Treatment preferences and the selection of acute migraine medications: results from a population-based survey

Abstract

Seven oral triptans are available for treating acute migraine. We surveyed US migraine sufferers on the relative importance of treatment attributes for choosing among oral triptans. A multiattribute decision model was used to combine data on the relative importance of efficacy, consistency, and tolerability of acute treatment (determined by 206 triptan-experienced and 209 triptannaive subjects) with data on the performance of the triptans across the attributes (derived from a recent meta-analysis). Efficacy was considered significantly more important than tolerability and consistency: tolerability was significantly more important than consistency for triptan- naive but not triptan-experienced subjects. The multiattribute decision model found that almotriptan, eletriptan, and rizatriptan were significantly closer to the hypothetical ideal triptan than the reference product, sumatriptan 100 mg, for both triptan-naive and triptan-experienced migraine sufferers. Almotriptan, eletriptan, and rizatriptan were the preferred triptans selected on the basis of patients' own priorities and product performance data. [PUBLICATION ABSTRACT]

J Headache Pain (2004) 5:123130

DOI 10.1007/s10194-004-0080-yORIGINALRichard B. Lipton

Joshua LibermanF. Michael Cutrer

Peter J. Goadsby

Michel Ferrari

David W. Dodick

Douglas McCrory

Paul WilliamsTreatment preferences and the selection of

acute migraine medications: results from a

population-based surveyAbstract Seven oral triptans are

available for treating acute migraine.

We surveyed US migraine sufferers

on the relative importance of treatment attributes for choosing among

oral triptans. A multiattribute decision model was used to combine data

on the relative importance of efficacy, consistency, and tolerability of

acute treatment (determined by 206

triptan-experienced and 209 triptannaive subjects) with data on the performance of the triptans across the

attributes (derived from a recent

meta-analysis). Efficacy was considered significantly more important

than tolerability and consistency: tolerability was significantly more

important than consistency for triptan-naive but not triptan-experienced

subjects. The multiattribute decision

model found that almotriptan,

eletriptan, and rizatriptan were significantly closer to the hypothetical

ideal triptan than the reference product, sumatriptan 100 mg, for both

triptan-naive and triptan-experienced

migraine sufferers. Almotriptan,

eletriptan, and rizatriptan were the

preferred triptans selected on the

basis of patients own priorities and

product performance data. Key words Meta-analysis

Migraine Surveys Treatment

attributes TRIPSTAR TriptansReceived: 25 November 2003

Accepted in present form: 14 January 2004R.B. Lipton ()Departments of Neurology, Epidemiology

and Population Health,

Albert Einstein College of Medicine,

1300 Morris Park Avenue (Russo-3rd Floor),

Bronx, NY 10461, USA

e-mail: [email protected]

Tel.: [phone]Fax: [phone]J. LibermanIMR, an Advance PCS Company,

Hunt Valley, USAF.M. CutrerMayo Clinic, Rochester, USAP.J. GoadsbyInstitute of Neurology

The National Hospital for Neurology and

Neurosurgery,London, UKM. FerrariDepartment of Neurology,

Leiden University Medical Centre,

Leiden, The NetherlandsD.W. DodickDepartment of Neurology,

Mayo Clinic, Scottsdale, AZ, USAD. McCroryDuke University Center for Clinical Health,

Policy Research,Durham, USAP. WilliamsPAREXEL International,

Uxbridge, Middlesex, UK124IntroductionAccording to the American Migraine Study II, a large population-based survey, approximately 28 million Americans

suffer from migraine [1]. Migraine is more prevalent in

whites than in other ethnic groups, more common in lowerincome than higher-income groups, and has a peak prevalence from ages 25 to 55 years, the most productive years

of an individuals life [1].Migraine is characterized by severe pain and often by

headache-related disability, which causes absenteeism as

well as reduced productivity, imparting a substantial diseaserelated burden on the individual and economic burden on

