Research Submission

Stability, Responsiveness, and Reproducibility of a VisualAnalog Scale for Treatment Satisfaction in Migrainehead_2157 1005..1018

Christian Lucas, MD, PhD; Sophie Romatet, MD; Claude Mekiès, MD; Bashar Allaf, MD;Michel Lantéri-Minet, MD, PhD

Objectives.—To evaluate the stability, responsiveness, and reproducibility of a simple visual analog scale (VAS).Background.—In order to help physicians in the management of migraine in everyday general practice and assess whether

the treatments that they are currently prescribing are actually effective, a VAS of treatment satisfaction with acute migrainetreatments has been developed.

Methods.—The study used an open-label, multicenter, prospective design. Adult patients fulfilling diagnostic criteria formigraine and who consulted a participating hospital or community neurology clinic were eligible. At inclusion, patients ratedtheir satisfaction with their current treatment on the VAS. Those scoring 7-10 (satisfied) on the VAS were allocated to theVASCO cohort, and those scoring 0-4 (dissatisfied) were switched to almotriptan and allocated to the ALMO cohort. Patientsscoring between 4 and 7 were assigned to 1 or other cohort at the physician’s discretion. The VAS was re-administered at homethe next day and also after the treatment of 3 further headaches, both at home and at a follow-up visit.

Results.—Ninety-eight patients in the VASCO cohort and 102 in the ALMO cohort were analyzed. Stability was evaluatedin the VASCO cohort: 55/98 patients initially satisfied with treatment remained so at study end, whereas 7/98 became dissatisfied.Responsiveness of the VAS to a change in treatment was evaluated in the ALMO cohort: 64/102 patients moved to a highertreatment satisfaction category, whereas 6/102 moved to a lower one. Reproducibility of the VAS was determined in 4 settings(both at the inclusion visit and at study closure in both cohorts). In each setting,VAS scores were compared between consultationand at-home ratings. In 3 of the 4 settings (both measures in the ALMO cohort and at study closure in the VASCO cohort), goodagreement was observed between the 2 ratings (k = 0.62-0.69). At inclusion in the VASCO cohort, agreement was only fair(k = 0.33).

From the Neurology Clinic, Roger Salengro Hospital, Lille, France (C. Lucas); Neurology Department, Poissy-St Germain Hospital,Saint Germain-en-Laye, France (S. Romatet); Clinique des Cèdres,Toulouse, France (C. Mekiès);Almirall SAS, Paris, France (at thetime of the study) (B.Allaf);Department of Pain Evaluation andTreatment,Clinical Neurosciences School,Nice University Hospital,Nice, France and INSERM/UdA, U1107, Neuro-DOL, University of Auvergne, Clermont-Ferrand, France (M. Lantéri-Minet).

Address all correspondence to C. Lucas, Clinique Neurologique, Hôpital Roger Salengro, Rue du 8 mai 1945, 59037 Lille Cedex,France, email: clucas@chru-lille.fr

Accepted for publication January 28, 2012.

Conflict of Interest:C.L. received honoraria for speaking from Allergan, Almirall SAS, AstraZeneca, Boehringer Ingelheim, Bouchara recordati,

GlaxoSmithKline, Menarini, Merck, Pfizer, Sanofi, and Servier.S.R. received honoraria for speaking from Almirall SAS, AstraZeneca, Merck, Boehringer Ingelheim, and Eisai.C.M. received honoraria for speaking from Almirall SAS, AstraZeneca, Bayer, Biogen Idec, Boehringer Ingelheim, Eisai,

GlaxoSmithKline, Janssen, Lundbeck SAS, Merck Serono, Novartis, Pfizer, Sanofi, Teva, and UCB.B.A. was an employee of the Almirall SAS at the time of the study, a company that manufactures almotriptan, a migraine

treatment.M.L.M. received honoraria for speaking from Allergan, Almirall SAS, AstraZeneca, GlaxoSmithKline, Grünenthal, Lilly,

Johnson & Johnson, Medtronic, Menarini, Merck, Laboratoires Pierre Fabre, Pfizer, Sanofi, UCB, and Zambon.

Financial support: This study was supported by Almirall SAS, Paris, France.

ISSN 0017-8748doi: 10.1111/j.1526-4610.2012.02157.xPublished by Wiley Periodicals, Inc.

Headache© 2012 American Headache Society

1005

Conclusions.—The VAS scale described here is a responsive and easy-to-use tool for evaluating treatment satisfaction andfor monitoring changes to treatment if these are required.

Key words: migraine, treatment satisfaction, visual analog scale, almotriptan, validation

(Headache 2012;52:1005-1018)

Migraine is a chronic neurological disordermanifesting as episodic headaches characterized bysevere unilateral pulsatile pain, often accompanied bynausea, vomiting, and photosensitivity and phonosen-sitivity, which are frequently incapacitating. Migraineis associated with a marked reduction in quality oflife,1,2 not only during the headaches themselves, butalso between headaches, and carries a significant eco-nomic burden.3-6 The prevalence of migraine in theadult French general population has been estimatedat between 17%7 and 21%,8 corresponding to around8 million individuals, young women being the demo-graphic group most frequently affected.

Migraine is frequently underdiagnosed andundertreated. The American Migraine Study,9

performed in 1999, found that 52% of the US popu-lation fulfilling diagnostic criteria for migraine do notreceive a diagnosis of migraine. Two general popula-tion studies in France (GRIM2000,5 performedin 1999, and FRAMIG 3,8 performed in 2003) havereported similar low rates of diagnosis. Indeed, inFRAMIG 2000, only 1 migraineur in 5 was consult-ing a physician for their headaches,10 and in theGRIM2005 study, only 3 migraineurs in 10 were inactive consultation.11 In the Migraine And Zolmitrip-tan Evaluation (MAZE) study,12 conducted in 5 coun-tries in 1999, the proportion of migraineurs consultinga physician ranged from 41% in the USA to 63% inFrance. The most frequent reasons given for non-consultation were that over-the-counter medicationswere sufficient and that headaches did not warrantseeing a doctor. Even when patients do consult aphysician, most do not receive an appropriatemigraine treatment,8,12 use over-the-counter medica-tion,5,13 and are unsatisfied with their treatment.13 Thissituation is unsatisfactory because specific migrainetreatments exist that can be prescribed effectively andsafely to the vast majority of patients with migraine.

