Patient preference survey for attributes of locally advanced or metastatic urothelial carcinoma treatment in Japan: a discrete-choice experiment Authors: Hiroshi Kitamura, Hiroki Uchida, Yusuke Tanaka, Takeshi Akiyama, Ayumi Morishita, Anzu Yakushiji, Masaya Sakurai Published: 2025 ABSTRACT ABSTRACTObjectiveTo investigate treatment preferences of Japanese patients with locally advanced or metastatic urothelial carcinoma (la/mUC).DesignA discrete-choice experiment survey methodology was used to elicit patient preferences for attributes of la/mUC treatment, including adverse events (nausea/vomiting, neuropathy, alopecia and maculopapular rash), hospital stay requirements for treatment administration and overall survival. A multinomial logistic regression model was used to estimate treatment preferences. Coefficients of the model were then used to calculate the relative importance of each treatment attribute.Setting and participantsParticipants were recruited through healthcare facilities with urology departments across Japan.ResultsThe study included 109 patients (72.5% male; mean (SD) age, 71.3 (8.9) years). Patients preferred treatments that minimised adverse events that could affect their daily activities, had a more favourable survival profile and did not require hospital stays for administration. Neuropathy emerged as the most important attribute to patients when making treatment decisions (relative attribute importance (RAI), 27.7%), followed by nausea/vomiting (RAI, 27.3%), maculopapular rash (RAI, 16.5%), hospital stay requirements (RAI, 12.1%), alopecia (RAI, 8.2%) and overall survival (RAI, 8.1%). Findings were similar across various subgroup analyses, though patients who were chemotherapy-experienced prioritised avoidance of neuropathy and nausea/vomiting more highly than those who were chemotherapy-naïve.ConclusionsIn this study, Japanese patients with la/mUC were more concerned about reduced quality of life due to adverse events than extending survival by 6 months. Patients, clinicians and other members of the care team need to communicate frequently and candidly about a patient’s well-being. STRENGTHS AND LIMITATIONS OF THIS STUDY Attributes and their levels were carefully determined by preceding studies, literature reviews and Japanese medical experts. Attributes and their levels were carefully determined by preceding studies, literature reviews and Japanese medical experts. Only a small sample size of survey responders was available, which may have affected the generalisability of this study. Only a small sample size of survey responders was available, which may have affected the generalisability of this study. As a cross-sectional study with a one-time execution for each patient, changes in patient preference over time were not captured in this study. As a cross-sectional study with a one-time execution for each patient, changes in patient preference over time were not captured in this study. An initial pilot study in this patient population was not conducted, and no validation questions were included in the survey to identify unreliable responders. An initial pilot study in this patient population was not conducted, and no validation questions were included in the survey to identify unreliable responders. INTRODUCTION Urothelial carcinoma (UC) accounts for nearly all cases of bladder cancer.1 It is the 10th and 13th most commonly diagnosed type of cancer in the world and Japan, respectively,2 3 with approximately 573 000 new cases diagnosed globally in 2020.3 Patients diagnosed with localised disease generally respond well to treatment, with 5-year survival rates across all stages above 60% in Japan.4 However, the prognosis for metastatic UC (mUC) is generally much poorer, with 5-year survival rates of approximately 8% in the USA depending on the extent of metastases.5 Combined data from the USA, Europe and Japan also suggest a 5-year survival rate of 5%.6 However, the prognosis for metastatic UC (mUC) is generally much poorer, with 5-year survival rates of approximately 8% in the USA depending on the extent of metastases.5 Combined data from the USA, Europe and Japan also suggest a 5-year survival rate of 5%.6 Platinum-based chemotherapy represents the historic standard of care in the first-line setting for patients with locally advanced or mUC (la/mUC).79 However, given that most bladder cancers are diagnosed in elderly populations, older patients are often not eligible to receive standard chemotherapies because of poor performance status and comorbidities.7 8 Evidence suggests that 30%–50% of patients with UC are ineligible for cisplatin.10 Additionally, patients who are refractory to or relapse after platinum-based chemotherapy have historically faced even more limited treatment options.8 11 The availability of