patient preferences for drug treatments …...patient preferences for drug treatments for multiple...

148
PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS M a r k Stephtn Rolnick A thtsis submitted in conformity with the requirements for the degree of Masters of Science Graduate ûepartment of Pharmaceuticai Sciences University of Toronto O Copyright by M a r k Stephen Rolnick 1999

Upload: others

Post on 28-Jul-2020

2 views

Category:

Documents


0 download

TRANSCRIPT

Page 1: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS

M a r k Stephtn Rolnick

A thtsis submitted in conformity with the requirements for the degree of Masters o f Science

Graduate ûepartment o f Pharmaceuticai Sciences University of Toronto

O Copyright by M a r k Stephen Rolnick 1999

Page 2: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

National Library lJl1 OfCamda Bibliothèque nationale du Canada

Acquisitions and Acquisitions et Bibliographie Services services bibliographiques 395 Wdlingtm Street 395. rue We~WlgtOri OttawaON K1A ON1 OnawaON K t A W canada canada

The author has granted a non- exclusive licence allowing the National Library of Canada to reproduce, loan, distriôute or seU copies of this thesis in microfom., paper or electronic formats.

L'auteur a accordé une licence non exclusive permettant à la Bibliothèque nationale du Canada de reproduire, prêter, distribuer ou vendre des copies de cette thèse sous la fome de microfiche/^ de reproduction sur papier ou sur format électronique .

The author ret- ownership of the L'auteur conserve la propriété du copyright in this thesis. Neither the droit d'auteur qui protège cette thèse. thesis nor substantial extracts fkom it Ni la thèse ni des extraits substantiels may be p ~ t e d or othewise de celle-ci ne doivent être imprimés reproduced without the author's ou autrement reproduits sans son permission. autorisation.

Page 3: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

ACKNOWLEDGEMENTS

I would like to thank Dr. Linda MacKeigan for her support, guidance and commitment

over the last two years. Her critical thinking and writing skills were crucial to the

completion of this project. 1 feel privileged to have had her as my supervisor. Special

thanks to the members of my Advisory Cornmittee; Dr. Paul O'Connor, Dr. Bernie

O'Brien, Dr. Joan Marshman and to Dr. Vinay Kanetkar-

I would also like to thank the neurologists and s t a f f of the MS clinics at St. Michael's

Hospital and McMaster University for their assistance during the recruitment phase of

this project.

Final1 y. I would like to thank my parents, brother and sister and grandmother for their

moral support and Alyssa for her patience and understanding.

Page 4: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

TABLE OF CONTENTS

Page

AB STRACT 1

-*

LIST OF APPENDICES u

... LIST OF TABLES 111

1. ][NTRODUCTION 1.1 Background 1 2 Research Purpose, Objectives and Hypotheses 1.3 .Assurnptions of the Research

2. LITERATURE REVlEW 9 2 .1 Multiple Sclerosis 9 2.2 Disease Modifying Dmgs for Multiple Sclerosis (DMMSDs) 11 2.3 Utilization of DMMSDs in Canada 13 2.4 Economic Studies of Multiple Sclerosis and its Treatments 14 2.5 Patients' Treatment Preferences 16

2.5.1 Patient Participation and Preferences in Medical Decision Making 16

2.5.2 Effect of Drug Attributes on Patient Preferences for Treatment 18

2 -53 Effect of Disease Severity and Prior Experience on Patients' Treatment Preferences 20

2.6 Modehg Patient Preferences for Treatments 21 2.7 Conjoint Anaiysis 23 2.8 Ranking and Rating versus Choice-Based Conjoint Analysis 25 2.9 The Multinomial Logit Choice Mode1 27 2.10 Contingent Valuation 28 2.11 Summary 30

METHODS 3.1 S tudy Design 3.3 Subjects 3 -3 Recruitment Procedures 3 -4 Development of the Treatment Preference Questionnaire

3 -4.1 Attribute Determination 3 -4.2 Attribute Level Determination 3 -4.3 Number of Treatment Alternatives per Choice Task 3 -4.4 Number of Choice Tasks in the Questionnaire 3 -4.5 Expert Review 3.4.6 Pilot Test 3 -4.7 Final Design

Page 5: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

3 - 5 Sarnple Size 3.6 Demographic and MS-Related Medical History Questionnaire

3 -6.1 Prior DMMSD Expenence 3 -6.2 Disease Severity 3 -6.3 Wdhgness to Pay for Existing DMMSDs by Contingent

Valuation 3.7 Data Collection and Management Procedures 3. S Data Analysis

3.8.1 Screen for Rationality of Responses 3 -8.2 Statistical Model 3 -8.3 Effect of Disease Severity and DMMSD Experience

Interactions 3.8 -4 Merred Wdhgness-to-Pay for Improvements in

Attniute Levels and for Existing Dmgs 3 -9 Vaiidity of the MNL Choice Model

4. RESLZTS 4.1 Useable Surveys 4.2 Description of the Sample 4.3 Lexicographie Decisions 4.4 MuItinomial Logit Analysis of Choice Tasks

4.4.1 Drug Attribute Importance 4.4.2 Infiuence of Disease Severity and DMMSD Expenence

on Patient Preferences 4-5 Willingness-to-Pay for Changes in Existing Therapies 4.6 Willingness-to-Pay for Existing Therapies 4.7 Summary of Results

5. DISCUSSION AND CONCLUSIONS 5.1 Sarnple 5.2 EvaIuation of Survey Methods 5.3 Drug Attribute Importance 5 -4 Willingness-to-Pay for Changes in DMMSD Attributes 5 -5 Effect of DMMSD Experience and Disease Severity on Patient

Preferences 5 -6 Willingness-to-Pay for Existing DMMSDs 5 -7 Predictive Ability of the MNL Choice Model 5.8 Study Limitations 5 -9 Recommendations for Future Research 5.10 Conclusions

REFERENCES

Page 6: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical Sciences, University of Toronto.

Purpose: Disease-rnodifiing dmgs 0 s ) are a recent breakthrough in the treatment of

multiple sclerosis (MS),yet they have not been widely adopted The objectives of this

study were to detennine: 1) the relationship between drug attributes and patient choice of

a DMD, 2) the effects of disease severity and DMD-experience on this relationship, and

3) patients' willingness-to-pay (WTP) for exising DMDs and for improvements in

existing DMDs. Methods: A convenience sample of 100 patients tiom two MS clinics in

Southem Ontario completed a questionnaire consisting of 15 drug choice tasks. Each of

the 2 dmgs was described in terms of 7 attributes. Discrete choice modeling was

conducted using multinoniial logistic (MNL) regression analysis. Analyses were repeated

with subgroups based on disease severity and DMD-experience. Results: Changes on al1

7 drug attributes significantly affected patient choice. Cost changes were the most

important, followed by dosage form and chance of flu-like symptoms. Patients with more

severe disease were more willing to accept lesser effectiveness in reducing relapses.

Patients with DMD experience were not as influenced by cost as were DMD naïve

patients. WTP for a typical existing DMDs was $227 I $44 per month by CV; WTP was

$49 to $723 per month for improvements in individual attributes. The drug attribute

mode1 predicted 70% of respondents' choices. Conclusions: MS patients' choices of

hypothetical DMDs were more infiuenced by changes in negative effects (side effect

profile, dosage form and cost) than positive effects of therapy. Disease severity and pnor

experience with DMDs affected drug choice. Patients are willing to pay substantial

amounts for improvements in DMDs.

Page 7: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

LIST OF APPENDICES Page #

APPENDIX A:

APPENDiX B

APPENDK C:

APPENDCX D:

APPENDE E:

APPENDiX F:

APPENDiX G:

APPENDIX H:

APPENDK 1:

APPENDlX J:

APPENDlX K:

APPENDlX L:

Human Subjects Approval: University of Toronto

Human Subjects Approvd: McMaster University

Hurnan Subjects Approval: St. Michael's Hospital

Patient Information Sheet

Study Consent Form

Focus Group Report

Attri'bute Ranking S u ~ e y

Attribute Ranking Survey Report

Pilot Test Report

Code Book o f Variables

Treatment Preference Questionnaire

Demographic and Medical Questionnaire

Page 8: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

LIST OF TABLES

Table 1.

Table 2.

Table 3 .

Demographic Characteristics of Focus Group Participants

MS Related Medical Wstory of Focus Group Participants

Demographic Characteristics of Participants in the Attribute Rating Survey

Table 4. MS-Related Medical History of Participants in the Attribute Rating Survey

Table 5. Importance of Drug Attributes for Drug Choice: Results of a Preliminary Phase of the Study

Tabte 6. Frequency of Attribute Selection in Six Most Important: Results of a Preliminary Phase of the Study

Table 7.

Table 8.

Table 9.

Table 10.

Table 1 1.

Demographic Characteristics of Pilot Survey Participants

MS-Related Medical History of Pilot Survey Participants

Demographic Characteristics of Survey Participants

MS Related Medical H i s t o ~ of Survey Participants

Demographic Characteristics of DMMSD vs Non-DMMSD Experienced Survey Participants

Table 12. MS-Related Medical HÏstory of DMMSD vs Non-DMMSD- Experienced Survey Participants

Table 13. Most Common Reasons that DMMSD-Naïve Patients Had Not Tried Therap y

Table 14,

Table 1 S.

DMMSDs Used by Survey Participants

Demographic Characteristics of Participants with Mild MS vs. Moderate MS

Table 16. MS Related Medical History of Participants with Mild MS vs. Moderate MS

Table 17. Results of Maximum Likelihood Estimation: Main Effects Only Mode1

Page 9: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

Table 18. Results of Maximum Likelihood Estimation: Mode1 with Main Effects and DMMSD-Experience Interactions 65

Table 19. Results o f Maximum Likelihood Estimation: Mode1 with Main Effects and Disease Severity Interactions 66

Table 20. Lnferred Wiliingness-to-Pay for hprovements in Existing Therapies. 68

Table 21. Frequency Distribution o f Willingness-to-Pay for an Existing DMMSD

Table 22. Mean Wiltingness-to-Pay for an Existing DMMSD 69

Page 10: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

CHAPTER 1

INTRODUCTION

1.1 Background

Multiple sclerosis (MS) is a chronic, often disabling, neurological condition which

affects approximately 3 5,000 to 50,000 Canadians. Relapsing remitting MS (RRMS) is the

most common fonn of MS, presenting in 85% of patients at the time of diagnosis. It is

characterized by clearly defined episodes of acute worsening of neurological fùnction

(relapses), that last corn days to weeks, and manifest in symptoms ranging fiom blurred

vision and fatigue tu numbness and weakness in the extremities. Relapses are followed by

partial or complete recovery penods (remissions) that are fiee of progression of disease.

Secondary progressive MS, characterized by a steadily worsening disease course between

relapses, develops in approxirnately haif of RRMS patients between 1 O to 15 years after

disease onset (Weins henker 1 994).

Until 1995, only supportive therapy was available to MS patients. This included

corticosteroids to treat relapses, and muscle relaxants, antidepressants, anticonvulsants, and

CNS stimulants to manage associated symptoms. Since then, four disease-modifjing M S

drugs (DMMS Ds) have become avaiiable in Canada: Interferon beta- 1 b (BetaseronB),

copolymer-1 (CopaxoneB) and two interferon beta-la drugs (AvonexB and RebitlB). The

recent availability of DMMSDs has brought a new sense ofoptimism to patients, fnends,

relatives and neurologists alike, because the drugs represent the first step in the treatment of

MS. The DMMSDs are capable of preventing relapses and slowing progression of disease,

but they do not treat active symptoms or reverse existing nerve damage.

Page 11: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

2

Only approxhately 35% of Canadian RRMS patients who are suitable candidates for

therapy (ambulatory and at least two relapses in the past two years) are using a DMMSD

(EUS Canada 1999). The reasons why utilkation rates are low are not clear.

To fÙUy interpret the DMMSD utiiization, we must understand the process by which

medical decisions are made. The current trend is for patient-physician relationships to move

away fkorn a paternalistic model of patient care and toward a collaborative model; that is,

patients are taking on a more active role in their own medical decision- making. MS patients

are Likely to participate in their medical decision-making because the disease causes

signïfkant detenoration of their quality of üfe. Aiso, research shows that Young, female and

Caucasian patients prefer to participate in their medical decision-making (Struii 1984,

Stiggelbout 1997). These characteristics are generally found amongst the MS population

(Kesselring 1998, O'Connor 1998).

The treatment context for MS patients supports a coilaborative decision-makuig

process. Ln Ontario, local neurologists generdy care for MS patients. However,

neurologists, in tertiary care MS clinics, see patients for speciaiiied treatment of reiapses, for

semi-annuaYmual check-ups, and for purposes of research including clinical dmg trials. Any

medical doctor can write a prescription for a DMMSD; however, neurologists who specialize

in MS are most commonly the prescribers as they are familiar with the drugs and the

necessary paperwork for insurance coverage. In addition, they can detemine ifpatients meet

the ciinical indications of the drugs. These neurologists may introduce the concept of

DMMSDs to suitable patients seen in the MS clinic, or altematively, patients may initiate the

discussion about treatment options. Regardless of who broaches the subject of disease-

rnodiijing therapy, neurologists generally offer therapeutic opinions to patients, but leave the

patient with the final decision ofwhether or not to try a disease-modifjing dnig.

Page 12: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

3

The facts that MS patients are puiicipating in their medical decision-making and that

they are not hi& usen ofDMMSDs imply that most patients prefer not to try existing

therapy. A patient preference is defined as "an action a patient would choose in a particular

medicai situation at a particular tirne, given a set of alternatives7' (Herman 1985). Research

on which drug treatment options MS patients prefer has not been conducted because MS-

related research has traditionaily focussed on finding treatments for the disease itself: rather

than exploring social/psychologicaI aspects of treatment decision-making. Now that four

DMMSDs are on the Canadian market, each one similar, but with some distinct

characteristics, and with more products soon to arrive, the time has corne to assess MS

patient preferences for DMMSDs

One way of enhancing our understanding of why some patients are deciding to try a

DMMSD while others are not, is to examine the effect of dnig attributes (i-e. route of

administration, side effects, efficacy, cost, etc) on patient preferences for treatment. Several

studies have examined the importance or influence of drug attributes on patient preferences

for medical treatment (Stanek 1997, Ludwig et al. 1997, Luciani et al. 1995, Rosenfeld 1997,

Reardon 1990, Gabriel 1992, Ryan and Hughes 1997, Singh 1998). Generally, these

att nbutes include positive effects, side effects and cost.

It is conceivable that the attributes of available DMMSDs are afEecting patients'

decisions to try them. In terms of positive effects, the DMMSDs reduce relapse fiequency, as

well as the seventy of relapses. Relapses are rare events that occur an average of once per

year to once every four years (Kesselring 1997), and tend to decrease in fiequency as disease

duration increases. According to clinical trials, the DMMSDs reduce the relapse rate by

approximately one-third (Wienstock-Guttman and Cohen 1996). Some of the drugs have also

been shown to successfully slow the progression of disease, which often develops in tandem

Page 13: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

1

with incomplete recovery fiom relapses, over a penod of years (European Study Group on

Interferon Beta-lb in Secondary Progressive MS 1998 and Jacobs et al. 1996).

Ail of the existhg dnigs require d d y to weekly injections (either subcutaneously or

intramuscularly). Aversion to the drugs because of fear of needles is exacerbated by the fact

that injection site reactions, presenting as red areas of lasting soreness, occur with some of

the DMMSDs. Other side enects of DMMSDs Uiclude: a chest tightness/flushing reaction

(with copolymer-1), pain on injection and flu-like symptoms including fever, chiils sweats,

myalgia and malaise (with interf'erons).

The DMMSDs are expensive, ranging corn $12,000 to $20,000 CDN per patient

year; however, in Ontario, the Dmg Benefit (ODB) Program and other third party payors are

currently covering most of the cost for ambulatory patients who have had at least two

relapses in the last two years. Coverage is restricted to these patients because, untif very

recently, the DMMSDs have only been studied in this group.

The value that patients place on the DMMSDs is currently unknown, but may be

reflected in patients' preferences for the attributes of the DMMSDs. In theory, improvements

in drug attributes should increase the value of the drugs. However, we do not know how

much these improvements are worth to patients. One approach to assessing value is to survey

patients regarding their willingness-to-pay (WTP) to use a DMMSD. Donaldson and

Shackley (1 997) daim that conventional WTP methodology (contingent valuation), eliciting

how much respondents are willing-to-pay for an outcome, may not be sensitive enough to

capture the value associated with the process of treatment, an important detenninant (in

addition to health outcornes) of overall treatment value. ln the case of DMMSD valuation,

simply asking patients directly how much they are willing-to-pay for DMMSDs may not

Page 14: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

convey the disutility associated with regularly scheduled injections that the DMMSDs

necessitate-

Donaldson and Shackley (1997) suggest that conjoint analysis may be an alternative

method of m e a s u ~ g WTP, that may be more sensitive to process attributes. Conjoint

analysis (CJA) is a survey technique commonly used in marketing research to explain and

predict consumer preferences for products and their attributes. CJA is defined as "any

decompositional mode1 that estimates the structure of a consumer's preferences given hislher

overall evaluations of a set of alternatives that are pre-specified in terms of levels of daerent

attributes" (Green and Srhivasan 1978). Through CJ& the importance of attributes that

describe products or services can be detennined. In addition, when cos is included as an

attribute in the profiles being evaiuated, WTP for changes in attribute levels (including

process attributes) and for entire treatments cm be inferred. Although CJA was developed in

the 1960's, it has seldom been used in health care and even less often used to address patient

preferences for pharmaceuticals. Although Ryan and Hughes (1997) have promoted CJA as a

tool to be added to an economist's tool-box, cornparisons between WTP estimates obtained

from contingent valuation and fiorn CJA have not been made in healthcare.

CJA can be used to assess the tradeoffs that patients make, based on their

preferences. IntuitiveIr, MS patients would prefer treatments that are administered orally,

have high effectiveness and have low risk of side effects, compared to treatments that are

injected, have poor effectiveness and have high risk of side effects. However, we do not

know what tradeoffs MS patients are willing to make among attributes (e-g. ùnproved

effectiveness at the cost of increased side effects).

The effect that patient characteristics have on patient preferences can dso be

explored with CJA This is important because seventy of disease and DMMSD-experience

Page 15: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

6

rnay influence the effect of a drug's attniutes on patient preferences. Many patients, with a

mild form of RRMS, have a 'wait and see' attitude towards trying disease-rnodmg therapy

because they feel that MS is not impairing their Iives enough to warrant invasive dmg

therapy. AIso, R R M S patients who are in remission and asyrnptomatic, may have a fdse

sense of security about the probability of recurrence of fbture relapses. Patients whose MS

has progressed rnay be willing to accept more risk when making treatment choices. These

patients rnay be more willing to try a new or experimental dmg therapy with a more

bothersome side effect profile than those patients with mild disease. Thus, patients fiom

opposite ends of the clinical spectrum of MS rnay d ï e r in their overali preferences for

treatment and specifically on the importance that specitic dmg attributes have on their

treatment choices.

Evidence f?om the literature suggests that health care experiences affect patients'

preferences for treatment (Ryan et al. 1998). Patients who have taken a DMMSD likely view

the attributes of dmg therapy differently fiom those who have not tned a DMMSD.

In summary, patient preferences for DMMSDs have not been studied. Further, this

thesis postulates that patient preferences for DMMSDs are based on drug attributes and rnay

be modified by disease severity and DMMSD-expenence. CJA rnay prove to be a usefiil

technique to help expiain patient preferences for DMMSDs.

Page 16: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

1.2 Research Purpose, Objectives rad Hypothcscs

The purpose of this study is to assess the inauence of drug attributes on patient

choice of disease-modifling dmgs.

Primary Research Objecrives

1. To determine the relationship between drug attributes and patient choice of a DMMSD.

2. To estimate how much patients are wiiiing to pay for a change in a specific attribute of a

DiMMSD.

3. To determine the eEect of experience with DMMSDs on the relationship between

specific dnig attributes and patient choice ofa DMMSD.

4. To determine the effect of disease severity on the relationship between specific drug

attributes and patient choice of a DMMSD.

Secortdav Reseurch Objective

1 . To compare patients' willingness to pay for existing DMMSDs estimated by contingent

valuation and iderred by CJA

Hypo theses

1 . Change in the level of a drug attribute wiil influence patient preference for a DMMSD.

2. Disease severity will interact with drug attributes to affect patient choice of DMMSD.

3. Experience with DMMSDs wiii interact with dmg attributes to affect patient choice of

DMMSD.

Page 17: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

8

1.2 Assumptions of the Rcsearch

In conducting this research several assumptions were made. The main assumption of

was that patients' choices of hypotheticd DMMSDs are indicative of their real choices. A

second assumption was that patients have the cognitive capacity to process the amount of

information presented to them on multiple drug attributes, in order to make a treatment

decision. A third assumption was that physicians present treatment options to patients, but

patients always make the decision to try a DMMSD or not, and which DMMSD to try.

