Abstract
Aim of this study was to determine whether menstrual attacks differ from non-menstrual attacks (NMA) as regards clinical features or response to abortive treatment in women affected by menstrually related migraine (MRM) referred to tertiary care centres. Sixty-four women with MRM were enrolled in a 2-month diary study. Perimenstrual attacks were split into three groups – premenstrual (PMA), menstrual (MA) and late menstrual (LMA) – and compared to nonmenstrual ones. Perimenstrual attacks were significantly longer than NMA. No other migraine attack features were found to differ between the various phases of the cycle. Migraine work-related disability was significantly greater in PMA and MA than in NMA. Acute attack treatment was less effective in perimenstrual attacks. Pain-free at 2 h after dosage was achieved in 13.5% of MA (OR 0.41; 95% CI 0.22, 0.76) vs. 32.9% of NMA. We concluded that, in MRM, perimenstrual attacks are longer and less responsive to acute attack treatment than NMA.
Introduction
Menstrually related migraine (MRM), i.e. the occurrence of migraine attacks preferentially but not only during the perimenstrual period, is a common form of migraine without aura (MO) in women, affecting about 40–70% of female migraineurs (1–6), while true menstrual migraine, i.e. the occurrence of migraine attacks only during menses, is far less frequent, accounting for 7–12% of women with MO (1, 3, 6). Many clinicians are under the impression (7–11), supported also by recent headache textbooks (12–16), that MRM attacks are more severe, longer-lasting and more refractory to both acute and prophylactic treatment than nonmenstrual ones.
Theoretically, the pathogenesis of menstrual and nonmenstrual attacks could differ, resulting in different clinical features or a different response to abortive and prophylactic therapy (17). Even though the pathophysiology of menstrual migraine is not yet fully understood, it is believed to involve abnormal neurotransmitter and neurohormonal responses (18–25) or abnormal release of prostaglandins (25, 26) triggered by the cyclical drop in oestrogen levels (27).
The feeling that menstrual migraine is more severe and more difficult to treat, however, is not fully substantiated by scientific evidence. In a recent study on a representative Dutch population sample, women with MRM reported that their menstrually related attacks were more severe, of longer duration, and more resistant to treatment (28). On the contrary, Stewart et al. in a rigorous prospective survey of 81 migraineur women from the general population (29), did not find significant differences between menstrual and nonmenstrual MO attack characteristics, except for a slight increase in the severity of attacks occurring the first 2 days of menses. Moreover, one randomized controlled trial of subcutaneous sumatriptan (30) and several retrospective analyses of phase II/III clinical studies (mainly involving the triptans) (31–36), have shown that menstrual and nonmenstrual migraine respond to acute attack treatment in a similar way.
The purpose of this study was to determine whether menstrual attacks differ from nonmenstrual attacks (in terms of clinical features or response to abortive treatment) in women affected by menstrually related MO referred to tertiary care centres, presumably a sample of patients reasonably representative of those seen by many headache specialty clinics.
Patients and methods
Patients
The study was carried out on women affected by MRM without aura consecutively referred for a first observation to five Italian headache centres.
Inclusion criteria were
childbearing age;
occurrence of regular menstrual cycles, i.e. 24–32 day cycles throughout the previous 6 months (women taking oral contraceptives were enrolled only if they had been taking the same compound for the previous 6 months and were expected to continue the treatment throughout the course of the study);
diagnosis of MO according to the criteria of the International Headache Society (IHS) (37);
a minimum 1-year history of MO;
the occurrence of two to six MO attacks per month during the previous 6 months;
the occurrence of at least one perimenstrual migraine attack per month during at least 5 of the previous 6 months.
For the purposes of this study, the perimenstrual period was defined, according to McGregor (38), as a 5-day period encompassing the first day of the menses (i.e. day 1 ± 2 of the menstrual cycle). A menstrual period was defined as 2 or more consecutive days of bleeding, according to the World Health Organization definition (39).