programmed cell death protein 1/ligand 1 (PD-1/L1) inhibitors in recent years has shifted treatment approaches for bladder cancer.12 PD-1/L1 inhibitors (including nivolumab, pembrolizumab and avelumab) provide more options for patients with la/mUC as 1L, 1L+ and maintenance therapy, respectively.1315 More recently, the antibody–drug conjugate directed against Nectin-4, enfortumab vedotin, has become available for patients previously treated with platinum-based chemotherapy and PD-1/L1 inhibitors.16 The enfortumab vedotin and pembrolizumab combination is now also recommended as the new standard of care in first-line advanced UCs.17 As the suite of treatment options expands, patients and their care teams may be faced with increasingly complex considerations when making treatment decisions. New therapies often have distinct treatment characteristics, including expected effectiveness, potential adverse events (AEs), dosing frequencies and administration logistics. Quantitative methodologies, including discrete-choice experiments (DCEs), can be used to measure patient preferences for attributes of cancer treatment.18 Globally, patient preference research is emerging as a topic of interest as regulatory agencies and health technology assessment bodies emphasise patient-focused drug development and incorporate patient preference information into healthcare decision-making.18 19 Research also suggests that patient preference has implications for the choice of treatment, such as the likelihood of patient adherence to prescribed therapy.20 Although literature on patient preferences for cancer treatment has increased in recent years,18 19 little has been published on the preferences of patients with UC in Japan; however, a recent discrete-choice survey in the USA showed that patients with advanced UC determined treatment experience to be most important when considering options, followed by overall survival (OS).21 This study aimed to understand the importance of attributes associated with treatment, including expected survival, AEs and requirements for hospital stays, that are most important to patients with la/mUC in Japan. Methods Study design This study was a non-interventional, cross-sectional, multicentre survey conducted in Japanese patients with la/mUC from 10 healthcare facilities (online supplemental table 1) between June and November 2022. A DCE methodology was employed to quantify preferences for attributes of treatment, where respondents choose between hypothetical treatments with different combinations of levels of attributes across a series of questions determined by an orthogonal design.22 In this methodology, attributes refer to key characteristics associated with treatment (eg, OS), and levels refer to the specific value or category that an attribute can take (eg, 14-month survival and 20-month survival). In this study, respondents chose between two hypothetical cancer treatments. This study was reviewed and approved by an ethics committee before data collection (Non-Profit Organisation MINS Institutional Review Board (http://www.npo-mins.com/irb), first approved in February 2022, with approval ID 220206). All respondents provided informed consent after reviewing a detailed explanation of the study and before participating in any study activities. All healthcare facilities included in this study were allowed to recruit patients following approval by the head of the facility. Survey instrument and experimental design A key initial step in the development of a DCE survey is to select appropriate attributes and levels based on existing or emerging treatments in the therapeutic area (in this study, la/mUC).23 A targeted literature review via Medline for English-language publications and the Igaku Chuo Zasshi database for Japanese-language publications was conducted to identify attributes relevant to patients with la/mUC. Studies identified in the literature review included previously conducted patient-preference research, patient-reported outcome studies including quality of life (QOL) research and clinical trials for la/mUC populations. A full list of all identified publications is provided in online supplemental table 2. To aid the development of the questionnaire, an advisory board meeting was convened, during which urology experts provided feedback on the attributes and levels. Based on this, the questionnaire was drafted, after which a pilot study with a physician and proxy responder, a different physician, was also conducted to determine whether the survey questions could pose a psychological burden or ethical issue to patients and whether patients would be able to understand the survey questions. As a final step, feedback from the pilot study was discussed during an advisory board to finalise the questionnaire, including attributes