How Treatment Priorities Influence Triptan Preferences in Clinical Practice: Perspectives of Migraine Sufferers, Neurologists, and Primary Care Physicians

Key words: Migraine – Neurologists – Primary care physicians – TRIPSTAR – Triptans

ABSTRACT

Background: In treating migraine sufferers, physicians can choose from among seven triptans with different attributes.

Objective: To develop a system for selecting an oral triptan based on treatment priorities of migraine sufferers, neurologists, and primary care physicians (PCPs) in the United States, and evidence-based performance of triptans in clinical trials.

Methods: The TRIPSTAR project combines data on the treatment preferences of migraineurs and physicians with results from a meta- analysis of individual triptans, which evaluated their effectiveness on various clinical endpoints. Telephone interviews with migraine sufferers, neurologists, and PCPs were conducted to elicit individual views on the relative importance of a prespecified set of acute treatment outcomes. Four hundred and fifteen migraine sufferers, both triptan-experienced and triptan-naive, were interviewed. Also, 200 board-certified neurologists and 200 PCPs provided information on migraine patients from their clinical practice. A multiattribute decision model for selecting an oral triptan was constructed using attribute importance weights collected at telephone interview and the meta-analysis data, which were drawn from 53 clinical trials of 6 oral triptans.

Results: Efficacy attributes were rated significantly more important than tolerability or consistency in selecting an oral triptan, according to migraine sufferers and physicians. Freedom from cardiovascular adverse events was the most important tolerability attribute, according to migraine sufferers and physicians alike. Pain free at 1 h was the most important lower- level efficacy attribute for migraine sufferers, while sustained pain free was most important for physicians. When weighted treatment attributes were combined with meta-analysis data in a multi- attribute decision model, almotriptan 12.5mg, eletriptan 80 mg, and rizatriptan 10mg were significantly closer to the hypothetical ideal triptan than was sumatriptan 100 mg. Triptans selected by the model were generally closer to the patient-specific ideal triptan than were the triptans prescribed by physicians.

Conclusions: Almotriptan, eletriptan, and rizatriptan were the three triptans closest to the ideal, from the perspectives of migraine sufferers, PCPs, and neurologists alike. The TRIPSTAR model may be a potentially useful decision-support tool to help physicians select the triptan most likely to produce a successful outcome in migraine sufferers.

Introduction

Surveys in the United States and elsewhere have shown that migraine is a common condition occurring in about 12% of the US population and is associated with severe pain and disability1-3. Only 50% of survey participants who met International Headache Society (IHS) criteria for migraine have received a medical diagnosis of migraine, and less than 20% received a prescription medication for their migraine1,4. Over 50% of patients reported moderate or severe disability or the presence of symptoms consistent with Grade III or IV migraine disability assessment score (MIDAS)1. These findings on prevalence, disability, and treatment patterns are consistent across a range of studies from the general population2,4.

Surveys of migraine patients have consistently reported that the most important attributes of migraine treatment are rapid and sustained relief from pain, and absence of side effects1,5,6. The older anti-migraine drugs, including simple and combination analgesics, nonsteroidal anti-inflammatory drugs (NSAIDs), antiemetics, and ergot compounds, produce less than optimal efficacy and tolerability7,8. The introduction of the selective 5-HT^sub 1^ agonists, or triptans, dramatically expanded the options available to migraine sufferers and the American Academy of Neurology recommends triptans as first-line agents for moderate to severe migraine headaches or headaches of any severity when nonspecific medications did not provide adequate relief in the past9. Seven oral triptans are now available in the US: almotriptan, eletriptan, frovatriptan, naratriptan, rizatriptan, sumatriptan, and zolmitriptan. Although pharmacologically similar, these drugs show clinically meaningful differences in efficacy, onset of action, rate of headache recurrence, and tolerability, notably chest symptoms and central nervous system (CNS) adverse events (AEs)10,11. While non- oral administration can provide rapid absorption, high bioavailability, and increased response rates8,12 patients favor oral formulations for convenience and tolerability1,6,10. When migraine sufferers seek medical care, an individualized treatment plan which reflects the patients’ headache characteristics and preferences should be developed9. However, no specific treatment algorithm exists to guide physicians in matching their migraine patient with a particular triptan. The TRIPSTAR project was conceived to provide an answer to the question: from the treatment options available, how can physicians best choose from among the oral triptans to optimize the match between product attributes and patient needs?

