- Top of page
- POPULATION AND METHODS
- Supporting Information
Background.— Various tools now exist to define migraine and its impact on everyday life. Those are scales that help diagnosing the disease and managing treatment. However, there is no such emphasis on defining individual migraine attacks and their variability. Yet, as a chronic disease with variable frequency and intensity among individuals, there should be a way of developing the management of the disease based on knowledge of the attacks.
Objectives.— To determine the variability of the severity of migraine attacks over time in individual patients using the MIGSEV scale.
Methods.— A set of migraineurs recruited in general practice filled a diary about their attacks during a 3-month period. The MIGSEV scale was used to determine the severity of their attacks, which could be mild, moderate, or severe. A Markov model was then adjusted with those 3 states and explanatory variables were considered as possible covariates.
Results.— Probabilities were estimated that enabled to expect the severity of an attack to come, given the severity of the previous one. For instance, no matter the profile of the patient, the severity of an attack had more chances of being similar to the previous one: 74% of remaining mild, 60% of remaining moderate, and 55% of remaining severe. When it was not similar, having an attack milder than the previous one was more likely than having a more severe attack. For example, the probability of going from moderate to severe was 0.105 whereas the probability of going from moderate to mild was 0.219. Covariates had no significant effect on estimated probabilities.
Conclusions.— An important aspect of the illness was defined here: the severity profile of patients. The MIGSEV is stable over time and severity of individual attacks represents a significant indicator in the characterization of the disease.
Migraine is a widespread and highly debilitating neurological disorder that was for a long time poorly characterized from a clinical standpoint. In 1988, the development of diagnostic criteria by the International Headache Society (IHS),1 revised in 2004,2 rep resented a major advance for the diagnosis of migraine and has also been important in standardizing entry criteria and outcome measures in clinical trials. In parallel to improved diagnosis, it has been important to identify and quantify the impact of migraine pathology on everyday life. To this end, a number of internationally validated scales have been developed in order to determine the impact of migraine on work productivity, health-related quality of life or more generally, burden of disease. Such tools can be used to optimize treatment strategies and to individualize these as much as possible to the state of each patient.
Most of these tools address the impact of migraine pathology in general but not the disability related to individual migraine attacks. In this respect, the MIGSEV3 scale, a 4-item questionnaire classifying the severity of migraine attacks based on patient selfreport, is particularly interesting. Scores on the MIGSEV scale are well correlated with other more global patient-reported outcome measures, and the severity of migraine attacks is now widely recognized as being relevant for the management of the disease.4 For instance, a study by Pradalier et al.5 demonstrated that healthcare expenditure for migraine is associated with greater disability and poor quality of life. The same study indicated that healthcare expenditure was also correlated with MIGSEV scale scores for the severity of the latest attack. This suggests that MIGSEV scores could have a more universal relevance for migraine pathology beyond the severity of individual migraine attacks.
The Migraine Disability Assessment (MIDAS) questionnaire,6 a patient-reported outcome measure that measures overall disability in the previous 3 months has been used in the Disability in Strategies of Care (DISC) study to stratify intensity of treatment according to MIDAS grade. This study demonstrated that such a stratified care strategy led to better clinical outcomes.7 Leading on from these findings, it has been suggested8 that the MIGSEV scale could be used to orientate stratified care such that patients would take a given treatment according to the severity of the attack. However, since migraine is a chronic disease whose intensity and frequency varies over time, it is important to determine how severity varies from attack to attack within a given individual in order to calibrate a severity-based stratified care proposal.
Currently, little information is available on the variation of the severity of migraine attacks over time. One study by Nachit-Ouinekh et al.9 has shown that certain symptoms defined in the IHS classification are relatively stable over time. In this study, criteria A (unilaterality, pulsatility, interference with activities of daily living and aggravation with effort) and B (nausea or vomiting, photophobia, and phonophobia) of the IHS classification were the most stable over a 12-month period. These criteria match 2 of the 4 items of the MIGSEV scale, namely interference with activities of daily living and nausea. This result would be consistent with the stability of attack severity over time.
The present study assesses the hypothesis that the MIGSEV scale could characterize illness severity over time for a given patient and thus be used to support a severity-base stratified care strategy for migraine management. A multi-state Markov model was used to assess stability of severity over time. Markov models are widely used in modeling outcome in chronic diseases, since these can easily be explained in terms of stages of progression. However, such models have never been applied to the severity of migraine attacks.
- Top of page
- POPULATION AND METHODS
- Supporting Information
The population of migraineurs evaluated in this study was representative of the patients consulting general practitioners for the management of their migraine. Data on migraine attacks were obtained from these patients during a 3-month follow-up period.
The principal finding of the study was that knowledge of the severity of one migraine attack could help predict the severity of the next headache. Notably, patients were more likely to experience a second attack of the same severity as the first one. The transition probabilities of experiencing 2 consecutive mild, moderate, or severe attacks were all above 50% and superior to those associated with the 6 other possible transitions. We also found that milder attacks were more probable than more severe ones and that the probability of having an attack less severe than the previous one was always higher than the opposite.
These findings were independent of the demographic and clinical profile of the patient. Although most of the potential explanatory variables tested were relevant since the univariate models were more likely than the null model, their effects on transition probabilities were found to be nonsignificant.
It is important to note that the majority of patients (75.9% by acute treatment, 23% prophylactic) in the study were treated. It would be ethically difficult to compare treated and untreated headaches, at least in a sample of consulting patients. Nevertheless, the lack of association between prophylactic treatment and severity transition probabilities suggests that differences between treated and untreated headache stability may be relatively small.
In addition, the lack of association between headache severity and acute treatment suggesting that physicians did not take into account the severity of individual headache attacks when deciding treatment. In the particular case of triptan use, this lack of association is at first sight surprising, especially since 75.9% of our population were prescribed these drugs. However, in order to observe a measurable association with triptan use, it would have been necessary to record whether a triptan had been taken and had relieved the headache before filling the headache diary. Without this information, it is not possible to conclude whether any such association exists. It thus would be interesting to repeat this study and collect this type of information.
Most of the migraine attacks in our dataset were of mild severity. Patients for whom the first attack reported was mild mainly experienced mild attacks thereafter. Even in patients experiencing a moderate or a severe attack, the probability that the next attack would be mild was 20% or more. However, frequent transitions were observed between moderate and mild attacks and moderate attacks were almost as frequent as mild attacks. In contrast, severe attacks were rare and less predictive of the severity of the following attack. The results obtained would be consistent with the hypothesis that some patients with migraine in primary care usually experienced mild attacks and from time to time switched between mild and moderate headaches. Occasionally, these patients might experience severe headaches but then return to a milder severity level for subsequent headaches. It would be interesting to conduct further studies to evaluate this hypothesis. The possibility of predicting headache severity more than 1 step ahead would also be interesting to explore.
This study has a number of limitations. First, data on several variables that are believed to be associated with headache severity were not collected. These include migraine with aura, psychiatric comorbidity, and headache chronicity. Such variables may be important determinants of severity transitions as well, and it would be interesting to evaluate this in future studies. Second, collection of information on severity from patient self-report may introduce a bias leading to overrepresentation of headache severity. However, it is not possible to obtain information on severity other than by self-report. If such bias does occur, it may be constant over the duration of the study for a given patient, and thus have a little influence on the estimate of severity transitions.
In conclusion, this study identified probabilistic patterns in the variability of migraine headache. The findings indicate that the MIGSEV could be used not only for characterizing individual attacks but also for identifying patterns of severity across the course of the disease.