and levels. Table 1 shows the final six attributes and levels chosen for inclusion, and figure 1 depicts how attributes and levels were presented to patients. The original Japanese version of the chart is provided in online supplemental figure 1. Table 1Attributes and corresponding levels included in the discrete-choice experimentAttributeLevelOverall survival, monthsL1: 20L2: 14Nausea/vomitingL1: controlled by medication; no influence on daily life activitiesL2: controlled by medication; occasional symptoms with some influence on activities of daily livingL3: poorly controlled with medication; frequently results in difficulty in activities of daily livingNeuropathyL1: peripheral neuropathy (eg, muscle weakness, numbness and paraesthesia) but no influence on activities of daily livingL2: peripheral neuropathy (eg, muscle weakness, numbness and other sensory abnormalities), tripping and difficulty grasping with some influence on activities of daily livingL3: peripheral neuropathy (eg, muscle weakness, numbness and paraesthesia) results in difficulty in activities of daily livingMaculopapular rash/rashL1: itching, bleeding and scratchy skin eruptions in ≈1% of patients treated with this medicationL2: itching, bleeding and scratchy skin eruptions in ≈20% of patients treated with this medicationL3: itching, bleeding and scratchy skin eruptions in ≈50% of patients treated with this medicationAlopeciaL1: more hair loss than usual but not noticeableL2: level of hair loss somewhat noticeableL3: large amount of hair loss concerning for one’s appearanceHospital stay for drug administrationL1: three visits with 21 days of administration; hospital stay not requiredL2: three visits with 21 days of administration; hospital stay requiredAttributes and levels were decided based on the results of a targeted literature review and feedback during advisory board meetings.L, level. Attributes and corresponding levels included in the discrete-choice experiment Attributes and levels were decided based on the results of a targeted literature review and feedback during advisory board meetings. L, level. Depiction of attributes and levels presented in the discrete-choice experiment.*,† *The information in this figure was translated from Japanese to English; patients viewed this information in Japanese. †This figure depicts the iconography and language used to describe the discrete-choice experiment attributes; it does not show a direct example of one of the choice tasks patients directly answered. Details for all attributes and levels were explained in an instruction flyer describing the content shown in table 1 for all participants. After choosing attributes and levels and developing the questionnaire, the next step was to construct the DCE design.23 In designing the discrete-choice questionnaire, the variation of attributes was balanced and independent to ensure that each level appeared equally as often, consistent with the International Society for Pharmacoeconomics and Outcomes Research Good Research Practices for Conjoint Analysis.23 The design was constructed using conjoint analysis capabilities in the Qualtrics platform (Qualtrics, Provo, Utah).24 In this study, six attributes with a maximum of three levels per attribute were identified. Published literature suggests that 7–16 is a feasible number of choice tasks, as more tasks may increase respondent burden.25 After reviewing information about treatment attributes, patients responded to 15 choice sets as dictated by the experimental design. Online supplemental figure 2 depicts two examples of DCE choice sets answered by respondents. In addition to the DCE, respondents reported demographic and clinical characteristics, including age, sex, level of education, marital status, household income, smoking status, time since UC diagnosis and UC treatment history. These variables also formed the basis of subgroup analyses to assess differences in treatment preferences across prespecified subgroups. Study size and participant recruitment The required sample size was calculated using Orme’s equation, which is commonly used for DCEs,23 26 N ≥ (750×C) ÷ (T×A), where T is the number of choice tasks, A is the alternatives per task and C is the highest number of levels across attributes. For this study, T=15, A=2 and C=3, yielding N = (750×3) ÷ (15×2) = 75. To account for incomplete responses, the recruitment target was increased to 100. Although no minimum sample size is mandated for DCEs, evidence suggests that samples over 100 generally support robust preference modelling.27 Participants were recruited through healthcare facilities with urology departments across Japan. No specific sampling techniques were applied to the enrolment of patients. Physicians invited patients with la/mUC to participate in this study after