society. Of all migraineurs in the American Migraine Study

II, 53% reported severe impairment of their daily activities or

a need for bed rest with severe headaches [1]. Moreover, the

Global Burden of Disease Study, sponsored by the World

Health Organization, ranked the disability associated with

severe migraine among the worlds 20 most disabling chronic disorders [2] and lost labor costs in the United States have

been estimated at $13 billion per year [3].Headache-related disability is an important target for

migraine therapy. Yet despite an increase in the diagnosis of

migraine over the past decade, the majority of migraineurs

remain undiagnosed [4] and the proportion of migraine sufferers treating with a prescription medication only increased

modestly from 37% in 1989 to 41% in 1999 [4]. Factors

contributing to these low rates of diagnosis and treatment

include failure to consult a physician, poor patient-physician

communication, and patient reluctance to take a prescription

migraine medication due to safety concerns [4].Treatment patterns, reported in a more recent survey,

show that about half of migraineurs use over-the-counter

medications exclusively, 21% use prescription medications

exclusively, and 23% use both [5]. In the prescription-only

group, 36% used butalbital-caffeine combinations and only

18% used triptans [5]. This underuse of triptans prevails

despite American Academy of Neurology guidelines recommending triptans as the treatment of choice for patients

with moderate-to-severe migraine and for those with

migraine of any severity who have not previously responded to nonspecific medications (e.g. non-steroidal antiinflammatory drugs [NSAIDS], analgesics) [6].Seven orally administered triptans are now available to

treat patients with acute migraine: almotriptan, eletriptan,

frovatriptan, naratriptan, sumatriptan, rizatriptan, and

zolmitriptan. A recent meta-analysis evaluated 6 of these

triptans in more than 24 000 patients in 53 published and

unpublished controlled trials; the conclusions were that

these agents differed with respect to efficacy, consistency

of effect, rate of recurrence, and tolerability, and that these

differences were clinically relevant to the individual [7, 8].

Matching triptans to the needs of individual patients is a

challenging task. Indeed, a study of neurologists learning

needs (when only 4 of the 7 currently available triptans

were on the market) found that while all felt that they used

evidence-based medicine in their daily practice, there was

considerable uncertainty about how to appraise the triptans

and select which to prescribe [9].This issue can be framed as a multiattribute decision

problem, in which a decision maker must select from among

competing alternatives on the basis of multiple criteria and

points of view [10]. Multiattribute decision-making

(MADM) methods, frequently used in public sector and business settings [1113], have only rarely been applied to the

issue of selecting medical interventions, but recent examples

include the Star Systems, used to aid physicians in choosing

among antiepileptic drugs [14], and an evaluation of the performance of imaging techniques for breast cancer [15].In this article, we first report the relative importance of

prespecified treatment attributes for selecting an oral triptan, from the perspectives of 2 groups of headache sufferers. The first group comprised triptan users, while the second group consisted of triptan-naive subjects with high

levels of disability (Migraine disability assessment

[MIDAS] grades III or IV), who might be considered to be

candidates for triptans. We then combine these data with

information on the relative performance of the triptans as

reported in a recent meta-analysis [7, 8] using a multiattribute decision model to identify the preferred triptans.MethodsPopulation surveyData on the relative importance to migraineurs of selected triptan

treatment attributes were collected in a cross-sectional population

survey. A random sample of us households with a telephone were

contacted, and a computer-assisted telephone interview (CATI)

was conducted among eligible and willing headache sufferers

between the ages of 18 and 60 years. We administered both a diagnostic interview and the MIDAS questionnaire by phone and after

eliminating the ineligible candidates, selected 2 groups: the triptanexperienced group (n=206) comprised subjects who treated at least

one headache with a triptan in the past 12 months, and the triptannaive group (n=209) consisted of individuals who never used a

triptan to treat a headache but who reported significant headacherelated disability, as measured by MIDAS scores of III or IV [16].Respondents were questioned about their medical and treatment

history and their headache characteristics. They were asked to evaluate the relative importance in selecting a new treatment for your

