`
`Dovepress
`
`ORIGINAL RESEARCH
`
`Patient preferences for important attributes of
`bipolar depression treatments: a discrete choice
`experiment
`
`This article was published in the following Dove Press journal:
`Patient Preference and Adherence
`
`Daisy Ng-Mak'
`Jiat-Ling Poon?
`Laurie Roberts?
`Leah Kleinman?
`
`Dennis A Revicki?
`Krithika Rajagopalan!
`'Global Health Economics and
`OutcomesResearch, Sunovion
`Pharmaceuticals Inc., Marlborough,
`MA, *Patient-Centered Research,
`Evidera, Bethesda, MD, USA
`
`Correspondence: Daisy Ng-Mak
`Sunovion Pharmaceuticals Inc.,
`84 Waterford Drive, Marlborough,
`MA 01752, USA
`Tel +1 774 369 7010
`Email daisy.ng-mak@sunovion.com
`
`submit your manuscript
`woes HV OO
`hep
`
`Purpose: The purpose ofthis study wasto assess patient preferences regarding pharmacological
`treatmentattributes for bipolar depression using a discrete choice experiment (DCE).
`Methods: Adult members ofan Internet survey panel with a self-reported diagnosis of bipolar
`depression were invited via e-mail to participate in a web-based DCE survey. Participants were
`asked to choose between hypothetical medication alternatives defined by attributes and levels
`that were varied systematically. The six treatment attributes included in the DCE were time to
`improvement, risk of becoming manic, weight gain,risk of sedation, increased blood sugar, and
`increased cholesterol. Attributes were supported byliterature review, expert input, and results
`of focus groups with patients. Sawtooth CBC System for Choice-Based Conjoint Analysis was
`used to estimate the part-worthutilities for the DCE analyses.
`Results: The analytical sample included 185 participants (50.8% females) froma total of
`200 participants. The DCE analyses found weightgain to be the mostimportant treatmentattri-
`bute (relative importance =49.6%), followed by risk ofsedation (20.2%), risk ofmania (13.0%),
`increased blood sugar (8.3%), increased cholesterol (5.2%), and time to improvement(3.7%).
`Conclusion: Results from this DCE suggest that adults with bipolar depression considered
`risks of weight gain and sedation associated with pharmacotherapyas the most importantattri-
`butesfor the treatmentofbipolar depression. Incorporating patient preferences in the treatment
`decision-making process maypotentially have an impact on treatment adherence and satisfaction
`and, ultimately, patient outcomes.
`Keywords: bipolar depression, treatment preference, adverse events, weight gain
`
`Introduction
`
`Bipolar | disorder is characterized by periods of severe mood episodes thatfluctuate
`between clinical depression, mania, and mixed episodes and are associated with sig-
`nificant disability and functional impairment.' In the US,the lifetime prevalence of
`bipolar I disorderis 1%, with a ]2-month prevalence of 0.6%.’ Patients diagnosed with
`bipolar I disorder have been found to experience depressive symptoms three times
`more often than manic symptoms.** Depressive episodesin bipolar disorder(ie, bipolar
`depression)tendto last longer, occur more frequently, and are associated with higher
`suicide rates and work-related disability compared to manic episodes.*
`Althoughseveral! treatment options are available for the managementofbipolar I
`disorder, there are currently only three US Food and Drug Administration (FDA)-
`approved atypical antipsychotic treatments for bipolar depression: quetiapine,
`olanzapine in combination with fluoxetine, and lurasidone.* These medications have
`also received regulatory approval for the treatment of bipolar depression in other
`
`35
`Patient Preference and Adherence 2018:12 35-44
`© 2018 Ng-Mak ot al This work publithed and licenced by Dove Medical Press Limited. The fall terms cf this cence are available at hippc//wwwdovepress.coe/termipip
`B A and incorporate the Creaive Common Auribetion ~ Nor Commercial (wnported, 13.0) Licerwe (tnip//ceadvecormmans.crplicenses/by-nc/).0/). By accessing the work you
`hereby acaept the Terra.Men-conteneraal user of the werk are permitted wichwu! any farther pemisaon from Dove Medical Press Limned, previded the work i property attrbeted. Fer perraizicn
`for commercial use of this work, please see parngraghs 42 and 5 ef our lemm fhiapes/www.dovepreiscom/termsphp).