Multi-Attribute Decision Making (MADM) methods were developed to assist decision-makers in selecting among competing alternatives on the basis of multiple, often conflicting criteria and different points of view13. These have been widely used in public-sector and business settings, but their use in selecting drug products appears to be relatively limited, although a number of reports have appeared in the literature over the past decade or so14-19. MADM problems can be represented as a matrix in which, in the current context, each column represents a treatment attribute (efficacy, tolerability, etc.) while each row represents an alternative treatment (triptan). The cells contain data for the relative performance of the triptans on the attributes. A final row is added, containing weights which express the relative importance of the attributes. The data on performance of the alternatives and the information on the relative importance of the attributes can then be systematically combined according to prespecified decision rules, to result in an overall evaluation of the desirability of or preference for each alternative triptan.

The TRIPSTAR project has the following objectives: (1) to solicit the viewpoints of neurologists, primary care physicians (PCPs), and migraine sufferers on the relative importance of a set of prespecified treatment attributes of oral triptans; (2) to collect information on the relative performances of the oral triptans by reference to a comprehensive meta0 -analysis10,11; and (3) to combine both sets of data in a multi-attribute decision model to identify the preferred triptans.

Design and methods

Relative importance of treatment attributes: headache sufferers

To solicit opinions of headache sufferers, a sample of US adults (at least 18 years of age) was contacted by telephone using random digit dialing and asked to participate in a study of headaches. Those who had used a triptan for at least one headache in the previous 12 months were eligible for inclusion in the triptan- experienced group, while those whose headache experience put them into Grades III or IV (moderate or severe migraine-associated disability) on the Migraine Disability Assessment Scale (MIDAS20), but had never used a triptan, were eligible for inclusion in the triptan-naive group. At least 200 eligible and consenting participants were obtained for each group. These two groups were selected to permit us to contrast a group that was using triptans with a group that may well benefit from them21,22.

Participants were interviewed regarding their medical and treatment history and headache characteristics (details have been presented elsewhere23). They were asked to rank the relative importance of a prespecified set of treatment attributes encompassing efficacy, consistency of effect, and tolerability that were carefully defined (Table 1). These attributes were selected based on the outcomes of Ferrari et al.’s meta-analysis10,11 and were focused on three top-level (general) attributes: efficacy, tolerability, and consistency. Two sets of lower-level (specific) attributes for efficacy (pain free at 1 h, pain free at 2 h, and sustained pain free) and for tolerability (freedom from CNS, CV, and other adverse events) were defined. The relative importance of top- level and lower-level attributes was assessed by pairwise comparisons. Using a matrix-multiplication algorithm24, the rankings from each participant were transformed into importance weights, scaled to sum to 100% within each group24. Confidence intervals for the mean importance weights were estimated in a nonparametric bootstrapping exercise with 10000 resamples25.

Table 1. Treatment attributes

Relative importance of treatment attributes: neurologists and primary care physicians (PCPs)

Surveys of US neurologists and PCPs were conducted based on patients in their routine clinical practice. A sample was drawn from a commercial database of US physicians to conform to prespecified quotas for gender26, regional distribution27, and duration of experience, to ensure representation across the experience range. Pote\ntial participant physicians were contacted by telephone and fax, and those who reported that they had prescribed an oral triptan for one or more migraine patients each week over the past 3 months were considered eligible to participate. Recruitment was complete when there were 200 eligible and willing participants in each physician group. Participating physicians were informed that the aim of the study was ‘to understand factors that doctors take into account when they select drug treatment for patients with migraine’. They did not receive any further information regarding the research objectives or how the data were to be analyzed. Participants received reimbursement for their time.

Participating physicians were asked to select and review 2 patients in advance of the telephone interview: the migraine patient most recently started on an oral triptan for the first time (a triptan-naive patient), and the migraine patient who most recently exchanged one oral triptan for another (a triptan-experienced patient). During the interview, physicians were asked to profile the 2 selected patients according to headache characteristics and medical treatment history. The physician participants were then asked to review their recent triptan selection decisions for each patient and to evaluate the relative importance (for triptan choice) of the triptan treatment attributes (Table 1). The physicians were queried only about the relative importance of treatment attributes; they were not asked to rate the attributes of any specific triptan. Details of the neurologist and PCP interviews have been presented elsewhere28,29.