headache of a prespecified set of treatment attributes, which were

taken from those used to evaluate oral triptans in the meta-analysis,

thus ensuring the availability of controlled clinical trial data for subsequent comparison [7]. Queries were restricted to the relative

importance of treatment attributes; migraineurs were not asked to

rate the attributes of specific drugs. Treatment attributes were

arranged into a hierarchy of 3 top-level attributes (efficacy, consis-125Fig. 1 Treatment attributes that survey participants were asked to evaluate regarding the choice of a

headache medication. Sustained

pain-free status was defined as no

pain 2 hours after dosing, with no

recurrence of moderate or severe

headache, and no use of any rescue

headache medication 224 hours

after dosing. Central nervous system

(CNS) adverse events included sedation and dizziness. Cardiovascular

(CV) adverse events included chest

pressure, chest pain and palpitations.

Other adverse events included

paresthesias, flushing and limb heavinesstency of effect, and tolerability) and 2 sets of lower-level efficacy

and tolerability attributes (Fig. 1). The lower-level efficacy attributes were pain-free status at 1 hour, pain-free status at 2 hours, and sustained pain-free status (no pain 2 hours postdose, without recurrence

of moderate or severe headache or use of rescue headache medication 224 hours postdose). The lower-level tolerability attributes

were freedom from cardiovascular (CV) adverse events (side effects

such as chest pressure or pain), freedom from central nervous system

(CNS) adverse events (side effects such as such as sleepiness or

dizziness), and freedom from other adverse events (Fig. 1).The relative importance of the attributes was assessed in a

series of pair-wise comparisons, and the ratings from each participant were transformed into importance weights using a matrixmultiplication algorithm [17] and scaled to sum to 100% within

each group. Confidence intervals for the mean importance weights

for the entire sample (separately for the triptan and MIDAS

groups) were estimated in a nonparametric bootstrapping exercise

with 10 000 resamples [18].Triptan performancePerformance data on the efficacy and tolerability attributes of 6 oral

triptans were obtained from a meta-analysis of 53 published and

unpublished, randomized, double-blind, placebo- or active-controlled migraine treatment trials in more than 24 000 patients [7, 8].Multiattribute decision modellingAll MADM problems can be expressed as a matrix. In the MADM

matrix used here, triptan treatment attributes such as efficacy, consistency of effect, and tolerability were plotted against 6 triptan alternatives. The performance of the triptans on these attributes, and weights

reflecting the relative importance of the attributes, were combined

according to prespecified decision rules to generate an overall evaluation of the desirability of, or preference for, each of the triptans. The

attribute importance weights from the survey were combined with the

performance data from the meta-analysis using a modified version of

the MADM model Technique for Order Preference by Similarity to

Ideal Solution (TOPSIS) [19, 20]. In the TOPSIS model, data are first

standardized to take into account differences in measurement units.

Then, in keeping with Zelenys axiom of human choice

(Alternatives closer to the ideal are preferred to those further away.

To be as close as possible to the perceived ideal is the rationale of

human choice.) [21], the model identifies a hypothetical ideal triptan: a composite that, if it really existed, would perform best on every

attribute. The definition of the ideal for the purpose of this model is

best achievable with current technology or optimal. Next, the

model measures the distance of each of the 6 real triptans from the

hypothetical ideal (by calculating the weighted Euclidean distance)[13]. These distance measures are scaled such that 0 is identical to the

anti-ideal (i.e. the worst possible) and 100% is identical to the ideal

(the best possible), while intermediate values measure similarity to

the ideal triptan, providing the basis for comparing the triptans.The models were run twice, first using absolute, and then

placebo-corrected efficacy and consistency data (the tolerability

data were always placebo-corrected). A correction was introduced to remove the effect of sustained pain-free rates for placebo that were significantly higher in the almotriptan studies and

significantly lower in the eletriptan studies [22]. Because there

were no consistency data for zolmitriptan, a consistency level

equal to the mean of all the other products was assumed.To account for the variance in the meta-analysis results and in the