Page 18: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

The first part of this titerature review provides background uiformation on multiple

sclerosis, its treatments and its costs. Then, patient participation in medical decision making

and patient preferences for drug treatrnents are discussed. The effects of drug attributes,

disease severity and prior drug therapy experience on patients' treatment preferences are

presented next. in the Iast part ofthis chapter, decision models and methods of estimating

willingness-to-pay for healthcare services are reviewed, focussing on conjoint analysiq the

mode1 selected for this study.

2.1 Multiple Sclerosis

Multiple sclerosis (MS) is a chronic, and often disabling, neurological condition.

Classified as an autoimmune disease, the cause of MS remains largely unknown. The immune

systern of a person with MS recognizes myelin, a fatty protein which insulates nerves of the

central newous system, as a foreign protein, and initiates an attack on it. This destroys the

myelin coating (demyelination), ieaving the nerve cells idamed and scarred. The

demyelination happens sporadicaliy, and the intensity and the location of the attacks are

unpredictable, varying greatly within and between patients (O'Connor 1998).

MS manifests through various distinct clinical patterns. Three disease classifications,

Relapsing remitting (RRMS), Secondary Progressive (SPMS) and Primary Progressive

(PPMS) make up the majority of aii cases. RRMS is characterized by cleary defined episodes

of acute worsening of neurological fùnction known as relapses, attacks or exacerbations.

Relapses Vary in severity and duration (days to weeks) and manifest in symptorns such as

Page 19: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

10

visual disturbances, fatigue, muscle incoordination, and numbness in arms and legs. Partial or

complete recovery penods known as remission follow these exacerbations. RRMS is the

most common form of the disease at the t h e of diagnosis, with 85% of patients falling into

thi s category. S econdary progressive MS, characterized by a steadily worsening disease

course, develops in approxhately haif of RRMS patients within 10- 15 years of disease

onset. Primary progressive disease, occumng in 1 5% of patients, involves progression from

the onset, with or without relapses (Weinshenker 1994).

Although the manifestations of MS differ in each patient, some general observations

have been made. The onset of MS peaks in people in the late twenties and early thirties

(Sadnovick et al. 1993). MS affects nearly twice as many women as men (Kesselring 1998).

The disease is more cornrnoniy found in Caucasians than in any other race (O'Connor 1998).

Life expectancy for MS patients is approxhately six years less than that of the general

population (Sadnovick et al. 1992). M e r having M S for 15 years, 50% of patients are

disabled, to the extent that they require a cane, walker or wheelchair to move 100 meters

(Weinshenker 1994).

Until the-mid 1990s, management of symptoms was the only treatment strategy

available for MS patients. Anticholinergics, antispasmodics, and unnary suppressants treat

bladder dysfllnction. Antidepressants, anticonvulsants and muscle relaxants treat pain while

muscle relaxants, CNS stimulants and antiemetics treat spasticity, fatigue and vertigo

respectively. Drugs are also available to treat impotence and depression which are other

symptoms of MS (O'Connor 19983.

High dose corticosteroids are the dmgs of choice to treat relapses. They have been

shown to accelerate recovery h m a relapse; however, they have no effect on the level of

disability that remains following an exacerbation (Weinstock-Guttman et al. 1996). Although

Page 20: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

11

the corticosteroids are typically a d d s t e r e d intravenously, the setting in which they are

administered can vary. Patients who are suffering nom a major relapse may be hospitalueci,

while those who expenence mild or moderate relapses may be treated as outpatients or in

their own homes, through home care.

2.2 Disease-modifying Drugs for Multiple Sclerosis (DMMSDs)

Since 1995, four disease-modwg drugs have become available to treat RRMS.

The DMMSDs are similar in that they ail must be administered by subcutaneous or

intrarnuscular injection and they al1 reduce relapse fiequency b y approximately one-third for

penods of up to two years. However, each drug product has some unique characteristics.

Interferon Beta- 1 b (BetaseronB), produced by recombinant DNA technology, and

approved in Canada in 1995, was the first disease-modifling drug approved to treat multiple

sclerosis. In a double-blind, placebo-controlled trial of 372 RRMS patients, patients treated

with BetaseronB for two years experienced 3 1% fewer relapses, and had a significant

reduction in MS brain lesions on magnetic resonance imaging (INFB Study Group 1995).

Nso, the number of hospitalizations and number of patients hospitalized were significantly

reduced. Another double-blind, placebo-controlled study of 7 18 patients with secondary

progressive MS found a sipificant difEerence in the t h e to codrmed progression of

disability in favour of patients on BetaseronB (European Study Group 1998). Over three

years. 50% of the placebo group progressed versus 39% of the treated group. The

recommended dose for BetaseronB is 8 million international units administered

subcutaneously, every other day. The most common side effects are flu4ke symptoms (chills,

sweats, headache, fever and malaise) and injection site reactions (redness, sweiiing, areas of

lasting soreness). Betaserona costs S 16,920 CDN per year.

Page 21: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

12

Copolymer- 1 or glatiramer acetate (CopaxoneO) was approved in Canada in

Se ptember 1 997. It is a mixture of synthetic polypeptides containing four naturall y occurring

amino acids. In a placebo-controlled, double-blind study of 25 1 RRMS patients, patients

treated with CopaxoneB experienced 29% reduction in relapses over a two-year period

(Johnson et al. 1995). Effects on brain lesions and severity of relapses were not reported in

the cli~cal trial. Furthermore, the trial did not assess the dnrg's impact on hospitalizations.

Recent reports have indicated that CopaxoneB has a beneficial effect on brain lesions

detected by MRI (Mancardi et al. 1998). The recornmended dose for CopaxoneB is 20 mg,

administered once daily by subcutaneous injection. CopaxoneB is better tolerated than

interferon. The skin reaction is less severe and there are no flu-like symptoms. However,

10% of users, on at least one occasion, have experienced a severe chest tightening reaction,

that is transient and not life threatening. CopaxoneB costs $12,760 CDN per year.

Interferon beta- 1 a (Rebi£B), a recombinant DNA product, was approved in Canada,

in February 1998, for RRMS. In a placebo-controiied double-blind study in 560 patients,

patients treated with low-dose Rebim for two years had 29?% fewer relapses. Patients

treated with high dose had 32% fewer relapses (PRISMS Group 1998). Treatment with

RebifD resulted in a significant reduction in brain lesions measured by MRI and slowing of

disease progression. Rebim's side effect profile is similar to that of BetaseronB. The

recornmended dosage for R e b D is 6 or 12 million international units, administered

subcutaneously, three times each week. The higher dose may be preferable for patients with

more advanced disability. Depending on the dose, the cost of Rebim ranges fiom % 16,560 to

$20,5 10 CDN per year.

A second interferon beta- 1 a (AvonexB), manufactured with recombinant DNA

technology, was approved in Canada, in April 1998, to treat relapsing forms of MS. ln a

Page 22: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

double-blind, placebo controiied trial of 30 1 relapsing-remitting or relapsing-progressive

patients, patients treated with AvonexO for two years experienced 32% fewer relapses and

13% fewer patients experienced disease progression (Jacobs et al. 1996). A h , the number

and volume of idammatory brain lesions were reduced. AvonexB was associateci with flu-

like symptoms iike the other interferon products but no injection site reaction. The

recomrnended dosage is 6 million international units injected, intramuscularly, once a week-

AvonexB costs $16,360 CDN per year.

2.3 Utilization of DMMSDs in Canada

Now that DMMSDs are available it is important to stakeholders (physicians, dmg

manufacturers, and insurers) to know how otten they are being used. To estimate the percent

of eligible RRMS patients who are using DMMSDs we must know the number of RRMS

patients who are suitable candidates for therapy (have had at Ieast hivo relapses in the last

two years, and are ambulatory) and the number of prescriptions filied each month.

Conservatively speaking, 35,000 Canadians have MS,' of which approximately 40% have the

RRMS form (Weinshenker 1994). If approxirnately 70% of al1 RRMS patients are suitable

candidates for therapy,* then 9,800 Canadian MS patients should at Ieast consider and/or be

considered for disease-mod-g treatment.

The market for DMMSDs in Canada has grown every year since the f im one became

available in 1995. Sales of DMMSDs were $28.7 million in 1998, up fiom 3 16.0 million in

1997. The sales figures for 1999 are expected to top $42.3 million ( IMS Canada 1999). The

growth of this market measured in sales dollars is paralleled by increases in the number of

' Estimates range fiom 35.000 to 50,000 (Sadnovick and Ebers 1993)

' Communication with Dr. P. O'Connor

Page 23: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

14

prescriptions. The number of prescriptions filleci in 1999 is expected to exceed 4 1,000.

Assuming monthly refiils of prescriptions, approximately 3,400 patients are getting a

DMMSD prescription filled each month. This represents 35% of suitable RRMS patients.

With continued growth in this market expected in the fiiture, as better, more convenient and

possibly more expensive products become available, it is important to look at the economics

behind MS and its treatments.

2.4 Economic Studies o f Multiple Sclerosis and its Treatments

Within 10 years of being diagnosed with MS, over one-halfof patients become

unemployed (Bourdette et al. 1993). As one might expect, as MS patients become more and

more disabled over time, their cost to society increases because of increasing unemployment

and increasing use of the hedth care system (Inman 1984, Bourdette at al. 1993, Canadian

Burden of Illness Study Group 1998). Thus, ifdisability can be prevented by slowing the

progression of disease, people with MS will cost society less.

Two burden of illness studies have been done in Canada to assess the cost of MS.

Both used the human capital approach, which places monetary weights on healthy time using

market wage rates, and estirnates the costs of disability and premature mortality by the

present value of fiiture earnings lost (Drummond et al. 1997). Asche et ai. (1997) concluded

that the total costs of MS for Canadians in 1994 were $502.3 million, Of this, $188.6 million

were direct costs (hospitals, other institutions, medical services, non-DMMSD drugs and

prescription fees), while $3 13.8 million were costs of disability and premature mortality. The

Canadian Burden of Iihess Study Group (1998) estimated that the lifetime cost of MS,

including patient institutionalkation, was S 1,608,000 CDN per patient. The cost of

DMMSDs was not included in these studies.

Page 24: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

15

Although the DMMSDs have some effea on slowing disease progression, their most

clinically significant effects have been on relapse prevention. Research at the MS clinic in

Calgary found that it cost $570 CDN per day in 1993 to treat an acute exacerbation on an

inpatient basis (Harris and Metz 1993). Assuming an average of nine days in the hospital, the

total treatment cost was estimated at $5,130. The total cost of treating a relapse on an

outpatient bais was $558 CDN. Using the assumption that, on average, three days of

hospitalization are attributable to treatment of a relapse, Robson et al. (1998) estimated the

cost of treating patients for a relapse at $1,182 CDN and $7 1 5 CDN, including both inpatient

and outpatient costs. These studies underestimate the total cost associated with a relapse

because they were conducteci fiom a hospital perspective, omitting the costs of lost

productivity.

Several recent studies have attempted to address the cost effectiveness of the

DMMSDs. In a study by the Canadian Coordinating Office for W t h Technology

Assessrnent (CCOHTA), Betaseronm was found to cost between $24,073 and $76,795 per

relapse avoided over a four-year period, depending on whether or not patients were admitted

to hospital for relapse treatment (Otten 1996). It is clear that these dnigs are quite expensive

when relapse avoided is the designated effectiveness measure. This is due to three factors: 1)

the acquisition costs of these dmgs are very high, 2) relapses generally occur infiequently (on

average, less than 3 relapses every 2 years) (INFE3 MS Study Group 1995) and 3) relapses

naturally occur less often as the disease progresses over time (Kesselring 1997). Costs

associated with premanire disabiiity and mortaiity are the largest cost drivers in MS,

representing 62.5% of the total coa of MS in Canada (Asche et ai. 1997), not relapses. Thus,

it may be inappropriate to use relapses as the measure of effectiveness for an economic

evaluation of the DMMSDs.

Page 25: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

Brown et al. (1996) estirnated the cost in Canada of Betaseron8 per 'normaiized

EDSS disability year avoided' to be $2 l9,O6 1. The model they used took the average lifetirne

of an MS patient and assumed that the patient was treated until hekhe reached an Expanded

Disability Status Scale (EDSS) score of 6.0 out of 10 (zero being no MS symptoms and 10

being death fiom MS). A tùrther assurnption of this model was that BetaseronB delayed the

progression of disability by approximately 15%. The model has several Limitations, most

notably that it incorrectly assumes that EDSS scores, commonly used to measure disability,

are interval level.

In 1998, the Canadian Coordinathg Office Health Technology Assessment undertook

a cost-utility assessment of Rebi i . RebW was found to cost $406,400 per quality adjusted

life year (QALY) gained (Otten 1998). According to Laupacis et al. (1992)' a cost exceeding

$100,000 per QALY is 'weak evidence for adoption and appropriate utilization'.

One contingent valuation study has been undertaken on treatments for MS. It asked

MS patients how much they were willing to pay to achieve certain improvements in health

outcomes (Wilson et al. 1993). However, specific risks of therapies that might lead to those

health outcornes were not identified to study participants. The estimated WTP for a cure of

MS was $22,875 US per year.

2.5 Patients' Treatmcnt Prtferences

2.5.1 Patients ' Participation îmd Pre ferences in Medical Decision Making

The importance of incorporating patient preferences into medical decision-making has

been increasingly emphasized in recent years. Evidence suggests that patients who take an

active role in their treatment often perceive better health outcumes (Greenfield 1985,

England 1992), are more satisfied with care (Leman 1990) and are more cornpliant with

Page 26: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

therapy (Herman 1985).

Patient participation in medicd decision-making is consistent with the beiief that

patients should have some degree of control over theu bodies and health (Herman 1985).

Kassirer ( 1994) has identitied seven scenarios where assessing patient preferences is

important: 1) when there are major dserences in the kinds of possible outcomes, 2) when

there are major differences between treatments in the Likeiihood of and impact of

complications, 3) when making choices where trade-offs have to be made between near-term

and long-term outcomes, 4) when one of the choices can result in a small chance of a grave

outcome, 5) when the apparent difEerence between options is marginal, 6) when the patient is

particularly averse to taking risks, and 7) when a patient attaches an unusual importance to

certain possible outcomes. The third and fifth scenarios apply to the DMMSDs, wMe the

sixth and seventh scenarios are patient specific.

People differ with respect to the extent that they want to participate in their own

therapy decisions. Patients with chronic disease are more active in their care than those with

acute conditions (Brady 1998) and young people are more active in decision-making than old

(Stiggelbout 1 997). Some evidence suggests that women participate more in their medical

decision making than men, but the literature is somewhat divided (Stiggelbout 1997).

Caucasians have been shown to more actively participate in medical decision rnaking than

non-Caucasians (Stmll 1984). The demographic characteristics of MS patients (mostly

young, white and female) match those of patients who are more Uely to participate in their

medical decision-making (Brady 1998 and Stiggelbout 1997).

It is intuitive that patients who participate in their medical decision making are more

likely to receive treatments that meet their preferences. A patient's preference has been

defined as "an action a patient would choose in a particular medical situation at a particular

Page 27: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

tirne, given a set of alternatives" ( H e m 1985).

2.5.2 Eflect of Dmg Attributes on Parient Preferences for Tre~metat

Although the study of patient preferences has gained popularity in recent years, a

literature review did not yield any study that describes the effect of drug attributes (Le. route

of administration, side effects, efficacy, cost, etc) on MS patients' dmg preferences. The oniy

patient preference research related to MS has been on hyperbarïc oxygen therapy (Wynne

and Monks 1989). One reason why preferences of MS patients for disease treatment have

not been studied may be the lack of specific (disease-targeted) treatments for MS patients,

until recently.

Treatment preferences have been studied in other medicai conditions: migraine

management, A I D S treatment, thrornboiytic therapy &er myocardial infiarction and

antidepressant medication. Some noteworthy studies are described here. Luciani et al. (1995)

explored patient preferences for migraine management. Six hundred and forty eight migraine

sufferers, who had participated in a clinical tria1 of surnatriptan, were asked to rate the

importance of drug attributes (on a Likert-type scale) in inf'iuencing their choice of treatment.

In decreasing order of importance, the drug attributes were: 1) how weU it works, 2) how

safe it is, 3) how fast it works, 4) side effects, 5) physician recommended, 6) number of

doses to relieve pain, 7) total treatment cost, 8) how easy to buy, 9) how easy to take and

10) cost of the drug. The study found that there were significant differences in importance

ratings among the 10 drug attributes.

Rosenfeld (1 997) elicited patient preferences for marketed drugs, experimental dmgs

and alternative (naturopathic) treatments for AIDS. Eight treatments were evaluated.

Respondents (n=28) were presented with two treatments at one tirne and were asked to

Page 28: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

19

select the treatment option that they preferred. The treatments were descnied in terms ofl- 1)

known risks, 2) benefits, 3) FDA-approval status, 4) cost, 5) dosing schedules and 6) extent

o f empirical research regarding treatment effectiveness. Patient preferences for treatments,

analyzed using a multidimensional scaling technique, were significantly correlated with

benefits o f therapy and dosage frequency, but less so with FDA approval status, volume of

evidence and harmfiil side effects.

Stanek et al. (1997) surveyed 10 1 cardiac inpatients about preferences for

thrombolytic therapy for acute myocardial infarction (AMI), described in terms of stroke and

mortaiity risk. Under conditions of zero cost, patients preferred tissue plasminogen activator

(PA) over streptokinase (SK) due to its lower mortality rate. When cost was introduced in a

patient-payor scenario, the preference shifted towards the less expensive and slightly riskier

SK. Sirnilar trends were seen when the insurance Company or the government was the payor.

The authors concluded that patients traded offbetween c l i c a l attributes and cost in

selecting thrombolytic therapy in AMI.

Side effects of drug treatment are important considerations in patients' medical-

decision making. O'Brien et al. (1995) showed that depressed patients would be willing to

pay more (out of pocket) each month for a dmg that was equally efficacious as an existing

therapy, but had an improved side effect profiIe.

Negative attitudes towards needles have been documented in ciïnical conditions

treated with parenteral drug therapy. The discodort associated with intercavernosal

injections to treat male erectile dystùnction and the fear of subcutaneous insulin injections to

treat diabetes have sparked research into alternative drug delivery systems for these

conditions (Trehan and Ali 1998, Seyam et al. 1997, Hunt et al. 1997, Sumra and Gupta

1995, Kim et al. 1995). Needle phobia is a scientifically documented condition (Hamilton

Page 29: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

1995) that may affect patients' decisions to try a new DMMSD.

2.5.3 Effecr of Disease Severiiy and Prior Erperience on Patients' Treuîntent Preferences

Evidence from the literature suggests that disease severity and prior treatment

expenence affect patient preferences for treatment. Ludwig et ai. (1997) surveyed patients

with multiple myeloma on preferences for a disease-modïg drug (interferon alpha- 1 a).

Preferences for this drug are relevant to multiple sclerosis, which is aiso treated with

interferon. The disease-rnoditjhg dnig was described in tenns of its positive and negative

efFects, but its name was not disclosed to respondents. The authors found that patients who

were interferon naïve (comprising two-thirds of the study group) were more likely to reject

therapy than those expenenced with interferon Patients who rejected therapy did so largely

because the potential increase in life expectancy was less than satisfactory. Furthemore,

patients with advanced disease and with a reduced sense of weU being were more wilhg to

accept therapy.

Ryan et al. (1 998) explored the importance that medical outpatients place on the

availability of an electronic patient health card. They found that oniy people who actually had

such a card considered it to be a benefit. This implies that experience with a seMce affects

one's perception of i t Work by Porter and Macintyre (1984) on pregnant women's

satisfaction with antenatal care aiso suggested that patients have a preference for the care

they have already experienced. The authors offer patient wnservatism, deference andor

politeness as possible explanations of this finding.

Page 30: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

2.6 Modeling Patient Preferences for Treatments

The modeling of decision making is undertaken in fields of psychology, sociology,

economics and market research. Decision-making models address research objectives about

the process o r structure of decisions. Process models provide insight into how choices are

made; structural models are focussed on what the decision-maker chooses (Abelson and Levi

1985). Structural models were more applicable to this study of patient preferences for

disease-modwg multiple sclerosis dmgs @MMSDs) because Our purpose was to assess

what changes to these drugs would influence patients' drug choices.

Decision models can also be classified as descriptive o r prescriptive. Prescriptive

models provide information about how a decision should be made, while descriptive models

offer idormation on how decisions are made, with the goal of understanding and predictkg

choices (Green and Wind 1973). The purpose of this study, to explain patients' choices of

DMMSDs, indicates a descnptive approach.

