The Natural Course of Migraine Attacks. A Prospective Analysis of Untreated Attacks Compared with Attacks Treated with a Triptan

Abstract

This study was designed to document prospectively and explore scientifically the natural course of untreated migraine attacks in detail. A new, integrated, time-intensity method for self-assessment of the intensity of symptoms was tested on 18 adult International Headache Society migraineurs who volunteered to refrain from treatment during one attack. The area under the curves (AUC) during 72 h of untreated attacks was compared with attacks treated with a triptan. Migraine attacks are heterogeneous both inter- and intra-individually. In untreated attacks, the pain can stabilize and fluctuate around a plateau with a wavelength of hours. In general, the symptoms of each separate migraine attack follow a similar temporal course, with only moderate deviations. In some cases photo- and/or phonophobia (hyperexcitability) were not experienced at all, despite severe pain and nausea. Moreover, there was sometimes no nausea despite severe pain and hyperexcitability. Vomiting does not always correlate to the intensity of nausea and is not always followed by decreased headache intensity. Treatment with a triptan usually only temporarily distorts the basic pattern of attacks. Hyperexcitability can respond before pain to treatment. These genuine findings of the classic symptoms of migraine attacks support the notion of a mutual underlying pathophysiological mechanism.

Introduction

Migraine is an inherited, chronic, neurological disorder characterized by recurrent episodes of severe throbbing headaches associated with nausea and increased sensitivity to sound and light, but with freedom from symptoms between the attacks (1–3). It is considered possible to identify five phases of a migraine attack: the prodromal phase, aura, the headache phase, resolution and postdromal phase (4). Of these, only the classic neurological aura has a distinct onset and end. According to the international classification, a migraine attack lasts from 4 to 72 h (3).

A long tradition of clinical descriptions of migraine has resulted in an immense amount of literature which has been reviewed elsewhere (1, 5–23). From a modern perspective, mainly the premonitory symptoms and the aura are well documented in several retro- and prospective studies, and particular interest has been paid to their temporal relationship to the headache phase (4, 12, 24–29).

Considering the fact that the headache phase is regarded by a majority of patients as the worst feature of the attacks, the corresponding precise knowledge of its natural course is disproportionately limited. The primary reason is that most patients use acute medication and seldom let the attack continue untreated. When asked retrospectively to describe the characteristics of the attack, migraineurs find it difficult to comment on what is relief of symptoms due to treatment and what is a spontaneous regression. Furthermore, it is difficult to differentiate possible side-effects and weaknesses of the drugs from symptoms caused by the attack per se (30). This gives rise to clinical problems, such as estimating what is recurrence due to insufficient duration in time of acute treatment and what is a new attack or perhaps even a drug-induced headache.

Second, apart from short statements as to the duration of attacks, even the most detailed accounts of the symptomatology of the headache phase lack information on the important fourth dimension—time (11, 21, 30). Mostly, when searching for specific data on the dynamic change over time in intensity of the characteristic symptoms of the headache phase, one finds vague and/or weakly supported information. These are mainly retrospective outlines, descriptions of a single point in time during the migraine attack and/or treated attacks. Not uncommonly, the patients are interviewed about an atypically difficult attack (12, 13, 15, 31, 32). The authors as a rule try to estimate the average intensity of the symptom, which does not adequately reflect the studied phenomenon (33). Since the dynamic complexity of the attack renders it difficult to extrapolate experiences from the whole course, the patient may retrospectively let, for example, the highest intensity reached represent the whole headache phase (34).

Third, it is considered difficult to verbalize and communicate changes in the intensity of symptoms as a consequence of variations in language, experience, expectations and frames of reference (5, 33). For example, when asked about the presence of nausea (at least in Swedish), patients commonly respond ‘I do not vomit’. Since nausea can exist without vomiting, this symptom needs to be further investigated.