consideration of each patient’s status to meet the inclusion criteria. The inclusion criteria were as follows: a physician-confirmed diagnosis of la/mUC, age 18 years or older, Japanese residence and ability to participate in a web-based survey in Japanese. Patients could participate independently of the line of treatment they were receiving. Patients who indicated interest in the study received a leaflet from their primary doctor that included a link to access the survey. A detailed explanation of the study was provided to respondents on the survey site. After providing consent, they were directed to the screening questionnaire to confirm study eligibility and proceeded with the full survey. The survey could be completed at home on an electronic device or at the healthcare facility on a facility-owned device. Survey participants received a prepaid coupon of JP¥5000 (approximately $35) as compensation. Patient and public involvement Patients were initially contacted by their primary doctors, who introduced them to the study. Patients were not directly involved in the design, recruitment or conduct of this study; however, patient feedback related to the burden of treatment management was still integral to the choice of questions and primary outcome of the study. The time required to participate in the study was not considered a burden that required specific evaluation as part of the study. Patients were not requested to provide feedback regarding the presentation and distribution of study data but consented to disseminate the study results via conferences and/or publications. Statistical analysis Descriptive statistics were used to describe patient demographic and clinical characteristics. A multinomial logistic model was used to estimate treatment preferences from the DCE, with both hierarchical Bayes and Markov-chain Monte Carlo simulation methods employed for the estimation process. Coefficients of the model represent preference-based utilities that were used to calculate the relative importance of each attribute (relative attribute importance), representing the degree of influence each attribute has versus all other attributes on the overall treatment decision in the DCE. Subgroup analyses were conducted to assess potential differences in preferences across selected groups: age (<65 vs ≥65 and <75 vs ≥75 years), sex (male vs female), marital status (married vs not married), employment status (retired/unemployed vs others), time since la/mUC diagnosis (<1 vs ≥1 and <2 vs ≥2 and <5 vs ≥5 years), history of chemotherapy (yes vs no), history of PD-1/L1 inhibitor use (yes vs no) and presence of metastatic disease (yes vs no). The online survey system prevented respondents from omitting any responses before going to subsequent sections, and data from those who did not complete all the questionnaires and the DCE were deleted from the study dataset; as such, no data were missing in this study. Analysis of data for the DCE was performed using Qualtrics software. Demographic characteristics were analysed descriptively using R version 4.0.5 (R Foundation for Statistical Computing, Vienna, Austria). Results Survey participation and sample characteristics A total of 111 patients consented to participate in the study; however, two withdrew their consent, and 109 patients completed the survey. Data on the number of patients invited to participate in the study were not available. Most (72.5%) were men with a mean (SD) age of 71.3 (8.9) years (table 2). About one-third (33.0%) of the respondents were diagnosed with la/mUC between 2 and <5 years before study enrollment; 96.3% were undergoing treatment. Nearly all (90.7%) respondents had received chemotherapy, 38.9% had received pembrolizumab, 13.0% had received avelumab and 2.8% had received nivolumab. Notably, as treatments were not mutually exclusive, several respondents had a history that included multiple different treatments (table 3). Table 2Demographic and clinical characteristics of the study participantsCharacteristic, n (%)Participantsn=109Age, mean (SD), years71.3 (8.9)Age, years <6526 (23.9) 65 to <7538 (34.9) ≥7545 (41.3)Male79 (72.5)Marital status Married84 (77.1) Single12 (11.0) Divorced/bereaved13 (11.9)Education Junior high school24 (22.0) High school40 (36.7) Vocational school/junior college20 (18.3) University/graduate school25 (22.9)Employment status* Full time26 (23.9) Part time6 (5.5) Retired31 (28.4) Leave of absence1 (0.9) Unemployed35 (32.1) Homemaker9 (8.3) Student0Percentage of medical expenses† 0% (exempted from payment)6 (5.5) 10%31 (28.4) 20%27 (24.8) 30%39 (35.8)Household income, JP¥PI‡ 0 to <2 million30 (27.5) 2 to <5 million46 (42.2) ≥5 million23 (21.1)Smoking status Currently smoking11 (10.1) History of smoking75 (68.8) Never smoked23 (21.1)Time since urothelial carcinoma diagnosis <3 months6 (5.5) 3 to <6 months12 (11.0) 6 months to <1 year16 (14.7) 1 to <2 years21 (19.3) 2 to <5 years36 (33.0) ≥5 years18 (16.5)Site of metastasis§,¶ Bone7 (8.4) Lymph node33 (39.8) Lung/liver/brain27 (32.5) Other organ13 (15.7) None34 (41.0)Comorbidity§,** Heart disease5 (4.6) Liver disease3 (2.8) Stroke2 (1.8) Rheumatoid arthritis0 Palsy2 (1.8) Cancer other than urothelial carcinoma5 (4.6) Diabetes12 (11.0) Kidney disease12 (11.0) None67 (61.5)*One patient did not answer.