Multi-attribute decision modeling

Data from the 3 inquiries (neurologists, PCPs, and headache sufferers) were analyzed separately. In each analysis, the attribute importance weights elicited from the participants were combined with data on the relative performance of the triptans, taken from a metaanalysis of 53 controlled clinical trials of 7 oral triptans in the treatment of migraine, involving more than 24000 patients10,11. A modified version of the multiattribute decision model TOPSIS (Technique for Order Preference by Similarity to Ideal Solution)30,31 was used. The rationale and decision rule for the TOPSIS model is based on the concept of a hypothetical ideal alternative, a composite, which, if it really existed, would have the highest rank on every attribute32. In the model used in this study, the definition of the ‘ideal’ is ‘best achievable with current technology’, and thus may be described as ‘optimal’. The differences between each triptan and the hypothetical ideal were calculated33 and converted into a similarity score for each triptan, scaled so that 100% is equivalent to the ideal (best possible) and 0% is equivalent to the anti-ideal (worst possible), while intermediate values measure the degree of similarity to the ideal triptan. Comparisons among triptans can then be made systematically on the basis of their similarity scores.

Efficacy and consistency data were input into the model, first as absolute values, and then as placebocorrected values (tolerability data were always placebocorrected). Because the placebo response rates were significantly higher than average in the almotriptan studies and significantly lower than average in the eletriptan studies34, the calculations incorporated a correction to account for this. Because there were no consistency data for zolmitriptan, a consistency level equal to the mean of all the other products was assumed.

To address the variation in interview and metaanalysis data, 95% confidence intervals for the similarity scores were estimated by incorporating probabilistic uncertainty analysis into the bootstrapping35. As in the meta-analysis10, statistical significance was assessed using sumatriptan 100mg as the reference product. As the similarity scores for the different triptans were within- subject measures, significance could not be assessed by comparing confidence intervals. Instead, the similarity score for each triptan was compared to that of sumatriptan 100 mg, for each of the 10 000 uncertainty analysis iterations. Products that were closer to the ideal on ≥ 95% of the resamples were considered significantly superior to the reference product.

Results

Study participants

Headache sufferers who were interviewed, whether in the triptan- naive group (N = 209) or the triptan-experienced group (N = 206) were predominantly white females in their late 30s (Table 2). As might have been expected, significantly more time had elapsed since initial diagnosis in the triptan-experienced group than in the triptan-naive group (medians 16 years and 12 years, respectively, Mann-Whitney z = 2.42, p < 0.02). In addition, more sufferers in the triptan-experienced group than in the triptan-naive group described the majority of their headaches as moderate-to-severe in intensity (χ^sup 2^ 7.69, p < 0.01), unilateral (χ^sup 2^ 15.52, p < 0.001), and involving nausea/vomiting (χ^sup 2^ 28.64, p < 0.0001), phono/photophobia (χ^sup 2^ 11.05, p < 0.001), and/or changes in vision (χ^sup 2^ 6.79, p < 0.01). Only a minority (9%-13%) felt completely satisfied with their usual headache treatment (Table 2). Dissatisfaction with all efficacy parameters (rate, speed, consistency of pain relief, and rate of headache recurrence) was considerably more prevalent than dissatisfaction with tolerability.

A total of 984 neurologists and 922 PCPs were contacted to provide the samples of 200 participants for each physician group. The samples for participating neurologists (N = 200) and PCPs (N = 200) were comparable and similar to the prespecified quotas, with the exception that physicians in practice for less than 10 years or for 20 years or more were somewhat overrepresented in both samples (Table 3). However, as these were arbitrary quotas set simply to ensure a broad representation of experience, this was not considered important.

Physicians were asked to select the migraine sufferers who were triptan-naive and triptan-experienced. The characteristics of these physician-selected patients (Table 4) were broadly comparable to the headache sufferers in the population identified by random digit dialing (Table 3). Physician-selected migraine sufferers were somewhat younger (mean age 33 years-34 years vs 37 years for triptan- naive, and 36 years-37 years vs 39 years for triptan-experienced), were diagnosed at a later age (25 years-27 years vs 22 years of age for triptan-naive, and 24 years-26 years vs 22 years of age for triptan-experienced), and were less likely to have a high grade of headache-related disability in the triptan-naive group (63%-77% vs 100% MIDAS Grades III/IV). Relative to patients selected by neurologists, aura was substantially overrepresented among patients selected by PCPs.

Table 2. Characteristics of headache sufferers identified by telephone interview

Table 3. Characteristics of participating physicians

Table 4. Characteristics of migraine sufferers selected by physicians

Relative importance of triptan treatment attributes

The mean importance weights (and their bootstrap 95% confidence limits), as judged by the neurologists, PCPs, and headache sufferers for 3 categories of triptan attributes were determined separately for triptan-naive and triptan-experienced migraine sufferers (Table 5). We rely on the fact that ‘if two confidence intervals do not overlap, a comparable statistical test would always indicate a statistically significant difference’36. This avoids the need for multiple significance testing and consequent Bonferoni-type adjustment.