We here report the first open cross-over study in clinical practice designed to document prospectively the natural course of untreated migraine attacks in detail. A preliminary account of the project was presented earlier (35). A new integrated time-intensity method for self-assessment of the time course and grading of the key symptoms aura, headache, photo/phonophobia, nausea and vomiting was tested. In each participant, we have collected data on spontaneous recovery and, in addition, different strategies of usual acute treatment as a comparator. The principal goal was to explore scientifically important questions such as: is there a simple relationship between the point in time and pain intensity in the natural course of still untreated migraine attacks? Is there a clear-cut sequence in which the characteristic symptoms appear during the attack? We will also discuss how different pathophysiological models perform in the light of this new genuine clinical picture of migraine.

Methods

Design

This was an open, prospective, cross-over study conducted between November 1998 and November 2004 in an out-patient setting at the Gothenburg Migraine Clinic (GMC). The protocol was approved by the Ethics Committee of Göteborg University.

Subjects

All participants were recruited from among patients at the GMC. Using a method of randomized clustering, potential subjects (n = 400) were invited to take part in the study. Eligible patients were aged 18–65 years and had used their acute medication regularly with no contraindications for at least 6 months in acute migraine, with or without aura according to the International Headache Society (IHS) criteria (36). Patients were excluded if they had known or probable drug-induced headache, or had any severe concurrent medical condition which affected data interpretation. The required number of participants to achieve adequate power could not be calculated beforehand. A sample size considered large enough to detect a clinically relevant finding was set at approximately 20 (n = 20). A total of 30 (n = 30) patients were included in the study. A total of nine (n = 9) participants dropped out. Explanations given for this were: ‘too difficult to refrain from medication during untreated attack’ (n = 6), ‘lack of motivation’ (n = 1), and withdrawn consent for no specified reason (n = 2). An additional three (n = 3) patients were excluded from the analysis owing to inadequate data (n = 1), pregnancy (n = 1) and freedom from attacks since inclusion in the study (n = 1).

Protocol

All patients registered data from three attacks in their everyday environment, each time during a period of 3 × 24 h with at least 9 days between registered attacks. The first attack (A1) was treated with their usual medication to give them an opportunity to practice filling in the forms. During the second attack (A2) patients had been instructed to refrain from using any treatment. If the situation became unbearable, they were, however, allowed to use rescue medication (excluding ergotamine-containing products). Those who used rescue medication were later asked if they would be willing to make another attempt. All patients treated the third attack (A3) with a conventional rizatriptan 10 mg tablet. During treated attacks, patients took acute medication at the time of their own preference. A maximum of six open-label doses were taken per attack, and not more than two triptan doses were permitted in any 24-h period. Using the parameters registered, it could be confirmed retrospectively that all registered attacks were, in fact, migraine.

All characteristic symptoms were documented using 100-point visual analogue scales (VAS) (Fig. 1) filled in hour-by-hour during the awake portion of 72 h from the start of attacks. The report is based on aura, headache intensity, nausea, vomiting, photophobia and phonophobia. Using an electronic digitizer (IS/One Input system™; Kurta Corp., Phoenix, AZ, USA), the area under the curves (AUC) was measured. This parameter constitutes the (changing) symptom intensity multiplied by suffered migraine time. Data are reported as AUC-units (25 mm²) equal to 10 VAS steps multiplied by 1 h.

Figure 1

The 100-point visual analogue scale (VAS) filled in hour by hour during the awake portion of 72 h from the start of attacks. Examples of information extracted below are the time from first occurrence of pain to treatment (A), pain intensity at the time of treatment (B), pain intensity at 1 and 2 h after treatment (C, D), time from treatment to freedom from pain (E), area under the curves from treatment to pain free (F) and number of recurrences after treatment (G).