survey data, 95% confidence intervals for the similarity scores were

estimated by incorporating probabilistic uncertainty analysis [23] into126the bootstrapping. As in the meta-analysis, statistical significance was

assessed using sumatriptan 100 mg as the reference product. In each

of the 10 000 uncertainty analysis iterations, the similarity score of

each product was compared to that of the reference product, and triptans were considered significantly superior to sumatriptan 100 mg if

they were closer to the ideal on 95% of the resamples. ResultsThe participants in both the high disability triptan-naive group

and the triptan group were predominantly white women in

their late 30s (Table 1). The time since initial diagnosis was

significantly longer for the triptan-experienced group than for

the triptan-naive group (16 and 12 years, respectively; Mann-

Whitney z=2.42, p<0.02). Subjects in the triptan-experienced

group were more likely to describe the majority of their

headaches as moderate or severe in intensity (2=7.69,

p<0.01), unilateral (2=15.52, p<0.001), and involving

changes in vision (2=6.79, p<0.01), nausea or vomiting

(2=28.64, p<0.0001), and phono- or photophobia (2=11.05,p<0.001). Only a small percentage of participants in either

group felt completely satisfied with the usual treatment (Table1). Respondents were much more likely to be dissatisfied with

an efficacy parameter than with tolerability. Relative importance of triptan treatment attributesTable 2 gives the mean importance weights (and their bootstrap 95% confidence limits) for the top-level efficacy and

tolerability attributes, reported separately for the study

groups. Efficacy attributes were considered to be the most

important of the top-level attributes by both the triptan-naive

group and the triptan-experienced group. For the triptannaive group, but not the triptan-experienced group, tolerability attributes were significantly more important than consistency of effect. This pattern (efficacy>tolerability>consistency for the triptan-naive group; efficacy>[tolerabilityconsistency] for the triptan-experienced group) was consistent

across subgroups defined according to personal and

headache characteristics set out in Table 1 (data not shown).Table 1 Characteristics of survey respondentsTriptan-naive group Triptan-experienced group(n=209) (n=206)Women, n (%) 175 (84) 186 (90)

Men, n (%) 34 (16) 20 (16)Age, yearsa 37 (11.2) 39 (10.6)

Caucasian-white race, n (%) 166 (79) 176 (85)Age at diagnosis, yearsa 22 (9.9) 22 (10.6) Years since diagnosisb 12 (621) 16 (825)Headache days in previous 3 monthsb 8 (520) 8 (320)

50% of headaches of moderate-severe intensity, n (%) 166 (79) 184 (89)Headache duration, with treatment, hb 3 (112) 4 (124)

Headache symptoms, n (%)Changes in vision on 50% of occasions 77 (37) 102 (50)

Unilateral headache on 50% of occasions 111 (53) 148 (72)

Nausea or vomiting on 50% of occasions 95 (45) 147 (71)

Phono- or photophobia on 50% of occasions 171 (82) 191 (93)

Headache-related disability, MIDAS grades III or IV 209 (100) 118 (57)

Satisfaction with usual treatment, n (%)Completely satisfied 18 (9) 26 (13)

Inadequate level of pain relief 36 (17) 16 (8)

Relief takes too long 47 (22) 31 (15)

Doesnt always work 46 (22) 51 (25)

Recurrence 50 (24) 42 (20)

Adverse events 7 (3) 18 (9)

Other/not stated 5 (3) 22 (10)MIDAS, migraine disability assessmenta Values are mean (SD)b Values are median (range)127Table 2 Relative importance of the triptan treatment attributes. Values are mean importance weights (bootstrap 95% confidence intervals)Attribute Triptan-naive group Triptan-experienced group(n=209) (n=206)Top-levelEfficacy 45 (4151) 49 (4152)

Consistency of effect 23 (1824) 28 (2231)

Tolerability 32 (2940) 23 (2229)

EfficacyPain-free status at 1 h 58 (5263) 56 (5562)

Pain-free status at 2 h 13 (1116) 14 (1015)

Sustained pain-free status 29 (2534) 30 (3140)