`
`1
`
`Exhibit 2084
`Slayback v. Sumitomo
`IPR2020-01053
`
`
`
`Ng-Maket al
`
`Dovepress
`
`localities such as Canada (lurasidone and quetiapine), the
`European Union (quetiapine), and Japan (olanzapine). As a
`class, atypical antipsychotics have unique efficacy and toler-
`ability profiles but are usually associated with considerable
`adverse effects, including weight gain, type 2 diabetes, and
`hyperlipidemia.”
`Approximately 60% of patients with bipolar disorder
`do not sufficiently adhere to their medication.*!! Accord-
`ing to a recent systematic literature review of observational
`studies, one of the most commonly reported reasons for
`medication nonadherence in bipolar disorder is adverse
`effects of treatment, such as weight gain, sedation, tremors,
`and perceived cognitive impairment.'*"" In addition,residual
`depressive symptoms mayalso negatively impact medica-
`tion adherence.'* Using a stated-preference approach, side
`effects of weight gain or cognitive impairment weresimilarly
`identified as major considerations for the treatment of non-
`adherence in bipolardisorder.'> The managementof bipolar
`disorder includes proactive monitoring of these adverse
`effects, such as weight gain, through encouragement of
`lifestyle and behavioral modifications.'°
`Treatment nonadherence in bipolar disorder remains
`a continuous challenge with both clinical and economic
`consequences*!'"’ Nonadherenceis associated with decreased
`treatment effectiveness, increased relapses, escalated mor-
`bidity, and increased hospitalizations and other health care
`utilization'""” which can lead to higher health care costs
`and decreased quality of life. Identifying patient treatment
`preferencesby allowing patients to trade-offthe benefits and
`risks associated with the treatment ofbipolar depression may
`lead to a better understanding of the patients’ perspective
`for both physicians and patients and, ultimately, increase
`medication adherencerates. '*
`
`One way to assess patient preferences is to conduct
`a discrete choice experiment (DCE), a methodology that
`resernbles real-life decision making.'* Ina DCE,participants
`are asked to choose between scenarios describingrealistic
`treatment options and where they maketrade-offs between
`different treatmentattributes. This differs froma survey-only
`approach where patients may be asked to answer questions
`about independenttreatment features, includingsideeffects,
`withouttaking into accountthe trade-offs required to choose
`between multiple treatment characteristics at once.
`A few published studies of DCEs were conducted in
`mental health populations. However, a previous research has
`foundthat patients with severe mentalillness, such as schizo-
`phrenia or major depressive disorder, may be able to appro-
`priately complete DCEtasks and make meaningful decisions
`
`36
`
`submit
`you
`Dovepr
`
`your manuscript
`P
`
`about preferred treatment scenarios based on different
`attributes.'” DCE methodology has previously been used ina
`bipolar disorder populationto assessfactors associated with
`nonadherence to treatment. The results demonstrated that
`
`patients were morelikely to be adherent to medications ifthey
`reduced the severity oftheir depressive episodes and did not
`cause weight gain or cognitive side effects.!’ However, this
`prior work did not focus on bipolar depression.
`The objective of this study was to assesspaticnt prefer-
`ences regarding pharmacological treatment attributes for
`bipolar depression via a DCE.”
`
`Methods
`The DCEinvolved a series of systematic steps, including
`1) developmentoftreatmentattributes, and 2) implementation
`of the DCE. All study activities were conducted in English.
`
`Developmentof treatment attributes
`Relevant treatmentattributes and conceptualizationsoftreat-
`ment scenarios were developed through literature reviews
`and focus groups. A targeted literature review ofarticles
`that described bipolar depression treatments and a review
`of recent product inserts were conducted. PubMed and
`Embase were used to conduct the literature search, and the
`
`search strategies are included as a supplementto the manu-
`script. Product inserts for 29 medications used to manage
`bipolar disorder/depression were reviewed, including nine
`typical antipsychotics, 12 atypical antipsychotics, seven
`anticonvulsants, and lithium. Information related to dosing
`characteristics, need for monitoring, efficacy (eg, time to
`improvement, remission rates), adherencerates, and common
`adverse events was extracted from these review sources.