Decision models assume that people make decisions based on characteristics of a

product or service, known as attributes. Many different rules have been proposed to mode1

how people use information about attributes to make decisions. Such 'choice rules' can be

classified as either non-compensatory or compensatory. Non-compensatory niles are

characterized by the exclusion of alternatives that do not meet pre-specified levels on specific

attributes. Non-compensatory models do not allow for tradeoffs between attributes whereby

a low value on one attribute can be compensated by a high value on another attribute. An

example of a non-compensatory decision rule is a lexicographie decision nile, which

prescribes that the alternative with the best level on the most important attribute be chosen.

People tend to use a combination of different 'choice rules' depending on the task a t hand,

Page 31: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

22

aithough generdy compensatory choice rules are more applicable to decision-making

(Abelson and Levi 1985).

Compensatory models alIow for tradeoffs to take place across attributes. By

definition, compensatory models imply that products are constmcted fiom different

attributes, each of which contributes a part-value to the overail-value of the product.

Decision-makers either consciously or unconsciously make tradeoffs on these attributes and

select the option with the greatest amount of total value.

Compensatory models are generally Iinear or multiplicative in combination of

attributes and their part-values (weightings). By including error terms, they allow for

variability due to the inability of individuals to consistently apply their 'choice rules'. Linear

rnodels are most commonly used. Even if respondents make sorne non-compensatory

decisions, lineu models are excellent predictions of individual judgernents. Linear models cm

also provide information about the relative importance of attributes (through inference or

elicitation of weights on each attribute). A iïmitation of linear models is that h conclusions

about psychological process cannot be made. (Abelson and Levi 1985).

Linear models can be compositional or decompositional in nature. Compositional

models use a "build up" approach to determine relative value for a multiattribute product or

service. Preferences for different attribute levels and importance weights for attributes are

elicited ftom subjects; total utility is then estimated fiom these values with a composition

rule. Decompositional models use a "break down" approach, in that preferences for

alternatives are first elicited and then utilities of specific attribute levels are statistically

inferred through various foms of regession analysis (Green and Snnivasan 1978).

Compositional and decompositional models each have ümitations. Limitations of

compositional models include the cognitive difficulty of rating one attribute level when other

Page 32: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

23

attributes are not independent, and the model's inability to determùie the Likelihood of trying

a product or service because respondents' judgements are of individual attributes and not of

completely described products (Green and Srinivasan 1990). A limitation of decompositional

models is that the array of idormation that is provided to respondents may exceed

respondents' cognitive capacity. A decompositionai mode1 was chosen for this study because

it can be used to infer willingness-to-pay (WTP). Also, it is more intuitive that people make

judgements on entire products descnbed by specific levels, rather than on individual

attributes and levels without seeing a compIete product description.

Two decornpositional approaches have been used to assess consumer preferences.

Multiciirnensional scaling (MDS), also known as perceptual mapping, is a technique used to

transfonn consumer judgements of products/services ont0 a map that displays the perceived

relationship arnong products. The strength of MDS lies in its ability to infer the dimensions

(combinations of attributes) aEecting behaviour, without the researcher having to define

specific attributes. A downside of MDS is that you have to identifL the dimensions which is

not an easy task. Although this technique can be used to determine which alternatives people

think are similar or different fiom each other, additional data is needed to assess which

attributes predict the position of each dternative on a s a l e ('Hair 1995). Conjoint analysis

(CJA) is the other decompositional mode1 that has been used to assess consumer preferences.

It was chosen for this study.

2.7 Conjoint Analysis

Conjoint analysis (CJA) is the most widely applied methodology for analyzing

consumer preferences (Carroll and Green 1995). Developed in the 19603, in the field of

mathematical psychoiogy (Luce and Tukey 1964; Green and Srinivasan 1 W 8 ) , it is defined as

Page 33: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

"any decompositional mode1 that estimates the structure of a consumer's preferences given

hidher overall evaluations of a set of alternatives that are pre-specified in tems of levels of

different attributes" (Green and Srinivasan 1978). In a CM, a standardized set of attributes,

each with predetermined levels, is used to descnbe comparator products (e-g. in describuig a

car, colour is an attribute while red and blue are 1eveIs)- Each product is then described in

tems of a profile of anribute levels. Thus, a factorial design is created in which the factors

are equivalent to product attributes. Utilities (part-worths) for dEerent levels of a product

attribute are obtained by decomposing subjects' holistic preferences for diEerent products

t hrough the st atistical process of regession analysis (Green and Srinivasan 1 978). B y

manipulating the attribute levels across profiles, analysts can explore the tradeoffs that

subjects make between levels of attributes and detennine the relative importance of

attributes.

Ryan and Hughes (1997) have promoted the use of CJA as a device to be included in

an economist's 'tool-box'. If cost is included as one of the drug attributes, then CJA can be

used to infer willingness-to-pay (WTP) for changes in attribute levels. This is an important

advantage of CJA, in that the value of improvements in specific attributes of a product or

service can be assessed.

The use of conjoint analysis to assess patient preferences for healthcare services is a

relatively new development. Studies of conjoint analysis in health-related fields include:

consumers' preferences for dental seMces (Chakraborty et al. 1993), consumers' preferences

for health insurance (Chakraborty, Ettenson and Gaeth 1994), physician prescribing

preferences for antihypertensive medicines (Chinburapa and Larson 1988), patients'

preferences for in vitro fertilization services (Ryan 1999)' and women's preferences for

rniscarriage management (Ryan and Hughes 1997).

Page 34: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

25

Two studies have used CJA approaches to assess patient preferences for prescription

pharmaceuticals. Reardon et ai. (2990) compareci preference ratings versus open-ended WTP

evaluations of antihistarnine products in a survey of 143 university employees who used

allergy medicines to treat aiiergic rhinitis. CJA deterrnined that the importance of the seven

attributes describing each product was, in decreasing order: 1) effectiveness of de rgy reliet:

2) drowsiness as a side effect, 3) duration of ailergy reliec 4) onset of action, 5) dryness as a

side effect, 6) dmg interactions with tranquilizers or alcohol, and 7) prescription status. Price

was not included in the descriptions, because its presence would have biased subjects'

responses on the contingent valuation portion of the survey.

Singh et al. (1 998) surveyed 159 parents of short stature children on their preferences

for actual and hypothetical growth augmentation therapies. Based on rankings of dmg

profiles, the relative importance of the attributes in decreasing order were: long-term side

effects, out-of-pocket cost, child's attitude, certainty of effect, arnount of effect, and route of

treatrnent. Although price was included as an attntute, the authors did not infer WTP.

2.8 Ranking and Rating versus Choice-Based Conjoint Analysis

Ranking and rating of product profiles are two conventional methods of assessing

preferences for products. Ranking methods are Iikely to be more reliable than rating methods

because it is easier to rank products fiom most to least preferred than to rate products on a

scale. However, one of the main drawbacks of ranking tasks is that a personai i n t e ~ e w is

ofien needed, making them burdensome to administer (Hair 1998). A rathg task is easy to

administer, but respondents may be less discriminating in their judgements wrnpared to

ranking exercises (Hair 1998).

Page 35: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

26

A major disadvantage of ranking and rating-based CJA is that a person who highîy

ranks or rates a product may not choose to use it. Ranking and rating exercises represent

judgrnents as opposed to choices. In brief, judgements imply forming opinions or estimates

that comprise one cornponent of the choice process. Choices on the other hand require the

selection of an alternative, even if the alternative is deciding 'not to decide', fiom a set of

alternatives (Abelson and Levi 1985).

Choice-based conjoint analysis (CBCJA), aiso known as discrete choice modeling, is

a new development in which the dependent variable is a choice amongst a set of alternatives.

This method of analysis more accurately simulates consumer choice of products.

Furthemore, this method is more consistent with random utility theory and economic theory

which link preferences to choice (Ryan and Hughes 1997). Economic theory assumes that

people behave, in a market place, in such a way as to maximize their expected utility and that

their expected utility is consistent with their preferences. Random utility theory assumes that

people's choices reflect their maximum utility at a given moment, recogninng that al1 factors

that affect utiiity may not be witnessed by an observer (McFadden 1986, Ben-Aiciva 1985).

A unique advantage of CBCJA is that the option of choosing none of the presented

alternatives can be included. This option adds an element of realism to the choice tasks as it

better simulates real decisions. Ryan (1 999) has suggested that researchers should consider

including a 'no choice' option in fbture applications of conjoint methodology in health care.

CBCJA has limitations too. Uniike conventional CJA, which is capable of analyzing

data at the individuai level, data can only be analyzed in the aggregate. This is because it uses

logistic regression analysis, which is based on the odds or likelihood of the dependent

variable, to estimate the attribute level parameters. By their nature, odds can only be

estimated at the group level (Struhl 1994).

Page 36: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

2.9 The Multinomial Logit Choice Model

The most widely used logistic regression technique in choice modeling is multinomial

logistic regression, known as multinomial logit (MNL) (McFadden 1986). Developed £kom

random utility theory, MNL is used for choices arnong more than two alternatives (Struhl

1994).

The theory of MNL has several assumptions: 1) a choice set exists fiom which

individuals make choices; 2) individuals' choices reflect their preferences; 3) for estimation

purposes errors are assumed to be random and independently and identically distnbuted with

a mean of zero and constant variance; 4) the utility of the alternative being evaluated is

denved fiom attributes levels and the importance of each attribute level (Gensch and Recker

1979); and 5) independence fiom helevant alternatives, meaning that the odds of choosing

one alternative over another is constant regardless of the presence of what other alternatives

are present (Louviere and Woodworth 1983).

The theory of MNL is that the probability of choosing alternative (a) fiorn some

choice set A, is equal to the probability that its mean value (Li,) plus its random error (eJ is

Iarger than the mean vaiues and associated errors of each of the otherj alternatives in set A.

This can be described by the following equation (Louviere 1984):

P(a/A)=P(U, +%>&+a> ... >UN+%), foralljinA

Using the aforementioned assumptions of MNL and manipulating the above equation

algebraically, the following equation is neated:

P(a/A) = exp U, / Ci exp Uj, for ail j in A (Louviere 1984)

Thus, the probability of choosing alternative (a) fiom choice set A is equal to the ratio of the

exponential utility of (a) relative to the sum of al1 the exponential utilities in set A.

Page 37: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

28

In order to complete the model, the U's must be specified. MNL models have been

assumed to fit iiiear models. In most applications the utility of a product being evaluated is

represented by the foIlowing equation:

Cr, = -+ X,~P~, -+ x&s +- --- -i- xmPm (McFadden 1986)

Ln the above equation, x's represent the predetennined attriiute levels and Bs represent the

parameter estimates that are inferred tiom the data (McFadden 1986). When sufficient

observations3 are made, the probability that a given option is chosen h m the choice task can

be approximated and the parameter estimates can be inferred. The parameter estimates are

the importance weights for each attribute level. The value of a product profile is then

calculated fiom the parameter estimates.

2. IO Contingent Valuation

Contingent valuation (CV) is the direct measurement of willingness-to-pay (WTP),

using survey methods to elicit stated dollar values for some non-market phenornenon (Le.

change in health outcorne) (O'Brien and Viramontes 1994). Researchers have typically used

CV to determine WTP for healthcare services. However, CV has potentiai limitations which

include: lack of sensitivity to the value associated with process attributes (i-e. route of

administration) (Donaldson and Shackley 1997) and poor discrimination between the similar

alternatives (Donaldson, Shackley and Abdalla 1997). CJA has been proposed as a possible

solution to these limitations.

We decided to use CBCJA to infer WTP for dmgs because choice modeling of MS

patients' preferences for dmg treatments had not been done since the DMMSDs have

become available. Furthemore, no information was available in the literature on the value of

Page 38: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

29

individual attributes. Also, we decided to compare existing WTP for DMMSDs estimated by

CV and by CBCJA to assess WTP.

in a CV study the researcher must decide: 1) who should be asked WTP questions

(current users, pnor users, future users), 2) what should be asked (wiliiigness to pay for a

gain or willingness to accept for a loss) 3) what technique should be used to describe the

health outcome (certain or expected health outcornes), and 4) how the question should be

asked (open ended vs. close ended questions) (O'Brien and Gafni 1996).

The rationaie for our decisions in each of these areas follows. We evaluated WTP for

existing therapies with a user-based approach because survey respondents were MS patients

who were users, prior users and likely friture users of DMMSDs. The user-based approach is

to ask people at the point of consumption what they are wiiling to pay to use the product or

service. We did not use an ex anfe (insurance based) approach, which is to ask a person how

much h&he is willing to pay for insurance coverage, because that approach is used to survey

healthy members of the general public (Drummond et al. 1997).

There are two ways to measure the value that people have for a product or service.

Ask people 1) how much they would need to be cornpensateci if an existing service was

removed (WTA) or 2) how much they would be willing to pay to use a product or service

(WTP). We used a WTP approach for two reasons: 1) we did not want to instill fear in

survey respondents that insurers could decide to no longer cover the DMMSDs and 2) given

that some respondents were DMMSD-naïve, it would have been illogical to provide them

with a WTA scenario.

Observations in MNL are choices which are a fiuiction of the number of choiœ tasks and the number of respondents that complete each choiœ task.

Page 39: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

The treatment being evaluated can be described either according to certain o r to

probabilistic health outcornes. The data collecteci fiom the chcal trials of the DMMSDs,

like ail drugs, indicated a probabilistic description of the MS treatment.

There are two types of question formats used to elicit WTP: open-ended and

closed-ended. With both formats, the respondent is iirst presented with a description of the

product or seMce being evaluated, then the WTP question is asked. Ln an open-ended

question format, respondents are s h p l y asked what is the most they are willing to pay for a

product or seMce (e.g. what is the most money you would spend on the described MS

treatment?). Their responses are taken as theù maximum WTP. This question is difficult for

respondents because people are not used to thinking this way. It can therefore lead to

imprecise responses (Dmrnmond et al. 1997).

Closed-ended formats involve asking respondents a series o f yes or no questions. For

example, would you pay !§ 100 a month to use the described MS drug? If the response was

yes, the respondent would be asked, would you pay $200 a month; if the response was no,

the respondent would be asked would you pay $50 a month; and so on. Although close-

ended questions introduce a starting point bias, they are easier for respondents to answer and

they may be more precise than open-ended methods (Drummond et al. 1997). A close-ended

WTP question was used in this study.

2.11 Summary

In surnmary, MS is a neurological disease of varying clinical presentation. Until

recently, the only drugs available to MS patients have been to manage their syrnptoms. Now

four drugs are on the market which modw the underlying disease course. These drugs have

modest efficacy, have bothermme side effects, require injection and are expensive. Aithough,

Page 40: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

31

the DMMSDs are costly, so is MS because it is a disabling disease. Economic evaluations

have been done on some of the DMMSDs, and their results have not been favourable,

Low DMMSD utilization seems consistent with the unfavourable results fiom the

economic evaluations of the DMMSDs. However, as most MS patients do not have to pay

full price for the DMMSDs, the precise reasons why relatively few MS patients are deciding

to try a DMMSDs are unknown.

Patient preference research is gaining popularity because of patients' increasïng desire

to be active in their medicai decision making. The titerature shows that patients7 treatrnent

preferences for therapy are tikely based on characteristics of treatment called "attributes".

Furthemore, it suggests that disease seventy and experience with healthcare sewice a f k t

patients' preferences for treatment. However, patient preferences for DMMSDs have never

been assessed.

Choice-based conjoint analysis (CBCJA) is a technique capable of inferring consumer

preferences for treatment attributes. It can also be used to infer what respondents are willing

to pay for existing and hypothetical products. CBCJA has been put forth as a better method

of assessing the value that respondents assign to process attributes (Le. mode of dmg

administration) than direct evaiuation o € W (contingent vaiuation); however this has not

been empincally tested in healthcare.

Page 41: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

CHAPTER 3

METHODS

3.1 Study Design

This survey of MS patients utililed a two-part, self-adminiaered questionnaire. The

first part elicited patient preferences for shulated DMMSDs, while the second obtained

dernographie, medical and hancial information used to investigate factors &ecting

preferences. The sample was dichotomized in two diierent ways. Two disease severity

groups were created based on level of disability as detennuied fkom the Expanded Disability

Status Scale (EDSS). The second grouping of patients was based on prior experience with

DMMSD therapy. Groups were compared on the basis of their treatment preferences.

3.2 Subjects

Study participants consisted of a convenience sample of 100 patients fiom the MS

clinics at St. Michael's Hospital, Toronto, Ontario and McMaster Health Sciences Center,

Hamilton, Ontario (sample size determination is described in section 3.5). Together, the MS

clinics have approximately 4,800 registered patients, representing approximately one third of

the MS population in Ontario.

Eligibility criteria for this study included: 1) registration with either the St. Michael's

or McMaster's MS chic, 2) English speaking and reading, 3) greater than 1 8 years of age,

4) EDSS 5 6.5 and 5) history of at least one relapse in the past two years.

Prior to initiation of patient recruitment, ethics approval was received fkom the

University of Toronto, St. Michael's Hospital and McMaster University Faculty of Health

Sciences (Appendices B and C).

Page 42: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

3 3

3.3 Recruitment Procdures

Study participants were recruited fkom the MS clinics fiom midoMarch to the end of

May 1999. Upon completion of a visit with a patient who met the eligibility criteria, the c h i c

neurologist asked if he/she would be willing to consider participating in a study on patient

preferences for treatments of multiple sclerosis. At St. Michael's Hospital, interested patients

were introduced to a research assistant in the waiting room, while at the McMaster Health

Sciences Center, interested patients were introduced to a chic nurse. The research assistant

or nurse gave the patient an information sheet (Appendix D), addressed questions and

obtained written consent (Appendix E) fiom patients who wished to participate.

Participation in any phase of the study precluded participation in subsequent phases.

Thus, prior to recruitment, the research assistant checked the names of potential participants

against a master List of names of patients who had participated in eariier phases of the study.

As al1 preliminary work was done at the St. Michael's Hospital MS chic, this check was not

necessary at the McMaster site.

3.4 Development of the Treatment Preference Questionnaire

The treatment preference questionnaire consisted of 15 choice tasks. In each task, two

drug treatment options and a 'no new dmg' option were described in terrns of 7 attributes.

Respondents were asked to indicate their choice.

Development of the questionnaire proceeded in six phases discussed below: 1)

selection of the attributes to be used in describing each treatment, 2) determination of the

levels used for each attribute, 3) determination of the number of treatment alternatives per

choice task, 4) determination of the number of choice tasks in the questionnaire, 5) review of

sample drug profiles by an expert panel, and 6) pilot testing of the questionnaire.

Page 43: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

3.4. l Atiribute Detennirtution

A literature search did not provide infiormation on what aspects of treatment MS

patients find important. With this purpose in rnind, two focus groups were held in November

1998 with multiple sclerosis patients at St. Michael's hospital. One group consisted of

current users o f the new disease-modüjing agents and the other had never used the dnigs.

The characteristics of the focus group participants are surnmarized in Tables 1 and 2

respectively. The primary objective of these sessions was to determine, in patient language,

what attributes of drug therapy patients think about when they consider new MS dmgs. The

full focus group report (objectives, methods, discussion and conclusion) can be found in

Appendix F.

The nine attributes identified by focus group members were: 1) hequency of relapses,

2) severity of relapses, 3) effect on energy kvel, 4) eEect on progression of disease, 5) cost,

6) skin reactions as a side effkct, 7) flu-like symptoms as side effect, 8) dosage f o m and 9)

fiequency of dosing.

Page 44: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

Table 1. Demographic Characteristics of Focus Group Participants

Sex Males

Femaies

Highest Level of Education Completed High school

Some college or university College di plorna / University degree

1 Graduate degree 1 3

Table 2. MS-Related Medicai History of Focus Group Participants

Years since diagnosis <5

5-8 >8 -

Number of relapses in last 2 yurs <2 - >2

Most recent relapse Less than 6 months ago

Greater than 6 months ago

More Limited This Year Compared to Last Year Yes

No Don't Know

Page 45: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

36

Green and Srinïvasan (1990) have recommended using a maximum of six attributes to

describe a product in a fidl profile CJA (where aii attn'butes of a product are presented at

once). To reduce the number of dmg attributes to the most important six, we surveyed 25

c h i c patients on the importance of nine dmg attributes, identified in the focus groups, to

choice of a new drug treatment. Characteristics of these participants are summarized in

Tables 3 and 4.

Table 3. Demographic Characteristics of Participants in the Attribute Ranking Survey

Males Fernales

20-29 30-3 9 4049 > 50

Highest Level of Education Completcd Less than high school

High school Some coUege or university

ColIege diploma l University degree Graduate denree

Page 46: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

Table 4. MS-ReIated Medical History of Participants in the Attribute Ranking Survey

Years since diagnosis* <5 5-8 >8 -

Number o f relapses in h s t 2 yean <2 - >2

Most recent relapse Less than 6 months ago

Greater than 6 months ago More Limited This Year Compared to Last Year

Yes 12 No - 13

*Data missing fiom one respondent.

