Stratified Care vs Step Care Strategies for Migraine: Title and subTitle BreakThe Disability in Strategies of Care (DISC) Study:
A Randomized Trial FREE

As the therapeutic armamentarium for migraine expands, clinicians and patients have an unprecedented range of treatment choices.^{1 - 4} Emerging treatment guidelines recommend strategies for selecting and sequencing acute treatments without empirical evidence.^{2 - 4} Ultimately, the goal of optimal migraine care is to initiate effective treatment strategies as early as possible, thereby minimizing pain and disability.⁴

At least 3 different acute treatment strategies have been proposed for migraine: step care across attacks, step care within attacks, and stratified care.^{2 - 6} In step care across attacks, patients begin with a nonspecific therapy (ie, a simple or combination analgesic). After treating several attacks, if the treatment result is unsatisfactory, patients recontact their clinician for treatment escalation. The process is repeated until a satisfactory treatment result is achieved. This approach, often advocated in managed care treatment guidelines, has a number of advantages and disadvantages.^{5 - 6}

In step care within attacks, patients initially treat an attack with a nonspecific therapy. At a specific timepoint posttreatment, usually 2 hours, patients assess their response and, if needed, take another medication, often migraine-specific. This approach provides patients with a real-time strategy for managing treatment failure.⁵

In stratified care, the initial treatment is selected based on the patient's treatment needs. Variants of stratified care are recommended in various treatment guidelines.^{3 - 4} We have suggested that headache-related disability (activity limitations in various domains of function) may serve as the basis for stratification in migraine care.^{5 - 6} In clinical simulations, physicians use headache-related disability to determine severity of illness and to inform the choice of treatment.⁷ Although this approach of matching treatment to severity of illness is intuitive⁷ and used for other conditions (eg, asthma⁸ ), it has not been tested empirically in migraine.

To date, stratified care based on the Migraine Disability Assessment Scale (MIDAS) grade and step-care strategies for the acute treatment of migraine have not been rigorously evaluated. Therefore, we conducted a randomized clinical trial of 3 treatment strategies. In this study, patients were randomized, not to a drug, but to a strategy of care.

This was a randomized, parallel group, open-label, multiple attack study conducted by the Disability in Strategies of Care Study group for patients with an established diagnosis of migraine based on the International Headache Society's criteria.⁹ Patients were recruited from 88 centers in 13 countries. During the course of the 15-month clinical trial (December 1997 to March 1999), each participant treated up to 6 moderate to severe migraine attacks. Aspirin plus metoclopramide was selected as the nonspecific treatment because it is widely used around the world and has proven efficacy in migraine.^{10 - 11} Zolmitriptan, 2.5 mg, a widely used oral triptan, was selected as the migraine-specific therapy.¹²

Patients provided written informed consent during the first clinic visit (baseline). The trial was designed and monitored in accordance with the ethical principles of Good Clinical Practice¹³ as required by the major regulatory authorities and in accordance with the Declaration of Helsinki.¹⁴

The MIDAS questionnaire measures lost time in 3 domains of activity due to headaches; it has been extensively validated.^{15 - 17} MIDAS was used to assign a disability grade to each patient (grade I [little or infrequent disability, score 0-5]; grade II [mild or infrequent disability, score 6-10]; grade III [moderate disability, score 11-20]; and grade IV [severe disability, score ≥21]), indicating overall illness severity during a 3-month recall period. (Disability assessment for each treated migraine attack was measured as disability time as described below). Patients with grade I headaches were excluded from the trial because we determined that the majority of patients in this category do not warrant treatment with a triptan.

Eligible patients with a MIDAS grade of II, III, or IV were randomized to 1 of 3 treatment groups.

Stratified Care. Patients with grade II headache received aspirin, 800 to 1000 mg (depending on the standard dose in each country), plus metoclopramide, 10 mg, as their acute treatment for all 6 attacks. Patients with grade III or IV headache received zolmitriptan, 2.5 mg, as their acute treatment for all attacks. Patients were asked not to take rescue medication within the first 4 hours of the attack.

