Adv Ther (2020) 37:1479–1495 https://doi.org/10.1007/s12325-020-01258-5 ORIGINAL RESEARCH What Matters Most to Patients and Rheumatologists? A Discrete Choice Experiment in Rheumatoid Arthritis Cesar Dı́az-Torné . Ana Urruticoechea-Arana . José Ivorra-Cortés . Silvia Dı́az . Tatiana Dilla . José Antonio Sacristán . José Inciarte-Mundo . Marta Comellas . Miriam Prades . Luis Lizán Received: January 16, 2020 / Published online: February 22, 2020 Ó The Author(s) 2020 ABSTRACT Seven attributes with 2–4 levels were presented in eight scenarios. Attribute utility and relative Introduction: To determine patient and importance (RI) were assessed using a condi- rheumatologist preferences for rheumatoid tional logit model. Patient preferences for SDM arthritis (RA) treatment attributes in Spain and were assessed using an ad hoc questionnaire. to evaluate their attitude towards shared deci- Results: Ninety rheumatologists [52.2% sion-making (SDM). women; mean years of experience 18.1 (SD: Methods: Observational, descriptive, explora- 9.0); seeing an average of 24.4 RA patients/week tory and cross-sectional study based on a dis- (SD: 15.3)] and 137 RA patients [mean age: crete choice experiment (DCE). To identify the 47.5 years (SD: 10.7); 84.0% women; mean time attributes and their levels, a literature review since diagnosis of RA: 14.2 years (SD: 11.8) and and two focus groups (patients [P] = 5; time in treatment: 13.2 years (SD: 11.2), mean rheumatologists [R] = 4) were undertaken. HAQ score 1.2 (SD: 0.7)] participated in the study. In terms of RI, rheumatologists and RA patients viewed: time with optimal QoL: R: 23.41%/P: 35.05%; substantial symptom Enhanced Digital Features To view enhanced digital features for this article go to https://doi.org/10.6084/ improvement: R: 13.15%/P: 3.62%; time to m9.figshare.11830479. onset of treatment action: R: 16.24%/P: 13.56%; severe adverse events: R: 10.89%/P: 11.20%; Electronic supplementary material The online version of this article (https://doi.org/10.1007/s12325- mild adverse events: R: 4.16%/P: 0.91%; mode 020-01258-5) contains supplementary material, which is of administration: R: 25.23%/P: 25.00%; and available to authorized users. added cost: R: 6.93%/P: 10.66%. Nearly 73% of C. Dı́az-Torné J. A. Sacristán  J. Inciarte-Mundo Hospital de la Santa Creu i Sant Pau, Barcelona, Medical Department, Lilly, Spain Spain M. Comellas  M. Prades  L. Lizán A. Urruticoechea-Arana Outcomes’10, Castellón de la Plana, Spain Hospital Can Misses, Ibiza, Spain L. Lizán (&) J. Ivorra-Cortés Department of Medicine, Jaume I University, Hospital Universitario y Politécnico La Fe, Valencia, Castellón de la Plana, Spain Spain e-mail: lizan@outcomes10.com S. Dı́az  T. Dilla HO&RWE Department, Lilly, Spain 1480 Adv Ther (2020) 37:1479–1495 RA patients were involved in treatment deci- sion-making to a greater or lesser extent; how- For patients, attributes such as time with ever, 27.4% did not participate at all. optimal quality of life and treatment Conclusion: Both for rheumatologists and mode of administration were found to be patients, the top three decision-making drivers determinants in treatment selection. are time with optimal quality, treatment mode Most of RA patients were willing to be of administration and time to onset of action, more involved in the treatment decision- although in different ranking order. Patients making process. Although nearly 73% of were willing to be more involved in the treat- RA patients indicated that in real-world ment decision-making process. practice, they were involved in treatment decision-making to a greater or lesser Keywords: Conjoint analysis; Discrete choice extent, 27.4% of patients were not experiment; Patient perspective; Preferences; involved in the decision, with the Rheumatoid arthritis; Rheumatologist rheumatologists driving the decisions. perspective; Rheumatology; Share decision Key Summary Points Why carry out this study? INTRODUCTION Treatment characteristics may determine its convenience and thereby have impact Rheumatoid arthritis (RA) is one of the most on patients’ adherence and health-related prevalent autoimmune chronic inflammatory quality of life. Since patients are more diseases [1, 2]. If sub-optimally treated, RA can likely to be satisfied and adhere to a lead to joint damage [2], causing different treatment that is in line with their degrees of disability, loss of quality of life (QoL) preferences, a patient-centered care and even increased mortality [3]. approach may have significant impact on In recent years, there have been important treatment outcomes. advances in the management and treatment of RA, which have resulted in better patient prog- Shared decision-making between noses [2, 3]. Several efficacious agents are physician and patient is a key component available for RA treatment, including conven- of the patient-centered care. The first step tional synthetic (cs) disease-modifying anti- in shared decision-making is determining rheumatic drugs (DMARDs), biologic treatment preferences from patients’ (b) DMARDs and targeted synthetic (ts) perspective. DMARDs [4]. RA therapies vary in their mechanisms of The aim of this study was to determine the action and in other characteristics, such as route preferences of patients and and frequency of administration or necessity for rheumatologists for the attributes of RA laboratory monitoring. These characteristics treatments and to evaluate their attitude may determine the convenience of the therapy towards shared decision-making. and thereby have impact on patients’ adherence What was learned from the study? and QoL. Since patients are more likely to be satisfied and adhere to a treatment that is in line Both for rheumatologists and patients, the with their preferences [5, 6], a patient-centered top three decision-making drivers were care approach, defined as ‘providing care that is found to be time with optimal quality, respectful of and responsive to individual treatment mode of administration and patient preferences, needs, and values, ensuring time to onset of action, although in that patient values guide all clinical decisions,’ different ranking order. may have significant impact on treatment Adv Ther (2020) 37:1479–1495 1481 outcomes [7]. In RA, patient and physician METHODS assessment of disease