Data extracted from all attacks are total duration of headache (excluding pain free intervals), the AUC_total of headache intensity, and the proportion of this AUC_total represented by the AUC during the first 2 h (pain only). Also, the relative relation between the characteristic symptoms—the sequencies in which they emerged and receded—were studied with and without treatment. Curves mainly from the untreated attacks (A2) were used to challenge traditional paradigms on the natural course of migraine attacks. In the treated attacks only (A1, A3), it was studied whether the symptomatology of the attack fulfilled the IHS criteria for migraine as early as at the time of treatment, whether the characteristic symptoms responded in different types of treatment, and if successful triptan treatment seemed to influence the fundamental pattern of the attacks or only to relieve the main symptoms during various periods of time. Finally, the number of recurrences (defined as return of migraine less than 24 h after pain free from first episode) were counted, and the symptomatology of recurrences compared with the core phase of the attack was documented.

Statistical analysis

Intraindividual comparisons are made between the untreated attack (A2) and the treated attacks (A1, A3). Sign-test for paired samples was used to compare non-parametric data (VAS). Wilcoxon signed ranks test was used for parametric data. All analyses were performed using the Statistical Package for Social Sciences (SPSS™) software program (version 11.0.2 for MacOSX; SPSS, Inc., Chicago, IL, USA). Significance levels were set at 0.05.

Results

The analysis is based on 18 (n = 18) patients (median age 48, range 25–65 years, 15 women) with migraine (12 migraine without aura only, two migraine with aura only and an additional four with a combination of migraine without aura and migraine with aura). The median number of years with migraine was 25 (range 12–45 years).

The medications used for usual acute migraine treatment were acetylsalicylic acid per os (n = 1), acetylsalicylic acid + caffeine per os (n = 1), acetylsalicylic acid + metoklopramide per os (n = 2), diklofenac per os (n = 1), ibuprofen per os (n = 2), paracetamol per os (n = 1), paracetamol + codeine per os (n = 1), rizatriptan 10 mg per os (n = 3), sumatriptan 20 mg intranasally (n = 3), sumatriptan 50–100 mg per os (n = 7), sumatriptan 25 mg rectally (n = 1), sumatriptan 6 mg subcutaneously (n = 4) and zolmitriptan 2.5 mg per os (n = 3). One patient had been using pharmacological prophylaxis (amitriptylin 40 mg daily) for at least 3 months before the start of the study.

General profile of the attacks

In broad outline, the time–intensity curves showed an apparent heterogeneity both inter- and intraindividually. Different features of each separate attack, however, accompanied each other in a similar temporal course with moderate deviations in an overall visual assessment (Fig. 2b).

Figure 2

(a) Typical registration of symptoms during an untreated migraine attack (day 1). (b) Typical registration of symptoms during an untreated migraine attack (day 2).

Headache

The pain commonly appeared before the other characteristic symptoms and was often manifest at the time of awakening. After an initial mild headache without associated nausea, photo- or phonophobia (hyperexcitability) during 10 h, the pain intensity rapidly increased and developed into full-blown migraine in one case (S012/A2). In untreated attacks, the pain usually followed a smooth and evenly distributed curve connecting a crescendo with a diminuendo. In some patients, the debut was more rapid, in others the resolution had a greater velocity. In several cases, most (awake part of days and nights) of the attack was characterized by a plateau where the pain fluctuated in a wave-like and slow (hours) oscillation between moderate and highest imaginable intensity (Fig. 2a,b). The pain intensity could be unaffected when waking up from sleep (S035/A2). Pain was sometimes the last remainder of the characteristic symptoms before spontaneous resolution of the migraine attack (S012/A2, S030/A2).

Photo- and/or phonophobia (hyperexcitability)

Photo and phonophobia were commonly ranked as uniformly intense with a mutually related temporal profile throughout the untreated attacks (S011/A2, S015/A2). Photophobia was commonly more intense than phonophobia, which could, in fact, be completely absent (S011/A3, S012/A2, S035/A2, S038/A2). Phonophobia often appeared after photophobia, and the dynamic course could be divergent between the two (S011/A1 + A2, S016/A1 + A2, S030/A1, S035/A1). However, in one patient, photophobia appeared after phonophobia (S015/A2), and in others phonophobia was present in attacks devoid of experienced photophobia (S012/A2, S026/A2). Hyperexcitability (as represented by photo- and/or phonophobia) was sometimes not experienced at all in attacks with severe pain and severe nausea (S018/A1, S023/A3).