Step Care Across Attacks. Patients treated the first 3 attacks with aspirin, 800 to 1000 mg, plus metoclopramide, 10 mg. Those who did not have a satisfactory headache response (defined as a reduction in pain intensity from severe or moderate at baseline to mild or no pain at 2 hours in at least 2 of the 3 attacks) were instructed to escalate ("step up") therapy to zolmitriptan, 2.5 mg, for the next 3 attacks. The remaining patients continued using aspirin plus metoclopramide for the remaining 3 attacks.

Step Care Within Attacks. Patients initiated treatment with aspirin, 800 to 1000 mg, plus metoclopramide, 10 mg, for all attacks. After 2 hours, if a satisfactory headache response was not achieved, patients were instructed to step up therapy and take zolmitriptan, 2.5 mg.

Patients were allocated to treatment groups in balanced blocks and allocated sequentially by sex and MIDAS grade.

Eligible patients (1) ranged in age from 18 to 65 years, met the International Headache Society's criteria for migraine with or without aura,⁹ and had onset of their first migraine before age 50 years; (2) had from 1 to 8 migraine attacks per month for the previous 3 months and had not taken a triptan (selective serotonin_1B/1D agonist) within the last 3 months; (3) had fewer than 10 nonmigraine headache days per month during the previous 6 months and could differentiate between migraine and nonmigraine headaches, if they had both; and (4) could not start or change preventive therapy during the study.

Patients were excluded from the study if they (1) had a medical or psychiatric condition that might increase the risk of study medication or interfere with efficacy or safety assessments; (2) had a history of basilar, ophthalmoplegic, or hemiplegic migraine; or (3) were lactating or pregnant.

During the baseline visit, eligible patients were randomized to 1 of the 3 treatment groups, provided with diary cards, study medications, and instructions. Patients used diary cards to track treatment response (intensity, duration, treatment, and rescue medication). Patients graded pain intensity of each attack as none, mild, moderate, or severe and were instructed to treat only moderate or severe migraine attacks. Attacks were not eligible for study treatment if they were present for more than 12 hours, if an acute analgesic was used during the previous 6 hours, or if ergotamine or opiates were used during the previous 24 hours.

A maximum of 2 additional doses of study medication were allowed for treatment of a single attack (maximum daily dose: zolmitriptan, 7.5 mg, aspirin, 3000 mg, and metoclopramide, 30 mg) for persistent or recurrent migraine. Rescue medications (eg, nonsteroidal anti-inflammatory drugs, antiemetics, analgesics, and sedatives) were not permitted within 4 hours, ergotamine was not permitted for 6 hours, and other triptans were not permitted for 12 hours following treatment with study medications.

Patients made follow-up visits after treating their third (visit 2) and sixth attack (visit 3), at which time headache diaries were reviewed. At visit 2 in the step care across attack group, treatment was escalated in patients who responded to aspirin plus metoclopramide in only 0 or 1 of the 3 attacks.

Patients recorded pain intensity (at baseline, 1, 2, and 4 hours) using a 4-point scale (0 = no pain, 1 = mild pain, 2 = moderate pain, and 3 = severe pain). Functional status was recorded on a 0- to 100-point scale in which 0 represented an inability to do any activities and 100 represented perfectly normal function. Headache response at 2 hours and disability time per treated attack at 4 hours were the primary outcome measures. Headache response was achieved if headache pain intensity was reduced from severe or moderate at baseline to mild or no pain within 2 hours. Disability time was calculated for each attack and defined as the area under disability vs time curve (AUC). Level of disability was defined as 100 minus the recorded level of functionality and was measured at baseline, 1, 2, and 4 hours. Secondary efficacy end points included 1-hour or 4-hour headache response and 2-hour pain-free response.