severity and treatment response often do not align, suggesting that Design they focus on different aspects of the disease [8–11]. Values assigned by patients to their This was an observational, descriptive, explora- health status may be strongly driven not only tory and cross-sectional study based on a DCE. by clinical aspects such as functional status or The study was performed within the Spanish symptoms, but also by their beliefs and expec- healthcare public system from September 2017 tations. Involving patients in the decision- to February 2018. The DCE was conducted in making process is crucial, as they must trade off accordance with International Society for the perceived benefits of the different treat- Pharmacoeconomics and Outcomes Research ments with the potential negative conse- (ISPOR) good practice recommendations for quences. In line with this argument, European conjoint analysis in healthcare [21]. League Against Rheumatism (EULAR) makes the A steering committee constituted by three case for shared decision-making between the Spanish experts in RA (CDT, JIC, AUA) led the patient and the rheumatologist, ensuring project. inclusion of the patient’s preferences when choosing a suitable medication [12]. In fact, several studies suggest that the number of Study Participants patients who prefer to participate in decisions during the medical encounter has increased The study population included RA patients and [13]. Hence, it is becoming increasingly impor- rheumatologists with experience in the man- tant to inform the patient about the different agement of RA. Study participants were invited therapeutic options and to offer them the pos- to participate by emails sent by the Spanish sibility to actively participate in the decision- Patient Advocacy Group Coordinadora Nacio- making process [14], taking into account their nal de Artritis (ConArtritis) (RA patients) and perspective and preferences. the Spanish Society for Rheumatology (SER) Discrete-choice experiments (DCEs) have (rheumatologists). become the most frequently applied approach Patients 18 years or older, on DMARD treat- to assess patients’ preferences for treatment ment for at least 12 months and who gave their characteristics in health care research [15, 16]. consent to participate were included in the In a DCE, individuals are asked to choose their study. The participating rheumatologists had to preferred option among different (hypothetical) have at least 3 years of experience in RA man- alternatives. This method is based on two agement and work in the Spanish Health assumptions: (1) that treatments can be descri- System. bed in terms of a set of attributes (characteris- As per the approach proposed by Orme [22], tics) with varying levels and (2) that the priority the minimum sample size necessary for the DCE given by an individual to treatments depends was based on an estimate of proportion. The on the nature and level of the attributes that criterion of maximum variability was applied, compose them [15]. This methodology has been with a 95% confidence interval and 10% margin used successfully in the past to determine of error. Patient sample size was estimated on patient and/or physician preferences for the the basis of the adult population in Spain in attributes of DMARDs in rheumatic diseases 2016 (37,408,739) [23] and RA prevalence [17–20]. (0.5%) [24]. The sample size for the rheuma- The aim of this study was to determine the tologists was determined using the estimated preferences of patients and rheumatologists for number of rheumatologists practicing in the the attributes of RA treatments in Spain and to public Spanish Health System (629) [25]. A evaluate their attitude towards shared decision- minimum sample of 96 RA patients and 83 making (SDM). rheumatologists was required. 1482 Adv Ther (2020) 37:1479–1495 Discrete Choice Experiment The interpretation of the qualitative analysis and the analysis of the ranking exercises Selection of Attributes and Levels allowed identifying the most important In DCEs, patients choose between two hypo- attributes. thetical treatment alternatives described by After the patient focus group, one with four attributes (characteristics) and their corre- experienced rheumatologists (including the sponding levels (different possible values of the members of steering committee) was con- attributes) [15]. To select the attributes and ducted. The objectives of this focus group were levels for the DCE, three consecutive steps were to discuss the relevance of the attributes iden- conducted: (1) literature review; (2) RA patient tified in the literature review and proposed by focus group discussion; (3) rheumatologist patients from the focus group and define the focus group discussion. attributes and levels to be included in the DCE. As a result of the literature review and the Literature Review two focus groups, seven attributes composed of Key terms related to the disease, treatment and two or four levels each were selected (Table 1). stated-preferences studies were used to search the international Pubmed/Medline database. Experimental Design Publications referring to patient and physician The combinations of attributes and levels that preferences in relation to RA treatment as well defined each treatment pair were determined by as those that referred to their perspectives on an experimental design developed according to the management of the disease were consulted. ISPOR recommendations. The DCE design Articles published in Spanish or English up to 9 encompassed two properties: orthogonality and March 2016 were reviewed. balance [21]. The orthogonal design guarantees The results of the literature review [26] were that all attribute levels vary independently, and used to provide inputs for discussion in both the balance design ensures that each attribute focus groups. level occurs the same number of times. The pairs of choice (Fig. 1) were generated by the Focus Groups (Patients and Rheumatologists) mix and match algorithm [27]. To avoid domi- Following the literature review, two focus nance between alternatives, the resulting sce- groups, one with RA patients and one with narios were evaluated for dominated rheumatologists, were used to validate and alternatives. assess the relevance of the attributes and levels A total of