (PS) in migraine, the co-occurrence of different PS, and their association with migraine-related factors.
Methods
In this cross-sectional study, a validated questionnaire was sent to Finnish migraine families between 2002 and 2013 to obtain data on 14 predefined PS, migraine diagnoses, demographic factors, and migraine characteristics. The estimated response rate was 80%.
Results
Out of 2714 persons, 2223 were diagnosed with migraine. Among these, 77% reported PS, with a mean number of 3.0 symptoms compared to 30% (p < 0.001) and 0.5 symptoms (p < 0.001) among 491 persons with non-migraine headaches. Yawning was the most commonly reported symptom (34%) among migraineurs. Females reported PS more frequently than males (81 versus 64%, p < 0.001) and experienced a higher number of different symptoms (mean 3.3 versus 1.8, p < 0.001). All measures of migraine severity were associated with a higher burden of PS. Light and sound sensitivity showed the highest co-occurrence (kappa = 0.51, 95% CI 0.47–0.55). In a generalized linear model, age, gender, higher frequency, duration and intensity of headache, reduced working capacity, most aura symptoms, and associated symptoms of the headache phase were significantly associated with an increased in the number of PS.
Conclusion
PS are experienced by a majority of migraineurs. More severe migraine is associated with a higher burden of PS. Since the material was not entirely representative of the general population of migraineurs, caution should be exercised in generalizing the results.
Introduction
Migraine is the most common neurological disease and is characterized by intermittent headache accompanied by nausea, vomiting, photophobia, and phonophobia (1). The premonitory symptoms (PS) of migraine are defined in the International Classification of Headache Disorders, 3rd edition (ICHD-III beta), as “symptoms preceding and forewarning of a migraine attack by 2–48 hours, occurring before the aura in migraine with aura and before the onset of pain in migraine without aura,” with fatigue, elation, depression, unusual hunger, and craving for certain foods given as examples (1). However, various PS and timespans have been used in previous studies (2–8).
The pathophysiologic basis to the PS is incompletely understood, but the involvement of hypothalamic and dopaminergic mechanisms has been suggested (9,10). Recently, glyceryl trinitrate has been used to provoke migraine-like attacks as well as PS, allowing them to be studied directly, which might increase our insight into the pathophysiology of migraine (11,12).
Previous studies have reported that between 8% and 87% of adult migraineurs (4–7) and 67% of child migraineurs (8) experience PS. Varying study populations and definitions of PS might have contributed to the wide range of reported frequencies.
In a prospective study, it was shown that, in migraineurs with PS, the onset of headache attacks can be predicted several hours in advance (3). Early treatment guided by PS might be beneficial (13,14).
Our objectives were to determine the frequency and number of PS among persons from Finnish migraine families, to explore the co-occurrence of different PS, and to examine the association between PS and age, gender, and migraine characteristics.
Methods
We employed a cross-sectional design, using the validated Finnish Migraine Specific Questionnaire for Family Studies (15), which was issued in connection to genetic studies, to obtain data on PS, migraine diagnoses, demographic factors, and migraine characteristics among migraineurs in Finnish migraine families.
Migraine patients seen by neurologists at outpatient headache clinics in six Finnish cities with at least three first-degree relatives with possible migraine were invited to participate in the study. Those who accepted received the validated questionnaire, as did their affected relatives and the first-degree relatives of any affected individual. The relatives, however, were not seen by a neurologist and had no physical examination.
The headache diagnoses were based on the ICHD criteria (1) for migraine and consisted of migraine without aura (MwoA) and migraine with aura (MwA), with the subtypes being typical aura with migraine headache, typical aura with non-migraine headache, typical aura without headache, and hemiplegic migraine. Headaches not fulfilling any of these criteria were denominated as non-migraine headaches. Participants were recruited between January 2002 and March 2013 and the study size was determined by the number of participants during this period (n = 2714).