Adverse events were recorded on patient diary cards and defined as mild, moderate, or severe. Any detrimental changes in the patient's condition that occurred between taking the first dose of study medication and up to 24 hours after taking the last dose of trial medication (for the treatment of the same migraine headache) was defined as an adverse event. A serious adverse event was defined as fatal, life-threatening, requiring prolonged hospitalization, resulting in disability or incapacity, resulting in a congenital abnormality, or requiring medical intervention to prevent permanent impairment or damage.

Power calculations indicated that 300 participants were needed per study group to detect a difference of 15% in headache response at 2 hours. This was based on 2 assumptions: that approximately 20% of the patients would not be evaluable or would be withdrawn from the trial and that headache response rates of 50% would be achieved for the 2 step care strategies. Treatment efficacy rates at 2 hours for patients with grade III or IV headaches in the stratified care group were based on published reports for zolmitriptan (62% to 65%)^{12 ,18} and for aspirin plus metoclopramide (45%).^{10 - 11}

Data were analyzed for the efficacy population, defined as all participants who were randomized and treated at least 1 migraine attack and reported at least 1 efficacy assessment. All statistical testing was 2-tailed at the 5% type I error.

A random effects analysis of variance model was used to analyze continuous measures. Discrete data were analyzed using a generalized linear mixed model (using SAS macro GLIMMIX¹⁹ ). Both analyses allowed for the fixed effects of treatment regimen (stratified care, step care within attacks, step care across attacks), baseline MIDAS grade, attack baseline headache intensity/disability as appropriate, country, attack group (1-3 vs 4-6), attack number within an attack group, interaction between treatment and attack group (this allowed separate treatment effects to be produced for each attack group), and a random patient effect. Analysis of data revealed the most appropriate variance/covariance matrix was unstructured. Effect sizes are presented as either odds ratios (OR) (for discrete data) or differences in least squares means (for continuous data), with 95% confidence intervals (CIs). In studies of common outcomes, the OR should not be interpreted as a measure of relative risk.

The efficacy population totaled 835 migraine patients (Figure 1) who completed 2-hour efficacy data taken from the patient diary data on at least 1 attack and excluded those patients who did not complete diary cards (n = 129), but who received the study medication, or who were affected by a MIDAS grade protocol amendment (n = 98). (This procedure for scoring and grading disability was modified based on reliability studies.^{15 ,17} )

The safety and tolerability population included all participants who were randomized to treatment and received at least 1 dose of study medication for a migraine attack.

The 3 treatment groups were well balanced with respect to demographic and headache characteristics (Table 1), and no differences were observed between treatment groups. Of the 1062 patients, a total of 219 patients (20.6%) were withdrawn from the study because of loss to follow-up (n = 90, 8.5%), withdrawal of informed consent (n = 35, 3.3%), protocol noncompliance (n = 33, 3.1%), adverse event/concurrent illness (n = 32, 3.0%), deterioration in patient condition (n = 2, 0.2%), and for other reasons (n = 27, 2.5%). No significant differences were observed in the number of withdrawals between stratified care and step care across attacks groups (P = .91) or stratified care and step care within attacks groups (P = .92). The reasons for withdrawal were evenly distributed across treatment groups and MIDAS grades.

Stratified Care vs Step Care Across Attacks. The proportion of all attacks responding at 2 hours was significantly greater for stratified care than for step care across attacks for all 6 attacks (OR, 1.67; CI, 1.31-2.12; P<.001; Figure 2). Similar results were evident for headache response at 1 hour (OR, 1.61; CI, 1.20-2.15; P = .001) and 4 hours (OR, 1.76; CI, 1.38-2.24; P<.001).

For the first 3 attacks, the stratified care group showed a significantly higher 2-hour headache response than the step care across attacks group (OR, 2.91; CI, 2.18-3.87; P<.001; Table 2). In the step care across attacks group, treatment was escalated for attacks 4 to 6 in those patients who did not respond (ie, in at least 2 of the 3 attacks treated with aspirin plus metoclopramide). Treatment was escalated in 56.4% of patients with MIDAS grade II, 68.9% of patients with grade III, and 74% of patients with grade IV headache. The proportion of patients needing to step up treatment significantly increased with increasing MIDAS grade (χ² for trend, P = .03). For attacks 4 to 6, stratified care vs step care treatment strategies did not differ for 2-hour headache response rates (P = .79; Table 2).