In order to help physicians in the management ofmigraine in everyday general practice, the French

national health observatory (Haute Autorité deSanté; HAS) and the French Society for the Study ofMigraine Headache set up a working group in 2002 todevelop guidelines for the diagnosis and managementof migraine.14 An integral part of these guidelines wasa 4-item structured questionnaire for evaluating theeffectiveness of acute migraine treatments. These 4items are as follows:

1. Do you obtain significant headache relief twohours after taking medication?

2. Do you require only a single intake of medicationfor headache relief?

3. Is your treatment well-tolerated?4. Does treatment allow a rapid return to normal

social, family or professional activities?

If a patient replies “no” to any one of thesequestions, it is recommended to switch treatment to anon-steroidal anti-inflammatory drug (NSAID) or atriptan, which can be used in a stepped treatmentapproach, starting with the NSAID and escalatingto the triptan in case of inadequate response.For patients replying “yes” to all questions, pursuit ofcurrent treatment is recommended. Since the devel-opment of this questionnaire, a number of pharma-coepidemiological surveys and clinical trials haveinvestigated the effectiveness of migraine treatmentsin routine care and revealed that a large proportion ofpatients are receiving treatments that are not actuallyeffective. For example, in the FRAMIG 3 study,15 halfof the study population reported use of acute head-ache treatments which are not recommended in theHAS guidelines. In the GRIM2005 study, less thanhalf of a general population sample of migraineurs(45.8%) fulfilled the HAS criteria for treatmentresponse.16 In this study, around one-third of nonre-sponders were in active consultation for their head-aches, which suggested to the authors that physiciansdo not use the HAS criteria systematically to monitortreatment responses.

1006 June 2012

Given this unsatisfactory situation, it is importantthat physicians treating patients with migraine taketime to assess whether the treatments that theyare currently prescribing are actually effective. Tothis end, we have recently developed a simple visualanalog scale (VAS) of treatment satisfaction whosescore is associated closely with the HAS criteria fortreatment effectiveness.17 Using receiver operatingcharacteristic curves, scores on this scale were dividedinto 3 categories predictive of treatment response.Below a lower threshold of 4, it was shown thattreatment should be considered inadequate and bemodified, conversely, above an upper threshold of 7,it should be considered adequate and continued, andin-between (4-7), no useful conclusion can be drawn.These thresholds demonstrated high specificity andsensitivity for determining treatment responsivenessaccording to the HAS criteria (97% and 84%, respec-tively, for the lower threshold, and 82% and 83%,respectively, for the upper thresholds). In light ofthese findings, this VAS score appears to offer a prac-tical tool for rapid appraisal of acute treatmentresponse.17 Further work is however required to vali-date this VAS for use in everyday practice, in particu-lar with respect to its stability over time and itsresponsiveness to changes in treatment.

The primary objectives of the present study wereto assess the stability of the VAS in patients who werewell satisfied with their treatment over a 3-monthinterval and to assess its responsiveness to change inpatients who were not satisfied and in whom a treat-ment switch was implemented. Secondary objectiveswere to evaluate the reproducibility of the scale overa 24-hour period and to compare treatment satisfac-tion with the HAS criteria.

METHODSThe study was an open-label, multicenter,

prospective study conducted in 200 hospital or com-munity neurology clinics in France between Novem-ber 2006 and January 2008. Participating centers wereidentified from a list of active migraine treatmentcenters in France held by the study sponsor.The studyincluded 2 independent cohorts of patients, theVASCO cohort, corresponding to patients who weresatisfied with their migraine treatment, used for the

evaluation of stability of the VAS, and the ALMOcohort, made up of patients dissatisfied with theircurrent treatment, who were switched to almotriptan,a triptan with proven efficacy and a good tolerabilityprofile, and used for the evaluation of the responsive-ness of the VAS.

Study Investigators.—Potential study investigatorswere identified from a list of neurologists regularlyvisited by representatives of the study sponsor(Almirall SAS). This list contained the names of 1260of all 2237 practicing neurologists in France (56.5%)and included both community- and hospital-basedneurologists. Each neurologist on the list was con-tacted in random order by telephone by a contractresearch organization (DCi, Bougival, France) towhom implementation of the study had been del-egated by the study sponsor. The objective was torecruit 200 centers that offer migraine consultations,50% of them in community practice and 50% in hos-pitals.The first 200 such centers that agreed to partici-pate were recruited into the study. Each participatingcenter included patients into both the VASCO andALMO cohorts.

Patient Selection and Assignment to StudyCohorts.—Adult patients (aged 18-65 years) fulfillingdiagnostic criteria for migraine with or without auraas defined in the revised International HeadacheSociety classification (diagnostic categories 1.1 and1.2)18 and who consulted a study center during therecruitment period were eligible for the study. Thesepatients were asked to rate their satisfaction withtreatment on a VAS. This consisted of a featurelessblack horizontal line of 10 cm between 2 extremeslabeled “very satisfied” and “very unsatisfied.” Thepatient was asked to place a tick on the line to indi-cate their satisfaction with their current migrainetreatment.