Nausea and vomiting

In only a small minority (1/18) of cases did the nausea follow a time–intensity profile different from that of the pain (S026/A2). Nausea was sometimes not experienced at all in attacks with severe pain and severe photo- and/or phonophobia (hyperexcitability) (S016/A1, A2, A3).

Vomiting commonly appeared at the intensity peaks of pain and was immediately followed by a significant improvement of the pain (S012/A2, S015/A2, S016/A2). Vomiting sometimes also appeared hours before the pain had reached its maximum (S011/A1). When vomiting appeared at the peak intensity of pain, it was not always followed by decreased headache (S035/A2). The occurence of vomiting did not always correlate to the intensity of nausea (S016/A2).

Effects of treatment

The median number of rizatriptan doses was one (n = 1) per treated attack (A3) (mean 1.8, range 1–6). Treatment with rizatriptan commonly abolished all registered symptoms within 2–3 h after intake, also when pain was ranked 50–80 on the 100-point VAS (S010/A3, S018/A3, S038/A1). This iatrogenic resolution was sometimes slower than the corresponding debut of the attack (S026/A3, S038/A3). Regularly, the effect of triptan treatment was short (hours) and incomplete, with symptoms returning within hours to their former intensity (Fig. 3). The resulting time–intensity curves from patients who had used repeated doses in the same attack were those of a fundamental attack profile only temporally depressed by the drug (S015/A3, S023/A1, S035/A1 + A3).

Figure 3

Typical registration of symptoms during a treated migraine attack (day 3).

Nausea commonly responded better than pain or photo-/phonophobia to treatment (S018/A1, S030/A3, S035/A3). Photo- and phonophobia could respond to treatment before pain was affected (S016/A3, S038/A3), and pain could be the last symptom to vanish after effective treatment with a triptan (S038/A3). As shown in Fig. 4, there was a significant difference (P = 0.016, Wilcoxon signed ranks test) in the total duration (excluding pain-free intervals between the core phase of attacks and recurrences where applicable) of headache in untreated attacks compared with attacks treated with rizatriptan.

Figure 4

Mean total duration (excluding pain-free intervals) of headache in untreated attacks vs. attacks treated with a triptan (n = 18). Error bars show 95% CI of means.

There was a highly significant (P < 0.0001, sign test) difference between untreated attacks and attacks treated with a triptan in the AUC_total of headache intensity (Fig. 5). The median AUC_total during 72 h (pain only) was 168 units (mean 185.7, range 63.6–438.0) for untreated attacks (A2) and 49 units (mean 67.7, range 10.0–251.3) for treated attacks (A3). The median intraindividual reduction of AUC_total of pain intensity between untreated (A2) and triptan-treated attacks (A3) was 70.6% (mean 60.3%, SD 0.45, range −110% to +96%).

Figure 5

Area under the headache intensity curves (AUC) in untreated attacks (n = 18) vs. attacks treated with a triptan (n = 18). Plots give the interquartile range (boxes), median (line within boxes), the highest and lowest values excluding outliers and extreme values (whiskers), outliers (s) and an extreme value (rhombus).

The median AUC during the first 2 h (pain only) was 11.2 units (mean 10.8, range 4.0–16.2) for untreated attacks (A2) and 7.5 units (mean 8.0, range 2.9–16.8) for treated attacks (A3). The median proportion of the AUC_total represented by the AUC during the first 2 h (pain only) was 7.3% (mean 7.2%, SD 0.03) for untreated attacks (A2) and 16.9% (mean 22.7%, SD 0.22) for treated attacks (A3). The median intraindividual reduction of AUC of pain intensity during the first 2 h between untreated (A2) and triptan-treated attacks (A3) was 30.8% (mean 26.8%, SD 0.17, range −7% to +50%).