Disability time (from 0-4 hours) averaged across all 6 attacks was significantly lower for stratified care (AUC = 185 mm · h) than step care across attacks (AUC = 209.4 mm · h; treatment effect, −21.25; CI, −31.44 to −1.07; P<.001; Figure 3). The difference was explained by a significantly greater disability time for attacks 1 to 3 in the step care across attacks group compared with the stratified care group (treatment effect, −42.61; CI, −54.01 to −31.21; P<.001). Following treatment escalation in this group, no differences were observed with stratified care for attacks 4 to 6 (P = .99).

The proportion of attacks achieving a 2-hour pain-free response was significantly greater for the stratified care group than the step care across attacks group (OR, 1.66; CI, 1.18-2.32; P = .003; Table 2). Again, the statistically significant difference during the first 3 attacks (OR, 3.05; CI, 2.01-4.62; P<.001; Table 2) accounts for much of the advantage of stratified care over step care treatment strategy.

Stratified Care vs Step Care Within Attacks. Across all 6 attacks, the proportion of attacks with a headache response at 2 hours was significantly higher for the stratified care treatment group than for the step care within attacks group (OR, 2.14; CI, 1.66-2.77; P<.001; Figure 2). This was evident for attacks 1 to 3 (OR, 2.05; CI, 1.55-2.72; P<.001) and attacks 4 to 6 (OR, 2.24; CI, 1.64-3.07; P<.001; Table 2). Patients who did not achieve a headache response at 2 hours had the option of taking zolmitriptan, 2.5 mg, as a rescue medication. Comparing the stratified care and step care within attacks groups, significant differences in the proportion of attacks that responded to treatment were evident at 1 hour (OR, 1.77; CI, 1.33-2.36; P<.001). Between-group differences were not evident at 4 hours (P = .21); patients in the step care group who did not respond by 2 hours could escalate their treatments to zolmitriptan, 2.5 mg (Figure 2).

Disability time from 0 to 4 hours was significantly lower in the stratified care treatment group (AUC = 185 mm · h) compared with the step care within attacks group (AUC = 200 mm · h; all attacks, treatment effect, − 19.43; CI, −29.73 to −9.14; P<.001). Separation between the 2 treatment groups was most evident up to 2 hours prior to rescue with zolmitriptan. By 4 hours (Figure 3), though the AUC was lower for step care, the between-group differences in disability time also had decreased.

Stratified care provided more effective treatment than step care within attacks for the proportion of attacks that were pain-free at 2 hours (6 attacks, OR, 1.68; CI, 1.20-2.36; P = .003). Similar trends were evident at 1 and 4 hours.

The safety and tolerability population included 930 patients treating 4945 migraine attacks. Adverse events were reported in 697 attacks (14.1%). Many of the most common adverse events (defined as those occurring with a frequency >2.0%), including asthenia, dizziness, and paresthesia, were those usually seen with triptan treatment (Table 3). Adverse events in all treatment groups were predominantly mild to moderate and not severe (Table 3). One patient reported 7 serious adverse events during the course of an attack, defined as the period within 24 hours of taking the last dose of study medication. This patient was in the step care within attack treatment group and experienced 2 episodes of abdominal pain, 3 episodes of parasthesia, and 2 reports of somnolence and was withdrawn from the study after the third attack.

Six patients reported serious adverse events that started more than 24 hours after receiving the initial dose of treatment medication: 2 in the stratified care treatment group, 3 in the step care across attacks group, and 1 in the step care within attacks group. Adverse events leading to withdrawal from the trial were similar across the 3 treatment groups (stratified care 3.3%, step care across attacks 2.9%, and step care within attacks 3.8%).