TolerabilityFreedom from CNS adverse events 20 (1728) 20 (1527)

Freedom from CV adverse events 47 (4155) 54 (4765)

Freedom from other adverse events 34 (2536) 27 (1931)CNS, central nervous system; CV, cardiovascularNo differences were observed between the triptannaive group and triptan-experienced groups with respect to

the relative importance ascribed to the 3 efficacy attributes

(Table 2). According to the respondents, pain-free status at

1 hour was the most important of the efficacy attributes,

followed by sustained pain-free status. This pattern of

results (pain-free status at 1 h > sustained pain-free status

> pain-free status at 2 h) was observed consistently across

subgroups defined according to the personal and headache

characteristics set out in Table 1.Similarly, no differences were observed between the

groups for the relative importance assigned to the 3 tolerability attributes (Table 2). About half the total weight

ascribed to tolerability was accounted for by the most

important tolerability attribute, freedom from CV adverse

events. The order of importance (freedom from CV

adverse events > other adverse events > CNS adverse

events) was consistent across subgroups defined according

to personal and headache characteristics (data not shown).Similarity to the hypothetical ideal triptanImportance weights elicited from survey participants were

combined with clinical trial data from the meta-analysis in

a TOPSIS model. Mean similarity to the ideal scores, 95%

confidence limits, and statistical significance levels were

estimated for each product as described previously. At the

group level of analysis, 3 oral triptans (almotriptan,

eletriptan, and rizatriptan) were significantly closer to the

hypothetical ideal triptan than was the reference product,Fig. 2a, b Similarity of

triptans avaiable for oral

use to the ideal triptan. a

Triptan-naive group. b

Triptan-experienced group.

Values are mean (95% CI)a b128sumatriptan 100 mg (Fig. 2). This finding was true for both

study groups. The 40-mg dose of eletriptan was significantly closer to the ideal than sumatriptan 100 mg in the

triptan-naive group when placebo-corrected or absolute

data were used, but significantly closer in the triptan-experienced group only when placebo-corrected data were

used. Almotriptan 12.5 mg, eletriptan 80 mg, and rizatriptan 10 mg were significantly better than sumatriptan 100

mg regardless of whether absolute or placebo-corrected

data were used. At the individual level of analysis,

almotriptan, eletriptan, and rizatriptan constituted the top

3 closest to the hypothetical ideal for 77%78% of the

triptan-naive group and 85%86% of the triptan-experienced group of migraine sufferers, based on the importance weights elicited from the participants.DiscussionAccording to the survey respondents, efficacy was significantly more important than tolerability or consistency of

effect in selecting an oral triptan. For the triptan-naive

(high MIDAS score) group, tolerability was significantly

more important than consistency, while the triptan-experienced group attached equal importance to tolerability and

consistency. Pain-free status at 1 hour and freedom from

CV adverse events were considered to be the most important lower-level attributes for both groups. This pattern

was consistent across subgroups defined according to personal and headache characteristics.Converging evidence indicates that complete and rapid

pain relief without recurrence is the most important treatment attribute for migraine sufferers, as is also observed in

the present survey. According to 648 migraineurs participating in a placebo-controlled sumatriptan clinical trial [24],

the 4 most important attributes of migraine therapy were

how well it works, how safe it is, how fast it works,

and side effects; the least important attribute was cost of

drug. In a population-based telephone survey, 688

migraineurs reported that complete pain relief (87%), no

recurrence (86%), rapid onset (83%), and no side effects

(79%) were the most important attributes of acute treatment.