`
`Following the literature review, one expert clinician
`interview and two focus groups with 16 adult patients”!
`were conducted. The purpose of the expert interview was
`to draw on the clinician’s experiencesto identify key issues
`and concerns in bipolar depression, with a greater emphasis
`on treatmentside effects and reasons for continuing or dis-
`continuing treatment. The interview was conducted using a
`semi-structured interview guide.
`The focus groups enrolled adult participants from two
`clinical sites in the US (n=8 per site). All focus group par-
`ticipants had a clinician-confirmed diagnosis of bipolar |
`disorder, a history of =1 major depressive episode within
`the last 12 months, a lifetime history of =1 manic or mixed
`manic episode, and currently or previously received antipsy-
`chotie drug therapy for bipolar disorder. Mean age of focus
`group participants was 47.9 years (SD =6.0 years), and
`
`Patient Preference and Adherence 2018:! 2
`
`
`
`Dovepress
`
`Patient preferences for the treatment of bipolar depression
`
`69% were females. Mean time since the participants’initial
`bipolar I disorder diagnosis was 15.7 years (SD =| 1.4 years),
`and their mean duration of atypical antipsychotic use was
`4.9 years (SD =4.7 years). The focus groups were conducted
`using a semi-structured interview guide to elicit information
`regarding expectations oftreatment, treatment experiences,
`and potential barriers to treatment for bipolar depression.
`Audio recordings of the focus groups were transcribed and
`analyzed for themes that patients described as being related
`to their expectations and preferences for bipolar depression
`treatment using ATLAS.ti (version 7.5.3).
`The most important medication attributes identified from
`the expert clinician interview and patient focus groups are
`given in Table 1. Efficacy and weight gain were reported
`as important treatment attributes for patients with bipolar
`depression. Patients also defined “time to improvement”
`as the time from treatmentinitiation to when they began to
`observe improvements in their symptoms. Findings from
`the qualitative research were used to determine the relevant
`attributes and attribute levels for the DCE scenarios to be
`
`used in the pilot and main DCEstudies. In determining the
`finallist of attributes, greater emphasis was placed on factors
`identified by patients as being important in influencing their
`treatment decisions. Levels of attributes were determined
`
`based on results of clinical trials reported in the product
`inserts, including incidence rates of each event and time to
`improvement of depressive symptoms.
`
`Implementation of DCE
`The DCE was implementedvia a one-time, cross-sectional,
`web-based survey.Prior to full implementation, one-on-one
`pilot interviews were conducted via web conference.
`
`Participants
`For the pilot and main web-based surveys, members of
`MedPanel,** an Internet survey panel, with self-reported
`bipolar depression were invited via e-mail to participate.
`
`Table | Important medication attributes for the treatment of
`bipolar depression identified via interviews*
`
`Expert clinician
`Patient focus groups
`Efficacy
`Efficacy
`Metabolic side effects
`Increased blood sugar
`Sedation
`Increased cholesterol
`
`Sexual dysfunction
`Weightgain
`
`Note: ‘Medication attributes arelisted alphabetically.
`
`Risk of becoming manic
`Sedation
`Time to improvement
`Weight gain
`
`MedPanelspecializes in the life science industry and main-
`tains a large patient panel across various diseases, includ-
`ing bipolar disorder. MedPanel members were originally
`recruited through patientassociations, patient support groups,
`and physician referrals. Interested patients answereda series
`of screening questions to determine studyeligibility.