This randomized clinical trial demonstrated that the stratified care strategy produced better outcomes than the step care across attacks and the step care within attacks strategies for migraine. Stratified care treatment is based on the principle that patient characteristics (ie, headache disability) can be used to help determine the patient's treatment need, thereby increasing the chance of successful therapy from the outset.^{4 - 6}

Step care across attacks is a strategy that is frequently recommended in treatment guidelines.^{3 - 4} In practice, step care usually includes multiple steps beginning with, for example, a simple analgesic, followed by various combination analgesics, a trial of isometheptene or a butalbital-containing agent, and then triptans. One difficulty with the step care treatment strategy is that effective treatment may be delayed far more than in the present study. In addition, patients may not follow-up with their physician when a treatment fails and instead they may lapse from medical care.^{5 ,18 - 20} This approach to discovering the most effective treatment may prolong morbidity and add to health care costs if several clinical contacts (in-person or by telephone) and several failed prescriptions are required.⁵ Stratified care increases the probability of selecting appropriate treatment as the first intent.

Step care within attacks can delay the use of effective medication within an attack. Among the 72% of persons with migraine who are less than fully satisfied with their migraine treatment, 87% say that pain relief takes too long.²¹ While step care within attacks may be an appropriate treatment strategy for patients who usually respond to nonspecific therapy, it delays the onset of pain relief and restoration of function for many patients.

In the stratified care group in our study, treatment was selected based on severity of illness, as measured by MIDAS grade. In the step care across attack group, the proportion of patients who needed to escalate treatment because of lack of response to aspirin plus metoclopramide significantly increased with MIDAS grade from 56% for patients with grade II to 74% for patients with grade IV headaches. This finding supports the hypothesis that MIDAS grade predicts treatment need and suggests that aspirin plus metoclopramide effectively treats only about one fourth of patients in the highest disability grade.

The 3 different treatment strategies were generally well tolerated. Adverse events were more frequent in the stratified care treatment arm. This was mostly attributed to an increase in mild or moderate events, reflecting the higher numbers of subjects exposed to zolmitriptan as opposed to aspirin plus metoclopramide.

This study was designed as an open-label, randomized trial because of the complexity of the treatment groups. As a consequence, knowledge of treatment may have influenced study results. Open-label, randomized designs have been used in studies of treatment strategy for other conditions²² and more closely simulate clinical practice than double-blind studies.

Aspirin plus metoclopramide and zolmitriptan were selected as study treatments based on their frequent use, proven efficacy, and worldwide availability. Choice of these specific medications is not an intrinsic aspect of the stratified care approach. Had other agents been selected, however, the results may have been different.

Stratification based on individual attacks was an alternative treatment strategy.⁴ Using this approach, patients would opt for a nonspecific therapy or triptan drug based on their assessment of their own treatment needs for each attack. Given that the patient's ability to optimally select treatment may change with experience, more than 6 attacks would have been required; despite difficulties in study design, this approach merits empirical testing.

In some respects, the treatment strategies used herein differ from those used in clinical practice. In clinical practice using stratified care, if an initial migraine-specific therapy failed after 3 attacks, treatment might be switched to another migraine-specific agent.²³ Similarly, rescue medication is often given at 2 hours instead of 4 hours after receiving the initial medication. Thus, we may have underestimated the relative benefits of stratified care. For step care across attacks, treatment escalation was probably more consistent and rapid than in routine practice, possibly overestimating the benefits of step care across attacks. Finally, in the future, stratification based on factors other than migraine-related disability may be possible. Ultimately, symptom profiles, biological markers, or genetics may identify patient subgroups likely to respond to particular therapies.^{5 - 6}

To the best of our knowledge, this is the first study comparing strategies of care in acute migraine management. The results favor stratified care based on MIDAS grades vs step care within and across attacks. The results generally support recommendations regarding the use of stratified care as the more effective treatment strategy compared with traditional step care approaches.^{4 - 6} Use of clinically tested treatment strategies should improve the outcomes of care for patients with migraine.