The 14 predefined PS in the questionnaire were based on previous studies and selected according to consensus in the research group (Figure 1). The participants were asked if some of the symptoms repeatedly preceded the headache attacks without any specified time window, giving a binary response for every symptom. The questionnaire also covered headache frequency, duration and intensity, associated symptoms of the headache phase (unilaterality of headache, pulsating headache, aggravation by physical activity, nausea and/or vomiting, photophobia, and phonophobia), aura symptoms (photopsia, blurring of vision, visual field defect, scintillating scotoma, hemiparesis, hemisensory aura, and speech aura), and disability (working capacity and lifetime number of attacks requiring bed rest). Demographic factors consisted of age and gender. Age was divided into four groups for group analysis: 5–17, 18–40, 41–65, and above 65 years.
The frequency of individual premonitory symptoms among migraineurs (n = 2219).
A composite variable, “sum of associated symptoms”, was created by adding together the associated symptoms, forming groups of persons with 0–2, 3, 4, 5, and 6 symptoms. The outcomes were the frequency of PS (i.e. the proportion of persons that experienced at least one PS) and the total number of PS per person.
Statistical analyses
Statistical analyses were performed using SPSS version 22.0.0. Numerical data were analyzed using the Kruskal–Wallis test and the Mann–Whitney U test due to non-normality. Categorical data were analyzed using the χ2 test. The co-occurrence of PS was tested using Cohen’s kappa with 95% confidence intervals (CI). A two-sided p-value < 0.05 was considered statistically significant. The p-values were not adjusted for multiplicity. In the case of missing data, the person was excluded from analyses involving that variable.
For the binary multiple logistic regression, we considered age, gender, self-rated working capacity, headache characteristics (frequency, duration, intensity, unilaterality, pulsating headache, nausea and/or vomiting, photophobia, and phonophobia), and aura symptoms to be valid predictors. The dependent variable was the presence of any PS. The following variables were not considered to be valid predictors: diagnosis, as their criteria were included in the analysis; aggravation by physical activity, due to the large number of missing data; and lifetime number of attacks requiring bed rest, in order to reduce the number of categorical variables. Non-significant predictors were removed one by one from the model, except for age and gender, which were kept to ensure adequate controlling for these variables. Multicollinearity was tested with a tolerance value threshold of 0.1. Correlations between predictors were tested using Spearman’s rank correlation and only one of any strongly correlated (rs > 0.8) predictors was included. Nagelkerke R2 was used to evaluate the goodness of fit.
A generalized multiple linear model with a negative binomial distribution with a log link function was used for the regression analysis of the number of PS, as it was determined to be the best fit for our data that severely deviated from normality. Valid predictors were the same as for the logistic regression, as was the procedure for creating the model.
Standard protocol approvals, registrations, and patient consent
The ethical committee of Helsinki University Central Hospital approved the study. The responders gave their informed consent to participate in the study. Each participant was assigned an individual identifier and the data were anonymized.
Results
In total, data were collected from 2714 persons, of whom 2223 fulfilled different migraine criteria. Four persons received the diagnosis of typical aura without headache and were excluded from all analyses except for the comparison between migraine and non-migraine headaches. Thus, the migraine sample consisted of 2219 persons (475 males and 1744 females, mean age 45 years, SD 17, range 5–96), of whom 1188 fulfilled the MwA criterion and 1031 fulfilled the MwoA criterion. A total of 137 persons (62 males and 75 females) were aged 5–17 years (mean 13 years, SD 3). Of the 491 persons who had non-migraine headaches, 206 were males and 284 were females, with a mean age of 41 years (SD 20, range 6–88 years). We did not track the number of questionnaires sent out, but estimate that the number was 3400. As we received 2714 completed questionnaires, we estimate that the response rate was approximately 80%.
Frequency and number of premonitory symptoms.
nmHA: non-migraine headache; MwA: migraine with aura; MwoA: migraine without aura; AwHA: typical aura with non-migraine headache; HM: hemiplegic migraine; MwA1.2.1: typical aura with migraine headache; HA: headache; associated symptoms: unilaterality of headache, pulsating headache, aggravation by physical activity, nausea and/or vomiting, photophobia, and phonophobia.