Recurrence (return of migraine less than 24 h after pain free from first episode)

Most patients (14/18, 78%) experienced at least one headache recurrence during the three attacks. The mean number of recurrences was 0.6 (range 0–4) in the untreated attacks and 1.28 (range 0–5) in attacks treated with rizatriptan. No significant difference was found in the number of recurrences in attacks treated compared with untreated (P = 0.113, Wilcoxon signed ranks test).

Recurrences appeared despite fast and complete effect of triptans (S010/A1, S038/A1). Among the characteristic symptoms, pain reappeared most often in these cases (S010/A1, S011/A3, S018/A3, S023/A1, S038/A1 + A3), but in one patient photophobia was earliest to return (S030/A1). Compared with the core phase of the attack, a crude delineation of recurrences with respect to the constellation and relative intensity of symptoms shows intraindividual homogeneity in this series. The maximum intensity of symptoms in recurrences was lower than before both successful treatment and spontaneous initial resolution (S06/A3, S011/A3, S012/A3, S023/A3, S038/A1 + A2 + A3).

Discussion

To our knowledge, this is the first prospective study of untreated migraine attacks. The design has made it possible to examine a common and apparently well-known disease from a novel scientific perspective and to either challenge or support earlier paradigms. The main finding is that migraine attacks have a basic but heterogeneous natural course on which current acute medication—although effective—often exercises only a temporary influence.

A naturalistic and analytical observation of a phenomenon in a small reference population such as this is a chance to shed some light on principles and problems concerning pathophysiological mechanisms and treatment of direct importance for the patients (37). One assertion that could be made as early as during the period of recruitment is that migraineurs are most unwilling to endure attacks without treatment. Collection of information was closed after 6 years when the sources were running short and the categories were reaching saturation point. Further data were considered to add disproportionately little information in relation to the energy they would require to include. Only 4.5% of the invited subjects were included and completed the study. This highly selected group, however, seems to be comparable to the source population in terms of general management of migraine. Since it was difficult to recruit the necessary number of patients owing to the design of the protocol, it was decided not to exclude the patient (n = 1) using pharmacological prophylaxis. Three patients (n = 3) were not able to go through an entire attack untreated and used rescue medication after >15 h. The rationale for including these patients in the analysis was that the treatments (paracetamol per os, sumatriptan per os and zolmitriptan per os) did not have an apparent influence on the symptoms of the attack.

The material does not permit a subgroup analysis entering deeper into earlier findings from epidemiological research on differences in the headache phase between sexes, age groups and type of migraine (24–26, 38–43). However, it opens a unique possibility to describe systematically and open-mindedly patterns, variations and interplay in subjective sensations unavailable to experimental methods. It is usually more challenging to analyse observational studies than randomized trials, since statistical methods cannot be applied. Outliers are, for example, of great importance, since they refute over-simplified explanations of patterns. The basic theory of ‘interpretation in context’ of cases to develop general statements on regularities and to broaden experience is derived from qualitative research (44).

Part of the study is also experimental, since untreated attacks are compared with treated with the participants acting as their own controls, thus creating a hybrid design by combining elements of different basal models. Furthermore, we wanted to include a ‘yardstick’ as reference. For ethical reasons, this could not be placebo, but instead a positive control (optimal dose of standard treatment, rizatriptan 10 mg per os). One should be aware, however, that a fixed dose of a drug varies in distribution and metabolism between individuals, leading to varying concentrations in plasma and at the site of action with different physiological effects. However, we avoided ‘confounding by indication’, which would have arisen if we had only studied treatments that had been tested beforehand as optimal for each individual (37, 45). Rizatriptan was chosen just for simplicity since the differences among the oral triptans are sufficiently small and it is not only difficult to predict the clinical response to a triptan for a given individual, but individual patient preference is complex and depends on a balance of attributes unique to each patient.