Of these patients, 29% were very satisfied with their acute

migraine therapy, 48% were somewhat satisfied, 7% were

neither satisfied nor dissatisfied, 9% were somewhat dissatisfied and 7% were very dissatisfied [25, 26]. Reasons for

dissatisfaction were pain relief taking too long (87%),

incomplete pain relief (84%), inconsistent relief (84%),

headache recurrence (71%), and too many side effects

(35%). These results are supported by post-hoc analyses of

clinical trials showing that fast and complete pain relief predicts satisfaction and health-related quality-of-life [2729].Weightings of the treatment attributes derived from

migraine sufferers in the present study can be compared

with those of neurologists and primary care physicians

(PCPs) who were surveyed as part of the TRIPSTAR project [30, 31]. Both physicians and migraine sufferers rated

efficacy as the most important top-level attribute. Like the

triptan-naive (high MIDAS score) respondents themselves,

physicians rated tolerability as more important than consistency for their triptan-naive patients. Physicians and

migraine sufferers agreed that freedom from CV adverse

events was the most important lower-level tolerability

attribute but disagreed as to the most important lower-level

efficacy attribute. Migraine sufferers weighted pain-free

status at 1 hour as more important than sustained pain-free

status, whereas physicians believed sustained pain-free

status was more important.The attribute importance weights obtained from the

migraine sufferers were combined in a TOPSIS model with

clinical trial data from a meta-analysis [7, 8]. Three triptans (almotriptan, eletriptan, and rizatriptan) were consistently found to be significantly closer to the hypothetical

ideal of the decision model than was the reference product,

sumatriptan 100 mg. This was true for both the triptannaive (high MIDAS score) group and the triptan-experienced group, using either absolute or placebo-corrected

data at both the group and individual analysis levels.These results must be assessed in the context of limitations imposed by the multiattribute decision-making

model, its inherent assumptions and its structure, as well

as the data used in the models. The TOPSIS model used

here was chosen not only because of its intuitive appeal

(similarity to the ideal), but also because it makes no

distributional assumptions beyond preference independence and additivity, which are common to all decision

models. An important model-related issue, however, is

that the outcome of a model depends on the inputs. In the

present context, a different selection of treatment attributes may well have led to a different set of results. For

example, if a familiarity of use attribute had been

included, this would have influenced the model in favor of

the older triptans. Also, the model used for the TRIPSTAR

project did not include any subjective attributes, and only

treatment-related attributes for which robust data are

available were included, i.e. those studied in the recent

meta-analysis [7, 8].The 3 oral triptans that emerged as superior in this

study also constituted the preferred subset based on the

surveys of US neurologists and PCPs carried out as part of

the TRIPSTAR project [30, 31]. Furthermore, attribute

importance weights elicited in response to a case-history

presented at a symposium yielded the same set of preferred

products (almotriptan, eletriptan, and rizatriptan) when

combined with the meta-analysis data in a TOPSIS model129[8, 32]. A reasonable degree of convergent validity for the

superiority of these 3 agents (based on the treatment attributes included in the model) can therefore be said to exist.The explanation for the consistency of this finding lies

in the dominance hierarchy found in the meta-analysis

results. As McCrory [33] pointed out, all other triptans are

dominated (in a decision-analytic sense) by one or more of

these 3 products, and within this subset of preferred products (almotriptan, eletriptan, rizatriptan), no one dominates

any of the others. The superiority of this trio was further

supported in a TOPSIS model which used computer-generated attribute importance weights representing the entire

range of possible values for the relative importance of the

treatment attributes [34].Multiattribute decision-making models such as TOPSIS have a potential role in improving the match between

patients needs and treatment choice. Computer-based clinical decision support systems are becoming increasingly

prevalent, ranging in complexity from simple patient recall

systems through drug-dosing calculators, to complex diagnostic tools linked to electronic medical records [35]. A

review of meta-analyses of trials of the effectiveness of

such systems concluded that they may provide significant

benefits in the process of care [35]. A model of the kind

used in the TRIPSTAR project has considerable potential

as a decision-support tool in the management of patients

with migraine, as it provides a convenient way of incorporating migraine sufferers own viewpoints with objective

data on the performance of the oral triptans.A recent study by Lipton et al. [5, 36] revealed that

while 48% of US migraineurs consulted with physicians

over the last year, 21% had lapsed from care, many believing that there was nothing a physician could do to help

them. Tailoring treatment to individuals priorities may

increase the level of satisfaction among migraineurs and

improve adherence to treatment. While the results of the

TRIPSTAR project should not influence physicians to

change the triptan therapy of migraineurs who are satisfied

with their current treatment, the evidence nevertheless suggests that selection from the subset of preferred triptans