Patients scoring between 7 and 10 (ie, adequatetreatment satisfaction) were eligible for the VASCOcohort, and their current acute headache treatmentwas retained for the duration of the study. Patientsscoring between 0 and 4 (ie, inadequate treatmentsatisfaction) were eligible for the ALMO cohort,and their acute headache treatment was switched toalmotriptan 12.5 mg. For patients scoring over 4 andunder 7 (ie, undetermined treatment satisfaction), the

Headache 1007

physician was recommended to evaluate the HAStreatment responsiveness criteria and, on this basis,had the option either to maintain the current treat-ment, in which case the patient was eligible for theVASCO cohort, or to switch to almotriptan, in whichcase the patient became eligible for theALMO cohort.

Additional inclusion criteria for the study werethe occurrence of at least 3 migraine headaches in theprevious 3 months, the capacity of the patient to rec-ognize a migraine headache and, in particular, to dis-tinguish this from a tension-type headache, the use ofan oral acute headache treatment, and the intellectualand physical capacity to complete a study diary.Exclusion criteria were a headache frequency ofgreater than 6 migraines per month or of more than15 days per month, onset of migraine disorder afterthe age of 50, basilar or familial hemiplegic migraine,use of analgesics on more than 15 days per month orof opiates or triptans on more than 10 days permonth, and change or initiation of prophylactictreatment in the past 3 months. For the ALMOcohort, patients with contraindications to almotriptanor receiving an ergot alkaloid-based prophylactictreatment were also excluded. Individuals meetingthese criteria were provided with written informationon the study and were required to sign an informedconsent form. Each participating center was expectedto include 2 patients fulfilling the inclusion criteria foreach of the 2 study cohorts and providing informedconsent into the respective cohort.

Study Procedures.—The study required 2 studyvisits to the center, one at the time of inclusionand a second follow-up visit 3 months later. Be-tween these 2 study visits, included patients com-pleted a study headache diary at home. At the inclu-sion visit, the physician verified the inclusioncriteria, obtained signed consent, evaluated the HAScriteria for treatment responsiveness, the baselinesatisfaction VAS score, and ensured that the patienthad access to sufficient acute headache medication(previous treatment for the VASCO cohort andalmotriptan for the ALMO cohort) for the durationof the study.

In addition, each patient was provided withstudy diary in which they were expected to recordthe occurrence of the next 3 consecutive migraine

headaches they experienced. In order to determinethe HAS criteria for treatment responsiveness, head-ache severity, duration and symptoms, features of theresponse to treatment, and any side effects of treat-ment experienced were recorded.The study diary alsoincluded 2 treatment satisfaction VAS. The first ofthese was to be rated 24 hours after the inclusion visitand the second 24 hours after the last headacherecorded in the study diary. It was specified that theVAS should not be rated if the patient was experienc-ing a migraine headache at the time.

Within 2 weeks of the last headache recorded inthe diary, or 3 months after the inclusion visit, if thepatient had not yet experienced 3 headaches, thepatient was asked to return to the study center for afollow-up visit, at which a fourth treatment satisfac-tion VAS was rated. Information on the HAS treat-ment responsiveness criteria was also obtained by thephysician asking the patient, and this was noted in thecase report form.The study diary was collected by thephysician and returned, together with the case reportform, to the data management center.

Outcome Measures.—The primary outcomemeasure was the VAS treatment satisfaction score ateach of the 4 evaluation points (ie, during the inclu-sion visit, at home 24 hours after the inclusion visit, athome 24 hours after the last headache, and at thestudy closure visit). Stability of the VAS was evalu-ated by comparing VAS score at the inclusionvisit with VAS score at the study closure visit in theVASCO cohort. Responsiveness was evaluated bycomparing VAS score at the inclusion visit with VASscore at the study closure visit in the ALMO cohort.Reproducibility was evaluated by comparing VASscores between the physician visit and the at-homeratings in each cohort, both at inclusion and at studyclosure.

Statistical Analysis.—The study population ofinterest were the analyzable population of theVASCO and ALMO cohorts. These are defined as allincluded patients without a major protocol deviationwho attended the study closure visit and completed aVAS, and who provided data on at least 2 headachesin their study diaries. The size of the study sampleswas determined by a priori power calculations inorder to determine the parameters of interest with a

1008 June 2012

precision of 5%, based on previous determinations ofthe distribution of the VAS score.17 This provided anestimate of 381 patients for the VASCO cohort and of359 patients for the ALMO cohort. Assuming a non-completion rate of 10%, it was decided to recruit 400patients into each cohort.

Demographic and clinical variables were ana-lyzed descriptively as mean � standard deviationand medians [range] for quantitative variables and asnumbers and frequencies (%) for categorical vari-ables. VAS scores in each cohort were stratified into 3groups on the basis of their VAS scores (0-4, >4-<7,and 7-10).

The stability of the VAS score in the VASCOcohort visits was evaluated by group (0-4, >4-<7, and7-10) using weighted kappa coefficients and their95% confidence interval (CI). Given the ordinalnature of the scale, a weighted kappa was consideredmore appropriate than Cohen’s unweighted kappa.The weighted kappa coefficient ranges from 0 to 1,with a k < 0.20 being considered to signify poor agree-ment, 0.20 to 0.40 fair agreement, 0.40 to 0.60 moder-ate agreement, 0.60 to 0.80 good agreement, and 0.80to 1.00 very good agreement.19

The change in VAS score in the ALMO cohortwas assessed by group using the Wilcoxon signed ranktest and weighted kappa coefficients. In addition, ananalysis of repeated categorical measurements wasperformed, with the VAS score at baseline as covari-ate. VAS scores were also compared with the numberof HAS treatment responsiveness criteria fulfilledusing the Kruskal–Wallis test.

The reproducibility of VAS scores between visitswas using a weighted kappa coefficient for compari-son by class, as described earlier for the stability data,and by comparison of individual patient scores. Forthe latter, the test–retest correlation of the rankedVAS scores was evaluated with the Spearmanrank-order correlation coefficient (rs) and associatednondirectional probability values. Conventionally,a coefficient >0.60 is considered to indicate a strongcorrelation and a coefficient >0.80 a very strongcorrelation.