Top-level attributes

Efficacy was considered the most important of the top-level attributes, for triptan-naive and triptan-experienced migraine sufferers alike, whether based on data from headache sufferers, neurologists or PCPs (Table 5). For triptan-naive individuals, tolerability attributes were more important than consistency of effect, regardless of whether the data came from migraine sufferers or physicians (Table 5). For triptanexperienced migraineurs, consistency and tolerability were of approximately equal importance. Attribute ranking (efficacy > tolerability > consistency for triptan- naive; efficacy > [tolerability [asymptotically =] consistency] for triptan-experienced) did not vary according to personal or headache characteristics (as listed in Tables 2 and 4).

Efficacy attributes

Migraineurs, whether triptan-naive or triptan-experienced, considered pain free at 1 h to be the most important efficacy attribute, followed by sustained pain free (pain free by 2 h postdose, without recurrence of moderate or severe headache and without use of headache rescue medication 2h-24h postdose) (Table 5). By contrast, participating physicians considered sustained pain free to be the most important efficacy attribute, followed by pain free at 1 h, whether they were reporting on triptan-naive or triptan- experienced patients. This ranking of attributes (pain free at 1 h > sustained pain free > pain free at 2 h, for migraine patients; sustained pain free > pain free at 1 h > pain free at 2h, for physicians) did not vary according to patient and headache characteristics.

Tolerability attributes

Both headache sufferers and physicians were informed that typical triptan-induced CV adverse events are chest pressure and chest pain10. Study participants – headache sufferers, neurologists, and PCPs – all considered freedom from CV adverse events to be the most important of the tolerability attributes for both triptan-naive and triptan-experienced migraine sufferers (Table 5). Approximately half of the total weight ascribed to tolerability was accounted for by this single attribute.

Table 5. Relative importance of the triptan treatment attributes

Typical CNS adverse events, asthenia, abnormal dreams, agitation, aphasia, ataxia, confusion, dizziness, somnolence, speech disorder, abno\rmal thinking, tremor, vertigo, and other focal neurologic symptoms, were defined during the interview10. While participating physicians ranked freedom from CNS adverse events as more important than freedom from other adverse events, participating individuals with migraine reversed this order of importance. This ranking of attributes (freedom from CV adverse events > freedom from other adverse events > freedom from CNS adverse events, for migraine sufferers; freedom from CV adverse events > freedom from CNS adverse events > freedom from other adverse events, for physicians) did not vary according to triptan experience or according to personal or headache characteristics.

Similarity of oral triptans to the hypothetical ideal triptan – population analysis

The importance weights elicited from the participants were combined with the meta-analysis data in a TOPSIS model, and mean ‘similarity to the ideal’ scores, 95% confidence limits, and statistical significance levels were estimated for each triptan as described above. Three oral triptans – almotriptan 12.5mg, eletriptan (40mg or 80 mg), and rizatriptan lOmg – were significantly closer to the hypothetical ideal triptan than was the reference product, sumatriptan 100mg for both triptan-nave (Figure 1) and triptan-experienced migraine sufferers (Figure 2). Because placebo rates vary among studies, we prefer to use placebo- corrected data for making cross-study comparisons. Eletriptan 40 mg scored significantly better than sumatriptan 100mg when placebo- corrected, but not absolute, data were analyzed. The other top scorers, (almotriptan 12.5mg, eletriptan 80 mg, and rizatriptan 10mg), were significantly better than sumatriptan 100mg, irrespective of whether absolute or placebo-corrected data were used.

Individualizing treatment choice

At the individual level of analysis, almotriptan 12.5mg, eletriptan 80mg, and rizatriptan 10 mg were the top 3 (closest to the hypothetical ideal) for 77%-78% of the migraine sufferers, 90%- 91% of the PCP-selected patients, and 94%-95% of the neurologist- selected patients in the triptan-naive category; and for 85%-86% of the migraine sufferers, 95%-96% of the PCP-selected patients, and 96% of the neurologist-selected patients in the triptan-experienced category, whether absolute or placebo-corrected data were used in the analyses.