(almotriptan, eletriptan, and rizatriptan) is a useful starting

point for newly diagnosed migraineurs and for those who

are dissatisfied with their current treatment.Acknowledgement This project was supported by a grant from

Pharmacia Corporation.References1. Lipton RB, Stewart WF, Diamond S,

Diamond ML, Reed M (2001)

Prevalence and burden of migraine in

the United States: data from the

American Migraine Study II.

Headache 41(7):6466572. Menken M, Munsat TL, Toole JF

(2000) The global burden of disease

study: implications for neurology.

Arch Neurol 57(3):4184203. Hu XH, Markson LE, Lipton RB,

Stewart WF, Berger ML (1999)

Burden of migraine in the United

States: disability and economic costs.

Arch Intern Med 159(8):8138184. Lipton RB, Diamond S, Reed M,

Diamond ML, Stewart WF (2001)

Migraine diagnosis and treatment:

results from the American Migraine

Study II. Headache 41(7):6386455. Lipton RB, Scher AI, Steiner TJ,

Bigal ME, Kolodner K, Liberman JN,

Stewart WF (2003) Patterns of health

care utilization for migraine in

England and in the United States.

Neurology 60(3):4414486. Silberstein SD (2000) Practice parameter: evidence-based guidelines for

migraine headache (an evidence-based

review): report of the Quality

Standards Subcommittee of the

American Academy of Neurology.

Neurology 55(6):7547627. Ferrari MD, Roon KI, Lipton RB,

Goadsby PJ (2001) Oral triptans

(serotonin 5-HT1B/1D agonists) in

acute migraine treatment: a metaanalysis of 53 trials. Lancet

358:166816758. Ferrari MD (2002) TRIPSTAR: A

comprehensive patient-based

approach to compare triptans.

Headache 42[Suppl 1]:S18S259. Purdy RA (2000) Evidence-based

migraine therapy: learning needs and

knowledge assessment. Cephalalgia

20[Suppl 2]:5910. von Winterfeldt D, Edwards W (1987)

Decision analysis and behavioral

research. Cambridge University Press,

Cambridge11. Barnes JA, Rutherford GS (1997)

Analysis of the initial application of

the State of Washington highway

mobility project: ranking procedure

and recommended revisions for the

upcoming biennium. Washington State

Department of Transportation Report

Number WA-RD 428.1. Abstract available at http://www.wsdot.wa.gov/

research/onepages/WA-RD4281.htm

(accessed 17 June 2004)12. Jackson HV (1999) A structured

approach for classifying and prioritizing product requirements. PhD thesis,

North Carolina State University.

Available at http://www.lib.ncsu.edu/

etd/public/etd 354921279912131/

etd.pdf (accessed 17 June 2004)13. Deng H, Yeh CH, Willis RJ (2000)

Inter-company comparison using modified TOPSIS with objective weights.

Comput Operations Res 27:96397314. Brodie MJ, Kwan P (2001) The Star

Systems: overview and use in determining antiepileptic drug choice. CNS

Drugs 15(1):11213015. Azar F (2000) Multi-attribute decision

making: use of three scoring methods

to compare the performance of imaging techniques for breast cancer

detection. Avaiable at

http://www.cis.upenn.edu/~fredazar/pu

blications/pdf/fredazar_techreport_M

S-CIS-00-10.PDF (accessed 17 June

2004)16. Stewart WF, Lipton RB, Dowson AJ,

Sawyer J (2001) Development and

testing of the Migraine Disability

Assessment (MIDAS) questionnaire

to assess headache-related disability.

Neurology 56[6 Suppl 1]:S20S2817. Dolan JG (1989) Medical decision

making using the analytic hierarchy

process: choice of initial antimicrobial

therapy for acute pyelonephritis. Med

Decis Making 9(1):515618. Mooney CZ, Duval RD (1993)

Bootstrapping: a nonparametric

approach to statistical interference.