The results are reported with 2-sided P values,with a probability level of 5% being consideredsignificant throughout. All data were analyzed

centrally using SAS software (SAS Institute, Cary,NC, USA).

Ethics.—The study was conducted according to theDeclaration of Helsinki (Hong Kong Amendment),Good Clinical Practices (European Guidelines), andFrench regulatory requirements (Loi Huriet).Writteninformed consent was obtained from each patient.Patients were free to withdraw from the study at anytime for any reason, without effect on their medicalcare. No patient was offered financial inducementto participate in the study. The protocol was submit-ted to and approved by the Ethics Committee of SaintGermain-en-Laye. Procedures for data collection andmanagement was approved by the Conseil Nationald’Informatique et Liberté.

Role of the Funding Source.—The study wasinitiated by Almirall SAS (the study sponsor) andcoordinated by its medical advisor (B.A.). Thestudy sponsor designated a scientific committeeof migraine specialists (C.L., S.R., C.M., M.L.M.) toadvise on the design, implementation, and inter-pretation of the study. The committee assisted indeveloping the study protocol and the statisticalanalysis plan, advised on the analysis and exploita-tion of the study results, and contributed to thewriting of the present article. The committee had fullaccess to all data from the study and could requestfurther data analyses during meetings of the scien-tific committee. The committee members receivedhonoraria from the sponsor in return for their par-ticipation. Operational management of the study,data collection, and management were delegatedto DCi and statistical analysis to Altizem (Nanterre,France). Supplementary statistical analysis for thespecific purposes of this article was commissionedfrom STAT-Process (Port-Mort, France). The inves-tigators recruiting patients into the study receivedfees for their participation in the study on a perpatient basis. The study sponsor funded editorialsupport from a medical writing agency (Foxymed,Paris, France) for the preparation of the presentarticle and contributed, together with the scientificcommittee, to the revision of the different drafts ofthe manuscript. The corresponding author had finalresponsibility for the decision to submit the manu-script for publication.

Headache 1009

RESULTSPatient Disposition.—A total of 368 patients

were screened for the study, of whom approximatelyone-third fell into each VAS score category (Fig. 1).The VASCO cohort consisted of 182 patients and theALMO cohort of 186 patients. Twenty-one patients(5.7%) were incorrectly assigned (5 patients with noVAS rating reported at inclusion to the VASCOcohort, 10 patients with a score of 0-4 to the VASCOcohort, and 6 patients with a score of 7-10 to theALMO cohort). These patients were consideredto be protocol deviations. Thirty patients (16.5%) inthe VASCO cohort and 31 (16.7%) in the ALMOcohort reported less than 3 headaches in the 3-monthstudy period and were excluded from the analysis.In addition, 16 patients (8.8%) in the VASCOcohort and 29 (15.6%) in the ALMO cohort did not

complete the study, as they either did not attendthe study closure visit or failed to provide a VASrating. At least 1 other protocol violation was identi-fied in 64 patients in the VASCO cohort (35.2%) andin 36 patients in the ALMO cohort (19.4%). Theseprotocol deviations related principally to completionof the VAS or study closure visit outside the autho-rized time window (44 patients in the VASCO cohortand 29 in the ALMO cohort), or, in the VASCOcohort, to the use of multiple acute headache treat-ments (21 patients). The analyzable population thusconsisted of 98 patients (53.8%) for the VASCOcohort and 102 patients (54.8%) for the ALMOcohort. The mean time elapsed between the inclusionvisit and the study closure visit was 62.7 � 31.7 daysin the ALMO cohort and 59.3 � 29.7 days in theVASCO cohort.

Fig 1.—Patient disposition. †For 5 patients who were screened, the visual analog scale (VAS) score at inclusion was not entered intothe case report form. These patients, who were all assigned to the VASCO cohort, were subsequently excluded from the analyzablepopulation as protocol deviations. ‡Six patients with a VAS score at inclusion of 7-10 and 10 patients with a score of 0-4 wereincorrectly assigned to the wrong cohort. These patients were subsequently excluded from the analyzable population as protocoldeviations. §The subgroups of non-analyzable patients are not mutually exclusive.

1010 June 2012

Patient Characteristics.—The median age of thesubjects enrolled was 39 years, and 84.6% werefemale (Table 1). The disease duration was relativelylong (mean: 17 years), and patients reported a medianof 7 headaches in the previous 3 months.The majority

presented with migraine without aura and aroundhalf of the women reported experiencing migrainesduring their periods. In terms of demographicsand clinical features, the VASCO and ALMO cohortswere very similar, although the VASCO patients

Table 1.—Patient Characteristics

VASCO ALMO TotalN = 182 N = 186 N = 368

AgeMean � SD 40.1 � 11.4 37.0 � 12.4 38.5 � 12.0Median [interquartile range] 40.5 [33-48.5] 38 [29-46] 39 [30-47]

GenderMale 30 (17.1%) 24 (13.6%) 54 (15.4%)Female 145 (82.9%) 152 (86.4%) 297 (84.6%)

Time since diagnosisMean � SD 18.5 � 11.8 15.5 � 11.3 17.0 � 11.7

Headaches in last 3 monthsMedian [interquartile range] 7 [4-10] 6 [4-10] 7 [4-10]

Migraine typeOnly with aura 9 (5.0%) 11 (5.9%) 20 (5.5%)Only without aura 133 (73.9%) 136 (73.1%) 269 (73.5%)Mixed 38 (21.1%) 39 (21.0%) 77 (21.0%)Menstrual migraine† 74 (51.4%) 92 (61.0%) 166 (56.2%)