For each patient (triptan-nave or -experienced) described by a participating physician (neurologist or PCP), we used the TOPSIS model to select the triptan which best approximated that patient’s ideal. We then compared the mean similarity to the ideal of the physician-selected triptan (Table 6). Mean similarity to the ideal, as defined by the model, was 72%-74% for the model-assigned triptans, compared with only 51%-57% for the prescribed triptans, whether the prescriptions were written by neurologists or PCPs for triptan-naive or triptan-experienced patients. The frequency with which a prescribed triptan matched one of the triptans assigned by the model was somewhat higher for patients who were triptan- experienced (32%) than for those who were triptan-naive (18%-20%).

Discussion

The 3 preferred triptans in this MADM were almotriptan 12.5mg, eletriptan 40mg or 80mg; and rizatriptan 10mg. This finding applied across different analyses including triptan-naive vs triptan- experienced migraineurs, using placebo-corrected data, at the group or individual level, and was independent of the origin of the attribute importance weights (neurologists, PCPs or migraine sufferers themselves). Furthermore, attribute importance weights elicited in response to a case history presented to an audience of physicians and combined with the meta-analysis data in a TOPSIS model yielded the same preferred triptans – almotriptan 12.5mg, eletriptan 40mg and 80mg, and rizatriptan 10mg37. This combined body of evidence suggests that a reasonable degree of convergent validity exists for the superiority of almotriptan, eletriptan, and rizatriptan based on the treatment attributes included in the model. The explanation for the consistency of this finding lies in the dominance hierarchy observed in the meta-analysis results. McCrory and Williams38 demonstrated that in a decision-analytic sense, all other triptans are dominated by one or more of these 3 products, and within this subset of almotriptan, eletriptan, and rizatriptan, no one drug dominates any of the others. The superiority of almotriptan, eletriptan, and rizatriptan was further supported in a TOPSIS model which used computergenerated attribute importance weights representing the entire range of possible values for the relative importance of the treatment attributes16,39.

Figure 1. Similarity to the ‘ideal triptan’ (triptan-nave migraineurs). (a) Using absolute data; (b) using placebo-corrected data, p-values indicate that a triptan is statistically significantly closer to the ideal (at the p < 0.01 level or beyond) than is the reference product, sumatriptan 100 mg. The hypothetical ideal triptan is a composite, which, if it really existed, would have the highest ranking on every attribute

Figure 2. Similarity to the ‘ideal triptan’ (triptan-experienced migraineurs). (a) Using absolute data; (b) using placebo-corrected data, p-values indicate that a triptan is statistically significantly closer to the ideal (at the p < 0.01 level or beyond) than is the reference product, sumatriptan 100mg. The hypothetical ideal triptan is a composite, which, if it really existed, would have the highest ranking on every attribute

Table 6. Agreement between prescribed triptan and model-assigned triptans (based on placebo-corrected data)

A number of limitations should be kept in mind when considering the conclusions from this study. First, the physician interviews were based on quota, rather than random samples. While the samples reflect a broad range of medical opinion, as evidenced by the regional distribution and range of experience of the participants, they cannot be claimed to be statistically representative. Our response rate, however, is very similar to response rates obtained in random sample surveys of physicians. For example, Lipton et al. could achieve a response rate of only 25% in a survey of members of the American Academy of Neurologists, even with sponsorship by the Academy40.

A second limitation is that physicians’ judgments of the relative importance of attributes were based on recall. Inaccuracy of recall continues to be an issue for survey researchers41. It may be that in the present study, the attribute importance weights reflected the general view of physicians, rather than patient-specific judgments. This would account for the consistency, or lack of variation, in the pattern of attribute importance across subgroups defined according to patient and headache characteristics.

A third limitation is that imposed by the assumptions and structure inherent in a MADM. Multi-attribute decision models are similar in that they operate on a decision matrix, but they differ in the mathematical approach used. A recent review from the Canadian Air Force identified 30 different MADM models42. We chose the TOPSIS model because the decision rule, similarity to the ideal, can be clearly understood and has obvious real-life relevance.

An important model-related issue is that the outcome depends on the inputs. In the present context, a different selection of treatment attributes may have led to a different set of results. A familiarity of use attribute, for example, could result in a disadvantage for recently available products compared with products that have been marketed for many years. We decided to include only those product-related attributes for which robust data are available from controlled clinical trials10.

The consistency in the results obtained from headache sufferers, neurologists, and PCPs was a somewhat unexpected finding. Efficacy attributes emergedas significantly more important than tolerability or consistency of effect in all 3 surveys. For triptan-naive but not triptan-experienced migraine sufferers, tolerability was significantly more important than consistency. Triptan-na’ive migraineurs may value tolerability over consistency as they have not yet experienced the need to deal with inconsistent treatment responses. The refractory migraine sufferer who has undergone several medication trials may value better control of their headaches and be more willing to accept tolerability issues.

