Letter to the editor regarding proposed new diagnostic criteria for migraine

To the Editor:

We thank Torres-Ferrus and Pozo-Rosich, Kuan-Po Peng et al., and Guglielmetti et al. for their interest in our study, Proposed new diagnostic criteria for chronic migraine (1) and for their thoughtful comments (2–4). We are pleased that Torres-Ferrus et al. (5) agree with us that the third edition of The International Classification of Headache Disorders (ICHD-3) criteria for chronic migraine (CM) (6) do not account for patients who have a high frequency of migraine, and a high level of suffering. Also, that it is time for a revaluation of the existing ICHD-3 criteria for CM. We are sorry that the important article by Torres-Ferrus et al. escaped our notice, since it is highly relevant and represents a major effort to analyse the concept of chronic migraine. Torres-Ferrus et al. show that high-frequency episodic migraine (HFEM) does not differ practically from CM but differs a lot from low-frequency episodic migraine, suggesting that HFEM and CM should be lumped in a future definition of CM. Our cut-off for HFEM was eight or more days of migraine per month for more than 3 months, while the cut-off of Torres-Ferrus et al. was 10 or more days of headache per month. Since they do not distinguish between headache days and migraine days, it is difficult to compare the number of days. Most likely their cut off corresponds to the one suggested by us, since some of their 10 days are likely not migraine but tension-type headache. Guglielmetti et al. have made some interesting points about public health considerations of widening the definition of chronic migraine, which we shall elaborate on later in this response.

Based on our extensive, clinical semi-structured interview performed on all patients and the unique Danish registries, we showed that HFEM did not differ from CM with regard to demographic profile, migraine attack frequency, medication use, and social parameters, and the overall prevalence of comorbid diseases was similar in both groups. Furthermore, patients with HFEM purchased significantly more triptans than patients with CM. Finally, we found that the proposed new diagnostic criteria for CM doubled the number of patients with CM. These proportions were similar in a Russian population of medical students who had the same semi-structured interview as the Danish cohort.

Below, we answer some specific questions asked by Torres Ferrus and Pozo-Rosich (2).

1) In the headache field there are many patient related outcome measures (PROs) which can be used to evaluate disability and quality of life which have not been used. So, we believe that this study correlates migraine with costs, not disability.

There are valuable data on the issues in the paper by Torres-Ferrus et al. (5). We did not measure costs as they propose but a number of clinically and socially relevant issues.

2) We are not sure if both samples [Danish cohort and Russian cohort of medical students] can be compared as the Russian sample is smaller and from a different economic background.

The Russian population consists of medical students and serves to determine the generalizability of our results about the number of patients who fulfil the current and proposed criteria for CM, but not the results about the social impact. We find that the new modified criteria, which we suggest, approximately double the number of patients with CM in both the Danish cohort and the Russian cohort. It is important to show in different populations that the results are the same and we encourage future studies in this field.

3) Also, this study did not include low-frequency episodic migraine (LFEM) or controls in the analysis to compare.

Again, we refer to the study by Torres Ferrus et al. (5), which showed a big difference between LFEM and HFEM. Our aim was only to compare HFEM and CM.

4) In Chalmer et al., 8 days per month is arbitrarily established.

It is absolutely not arbitrarily decided. We chose the eight migraine days per month required in the diagnostic criteria for CM. Diagnostic criteria should always change as little as possible. The limit of eight days per month, which corresponds to two days a week, is well known by all clinicians. It is intuitive that such patients are severely affected by their migraine, and the data we generated support the limit of eight migraine days. To talk of headache days is, on the other hand, worrying because it does not distinguish between migraine and tension-type headache.

Below we answer some specific questions asked by Guglielmetti et al. (4).

1) Chalmer and colleagues did not report data on headache frequency in the two groups, which might therefore be either greatly different (e.g. 8–10 vs. 22–24 headache days/month) or little different (e.g. 12–13 vs. 16–18 headache days/month).

We distinguished between tension-type headache and migraine. We reported migraine frequency for the last year and migraine lifetime frequency, which did not differ between CM and HFEM.

2) Future studies refining the specific mechanisms of CM will likely contribute to define diagnostic criteria based on pathophysiology rather than numerical thresholds.

We agree that it would be wonderful to have specific mechanisms based on pathophysiology or clear biomarkers to define diagnostic criteria, but since these are not available we rely on diagnostic criteria.

3) As an example, we ran a secondary analysis based on data from 362 patients for two disability tools, the Migraine Disability Assessment (MIDAS) and the 12-item World Health Organization Disability Assessment (WHODAS-12), and a QoL measure like the Migraine Specific QoL (MSQ) (9). A total of 193 patients with low-frequency episodic migraine (LFEM), 95 with HFEM and 74 with ICHD-3-defined CM were compared using one-way ANOVA and Bonferroni post hoc test: Results show that patients with CM had 8–10 headache days/month more than those with HFEM. For all outcomes measures, patients with CM scored worse than those with HFEM which, in turn, scored worse than those with LFEM (Table 1).