eight scenarios were created, which identified in the literature as well as to identify formed a single block. Additionally, an initial attributes not previously described but that control scenario, in which one treatment was were relevant to the Spanish RA population. clearly superior to the other (dominant option), A total of five RA patients, invited by the was included. Participants who answered this patient advocacy group ‘‘ConArtritis,’’ partici- question incorrectly were excluded from anal- pated in the patient focus group. After the ysis as this indicates that they did not compre- completion of a brief questionnaire on hend what is required from them in this study sociodemographic and clinical characteristics, [28]. the list of attributes and levels derived from the literature review was presented to the patients Survey Instrument to discuss them. During the discussion, patients Two online surveys were generated, one for were encouraged to add new attributes and patients and one for rheumatologists. Both levels not previously identified in the literature contained the same DCE choice scenarios and review, but relevant for them. When all attri- included an information form and an electronic butes and levels were identified, a ranking informed consent form that had to be read and exercise was then performed to determine the accepted before completing the questionnaire. relevance of the attributes and levels proposed. In addition, both questionnaires initially Adv Ther (2020) 37:1479–1495 1483 Table 1 Attributes and levels used in the discrete choice experiment Attribute Level 1 Level 2 Level 3 Level 4 Time with optimal QoL 10 years 7 years 5 years 2 years Substantial improvement of RA 9 out of 10 patients 7 out of 10 patients 5 out of 10 patients 3 out of 10 symptoms patients Onset of treatment action 7 days 15 days 1 month 3 months Severe adverse events 1 out of 100 5 out of 100 patients patients Mild adverse events 10 out of 100 30 out of 100 patients patients Mode of administration Daily oral Weekly Monthly Monthly subcutaneous subcutaneous intravenous Additional cost/month for treatment 0 10 20 50 QoL quality of life, RA rheumatoid arthritis included a series of questions to verify that To assess the utility and the relative impor- participants met the selection criteria. tance (RI) value given to the attributes of RA The rheumatologist questionnaire included a treatments by patients and rheumatologists, a set of sociodemographic and professional vari- conditional logit model [31] was used. Respon- ables to characterize them. The patient ques- dents who did not select the dominant option tionnaire included sociodemographic and in the control scenario were excluded. Sub- clinical variables, and a Health Assessment stantial improvement of RA symptoms, time questionnaire (HAQ) to assess the patient’s with optimal QoL, severe and mild adverse functional status [29]. A set of ad hoc questions events and additional cost per month attributes was also included to collect the patient’s per- were linearly transformed. Coefficients ception of: (1) the current degree of involve- obtained in the conditional logit model repre- ment in treatment decision-making and their sented the partial utilities, i.e., the preference expectations about their involvement [30]; (2) for each level within each attribute. A statisti- the satisfaction with the information received cally significant coefficient indicates that the about the disease, current treatment and ther- attribute level influences the respondents’ apeutic alternatives (Likert scale: 1 = not at all treatment decisions. The RI of each attribute, satisfied; 5 = very satisfied). defined as the relative preference weight for the attribute over all attributes, was calculated as Statistical Analyses the quotient between the range of the partial utility values of the attribute and the sum of the Stata version 14 and R version 3.4.1 were used partial utility values ranges of the whole set of for the statistical analysis. A value of p \ 0.05 attributes. The greater the RI among the seven was considered significant for all statistical tests. attributes, the more significant the attribute was For the descriptive analysis of the qualitative for decision-making. variables, the relative and absolute frequencies To evaluate which characteristics of the par- were calculated, and for the quantitative vari- ticipants influenced decision-making, a hierar- ables central tendency and dispersion measures chical cluster analysis was applied to each group were used for each group of participants. of participants based on DCE response, i.e., the scenario selected in each pairs of choice [32]. Since scenario choices were dichotomous, a 1484 Adv Ther (2020) 37:1479–1495 Fig. 1 Example of pair of choice (scenario) presented to RA patients binary distance was applied. Stats and Nbclust associated with a clinical benefit attribute packages were used to determine the optimal (substantial improvement of RA symptoms) and number of clusters [33, 34]. To assess differential the utility associated with risk (severe adverse characteristics between the clusters obtained, events) [35]. sociodemographic and clinical characteristics of To establish differences between the each cluster were compared. patient’s current role in the decision-making Maximum acceptable risk (MAR) was esti- process and their expectations about their mated as the quotient between the utility involvement in it, the answers given in the ad Adv Ther (2020) 37:1479–1495 1485 hoc questionnaire [30] were compared using Table 2 Patients sociodemographic and clinical charac- McNemar Bowker’s test [36]. teristics (n = 137) Characteristic % of patients or Statement of Ethics Compliance mean Age, years, mean (SD) 47.5 (10.7) This study was conducted in accordance with the principles of the Declaration of Helsinki. It Female, % 84 was developed to ensure that Good Clinical Marital status, % Practices were observed, in keeping with ICH Harmonized Tripartite Guideline principles. Married/in a relationship 67.9 The study protocol was submitted to the Span- Single/separated/divorced/widower 32.1 ish Agency of Medicines and Medical Devices and to the Clinical Research Ethics Committee Level of education, % of Puerta de Hierro-Majadahonda Hospital for Primary school 10.2 