Exclusion criteria were:
assumption of a prophylactic treatment for migraine or any drug potentially able to modify the migraine course (beta-blockers, calcium antagonists, antidepressants, antiepileptic drugs, etc.) in the 3 months preceding the study;
occurrence of tension-type headache attacks (only if women could not reliably distinguish these from migraine attacks).
The study protocol and informed consent procedures were approved by the institutions’ Ethics Committees.
Data collection
Patients meeting the inclusion criteria and giving their written informed consent to participate in the study were instructed to complete a 2-month daily diary. In it, they were asked to state
the first day of their menstrual cycle (e.g. if the first day of a patient's last menses had occurred 13 days before the initial examination, she had to write ‘13’);
whether or not she was bleeding that day;
the occurrence of any migraine headache.
When a migraine headache did occur, patients had to fill in a diagnostic headache diary, detailing all the features (duration in hours, side, quality and intensity of pain, aggravation by routine physical activity, accompanying phenomena) used to diagnose MO according to the IHS criteria (37). Pain intensity was graded hourly on a 4-point scale, in which 1 = mild, does not impair daily activities; 2 = moderate, may inhibit, but does not prohibit daily activities; 3 = severe, prohibits daily activities; 4 = very severe, confines to bed. On the first day of each attack, information was collected about the following prodromal phenomena (occurring within the 24-h period preceding headache): anorexia, depressed mood, somnolence, nausea, yawning, asthenia/fatigue, impaired concentration, irritability, craving for food.
Another section of the headache diary investigated lost work due to migraine: both work hours missed and hours worked with reduced efficiency because of migraine.
The last section of the diary collected information about the use and efficacy of symptomatic drugs. For each attack, the number, dosage, route of administration and commercial name of the medications employed had to be recorded. Patients were also required to note headache severity (on a 5-point scale, in which 0 = no headache, 1 = mild headache, 2 = moderate headache, 3 = severe headache, 4 = very severe headache) immediately before and 2 h after the administration of the first symptomatic drug used, and both the occurrence and number of recurrences.
Diaries were collected in at two subsequent visits, 30 and 60 days after the initial enrolment. They were reviewed by the investigators together with the patients; incomplete and inconsistent items were discussed and, if possible, corrected. To ensure the highest possible uniformity of diary compilation and interpretation, only one investigator per centre (FG, GS, GA, BB, AT), previously instructed at a full-day dedicated meeting, was allowed to follow the patients and review the diaries. Moreover, before starting the survey, 10 trial diaries (two per centre) filled in by women not participating in the study, were thoroughly discussed by the five above-mentioned investigators and ambiguous items were changed. Finally, throughout the 10-month (from March 1 to December 31 2000) recruitment period, any questions that arose were put to a web-based steering committee (FG, REN, CB, EF, GN), which promptly provided answers and clarified doubts.
Data entry and analysis
Entry of the diary data into an electronic sheet (Excel 2000, Microsoft) was always performed, at each centre, by the investigator who had previously reviewed the diaries. Headache was considered to be MO only if the IHS diagnostic criteria were satisfied (on at least 1 day in the case of multiday attacks). If a migraine attack lasted more than 1 day, it was considered a single multiday attack, unless an interval of 24 fully pain-free hours separated headache on 2 consecutive days. For instance, if headache started at 0800h and ended at 1600h on day 1 and recurred at 1000h on day 2, it was defined as a single attack, but if it recurred at 1800h on day 2, two separate attacks were counted.
The objective of the analysis was to determine whether migraine features and response to symptomatic drugs differed between menstrual and nonmenstrual attacks. To define the boundaries of perimenstrual attacks, we undertook a univariate analysis, comparing the characteristics and response to acute treatment of attacks starting from 7 days before to 6 days after the first day of menses to those of attacks starting on the other days of the menstrual cycle. In this way we established four types of migraine attack
premenstrual attacks (PMA) starting the 2 days before the onset of bleeding;
menstrual attacks (MA) starting the first 2 days of the menstrual cycle;
late menstrual attacks (LMA) starting from day 3 to day 7 of the menstrual cycle;
nonmenstrual attacks (NMA) starting on any other day of the menstrual cycle.