The duration of the attack and the prevalence of symptoms have been described in numerous publications. What is innovative here is that we have prospectively confirmed the traditional statements that the time–intensity curves of treated and untreated attacks are heterogeneous both inter- and intraindividually (18, 46). The finding that the classical symptoms of each separate attack in general accompany each other in a similar temporal course with moderate deviations supports the notion of a mutual underlying mechanism. The anatomical structure driving the attack is not likely to be the somatosensory (trigeminal) system, since, as has previously been described, other features of the attacks such as photo-/phonophobia and nausea commonly appear before the pain. It is therefore apparent, that the attack per se starts during what is traditionally referred to as the ‘premonitory’ or ‘prodromal’ phase and continues after the resolution of pain during the ‘postdromal phase’, which probably takes place in the central nervous system (6, 29, 47, 48). The aura may, however, appear after the onset of headache (5, 24).

The frequent statement that although attacks can begin at any hour, they most commonly manifest as mild pain when the patient awakens in the morning, was confirmed (14, 15, 18, 20, 49, 50). It was also confirmed that they gradually, within hours, build up with great variability of speed to reach a peak which usually has moderate to the highest imaginable intensity (1, 5, 6, 10, 12, 15, 16, 20, 46, 51, 52). In some untreated patients, the pain then fluctuates around a plateau in a wave-like and slow motion. This has been mentioned but not documented earlier, and the wavelength has been said to last minutes, not hours as was the case in this series (1, 12, 14, 17). All incoming stimuli have been said to accentuate the headache, but for practical reasons this could not be tested here (6, 16). In our pharmacologically oriented era, knowledge on the natural resolution of attacks is limited (5). In the majority of these patients, the headache usually faded from maximum intensity over the course of hours, as has been generally considered to be the case (12, 16, 50). It was also observed that the headache commonly, but not always, receded during what has been referred to as a ‘critical’ sleep (4–6, 10, 14, 20, 51).

Considering the fact that photo- and phonophobia are part of the diagnostic criteria for migraine, it is noteworthy that very little has been said about their natural course in the many otherwise colourful and detailed descriptions of migraine attacks. It has been stated that such associated symptoms usually evolve after the debut of severe pain (15, 50), but photo- and phonophobia have also been demonstrated interictally and during the premonitory phase of the attacks. They have been found to correlate neither to the intensity of pain, nor to the intensity of nausea (5, 12, 17, 20, 29, 53–55). As a counter-claim to this, we usually saw a clear trend towards compatible time–intensity courses of photo-/phonophobia and headache, and also a high synchronicity between photo-/phonophobia and nausea in cases reporting these symptoms. In contrast to earlier suggestions, photo- and/or phonophobia (hyperexcitability) was, in some of our cases, not experienced at all in attacks, despite severe pain and severe nausea (46). Although the dynamical course could be divergent between the two, photo- and phonophobia were commonly ranked uniformly intense with a mutually related temporal profile throughout the untreated attacks. The latter has previously been reported in studies using different methodologies, and put forward as support of a common pathophysiological mechanism between photo- and phonophobia (17, 54, 55).

None of the participants reported nausea before pain, which one sometimes hears of in clinical practice (1, 5, 12, 50). In line with earlier experience, the nausea followed a time–intensity profile similar to that of the pain in the absolute majority of cases (12, 18, 50). But as has been reported earlier, nausea was sometimes not experienced at all in attacks with severe pain and severe photo- and/or phonophobia (hyperexcitability) (5, 13).

The occurrence of vomiting did not always correlate to the intensity of nausea (18). Vomiting did, however, commonly appear at the intensity peaks of pain and was often immediately followed by a significant improvement of the pain, but could also emerge hours before the pain had reached its maximum (6, 20, 46, 51). However, particularly in contrast to the earliest reports, when vomiting appeared at the peak intensity of pain, in our study it was sometimes not followed by decreased headache (6, 10, 11, 14, 16). This, in combination with studies of optokinetic stimulation of migraineurs, suggests that the headache may increase nausea/vomiting but not vice versa (56). Liveing was early to point out that the resolution of headache may not be a result of the sometimes preceding vomiting (14).