Acute headache treatment‡Paracetamol 12 (6.7%) 44 (24.0%) 56 (15.4%)Aspirin 7 (3.9%) 19 (10.4%) 26 (7.2%)Opioid analgesics 7 (3.9%) 17 (9.3%) 24 (6.6%)NSAIDs 34 (18.9%) 49 (26.8%) 83 (22.9%)Ergot alkaloids 1 (0.6%) 1 (0.5%) 2 (0.6%)Triptans 128 (71.1%) 74 (40.4%) 202 (55.6%)Others 10 (5.6%) 10 (5.5%) 20 (5.5%)

Prophylactic treatment‡,§Any prophylaxis 93 (52.2%) 46 (25.0%) 139 (38.4%)b-blockers 26 (28.0%) 18 (39.1%) 44 (31.7%)Antiserotonergic drugs 23 (24.7%) 8 (17.4%) 31 (22.3%)Antidepressants¶ 11 (11.8%) 7 (15.2%) 18 (12.9%)Alpha-adrenergic drugs†† 9 (9.7%) 5 (10.9%) 14 (10.1%)Anti-epileptic drugs 27 (29.0%) 13 (28.3%) 40 (28.8%)Others 11 (11.8%) 1 (2.2%) 12 (8.6%)

VAS score at inclusionMean � SD 7.45 � 1.90 3.53 � 2.15 5.44 � 2.82Median [interquartile range] 8.00 [6.90-8.50] 3.25 [2.00-5.00] 6.00 [3.00-8.00]Class

0-4 10 [5.6%] 123 (66.1%) 133 (36.6%)4-7 48 [27.1%] 57 (30.6%) 105 (28.9%)7-10 119 (67.2%) 6 (3.2%) 125 (34.4%)

†Patients replying that they had migraines only or sometimes during their periods. The percentages are calculated with respect tothe female population only.‡Patients could take more than 1 treatment, so categories are not exclusive.§Percentages are calculated with respect to the number of patients using prophylactic treatments.¶Amitriptyline was the only member of this class cited.††Indormaine was the only member of this class cited.NSAIDs = non-steroidal anti-inflammatory drugs; VAS = visual analog scale; SD = standard deviation.

Headache 1011

were slightly older and had slightly longer diseaseduration.

All but 5 patients reported using an acuteheadache treatment (Table 1). This was most fre-quently a triptan, either alone or in association,in a little over half of the participants, followedby NSAIDs, and paracetamol. Around one-thirdof patients reported using a prophylactic treat-ment, principally b-blockers and anti-epileptic drugs.However, marked differences were observed betweenthe 2 cohorts in terms of migraine treatments used.Notably, the use of triptans and of prophylactic treat-ments was around 2 times higher in the VASCOcohort, who were satisfied with their treatment, thanin the ALMO cohort, who were not. Conversely, useof non-specific headache relief medication (paraceta-mol, aspirin, and opioid analgesics) was higher in theALMO cohort.

Treatment Satisfaction Ratings at Inclusion.—Mean VAS treatment scores at the inclusion visit arepresented in Table 1. These were 7.45 in the VASCOcohort and 3.53 in the ALMO cohort. Around 1 in 3patients in both cohorts were derived from the unde-termined treatment satisfaction group (scores 4-7).These were assigned by the physician to 1 or other ofthe cohorts on the basis of the HAS criteria. Of thesecriteria, the most discriminating in the choice ofcohort assignment by the physician was whether thetreatment provided significant headache relief after 2hours, and to a lesser extent, whether it permitteda rapid return to normal activity (Table 2). Of thepatients assigned to the VASCO cohort, nearly halffulfilled all 4 HAS criteria, compared with only 2patients assigned to the ALMO cohort.

Stability of the Treatment Satisfaction VAS.—Stability of the treatment satisfaction VAS over a3-month period has been evaluated in the analyzablepopulation of the VASCO cohort. In this group, meanVAS scores were 7.7 � 1.3 at the inclusion visit to7.4 � 1.8 at the study closure visit. On an individualpatient basis, the difference in VAS score betweeninclusion and study end was not statistically significant(P = .111; Wilcoxon signed rank test). Although thecategorical distribution of VAS scores at the 2 time-points was generally similar (Table 3), 37 patients(37.8%) changed score category between the initialand final visit, such that overall score consistencybetween visits was poor (weighted kappa coefficient:0.095). The majority of this instability consisted ofdrift between the �7 “satisfied” zone and the >4 to7 “ambiguous” zone. Seven patients (7.1%) in thiscohort scored �4 (“unsatisifed”) at the second visit.

Responsiveness of the Treatment SatisfactionVAS.—Responsiveness of the treatment satisfactionVAS to a change in treatment was evaluated in theanalyzable population of the ALMO cohort. In thisgroup, mean VAS scores increased from 3.5 � 1.9 atthe inclusion visit to 6.1 � 2.5 at the study closurevisit following a change in treatment. Examination ofthe distribution of VAS scores at the 2 time pointsrevealed that the proportion of patients rating treat-ment satisfaction as 0-4 (unsatisfied) declined from65.7% to 24.5%, whereas 48.0% of patients werenow satisfied with treatment (Table 3). Sixty-fourpatients transited from a lower to a higher treatmentsatisfaction category, whereas 6 transited from ahigher to a lower one. The difference in VAS scoresbetween the 2 study visits was significant (P < .01;

Table 2.—Haute Autorité de Santé Treatment Responsiveness Criteria for Patients With an Undetermined TreatmentSatisfaction Visual Analog Scale Score (4-7)

VASCO ALMO TotalN = 48 N = 57 N = 105

Significant headache relief after 2 hours 33 (80.5%) 16 (28.1%) 49 (50.0%)Single intake of medication sufficient 27 (65.9%) 26 (45.6%) 53 (54.1%)Treatment well tolerated 39 (95.1%) 45 (78.9%) 84 (85.7%)Rapid return to normal activity 32 (78.0%) 19 (33.3%) 51 (52.0%)All 4 criteria fulfilled 22 (45.8%) 3 (5.3%) 25 (23.8%)

1012 June 2012

Wilcoxon signed rank test) and between-visit scoreconsistency poor (weighted kappa = 0.076; 95% CI[-0.006-0.159]). This difference in score remainedstatistically significant after adjustment for VASscore at baseline.