†Six patients did not answer.‡Ten patients did not answer.§Patients could select >1 response, so percentages sum to >100%.¶Site of metastasis was unknown for nine patients.**Comorbidity status was unknown for six patients.JP¥, Japanese Yen. Demographic and clinical characteristics of the study participants One patient did not answer. Six patients did not answer. Ten patients did not answer. Patients could select >1 response, so percentages sum to >100%. Site of metastasis was unknown for nine patients. Comorbidity status was unknown for six patients. JP¥, Japanese Yen. Table 3Treatment status and history of the study participantsCharacteristic, n (%)Participantsn=109Treatment status Before treatment (considering future treatment)0 Undergoing treatment105 (96.3) After treatment (no further treatment)0 On follow-up after treatment3 (2.8) No treatment in the past or anticipated in the future1 (0.9)Treatment history* Surgery79 (73.1) Radiotherapy24 (22.2) Chemotherapy98 (90.7) None of these0 Unknown0 Not applicable†1 (0.9)PD-1/L1 inhibitor therapy* Pembrolizumab42 (38.9) Avelumab14 (13.0) Nivolumab3 (2.8) Other chemotherapy52 (48.1) No drug treatment for this cancer7 (6.5) Unknown19 (17.6)*Patients could select >1 response, so percentages sum to >100%.†One person did not answer this question as he/she answered receiving ‘no treatment in the past and in the future’ in the previous question. PD-1/L1, programmed cell death protein 1/ligand 1. Treatment status and history of the study participants Patients could select >1 response, so percentages sum to >100%. One person did not answer this question as he/she answered receiving ‘no treatment in the past and in the future’ in the previous question. PD-1/L1, programmed cell death protein 1/ligand 1. Patient preferences and attribute importance The study patients placed the greatest importance on avoiding any AEs when making treatment decisions, followed by a longer OS. Neuropathy (relative attribute importance (SD)) emerged as the most important attribute (27.7% (9.7)), followed by nausea/vomiting (27.3% (7.1)), maculopapular rash/rash (16.5% (9.4)), required hospital stay for administration (12.1% (9.7)), alopecia (8.2% (2.6)) and lastly OS (8.1% (6.0); figure 2). For neuropathy, moving from ‘symptoms having no influence on daily life’ to ‘symptoms leading to difficulty in daily life’ was associated with a utility decrement of 2.78. In contrast, moving from an OS of 20 months to 14 was associated with a utility decrement of 0.74. These findings are displayed in figure 3, where higher utility values indicate levels of each attribute that are more preferred. Relative importance of treatment attributes to patients. OS, overall survival. Utilities across attribute levels. OS=L1, 20 months; L2, 14 months. Nausea/vomiting=L1, symptoms having no influence on daily life; L2, symptoms leading to some influence on daily life; L3, symptoms leading to difficulty in daily life. Maculopapular rash/rash=L1, 1% probability of occurrence; L2, 20% probability of occurrence; L3, 50% probability of occurrence. Neuropathy=L1, symptoms having no influence on daily life; L2, symptoms leading to some influence on daily life; L3, symptoms leading to difficulty in daily life. Alopecia=L1, hair loss not noticeable; L2, hair loss somewhat noticeable; L3, large amount of hair loss. Hospital stay required=L1, no hospital stay; L2, 1-week hospital stay. L1, level 1; L2, level 2; L3, level 3; OS, overall survival. Subgroup analysis Subgroup analyses revealed some differences in preferences by demographic and clinical characteristics. These were generally shown by changes in the ordering of overall attribute importance and were not assessed for statistical significance. In respondents aged <65 years (relative attribute importance (SD)), neuropathy contributed the most to the treatment decision (31.6% (9.0)), followed by nausea/vomiting (27.8% (7.0)), whereas those aged ≥65 years placed the most importance on nausea/vomiting (27.2% (7.1)), followed by neuropathy (26.5% (9.7)) (online supplemental figure 3a). This switch in the order of the most versus second-most important treatment attribute remained when age group was dichotomised as age <75 years versus ≥75 years (online supplemental figure 3b). Respondents diagnosed with UC less than 1 