Yawning, mood change, and lethargy were the most common PS, each affecting about one third of all migraineurs (Figure 1).
Figure 2 shows the distribution of the number of symptoms experienced per person. Only 9% experienced more than seven PS (Figure 2).
The number of premonitory symptoms experienced per person among migraineurs (n = 2219).
Compared with persons with MwoA, those with MwA experienced PS slightly more frequently (79% vs. 75%) and reported a higher number of PS (mean 3.3 vs. 2.7) (p < 0.05) (Table 1). Among the subtypes of MwA, hemiplegic migraine was associated with the highest frequency (93%) and number of PS (mean 5.0), and typical aura with non-migraine headache was associated with the lowest frequency and number of PS (41%; mean 0.8) (p < 0.001). Females reported PS more frequently and with a higher number of different symptoms than males (Table 1). The frequency of PS in different age groups did not differ significantly (p = 0.25), but the difference in the number of different symptoms was significant, being the lowest in the 5–17 age group.
Higher levels of headache frequency, headache duration, headache intensity, working capacity, and lifetime number of attacks requiring bed rest were all associated with a higher frequency and number of PS (Table 1).
All of the associated symptoms of the headache phase (unilaterality of headache, pulsating headache, aggravation by physical activity, nausea and/or vomiting, photophobia, and phonophobia) were individually associated with a higher frequency and number of PS (Table 1). When adding together the associated symptoms, the frequency and number of PS increased with the number of symptoms.
Kappa coefficients for the co-occurrence of premonitory symptoms (n = 2219).
The highest values are shown in bold.
Predictors in the regression analyses.
Logistic regression (n = 1959).
Generalized linear model (n = 1911).
Not included in the model.
Exp (b) is the exponent.
In the generalized linear model, age, gender, headache frequency, headache duration, headache intensity, working capacity, unilaterality, nausea and/or vomiting, phonophobia, photopsia, blurring of vision, hemisensory aura, and speech aura were associated with an increased number of PS, while scintillating scotoma was associated with a decreased number of PS (Table 3). For instance, age increased the number of PS by 0.5% (95% CI 0.3–0.7%) per year (p < 0.001). The model was significant when tested against the null model (likelihood ratio χ2(21) = 756.5, p < 0.001, deviance/df = 1,21). Due to missing observations, 1911 valid observations were used in the analysis.
Discussion
This study is, to date, the largest study of PS, including altogether 2714 persons with headache. The frequency of PS of 77% among migraineurs in this material is within the previously reported range of 8–87% (4–7), comparable to the frequency found in 100 persons seeking a general practitioner for migraine headache (84%) (5) and 374 migraine patients from a neurology outpatient clinic (87%) (7), but higher than in a study of 484 migraineurs in the general population (8%) (4) and 893 at a tertiary care clinic (33%) (6). Among the 137 persons below 18 years of age, the frequency of PS was 71%, which is in line with a previous report of 67% in 103 pediatric–adolescent migraine patients (8). The mean number of different PS per person of 3.0 out of a possible 14 is similar to a previously reported mean of 3.2 symptoms out of 12 (7).
The most commonly reported individual symptoms in the present study, namely yawning, mood changes, lethargy, neck symptoms, and light sensitivity, have all previously been shown to be frequent, with some variation between studies (3,5–8).
As expected, PS were more than twice as common in migraine headaches as in non-migraine headaches. Persons with MwA reported PS significantly more often and experienced a higher number of PS compared to those with MwoA, which has not been reported previously (4,5,7,8). The differences were much larger between the subtypes of MwA, with hemiplegic migraine having by far the highest frequency and number of PS. This is in agreement with hemiplegic migraine being at the very end of the spectrum of migraine severity.
Sound and light sensitivity showed the highest co-occurrence, which might represent a common origin, as both symptoms are aspects of sensory hypersensitivity. Yawning showed a high co-occurrence with several other symptoms; for instance, lethargy, craving, and mood changes. In a previous study exploring co-occurrences of PS, associations between these symptoms were not as strong as those between depression and irritability, concentration problems, and fatigue (7).