The authors did not define LFEM and HFEM; thus, it is not possible for us to comment further on their table. But, of course, ICHD-3 CM patients had more headache days than HFEM patients, this is the premise for the diagnostic criteria of CM. We notice that the authors have recently co-authored studies where they discuss the validity of MIDAS and question whether it is a reliable tool when it comes to patients with CM and HFEM (7), they conclude that “the value of MIDAS seems problematic in those patients with high frequency migraine and CM, because patients are likely to approximate responses to MIDAS questions by multipliers of 5 or 10” (8).

4) Following the proposal of Chalmer and colleagues, if the category of CM is to include those that are now defined as HFEM, then the amount of persons that may require access to specific prophylactic treatments is reasonably expected to increase. This would make the cost of migraine disorders rise dramatically.

We acknowledge the authors’ concern regarding the possible economic hurdles that may come with increasing the population of patients with CM, but we do not agree that this should hold us, as clinicians, from providing the appropriate treatment options for our patients with a high frequency of migraine. As we show, they are clinically similar to and as disabled as patients with ICHD-3-defined CM. We strongly believe that diagnostic criteria should rely on clinical characteristics and not economic considerations.

5) Based on the data herein presented, the alternative hypothesis that was also accounted by Chalmer and colleagues; that is, to pursue new studies and research able to define HFEM as an independent entity, may be a viable path to follow.

To define HFEM as an independent entity would in our opinion complicate things and would not assure patients with HFEM the access to novel and existing drugs as well as reimbursements that are currently reserved for patients with ICHD-3-defined CM.

Below, we answer some specific questions asked by Kuan-Po Peng et al.³

1) The comparison of the demographic profile was limited to the following variables: Sex, age, age of onset, and self-reported provocation by triggers. Several variables previously shown to be different between patients with CM and episodic migraine (EM), including body-mass index, education, ethnicity, household income, or marital status, all have not been included in the current study for comparison.

Other studies have not separated EM into LFEM and HFEM and these variables are therefore not relevant to our study. Our purpose was to compare HFEM and CM. All patients in the Danish cohort were of Danish descent.

2) In the CM group, only 31% of patients had > 36 migraine attacks in the previous year. That means, 69% of the CM patients had on average three or more migraine attacks monthly over the previous year?

These proportions are referring to the number of migraine-specific days and not headache days. The proportions suggest that migraine days are clustered in periods and not evenly distributed throughout the year. This is in accordance with the current criteria for CM. These are patients with chronic migraine and not persistent headache.

3) The authors claimed that HFEM is comparable to CM in disability. The four disability measures in the current study are defined by the receipt of social security benefits, including early retirement pension, receipt of sickness benefit, cash assistance, and rehabilitation benefit. For the following reasons, this claim requires further evidence:

CM patients are in a higher proportion in receipt of an early retirement pension than HFEM patients (33.5% vs. 20.8%, p = 0.001).

The odds ratios for all four disability measures are actually similar, ranging from 1.59 to 1.78. The case number may be simply too small to have the power to allow any statistical difference for the following: Sickness benefit, 15 (9.9%) vs. 9 (5.7%), p = 0.207; cash assistance, 25 (16.4%) vs. 13 (9.2%), p = 0.139; rehabilitation benefit, 18 (11.9%) vs. 12 (7.5%), p = 1.

The claiming of social security benefits strongly depends on the social security system in a country and is influenced by many other confounding factors besides disability. An alternative of using Migraine Disability Assessment (MIDAS) or Headache Impact Test (HIT-6) would have been more convincing, whereas both have been well established for the comparison between EM and CM.

We agree that the influence of sociodemographic factors was moderately but not statistically significantly greater in CM than in HFEM. The difference may be caused by political/administrative reasons. The overall picture, taking all our results into consideration, was that there is no or only a slight difference between HFEM and CM. HFEM is a disabling disorder and should be lumped with CM.

4) When the efficacy is comparable and the patients with CM have definitely more migraine attacks than those with HFEM, why would they consume less triptans?

CM had more headache days but not more migraine days than HFEM. We can only speculate on why CM patients consume less triptans than HFEM patients as we have not asked the patients about this. But these data may suggest that because ICHD-3-defined CM is a mixture of migraine and tension-type-like headaches, while HFEM includes patients with pure migraine attacks, HFEM patients may benefit more from triptans and consequently purchase more triptans than ICHD-3-defined CM.

5) Whether potential confounding factors including the use of preventive medication or comorbid medication overuse headache (MOH) are comparable should be examined.

In our cohort, none of the CM or HFEM patients had MOH. Such patients were excluded. There was no difference in the treatment effect of preventive medication in the two groups.