`Inclusion criteria were adult subjects (18-75 years), self-
`reported diagnosis of bipolar depression (bipolar I disorder
`with most recently documented depressive episode within the
`last 12 months), lifetime history of =1 manic or mixed manic
`episode, and currently or previously received antipsychotic
`drug therapy for bipolar disorder. Exclusion criteria were hos-
`pitalization for a manic or mixed episode within the 60 days
`prior to screening, participation in any otherclinicaltrial, or
`had received study medication =45 days from study screen-
`ing. Diagnosis of bipolar depression was self-reported by
`paticnts, and symptoms of depression were further verified
`throughpatients’ responses to screening questions related to
`use of medications to managebipolar disorder and symptoms
`of depression and/or mania.
`Institutional review board approval was obtained from
`Ethical and Independent Review Services on October 16,
`2015 (study number: 15127-01) for the study protocol and
`recruitment materials. All participants provided electronic
`informed consent, and eacheligible participant received $20
`for completing the study.
`A pilot study with four participants was conducted using
`the preliminary DCE scenarios to assess the clarity and
`understanding of the web-based survey questions. Based
`onparticipants’ feedback, minor changes were madeto the
`attribute names and the order of tasks. Participants under-
`stood the DCEtask and wereable to complete the web-based
`survey with minimaldifficulty. The final treatmentattributes
`and levels used for the DCE scenarios are given in Table 2.
`In the main DCE,eligible participants completed only the
`web-based survey.
`
`Survey instrument
`Eligible participants were asked to complete upto 10 sets of
`DCEscenarios, sociodemographic andclinical questions,the
`self-reported Montgomery—Asberg Depression Rating Scale
`(MADRS-S),?** the WHO-5 Well-Being Index (WHO-S),>
`and the Patient-Reported Outcomes Measurement Informa-
`tion System (PROMIS) Global Health Instrument (GHI).**
`In each DCE scenario, two hypothetical bipolar depres-
`sion medications comprising different attributes (time to
`improvement, risk of becoming manic, weight gain, risk of
`sedation, increased blood sugar, and increased cholesterol)
`
`Patient Preference and Adherence 20 18:12
`
`submit your manuscript
`
`37
`
`
`
`Ng-Maket al
`
`Table2 Final DCEattributes and levels
`
`Dovepress
`
`Attribute
`Time to
`improvement
`
`Risk of becoming
`manic
`
`Description
`The time until you feel an improvement in your
`depressive symptoms (ie, sadness, crying, feeling
`of isolation) after you start taking the medication
`The chancethat taking the medication when you
`are in a depressive episode will cause you to
`become manic instead
`
`Weightgain
`
`The amountof weight gain you will experience
`after taking the medication
`
`Levels
`| week
`2 weeks
`4 weeks
`Fewer than one in 100 of depressed patients will switch to
`being manic after taking the medication
`Five in |00 of depressed patients will switch to being manic
`after taking the medication
`Eight in 100 of depressed patients will switch to being
`manic after taking the medication
`Patients experience a minimum weight gain of less than
`3lbs after taking the medication
`Patients experience an average weightgain of 3-10 Ibs after
`taking the medication
`Patients experience an average weight gain of 10-20 Ibs
`after taking the medication
`Patients experience an average weight gain of more than
`20lbs after taking the medication
`Fewer than |0in 100 of patients will experience excessive
`sleepiness or drowsiness after taking the medication
`10-24in 100 patients will experience excessive sleepiness
`or drowsinessafter taking the medication
`25-50in 100 patients will experience excessive sleepiness
`or drowsiness after taking the medication
`More than 50 in |00 patients will experience excessive
`
`sleepiness or drowsiness after taking the medication
`Fewer than five in 100 patients will experience increased
`bloodsugar (glucose) after taking the medication
`10-15 in 100 patients will experience increased blood sugar
`(glucose) after taking the medication
`Fewer than five in 100 patients will experience increased
`cholesterol levels after taking the medication
`10-15 in 100 patients will experience increased cholesterol
`levels after taking the medication
`Abbreviation: DCE, discrete choice experiment.