Reproducibility of the Treatment SatisfactionVAS.—Reproducibility of the treatment satisfactionVAS was determined in 4 settings, namely at the timeof the inclusion visit and at the time of study closurein both the VASCO and the ALMO cohorts. In eachsetting, the mean scores were compared between arating performed during the consultation and oneperformed at home at an interval of 24 hours usinga weighted kappa coefficient for comparison byclass and a Spearman correlation coefficient forcomparison of individual patient scores. Agreementbetween the 2 ratings was good in the ALMO cohortat both the inclusion visit (k: 0.62 [95% CI: 0.46-0.78])and the follow-up visit (k: 0.79 [0.69-0.90]), and in theVASCO cohort at the follow-up visit (k: 0.63 [0.47-0.780]). However, at the inclusion visit in the VASCOcohort, agreement was only fair (k: 0.33 [95% CI:

0.14-0.527]). Spearman correlation coefficients at theinclusion visit were 0.57 in the VASCO cohort and0.65 in the ALMO cohort. The corresponding valuesfor the follow-up visit were 0.86 for the VASCOcohort and 0.88 for the ALMO cohort. All these test–retest correlations were highly statistically significant(P < .0001). Agreement between ratings at home andduring the consultation is presented in Figure 2 on anindividual patient basis.

VAS Scores and Treatment Response Criteria atStudy End.—The relationship between the VAStreatment satisfaction score at study closure and thenumber of HAS treatment response criteria fulfilledat the end of the study was determined. In the “unde-termined” treatment satisfaction category (VAS score4-7), a clear positive relationship between the meanVAS score and the number of HAS criteria fulfilledwas observed (Table 4). However, in the other 2 cat-egories (VAS scores 0-4 and 7-10), there was littlevariation in VAS score according to the number ofHAS criteria fulfilled, although the number of HAScriteria fulfilled clearly segregated between the 2 cat-

Table 3.—Stability (VASCO Cohort) and Responsiveness (ALMO Cohort) of the Treatment Satisfaction Visual Analog Scale(VAS)

VASCO cohort

VAS score at study closure

VAS Score at Inclusion

�4 (None) >4-7 (N = 33) �7 (N = 65) Total (N = 98)

�4 (N = 7) 0 3 4 7>4-7 (N = 22) 0 6 15 21�7 (N = 69) 0 15 55 70Total (N = 98) 0 24 74 98

ALMO cohort

VAS score at study closure

VAS Score at Inclusion

�4 (N = 67) >4-7 (N = 32) �7 (N = 3) Total (N = 102)

�4 (N = 25) 19 6 None 25>4-7 (N = 28) 18 10 None 28�7 (N = 49) 30 16 3 49Total (N = 102) 67 32 3 102

Headache 1013

egories. Of the patients scoring between 7 and 10 onthe VAS, 87/115 fulfilled all 4 categories and 26/115fulfilled 3 out of 4 categories. In contrast, for thepatients scoring between 0 and 4, only 2/32 fulfilled 3of 4 HAS criteria and 21/32 only fulfilled 1 criterion.

In the ALMO cohort, only 1 patient at the studyclosure did not fulfill any of the HAS criteria, while 54patients (52.9%) fulfilled all 4 criteria. This propor-tion can be compared with the 68.0% of patients in

the VASCO cohort who fulfilled all 4 HAS criteria atstudy end (Fig. 3). These percentages are not signifi-cantly different (P = .053; c2 test).

DISCUSSIONThe objective of this study was to evaluate the

stability and responsiveness of a VAS for measuringtreatment satisfaction in patients with migraine. Con-cerning responsiveness, a significant increase in VAS

Fig 2.—Correlations between visual analog scale (VAS) rating scores at home and during the consultation. A and B: ALMO cohort;C and D: VASCO cohort. White symbols: rating scores at inclusion; black symbols: rating scores at study closure. rs = Spearmanrank-order correlation coefficient.

1014 June 2012

satisfaction was observed following switching initiallyunsatisfied patients in the ALMO cohort to almotrip-tan. Around two-thirds of patients passed from alower to a higher treatment satisfaction category,although one quarter still remained unsatisfied at theend of the treatment period. In addition, half thepatients fulfilled all 4 HAS criteria for treatmentresponse after the switch. Dissatisfaction with treat-ment is thus not inevitable and can be improvedby switching to a recommended treatment suchas a triptan. The data obtained in the ALMO cohortare consistent with those from previous studies

evaluating the effectiveness of almotriptan, such asthe Reality with Almogran and Standardized Studywith Almotriptan in Early Treatment of Migraine(START) studies.19,20 The finding that increasedpatient-reported treatment satisfaction is associatedwith better treatment response as determined withthe HAS criteria being fulfilled is consistent withsimilar findings from the GRIM2005 study.16 Thisassociation reinforces the external validity of the VASfor measuring treatment satisfaction.

With regard to stability, 3 out of 4 patients inthe VASCO cohort who were satisfied with treatment

Table 4.—Mean Visual Analog Scale (VAS) Scores at Study Closure as a Function of the Number of Haute Autorité de Santé(HAS) Treatment Response Criteria Fulfilled.

HAS criteria fulfilled

Mean VAS Score at Study Closure � SD

0-4 (N = 32) 4-7 (N = 52) 7-10 (N = 115)

0 3.30 (n = 1) None None1 2.20 � 1.06 (n = 21) 4.80 � 0.14 (n = 2) 8.50 (n = 1)2 3.28 � 0.66 (n = 8) 5.63 � 1.03 (n = 8) 7.20 (n = 1)3 None 5.46 � 0.70 (n = 11) 8.10 � 0.64 (n = 26)4 2.05 � 0.78 (n = 2) 6.26 � 0.60 (n = 31) 8.30 � 0.81 (n = 87)P .046 .005 .276

These data are derived From both the ALMO and VASCO cohorts. SD = standard deviation.