approval. An electronic informed consent form Secondary school 32.1 was read and accepted by all participants before completing the questionnaire. University or higher 51.1 Other 6.6 RESULTS Employment status, % Employed, full or part time 47.7 Participants’ Characteristics Non-employeda 52.3 A total of 94 rheumatologists and 153 RA Incomes, % patients participated in the study. Due to incomplete data or incorrect responses to the B 2500€/month 83.9 control scenario, 4 rheumatologists and 16 RA [ 2500€/month 1.5 patients were excluded from the analysis. A total of 90 rheumatologists [52.2% women; Unknown/not answered 14.6 mean years of experience in RA 18.1 years (SD: Time since diagnosis, years, mean 14.2 (11.8) 9.0); seeing an average of 24.4 RA patients/week (SD) (SD: 15.3); 75.6% working in a 200 - 1000 bed hospital] and 137 RA patients [mean age: Time in RA treatment, years, mean 13.2 (11.2) 47.5 years (SD: 10.7); 84.0% women; mean time (SD) since diagnosis of RA 14.2 years (SD: 11.8) and Time in current treatment, years, 6.2 (7.8) time in treatment 13.2 years (SD: 11.2), mean mean (SD) HAQ score 1.2 (SD: 0.7)] were included in the final data analysis (Table 2). Mode of administration, % Oral 21,9 Rheumatologists and Patients’ Preferences Subcutaneous 18.3 for RA Treatment Attributes Intravenous 4.4 Partial Utilities Oral ? subcutaneous 46.7 The main results of the DCE are shown in Oral ? intravenous 8 Table 3. Partial utilities denote the importance assigned to each level within an attribute. A Subcutaneous ? intravenous 0.7 positive partial utility for an attribute level Other medicines received/day, mean 4.0 (3.4) indicates a preference for that level over the (SD) reference level, while a negative partial utility 1486 Adv Ther (2020) 37:1479–1495 Table 2 continued The onset of treatment action, probability of severe adverse events and additional cost/- Characteristic % of patients or month for treatment had a similar weight in the mean choice (RI: 13.6%, 11.2% and 10.7%, respec- tively). Overall, during the treatment decision- HAQ, mean (SD) 1.2 (0.7) making process, patients conferred much more RA rheumatoid arthritis, SD standard deviation, HAQ importance to treatment benefit-related attri- Health Assessment Questionnaire butes (time with optimal QoL ? substantial a Included: long-term sick-leave/disabled/unemployed/ improvement of RA symptoms) than to safety student/household/retired attributes (severe adverse events and mild adverse events). For rheumatologists, the mode of administration was the most important implies a lesser preference. Accordingly, the attribute followed by time with optimal QoL higher the partial utility, the greater the pref- and the onset of treatment action (Fig. 2). erence. It was not possible to directly compare For rheumatologists and patients, treatment partial utility values between attributes. The benefits (‘time with optimal QoL’ and ‘sub- attribute levels with a statistically significant stantial improvement of RA symptoms’) influence in patient/rheumatologist treatment accounted for 38.6% and 36.6% of the decision, choice are those with a p value \ 0.05. respectively. For both, these attributes had a Partial utility of the attributes that were then greater weight in the decision than those rela- linearly transformed (time with optimal QoL, ted to safety (patients: 38.7% vs. 12.1%; RA symptoms, severe and mild adverse events rheumatologists: 36.6% vs. 15.1%). It is impor- and additional cost per month) must be inter- tant to note that, for patients, time with opti- preted as one unit increases, 1 year for time with mal QoL provided by the treatment, and its optimal QoL attribute or a 1% increase in the mode of administration determined 60% of the proportion of patients for the rest. decision, while for rheumatologists, the deci- Partial utilities of the levels of onset of sion was driven mainly by time with optimal treatment action and mode of administration QoL, mode of administration and onset of are negative. This means that these levels are action (Fig. 3). less preferred than the level of reference: 7 days and daily oral administration, respectively. Cluster Analysis With respect to the onset of treatment action, Based on the choices made in the DCE, i.e., 1 month is statistically less preferred by patients scenario selected by participants in each pairs of and rheumatologists. For the latter, 3 months is choices, the patients’ sample was segmented also statistically significant. Notice that for into two subgroups (clusters). The analysis of patients, weekly subcutaneous and monthly the patients’ characteristics of the two clusters intravenous administrations were statistically only showed statistically significant differences less preferred than daily oral administration, for the route of administration (p = 0.008) while for rheumatologists only weekly subcu- (Supplementary data Table S1). The estimates of taneous administration was significantly less the utility values of the different routes of preferred. treatment administration according to their The probability of mild adverse events rela- current route of administration suggest that the ted to treatment and the associated additional current route of treatment administration may cost/month were not identified as treatment influence patient preference, showing a trend decision-making drivers. towards a greater preference for the route of administration in which the patient is currently Relative Importance receiving treatment (Supplementary data For patients, time with optimal QoL was the Figure S1). most important attribute in RA treatment No clusters were identified for selection followed by mode of administration. rheumatologists. Adv Ther (2020) 37:1479–1495 1487 Table 3 Patient and rheumatologist DCE results Attribute Level Patients (n = 137) Rheumatologists (n = 90) Partial SE p value* Partial SE p value* utility utility Time with optimal QoL 1 year 0.244 0.036 < 0.001 0.195 0.045 < 0.001 10 years 2.44 1.951 7 years 1.708 1.366 5 years 1.22 0.976 2 years 0.488 0.39 a Substantial improvement of RA 1% patients 0.003 0.003 0.251 0.015 0.004 < 0.001 symptoms 9 out of 10 patients 0.303 1.35 (90%) 7 out of 10 patients 0.235 1.023 (70%) 5 out of 10 patients 0.168 0.731 (50%) 3 out of 10 patients 0.101 0.438 (30%) Onset of