`
`Risk of sedation
`
`The chance that you will experience excessive
`sleepiness or drowsiness after taking the
`medication
`
`Increased blood
`sugar (glucose)
`
`The chance that your blood sugar (glucose) levels
`will change from normal to high after taking the
`medication
`
`Increased cholesterol
`(fat in the blood)
`
`The chance that cholesterol levels will change
`from normal to high after taking the medication
`
`and correspondinglevels for each attribute were presented
`(Table 2 and Figure 1). Participants wereinstructed to review
`the treatment pairings and select the medication they would
`prefer to take at the present time given the options. Each
`participant respondedto only 10 choicepairs in order to both
`minimize the cognitive burden on participants and maximize
`the efficiency of the study design, given the numberofattri-
`butes and levels included in the DCE. Oneofthe discrete
`
`choice scenarios presented was a fixed-choice question and
`was not included in the final analysis. The fixed-choice
`question presented a clearly favorable medication choice to
`establish that participants understood the DCE task. Those
`who respondedincorrectly were excluded from the analysis.
`To prevent potential biases in responses, the fixed-choice
`question waspresented as part of the full set of scenarios.
`In addition to the discrete choicetask, participants were also
`asked to directly rank the six attributes in order of importance
`on a scale of | (most important) to 6 (least important).
`
`The MADRS-Sis a nine-item self-report scale assessing
`depressive symptoms overthepast 3 days.”*** Patients were
`asked to rate the severity ofeach of the symptomsassessed on
`a scale ranging from 0 to 6. The total score for the MADRS-S
`wasthen calculated by summingthe ratings ofthe nine items,
`which ranged between 0 and 54, with higher scores indicating
`greater impairment.
`The WHO-Sis a measure ofemotional well-being devel-
`oped from the World Health Organization-Ten Well-Being
`Index”and consists offive positively worded itemsassessing
`emotional well-being over the past 2 weeks. Each item is
`rated ona 6-pointLikert scale rangingfrom0 (not present) to
`5 (constantly present). Individualitem ratings are summed to
`obtain a raw score ranging from 0 (worst possible quality of
`life) to 25 (best possible quality oflife), which may be trans-
`formedinto a percentage score ranging from 0 (worst possible
`quality of life) to 100 (best possible quality of life). A raw
`score =13 has been found to be indicative of depression.
`
`38
`
`submit your manuscript
`Dovepr
`
`Patient Preference and Adherence 2018:! 2
`
`
`
`Dovep
`
`Patient preferences for the treatment of bipolar depression
`
`Now keeping in mind the features you just read through, please read each option carefully and choose which medication you would prefer for the treatment of your bipolar depression.
`
`Ifyou need to review the description of the medication features again, click here: Glossary
`
`Four weeks
`
`Time to improvement
`
`Risk of becoming manic
`
`Weightgain
`
`Risk of sedation
`
`Increased blood
`sugar (glucose)
`
`Increased cholesterol
`
`
`
`Less than 1% (fewer than 1 in 100) of
`depressed patients will switch to being manic
`after taking the medication.
`
`8% (8 in 100) of depressed patients will
`switch to being manicafter taking the
`medication.
`
`Patients experience an average weight gain
`of 10-20 Ibs after taking the medication.
`
`Patients experience an average weight gain
`of 10-20 Ibs after taking the medication.
`
`Less than 10% (fewer than 10 in 100) of
`patients will experience excessive sleepiness
`or drowsiness after taking the medication.
`Less than 5% (fewer than 5 in 100) of
`patients will experience increased blood
`sugar (glucose)after taking the medication.
`10%-15% (10 to 15 in 100) of patients will
`experience increased cholesterollevels
`after taking the medication.
`
`25%-50% (26 to 50 in 100)ofpatients will
`experience excessive sleepiness or
`drowsiness after taking the medication.
`Less than 5% (fewer than 5 in 100) of
`patients will experience increased blood
`sugar (glucose)after taking the medication.
`Less than 5% (fewer than § in 100)of
`patients will experience increased cholesterol
`levels after taking the medication.
`
`Figure | DCE question sample.
`Abbreviation: DCE, discrete choice experiment.
`
`The PROMIS Global Health Questionnaire** comprises
`10 questions covering the global domainsof physical health
`and mental health. Severity questions assess the respondent's
`currentstate using a responsescale of “excellent, very good,
`good, fair, and poor”. Frequency questions assess the past
`7 days using a responsescale of “never, rarely, sometimes,
`often, and always”.