Fig 3.—Number of patients fulfilling Haute Autorité de Santé (HAS) criteria at study closure. Gray columns: ALMO cohort; blackcolumns: VASCO cohort. Data are expressed as absolute numbers of patients.

Headache 1015

at inclusion (VAS score �7) remained satisfied attheir third treated headache, and only 7 patients in all(<5%) became dissatisfied (VAS score �4). Thisimplies that patients who are satisfied with theirmigraine treatment tend to remain so over time.Nonetheless, significant drift was observed betweenthe �7 “satisfied” zone and the >4 to 7 “ambiguous”zone. This may reflect to some extent variability intreatment response between individual headaches,either due to intrinsic factors, such as variability inheadache severity, or to behavioral factors, such as therapidity with which nascent headaches were treated.In this respect, it may be informative to use the VASacross a series of headaches to obtain a more preciseidea of treatment satisfaction. For example, inclusionof the VAS in patient’s headache diary may be ofinterest, and future studies could address this. In addi-tion, a study using headache diaries would be usefulto determine the stability of the VAS score in patientswho are dissatisfied with treatment, which was notaddressed in the present study.

With respect to reproducibility, this was demon-strated in 3 out of the 4 conditions tested, namely inthe ALMO cohort both at inclusion and closure andin the VASCO cohort at closure, for which between-setting agreement was good. On the other hand, inthe VASCO cohort at inclusion, agreement was onlyfair. For the individual patient analysis, test–retestreproducibility was very strong when the test wasadministered first at home and then in the consulta-tion (rs > 0.85) and strong when administered theother way round (rs ª 0.6). In both cohorts, reproduc-ibility was less strong in the latter setting, with someevidence that subjects rated treatment satisfactionhigher when the VAS was completed in the presenceof the physician than when completed at home.This may reflect the fact that patients tend to telltheir physicians that treatment is more satisfactorythan it really is. This “trying to please the doctor” biashas been observed previously in other areas of medi-cine. This observation would lead us to recommendthat the VAS should be completed by patients ontheir own, for example in the context of a headachediary.

An original feature of this study was the stratifi-cation of the patients into 2 cohorts (VASCO and

ALMO) on the basis of their level of treatment sat-isfaction at inclusion. The 2 cohorts were very similarin terms of their demographics and migraine featuresbut differed markedly in the type of treatmentthat they were using. In the VASCO cohort (satisfiedpatients), more patients were using recommendedtreatments, with 70% taking a triptan, and the use ofprophylactic treatments was twice as high as in theALMO cohort. On the other hand, in the ALMOcohort (initially unsatisfied patients), less than halfwere using a triptan, and use of non-recommendedtreatments such as paracetamol and opioid analgesicswas 3 times higher than in the VASCO cohort. Thesefindings are consistent with surveys such as MAZE,20

FRAMIG 3,15 and START-1021 which have demon-strated that patients using triptans are more satisfiedwith their treatment than patients taking non-recommended medication. However, it is noteworthythat 40% of patients in the ALMO cohort weredissatisfied with the triptan that they were taking atinclusion. Again this is coherent with data from pre-vious clinical trials which show that around one-thirdof patients fail to respond adequately to a giventriptan, but that these patients with a poor response toa given triptan can frequently benefit from switchingto a different one.22

In the light of these observations, and also giventhe relatively small proportion of migraine sufferersin the general population who regularly use trip-tans,5,8,23,24 there is clearly room for improvement instandards of care by more proactive switching ofpatients whose treatment response is unsatisfactoryfrom non-recommended to recommended treat-ments, and if necessary from one triptan to another.For this purpose, treatment satisfaction needs tobe tested systematically and regularly so that adjust-ments can be made in a timely manner. The VASdescribed here provides a simple and quick way forphysicians to identify dissatisfaction with treatment,and thus inadequate treatment response.

This study has a number of limitations. First,the study centers may not be representative ofmigraine care in France, because they were restrictedto neurologist practices, and in particular to thosesufficiently interested in migraine to participate in thestudy. Such centers may provide better-than-average

1016 June 2012

care which could lead to an overestimate of overalltreatment satisfaction and an underestimate of anymismatch between VAS score ratings made in frontof the physician and at home. Second, the proportionof protocol violations and the non-completion ratefor the study were relatively high, which may havecompromised the sensitivity of the study, because thetarget sample size was not achieved. Nonetheless,even with a smaller sample size than planned, it wasstill possible to detect a significant change in scoresfollowing a treatment switch in the ALMO cohort.For the majority of these non-completers, the reasonsfor not including them in the analysis related toconstraints of the study protocol (for example,the requirement for a minimum number of treatedheadaches or for a maximum permitted intervalbetween last headache and follow-up visit), ratherthan being inherent to the questionnaire. In fact,only 45 patients overall (12%) failed to completetheir end-of-study questionnaire, suggesting that itwas relatively acceptable for patients.

Finally, no attempt was made in this study to vali-date the VAS rating with some independent measureof treatment satisfaction, which means that artifactscannot entirely be excluded. However, during thedevelopment of the VAS scale, it has been validatedagainst independent satisfaction measures in a largepopulation of migraineurs17 and, in the present study,there was good concordance between the VAS ratingand the HAS criteria for treatment response. In thiscontext, further studies could perhaps be useful tocharacterize more robustly the psychometric proper-ties of this scale, perhaps looking at long-term varia-tion in scale metrics, using multimodal assessments,externally validated methods, and a priori identifiedcriterions of performance.

In conclusion, the VAS scale described here is aresponsive and easy-to-use tool for evaluating treat-ment satisfaction and for monitoring changes to treat-ment if these are required. This scale could be usedin more proactive management of migraine and itstreatment by physicians.