treatment action 7 days (ref) 0.000 – – 0.000 – – 15 days - 0.372 0.285 0.193 - 0.602 0.356 0.091 1 month - 0.562 0.255 0.027 - 1.083 0.326 0.001 3 months 0.194 0.275 0.481 - 0.715 0.342 0.036 Severe adverse events 1% patientsa - 0.156 0.058 0.007 - 0.182 0.074 0.014 1 out of 100 patients - 0.156 - 0.182 (1%) 5 out of 100 patients - 0.781 - 0.908 (5%) Mild adverse events 1% patientsa 0.003 0.014 0.856 - 0.014 0.018 0.444 10 out of 100 patients 0.025 - 0.139 (10%) 30 out of 100 patients 0.076 – 0.416 (30%) Mode of administration Daily Oral (ref) 0.000 – – 0.000 – – Weekly subcutaneous - 1.394 0.372 < 0.001 - 1.683 0.461 < 0.001 Monthly subcutaneous - 0.375 0.364 0.302 - 0.249 0.442 0.574 Monthly intravenous - 0.956 0.366 0.009 – 0.446 0.458 0.331 1488 Adv Ther (2020) 37:1479–1495 Table 3 continued Attribute Level Patients (n = 137) Rheumatologists (n = 90) Partial SE p value* Partial SE p value* utility utility Additional cost/month for Additional 1€/month - 0.012 0.007 0.102 – 0.009 0.009 0.296 treatment 0 0 0 10 - 0.119 – 0.092 20 - 0.238 - 0.185 50 - 0.594 - 0.462 Bold values indicate that there is statistical significance (p\0.05) QoL quality of life, RA rheumatoid arthritis, SE standard error a Refers to a 1% of increase of the proportion of patients Fig. 2 Relative importance of attributes for patients and rheumatologists Maximum Acceptable Risk Patients’ Perception of Their Involvement Differences between patients and rheumatolo- in Treatment Decision-Making gist were observed regarding the maximum acceptable risk. While rheumatologists were Statistical differences were observed between willing to accept an increase of 0.08% of risk to the RA patients’ current role and the expected suffer a severe adverse event for a 1% increase in role in the treatment decision-making process the chance of a mayor symptom improvement, (McNemar-Bowker’s test; p \ 0001). In real- patients’ maximum acceptable risk was lower word practice nearly 73% of patients were (0.02%). involved in the treatment decision-making to a greater or lesser extent: 17.8% made the deci- sion together with their doctor, 23.0% Adv Ther (2020) 37:1479–1495 1489 Fig. 3 Relative importance of attributes for patients and rheumatologists evaluated the different treatment options toge- DISCUSSION ther with the doctor and the patient was the one who made the decision, and 31.8% evalu- Patients and physicians have different percep- ated the different treatment options together tions and beliefs about health and illness [8–11]. with the doctor but the doctor was the one who Adequate understanding of patients’ health made the decision. However, 27.4% of patients perceptions and risk tolerance may assist were not involved in the decision (the physicians in the decision-making process at rheumatologists made decisions). Related to the time of the clinical encounters by helping their expectations about their involvement, it is them take into account benefit-risk ratios that important to note that 98.5% of patients were are important to patients [11]. Therefore, in willing to be involved in the treatment deci- recent years, the healthcare systems in devel- sion-making process, and 71.1% of them would oped countries have been evolving into a like to be the ones who make the decision patient-centered care model, in which patients (Fig. 4). take a more active role in making decisions that affect them [37], promoting their involvement Patients’ Satisfaction in the healthcare decision-making process. with the Information Received Their participation can occur at multiple levels: individual (SDM), policy (patient expert on The majority of patients were satisfied with the panels) and commissioning (incorporating information received about the disease (60.3% patient preferences in health technology of patients very satisfied/satisfied) and their assessment or health state evaluation) [16]. current RA treatment (64.7% very satisfied/sat- Stated preference studies, such as DCE, are isfied). However, less satisfaction was reported increasingly advocated as one of the most reli- by patients with information received regarding able and valid techniques available for quanti- therapeutic options (35.8% very satisfied/ fying preferences in health care [15, 16]. satisfied). In RA, the treat-to-target approach has led health care professionals to focus on inflam- matory disease, whereas patients are mainly 1490 Adv Ther (2020) 37:1479–1495 Fig. 4 RA patients’ current and expected role in the decision-making process concerned about pain and fatigue reduction rheumatologists and patients, attributes related and maintaining physical and mental function to treatment benefits are key drivers. Mode of [38]. Better knowledge of patients’ preferences administration has also been identified as a may help healthcare professionals improve dis- decision driver both for RA patients and ease management. By providing information rheumatologists. This result supports findings related to patients’ most salient concerns, of previous studies that show that mode of treatment selection can be in line with their administration (route and frequency) has a preferences and therefore, improve their satis- strong impact on patients’ decisions, with oral faction and adherence to treatment. administration being the most preferred Several studies have investigated preferences [17, 19, 39]. Few studies have studied the of RA patients [17–20, 39–43]; however, none of influence of onset of treatment action on them include an attribute related to treatment treatment decision-making [46]. Our results benefits defined as time with optimal QoL. suggested that a faster time for the drug to start There is ample evidence that RA negatively working is preferred, and this attribute may impacts patients’ QoL [44]. In fact, previous impact treatment selection, mainly for works have found that many RA patients value rheumatologists. QoL more than disease-related variables such as Other studies available in the literature inflammatory biomarkers or joint counts [45]. identified cost [20, 40] and safety [17, 19, 20, 43] Therefore, it is not surprising that our results as main drivers of treatment selection. How- pointed out ‘time with optimal QoL’ as one of ever, in our study, additional treatment cost the most important attributes, achieving a and safety seemed to be less valued, with these higher relative importance than ‘substantial attributes being the ones with the least relative improvement of RA symptoms.’ These results importance. One of the possible reasons for the suggest that patients place greater value on discrepancy observed in relation to the impor- medium- to long-term improvement than a tance of cost in decision-making is the magni- one-off improvement in symptoms. For tude of the difference in costs proposed in the Adv Ther (2020) 37:1479–1495 1491 DCE. In our study, the additional cost ranged its treatment options and guiding patients from €0 to €50/month, while in previous studies through this process [50]. Since PDA provide the cost ranged from €500 to €1500/month [20] detailed information about the therapeutic or from €0 to a threefold increase on health care options, its use in clinical practice would cover taxes (€260/month) [40]. Regarding safety, dif- the demand for more information on thera- ferences may be due to the attributes used to peutic options found in the study. describe safety in previous studies (‘probability This study has some limitations, some of of suffering serious infections’ [20], ‘generalized which are inherent to its design. First, although adverse events’ [20, 40], ‘local adverse events’ DCE is the recommended approach and is [20], ‘reactions at the site of drug administra- widely used to assess patient preferences for tion’ [40], ‘to stop the medication due to a side treatment characteristics, real-world treatment effect by 6 months’ [43], ‘chance of serious decision-making may differ from the stated treatment reaction’ [18, 19] or ‘mild treatment choices provided during DCE because of the reaction’ [18]), which varied from the ones used influence of treatment attributes that were not in this survey (‘percentage of patients that suffer included in the study as well as other influences severe or mild adverse events’) and may be on decision making such as lifestyle or family interpreted in a different way. The lesser environment. Therefore, although the attri- importance attached to safety attributes may butes presented were confirmed as being the also be explained by the inclusion of the ‘opti- most relevant for treatment decision-making in mal QoL’ concept of the lack of side effects. RA, it cannot be excluded that attributes not The information provided by preferences included may also be relevant and play a role in studies is not only relevant to improving health treatment decision-making. Moreover, there is decision-making, but can also be useful for some uncertainty associated with the interpre- defining new drug development strategies, tation by patients of the attributes presented more closely aligned with patients’ preferences. within the scenarios, mainly time with optimal Preference studies conducted during preclinical QoL and onset of treatment action, and how it development may contribute to seeking early may affect the results. Finally, since RA patients patient input on what outcomes are important; were invited to participate by a Patient Advo- during clinical trial design they may help define cacy Group, they may not be representative of patient-relevant endpoints and study enrol- the RA population and limitations in the gen- ment criteria; or in post-market approval stud- eralizability of results cannot be excluded. ies, they may support the formulation of product communication and marketing strate- gies [47]. CONCLUSIONS Shared decision-making (SDM) is the pinna- cle of patient-centered care [48]. SDM increases Our results provide knowledge related to patient knowledge, reduces anxiety over the patient and rheumatologist preferences for care process and improves health outcomes [49] treatment, showing some differences between by ensuring that medical care better aligns with both perspectives. For rheumatologists, in patients’ preferences and values. The results of addition to efficacy, treatment mode of the study highlighted that RA patients were administration and time to onset of action are willing to be involved in treatment decision- decision-making drivers. For patients, efficacy making. Although most of them are involved in defined as time with optimal QoL and treat- the decision-making process, nearly 30% ment mode of administration are determinants maintain a passive attitude, with the rheuma- in treatment selection. tologist being the one who makes the decision The knowledge of patients’ preferences and for them. The use of patient decision aids (PDA) the evidence that patients are willing to be in RA may help rheumatologists to involve involved in treatment decision-making that patients in decision-making by providing have been gained by this study may contribute detailed information related to the disease and 1492 Adv Ther (2020) 37:1479–1495 to and facilitate the adoption of a patient-cen- Data Availability. The data that support the tered care model. findings of this study are available from the corresponding author upon reasonable request. Open Access. This article is licensed under a ACKNOWLEDGEMENTS Creative Commons Attribution-NonCommer- cial 4.0 International License, which permits The authors would like to thank the Spanish any non-commercial use, sharing, adaptation, Patient Advocacy Group Coordinadora Nacional distribution and reproduction in any medium de Artritis (ConArtritis) and the Spanish Society or format, as long as you give appropriate credit of Rheumatology for their support during the to the original author(s) and the source, provide conduct of the study. Also, they are very grate- a link to the Creative Commons licence, and ful to all the rheumatologists and patients who indicate if changes were made. The images or participated in the study. other third party material in this article are included in the article’s Creative Commons Funding. This research, the Rapid Service licence, unless indicated otherwise in a credit and Open Access Fees were supported by Eli line to the material. If material is not included Lilly & Co, Madrid (Spain). in the article’s Creative Commons licence and your intended use is not permitted by statutory Authorship. All named authors meet the regulation or exceeds the permitted use, you International Committee of Medical Journal will need to obtain permission directly from the Editors (ICMJE) criteria for authorship for this copyright holder. To view a copy of this licence, article, take responsibility for the integrity of visit http://creativecommons.org/licenses/by- the work as a whole, and have given their nc/4.0/. approval for this version to be published. Disclosures. Silvia Dı́az, Tatiana Dilla, José Antonio Sacristán and José Inciarte-Mundo are REFERENCES employees of Eli Lilly & Co. Marta Comellas, Miriam Prades and Luis Lizán work for an 1. Smolen JS, Aletaha D. Rheumatoid arthritis. Lancet. independent research entity that received 2016;388:2023–38. funding from Eli Lilly & Co to coordinate and 2. Smolen JS, Aletaha D, Barton A, Burmester GR, conduct the study and to write up the manu- Emery P, Firestein GS, et al. Rheumatoid arthritis. script. Cesar Dı́az-Torné, Ana Urruticoechea- Nat Rev Dis Prim. 2018;4(18001):1–23. Arana and José Ivorra-Cortés have nothing to disclose. 