year before study enrolment (relative attribute importance (SD)) placed the most importance on nausea/vomiting (27.9% (6.3)), followed by neuropathy (27.0% (9.6)), whereas patients diagnosed 1 year or longer before study enrolment placed the most importance on neuropathy (28.1% (9.8)), followed by nausea/vomiting (27.1% (7.4)). However, when time to diagnosis was dichotomised differently (as diagnosis <2 vs ≥2 and <5 vs ≥5 years), this reversal in the order of the first and second-most influential attributes was no longer observed, and the trend of attribute importance mimicked the overall pooled results. The time to diagnosis subgroup results are presented in online supplemental figure 4. Some differences in preferences were detected between respondents with and without a history of chemotherapy. Patients without a history of chemotherapy (relative attribute importance (SD)) placed the most importance on neuropathy (24.4% (10.4)), followed by nausea/vomiting (23.2% (7.9)) and occurrence of maculopapular rash/rash (23.3% (12.4) in the treatment decision. Respondents with a history of chemotherapy highly prioritised neuropathy (28.1% (9.6)) and nausea/vomiting (27.8% (6.8)) but found maculopapular rash to be less important (15.7% (8.8)). However, the number of study patients who were naïve to chemotherapy was small (n=11), so findings should be interpreted with caution. Chemotherapy history subgroup results are presented in online supplemental figure 5. Attribute importance was similar between males and females and in marital status, employment status and presence of metastasis subgroups (online supplemental figure 6). Discussion To our knowledge, this is the first study that quantified treatment preferences of Japanese patients with la/mUC. Patients preferred treatments that minimised AEs potentially affecting activities of daily living, had a more favourable OS (20 vs 14 months) and did not require hospital stays for administration. Neuropathy emerged as the most important attribute to patients (driving 27.7% of the treatment decision), followed by nausea/vomiting, maculopapular rash/rash, hospital stay requirements for treatment administration, alopecia and OS. In Japan, bladder cancer is approximately three times more prevalent in males than in females,28 whereas the incident age is approximately 70–74 years old.29 This is consistent with what was observed among the study participants. In 2021, a systematic review of DCEs in oncology found that survival attributes (eg, OS and progression-free survival) were most often ranked as most important to patients.30 Most surveys of patient preference for cancer treatment have been conducted in breast, prostate, skin, blood and lung cancer, and evidence is limited on the treatment preferences of patients with cancer residing in East Asian countries, such as Japan, or how these preferences may be related to other factors (ie, cultural or health system factors). Furthermore, differences in patient preferences have been reported. A study in Japan evaluating patients with prostate cancer found that avoiding fatigue and bone pain was more important than extending OS.31 In contrast, another study of Japanese patients with lung cancer found that OS was the most important attribute in the treatment decision, superseding diarrhoea, nausea and rash.32 However, there has been limited research on the preferences of patients with UC in Japan. In other regions, the preferences of patients with UC differ across published studies. In two studies from the USA, Apolo et al reported that OS was twice as important to patients with UC and their caregivers as reducing the risk of AEs when making treatment decisions.33 In contrast, Grivas et al reported that the nature of the treatment regimen (ie, number/duration of medications) and avoidance of AEs were more important than OS to patients with UC when deciding between first-line treatments.21 Any comparisons between the present study and these studies need to be interpreted cautiously because of the differences in attribute and level selection. Previous studies have indicated that older age may be associated with a greater preference for maintaining good QOL and avoiding AEs and with a lesser emphasis on extending survival34 35; this may explain the results of this study, considering that the mean age of patients in our sample was 71.3 years. Unfortunately, there does not appear to be any consensus on patient age across similar studies on patient preferences in patients with cancer across both Western21 and East Asian countries.31 32 As such, understanding at what age patients may begin to change their treatment priorities and identifying other factors affecting such a shift require further research. Treatment experience may also influence preferences; although the