Sage, Newbury Park19. Hwang CL, Yoon KP (1981)

Multiattribute decision making: methods and applications. Springer, Berlin

Heidelberg New York, pp 12814020. Yoon KP, Hwang CL (1995) Multiple

attribute decision making: an introduction. Sage, Thousand Oaks, pp

384021. Zeleny M (1982) Multiple criteria

decision making. McGraw Hill, New

York22. Roon KI, Lipton R, Goadsby PJ,

Ferrari M (2001) Placebo in triptan

trials: efficacy, tolerability and consistency. Cephalalgia 21:40543223. Baltussen RM, Hutubessy RC, Evans

DB, Murray CJ (2001) Uncertainty in

cost-effectiveness analyses: probabilistic uncertainty analysis and stochastic league tables. GPE Discussion

Paper Series, no. 34. World Health

Organization, Geneva Available at

http://www3.who.int/whosis/statistics/discussion_papers/pdf/paper34.pd

f (accessed 17 June 2004)24. Luciani RJ, Osterhaus JT, Gutterman

DL (1995) Patient preferences for

migraine therapy: subcutaneous sumatriptan compared with other medications. J Fam Pract 41:14715225. Lipton RB, Stewart WF (1999) Acute

migraine therapy: do doctors understand what patients with migraine

want from therapy? Headache

39[Suppl 2]:S20S2626. Lipton RB, Hamelsky SW, Dayno JM

(2002) What do patients with migraine

want from acute migraine treatment?

Headache 42[Suppl 1]:S3S927. Davies GM, Santanello N, Lipton R

(2000) Determinants of patient satisfaction with migraine therapy.

Cephalalgia 20(6):55456028. Santanello NC, Davies G, Allen C,

Kramer M, Lipton R (2002)

Determinants of migraine-specific

quality of life. Cephalalgia

22(8):68068529. Loder E, Brandes JL, Silberstein S,

Skobieranda F, Bohidar N, Wang L,

Boyle D, Kolodny A, Guerra F,

Santanello N, Johnson-Pratt L (2001)

Preference comparison of rizatriptan

ODT 10-mg and sumatriptan 50-mg

tablet in migraine. Headache

41(8):74575330. Dodick D, Cutrer FM, Ferrari M,

Goadsby PJ, Lipton RB, McCrory DC,

Williams P (2002) Prioritization of

treatment attributes in selecting an

oral triptan: a survey of US neurologists. Headache 42:39231. Cutrer FM, Goadsby PJ, Ferrari M,

Lipton R, Dodick D, McCrory D

(2002) Prioritization of treatment

attributes in selecting an oral triptan: a

survey of U.S. primary care physicians. Headache 42:39239332. Dodick DW, Lipton RB, Goadsby PJ,

McCrory D, Ferrari M, Williams P,

Cutrer FM (2002) Prioritizing triptan

treatment attributes: a pilot study.

Neurology 58[Suppl 3]:A12933. McCrory D (2002) Dominance relationships in the triptan meta-analysis.

In: 14th Annual Migraine Trust

International Symposium, September

2326, London, UK34. Ferrari MD (2002) Tripstar: a comprehensive patient-based approach to

compare triptans. Headache 42[Suppl1]:1825 35. Trowbridge R, Weingarten S (2001)

Clinical decision support systems. In:

Making health care safer: a critical

analysis of patient safety practices.

Evidence Report/Technology

Assessment, no. 43. Agency for

Healthcare Research and Quality,

Rockville (AHRC publication no. 01-

E058, Chapt. 53). Available at

http://www.ahcpr.gov/clinic/ptsafety/c

hap53.htm (accessed 17 June 2004)36. Lipton RB, Scher AI, Kolodner K,

Liberman J, Steiner TJ, Stewart WF

(2002) Migraine in the United States:

epidemiology and patterns of health

care use. Neurology 58(6):885894