Abu-Arafeh prospectively studied (usually treated) attacks of migraine in children (n = 10). As here, it was concluded that the attacks follow a similar pattern with or without treatment (57). In contrast to an earlier retrospective analysis, nausea commonly responded better than pain or photo-/phonophobia to treatment (58). Photo- and phonophobia could also respond to treatment before pain was affected, and pain could thus be the final symptom to vanish after effective treatment with a triptan.

After complete resolution, the headache commonly recurs. If the pain-free interval is shorter than 24 h, it is commonly perceived by researchers as an extreme fluctuation in headache intensity in one and the same attack (5, 59). This notion is supported in this study by the similar constellation of symptoms in recurrences compared with the core phases of the attacks. A patient would generally say that recurrence had happened in any migraine attack that worsened after initial improvement. Based on this, any detailed clinical history reveals that about every second migraineur experiences at least one relapse after the first dose of a triptan. In clinical trials, however, we use sophisticated definitions, such as ‘subjective significant worsening of migraine headache (to grade 3/2) within 24 (48) h of treatment following initial relief (grade 3/2 improving to grade 1/0) at 2 (4) h after treatment’. These randomized controlled trials indicate that headache recurrence within 24 h of initial dosing occurs in approximately 35% of patients. From the information available so far, it seems that the patient- and attack-related factors associated with fewer recurrences are: attack duration of <24 h; less severe attacks; and early or late treatment. This pattern indicates that headache recurrences are influenced by the patient as well as by the characteristics of the migraine attack. There was a trend (P = 0.1) of having more recurrences after active treatment with rizatriptan. This finding probably confirms the notion that triptans can not turn off the migraine attack but relieve the symptoms as long as there is sufficient plasma concentration of the drug.

The median intraindividual reduction of AUC_total of pain intensity between untreated and triptan-treated attacks was substantial (about 70%) but much less (about 30%) if only the first 2 h after administration was taken into an account. This fits well with a migraine-induced delay of the gastric emptying time and impaired absorption of a triptan taken orally during the migraine attack. ‘Sustained pain free’ (pain free within 2 h and no recurrence nor additional drug use within 24 h after successful treatment) has been recommended as the most stringent outcome measure but is limited to specific time intervals (2 and 24 h). The median proportion of the AUC_total represented by the AUC during the first 2 h (pain only), however, was only 17% (mean 22.7%, SD 0.22) for treated attacks. Obviously, the AUC obtained by repetitive, continuous measurements on time–intensity scales over the whole migraine attack provides further relevant information. Needless to say, however, this ought to be confirmed in larger randomized controlled trials.

It can thus be concluded that there is no simple relationship between the point in time and intensity of symptoms in the natural course of untreated migraine attacks. The time–intensity curves of migraine attacks are heterogeneous both inter- and intraindividually. In untreated attacks, the pain can, for example, stabilize and fluctuate around a plateau with a wavelength of hours. It is clear from the present observations that the IHS criteria for migraine are not always met by the individual attacks, in particular if treatment has been administered during the early phase of the attack. When the physician is taking a detailed clinical history for diagnosis, however, the headache patient gives their average picture of all attacks/episodes they have experienced, which then more easily meet the IHS criteria for migraine. The classical symptoms of each separate attack in general accompany each other in a similar temporal course with moderate deviation. This supports the notion of a mutual underlying pathophysiological mechanism. Usually, compatible time–intensity courses of photo-/phonophobia and headache were seen, and also a high synchronicity between photo-/phonophobia and nausea. But in contrast to earlier suggestions, photo- and/or phonophobia (hyperexcitability) was not experienced at all in attacks in some of our cases, despite severe pain and severe nausea. Nausea was also sometimes not experienced at all in attacks with severe pain and severe photo- and/or phonophobia (hyperexcitability). Vomiting did not always correlate to the intensity of nausea and was not always followed by decreased headache. Photo- and/or phonophobia could respond to treatment before pain was affected. Although effective mainly in reducing the total area under the pain–intensity curves, triptan treatment commonly only temporarily distorted the basic pattern of attacks. The high sensitivity and obvious clinical relevance of measuring the AUC_total should be taken into account in trials of migraine treatments.