sample size was small, this study found some differences in the relative attribute importance of AEs between chemotherapy-naïve and chemotherapy-experienced patients. The format and setting for presenting DCE attributes and levels may also have affected how patients viewed survival in relation to AEs. However, attribute levels in this study were determined by a targeted literature review of studies relevant to UC treatment and refined based on feedback from urology specialists. Results of this study provide insights that may help guide clinical conversations, specifically regarding AEs associated with la/mUC treatment. The study patients placed the most importance on avoiding AEs that would cause difficulties in their daily lives, underscoring that minimising the effect of treatment on daily life activities is of utmost importance to Japanese patients with la/mUC. Although it is likely that participants’ experiences with specific AEs affected their preferences, this was not analysed in this study. Research in patients with other tumour types has shown that cancer treatment-induced neuropathy is associated with impaired physical functioning and decreased ability to engage in activities of daily living, which are factors contributing to worsened QOL.36 37 Nausea and vomiting resulting from cancer treatment also have a negative effect on physical functioning and nutritional status.38 Clinicians should proactively engage with patients to monitor for and manage treatment-emergent AEs. Additionally, clinicians can support patients in weighing the benefits (eg, survival and disease control) of different treatment options relative to potential risks for AEs to maintain their desired QOL. Patient preferences and QOL are highly individual and dynamic, necessitating that patients, clinicians and other members of the care team communicate frequently and candidly about a patient’s well-being. This is particularly notable as previous studies have highlighted key differences in treatment preference between patients and clinicians, with the latter tending to prioritise survival over treatment experience.21 This study used a well-established methodology to measure treatment preferences; still, several limitations should be considered. Although choice questions were designed to be as clinically realistic as possible, the DCE approach presents hypothetical scenarios that may not be able to accurately reflect real-world treatment choices; a multitude of other factors (eg, comorbidities and clinician recommendation) may be present in real-world treatment decision-making that could not be accounted for in this study design. This study also only included OS as a measure of treatment benefit, although disease control and reduction of cancer-related symptoms may also be considered relevant by patients when making treatment decisions. 6-month improvements in OS may also be insufficient to differentiate from the patient perspective; this improvement is also shorter than the ≥12-month increases in OS detailed in other similar DCEs,31 32 further highlighting the differences in these data. Patients may also change their treatment preferences over time; however, this trend could not be evaluated owing to the cross-sectional nature of the study. As with all survey-based research, results are subject to potential self-report biases (eg, patients reported their own medical history and demographic characteristics), which could have affected the subgroup analyses based on patients’ characteristics. However, all participants were recruited for the study by their healthcare professional, and their diagnosis of la/mUC was confirmed. No preliminary pilot study was carried out for this patient group, and the survey lacked validation questions to detect potentially unreliable responses. Although some subgroup analyses were performed and minor differences in attribute importance were detected, sample sizes in each subgroup were small, and statistical significance was not assessed. Further research is needed to assess how treatment preferences of patients with la/mUC may differ by patient demographic and clinical characteristics. Conclusions In this study, Japanese patients with la/mUC were more concerned about AEs potentially reducing QOL than extending survival by 6 months. This study represents the first assessment of treatment preferences among patients with UC in Japan. These results may inform shared decision-making; patients, clinicians and other members of the care team should communicate frequently and candidly about a patient’s well-being. However, further research is still needed to deepen our understanding of how patients with cancer outside of the Western world view and prioritise the expected effectiveness and safety of their treatment options. Supplementary material