`
`Statistical analyses of DCE
`Descriptive analyses were conducted on sociodemographic
`and patient-reported outcome questionnaire data. For the
`DCEdata, preference weights (part-worth utility values) were
`estimated using a random-effects multinomial logit model.”’
`The model estimated the probability of a patient choosing
`an alternative 7 (over a set of possible alternatives / in the
`given choice set) with the Bs representing the estimated
`part-worthutilities.
`
`Exp (V(B,X,))
`<=
`>. Exp(ViB,X,))
`
`were scaled to have a mean value of zero and then used to
`
`calculate the relative importance of eachattribute. The rela-
`tive importance of each attribute was then calculated using
`the following formula:
`
`Relative importanve = Overall utility for each attribute
`Total utility
`
`Overall utility value for cachattribute equaled the range
`of part-worth utilities within each attribute, andtotal utility
`value equaled the sumofoverall utility values acrossall attri-
`butes. The relative importance ofeach attribute was expressed
`as a percentage,reflecting the proportionofthe variancein the
`overall medication decision that was accounted for by each
`attribute. Utilities and relative importance were evaluated for
`each DCEattribute. Sawtooth CBC System for Choice-Based
`Conjoint Analysis (version 7; Sawtooth Software, Inc., Provo,
`Utah, USA) wasused to generate the DCE survey questions
`and to estimate the part-worthutilities for the DCE analyses.
`SASstatistical software version 9.4 (SASInstitute Inc., Cary,
`NC, USA) wasused to conductall other analyses.
`
`A positive part-worthutility indicated that the attribute
`level was preferred over levels with negative values, and
`larger part-worth utilities indicated a higher degree ofpref-
`erence for one level over another. The part-worth utilities
`
`Subgroup analyses
`Participant preferences were stratified by gender and age
`(using a median split). The subgroups were determined based
`on a priori hypotheses that there may be genderdifferences
`
`Patient Preference and Adherence 20 18:12
`
`submit your manuscript
`
`39
`
`
`
`Ng-Maket al
`
`Dovepress
`
`Table 3 Demographic andself-reported clinical characteristics of
`DCE web-based survey participants
`Characteristic
`
`N=185
`
`41.7 (13.4) [18-72]
`94 (50.8)
`
`in preferences for the attributes included in the DCE, espe-
`cially weight gain.In addition, it was hypothesizedthat older
`patients may be diagnosed asbipolar for longer and have
`more experience with different bipolar medications, while
`youngerpatients may place a greater emphasis on attributes
`suchas risk of sedation due to work or school productivity
`concerns. Relative importance was calculated separately
`for each age and gender subgroup. Chi-square tests were
`used to determine any significant differences in the relative
`importanceofthe attributes between subgroups.
`
`Results
`Sample characteristics
`A total of 200 eligible participants provided informed con-
`sent and completed the main web-based DCEsurvey.In all,
`11 (5.5%) participants provided an incorrect response to the
`fixed-choice question andfourparticipants (2.0%) “straight-
`lined” their responses (ie, they selected the same response
`option for all questions, indicating that they may not have
`been making decisions but instead trying to complete the
`survey quickly); these participants were excluded from
`the analyses, resulting in 185 participants being included in
`the analytical sample.
`Demographic and self-reported clinical characteristics of
`the participantsare presented in Table 3. The majority of the
`participants (88.6%) were currently receiving medication for
`bipolar depression, most commonly atypical antipsychotics
`(75.1%). Mean MADRS-S total score was 23.9 (SD =9.9),
`
`indicative of moderate depression, and mean WHO-5 raw
`score was 8.9 (SD =5.0), indicative of poor well-being.