The following features of attacks were noted:
Duration (in hours);
Severity of pain (we only considered attacks in which, at some point, the pain was described as severe or very severe);
Occurrence of associated phenomena (photophobia, phonophobia, nausea, vomiting);
Occurrence of prodromal phenomena (in the 24 h preceding the headache onset);
Use of symptomatic drugs and number of drugs used;
Occurrence and number of recurrences.
Work-related disability (attacks occurring during paid work) was measured on the basis of work hours lost because of migraine and hours of reduced efficiency when working during a migraine attack. A composite measure of work-related disability – lost work hour equivalents (LWHE) – was calculated, combining information on missed hours and hours at work with reduced efficiency due to migraine. This was done using the following formula (40):
LWHE = W1i + W2i(1-Pi)
in which W1i = the number of hours that could not be worked because of migraine; W2i = the number of hours spent working with a migraine attack; Pi = percentage residual efficiency when working during a migraine attack.
The patients were instructed to take their usual acute attack treatment when the pain reached at least grade 2 on the 4-point scale mentioned earlier. No rescue drugs were allowed within the first 2 h postdose. The efficacy of the first symptomatic drug used in each attack was evaluated using a series of parameters
Headache response at 2 h (the percentage of patients whose headache severity was reduced from very severe, severe or moderate at baseline to mild or none 2 h postdose);
Pain-free at 2 h (the proportion of patients whose headache severity was reduced from very severe, severe or moderate at baseline to none 2 h after taking a dose);
Recurrence (the percentage of patients who, having obtained a response or freedom from migraine by 2 h postdose, developed a worsening to moderate/severe/very severe pain within 24 h of the initial dose);
Sustained headache response (the percentage of patients who, having obtained a response or freedom from migraine by 2 h postdose, did not have a recurrence and did not use any rescue medication within 24 h of the initial dose);
Sustained pain-free (the proportion of patients who, having obtained freedom from migraine by 2 h postdose, did not have a recurrence and did not use any rescue medication within 24 h of the initial dose).
Univariate and multivariate (logistic regression) analyses were performed to compare the characteristics of migraine attacks during premenstrual, menstrual and late menstrual periods vs. the nonmenstrual period, taken as reference. Separate models were fitted to estimate the differences between migraine features and prodromal phenomena. Because we had multiple observations for each subject, repeated measures logistic regression was used to account for intrasubject correlation. All analyses were performed with the SAS (Cary, NC, USA) package (release 8.0).
Results
Subject participation and characteristics
One hundred and nine eligible women were referred to the participating centres for a first consultation during the recruitment period. One hundred (91.7%) of them agreed to participate in the study. Eighty-two women (75.2% of the eligible subjects) completed the 2-month diary, while the others were lost to follow-up (n = 7) or did not return the diary (n = 11). Another 18 women were subsequently excluded for various reasons: four did not experience a menstrual period or had irregular cycles, four had more than six attacks per month, four did not experience perimenstrual attacks, and six returned inadequately completed diaries. Thus, the final sample numbered 64 women (58.7% of the eligible subjects). No significant differences in age, marital status and level of education emerged between the 64 participants and the 45 women who refused to participate, were lost to follow-up or excluded for other reasons.
The 64 women included in the analysis had a mean age of 34.7 ± 6.3 years (range: 21–47 years), were mostly married (n = 50; 78.1%) and had a high level of education (n = 48; 75.0%). Thirteen women (20.3%) were taking oral contraceptives.
Migraine attack characteristics by phase of cycle
In the 2-month study period, the 64 women reported a total of 459 MO attacks, 231 of these (50.3%) being nonmenstrual and 228 (49.7%) perimenstrual attacks. PMA (n = 60) accounted for 13.1% of the whole number of attacks (26.3% of the perimenstrual attacks), MA (n = 78) for 17.0% (34.2% of the perimenstrual attacks), and LMA (n = 90) for 19.6% (39.5% of the perimenstrual attacks).