Participants made a checkmark next t o the six attributes, on a list of nine, that they

found to be most important. After nine of the participants had completed the survey we

realized the check mark system could not account for infiequently selected attributes but that

were very important to the few that selected them. The survey was then changed to ask

patients to rank the attributes fiom I (most important) to 9 (least important). The amended

survey is found in Appendix G. The overd importance of each attribute was detennined by

taking the mean of the rank assigned to each attribute by the 16 subjects who completed the

revised survey. The six most highly ranked attributes, in decreasing order of importance, were:

1) effect on number of relapses, 2) effect on severity of relapses, 3) eflFêct on progression of

MS, 4) effect on energy level, 5) coa, and 6) chance of flu-like syrnptoms. The overall ranking

of each attribute can be found in Table 5, the tiequency distribution in Table 6, and the

attribute importance survey report in Appendk H.

Page 47: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

Table 5 . Importance of Dmg Attributes for Dmg Choice (Results of Attribute Ranking Survey)

Effect on severity of relapses 1.3 1 1 -63 1 .O0

Effect on progression of disease 1 1.75 1 2.25 1 1-25 L 1 1

Effect on energy level 1 2.25 1 2.75 1 1.75 1 1 I

Out of pocket cost 1 3 .O6 1 3 -50 1 2.63

Chance of flu like symptoms 4.3 1 5 .O0 3 -63

Injection site reaction 5 .O0 5.13 4.88

Doses per week 5.50 6.63 4.38

* I = Most important, 10 = Least important

Table 6. Frequency of Attribute Selection in the Six Most important (Results of Attnbute Ranking Survey)

Attributes . ..

IEffect on progression of disease 1 24 1 13 1 11

Effect on severity of relapses

Pmiuency ob Sdectim ... . . . . . .

AU . ricspondeab (n-w

25

Effect on your energy level

Effect on number of relapses

Out of pocket cost

DMMSDNabe (n44)'

14 25

14

24

Chance flu like symptoms

L 1

Doses per week 1 4 1 2 1 2

.' DMMSD Es@erienceâ @=LI)

Il

Il

16

Injection site reaction

14

12

10

7

1

5 l 4

Other (neutralizing antibodies)

9

7

1

5

1 1 O

Page 48: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

39

Subsequent to the importance sumey, the attributes of number and severity of

relapses were combined to form one atîribute. This was in order to describe relapse by only

one attribute. The justification was that they are clinically related (Weinstock-Guttman and

Cohen 1996). Thus, we were able to add dosage form as the sixth attribute.

3 - 4 2 A ttribtrfe Level Detemination

Next, levels for each attribute were created, based on characteristics of marketed

products (AvonexB, BetaseronB, CopaxoneB, and RebifW) and on anticipated

improvements to these products. We did not want to present unredistic drug profiles;

therefore, only attribute levels that were equal to or better than avaiiable products were

included. Levels were deliberately selected to describe a broader range of attribute levels

than represented in currently marketed products. This was done to increase the likelihood

that respondents woutd make tradeoffs between alternatives.

3.4- 3 Num ber of Treatmenî Ahematives per Choice Tmk

It has been shown that the greater the number of alternatives that are presented in

each task the less likely respondents are to consider al1 of the attributes that are presented.

(Abelson and Levi 1985). Although there are currently four DMMSDs on the Canadian

market we thought that presenting four diEerent drug options including an option of 'no

drug' in each choice task would have been too much information for respondents to process.

To reduce the cognitive burden on respondents we decided to use two drug alternatives and

the 'no new drug' option in each task (the least taxing combination of options that allowed a

'no new drug' option and a cornparison of DMMSDs).

Page 49: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

3.4.4 Number of Choice Tasks in the Ques!ionmire

Choice-based conjoint studies with up to 20 choice tasks and no more than six

attributes per option have been undertaken 4 t h no degradation in data quality due to

respondent boredom or fatigue (Johnson and Orme 1996). We conservatively selected 15

choice tasks for the study, kwping in mind the possibility that expert review or pilot testing

might reveal the need to include additiond drug attnbutes.

3.4 5 Expert Review

A sample questionnaire was developed and reviewed by clinical experts in MS,

including two neurologists and two nurses fiom the St, Michael's MS clinic and one

pharmacist, f?om the Ottawa General Hospital, who specialues in MS. The experts were

asked to attempt three choice tasks and determine whether: 1) selected levels within

attributes were sufficiently difEerent fkom each other, 2) any combinations of attribute levels

were unrealistic, and 3) the attribute levels were described in words that patients would

understand-

Although the reviewers seemed to think that the levels were sufficientiy different

from each other, they thought that the combination of 'no efféct on relapses' and 'prevents

any worsening of the disease' was unrealistic. Based on that comment, we decided to

eliminate the 'no effect on relapses' fevel in the drug descriptions. Two of the reviewers

suggested that fiequency of dosing be included as an attribute, as it represents one of the

main differences amongst the currently marketed products. Although patients who

participated in the focus groups and attribute rating survey did not indicate that dosage

fi-equency was particularly important, we included fkquency of dosing as the seventh drug

Page 50: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

41

attnbu t e based on the experts' recomrnendation. Finally, the reviewers identifieci tenns that

might have been confiising for patients and so Mnor word changes were made.

3.4 6 Pilot Test

To assess respondent burden, clarity of instructions and appropriateness of attribute

levels chosen, the survey was pilot tested in 16 RRMS patients (who met the inclusion

criteria of the study) in February 1999 at the M S clinic at St. Mchael's hospital. The

characteristics of pilot test subjects are sumrnarized in Tables 7 and 8.

Table 7. Demographic Characteristics of the Pilot Survey Participants

Males Females

Age (years) 20-29 30-39 40-49

>50 Highest level of education completed

Less than high school High school

Some college or university College dipioma / University degree

Graduate degree

Household Incorne** < $34,999 per year

$3 5,000 to $49,999 per year $50,000 to $74,999 pet year

** 4 respondents omitted this question

Page 51: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

Table 8. MS-Related Medical History of Pilot Survey Participants

Years since diagnosis €4 -

5-8 >8 -

Number of relapses in Iast 2 years* <2 - >2

Most recent relapse Less than 1 month ago

1-5 months ago 6- 12 months ago

Greater than 1 year ago EDSS Scoret

<3 - 3-5 to 6-5

*One respondent ornitted this question tEDSS is the Expanded Disabitity Status Scaie. Possible scores range Erom O (no disability) to 10 (death due to multiple sclerosis).

The pilot test report (objectives, methods and results) c m be found in Appendix 1.

Salient findings were that the once a week capsule regirnen was thought to be too good to be

truc by several respondents and that the 15 choice tasks and three treatment alternatives (two

dmg options and a 'no drug' option) per task were not too burdensome for participants.

Based on these findings. we prohibited the combination of the capsule level and the once a

week level within any drug profile in the actual survey.

3.4 7 Final Design

Of the seven drug attributes used in the rnodel, cost was described by four levels,

relapses by two levels and the other five attributes were described by 3 levels. A complete list

Page 52: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

43

of attributes and their levels, dong with indicator coding for the MNL mode1 is attached

(Appendix J).

Twenty different fiactionai faaorial design8 of 30 dmg profiles were created using

the PROC OPTEX cornmand in SAS (6.12 version) (Kuhfield 1996). This comrnand

randody selects a design and aiters it until the efficiency wmpared to a f ù U factorial design

ceases to improve. Although, the default is set to repeat this efficiency procedure with 10

different designs, we set 20 designs as the default in an effort to increase the Likelihood of

generating a more efficient design. Using the BLOCK comrnand in SAS, the 30 profiles

selected were randomly divided into two groups.

Knowing that the treatment preference questionnaire was cognitively challenging, we

wanted to identiQ participants who did not ration* respond. Three of the survey's 15

choice sets (nurnbers 2, 10, 13 or dnig pairs AB, ST and YZ) were deliberately constmcted

so that one drug alternative clearly dominated the other.' Irrational choices on these three

tasks (according to rules outlined in the data analysis section) resulted in exclusion of the

respondent's entire data set fiom data analysis.

Once the three choice sets with dominated alternatives were established, the

remaining 12 choice sets were selected by the primary investigator in such a way as to: 1)

minirnize the number of times levels on an attribute were identical across pairs and 2) ensure

that the number of times levels on an attribute were identical across pairs was similar across

attributes. The objective was to ensure that respondents would see each level of an attribute

an equal number of times, thus equating the number of tradeoff opportunities for each

' A mcthod of designing stimuli for evaluation by generating a subset or fraction of al1 possible combinations of levels.

Within a choice task an alternative is dominant with respect to another if it is better on at least one attribute and not worse on al1 othcr attributes (Ben-Akiva 1985).

Page 53: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

44

attribute level.

With 15 choice tasks, the respondent is faced with 15 pairs of levels on each attribute.

If on average there are 3 levels per attribute then the probabiiity is 0.33 that the levels will be

identical between the two drug options within one choice task. This works out to 5 identical

pairs per attribute, over the entire questionnaire. In order to select the final set of choice

tasks for the qiiestiomaire, the foilowing method was used- Any set of 15 choice tasks in

which identical levels across attributes occurred less than 4 times or more than 6 times were

excluded. The exception was the cost attribute, which was allowed to have fewer than 4

identical pairs because 4 levels describe it. Once the 15 choice sets were selected, their order

was randomized. A copy of the treatrnent preference questionnaire is attached (Appendix K).

3.5 Sample Size

In a choice-based conjoint study, sample size depends on the number of choice tasks,

the number of levels in the attributes and the number of alternatives per choice task (Orme

1998). In a main-effects choice based conjoint analysisy6 sarnple size can be derived by the

following formula (Orme 1998):

{sam~le sizel(# of choice tasksW alternatives Der task excludine "none"') 2 500 the largest number of levels in any one attribute

With 15 choice tasks, two drug alternatives per task and the Iargest number of levels in

-

Conjoint studies typically estimate main effects. Models Hith interaction te- ofien lead to lowcr predictive validity. That is. the increase in reaiism in the model, obtained by including interactions, is srnafi compared to the deterioration in predictive accuracy causai by including additional parameters (Green and Srinivasan 1990). Additionally, including ail twbway interactions between attribute levels would have quired a four- fold increase in sample size. quiring 10 months of recnritment. For these reasons and for ease of design, a maineffects mode1 was designd

' Here, "none" refers to the 'no new dmg alternative'

Page 54: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

45

any attribute being four, a minimum of 67 respondents were required by the above equation.

In order to exceed the minimum sample size necessary and to allow for unusable surveys, we

stopped data collection when one hundred patients had completed the survey.

The adequacy of this sample size can be judged by the accuracy it provides in

predicting respondents' dmg choices. In this study design, two DMMSDs were presented in

each choice task. Thus, the probability (p) that a participant (who selected a drug) would

select either of the DMMSDs by chance alone was 0.5. Assuming 85 respondents produce

useable data, the standard enor of this proportion is 0.020, yielding a 95% confidence

interval of 0.46 1 to 0.539.' Should the proportion selecting a treatment lie outside of this

range, it can be said that the dïerence is not likely due to chance. The 95% confidence

intervai of +/- 0.039 around a proportion of 0.5. was an acceptable level precision in our

judgement.

3.6 Demographic and MS-Related Medical History Questionnaire

The following information was coliected in the demographic and MS related medicai

history questionnaire: 1) age, 2) sex, 3) year of diagnosis, 4) education, 5) number of

relapses in the 1st 2 years, 6) time since most recent relapse, 7) experience with DMMSDs,

8) reasons for not having tried a DMMSD [ifapplicable], 9) household income, 10) number

of people per household, I 1) willingness-to-pay for existing DMMSDs and 12) disability as

measured by a self-administered version of the Expanded Disability Status Scale (EDSS). A

copy of the demographic and MS-related medical questionnaire is attached (Appendk L).

-- - -

' Standard error m a s calculated using the foiiowing formula: SE = v@)(l-p)/n . where p = probability that the h g option is chosen and n is the number per cell, that is: (85 respondeats)(tS choiœ tasks)(2 DMMSD alternatives) / (4 masimum atuibute levels per attribute)

Page 55: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

3.6-1 P rior DMUSD Experience

Subgroups of participants were created based on responses to questions in the

Demograp hic and MS-Related Medical History Questionnaire. S pecifically, patients who

indicated that they had tried a DMMSD in question 8 were classified as DMMSD-

experienced (dummy coded O), and those who had never tried a DMMSD were classifieci as

DMMSD-naive (dummy coded 1). Participants who indicated in question 9 that they had

received a prescription for a DMMSD tiom their neurologist, but had not gotten it fiiled

were classified as DMMSD-naïve (dummy coded 1).

3.6.2 Disease Severity

Sub-grouping of disease severity was based on scores on a self-adrninistered version

of the EDSS questionnaire. The inter-rater reliability of EDSS scores corn a self-

administered instrument has been estimated at 0.84, using an intraclass correlation coefficient

(SoIari et al. 1993). Participants' responses on the questionnaire were converted to a

composite EDSS score by the director of the MS clinic at St. Michael's Hospital.

Participants with an EDSS score of less than four were classined as rnildly disabled (dummy

coded 0) and those with EDSS score of 4 to 6.5 were classified as rnoderately disabled

(dummy coded 1). This division was selected for two reasons: 1) because it created create

similady sized groups and 2) because the EDSS score of 4.0 represents the beginning of

significant disability.

3.6.3 WiIIingness-tu-Pay for Ejtisting DMMSDs &y Contingent Vuhatiotorl

In question 1 1 of the Demographic and Medical Questionnaire (Appendix L),

participants were provided with a generic description (general description of the typical

Page 56: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

47

DMMSD on the market) of the existing DMMSDs according to the same attributes used in

the treatment preference questionnaire. They were asked to consider the hypothetid

scenario in which their doctor recomrnended that they try a DMMSD. A dose-ended

technique was used to estimate WTP, whereby participants were asked to indicate if they

would be willing to pay the amount presented in each of several ranges of monthly cost by

placing a checkmark in the 'yes' or 'no' box next to each range, the last of which had no

upper limit.

The rnidpoint of the highest range that each participant selected was taken as his/her

WTP. #en the last response option (Le. greater than $1,000 per month) was chosen, the

participant's WTP was taken to be $1,000 per month- Each participant's WTP was then

divided by the midpoint of hidher household income range to get the percent of household

income he/she was willing t o spend on a DMMSD. Seventy-five thousand douars was taken

as the househoId income o f participants who indicated that their total household income was

greater than $75,000 per year. Mean WW was calculated by summing each respondent's

WTP per month and dividing the sum by the number of respondents.

3.7 Data Collection and Management Procedures

Upon receiving signed consent, the research assistant or nurse gave participants two

paper and pencil questiomaires to complete (choice tasks and demographic/MS-related

medical history), dong with a definition sheet (Appendix M) explaining the meaning of each

attribute. Participants had the option of completing the questiomaire in a private room at the

MS clinic or taking the questionnaire home to complete. Participants who took the survey

home scheduled a phone appointment with the research assistant or nurse. Participants were

instructed to complete the survey prior to the phone call. During the phone appointment, the

Page 57: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

18

nurse or research assistant obtaïned participants' responses to the previously completed

questionnaire.

The primary investigator entered the data collected fiom both parts of the

questionnaire into an ExcelB spreadsheet and the research assistant double-checked the

entered data against the completed questionnaires.

3.8 Data Analysis

3.8-1 Screen for Rationafity of Repmes

Before data were analyzed, unretiable respondents were identified and excluded. This

was done by examining responses on the 3 choice sets in which one alternative was designed

to be dominant. If participants selected the cleariy inferior dnig more than once (as might

occur by chance), then it was possible that they either rnisunderstood the task or did not take

the survey senously. Their data were excluded from analysis.

3.8.2 Sfaristical M d e l

Data analyses were performed with LIMDEP Execution Trace (Version 7.0) (Greene

1995). Treatment choice (arnong two dmgs and a no dnig option) was modeled as a

multinomial togit tùnction of seven drug attributes and their interactions with disease severity

and DMMSD experience. Parameters of the logit model were estimated using maximum

Iikelihood estimation procedures. These main effects and interaction effects were expressed

by parameter estimates, which were considered signifiant at p<0.05. The parameter estimate

associated with each attribute level represents the attribute level's influence on treatment

choice relative to the base level on the attibute.

Page 58: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

Choice of a drug option was modeled diierently from choice of a 'no new drug'

option. Specifically, the utility associated with a drug option was decomposed as a hear

fùnction of its attribute levels, specified as:

Udmg = Z G B i + e

for the main effects model.

where;

Udmg = the utility of a drug pronle Bi = the part-worth associated with attribute level i xi = the attribute level of the drug e = a random error tenn

The 'no new drug' option was modeled without attributes or levels. Its utility was based on a

single parameter estimate (&,,,A according to the equation below:

- Unodrug-Bnodnig+e

To test the hypotheses that disease severity and expenence with DMMSDs influence

treatment choice through interactions with drug attributes, the utilities of the drug

descriptions were modeled in expanded models as a ninction of these variables, as follows:

Udmg = X x i B i + CXiPiaiss+ e

and

Udmg = % P i + b i p i d m w u d + e

where;

Udmg = the utility of a drug profile

B; cdss = the interaction effect of disease severity with an attribute level

pi dm&= the interaction effect of DMMSD-experience with an attribute level

Page 59: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

50

The chi square statistic was used to test whether the main effects model was a

significantly better predictor patient choice than a base model (where all parameter estimates

equal zero). For each drug attribute level, the parameter estimate and t statistic reflect the

iduence of the attribute level on drug choice. The larger the t statistic for an attribute level,

the greater its influence on drug choice. Therefore, the order of importance of attributes was

determined by the t statistic of highest magnitude for that attribute.

For the purposes of inferring WTP, cost was modeled as a c o n ~ u o u s variable.

However, cost was also modeled as a categorical variable to determine ifhaving no cost was

actually an incentive to try a DMMSD (i-e. zero cost was dummy coded O and cost greater

than zero, was dummy coded 1).

3.8.3 Eflect of Disease Severiiy and D M S D Experience Interaclions

The chi square statistic was also used to test whether the expanded models were

statistically different tiom the main effects model. In logistic regression, the chi square

statistic is determined by multiplying the difference between log-likelihood statistics of two

models (Le. main effects model and expanded model) by two (Chow 1983). Although an R~

statistic can be computed with logistic regression, it is an inappropriate measure of explained

variance because it was designed for linear models (Gensch and Recker 1979).

Unfortunately, in logistic regression there is no clear way to determine how much of the

variance is explained by a model.

To test if the expanded models (with the disease severity and DMMSD experience

interaction terms) explained more variance than the main effects only model, the log-

likelihood statistics for each model were compared statistically with the chi square statistic.

Once the models were compared, the parameter estimates for the interaction ternis were

Page 60: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

51

examined on each attribute level with the purpose of determining if disease severity andor

DMMSD experience interacted with the effect of specific drug attributes, on drug choice.

3.8.4 6Iferred Wiffingness to Pay for hprovements in Atcribute Levefs and for Existing Dmgs

In order to examine the tradeoff between cost and Unprovernent in an attribute level,

the difference between parameter estimates of levels of an attribute are divided by the

difference in parameter estimates of levels of the cost attrîbute. To infer WTP for

improvements in each attribute, the dflerences in parameter estimates between ievels on the

cost attribute must be the same. The only way to accomplish this is to assume that the

influence of cost is linear across its levels, and to model cost as a continuous variable, Thus,

when differences in parameter estimates of two levels of a drug attribute are divided by the

parameter estimate of cost, WTP for the changes in the dmg attribute are inferred.

From the conjoint analysis it was also possible to infer how much patients were

willing to pay for existing therapies. This was done by creating two hypothetical DMMSDs.

One consisted of a11 the worst attribute levels across existing products (Le. a drug that had no

effect on progression of disease, reduced energy level for the first few months of therapy,

reduced relapse fi-equency and severity by one-third, had a 60Y0 chance of causing flu-like-

syrnptoms and was administered daily with a injection that causes a skin reaction) and the

other consisting of the best attniute levels across existing products (Le. a drug that slowed

the progression of disease, had no effect on energy level, reduced relapse tiequency by one-

third and relapse severity, had no chance of causing flu-lie-symptoms and was administered

one a week by injection that did not cause a skin reaction). We assumed that each existing

DMMSD lay somewhere between these two extremes, and that patients would be willing to

pay $O for the worst case DhIIMSD. Respondents' WTP for the best case DMMSD

Page 61: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

52

(consisting of the best levels of exïsting DMMSDs) was infierreci by summing the pararneter

estirnates for each of these levels and dividing the sum by the parameter estimate of the cost

attribute (modelled as a continuous variable). The midpoint of these extremes was taken as

the inferred WTP for existing DMMSDs.