3. Sanmartı́ R, Garcı́a-Rodrı́guez S, Álvaro-Gracia JM, Andreu JL, Balsa A, Cáliz R, et al. Actualización 2014 del Documento de Consenso de la Sociedad Espa- Compliance with Ethics Guidelines. This ñola de Reumatologı́a sobre el uso de terapias study was conducted in accordance with the biológicas en la artritis reumatoide. Reumatol Clı́n. principles of the Declaration of Helsinki. It was 2015;11(5):279–94. developed to ensure that Good Clinical Prac- 4. Smolen JS, Landewé R, Bijlsma J, et al. EULAR rec- tices were observed, in keeping with ICH Har- ommendations for the management of rheumatoid monized Tripartite Guideline principles. The arthritis with synthetic and biological disease- study protocol was submitted to the Spanish modifying antirheumatic drugs: 2016 update. Ann Rheum Dis. 2017;76(6):960–77. Agency of Medicines and Medical Devices and to the Clinical Research Ethics Committee of 5. Pasma A, van’t Spijker A, Hazes JMW, Busschbach Puerta de Hierro-Majadahonda Hospital for JJV, Luime JJ. Factors associated with adherence to approval. An electronic informed consent form pharmaceutical treatment for rheumatoid arthritis patients: A systematic review. Semin Arthritis was read and accepted by all participants before Rheum. 2013;43(1):18–28. https://doi.org/10.1016/ completing the questionnaire. j.semarthrit.2012.12.001. Adv Ther (2020) 37:1479–1495 1493 6. Bemt BJ van den, Zwikker HE, Ende CH van den. experimental designs for discrete-choice experi- Medication adherence in patients with rheumatoid ments: report of the ISPOR conjoint analysis arthritis: a critical appraisal of the existing litera- experimental design good research practices task ture. Expert Rev Clin Immunol [Internet]. 2014; force. Value Heal. 2013;16(1):3–13. 337–51. Disponible en: http://www.tandfonline. com/doi/abs/10.1586/eci.12.23?journalCode=ierm 16. Soekhai V, de Bekker-Grob EW, Ellis AR, Vass CM. 20#.Vegj7PmeDGc. Accessed 8 Apr 2019. Discrete choice experiments in health economics: past, present and future. Pharmacoeconomics. 7. Voshaar MJH, Nota I, Van De Laar MAFJ, Van Den 2019;37(2):201–26. https://doi.org/10.1007/s402 Bemt BJF. Patient-centred care in established 73-018-0734-2. rheumatoid arthritis. Best Pract Res Clin Rheuma- tol. 2015;29(4–5):643–63. https://doi.org/10.1016/j. 17. Nolla J, Rodrı́guez M, Martin-Mola E, Raya E, Ibero berh.2015.09.007. I, Nocea G, et al. Patients’ and rheumatologists’ preferences for the attributes of biological agents 8. Wen H, Ralph Schumacher H, Li X, Gu J, Ma L, Wei used in the treatment of rheumatic diseases in H, et al. Comparison of expectations of physicians Spain. Patient Prefer Adherence. 2016;10:1101. and patients with rheumatoid arthritis for rheumatology clinic visits: a pilot, multicenter, 18. Poulos C, Hauber AB, González JM, Turpcu A. international study. Int J Rheum Dis. 2012;15(4): Patients’ willingness to trade off between the 380–9. duration and frequency of rheumatoid arthritis treatments. Arthritis Care Res (Hoboken). 9. De Mits S, Lenaerts J, Vander Cruyseen B, Mielants 2014;66(7):1008–15. H, Westhovens R, Durez P, et al. A Nationwide Survey on Patient’s versus Physician’s Evaluation of 19. Louder AM, Singh A, Saverno K, Cappelleri JC, Aten Biological Therapy in Rheumatoid Arthritis in AJ, Koenig AS, et al. Patient preferences regarding Relation to Disease Activity and Route of Adminis- rheumatoid arthritis therapies: a conjoint analysis. tration: The Be-Raise Study. Plose One. 2016;11(11): Am Heal Drug Benefits. 2016;9(2):84–93. 1–10. 20. Augustovski F, Beratarrechea A, Irazola V, Rubin- 10. Markenson JA, Koenig AS, Feng JY, Chaudhari S, stein F, Tesolin P, Gonzalez J, et al. Patient prefer- Zack DJ, Collier D, et al. Comparison of physician ences for biologic agents in rheumatoid arthritis: a and patient global assessments over time in patients discrete-choice experiment. Value Heal. 2013;16(2): with rheumatoid arthritis: a retrospective analysis 385–93. from the RADIUS Cohort. J Clin Rheumatol. 2013;19:317–23. 21. Bridges JFP, Hauber AB, Marshall D, Lloyd A, Prosser LA, Regier DA, et al. Conjoint analysis applications 11. Suarez-Almazor ME, Conner-Spady B, Kendall CJ, in health—a checklist: a report of the ISPOR Good Russell AS, Skeith K. Lack of Congruence in the Rat- Research Practices for Conjoint Analysis Task Force. ings of Patients’ Health Status by Patients and Their Value Heal. 2011;14(4):403–13. Physicians. Med Decis Mak. 2001;21(2):113–21. 22. Orme B. Getting started with conjoint analysis: 12. Smolen JS, Landewé R, Breedveld FC, Buch M, strategies for product design and pricing research. Burmester G, Dougados M, et al. EULAR recom- 2nd ed. Madison, WI: Research Publishers LLC; mendations for the management of rheumatoid 2010. p. 57–66. arthritis with synthetic and biological disease- modifying antirheumatic drugs: 2013 update. Ann 23. Instituto Nacional de Estadı́stica (INE). Estima- Rheum Dis. 2014;73(3):492–509. ciones de la población actual de España a 1 de enero de 2016. Available at: https://www.ine.es/dyngs/ 13. Chewning B, Bylund CL, Shah B, Arora NK, Gue- INEbase/es/operacion.htm?c=Estadistica_C&cid=12 guen JA, Makoul G. Patient preferences for shared 54736176951&menu=ultiDatos&idp=1254735572 decisions: a systematic review. Patient Educ Couns. 981. Accessed 21 Sept 2016. 2012;86(1):9–18. https://doi.org/10.1016/j.pec.2011. 02.004. 