Acknowledgments: The authors would like to thank

Florence Mercier (STAT-Process, Port-Mort, France) for

her valuable contribution to the statistical analysis of data

of this study.

STATEMENT OF AUTHORSHIP

Category 1(a) Conception and Design

Christian Lucas; Bashar Allaf; Michel Lantéri-Minet

(b) Acquisition of DataBashar Allaf

(c) Analysis and Interpretation of DataChristian Lucas; Michel Lantéri-Minet; SophieRomatet; Claude Mekiès

Category 2(a) Drafting the Article

Christian Lucas; Michel Lantéri-Minet(b) Revising It for Intellectual Content

Christian Lucas; Michel Lantéri-Minet

Category 3(a) Final Approval of the Completed Article

Christian Lucas; Sophie Romatet; ClaudeMekiès; Bashar Allaf; Michel Lantéri-Minet

REFERENCES

1. Dahlof CG, Dimenas E. Migraine patientsexperience poorer subjective well-being/quality oflife even between attacks. Cephalalgia. 1995;15:31-36.

2. Duru G, Auray JP, Gaudin AF, et al. Impact ofheadache on quality of life in a general populationsurvey in France (GRIM2000 Study). Headache.2004;44:571-580.

3. Burton WN, Landy SH, Downs KE, Runken MC.The impact of migraine and the effect of migrainetreatment on workplace productivity in the UnitedStates and suggestions for future research. MayoClin Proc. 2009;84:436-445.

4. Stang P, Cady R, Batenhorst A, Hoffman L.Workplace productivity. A review of the impact ofmigraine and its treatment. Pharmacoeconomics.2001;19:231-244.

5. Pradalier A, Auray JP, El Hasnaoui A, et al. Eco-nomic impact of migraine and other episodic head-aches in France: Data from the GRIM2000 study.Pharmacoeconomics. 2004;22:985-999.

6. Stovner LJ, Andree C. Impact of headache inEurope: A review for the Eurolight project. J Head-ache Pain. 2008;9:139-146.

Headache 1017

7. Henry P, Auray JP, Gaudin AF, et al. Prevalenceand clinical characteristics of migraine in France.Neurology. 2002;59:232-237.

8. Lantéri-Minet M, Valade D, Géraud G, ChautardMH, Lucas C. Migraine and probable migraine –results of FRAMIG 3, a French nationwide surveycarried out according to the 2004 IHS classification.Cephalalgia. 2005;25:1146-1158.

9. Lipton RB, Diamond S, Reed M, Diamond ML,Stewart WF. Migraine diagnosis and treatment:Results from the American Migraine Study II. Head-ache. 2001;41:638-645.

10. Lucas C, Chaffaut C, Artaz MA, Lantéri-Minet M.FRAMIG 2000: Medical and therapeutic manage-ment of migraine in France. Cephalalgia. 2005;25:267-279.

11. Lantéri-Minet M, Massiou H, Nachit-Ouinekh F,et al. The GRIM2005 study of migraine consultationin France I. Determinants of consultation formigraine headache in France. Cephalalgia. 2007;27:1386-1397.

12. Brandes JL. Global trends in migraine care: Resultsfrom the MAZE survey. CNS Drugs. 2002;16(Suppl.1):13-18.

13. MacGregor EA, Brandes J, Eikermann A. Migraineprevalence and treatment patterns: The globalMigraine and Zolmitriptan Evaluation survey.Headache. 2003;43:19-26.

14. Géraud G, Lantéri-Minet M, Lucas C, Valade D.French guidelines for the diagnosis and manage-ment of migraine in adults and children. Clin Ther.2004;26:1305-1318.

15. Lucas C, Géraud G, Valade D, Chautard MH,Lantéri-Minet M. Recognition and therapeutic man-agement of migraine in 2004, in France: Results ofFRAMIG 3, a French nationwide population-basedsurvey. Headache. 2006;46:715-725.

16. Lucas C, Lantéri-Minet M, Massiou H, et al. TheGRIM2005 study of migraine consultation in FranceII. Psychological factors associated with treatmentresponse to acute headache therapy and satisfactionin migraine. Cephalalgia. 2007;27:1398-1407.

17. Lantéri-Minet M, Massiou H, Romatet S, Barba A,Lucas C, Allaf B. An instrument to assess patientperceptions of satisfaction with acute migrainetreatment (EXPERT study). Headache. 2011;51:590-601.

18. Headache Classification Committee of the Interna-tional Headache Society. Classification and diagnos-tic criteria for headache disorders, cranial neuralgiasand facial pain. Second edition. Cephalalgia. 2004;24:1-160.

19. Altman DG. Practical Statistics for MedicalResearch. London, UK: Chapman and Hall; 1991.

20. MacGregor EA, Brandes J, Eikermann A,Giammarco R. Impact of migraine on patientsand their families: The Migraine And ZolmitriptanEvaluation (MAZE) survey – phase III. Curr MedRes Opin. 2004;20:1143-1150.

21. Ceballos Hernansanz MA, Sanchez Roy R,Cano Orgaz A, Lopez-Gil A. Migraine treatmentpatterns and patient satisfaction with prior therapy:A substudy of a multicenter trial of rizatriptaneffectiveness. Clin Ther. 2003;25:2053-2069.

22. Dodick DW. Triptan nonresponder studies: Implica-tions for clinical practice. Headache. 2005;45:156-162.

23. Panconesi A, Pavone E, Vacca F, Vaiani M, Banfi R.Triptans in the Italian population: A drug utilizationstudy and a literature review. J Headache Pain.2008;9:71-76.

24. Williams D, Cahill T, Dowson A, et al. Usage oftriptans among migraine patients: An audit in nineGP practices. Curr Med Res Opin. 2002;18:1-9.

1018 June 2012