3.9 Validity of the MNL Choice Mode1

The predictive validity of the MNL choice model was determined two ways: 1) the

percentage of the choice tasks in which the no drug option was chosen was compared with

the probability predicted by the model that the no drug option would be chosen, and 2) the

percentage of participants' DMMSD choices that were correctly predicted by the model.

Specifically, the utilities for the DMMSDs in each choice task were calculated as the surn of

the pararneter estimates for their attribute levels (according to the main effects model). The

DMMSD alternative with the higher inferred utility in each choice set was the predicted

choice. The percent of correct choices predicted by the model was calculated by dividing the

number of times respondents selected the predicted DMMSD by the sum of ail the times a

DMMSD was chosen.

The face validity of the MNL choice model was assessed by examining the signs and

the magnitudes of the pararneter estimates for each drug attribute level. For example, a

positive pararneter estimate indicates that the level of the attribute is preferred to the level

dumrny coded zero and a negative parameter estimate indicates that the attribute ievel is less

preferred than the one dumrny coded zero. Given that the worst level on each attribute was

dummy coded zero, the parameter estimates for more preferred levels on a drug attribute

should have had a larger positive value than the parameter estimates of less preferred levels.

Specifically, attribute levels representing: improved energy, less deciine in ability to do daily

Page 62: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

53

activities, greater effect on relapses, orally administered dmgs, less fiequent dosing intervals,

and Iower cost, should have had larger positive parameter estimates.

Page 63: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

C-R 4

RESULTS

4.1 Useable Sutveys

One hundred and Gtty three patients were approached to participate in the survey:

129 at St. Michael's Hospital and 24 at McMaster Health Science Center. One hundred and

nineteen (78%) agreed to participate: 95 at S t Michael's Hospital and 24 at McMaster

Health Sciences Center. Sixty-six participants completed the survey at the clinic, while 5 1

took the survey home to complete. Responses were obtained fiom 38 (75%) of those who

took the survey home.

Of the 119 patients who signed the consent form, 100 provided useable data (65% of

patients who were approached). Of the 19 patients who did not provide useable data, 15 had

not completed the survey: 5 could not be reached by phone, 5 said they would fax or mail it

back and did not, 2 were too busy to complete the questionnaire, 2 lost it and 1 participant

started the questionnaire but did not finish because she felt that she was "too poor a

candidate to participate". Although 4 of the 19 patients completed the questionnaire, 3 were

excluded because they did not have at least one relapse in the past 2 years (a requirement for

inclusion in the study) and 1 was excluded because his drug choices were made with the

assumption that insurance covered 80% of the dmg costs (inconsistent with the survey

instructions).

Fifteen of the 100 respondents selected a dominated product once out of three times;

one participant seiected a dominated product twice out of three times and no participants

Page 64: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

55

selected a dominated product ail three t ime~ .~ Therefore, according to our pre-established

critena for excludimg irrational responses, one data set was exchded and 99 data sets were

used for anaiysis.

4.2 Description of the Sample

The demographic characteristics and MS-related medical history of the survey

participants are summarized in Tables 9 and 10 respectively.

Tabie 9. Demographic Characteristics of Survey Participants

Sex Males

Females

Highest levcl of education completcd Less than high school

High school Some college or university

College diploma /University degree Graduate degree

Household Income <% 10,000 per year

% 10,000 to $19,999 per year $20,000 to $34,999 per year $35,000 to %49,999 per year $50,000 to $74,999 per year

>$75,000 Der vear

Al1 15 respondents who selected the dominated alternative, did so in the second choiœ task in the questionnaire

Page 65: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

Table 1 0. MS-Related Medical History of Survey Participants

~ e a r s rince diagnosis* <4 years

4-8 years >9 years

Number of relapses in Iast 2 yurs <2 - >2

Most recent relapse Less than 1 month ago

1-5 months ago 6- 12 months ago

Greater than 1 year ago DMMSD Exptrience

Yes No

EDSS Score* <4

*Data missing fiom four respondents tEDSS is thi~xpanded ~ i i b ü i t ~ Status Scale. Possible scores range nom O (no disability) to IO (death due to multiple scIerosis).

Participants with DMMSD experience are compared to those without in Tables 11 and

Page 66: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

Table 1 1. Demographic Characteristics of DMMSD vs. non-DMMSD-Experienced Participants

Sex Males

Females

20-29 3 0-3 9 40-49

>50 Highest level of education completed

Less than high school High school

Some coliege or university College diploma /University degree

Graduate degree Household Income

<$IO, 000 per year $1 0,000 to !§ 19,999 per year %20,000 to $34,999 per year $35,000 to $49,999 per year $50,000 to $74,999 per year

>$75,000 per year

Although demographic characteristics are similar, dserences are noted with respect

to MS-related medical history. Specifically, DMMSD-expetienced participants had had a

mean of 3 -53 f 2.38 relapses in the last two years compared to 2.56 c 1.9 1 in the DMMSD-

naive group (p<O.OS). DMMSD-experienced participants were also more disabled, as

deterrnined by EDSS score. The average EDSS score of DMMSD-naïve participants was

3.13 compared to that of 4.5 1 for DMMSD-experienced participants (pc0.00 1).

Page 67: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

Table 12. MS-Related Medicd Hiaory of DMMSD vs. non-DMMSD Experienced Participants

Years since diagnosis* <4 years

4-8 years 9- 12 years >12 years

Number of relapses in lait 2 years <2 - >2

Most recent relapse Less than 1 month ago

1-5 months ago 6- 12 months ago

Greater than 1 year ago EDSS Scoret

<4

*Data rnissing for 4respondents tEDSS is the Expanded Disabiiity Status Scale. Possible scores range O (no disability) to 10 (death due to multiple sclerosis).

The rnost comrnon reasons given by the 5 1 DMMSD-naive participants for not

having tned a DMMSD are summarized in Table 13.

Page 68: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

Table 13. Most Common Reasons that DMMSD Naïve Participants Had Not Tried Therapy

Other reasons participants stated for not h a h g tried a DMMSD included: fear of flu-

Currently deciding whether or not to try a DMMSD Have 'rnild disease' and treatment is not yet warranted Has a prescription for a DMMSD but has not tilled it yet Doctor has not recommended a DMMSD Has not heard of the dnigs Financiai problems 1s pregnant, or planning to, becorne pregnant

like symptoms, concerns about the eff'ects of the dmgs on depression, not being weU enough

14 11 7 6 4 3

2 A

infomed about the dmgs, interaction with concomitant medication, and fear of needles. in

addition, one person said that she was "stiil able to fbnction and accepted her present

condition".

For those who had DMMSD experience, the distribution of specific product

experience is summarized in Table 14. BetaseronB, the product on the market for the longest

time, was most cornrnonly used.

Table 14. DMMSDs Used by Survey Participants

10 Oniy the 5 1 DMMSD-naîve respondents were asked to respond to this question. The 47 tesponses included in this table represent responses that were presented by at 1- 2 respondents. The remaining responses are listed in the paragraph following table.

Thirty-sis participants had tried only one DMMSD. 9 had vied two dinerent DMMSDs and thme had tri& 3 Merent DMMSDs,

Page 69: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

Characteristics of participants with mild (EDSS 0-3.5) and moderate (EDSS 4-6.5)

disease are compared in Tables 15 and 16. Those with moderate MS tended to be older, and

to have lowcr annual household income than those with mild MS, Aiso those with moderate

MS generally had had MS for a longer period of tirne and had had more experience with the

DMMSDs.

Table 15. Demographic Characteristics of Participants with Mild vs. Moderate MS

Sex Males

Females Age (years)

20-29 3 0-3 9 4049 >50

Highest levei of education completcd Less than high school

High school Some college or university

College diploma /University degree Graduate degree

Household Incorne <$IO, 000 per year

$10,000 to $ 19,999 per year $20,000 to $34,999 per year $35,000 to $49,999 per year $50,000 to $74,999 per year

1 >%75.000 oer vear

QEDSS score of 4 to 6.5

Page 70: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

Table 16. MS-Related Medical History of Participants with Mild vs. Moderate MS

Years since diagnosis * <4 years

4-8 y m s 9- 12 years > 12 years

Num ber o f relapses in Iast 2 years

<2 - >2

Most recent relapse < 1 month ago

1-5 months ago 6- 12 months ago

> 1 year ago DMMSD exptrience

Experience

* Datz missing for 4 respondents (2 in each group) OEDSS score of O to 3 -5 +EDSS score of 4 to 6.5

4.3 Lexicograp hic Decisions

Fifty-six participants made lexicographic treatment choices. Specifically, 53

participants consistently chose the drug with the best level on one attribute, in other words

used a lexicographic choice rule: 17 participants always selected the drug with the lowest out

of pocket cost, 10 always selected the drug with the highest level on relapses, 8 always

selected the capsule form, 7 always chose the lead progression of disease, 4 always selected

dmgs with no flu-like symptoms, 4 aiways selected the drug with the lowest dosing

frequency, and 3 always chose the least reduction in energy level. Also, 3 participants aiways

selected the 'no new drug' option.

Page 71: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

4.4 Multinomial Logit Anilysis of Choict Tuks

The sample consisted of 1477 observations [(99 participants x 15 choices) - 8

missing responses]. Respondent choices were analyzed with a non-nested logit anaiysis. In

cther words, a one step decision model which considered the choice to be among Drug A,

Drug B or No new drug was used. A nested-mode1 (Le. a two step decision model which

assumes the first choice is between new dnig therapy and no new drug therapy) could have

been used. The nested model was tested but rejected because the relative magnitude of

parameter estimates for the levels within the disease progression and drug form attributes

were coonterintuitive.

4.4. i Drug Aitribute Impor~ance

Parameter estimates for attribute levels are presented in Table 17. Changes in at least

one level on each attribute significantly dected dmg choice. Based on the largest t statistic,

changes within an attribute dEered in importance. Ln decreasing order, it was change I=n:

cost, dosage form, chance of flu-like symptoms, dose frequency, effect on progression of

disease, effect on energy level and effect on relapses.

The parameter estirnate for 'no cost each month' was statistically signincant, its

positive sign indicating that overall, a DMMSD that had no cost was preferred by

participants. The negative sign in front of the parameter estimate for cost when modeled as a

continuous variable means that the greater the cost the less likely participants were to select a

DMMSD. The negative sign in front ofthe parameter estimate for the 'no new dmg' option

indicates that it was not preferred relative to DMMSD options.

Page 72: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

Table 17. Results of Maximum Likelihood Estimation: Main Effects Only Mode1

NoNewDnijg N/A -2.84 1 0.24 - 1 1 -96 <O,OO 1 Cost As a continuous variable -0.00 15 0.00 -9.2 1 <O.OO 1

No cost each month 0.363 0.12 3 .O2 ~0.00 1 Some cost each month -

Dmg Form Capsule 1 .O86 O. 13 8.52 <O,OO 1 Injection without skin 0.377 0.1 1 3 -42 cO.00 1 reaction Iniection with skin reaction - . -

Flu Like No chance 0-742 0.1 1 6.57 CO-00 1 Symptoms 30a! chance 0.227 O. 10 2.27 0.023

60 % chance - DW3 Once each week 0.557 0.1 1 4.87 4-00 1

Frequency Every other day O. 186 O. 12 1.55 O. 121 Every day -

Progression of Prevents any decline in daily Disease activities 0.498 0- 14 3 -64 <O,OO 1

Slows declime in daily activities 0.082 O. 12 0.68 0.496 No effect on daily activities -

-

Energy Level Improves Energy Level 0.454 O. 13 3 -56 c0.00 1 No effect on energy Ievel 0.075 O. 12 0.625 0.529 Reduces energy level for the - first few months of thera~v

d

Relapses Less severe and 2/3 less 0.25 1 0.08 2.98 0.003 often Less severe and 1/3 less - often

Log-likelihood statistic main effects model (LLM) I -1436 Log-likelihood statistic base mode1 @LB) if al1 f3=0 - 1 623 I Chi-squared test statistic J = 2* (LLM - LLB) 374 R* = 1 - (LLM/LLB) 0.1 1

4.42 infience of Disease Severity and DMMSD-Fxperience on Putiemt Preferences

The effects of DMMSD experience and disease severity on the parameter

estimates for the attribute levels are presented in Tables 18 and 19 respectively. In

those tables, 'main eRects7 are represented by the parameter estimates of the attribute levels

Page 73: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

65

for the DMMSD-experienced participants and participants with mild MS respectively (each

coded O). The 'interaction effects' refer to the effects of the variable levels, DMMSD-naïve

and moderate MS (each coded l), on attribute level importance. Thus, when the parameter

estimate for each interaction effect is added to the parameter estimate for an attribute main

effect, the part-worths of attribute levels for DMMSD-naive participants and participants

with moderate disease are obtained.

Page 74: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

TabIe I S. Results of Maximum Likeiihood Estimation: Mode1 with Main Eff' and DlviMSD Experience Interactions

-

L C V ~ fb -3.276' (0.333) C o s No con. each month 0.492 (O. 175)

~~ E F F E ~ '

1 Somc cos I -

-

-

-

AS a conànuaus variable I 0.00 1. (0.000) Dnig Fonn Capsule L. LU' (O. 153)

I

-

I

1 N I E R -

ACflON

EFFECTS

Injection with slan --on Ru L i k 1 No chance

0.2 L7 (O. L37) 0.752. (O. L63)

30% chance 60 % c h y i u

0.695. (O. LS6) 0.224 (O. L70)

Dnig Regimea

Once uch wcck Evcryothcrday

1 L e s sevtrt and l f i Iess oftcn 1 No Dmq 1 Level frrt 1 0.69 1 (0.4861

-0,026 (0,172)

0.56j0(0. 153)

O. 1 L L(0.1761

Slows da& in d d y activiaes D& 1 Na &a on dlily ulivirrri

L e s scvect and V3 Icss often I - 1 3umber of Observations (cornpleted ehoicc tasks) 1 1477

Energy Levtl

fmprovts Energ Icvcl

No &cct on cncrgy ltvel Rcduccs encm I d for the fim rnonths of thenpv

Relapses

-0.273 (0.242) - -0.00 lb (0.000) 4.043 (0.359) 4.140 (0.133)

0.02s (0.130) 0.029 (0.196) - -0.296 (0.133) 4.082 (0.1U)

-0.089 (0.179) 0.171, (0.74L)

4.135 (0.16 1) 4.043 (0.244) -

=On l No cost each monrh Somc cost & a cnnllnuaus variabie

Lcss were and 23 lcss oftcn 1 0.40%' (O. 120)

Relapses 1 L a s wen and 213 Iess ohen 1 -0.293" (0. L70)

D m j F o m

Ru Like Syrnpcoms

Capsuie [njection without skin ruction [njecüon with skin ruction Nockmcz 30% c h m a 60 % chance

l Once =ch w c k Regmen Every other &y

Everv dav Progression

of D iseve

trier,- Level

Prevcncs riny dccline in daiiy acrivities Slows dcclinc in M y rictivities No 2 f f ~ t on dâiiy activities

Improves energy Iwcl No effect on encrgy levcl Reduçcs energy teveI for the lim rnonths of chenpv

Page 75: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

Table 19. Results of Maximum Likefihood Estimation: Modei with Main Effects and Disease Severity Interactions

No New Dmg Cost

Fonn Drug

Drug Rcgimcn

Energy LeveL

Relapses

h l 61CC No cos cach month Some C m As a conunwu vYiabie

injecrion withouc skin ceaction iajcctioa with skin rua ïon No chaau 30% chance 60 0% chana Once each week Evcry other âay Evcry day Ptcvents any decLine in daiiy 3Cfivitics Slows daline in M y 3Ctivitics No &àt on d d y 3criMtits Improves Energy Ltvel No eEect on energy levei Rcduces energy lcvcl for rhe fïm mon& of thenpy Less mre and 23 Iess oftcn

mil- A C n O N EFFECTS

Cost

Drug Farm

Flu Like Symptoms

Progression of D tscase

Relapses

&el frrc No con a c h month Somc cos As 3 concinuous variable Capsule Injcnion without skin rucuon Injection wirh skin rua ïon No chana 30% chance 60 % chancc

Once =ch \vcck Evcry othcr day Every &y No &cn on daiiy activiues Slows declinc in dYIy activitits Prcvcnts any declinc in dady activitics iinpmvcs énergy lmel No c f k t on energ lcvel Reduces energy kvtl for the fks months of rhenpy Lcss sevcre and 2 3 less oftcn Less and LE Iess ofien

'lumber cf Obsemitions (completcd choicc wkr) Log-likelihood sratistic main effecu mode1 (LL,M) Log-likeiihood nïtistic expÿnded modcl (LLE) Chi-squarcd test stïtirtic = 2' (LLE - LLAM) di= 14

' Mild MS codeci O and modentc MS mdai L

Page 76: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

Each of the expanded models (with disease severity and DMMSD experience

interaction terrns respectively) were better predictors of dnig choice than the reduced model

(main effects only), as is evident fiom the chi square statistics (p<O.OOl).

The specific interactions which were significant were relapse with disease severity

and with DMMSD experience, cost with DMMSD experience and energy level with disease

severity. The improvement in relapse fiequency and severity tiom one-third reduction to

two-thirds was more important to participants with DMMSD experience and participants

with mild MS than it was to those without DMMSD experience and those with moderate

MS. Cost was a bigger factor in the decision making of DMMSD-naïve participants than of

those with DMMSD experience. Participants with mild MS viewed a drug that had no effect

on energy level to be a significant improvement over one that reduced energy for the first few

months of therapy.

The 'no new dnig' option was selected 463 times out of the 1477 choices (3 1.3%).

As expected, the participants who had DMMSD experience selected a dmg therapy more

frequently than DMMSD-naive participants (73 -4% and 64.3%, %L13 -80, d e l , p<O.000 1).

Although the results of the expanded model showed a trend towards a similar finding, the

size of the sample was not sufficient to detect a statistically significant difference, in influence

of the 'no new drug' option on dnig choice, between DMMSD-experience groups.

4.5 WTP for Improvements in Existing Therapies

The amount participants were willing to pay for improvements in existing therapies was

inferred fiom the CBCJA and is sumarized in Table 20.

Page 77: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

Table 20. Inferred Willingness-to-Pay (WTP) for Improvements in Existing Therapies

Drug Form Injection with a skin reaction

Capsule

Injection without a sicin reaction

l Chance of Hu-like-Symptoms 60% chance No chance

30% chance Dosage Regimen

Every day Once each week 371 Every other day 124

Disease Progression No effect Prevents 332

Slows 55 - - - - - - -- - -

Energy Reduces energy level for the Irnproves energy level 302 fist few months of therapy

No effect on energy 49 level

I Relapses Less severe and one third less often

Less severe and two thirds less ofien

Participants were wiiiing to pay the most for improvements in dosage fonn foUowed

by improvements in chance of flu-iike-symptoms, dosage regimen, effectiveness against

progression of disease, energy level and relapse severity and fkequency.

Page 78: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

4.6 Willingnws-To-Pay for Eristing DMMSDs

In question 1 1 of the Demographic and Medical Questionnaire respondents indicated

how much they were willing to pay to use an existing DMMSD. A fiequency distnibution of

their responses is provided in Table 2 1.

Table 2 1 . Frequency Distribution of W-Uingness-to-Pay (WTP) for an Existing DMMSD

Participants were wiliing to pay a mean of $227 per month to use a typical DMMSD.

Patients stated WTP and WTP as a percentage of their household income is presented in

Table 22.

Table 22. Mean Wiiiingness-to-Pay for an Existing DMMSD

I I - --

DMMSD $269 ($195 - $343) 7 - 8 9 ! Ex~erienced

DMMSD I $187 ($139 - $235) I 7.74% Naïve

,

Moderate MS EDSS 4-6.5)

$2 13 ($148 - $278) 7.12%

Page 79: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

70

WTP for existing DMMSDs infemd by CBCJA ranged îkom a low o f $0 for the

hypothetical DMMSD composed of the worst attn'bute levels on exkting DMMSDs to a high

of $1,193 per month for the hypothetical DMMSD composed of the best levels on existing

dmgs. Respondents' WTP for the average existing DMMSD was estimateci fiom the

midpoint of this range, $597 per month.

4.7 Summary o f Results

The decreasing order of attribute importance was: cost, drug form, chance of nu-like

symptoms, dose fiequency, effect on progression of disease, effect on energy level and effect

on relapses. The expanded models including disease severity or DMMSD experience

interactions were better predicton of dnrg choice than the drug attribute main effects model.