Physicians believed sustained pain free to be the most important efficacy attribute, while individuals with migraine believed the most important attribute was early relief from pain (pain free at 1 h). This makes sense: physicians may select sustained pain free as the most important efficacy attribute as their goal may be to reduce frequency of migraine episodes. For migraineurs, in contrast, the priority is more immediate – pain relief as soon as possible, with long-term pain relief becoming a later priority.

Although previous studies have assessed physicians’ and migraine sufferers’ prioritization of efficacy and tolerability factors in acute migraine6,21,43, differences in methods and in attributes make comparisons difficult. The consistency of responses among the 3 types of respondents in this study adds strength to the finding that efficacy attributes are most important for selecting triptans.

In this study a comparison was made between triptan selection by physicians and the triptan assigned by the model on the basis of the attribute weightings. The agreement was higher for the triptan- experienced than the triptan-naive patients, but even so, the comparison showed that, on average, products selected by the model were closer to the hypothetical ideal for each particular patient than the triptan selected by the physician. Computer-based decision- support systems are increasingly prevalent, ranging in complexity from patient recall systems and drug do\sing calculators to complex diagnostic tools linked to electronic medical records. In reviewing these, Trowbridge and Weingarten44 concluded that most decision- support systems provide significant benefits by improving the process of care, preventing medical errors and prompting physicians to provide appropriate preventive care measures. They also noted, however, that in terms of guiding the treatment of individual clinical disorders, decision-support systems remain a matter of study. In the present context, a recent study of neurologists’ learning needs 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 use45. This was reported at a time when there were fewer oral triptans from which to choose than at present. Thus, a model of the kind used in the TRIPSTAR project has considerable potential as a decision-support tool in the treatment of acute migraine, as well as allowing for the possibility of incorporating individual patients’ preferences, where appropriate, into the prescribing decision44.

As the meta-analysis used in the TRIPSTAR project was based on clinical trials in which headache was of moderate-to-severe intensity prior to treatment, it is fair to ask whether the TRIPSTAR findings apply to patients who are advised to treat their migraine attacks at an earlier stage. Intervention at the onset of pain with oral sumatriptan 50mg and almotriptan 12.5mg led to a significantly higher percentage of aborted attacks (i.e., pain free at 2 h) and to significantly more consistent pain relief over multiple attacks46. Early intervention also was associated with lower rates of recurrence, diminished need for rescue medication, and fewer adverse events47. The question of whether the relative rankings of the triptans, as reported here, might change, based on the timing of the intervention, should be addressed in future studies.

Neither personal characteristics nor headache features altered the study recommendations. Almotriptan 12.5mg, eletriptan 40 mg and 80 mg, and rizatriptan 10mg remained the triptans selected by the model as closest to the ideal for all subgroups of patients and headaches.

Conclusion

On the basis of current knowledge, including the present study, it cannot be predicted whether a given patient or a given migraine attack will respond to a particular triptan10,12. As a result, clinical judgment will continue to be an inevitable part of treatment. To facilitate treatment, physicians need to have information on, and access to, more than one triptan10, but most physicians will probably not want to become familiar with 7 oral triptans. The TRIPSTAR project provides guidance for the selection of triptans most likely to yield a successful therapeutic outcome.

Acknowledgment

This project was supported by a grant from the Pharmacia Corporation.

References

1. MacGregor EA, Brandes J, Eikermann A. Migraine prevalence and treatment patterns: the global Migraine and Zolmitriptan Evaluation survey. Headache 2003;43(1):19-26

2. Lipton RB, Stewart WF, Diamond S, Diamond ML, Reed M. Prevalence and burden of migraine in the United States: data from the American Migraine Study II. Headache 2001;41 (7):646-57

3. Steiner TJ, Scher AI, Stewart WF, Kolodner K, Liberman J, Lipton RB. The prevalence and disability burden of adult migraine in England and their relationships to age, gender and ethnicity. Cephalalgia 2003;23(7):519-27

4. Lipton RB, Diamond S, Reed M, Diamond ML, Stewart WF. Migraine diagnosis and treatment: results from the American Migraine Study II. Headache 2001;41(7):638-45

5. Gallagher RM, Kunkel R. Migraine medication attributes important for patient compliance: concerns about side effects may delay treatment. Headache 2003;43:36-44