The attack duration was significantly shorter in NMA (16.2 ± 15.1 h) than in PMA (29.6 ± 24.1 h; P < 0.0001), MA (33.7 ± 24.8 h; P < 0.0001), and LMA (24.0 ± 21.8 h; P = 0.002). Table 1 details the occurrence of status migrainosus, defined according to the IHS criteria (37), the severity of the pain and the associated phenomena reported in the four phases of the menstrual cycle. Status migrainosus occurred more frequently in PMA and MA than in NMA. The features of the migraine attacks did not differ in the various phases of the cycle, except for the presence of a greater severity of pain in LMA.
Crude and adjusted OR of the features of migraine attacks in the various phases of the menstrual cycle. Non-menstrual attacks (occurring on any of the days but those included in the premenstrual, menstrual and late menstrual phases) were taken as reference
Defined, according to the International Headache Society criteria (37), as an attack of migraine lasting more than 72 h despite treatment,
Attacks in which, at least at some point, the pain was perceived as severe or very severe,
Attacks in which, at least at some point, the pain was perceived as very severe,
OR (95% CI) excludes 1.0.
The prodromal phenomena reported by the patients are detailed in Table 2. Some prodromic symptoms were more frequent in perimenstrual attacks than in NMA, in particular irritability (in PMA and MA) and fatigue (in MA and LMA). Nausea, on the contrary, occurred less frequently in LMA than in NMA.
Crude and adjusted OR of prodromes (occurring in the 24 h preceding headache onset) of the migraine attacks in the various phases of the menstrual cycle. Non-menstrual attacks (occurring on any of the days but those included in the premenstrual, menstrual and late menstrual phases) were taken as reference
OR (95% CI) excludes 1.0.
Drugs for the acute treatment of migraine attacks were taken in 437 out of 459 attacks (95.2%), in particular in 94.6% of PMA, 93.6% of MA, 100% of LMA, 93.7% of NMA (LMA vs. NMA: P = 0.01). In most attacks patients used NSAIDs; triptans were utilized in 26.4% of the attacks, without significant differences emerging between the various phases of the cycle. No patients used ergot derivatives. The mean number of symptomatic drug doses taken per attack was higher in perimenstrual attacks (PMA = 2.0 ± 1.4, p = 0.06 vs. NMA; MA = 2.3 ± 1.4, P < 0.0001 vs. NMA; LMA = 1.9 ± 1.2, P = 0.1 vs. NMA) than in NMA (1.6 ± 1.1). The mean number of recurrences per attack (n = 415) was 0.8 ± 0.8 in PMA (P = 0.02 vs. NMA), 0.9 ± 0.9 in MA (P = 0.001 vs. NMA), 0.5 ± 0.9 in LMA and 0.5 ± 0.9 in NMA.
Migraine work-related disability by phase of cycle
Three hundred and seventy-seven attacks occurred during paid work. Work hours were lost because of migraine in 51.3% of PMA (P = 0.006 vs. NMA), in 27.3% of MA, in 27.0% of LMA and in 27.3% of NMA. The mean number of lost work hours was 2.7 ± 3.4 in PMA (P = 0.02 vs. NMA), 2.1 ± 4.0 in MA, 1.0 ± 2.0 in LMA and 1.2 ± 3.2 in NMA. LWHE totalled 4.8 ± 3.5 in PMA, 8.6 ± 15.2 in MA (P = 0.05 vs. NMA), 4.2 ± 3.9 in LMA and 4.4 ± 4.4 in NMA.