Specific significant interactions were noted between DMMSD experience and the cost and

relapse attributes, and between disease severity and the relapse and energy attributes. The

inferred WTP for hprovements in an aîtribute level ranged corn a Iow of $49 per month for

energy level to a high of $723 per month for dosage forrn. WTP for existing DMMSDs was

estimated to be $227 + $44 per month by contingent vduation and approximately $597 per

month by choice-based conjoint analysis.

Page 80: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

DISCUSSION AND CONCLUSIONS

5.1 Sample

The patients who participated in the survey were predominantly young and femaie,

which is consistent with existing iiterature on the demographics of MS patients (Kesseiring

1998). Furthemore, they were a highly educated group, with 80% having some coiiege or

university education. Haif of the patients had DMMSD experience at some point in their

disease course, which exceeds the projected rate of DMMSD utilization in Canada for 1999

( M S Canada, 1999).

5.2 Evaluation of Survey Methods

The overaîi participation and response rates were high; 78% of patients who were

told about the study signed the consent form and 87% of these patients completed the study.

The telephone inteMew procedure worked weii; data were coiiected fiom 75% of those who

took the survey home to complete. These statistics indicate that MS ciinic patients are

interested in being surveyed about their medicai decision making.

Ody one participant's responses did not meet Our criterion for rationality. This

suggests that a wo-dmg choice task that includes seven drug attributes may not exceed most

participants' cognitive capacity. The fact that 15 participants made an irrational drug

selection in the second choice task of the survey may indicate that participants were

overwhelmed with information at the start of the survey.

People tend to make lexicographic decisions to simplQ tasks when too much

information is presented at once (Abelson and Levi 1985). The fact that 56 out of 99

respondents exhibited lexicographic choice rnay refute the conclusion Eom the rationality

Page 81: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

check or it may sirnply be that these participants always base their drug choices on a single

important attribute.

A recomrnendation for fùhtre work is to repeat this study with fewer attributes to

determine if lexicographic patterns of decision-making persist. Another recommendation is to

use a warm-up task to prepare respondents for the actual choice tasks

5.3 Drug Attribute Importance

The findings in this study regarding the relative importance of drug attributes are

supported by a survey of 594 DMMSD-expenenced patients fiom a North American MS

registry (Vollmer and Hadjimichaels 1999). The improvement most fiequently 'wished for' in

response to an open-ended question about desired improvernents in DMMSDs was a pill

forrn of the drug. The other 'wished for' improvements, in descending order of fiequency,

were: pre-rnixed preparations, a curdstop progression, fewer injections, lower cost and

fewer side effects. Support is also found in a study of growth hormone. Singh et al. (1998)

surveyed parents of short stature children, registered to see an endocrinologkt for short

stature but not on therapy. They found that that long-term side effects and cost were more

important to parents than the efficacy of therapy.

Our finding that changes on the efficacy attributes (relapse, progression, and energy)

were the least important to patients was contrary to clinical expectations and the results of

the attnbute rating survey. Not withstanding the low importance assigned to changes in

efficacy attributes by the majority, 20 subjects made lexicographic choices based solely on

efficacy or an efficacy attributes.

A study of patient preferences for migraine treatments by Luciani et al. (1 995) found

that migraine patients rated two efficacy attributes (how weU it works and how fast it works)

Page 82: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

73

in the top three most important of 10 drug attn'butes. This finding contradicts the findings of

the CBCJA component of our study but is congruent with the results of our attribute rating

suwey. It is understandable that an effecbve treatment of migraine is important to patients

because good drugs for migraine act quickly and can cure the migraine; however, the

DMMSDs are chronic therapies that do not offer comparable 'relief for MS patients.

ïhere are several general explanations for the low importance of efficacy amibutes

resuiting tiom the MNL analysis. One possibility is that the attnbute levels used to describe

efficacy were not attractive enough to cause participants to make tradeoffs in their favor.

S pecifically, prevention of any fùture relapses, curdreversal of disability and e l i i a t i on of

chronic fatigue were not included as attribute levels because they were thought to be

unrealist ic outcornes. Another possible explanat ion for the low importance of the efficacy

attributes is that, because al1 participants had mild or moderate MS, they may have been

optimistic regarding their own disease course and response to treatment. This may have

influenced their perceptions of the importance of the individual efficacy attributes in the

hypothetical drug profiles. A third reason is that attnbute importance may d z e r depending

on assessrnent methods (Jaccard et ai. 1986)-

There are also specific explanations for low importance of changes in the relapse and

progression attributes. It is possible that patients are not strongly duenced by relapse

fiequency because they are rare events (KesselMg 1997). It is also possible that patients feeI

that the dmgs are sufficiently reducing the number and severity of their relapses and so no

improvement is really needed or that reduction in relapse is not enough. A possible

explanations for the finding in Our study that changes in progression were of low importance

is that we descnbed 'progression', in terms of fùnctional status (Le. ability to do daily

Page 83: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

74

activities), as opposed to changes in the advancement of the disease. It may be that changes

in functionai status are perceived to be l e s serious than changes in disease progression.

Overail, it may be that improved efficacy is important to patients, but the levels we

selected, the descriptors we used, and the number of attributes representing efficacy

(relapses, progression and energy) affecteci the importance weights obtained. Future researc h

in this area might combine the efficacy attributes into one comprehensive attribute so as not

to difise the influence of efficacy on patient choice.

In tems of other attributes, our finding that an oral dosage hm was important

matches the finding reporeed by Vollrner and Hadjirnichael (1999) that a pi11 fom was the

most wanted improvement. The cost attribute appeared to dorninate many patients' treatment

decisions. This may have been an artifact of the range of cost levels used ($0 to $1,000 per

month). We considered reducïng this range to a maximum of $500 per month because of the

dominance of the cost attribute in the pilot study. We decided not to because some patients

have to pay large sums out of pocket for the drugs and some patients in the pilot test

indicated that they were willing to pay more than $500 per month for an existing DMMSD.

A recornmendation for fùture research is to survey current DMMSD users to identie the

range of monthly out of pocket costs for a DMMSD. Using this as a foundation for the levels

of the cost attribute in a CJA would reduce the dominance of cost, because it is likety that

the range would be narrower than the range used in this study. With a smaller range of cost,

respondents will likely pay more attention to the other attributes.

5.4 Willingness-to-Pay for Changes in DMMSD Attributes

The inferred amounts subjects were willùig-to-pay for improvements in attributes

were directly related to attribute importance. When summed over al1 attributes, the totai

Page 84: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

75

inferred WTP for a hypotheticai drug composed of the highest levels on each attribute was

$2,389 each month. This calculation has two assurnptions: 1) independence of drug attributes

in infIuencing overd preference for a drug and 2) additivity of the parameter estimates of

each attribute level for a given dmg protile. These assumptions were not tested; therefore,

the inferred WTP may be overestimated. The estimated WTP of $28,668 per annum for this

hypothetical best exceeds the average price of exkting DMMSDs by about one and a haif

tirnes.

5.5 Effect of DMMSD-Experience and Disease Severity on Patient Preferences

The finding that the effect of DMMSDs on relapses is more important to those with

DMMSD experience may be explained by the fact that DMMSD-experienced respondents

had more relapses in the iast two years compared to DMMSD-naïve respondents. However.

the finding that improvement in relapse reduction was more important to those with mild

disease cannot be explained in this way, because relapse fiequency was sirnilar between

groups. The explanation for the interaction between disease seventy and relapse is not clear.

5.6 Willingness-to-Pay for Existing DMMSDs

The responses to the WTP survey question may have been biased by DMMSD

experience, in that the price these respondents paid for DMMSD therapy might have

influenced their response to the CV question. Also, a strategic response bias may have been

present in that respondents rnight have thought that iftheir stated WTP exceeded their

existing cost to use DMMSD therapy, they might be required to pay the dserence in the

fùture. Although we do not know how much each survey respondent was paying for their

DMMSD. we do know the average amount that DMMSD-experienced participants in Our

Page 85: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

76

focus group and attribute rating suwey (n=19) were paying was S 173 per month (95%

confidence interval $166 to $426). This arnount is comparable to the WTP estimated fiom

the CV question which was $227 per month (95% confidence interval $183 to $271).

The inferred WTP for a typical existing DMMSD was found to be more than double

that detennined by contingent valuation. Only a loose comparison can be made between

these two results because the estimate of WTP inferred by CBCJA was made with the

assumption that a DMMSD composed of the worst levels on ail attn'butes was worth zero

dollars. With CBCJq absolute WTP is not estimable, ody WTP for improvements in an

attribute (Le. injection to capsule) can be estimated.

According to responses to the CV question, respondents were wiiiing to pay 7.8 1%

of their total household incorne (censured at $75,000) to use an existing DMMSD. This is

substantial considering that the typicai existing DMMSD descnïed offered deviation of

symptoms, not a cure.

5.7 Predictive Ability of the MNL Choice Model

Tliere are several ways to assess the predictive ability of the MNL choice model. The

statisticai model estimated that respondents would choose the 'no new drug' option 3 1.5%

of the time. This was remarkably close to the actual percentage of tasks in which participants

selected the no drug option (3 1 -3%). For the 10 14 times (out of the 1477 observations) that

respondents selected a drug alternative, the main effects model predicted 708 (69.8%) of

these choices correctly. This proportion is similar to those found in other studies. In a study

of preferences for in vitro fertilization by Ryan et al. (1999), 79% of choices were predicted

correctly, while a study of preferences for miscarriage management, predicted 70% of the

Page 86: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

77

choices correctly (Ryan and Hughes 1997). The other two CJA of preferences for drug

therapy (Reardon and Pathak 1990, Sin@ et al. 1998) did not empioy a choice task.

The percentage of correct predictions is ükely an overestimate because the model that

was used to derive the parameter estimates was also used to predict responses. A better

assessrnent of predictive ability would have been to incorporate a holdout sample, which

consists of a subgroup of the original sample, whose responses are intentionally not included

in the statistical estimation of model parameters. Predictive ability of the model would then

be tested on the responses of the holdout group. Due to sample size limitations, this exercise

could not be undertaken.

The face validity of the model was also examined. The signs of the parameter

estimates for each attribute level are consistent with intuition, that is, patients prefer lower

nsks of side effects, l e s out of pocket cost, better efficacy (on energy, relapses and

progression of disease), less fiequent dosing and oral vs. injectable administration.

It was beyond the scope of this study to determine whether patients' choices in the

CJA tasks predicted their real fùture dmg choices. Thus, we cannot comment cn the external

vaIidity of this study. Future methodological work on CBCJA in healthcare should attempt to

compare choices stated in a CJA with subsequent choices in clinical settings in an effort to

address the external validity of CBCJA

5.8 Study Limitations

The results of this study are subject to limitations. Fust, drug preference was assessed

by a series of choices in a paper and pencil task. Although such choices are indicators of

patients' intended behaviour, they do not represent real behaviour. Also, the dnig choices

that respondents made pertain only to initial use of a new dmg and do not represent choices

Page 87: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

78

that would continue over t h e . This is because repeat choices are partly dependent on

satisfaction with actual product experience (Louviere 1984).

A second limitation is that the results of this study can not be generalized beyond

patients with relapsing MS treated at MS clinics in Ontario.

Another limitation is that respondents were not asked about CO-existing medical

conditions. Thus, we could not determine whether comorbidities and their treatment Sected

participants' choice of DMMSD. A h , we did not collect data on the type of DMMSD

experience that patients had had. Specificaiiy, we don't know if their expenence was positive

or negative nor whether they were a current or former user of a DMMSD.

Although each of the 1477 observations are not tmly independent of each other

(because they corne f?om 99 people as opposed to 1477 dEerent people), conjoint analysis

assumes that each observation is independent. An additional assumption was that the drug

attributes were independent of each other, interactions among attributes were not included in

the model. The above two assumptions are typical of CJA studies and have not been tested in

this study. Failure to meet these assumptions would b i s the parameter estimates.

5.9 Recommendations for Future Research

It is likely that people have dEerent attitudes towards drug therapy. Some people

prefer to live with symptoms while others are eager to turn to a pharrnacological solution.

Our study did not specificaily address patients' attitudes to trying pharmacotherapy. A

recornmeiidation for fùture research is to assess the effect of attitude towards drug therapy in

general (vs. other modes of therapy) on patient treatment choice.

Another potentiai research idea is to foltow a group DMMSD-naive patients after

they complete the choice tasks and see how weii the estimated modei predicts actual drug

Page 88: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

79

choice over time. Research in this area would be quite valuable because Little published

information exists on the extemal validity of CJA studies.

S. 10 Conclusions

This study examined the relationship between cimg attributes and patient choice of

DMMSD using a CBCJA with seven drug attributes. Use of seven product attributes in a fùll

profile CBC JA provided reasonable prediction of respondents' choices. As hypothesized,

changes in levels of al1 drug attributes a f f i e d patient preference for DMMSDs. Overali,

patients' drug choices were more uifluenced by changes in negative aspects o f therapy (cost,

dose regimen and dosage form) than changes in positive ones (effect on relapses, progression

of disease and energy level); however, 20% of the participants made dnig choices based on

efficacy attributes alone.

Disease severity and DMMSD experience uifluenced the importance of drug

attnbutes on patient choice of a DMMSD. Patients with moderate MS placed less

importance on the effect of the dmg on relapses than did patients with mild MS- Patients

with DMMSD experience placed more importance on the effect of the drug on relapses and

less importance on the cost of the dmg than did DMMSD-naïve patients.

Patients are willing to pay substantial amounts for irnprovements in the attnbutes of

DMMSDs, with the inferred amount varying by attribute. The amount that MS patients were

willing to pay for existing therapies was less when deterrnined holistically, by contingent

valuation, than when infemed per attribute (by CBCJA) and composed to obtain an overail

estirnate.

The results of this study have marketing implications. While the primary objective of

drug manufacturers is to develop more effective dntgs, this study shows that there is also

Page 89: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

80

value to patients in developing non-parenteral forms and improving the side effect profiles of

DMMSDs.

Finaily, CBCJA is a usefùl tooI for inferring the vdue that patients assign to changes

in drug attributes, especially when the value associated with process of treatment is in

question.

Page 90: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

Abelson R and Levi A- Decision Making and Decision Theory. In: Lindzey G, Aronson E, editors. The Handbook of Social Psychology. New York: Random House; 1985.

Asche CV, Ho E, Chan B. Coyte PC. Economic consequences of multiple sclerosis for Canadians. Acta-Neurol-Scand. 1997;95:268-74.

Ben-Akiva M, Leman SR. Discrete Choice Analysis: Theory and Applications to Travel Demand. Cambridge (Mass): MIT Press; 1985.

Bourdette DN, Prochazka AV, Mitchell W, the VA Multiple Sclerosis Rehabilitation Study Group et al. Health care costs of veterans with multiple sclerosis: implications for rehabilitation of MS- Archives ofphysician Medical Rehabilitation. 1993;74:26-3 1-

Brown MG, Murray TJ, Fisk JD, et al. A therapeutic and economic assessrnent of Betaseron 0 in multiple sclerosis. Ottawa: Canadian Coordinating Office for Health Technology Assessrnent (CCOHTA). 1996.

Brady TJ. The patient's role in rheumatology care. Current Opinion in Rheumatology. 1998;lO:l46-lS 1.

Canadian Burden of Illness Study Group. Burden of Illness of multiple scierosis: Part 1 : Cost of illness. Canadian Journal of Neurological Sciences. 1998;25:23-30.

Carroll JD, Green PE, Psychometnc methods in marketing research: Part 1, Conjoint analysis. Journal of Marketing Research. 1 9%;32::3 85-9 1.

Chakraborty G, Ettenson R, Gaeth G. How consumers choose health insurance. Journal of Health Care Marketing 1994; 14:2 1-33.

Chakraborty G, Gaeth GJ, Cunningham M. Understanding consumer preferences for dental service. Journal of Health Care Marketing 1993 ;Fall:48-58.

Chinburapa V, Larson L. Predicting prescribùig intention and assessing drug attribute importance using CJA Journal of Pharmaceutical Marketing and Management l988;3 :3-18.

Chow GC. Econometrics. New York (NY): McGraw-Hill Book Company; 1983.

Donaldson C, Shackley P. Does "process utility" exist? A case study of wiiiingness-to-pay for laparoscopic cholecystectomy. Social Science and Medicine. 1997;44:699-707.

Donaldson C, Shackley P, Abdalia M. Using willingness-to-pay to value close substitutes: Carrier screening for cystic fibrosis. Health Economics. 1997;6: 145-59.

Drummond MF, O'Brien B, Stoddart GL, Torrance GW. Methods for the Economic Evaluation of Health Care Programmes. 2nd Edition. New York 0: Oxford University

Page 91: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

Press; 1997

England SL, Evans J. Patients choices and perceptions afker an invitation to participate in treatment decisions. Social Science and Medicine. 1992;34: 12 17-25.

European Study Group on Interferon Beta- 1 b in Secondary Progressive MS . Placebo- controlled multicentre randomiseci trial of interferon beta-1 b in treatment of secondas- progressive multipie sclerosis. Lancet- 1998;352; 149 1-7.

Gensch DK, ReckerWW. The multinomial, multiattribute iogit choice model. Joumal of Marketing Research. 1979; 16: 124-32.

Green PE, Srinivasan V. Conjoint analysis in consumer research: Issues and outlook- Journal of Consumer Research 1978;s: 103-23.

Green PE, Sri~vasan V. Conjoint analysis in marketing: New developments with implications for research and practice. Journal of Marketing 1990 October:3-19.

Green PE, Wind Y. Multiattrïbute decisions in marketing: A measurement approach. Hinsdale (IL): The Dryden Press; 1973.

Greene WH. LIMDEP Execution Trace (7.0). Econometrics Software Inc. NY. 1995

Greenfield S, Kaplan S, Ware JE. Expanding patient involvement in care. Annals of Internai Medicine. 1985; 102:520-8.

Hair FH, Anderson RE, Tatham RL, BIack WC. Multivariate Data Analysis: with Readings, 4Lh Ed., Englewood Cliffs, (NJ) Prentice Hall; 1995.

Hamilton JG. Needle phobia: A neglected diagnosis. Journal of Family Practice. 1995;4 1 : 169-75.

Harris C, Metz L. (1993). Home intravenous Solu-Medrol program for multiple sclerosis patients. (abstract). Proceedings of the Multiple Sclerosis Sateiiite Symposium to the World Congress of Neurology, September, 1993, Vancouver, British Columbia, pp. 36.

Heman JM. The use of patients' preferences in farniiy practice. Journal of Family Prtactice. 1985;20: 153-6,

Hunt LM, Valenniela MA, Pugh J A NIDDM patients' fears and hopes about insulin therapy. Diabetes Care. 1997;20:292-8.

ïNFB MS Study Group and the University of British Columbia MS/MRL Analysis Group. Interferon beta-1 b in the treatment of multiple sclerosis: Final outcome of the randornized control trial. Neurology 1995;45: 1277-85.

IMS Canada, unpublished data, 1999.

Page 92: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

Inman RP. Disability indices, the economic costs of illness and social insurance: the cost of MS . Act a Neurologica Scandinavia Supplement. 1 984;7OS :46-S -

Jaccard J, Brinberg D, Ackeman W. Assessing attribute importance: A comparison of six methods. Journal of consumer research, 1986; 12:463-8.

Jacobs LD, Cookf'air DL, Rudick et al. Intramuscular interferon beta-la for disease progression in relapsing multiple sclerosis. Arnerican Neurological Association. 1996;39:285- 94.

Johnson KP, Brooks BR, Cohen JA, et al. Copolymer 1 reduces relapse rate and improves disability in relapsing-rernitting multiple sclerosis: Results o fa phase III multicenter, double- blind, placebo-controlled trial. Neurology 1995;45: 1268-76.

Johnson RM, Orme BK. How many questions should you ask in choice-based conjoint studies? ART Forum, Beaver Creek, 1996. Sawtooth Sofhvare, inc.

Kassirer JP. Incorporating patients' preferences into medical decisions. Editorial. New England Journal of Medicine. 1994;330: 1895-6.

Keating MM, Ostby PL, Education and self-management of interferon beta- lb therapy for muItiple scIerosis. Journal of Neurosçience Nursing. l996;28:3 50-8.

Kesselring J, editor. Multiple Sclerosis. Cambridge: (NY): Cambridge University Press,: 1997.

Kesselring J. Clinical patterns of multiple sclerosis. European Journal of Neurology. 1998;5(~2):s 1 1-2.

Kim ED, El-Rashidy McVary KT. Papverine topical gel for treatment of erectile dysfùnction. Joumal of Urology. 1995; 153 :36 1-5.

Kuhfeld WF. Muitinomial Logit, Discrete Choice Mgdeling. SAS Institute Inc. Cary, NC. 1996.

Leman CE, Brody DS, Caputo GC et al. Patients perceived involvement in care scale: Relationship to attitudes about iliness and medical care. Journal of General Interna1 Medicine. 1 WO;5:29-33.