6. Lipton RB, Stewart WF. Acute migraine therapy: do doctors understand what patients with migraine want from therapy? Headache 1999;39(Suppl 2):S20-S26

7. Deleu D, Hanssens Y, Worthing EA. Symptomatic and prophylactic treatment of migraine: a critical reappraisal. Clin Neuropharmacol 1998;21(5):267-79

8. Dahlof C. Integrating the triptans into clinical practice. Curr Opin Neurol 2002;15(3):317-22

9. Silberstein SD. 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 2000;55(6):754-62

10. Ferrari MD, Roon KI, Lipton RB, Goadsby PJ. Oral triptans (serotonin 5-HT^sub 1B/1D^ agonists) in acute migraine treatment: a meta-analysis of 53 trials. Lancet 2001;358:1668-75

11. Ferrari MD, Goadsby PJ, Roon KI, Lipton RB. Triptans (serotonin, 5-HT^sub 1B/1D^ agonists) in migraine: detailed results and methods of a meta-analysis of 53 trials. Cephalalgia 2002;22:633- 58

12. Goadsby PJ, Lipton RB, Ferrari MD. Migraine – current understanding and treatment. New Engl J Med 2002; 346:257-70

13. von Winterfeldt D, Edwards W Decision analysis and behavioral research. Cambridge: Cambridge University Press; 1987

14. Brodie MJ, Kwan P. The Star Systems: overview and use in determining antiepileptic drug choice. CNS Drugs 2001; 15(1):1-12

15. Azar F. Multiattribute decision making: use of three scoring methods to compare the performance of imaging techniques for breast cancer detection. Available at: http://www.cis.upenn. edu/~fredazar/ publications/pdf/fredazar_techreport_MS-CIS-00-10.PDF [accessed October 21, 2002]

16. Ferrari MD. Tripstar: a comprehensive patient-based approach to compare triptans. Headache 2002;42( Suppl 1):18-25

17. Schumacher GE. Multiattribute evaluation in formulary decision making as applied to calcium-channel blockers. Am J Hosp Pharm 1991;48(2):301-8

18. Eriksen S, Keller LR. A multiattribute-utility-function approach to weighing the risks and benefits of pharmaceutical agents. Med Decis Mak 1993;13(2):118-25

19. Dolan JG, Bordley DR. Isoniazid prophylaxis: the importance of individual values. Med Decis Making 1994;14(1):1-8

20. Stewart WF, Lipton RB, Dowson AJ, Sawyer J. Development and testing of the Migraine Disability Assessment (MIDAS) Questionnaire to assess headache-related disability. Neurology 2001;56(6 Suppl 1):S20-S28

21. Holmes WF, MacGregor EA, Sawyer JP, Lipton RB. Information about migraine disability influences physicians’ perceptions of illness severity and treatment needs. Headache 2001;41(4): 343-50

22. Lipton RB, Stewart WF, Stone AM, Lainez MJ, Sawyer JP. Stratified care vs step care strategies for migraine: the Disability in Strategies of Care (DISC) Study: a randomized trial. J Am Med Assoc 2000;284(20):2599-605

23. Lipton R, Liberman J, Cutrer F, et al. Treatment preferences and the selection of acute migraine medications: results from a population-based survey. J Headache Pain 2004; 5 (2): 123-30

24. Dolan JG, Isselhardt Jr BJ, Cappuccio JD. The analytic hierarchy process in medical decision making: a tutorial. Med Decis Making 1989;9(1):40-50

25. Mooney CZ, Duval RD. Bootstrapping: a nonparametric approach to statistical interference. Newbury Park (CA): Sage Publications; 1993

26. Anon. American Medical Association Physician Characteristics and Distribution in the US. Chicago (IL): American Medical Association; 2001

27. Health Resources and Services Administration – Bureau of Health Professions. State health workforce profiles. Available at: http://bhpr.hrsa.gov/healthworkforce/reports/profiles/default. htm. [accessed January 14, 2003]

28. Dodick D, Lipton RB, Ferrari M, et al. Prioritizing treatment attributes and their impact on selecting an oral triptan: Results from the TRIPSTAR project. Curr Pain Headache Rep 2004;8:435-42

29. Cutrer FM, Goadsby PJ, Ferrari MD, Lipton RB, Dodick DW, McCrory D. Priorities for triptan treatment attributes and their implications for selecting an oral triptan for acute migraine: a study of US primary care physicians (the TRIPSTAR project). Clin Ther 2004;26:1533-45