Efficacy of the first symptomatic drug by phase of cycle
Efficacy of the first symptomatic drug could be analysed in 415/437 attacks (95.0%). In the remaining 22 attacks, data were incomplete and/or an additional medication was taken within 2 h postdose. Table 3 illustrates headache response at 2 h, pain-free at 2 h, recurrence, sustained headache response and sustained pain-free of the first drug taken during migraine attacks in the various phases of the cycle. Drugs for the acute treatment of migraine were less effective in PMA and MA than in NMA. Pain-free at 2 h was achieved in 23.1% of the PMA and in 13.5% of the MA, compared to 32.9% of the NMA, while sustained pain-free was attained in 7.7% of the PMA and in 5.4% of the MA, compared to 24.2% of the NMA.
Percentages and OR of the measures of efficacy of the first symptomatic drug taken during migraine attacks in the various phases of the cycle (n = 415). Non-menstrual attacks (occurring on any of the days but those included in the premenstrual, menstrual and late menstrual phases) were taken as reference
OR (95% CI) excludes 1.0.
In women taking oral contraceptives, the characteristics of the attacks and the response to acute attack treatment in the various phases of the cycle did not differ to those in women not using them (data not shown).
Discussion
Although MRM is very frequent, there is no unanimous consensus as regards its definition. MacGregor, in her prospective diary study of a small sample of female migraineurs (1), defined as MRM those migraines that occur at any time of the month, but show an increased attack frequency within the menstrual period. Silberstein (41), however, pointed out that, applying MacGregor's definition, some women with regular attacks during the menstrual period but also frequent attacks outside it, would not show an evident increase in attack frequency during menses and therefore should not be considered to have MRM. He proposed, instead, that the definition MRM should cover migraines triggered by menstruation on a regular basis but also occurring at other times of the month. In our study we adhered to this concept, including women with at least one perimenstrual attack per month in at least 5 of the previous 6 months, but also with one to five NMA per month. Our idea was, in fact, that in MRM the menstrual period is not an aggravating factor, but a triggering factor occurring on a regular basis. Having defined a conceptual framework for MRM, the first difficulty we faced was to define menstrual attacks. In the literature, in fact, the prevalence of menstrual migraine varies enormously (from 4 to 73%), mainly due to the lack of a standard definition of the timing of menstrual attacks (38). The time frame for menstrual attacks proposed by the IHS headache classification (from 2 days before menses to the last day of menses) (37) seems to be of little value, since the duration of bleeding depends on many nonhormone-related factors. To be conservative therefore we chose to include in our study women with menstrual attacks occurring on day 1 ± 2 of the menstrual cycle, according to MacGregor's strict definition (38). Given the lack of sound pathophysiological data, however, these limits are rather empirical and primarily based on the increased frequency of MO attacks during the 2 days prior to menstruation and, to a greater extent, the first 2 days of the menstrual cycle, a trend first observed in 55 female migraineurs from a headache centre (who were studied over three menstrual cycles) (1) and recently confirmed in a sample of 81 migraineurs from the general population (2). Other prospective studies, however, have given different results. Dalton (42) recording 512 migraine attacks in 52 women, found the peak incidence of migraine on the 2 days immediately before the onset of menses, while Johannes et al. (43), studying 74 women affected by migraine with aura, observed an increased number of MO attacks only during the first 3 days of menstruation. In our sample of women affected by MRM, we found an increase in attack number from day −3 to day +6 of the menstrual cycle, with three peaks on days − 2, +1 and +2 (Fig. 1). Moreover, in general practice many women maintain that they often experience severe MO attacks before, during, and after the menses. To define the boundaries of menstrual attacks in our patients, and having identified attack duration as the main parameter able to distinguish perimenstrual from nonmenstrual attacks, we performed several analyses (data not shown) applying different temporal limits – from day 7 before menses to day 7 of the menstrual cycle. We found that the duration of the attacks increased a couple of days before menstruation (from 19.4 ± 15.2 h on day − 3 to 30.8 ± 25.6 h on day − 2) and decreased sharply 1 week after the onset of bleeding (from 22.0 ± 24.8 h on day + 7 to 7.2 ± 4.2 h on day + 8). Having established the boundaries of perimenstrual period, we established three subgroups of perimenstrual attacks – premenstrual, menstrual and late menstrual, as defined in the patients and methods section – that could have different underlying pathophysiological mechanisms.