Laupacis A, Feeny D, Detsky A, Tugwell PX. How attractive does a new technology have to be to warrant adoption and utilization? Tentative guidelines for using clinical and econornic evaluations. Canadian Medical Association Journal, 19%; I46:473-48 1.

Louviere JJ. Using discrete choice experiments and multinomial logit choice rnodels to forecast t d in a competitive retail environment: A fast food restaurant iiiustration. Joumal of Retailing. 1984;4:8 1- 107.

Page 93: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

Louviere JJ, Woodsworth G. Design and analysis of simulated consumer choice or allocation experiments: An approach based on aggregate data Journal of marketing Research. 1983;20:350-66.

Luce RD, Tukey JW. Simultaneous conjoint measurement: A new type of findamental measurement. Journal of Mathematical Psychology 1964; 1 : 1-27.

Luciani RJ, Osterhaus JT, Gutterrnan DL. Patient preferences for migraine therapy: Subcutaneous sumatriptan compared with other medications. Journal of Family Practice. 1995;4 1 : 147-52.

Ludwig H, Fritz E, Neuda J, Dune BGM, Patient preferences for interferon alfa in multiple myeloma. Journal of Clical Oncology. 1997; 15: 1672-9.

McFadden D. The choice theory approach to market research. Marketing Science. l986;5 :275-97.

Mancardi GL, SardaneIli F, Parodi RC, Melani E, et al. Effect of copolyrner-1 on serial gadolinium-enhanced MRI in relapsing remitting multiple sclerosis. Neurology. I998;SO: 1 127-33.

O'Brien B, Gafhi A When do the "Dollars" make sense? Toward a conceptual fiarnework for contingent valuation studies in health care. Medical Decision Making. 1996; 16:288-99.

O'Brien BJ, Novosel S, T o m c e G, Streiner D. Assessing the economic value of a new antidepressant : A willingness-to-pay approach. P harmacoeconomics. 1 9S;8 :3445.

O'Connor P. Multiple sclerosis: The facts you need. Canadian Medical Association. Toronto (ON); 1998.

Orme B. Sarnple size issues for conjoint analysis studies. Sawtooth Software, hc. 1998

Onen N. Interferon beta- 1 b and multiple sclerosis. Ottawa: Canadian Coordinating Office for Health Technology Assessment (CCOHTA); 1 996.

Otten N. Cornparison of drug treatments for multiple sclerosis. Ottawa: Canadian Coordinating Office for Health Technology Assessment (CCOHTA); 1998.

Porter M, Macintyre S. 'What is, must be best': A research note on conservative or deferential responses to antenatal care provision. Social Science and Medicine. 1984; 19: 1 197- 1200.

PRlSMS Study Group. Randomiseci double-blind placebotontroUed study of interferon beta- 1 a in relapsing/remitting multiple sclerosis. Lancet. l998;3 52: 1498-504.

Reardon G, Pathak DS. Sepenting the antihistamine market: An investigation of consumer preferences. Journal of Health Care Marketing. 1990; 10:U-33.

Page 94: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

Robson LS, Bain C, Beck S, et al- Cost analysis of methyiprednisolone treatment of multiple sclerosis patients. Cnadia. Journal of Neurological Sciences. l998;25:222-9.

Rosenfeld BD, White M, Passi. SD. Making treatment decisions with HIV infection: A pilot study of patient preferences. Medicai Decision Making. 1997; 17:307-3 14.

Ross CK, Steward CA, Sinacore JM. The importance of patient preferences in the measurement of health care satisfaction. Medicai Care. t 993;3 1 : 1 138- 1 149.

Ryan M, Hughes J. Using conjoint anaiysis to assess women's preferences for miscarriage management. Health Economics. l997;6:26 1-73.

Ryan M, McIntosh E, Shackley P. Meihodological issues in the application of conjoint analysis in health care, 1998;7:373-78.

Ryan M. Using conjoint analysis to take account of patient preferences and go beyond heaith outcomes: an application to in vitro fertihtion. Social Science and Medicine. 1999;48:535- 46.

Sadnovick AD, Ebers GC. Epidemiology of multiple sclerosis: A critical overview. The Canadian Journal of Neurologicd Sciences 1993 ;20: 17-29.

Sadnovick AD, Ebers GC, Wilson RW, Paty DW. LXe expectancy in patients attending multiple sclerosis clinics. Neurology. 1992;42:99 1 4 .

Seyarn RM, Begin LR, Le-Mai T, et al. Evaluation of a no-needle penile injecter: A preliminary study evduating tissue penetration and its hemodynamic consequences in the rat. Uroiogy. l997;50:994-8.

Singh J, Cuttler L, Shin M, et ai. Medical decision making and the patient: Understanding preference patterns for growth hormone therapy using conjoint analysis. Medical Care. l998;36:AS3 1 -AS45.

Solai 4 Amato MP, Bergamaschi R, Logroscino G, et al. Accuracy of self-assessment of the minimal record of disability in patients with multiple sclerosis. Acta Neurologica Scandinavica. 1993;87:43-46.

Stiggelbout AM, Kiebert GM. A role for the sick: Patient preferences regarding uiformation and participation in clinical decision-making. Canadian Medical Association Journal. 1997; 157(4):3 83-9.

Stanek EJ. Cheng JWM, Peeples PJ, et al. Patient preferences for thrombolytic therapy in acute myocardial infarction. Medical Decision Making. 1997; 17:464-7 1.

Stnihl S. Discrete choice modeling: Understanding a "Better conjoint than conjoint". Quirk's Marketing Research Review. JundJuly 1994. (Reprint fiom Sawtooth Software)

Page 95: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

Strd WM, Lo B, Charles G. Do patients want to participate in medical decision m a h g ? Journal of the American Medical Association 1984; 252:2990-4.

Surnra M, Gupta VB. Insulin to human insulin. Indian Dnigs. 1995;32:356-6 1.

Trehan 4 Ali A. Recent approaches in insulin delivery. Drug Development and Industrial Phannacy. l998;24:589-97.

VolImer TL, Hadjhichael O. Adherence to injection therapy in multiple sclerosis: US patients survey. [abstract 14631 Proceedings of the 5 lL Annual Meeting of the American Academy of Neurology; 1999 April 17-24; Toronto, Ontario, Canada

Weinshenker BG. Natural history of multiple sclerosis. Annals of Neurology. 1994;3 6:s6- sll.

Weinstock-Guttman B, Cohen JA Newer versus older treatments for relapsing-remitting multiple sclerosis. Dnig Safety. 1996; 14: 12 1-30.

Whynne 4 Monks J. Patients' decisions about continuing with therapy in chronic illness: A study of hyperbaric oxygen therapy in multiple sclerosis. Family Practice. 1989:6:268-73.

Wilson Rh& Martin JR. Economic Costs to individuals with multiple sclerosis. A report to the National Multiple Sclerosis Society. 1993

Page 96: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

APPENDIX A

Human Subjects Approval: University of Toronto

Page 97: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

University of Toronto

OFFICE OF THE VICE-PRESfOENT - RESEARCH AND INTERNAT~ONAL REUTIONS

PROTOCOL REFERENCE # 4270

Prof L Mackeigan Fadty of Pharmacy 19 Russeil Sucet Toronto. ON M5S 2S2

Dear Prof Mackeigan,

'Parient preferences for disease modifiing dmgs for rn&ple scferosis '

We are wriring to advise you that Dr. T. Paton of the Human Subjects Review Comminee has granteci approval to the above-narned research study based on the SC. ~Michael's Hospital Research Erhics Board approvai.

The appmved consent forms are attached. Subjects should reccive a copy of thcir consent fotm.

During the course of the research. any significant deviation fkom the approved protocol (that is, any deviation wbch wouid lead to an increase in risk or a decrease in benefit to human subjects) ancilor any unanticipa?& Ceveiopments within the research should be brouet to the attention of the Office of Research Services.

Best wishes for the successful cornpietion of your projecr.

Yours sincerely ,

Executive Officer Human Subjects Review Committee

SPlrnr Enclosure(s) cc: Dean W. Hindmarsh

St. Michael's Hospirai

Page 98: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

Human Subjects Approvd: MCMastcr University

Page 99: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

McMASTER UNIVERSITY faculty of health sciences 1200 L\AlN STREET WEST, HAWLTON. ONTARIO L8N 37f

DATE: ~March 12, 1999

TO: W. Mark Rolnick CC: Dr. Paulseth

FROM: Marie Townsend Committee on Scientific Development

RE: "Patient Preferences for Drug Treatments for Multiple Sclerosis"

I am writing to notÏQ you that the above smdy has been revîewed and approved by the Research Ethics Board of the FacuIty of Health Sciences/Hamilton He& Sciences Corporation.

Please be sure to inform the Research Ethics Board of any changes to the study protocol.

Good luck with your project,

Page 100: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

b

Human Subjects Approval: St. Michael's Hospital

Page 101: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

Research Ethies Board Office of Research Administration Tel-: 416 864606û Ejd L= Fscsirnilt' 41 6 -0 e-rnaiI: pîLddQsmh.toronta on.ca

October 7. 1998

Dr Paul O'Connor Chief. Division of Neumlogy St Michaet's Hospital

Oear Dr O'Connor

Re; REB 98-99-026: Patient Prefemnces for Treatment for Multiple Sclemsis

Thank you for yaur communications of September 30, 7998 regarding the above narned study. You have adequately addressed the canums raised oy the Research Ethics Board at its meeting of September 9, 1990.

1 am happy to issue final approval for the study for a pen'od of 12 months from the date of this fetter. Continuation beyand that date will requim further annual review of REB approval.

During the course of this investigation, any significant deviations fmm the approved protocol and/or unanticipated developrnents of significant advene events should immediately ba brought ta the attention of the Research Ethics Board.

This letter serves as approval by the St Michael's Hospital Research Ethics Board for conduct of this study. hwever additional approvals are requimd as outlined on the Research Administration "Authon'zation Check List" fcrrn. I encfose a copy of this check list and have indicated REB autharization in the appropriate space. The fernainder of the approvals mu- be coordinated through the Office of Research Administration prior to initiation of this research.

Good iuck wrth your investigation.

VVith best wishes

u Ron Heslegrave Ph0 Co-Chair Research Ethics Board

End: R Hfdp

Toronto's Urban Angel

Page 102: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

Patient Informatioo Sbeet

Page 103: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

Investiaators: Mark Rolnick, B.Sc.Pharm. Master's Student, Faculty of P hamacy , University of Toronto

Linda MacKeigan, PhD Thesis Supervisor and Assistant Professor, Faculty of Phannacy. University of Toronto

Paul O'Connor, MD, MSc, FRCP(C) Chief - Division of Neurology St. Michael's Hospital and Associate Professor, Faculty of Medicine, University of Toronto.

Studv Title: Patient Preferences for Disease-modifying Dnigs

for Multiple Sclerosis

Within the past couple of years 4 drugs have become

available to treat multiple sclerosis, however, patients seem

reluctant to try them. In order to understand why, we need to

investigate what aspects of treatment are important to patients

when they make decisions about whether to try new drugs. We

need your help to do this.

S t u d ~ Pur~ose:

The purpose of this study is to find out what aspects of

treatment are important to multiple sclerosis patients when making

treatment choices.

Toronto's Urban Acgel

Page 104: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

Studv Descri~tion:

Should you wish to participate in this study, you will complete

a two-part questionnaire at the St. Michael's Hospital multiple

sclerosis clinic or at home. In the first section, you will be

presented with sets of 2 drug descriptions. You will then be asked

to choose the treatment that you rnost prefer. Should none of the

presented options be beneficial enough to you, you may select the

'no drug' option, indicating that you prefer to get treatment for

symptoms and relapses but not for your multiple sclerosis itself.

You will see a series of 15 choice tasks of this nature. The second

part of the questionnaire will ask you to provide personal

information on disease history, age, sex, household income,

multiple sclerosis history and your experience with new treatments.

This information will help us detennine if disease history and other

qualities of multiple sclerosis patients affect their preferences for

treatments.

Benefits:

You will not benefit directly from this study. However, this

study will offer you the chance to think about therapies for your

multiple sclerosis in ways you might not have before.

Information from this study can help drug manufacturers be

more in tune with patients' preferences, potentially leading to

drugs that are more appealing to patients like you.

Also, the findings from this study will increase doctors'

awareness of patients' treatment preferences helping them to

make better prescribing decisions for patients.

Page 105: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

Risks:

As this study is in the fonn of a questionnaire, there are no

foreseeable health risks to you. AI1 information obtained from the

questionnaire will be kept strictly confidential, it will only be

identified by a study number. You will not be personally identified

in any publication or report. Only the results from patients as a

group will be reported.

Voluntarv Nature of Studv - Freedom to Withdraw or Partici~ate:

Your participation in this study is cornpletely voluntary. The

care you receive at the ch ic will not be affected by your decision

to participate in this study. You may choose to withdraw from this

study at any time, for any reason.

Should you want more information, please feel free to cal1 the

primary investigator. Mark Rolnick, at (41 6) 978-6608.

Page 106: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

APPENDIX E

Study Consent Form

Page 107: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

Patient Preferences for Multiple Sclerosis Treatments Study

Investigator: Mark Rolnick. Graduate Student, Faculty of Pharmacy, University of Toronto

I understand that participating in this study means completing a

written questionnaire on personal information and treatment choices.

I understand that participating will take approximately 20 minutes

of my time and will occur at the multiple sclerosis clinic or at home. 1

realize that I may refuse to answer any questions. I understand that my

name will not appear on the questionnaire and that my specific answen

will remain confidential. I understand that I will not be identified in any

report or presentation that may anse from the study.

I understand the benefits and risks, to me, of participating in the study. All

questions that 1 presently have conceming my participation in this research

project have been answered to my satisfaction. Should I have any questions in

the future, I undentand that I may ask or cal1 the primary investigator, Mark

Rolnick, at (41 6) 978-6608.

I understand what this study involves. My signature below

indicates that I voluntarily agree to participate and that have been

offered a copy of this fom.

Date Signature

Printed Name

Witness

Page 108: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

Focus Group Report

Page 109: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

Ph'ltuuy Objective

1. To detemine, in patient language, what attributes of drug therapy patients consider when deciding to try a new MS dmg.

Rationale: To identfi dmg attributes important to patients. To describe these attributes, in language patients understand, for the survey questionnaire-

Sec0nd.y Objectives

1. To determine reasons why some patients decide to adopt therapy while other patients do not.

Rationale: To identfi appropriate response options for forced-choice questions addressing reasons for tryinghot trying dnig treatment in the demographic and medical.

2. To get descriptions, in patient language, of side effects experience(d) whiie on therapy. R a t i o d e : To ident* h g side effects that patients experience(d).

To describe these side effects, in language patients understand, for the survey questionnaire.

3. To Iearn how fmely MS patients distinguish side effects of drugs when they are making therapeutic decisions.

Rationale: To determine how to describe side effects as attributes of the dmg profiles.

4. To define in patient language relapse and disease progression. Rationale: To describe these terms on the survcy questionnaire in words patients can understand.

Two focus groups, each lasting 1 to 1.5 hours, were held on separate weeknights in

the waiting room of the St. Michael's Hospital MS clinic in early November 1998. The

groups were composed of relapsing remitting MS patients, who could walk fieely or with

assistance of a cane and who were over 18 years oid and could speak English. One group

consisted of DMMSD-naïve patients (n=6) while the other of DMMSD-experienced patients

(n=8). Upon completion of the focus group, each respondent was oEered $10 dollars to

cover their travel expenses.

Page 110: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

Prior to the initiation of the focus group, each participant completed a consent form

and a brief demographidmedical questionnaire. During the fonis groups several questions

were posed to the participants. These are presented in foiiowing box.

Questions Posed to Focus Group Pahcipants

DMMSD Experienced gr ou^: 1. What aspects of drug treatment of multiple sclerosis did you consider when deciding to

try a new drug? Contingency Question

If you could make any changes to your drug therapy at al1 what would they be? 2. What side effects have you experienced whiie on these drugs? 3. When you think about side effects of drugs what kind of information is usehl to you?

(Le. do you like to see the side effects listed dong with their fiequencies or is it enough to know generally what the side effkcts are and overall what are the chances that you'U experience them.)

4. For what reasons did you, at one point, decide to try one of the new multiple sclerosis drugs?

5 - What cornes to mind when you think of the phrase 'relapse of multiple sclerosis'?* 6 . What cornes to mind when you think of the phrase 'progression of multiple sclerosis'?*

* Time ran out before these questions could be asked

DMMSD Naïve gr ou^:

1. Ifyour doctor wanted you to consider a new drug treatment for your multiple sclerosis what would you like to know about the dnig? a What characteristics of a drug do you (would you) consider when deciding to try a

new drug for your multiple sclerosis? 2. For what reasons have you not tried one of the new multiple sclerosis modXying drugs? 3. What cornes to mind when you think of the phrase 'relapse of multiple sclerosis'? 4. What cornes to mind when you think of the phrase 'progression of multiple sclerosis'? 5. If your doctor wanted you to consider a new drug treatment for your multiple sclerosis

what sort of details would you like to know about the drug's side effects?

The focus groups were moderated by the prirnary ùivestigator (MR) and Linda Mackeigan

was the assistant. Data was coltected on audio-tape and on poster paper.

Page 111: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

Dernographic characteristics and MS-related medicai history of focus group

participants are summarked in Tables 1 and 2.

Drug attributes that focus groups participants considered when making treatment

choices were categorized as: 1) side effects, 2) benefits, 3) how the drug has to be taken (by

injection or by mouth), 4) duration of treatment, and 5) the cost. The benefits of interest

included less fiequent exacerbations, duration of treatment, curdrepair existing darnage. The

DMMSD-experienced group aIso identified the eEect of the drug on relapse duration as a

benefit of treatment. Frequency of dosing, dmg interactions and effect on pregnancy were

identified by the DMMSD-naïve gmup as attributes that would affect their treatrnent choices.

The reasons that the DMMSD-naïve group provided for not being on therapy

included: 1) the high cost of new agents, 2) the risk associated with new medicines that don't

have long term data on safety and effects 3) family pressures to stay off of medicines, 4)

negative attitude towards medicines in general, 5) symptoms of MS not yet severe enough to

warrant treatment.

The group of patients with DMMSD-expenence said that they decided to go on drug

therapy for four reasons: 1) they wanted a feeling of hope, 2) they wanted to take 'control'

of their MS, 3) they were influenced by their neurologist and 4) they had developed new

symptoms.

The side effects that the interferon-experienced participants cited were 'au-like

symptoms' (Le. fever, shakes, spasms, vomiting, chills, sweats, nausea, loss of appetite),

worsening symptoms of MS over the first few months of drug therapy including leg

weakness and fatigue, and injection site reactions (i.e areas of lasting soreness, redness, and

Page 112: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

slight hardness of skin). None of the participants mentioned any severe skin darnage

(necrosis) at the injection site.

The side effects that participants with experience with copolymer- l cited included:

pain on injection that lasted 10-15 minutes before subsiding and shooting pains in the

extrernities. The redness and soreness caused by the injection seemed to be less of an issue

and neit her of t he two participants on copolymer- 1 had experienced a chest tightness or

flushing reaction.

Some patients in the DMMSD-naïve group described a 'relapse' as an ernotional

period of time when patients experience new symptoms that are unexpected and that do not

completely go away. Others described it as a sense that something about them was changïng.

Symptoms that the group had experienced included: 1) muscle spasms, 2) insomnia, 3)

slurred speech and slowness in expressing thoughts, 4) numbness and lack of control in arrns

and legs, 5) eye pain and headache sometimes associated with blumed or double vision, 6)

incontinence or difficulty urinating, 7) weakness in legs, 8) disorientation (due to balance,

nurnbness and vision problems), and 9) being bedridden.

The DMMSD-naïve group described progression of multiple sclerosis as the process

whereby they begin accepting their limitations based on inability to do usual activities (Le.

dimbing stairs, riding the bus). Participants also felt that progression occurs when they

cannot cope as well and are constantly thinking of their MS. Finally, general muscle

weakness with symptoms that Iast longer with less time between relapses was also associated

with progression of disease.

Focus group participants wanted information about long term side effects, precise

descriptions of common side effects in patient language and a List of possible side effects with

frequencies/probabiiities of occurrence.

Page 113: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

Discussion and Acîiom Taken:

The ability of a new drug to be a cure o r repair existing damage was narned as one of

the benefits of treatment that patients in both groups would like to see. However, as this

Ievel of benefit, is far fiom beïng availabie, i t was decided not to include it.

Arnong other attributes pertaining to drug efficacy, the effect of the drug on relapse

duration was identified by the DMMSD-expenenced participants to be important. However,

relapse duration was not measured in any of the DMMSD clinîcal studies. As relapse severity

was measured in several DMMSD studies, we decided to include it in the attribute rating

suwey in lieu o f relapse duration.