`Physical health (T-score =39.1; SD =7.2) and mental health
`(T-score =35.6; SD =7.7) scores, as assessed by the PROMIS
`GHI, were lower than that of the general population (based
`on the standardized population mean of 50), also indicating
`poor health status.”*
`
`Relative importance weights for
`treatmentattributes
`The part-worth utility values are shownin Figure 2. The cor-
`respondingrelative importance indicated that, in the context
`of the included attributes, participants considered weight
`gain as the most importantattribute of a treatment(relative
`importance =49.6%; Figure 2). A treatment associated with
`less weight gain was preferred over a treatment associated
`with more weight gain. In the context of this study, the
`second-most importantattribute was risk of sedation (relative
`importance =20.2%). Treatment associated with lowerrisk
`of sedation waspreferred overtreatment with more sedation.
`
`Age, years, mean (SD) [range]
`Gender, female, n (%)
`Race/ethnicity, n (%)
`American Indian or Alaska Native
`Asian
`Black or African American
`White
`Others*
`More than one”
`Currentliving/domestic situation, n (%)
`Living alone
`Living with others (including friends,
`partner, parents, other family)
`Education, n (%)
`Somecollege and below
`College degree and above
`Household income, n (%)
`<$20,000
`$20,00 |-$40,000
`$40,00 |-$60,000
`$60,00 1|-$80,000
`$80,00 |-$ | 00,000
`>=$100,001
`Current treatment for bipolar depression,‘ n (%)
`Atypical antipsychotics
`Non-atypical antipsychotics
`(eg, antidepressants, mood stabilizers,
`anticonvulsants, stimulants)
`None
`Prior or current treatment with SSRIs or
`SNRIs, n (%)
`MADRS-S total score, mean (SD) [median,
`range]
`WHO-Stotal percentage score, mean (SD)
`[median, range]
`PROMIS global health scale, mean (SD)
`[median, range]
`39.1 (7.2) [39.8, 20-58]
`Physical health summary T-score
`Menta! health summary T-score
`35.6 (7.7) [36.3, 21-63]
`2.7 (0.9) [3, I-5]
`General health
`2.2 (1.0) [2, I-5]
`Social health
`Notes: ‘Other race referred to Mexican. Multiple races were selected:
`Alaska Native and White. ‘Responses are not mutually exclusive.
`self-reported
`Abbreviations: DCE, discrete choice
`experiment; MADRS-S,
`Montgomery-Asberg Depression Rating Scale; PROMIS, Patient-Reported Outcomes
`Measurement Information System; SNRI, serotonin-norepinephrine reuptake inhibitor;
`SSRI, selective serotonin reuptake inhibitor; WHO-5, WHO-5 Well-Being Index
`
`3 (1.6)
`4 (2.2)
`13 (7.0)
`158 (85.4)
`| (0.5)
`6 (3.2)
`
`42 (22.7)
`143 (77.3)
`
`124 (67.0)
`61 (33.0)
`
`64 (34.6)
`56 (30.3)
`24 (13.0)
`19 (10.3)
`9 (4.9)
`13 (7.0)
`
`139 (75.1)
`96 (51.9)
`
`21 (11.4)
`38 (20.5)
`
`23.9 (9.9) [24, 2-50]
`
`35.7 (19.9) [32, 0-84]
`
`Indian/
`
`The remaining attributes, in order of relative importance,
`were risk of becoming manic, increased blood glucose,
`increased cholesterol, and time to improvement. Part-worth
`utilities were in the expected direction for the levels within
`eachofthese attributes, with greater preference forless severe
`adverse events and faster improvement.
`Participant preferences were further stratified by gen-
`der and age (using the median split of =40 years [n=93]
`
`40
`
`submit your manuscript
`Dover
`
`Patient Preference and Adherence 2018:! 2
`
`
`
`Dovepress
`
`150
`
`125
`
`-75
`
`-100
`
`125
`
`-175
`
`Patient preferences for the treatment of bipolar depression
`
`
`
`Morethan20Ibs+
`
`
`
`Fewerthan10in100
`
`3-10Ibs
`
`10-20Ibs
`
`
`
`Lessthan3Ibs
`
`Weight gain
`(RI =49.6%)
`
`5in100
`
`8in100
`
`10-24in100
`
`25-50in100
`
`
`
`Morethan50in100
`
`
`
`Fewerthan1in100
`
`10-15in100
`
`
`
`Fewerthan5in100
`
`
`
`Fewerthan5in100
`
`10-15in100
`
`1week
`
`2weeks
`
`4weeks
`
`Risk of sedation
`(RI =20.2%)
`
`Risk of becoming
`manic
`(RI =13.0%)
`
`Increased Increased
`blood sugar cholesterol
`(RI =8.3%)
`(RI =6.2%)
`
`Time to
`improvement
`(RI =3.7%)
`
`Figure 2 Part-worth utility values (N=185).