All measures of severity (frequency, duration, intensity, working capacity, attacks requiring bed rest, and presence of associated symptoms) were related to a higher burden of PS. This is in accordance with clinical impressions and previous findings that severe pain is associated with a higher number of PS than mild–moderate pain (5) and that those who experience PS have a longer duration of headache and more nausea (i.e. more is more in migraine) (6). These factors also significantly and greatly affected the PS as predictors in the regression analyses. An unpredicted exception was that scintillating scotoma was associated with a lesser burden of PS.
In this material, a relatively large number of participants reported a typical headache duration of less than 4 hours. This may be due to the large number of children (in whom a shorter duration is more common) (16), the effects of successful treatment, or the uncertainty of self-reported time estimates. In addition, the ICHD criteria for MwA do not include a minimum duration (1).
The number of PS differed across age groups and age was a significant predictor in both regression analyses. Previously, in smaller studies, age has not been associated with PS (5,7,8).
As in this study, females have been associated with a higher number of PS previously (7), while there have been no gender differences in other studies (4–6,8). In the generalized linear model, females reported a 16% higher number of PS. Females have a higher prevalence of migraine and more severe migraines than males (17), thus our findings might be explained by a stronger migraine phenotype in females.
Strengths of the study include the vast material, the large pediatric–adolescent subset, and ample information on other characteristics of the migraines.
In terms of the limitations of the study, firstly, the material did not represent the general population because the questionnaire was only distributed to migraine patients and their relatives. While the study population did consist of persons from a wide range of ages with both genders being well represented, our sampling strategy caused the material to be skewed towards those with a larger hereditary burden and more severe migraine, which was shown in that MwA was more common than MwoA and in that hemiplegic migraine was considerably more prevalent than expected, at approximately one-tenth of all migraineurs compared to an estimated 1 in 10,000 in the general population (18). However, this allowed us to study the relationship between PS and measurements of migraine severity across their entire range.
Secondly, retrospective questions have the drawback of potential recall bias and lack the prospective recordings of the frequency and temporal aspects of the symptoms. We did not determine the frequency of the symptoms we considered to be PS during headache-free intervals, which is more readily done with a prospective approach, and this might have led to an overestimation of their rate, as several such symptoms have been shown also to be common outside the migraine attack (5).
Furthermore, all of the PS in the study were predefined, which could result in a lower reported frequency than with open questions in which participants are able to add their other observed symptoms. For instance, “face changes”, which in a previous report was the most common individual PS among children, was not included (8). However, open questions might have introduced unspecific symptoms and been unfavorable for determining co-occurrences.
In conclusion, PS are experienced by a majority of migraineurs, more commonly in females. A higher burden of PS is associated with a more severe migraine. Thus, further understanding of the pathophysiology of the premonitory phase is desired to enable the development of specific therapy for this early phase of migraine that might alleviate the subsequent attack.
Clinical implications
Increased knowledge of the occurrence of premonitory symptoms (PS) could help in understanding the pathophysiologic basis of migraine. Among 2223 individuals with migraine, 77% reported PS. More severe migraine was associated with a higher burden of PS.
Footnotes
Acknowledgments
Lars Söderström, statistician at the Unit of Research, Education and Development, Östersund Hospital, for statistical advice, and the following physicians for assisting in collecting patient data: Markus Färkkilä, Helsinki University Central Hospital; Markku Nissilä, Matti Ilmavirta, Suomen Terveystalo, and Erkki Säkö, Turku Headache Center and Suomen Terveystalo; Hannele Havanka, Länsi-Pohja Central Hospital; Heikki Teirmaa, Etelä-Pohjanmaa Headache Center; and Jarmo Liukkonen, Mikkeli Neurology Center.
Declaration of conflicting interests
The authors declared no potential conflicts of interest with respect to the research, authorship and/or publication of this article.
Funding
The authors disclosed receipt of the following financial support for the research, authorship, and/or publication of this article: Helsinki University Central Hospital, the Folkhälsan Research Foundation, The Academy of Finland (grant number 139795), Medicinska Understödsföreningen Liv och Hälsa.