Days of the menstrual cycle with the highest incidence of migraine without aura attacks according to our and other studies. — peaks of attacks; —— days with higher incidence of attacks than that recorded in the non-menstrual period.
As mentioned above, attack length was the main clinical feature distinguishing perimenstrual from nonmenstrual attacks and, in accordance with clinicians’ observations in daily practice (8–11), it was found to be higher in the attacks occurring around menses. Accordingly, the mean number of recurrences per attack was higher in PMA (0.8 ± 0.8) and MA (0.9 ± 0.9) than in NMA (0.5 ± 0.9). Visser et al. (8), analysing the determinants of headache recurrence after sumatriptan use, found that 63% of female migraineurs who complained of constant or almost constant recurrences after subcutaneous sumatriptan had MRM compared to 26% of patients not experiencing recurrences. Even though the mechanism of recurrence after acute attack treatment is not fully known, they attributed the higher rate of recurrences in MRM patients to their longer and more severe attacks, rather than to hormonal or pharmacokinetic factors (8). Not unexpectedly, the number of attacks that fulfilled the IHS criteria for status migrainosus (attacks lasting more than 72 h whether treated or not with continuous headache or free intervals lasting less than 4 h) (37) was higher in PMA and MA (Table 1), probably reflecting the longer mean duration of these attacks.
In accordance with Stewart et al. (29), we found that the characteristics of pain, such as unilaterality, throbbing quality, aggravation by routine physical activity (data not shown), and the accompanying phenomena (Table 1) were similar in perimenstrual and nonmenstrual attacks, demonstrating that perimenstrual attacks are qualitatively identical to nonmenstrual ones. Multivariate analysis showed pain severity to be significantly increased only in LMA vs. NMA. Other clues, however, indicate higher severity of perimenstrual attacks. First of all, migraine-related disability was higher both in PMA (in terms of missed work hours) and in MA (in terms of LWHE) than in NMA, even though we cannot exclude that symptoms associated with premenstrual syndrome or menstruation itself may have rendered the migraine less bearable and resulted in women stopping work or in a further reduction of their efficiency. Second, more symptomatic drug doses were needed to treat MA than NMA and more recurrences occurred both in PMA and in MA than in NMA. Third, the efficacy of acute attack treatment was lower in perimenstrual attacks, especially MA (see below). We can also hypothesize that our 5-point scale for pain severity (see methods) was not sensitive enough to detect possible differences between the various phases of the cycle, since all the attacks were deemed quite severe or worse (75% of all the attacks were rated as severe or very severe). A more detailed scale, e.g. the visual analogue scale, would probably have been better able to detect small differences.
Some prodromic phenomena, in particular fatigue and irritability, were more frequently reported in perimenstrual than in nonmenstrual attacks. It is difficult to establish, however, the extent to which these symptoms were related to the migraine attacks themselves rather than to the premenstrual/menstrual symptomatology.
At variance with the literature (30–36), perimenstrual attacks, in particular MA, seemed in our study to be less responsive than NMA to acute attack treatment. Even though our study was not a randomized controlled trial, and therefore the results must be treated with caution, we believe that they may reflect a real relative refractoriness of perimenstrual attacks to symptomatic drugs, at least in severely affected MRM patients. This divergence from the positive results presented in the literature – we must recall that only one study of subcutaneous sumatriptan was a randomized controlled trial (30), while the others were posthoc analyses of previous trials and therefore not randomized for menstrual attacks – could be explained by several factors:
While we studied a sample of MRM patients, comparing perimenstrual and nonmenstrual attacks, literature trials considered menstrual attacks, not patients with MRM. Therefore, they probably included patients with true menstrual migraine (whose attacks may not react to acute treatment in the same way as perimenstrual MRM attacks), MRM and patients with attacks that occurred by chance during the perimestrual period but were not triggered by it. About a third of female migraineurs, in fact, complain of MO attacks that can occur at any time in the menstrual cycle, but without a predilection for the perimenstrual period (1, 3). In a clinical trial that selected women with MRM (7) (at least 80% of attacks in the last 6 months occurring between 3 days before to 5 days after the onset of menstruation), the fraction of participants who in the final analysis fulfilled the entry criteria for menstrual migraine had less benefit from sumatriptan 100 mg inside the menstrual window (56% of patients reported headache relief 4 h after dosing) than outside it (81% of patients reported headache relief 4 h after dosing).