30. Hwang C-L, Yoon KP. Multiattribute decision making: methods and applications. Berlin: Springer; 1981. p. 128-40

31. Yoon KP, Hwang C-L. Multiple attribute decision making: an introduction. Thousand Oaks: Sage; 1995. p. 38-40

32. Zeleny M. Multiple criteria decision making. New York: McGraw Hill; 1982

33. Deng H, Yeh C-H, Willis PJ. Inter-company comparison using modified TOPSIS with objective weights. Comput Operations Res 2000;27:963-73

34. Roon KI, Lipton R, Goadsby PJ, Ferrari M. Placebo in triptan trials: efficacy, tolerability and consistency. Cephalalgia 2001;21: 405-32

35. Baltussen RM, Hutubessy RC, Evans DB, Murray CJ. Uncertainty in cost-effectiveness analysis. Probabilistic uncertainty analysis and stochastic league tables. Int J Technol Assess Health Care 2002;18(1):112-9

36. Washington State Department of Health. Guidelines for using confidence intervals for public health assessment. Website: http:// www.doh.wa.gov/Data/Guidelines/ConfIntguide.htm [accessed 16th December 2004]

37. Dodick DW, Lipton RB, Goadsby PJ, et al. Prioritizing triptan treatment attributes: a pilot study. Neurology 2002;58(Suppl 3): A129

38. McCrory D, Williams P. Selecting the preferred triptans. J Headache Pain 2004;5:247-50

39. Ferrari MD, Goadsby PJ, Lipton RB. The use of multiattribute decision models in evaluating triptan treatment options in migraine. J Neurol [in press]

40. Lipton RB, Bigal ME, Rush SR, et al. Migraine practice patterns among neurologists. Neurology 2004;62:1926-31

41. Moss L, Goldstein H. The recall method in social surveys. London: University of London Institute of Education; 1979

42. Abi-Zeid I, Belanger M, Guitouni A, Martel J-M, Jabeur K. A multicriteria method for evaluating courses of action in Canadian airspace violation situations. Paper presented at: Fourth International Symposium on command and Control Research and Technology; June, 1998; Nasby Slott, Sweden

43. Dayno JM, Lipton RB. Acute migraine treatment: factors affectingtherapeutic decision making. Cephalalgia 2001;21:320-1

44. Trowbridge R, Weingarten S. Clinical decision support systems, Chapter 53. In: Evidence report/technology assessment no. 43, making health care safer: a critical analysis of patient safety practices. Agency for Healthcare Research and Quality, Washington. Available at: http://www.premierinc.com/all/safety/ resources/ patient_safety/downloads/23_AHRQ_evidence_ report_43.pdf [accessed December 5, 2003]

45. Purdy RA. Evidence-based migraine therapy: learning needs and knowledge assessment. Cephalalgia 2000;20(Suppl 2):5-9

46. Pascual J. Clinical benefits of early triptan therapy for migraine. Headache 2002;42(Suppl 1):S10-S17

47. Mathew NT. Early intervention with almotriptan improves sustained pain-free responses in acute migraine. Headache 2003;43:1075-9

CrossRef links are available in the online published version of this paper:

http://www.cmrojournal.com

Paper CMRO-2774_3, Accepted for publication: 27 January 2005

Published Online: 28 February 2005

doi: 10.1185/030079905X36387

R. B. Lipton(a), F. M. Cutrer(b), P. J. Goadsby(c), M. D. Ferrari(d), D. W. Dodick(e), D. McCrory(f), J. N. Liberman(g) and P. Williams(h)

a Departments of Neurology, Epidemiology and Population Health, Albert Einstein College of Medicine, Bronx, NY, USA

b Mayo Clinic, Rochester, MN, USA

c National Hospital for Neurology and Neurosurgery, London, UK

c Leiden University Medical Centre, Leiden, The Netherlands

e Mayo Clinic, Scottsdale, AZ, USA

f Duke University Medical Center, Durham, NC, USA

g IMP, an AdvancePCS Company, Hunt Valley, MD, USA

h PAREXEL International, Uxbridge, Middlesex, UK

Address for correspondence: Professor Richard B. Lipton, Professor and Vice Chair of Neurology, Professor of Epidemiology and Population Health, Albert Einstein College of Medicine, 1300 Morris Park Avenue (Russo-3rd Floor), Bronx, NY 10461, USA. Tel.: +1 718 430 3886; Fax: +1 718 430 3857; email: rlipton@aecom.yu.edu

Copyright Librapharm Mar 2005