24. Carmona L, Ballina J, Gabriel R, Laffon A, EPISER Study Group. The burden of musculoskeletal dis- 14. Neame R, Hammond A, Deighton C. Need for eases in the general population of Spain: results information and for involvement in decision mak- from a national survey. Ann Rheum Dis. ing among patients with rheumatoid arthritis: a 2001;60(11):1040–5. questionnaire survey. Arthritis Care and Res. 2005;53:249–55. 25. González López-Valcárcel B, Barber Pérez P, Suarez Vega R. Oferta y necesidad de especialistas médicos 15. Johnson FR, Lancsar E, Marshall D, Kilambi V, en España (2010–2025).2011; 230. Muhlbacher A, Regier DA, et al. Constructing 1494 Adv Ther (2020) 37:1479–1495 26. Dilla T, Rentero M, Comellas M, Lizan L, Sacristán J. 38. Fautrel B, Alten R, Kirkham B, de la Torre I, Durand Patients’ preferences for rheumatoid arthritis treat- F, Barry J, et al. Call for action: how to improve use ments and their participation in the treatment of patient-reported outcomes to guide clinical decision-making process: a systematic review of the decision making in rheumatoid arthritis. Rheuma- literature. Value Heal. 2015;18(7):A652. tol Int. 2018;38(6):935–47. https://doi.org/10.1007/ s00296-018-4005-5. 27. Aizaki H. Basic functions for supporting an imple- mentation of choice experiments in R. J Stat Softw. 39. Alten R, Krüger K, Rellecke J, Schiffner-Rohe J, 2012;50:1–24. Behmer O, Schiffhorst G, et al. Examining patient preferences in the treatment of rheumatoid arthritis 28. World Health Organization. How to conduct a using a discrete-choice approach. Patient Prefer discrete choice experiment for health workforce Adherence. 2016;10:2217–28. recruitment and retention in remote and rural areas: a user guide with case studies. Geneva: World 40. Scalone L, Sarzi-Puttini P, Sinigaglia L, Montecucco Health Organization; 2012. C, Giacomelli R, Lapadula G, et al. Patients’, physicians’, nurses’, and pharmacists’ preferences 29. Esteve-Vives J, Batlle-Gualda E. Reig A (1993) on the characteristics of biologic agents used in the Spanish version of the Health Assessment Ques- treatment of rheumatic diseases. Patient Prefer tionnaire: reliability, validity and transcultural Adherence [Internet]. 2018;12:2153–2168. https:// equivalency. Grupo para la Adaptación del HAQ a la www.dovepress.com/patientsrsquo-physiciansrsq Población Española. J Rheumatol. 1993;20(12): uo-nursesrsquo-and-pharmacistsrsquo-prefere-peer- 2116–22. reviewed-article-PPA. Accessed 23 Apr 2019. 30. Solari A, Giordano A, Kasper J, Drulovic J, van 41. Hifinger M, Hiligsmann M, Ramiro S, Watson V, Nunen A, et al. Role preferences of people with Severens JL, Fautre B, et al. Economic considera- multiple sclerosis: image-revised, computerized tions and patients’ preferences affect treatment self-administered version of the control preference selection for patients with rheumatoid arthritis: a scale. PLoS One. 2013;8(6):e66127. https://doi.org/ discrete choice experiment among European 10.1371/journal.pone.0066127. rheumatologists. Ann Rheum Dis. 2017;76(1): 126–32. 31. Mcfadden D. Conditional logit analysis of qualita- tive choice behavior. New York: Academic Press; 42. Bolge SC, Goren A, Brown D, Ginsberg S, Allen I. 1974. p. 105–42. Openness to and preference for attributes of bio- logic therapy prior to initiation among patients 32. Everitt B, Landau S, Leese M, Stahl D. Cluster with rheumatoid arthritis: patient and rheumatol- analysis. 5th ed. Amsterdam: Wiley; 2011. p. 346. ogist perspectives and implications for decision making. Patient Prefer Adherence. 2016;10: 33. Core Team R. R: A language and environment for 1079–90. statistical computing R Foundation for Statistical Computing. Vienna: Core Team R; 2018. 43. Hazlewood GS, Bombardier C, Tomlinson G, Thorne C, Bykerk VP, Thompson A, et al. Treatment 34. Charrad M, Ghazzali N, Boiteau V, Niknafs A. preferences of patients with early rheumatoid NbClust: an R package for determining the relevant arthritis: a discrete-choice experiment. Rheumatol number of clusters in a data set. J Stat Softw. (United Kingdom). 2016;55(11):1959–68. 2014;61(6):1–36. https://doi.org/10.18637/jss.v061. i06. 44. Matcham F, Scott IC, Rayner L, Hotopf M, Kingsley GH, Norton S, et al. The impact of rheumatoid 35. Johnson FR, Özdemir S, Mansfield C, Hass S, Miller arthritis on quality-of-life assessed using the SF-36: DW, Siegel CA, et al. Crohn’s disease patients’ risk- a systematic review and meta-analysis. Semin benefit preferences: serious adverse event risks ver- Arthritis Rheum. 2014;44(2):123–30. https://doi. sus treatment efficacy. Gastroenterology. org/10.1016/j.semarthrit.2014.05.001. 2007;133(3):769–79. 45. Sanderson T, Morris M, Calnan M, Richards P, 36. Bowker AH. A Test for symmetry in contingency Hewlett S. Patient perspective of measuring treat- tables. J Am Stat Assoc. 1948;43(244):572–4. ment efficacy: the rheumatoid arthritis patient pri- orities for pharmacologic interventions outcomes. 37. Sacristán J. Patient-centered medicine and patient- Arthritis Care Res. 2010;62(5):647–56. oriented research: improving health outcomes for individual patients. BMC Med Inform Decis Mak 46. Fraenkel L, Bogardus ST, Concato J, Felson DT, 2013;13(1). https://doaj.org/article/a61e88786f38 Wittink DR. Patient preferences for treatment of 4c77a54c0bd21723e11f. Accessed 23 Apr 2019. rheumatoid arthritis. Ann Rheum Dis. 2004;63(11): 1372–8. Adv Ther (2020) 37:1479–1495 1495 47. Postmus D, Mavris M, Hillege HL, Salmonson T, 49. Oshima Lee E, Emanuel EJ. Shared decision making Ryll B, Plate A, et al. Incorporating patient prefer- to improve care and reduce costs. N Engl J Med. ences into drug development and regulatory deci- 2012;368(1):6–8. sion making: results from a quantitative pilot study with cancer patients, carers, and regulators. Clin 50. Pablos JL, Jover JA, Roman-Ivorra JA, Inciarte- Pharmacol Ther. 2016;99(5):548–54. Mundo J, Dilla T, Sacristan JA, et al. Patient decision aid (PDA) for Patients with rheumatoid arthritis 48. Barry MJ, Edgman-Levitan S. Shared decision mak- reduces decisional conflict and improves readiness ing—the pinnacle of patient-centered care. N Engl J for treatment decision making. Patient Patient- Med. 2012;366(9):780–1. https://doi.org/10.1056/ Centered Outcomes Res. 2019. https://doi.org/10. NEJMp1109283. 1007/s40271-019-00381-y.