The impact of drug interaction and the warning against becorning pregnant on

DMMSD therapy were two attributes identified by the DMMSD-naive group participants.

However, we decided not to include them in the next stage of the study as the drug

interactions are not known to be significant and the waniing against becoming pregnant is a

recommendation for many new drug treatments.

General worsening of weakness and fatigue was described as a side effect that some

users of interferon experienced over the first months of therapy. Also, participants associated

weakness with progression o f their disease. However as the effect on energy level seems to

be distinct fiom progression of disease, we decided to include it as a separate attnbute.

Although physician intluence was a main factor in patients' adoption of therapy, it

was not incorporated into the survey as an attribute. This is because without physician

support of the DMMSDs, patients would not have access to these treatments. An underlying

assumption in Our study was that the physician presents the drug treatment option to eligible

recipients (i.e. physician influence) and patients must decide to try a drug o r not t o try a

drus based on attributes of the drug itself.

Page 114: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

105

The dosage form (i-e. needle versus pill) is one such dnig attribute that is an issue for

people who cannot receive injections for medical rasons and for those who have needle

phobia. GeneraLiy speakuig, participants in both focus groups seemed to be willing to tolerate

injections for proven efficacy. Although one of the key difrences in the administration of

the DMMSDs is the frequency of injection, only the DMMSD-naïve group mentioned it as an

important attribute.

Based on the results of the focus groups, the attributes considered for inclusion in the

attribute rating survey were: 1) the effect of the dmg on the number of flue-ups that you

experience, 2) the effect of the dmg on the severity of your flare-ups, 3) the chances of

getting 'flu like symptoms' (these include sweats, chills, nausea, vomiting and fever) after

every dose, for the first few months, 4) the effect of the drug on the progression of

your multiple sclerosis, 5) the number of times per week you have to take the drug, 6) the

out of pocket cost to you and your family, 7) the chances of getting a lasting skin reaction

(red, sore, and slightly swollen blotches) o n your stomach, and thighs 8) how you have to

take the drug (e-g. by injection or by mouth), and 9) the effect of the drug on your energy

level.

The DMMSD-naïve group was interested in knowing for how long the effects of the

drug treatments last. As this information is not precisely known, we decided to use a blanket

statement in the real survey that addressed the general lack of long term information on both

positive outcornes and side effects of these dmgs.

Participants regarded worsening of old symptoms to be less alarming than appearance

of new ones. However clinical studies of DMMSDs defined relapse in terms of worsening of

old symptoms and the appearance of new ones. We decided to define relapse with the

definition used in the DMMSD clinicat trials.

Page 115: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

106

The focus group participants described the term 'relapse' similady to the way it was

descnbed in the demographic and MS-reiated medical history questio~aire patients füled out

p60r to the focus group and so, no significant changes will be made to the definition of

reiapse.

Page 116: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

Attribute Ranking Survcy

Page 117: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

S ~ M H Y- kt:<-! !:\LI ? 1 :c b \ l t l l # \ i

Attribute Ranking Suwey uniwarsty ot T m w

In group discussions with multiple sclerosis patients, we have identified 9

characteristics of multiple sclerosis dnigs that patients consider when making a

decision about dnig therapy.

Please rank the characteristics of treatment listed below from the most

important (1) to least important (9) to you, if you were deciding whether to start

a new multiple sclerosis drug.

d C ha racteristics

The effect of the dnig on the number of Rare-ups that you experience

The effect of the drug on the severity of your Rare-ups

The chances of getting Wu Iike symptoms' (these include sweats, chills, nausea, vomiting and fever) after every dose, for the first few months

The effect of the dnig on preventing reduction in your physical abilities

The number of times per week you have to take the dnig

The out of pocket cost ta you and your family

The chances of getting a lasting skin reaction (red, sore, and slightly swollen blotches) on your stomach, and thighs.

How you have to take the drug (e-g. by injection or by mouth)

The effect of the dnig on your energy level

Other (please write in)

If you were deciding whether to try a new, multiple sclerosis, dnig treatment, Toronto's Ur an Ar.

how many cAara8eristics of the drug would you consider at one time?

Page 118: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

Attributt Rirrnking Suwty Report

Page 119: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

Goal

To determine which attributes to include in the description of the DMMSDs in the questionnaire

Objectives

1. To reduce the number of attributes descnied in the focus groups to a maximum of six Rationale: The maximum number of attributes recornmended for inclusion into a &il profile conjoint analysis is six (Green and Srinivasan 1990).

2. To determine how many attributes respondents think they can consider at one time when undertaking the choice tasks in the questionnaire

Rationale: It is possible that participants may fînd it difFidt processing as many as six attributes at once. Requiring participants to perfonn tasks that exceed their capabilities might increase the survey dropout rate o r produce unretiable data.

See Chapter 3 'Attribute Determination'

Results:

Most respondents had difficulty ranking the importance of each attribute because

many patients ranked several attributes with the same importance. This inflated the rankings

with many attributes ranked one and two (most important) and few with ranking of 7,8, and

9 (least important). It was easier for participants to select the six most important attributes. If

repeating such a survey, participants should be instructed to use each number Erom 1 to 9

ody once.

The attributes describing the positive outcomes of therapy ,reduction in number of

relapses, severity of relapses, progression of disease and energy Ievel were most important to

participants. It is also evident that the fiequency of dosing and chances of developing a skin

reaction were not particularly important. As seen in Tables 5 and 6, the order of importance

of attributes obtained by the ranking exercise paralleleci the fiequencies obtained fiom the

check mark task.

Page 120: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

111

To see if the attributes which patients found important were different between

DMMSD experienced and naïve participants, the results fiom each group were analyzed

separately (Tables 5 and 6). The order of importance of the attributes appears to be almost

identicai between groups. The three least important attributes; 1) skin reactions, 2) form of

drug and 3) fiequency of dosing, were the only attributes found to be ranked differently by

each group.

Eleven of the 14 DMMSD naïve participants answered the final question of the

survey, on the number of dmg characteristics that patients could consider at one time, while

al1 1 1 DMMSD expenenced patients answered it. On average the DMMSD naïve group

would consider 7.2 attributes at one time while the experienced group would consider 6-3

attributes.

Discussion:

The results fiom the final question of the survey indicated that patients thought they

could consider on average greater than six attributes at one tirne. Thus, six, the maximum

number of attributes that has been recommended for inclusion into a tùU profile CJA was

used. The difference in number of attributes that DMMSD users vs. non-users would consider

is consistent with the results of the focus groups. Naïve patients tended to want more

information about the dmgs.

The foliowing strategy was used to select six attributes fiom the nine included in the

ranking survey. Two attributes, effect on the number of relapses, and effect on severity of the

relapses were combined to form one because they covaried. Additionally, three of the four

available agents have been s h o w to reduce severity of relapses while al1 of them reduce the

relapse rate. The combination attribute of relapse severity and frequency, in addition to effect

Page 121: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

of the drug on progression of MS, and effect of the drug on patients' energy levet represent

three of the six attributes to be used. The chance of getting flu like syrnptoms and the out of

pocket costs were also be included as drug attributes in the questionnaire. The injection site

reactions were not too important to either group. However, as DMMSD experienced patients

who participated in the focus group complained about them a lot and the severity of the

reaction differs considerably between available injected products, this attribute was combined

with the form of the drug. That is, two types of injections one with more bothersome skin

reactions than the other were used. Although the drugs currently on the market d s e r with

respect to the number of t h e s per week they have to be taken, patients, both with and

without DMMSD experience, did not think that it was particularly important in affecthg

treatment choice. Thus, the dosing fiequency attribute was not included in the drug

descriptions.

Page 122: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

Pilot Test Report

Page 123: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

1. To determine if the instructions for the choice task exercise are clear. 2. To determine ifthe information presented in the choice tasks is sufficient or too much for

participants to choose between treatrnent options. 3. To determine ifany words or terms used in the choice task are not understood. 4. To detennine ifany questions in part 2 are unclear 5 . To detennine if the respondent burden is reasonable 6. To detemiine if. the scenarios are sufficiently realistic 7. To detennine if the setting is suitable to complete the survey 8. To test recruitment and anaiysis procedures

The sutvey was pilot tested in 16 patients at the St. Michael's Hospital MS clhic Neurologists at the c l i c , told eligible patients (relapsing remitting MS, who are over 18 years old and not pemanently wheelchair bound) about the study dunng their appointment. The neurologist introduced interested patients to the primary investigator. The primary investigator gave an information sheet to interested patients, answered fiirther questions about the study and obtained written consent. The investigator took participants to a private examination room in the c h i c to complete the ~el~administered questionnaire (Appendices A and B). Participants were instructed to write down comments and suggestions as they completed the survey. The start and stop time was recorded. A one on one debriefmg session (participant and investigator) followed the completion of the survey. (Debriefing Questionnaire on foilowing box).

Debriefing Questionnaire 1. Were the instructions for the choice tasks clear to you? 2. Were there too rnany characteristics described for each dnig for you to make a choice? 3. Was there sufficient information in each drug characteristic for you to make a choice 4. Were there any words in the choice tasks that were unclear to you? 5. Were there any questions in part 2 that were unctear to you? 6 . On a scale of l(simple) to IO (difficult) how would you rate the difficulty in chooshg a

dnig in each choice task? 7. Did you find that the questionnaire took an u~easonable amount of your time to

complete? 8. On a scale of 1 (least) to 10 (most) how would you rate the level of reaiisrn of the dnig

description in this survey? 9. Was the setting is suitable for you to complete the survey? 10. Do you have any suggestions for how I can make this questionnaire easier for people

such as yourseif to complete?

Page 124: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

Resulis

Participants did not £hd the survey to be burdensome. On average it took participants

23 minutes to complete and they rated the choice tasks a 3.4 out of 10 on an ease to

complete scale (10 being difticult). The instructions were clear to rnost people, but a couple

of participants did not understand right away that they needed to choose one treatment

option per page as opposed to one therapy fiom the entire questionnaire. Also one woman

did not understand the cost attribute as being over and above what insurance pays. Generally,

patients found the survey to be realistic, as they rated the profiles at 6.3 out of 10 on a

redism scale. Results fiom the MM, analysis indicated that ali of the dmg attributes were

statistically significant, at one or more levels, in affeçting respondent choice. Although the

demographic section was eady completed and well understood by respondents, some said

the self-adrninistered EDSS was emotionally difticult to answer at times.

Discussion and Actions Taken

Patients claimed that the once a week capsule was too good a product to be tme.

Also patients did not seem to notice a difference between 50% relapse reduction and 75%

relapse reduction. Based on these comments, two changes were made to the attribute Ievels.

Firstly, the combination of capsule and once a week dosing was prohibited. Secondly, the

three levels of dnig effect on relapses (25%- 50% and 75%) were altered in to two levels of

30% and 60%.

Page 125: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

Code Book for Variables

Page 126: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

Code Book for Variables

Energy Level

Disability

Relapses

Flu-like sym ptoms

Drug Form

Dosage Regimen

Cost

Disease Severity

DMMSD- Experience

Reduces your energy level for the 6rst few months of therapy . No effect on your energy level hproves your energy level

No e f f i on your daiiy activities Slows the decline in daily activities Prevents any decline in daily activities

Less severe and less fiequent by one-third (1/3) Less severe and less fiequent by two-thirds (2/3)

60% chance of having flu-like symptoms 30% chance of having flu-like symptoms No chance of having flu-like symptoms

An injection that ofien causes a skin reaction An injection that does not cause a skin reaction A capsule to be taken by mouth

Once a day 3 times a week Once a week

No cost to you each month You pay $200 each month You pay $500 each month You pay $ 1,000 each month

Mild MS Moderate MS

Expenenced with therapy Naïve to theranv

Page 127: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

Treatment Preferenct Questionnaire

Page 128: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical
Page 129: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

PART 1

Imagine your doctor presents you with 3 treatment options for your multiple sclerosis. Two of

1 the options are drugs while the other option is not to take a drug at all. Although these drugs have

1 been shown to be effective over a Iwo-year period, no information is known about their long-term

1 side effects. Your doctor asks you to choose the option you most prefer and assures you that

1 whichever treatment option you choose, you will still be treated for the symptoms of your relapses

if they occur.

As you read the following pages, please keep the above scenario in mind.

On the leR side of each of the following pages, you will be given a list of drug characteristics

used to describe the 3 treatment options shown in middle of the page. Please place a checkmark

in the box under the treatment option that you would choose for yourself. If you feel that neither of

the drugs would be worth trying then you should select the 'No Drug' option. There are 15 of these

choice questions for you to answer.

Please turn over the page and begin the questionnaire when you feel that you are ready.

Page 130: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

Please place a checkmark in the box, under the ONE option that you think you would choose.

Energy

Oally ~ctlvltlea'

~elapaes'

Slde ~ f f e c t d

Drug ~orm'

~ e ~ l r n o n '

Drug A

Reduces your energy level for the lirst few months of therapy

qo effect on your abllity to do daily acllvlties

Less severe and one- thkd (1 /3) tess ohen

60% chance of gettlng flu-like symptoms

4n lnjectlon thal does not cause a skin reaction

Once each day

No cost to you each month

MY CHOICE IS:

-b

Drug B

Reduces your energy level for the first few months of therapy

JO effect on your ability to do dail) activitles

Less severe and two-thlrds (W3) less ahen -

30% chance of gettlng flu-llke symptoms

A capsule taken by mouth

Once each day

You pay $200 each month

No New Drug

No eff ect on your energy level

+JO effect on your abillty to do dally actlvitles

No effect on severlty or frequency

No chance of gettlng flu-#ke symptoms

Page 131: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical
Page 132: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical
Page 133: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical
Page 134: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical
Page 135: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical
Page 136: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

Please place a checkmark in the box, under the ONE option that you think you would choose.

Energy

Dally ~ctlvltlea'

~ e l a ~ a e r '

Sldr ~tfects'

Drug ~orm'

~eglrnen'

c o d

Drug M

No effect on your energy level

Slows decline in your ability to do daily activltles

Less severe and one-thitd (1/3) less often

60% chance of gettlng Ilu-like symptoms

An injection thal causes a skln reaction

Once each week

No cos1 to you each month

Drug N

Reduces your energy level for tht fint few months of therapy

Slows decline in your abllity to da daily activltles

Less severe and one-third (1/3) less often

pp --

60% chance of getting flu-like symptoms

A capsule taken by mouth

3 ttmes each week - -- --

You pay $500 each monlh

No New Orug

No elfect on your energy level

JO eflect on your ability Io do dail! activities

Uo effect on severity or frequency

No chance ol getting (lu-llke symptorns

MY CHOICE IS:

II,

Page 137: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical
Page 138: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

(CI

Page 139: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

Ptease place a checkmark in the box, under the ONE option that you lhink you would choose.

Energy

Dally ~ctlvltler '

~ e l r p s e 8 ~

Side €f(ects3

Orug ~orm'

~egglmed

art'

No New Drug

No ellecl on your energy level

Uo effect on your ablity to do dail! activities

No effect on severity or lrequency

No chance of getllng flu-lke symptoms

MY CHOICE IS:

II,

-

lmproves your energy level

Vo effect on your abllty to do dail! aclivities

Less severe and Iwo-thlrds (2i3) less often

60% chance of gettlng llu-like symptoms

-

An Injection thal causes a skln reaction

- -

Once each week

Vou pay $200 each monlh

Drug T

No effect on your energy level

Vo effect on your ability to do dail1 activittes

Less severe and Iwo-lhirds (213) less often

-

60% chance of getting flu-like symptoms

An injection that causes a skin reaction

Once each day

You pay $500 each monlh

Page 140: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical
Page 141: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical
Page 142: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

Please place a checkmark in the box, under the ONE option that you think you would choose.

Energy

Dally ~ctlvltled

~ekprer '

Slde ~ffectr'

Drug ~orm'

~ e ~ l n i e n ~

cost6

Drug Y

No effect on your energy level

Prevents any decline In your abllity to do daNy actlvlies

Less severe and two-thirds (2/3) less often

30% chance of getting flu-like sy mptoms

An lnjectlon lhat causes a skln reaction

3 times each week

- -

You pay $200 each month

MY CHOICE IS:

II,

Drug Z

No effect on your energy level

Prevents any decllne in your abillty to do dally activitles

Less severe and two-thirds (213) less often

No chance of gettlng flu-llke symptoms

4n injection that does not cause a skin reactlon

Once each week

No cost to you each rnonth

No New Drua

No effect on your energy level

Vo ellect on your ability to do dail! activitles

- - - - - . .

No chance of getting flu-like syrnptoms

Page 143: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

(II

~ -

(II al e

Page 144: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

Please place a checkmark in the box, under the ONE option that you think you would choose.

Dally ~ctlvltles'

flelapses2

Slde €ffectr3

Drug ~orm'

fteglmenS

corts

Dtua CC Druo DD

lmproves your energy level - - - - - - -

Prevents any decline In your ablllty to do daily aclkllies

Less severe and one-thlrd ( I D ) less otten

30% chance of geîîing flu-like sy mptoms

An injection that causes a skin reaction

Once each day

You pay $500 each rnonlh

MY CHOICE IS:

(II,

No effecl on your energy level

Slows decline in your aMllty to do daily activities

Less severe and Iwo-thhds (2/3) less often

No chance ol getting Ilu-llke sy mptoms

An Injection that causes a skh react Ion

Once each week

You pay $1,000 each month

No New Drug

No ellect on your energy level

Vo elfect on your abillty to do dail! activities

No chance of getting Ilu-like symptoms

Please Continue on to Part 2

Page 145: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

APf ENDIX L

Demographic and Medicd Questionnaire

Page 146: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

Demographic and Medicai Questionnaire

The following pages contain questions about yourself. and your history of multiple sclecosis. Please complete these questions ta help us describe the group of people who completed the questionnaire.

2. Sex: Male Female

3. Year in which a doctor first told you that you have multiple sclerosis

4. What is the highest level of formal education that you have completed?

Less than a high school diploma High school Some college/university College diplomaluniversity degree Graduate degree

-

In this questionnaire, when we Say relapse of multiplesclerosis we mean the appearance of new syrnptoms or the worsening of-old syrnptoms lasting longer than 24 hours, af€er a period in which your fiealth was stable for at least one rnonth. Symptoms migM include double vision, loss of balance, numbness of amis andor legs, or slurred speech. By "relapse" we do mean changes in symptoms, lasting less than 24 hours.

5. How many relapses as defined above have you experienced in the last two years? (If you are unsure, please give your best estimate)

6. When did you experience your most recent relapse? Less than a rnonth ago 1 6 months ago 6-1 2 months ago Greater than 1 year ago

Page 147: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

7. Which of the following new dnigs for multiple sclerosis have you tried?

vone ex^ ~etaseron' copaxonea ~ebif@. An experimental drug None of the above

Only answer question # 8 if you selected the 'None of the above' response to question # 7, otherwise please continue on to question # 9.

8. What is the main reason that you have not tried one of the new drugs named in Question 7 to treat your multiple sclerosis? (You may only choose one answer)

I have not heard of these drugs. I have heard of these dnigs, but I'm not well-enough inforrned about them. I'm currently deciding whether or not to try a new drug. I just received a prescription for a new drug from my doctor, but I have not filled it yet. I have discussed the new drugs with my doctor and I have decided not to try one because

9. What was your total household income before taxes from al1 sources this past year?

Less than $1 0,000 $1 0,000 to $1 9,999 $20,000 to $34,999 $35,000 to $49,999 $50,000 to $74,999 Greater than $75,000

Page 148: PATIENT PREFERENCES FOR DRUG TREATMENTS …...PATIENT PREFERENCES FOR DRUG TREATMENTS FOR MULTIPLE SCLEROSIS. Mark Sttphen Rolnick, MSc, 1999, Faculty of Pharmacy Department of Pharmaceutical

10. How many people live in your household?

--

The new drugs that ara currentiy avaitabE&tetr,eat muMpI8: sderosis must be injected (either every-day, 3 àmes each w d c or once each week). On average, they can reduce the number of relapses that patients experience by one-third (1/3). Also they can slow decline in your ability to do daiiy activities. The common side effects indude flu-like symptoms, reduced energy for the first few months of therapy. and skin reactions around the injection site. You would not necessari(y experience al1 of these side effects if you were taking one of these dnigs.

Suppose your doctor recommended that you take one of these drugs but it was not covered by private or public insurance, so you had to pay for it with your own money.

How much money would you be willing to pay each month for one of these drugs? Please keep in mind how much you are able to pay. - Please check one box in each row.

Yes No

O O O 0 O O 0 0 O O O O O 0 O 0

Less than $1 0 each month

$1 1 to $99 each month

$1 00 to $249 each month

$250 to $399 each month

$400 to $599 each month

$600 to $799 each month

$800 to $999 each month

Greater than $1,000 each month