The definitions of ‘menstrual migraine’ employed by the various studies are heterogeneous, covering periods of time spanning from 1 to 3 days before to 3–5 days after the onset of menstruation (30–32, 34, 35); in one study (33) ‘menstruation-associated’ migraine was migraine occurring at any time during bleeding. In our study, even though sustained pain-free was achieved in a significantly lower proportion of patients in all the perimenstrual attacks compared to nonmenstrual ones, the attacks least responsive to acute treatment were those occurring the first 2 days of the menstrual cycle.
For most attacks our patients took NSAIDs, as opposed to triptans in all the literature studies but one (33). We cannot exclude therefore that the more potent triptans might have acted differently in our patients. The only trial not involving triptans studied the effect of a combination of acetaminophen, aspirin and caffeine (33). It excluded, however, subjects with incapacitating disability (attacks requiring bed rest > 50% of the time) and those who experienced vomiting in ≥ 20% of the attacks. The population studied therefore is not comparable with ours and probably not representative of MRM patients.
Our study has several limitations. It was carried out on a selected sample (about 60% of the eligible subjects) of women affected by MRM referred to tertiary headache centres. No studies, to our knowledge, have investigated the characteristics of perimenstrual attacks in women affected by MRM in comparison with menstrual attacks in women with true menstrual migraine. Our results therefore apply only to MRM, and not to menstrual attacks in general. Moreover, it is well known that migraineurs seeking care in subspecialty centres constitute only a tiny fraction of the whole population of migraine patients (44), and represent its most severely affected component. The conclusions of this study therefore are not applicable to all the female migraineurs in the general population. Finally, the results concerning the efficacy of acute attack treatment were not obtained by a randomized controlled trial and therefore do not provide the level of evidence generated by double-blind placebo-controlled studies.
Even taking into account these limitations, we believe that the results of our study may be useful to clinicians, who, in their practice, often see MRM patients with a profile similar to that of the subjects we investigated. It is possible to draw several conclusions from our study: Firstly, even though, in accordance with previous studies (1, 29, 42), we found a main peak of menstrual attacks the first 2 days of the menstrual cycle and, as in the survey by MacGregor et al. (1), a second peak 2 days before menses, attacks that are longer-lasting and less responsive to acute treatment cover the entire perimenstrual period, from day − 2 to day +7. This finding is in remarkable agreement with what the women themselves say (i.e. that difficult-to-treat attacks can occur before, during and after the menses) and underlines the substantial unity of perimenstrual attacks, as opposed to further subdivisions; secondly, with the exception of duration, the clinical features of perimenstrual attacks do not differ from those of nonmenstrual ones, a conclusion also drawn by Stewart et al. (29) in their prospective study on a sample of migraineurs from the general population. This means that, in MRM occurring in women referred to specialty care, the perimenstrual period is only a trigger of typical (albeit longer, more disabling and harder-to-treat) migraine attacks; and finally, perimenstrual attacks are characterized by longer duration and, especially in the case of MA, by a relative refractoriness to acute treatment, a finding that supports the long-established impression of clinicians dealing with these patients.
Footnotes
Acknowledgements
This study was partially supported by a grant from the Italian Ministry of Public Health, RC2000 – Institute of Neurology ‘C. Mondino’ Foundation, Pavia.