`Notes: Part-worth utility values scaled within each attribute to have a mean of 0. Positive part-worthutility value indicates that the attribute level is preferred overlevels
`with negative values. Larger part-worth utility values indicate a higher degree of preference for one level over another. Error bars denote standard errors.
`Abbreviation: Rl, relative importance.
`
`and >40 years [n=92]). Within each subgroup,the relative
`importance of each attribute in rank order was the same as
`that of the overall sample. Descriptively, females placed
`greater relative importance on weight gain (53.2%) than
`males (45.9%), while slightly greater relative importance
`was placed on all other attributes by males than by females.
`Theattributes “increased cholesterol” and “time to improve-
`ment” werebothfifth in relative importance in the >40 years
`subgroup. A chi-square test comparing relative importance
`between subgroups (male vs female, =40 years vs >40 years)
`found nostatistically significant differences in the relative
`importance ofeachattribute across gender or age groups.
`Whenasked to directly rank the six attributes included
`in the DCE in order of importance (1= most important; 6=
`least important), the participants ranked weight gain as the
`most important, with a mean rank of 2.4 (SD =1.6), followed
`by risk of becoming manic (mean rank =3.0 [SD =1.7]),
`time to improvement (mean rank =3.1 [SD =1.6]), risk of
`sedation (mean rank =3.7 [SD =1.7]), increased blood sugar
`(mean rank =4.2 [SD =1.4]), and increased cholesterol (mean
`rank =4.6 [SD =1.3]).
`
`Discussion
`This is the first study to examine patient preferences in the
`treatment of bipolar depression using a DCE methodology.
`In this DCE study, weight gain, risk of sedation, and risk of
`becoming manic wereidentified as the three most important
`attributes for adult participants with bipolar depression.
`In contrast, when time to improvement wasconsidered along-
`side these side effects, it was ranked the least important. These
`findings were consistent across age and gender subgroups.
`While this study wasnot designed to assess factors that would
`increase patients’ adherenceto treatment,it identified factors
`that patients consider importantin influencing their treatment
`decisions. These factors can be used by health care provid-
`ers as a starting point to initiate treatment discussions, and
`discussions around treatment adherence, with patients.
`The importance of tolerability in influencing patient
`treatment preferences emerges as a consistent theme when
`examiningthe results of this study with other studies con-
`ducted in mental! health populations using DCE methodol-
`ogy, including bipolar disorder, major depressive disorder,
`and schizophrenia.'™'8?**! Consistent with the literature,
`
`Patient Preference and Adherence 20 18:12
`
`submit your manuscript
`
`4l
`
`
`
`Ng-Maket al
`
`Dovepress
`
`treatment compared to placebo were observed as early as
`week |. In contrast, the time to improvementin the ranking
`exercise did not specify the range oftime for improvement,
`and participants ranked eachattribute independently of the
`others, without having to maketrade-offs betweenattributes
`and their levels of importance. If the range of levels for
`the time to improvement in the DCE hadincluded a more
`extended time period to improvement, time to improvement
`may have beenidentified as relatively more important.
`
`our findings demonstrated that tolerability in general, and
`weight gainin particular, is an important treatmentattribute
`for patients with bipolar disorder or bipolar depression.'*2"
`As weight gain amongpatients taking atypical antipsychotics
`is a major concern, a personalized treatment witha relatively
`lowerrisk of weight gain may increase treatment adherence
`and thereby improve patient outcomes.*?*?
`Riskof sedation was identified as the second most impor-
`tant