Background: Treatment decisions in type 2 diabetes mellitus (T2D) require knowledge of patient preferences. The German National Health Care Guideline emphasizes preference-based decision making, with treatment goals negotiated in light of individual risk profiles, therapy-related effects, and life circumstances. Systematic preference heterogeneity suggests these contextual factors shape how patients prioritize attributes, posing a central challenge for medical decision making. Methods: Latent class analysis (LCA) was applied to data from a discrete choice experiment (DCE) conducted between August and November 2023 among 583 individuals with T2D in Germany. Participants completed 12 choice tasks using a partial profile design. Treatment options were described by risks of myocardial infarction, stroke, neuropathy, nausea, and severe hypoglycemia as well as weight change, intake mode, and schedule. Analyses were conducted for the overall sample and stratified by treatment experience (population 1: monotherapy, n = 292; population 2: second line, n = 291). Results: In a 3-class solution ( N = 583), one class (28%) strongly prioritized weight outcomes, while remaining classes prioritized nausea risk. Stratified 2-class solutions revealed significant differences in preferences for weight change, administration schedule, and nausea risk, with professional status and weight-related factors corresponding to class membership as contextual factors. Limitations: LCA findings should be interpreted as hypothesis-generating representations of distinct decision logics. Furthermore, limitations inherent to the DCE design, including attribute reduction and partial profile approach, as well as potential variance in response behavior, were carefully considered and addressed. Conclusions: Patients with T2D exhibit distinct preference structures rather than uniform tradeoffs, differing fundamentally in whether disease-related, treatment-related, or administration attributes dominate their evaluations. Implications: Identifying latent preference patterns may enhance the transparency of tradeoffs in participatory treatment deliberation and support structured, guideline-aligned medical decision making.
Background: Although current National Comprehensive Cancer Network guidelines recommend immunotherapies for advanced melanoma, variability in patients’ risk–benefit profiles and treatment characteristics makes treatment selection challenging. Given these differences, we aimed to assess patient and physician preferences for treatment attributes in advanced melanoma. Methods: This cross-sectional study included 2 surveys: one administered to patients with self-reported unresectable stage III or IV melanoma and another to treating physicians. A discrete choice experiment (DCE) was used to elicit preferences and the relative importance (RI) of treatment attributes of efficacy (progression-free survival, overall survival [OS], objective response rate), safety (grade 3/4 treatment-related adverse events [TRAEs], endocrine AEs, discontinuation due to TRAEs), and dosing. The DCE included 12 tasks, each with 2 options with varying levels. The same DCE was used across both surveys, with language tailored to each participant group. A hierarchical Bayesian model was used to estimate preference weights for each level. Results: Overall, 75 patients (mean age, 44.8 y; male, 65%) and 75 physicians (mean age, 45.5 y; male, 76%) were included. For both groups, improvement in safety attributes (overall RI, patients: 49%; physicians: 49%) was more important than improvement in efficacy attributes (overall RI, patients: 35%; physicians: 41%). Both groups were willing to accept a decrease in OS from 65% to 55% for a decrease in grade 3/4 TRAEs (60% to 10%) and discontinuation due to TRAEs (40% to 7%), respectively. Endocrine AEs were considered the least important. Conclusion: Patients and physicians prioritized safety over efficacy when efficacy differences were minimal in selecting first-line treatment for advanced melanoma. By eliciting preferences from these groups, this study identifies the treatment attributes most valued by each group and informs patient-centered treatment recommendations and benefit–risk assessment.
Objective: To examine preferences for allocating scarce health resources between age groups: 10, 40, and 75 y old. Methods: We fielded a set of Bognar-style allocation questions to a nationally representative sample of 1,122 US adults through the NORC AmeriSpeak Survey. Respondents allocated 1) life-extending or 2) life-saving treatments in 2 choice tasks: A) select a single age group to treat and B) allocate 100 treatments across the age groups. We calculated the proportion of respondents who selected each age group and the mean number of treatments allocated to each age group. Results: For task 1A, 60.4% of respondents selected age 10, 30.5% selected age 40, and 5.9% selected age 75, compared with 76.6%, 17.9%, and 1.8% for task 1B. For task 2A, an average of 51.5 treatments were allocated to age 10, 33.3 to age 40, and 15.2 to age 75, compared with 62.2, 27.9, and 9.8 for task 2B. In task 2B, 37.8% of treatments were allocated to older ages. Conclusions: Most respondents preferred treating children in both frames, but a minority of respondents maximized life-years gained. Results from this representative sample underscore the importance of considering age and the context-dependent nature of resource allocation decisions.
Background: Kidney transplantation is the treatment of choice for patients with kidney failure, but transplants may fail due to rejection. Emerging precision medicine approaches seek to reduce rejection and graft failure through improved donor–recipient matching. To evaluate such strategies, we developed KidneyTx-Sim, a discrete event simulation (DES) model that uses real-world data to replicate the current kidney allocation system in British Columbia (BC), Canada, and assess the clinical and economic implications of alternative allocation policies. Method: KidneyTx-Sim simulates the kidney transplant pathway, from waitlist entry to transplantation, rejection, graft failure, and death. The model incorporates patient- and donor-level characteristics using BC data (2008–2018). Its modular structure allows components to be updated as new data and policies emerge. The matching algorithm replicates current allocation rules and can be modified to evaluate alternative strategies. Additional modules represent patient and organ generation and pre- and posttransplant events. Time-to-event outcomes were estimated with Cox proportional hazards models, and health utilities and costs were embedded to enable economic evaluation from a health system perspective. Model parameters and assumptions were developed in collaboration with clinical experts. Results: The model accurately reflects patients with kidney failure on the deceased donor waiting list, aligning with real-world data. Simulated outcomes, including time to transplant, waitlist removal, and death, matched observed outcomes, with 95% confidence intervals encompassing empirical rates. The model reproduced BC Renal outcomes that were not used in its development, including annual waitlist size. Conclusion: KidneyTx-Sim provides a robust, validated framework for evaluating the clinical and economic impact of alternative deceased donor kidney allocation criteria. It can assess policy changes, supporting precision medicine integration and informed decision making in kidney allocation.
Background: Randomized controlled trials (RCTs) are needed to evaluate multicancer detection (MCD) tests. A novel “intended-effect” RCT design in which control group participants receive MCD testing but results remain undisclosed can increase statistical efficiency. Prospective participants may find this design unacceptable due to ambiguity aversion introduced by the existence of undisclosed but theoretically knowable test results. We evaluated this effect and tested if reducing ambiguity by ensuring the unknowability of test results mitigates this bias. Methods: US adults aged 45 to 70 y ( N = 1,576) were randomized to read about 1 of 4 RCT designs evaluating MCD testing: 1) RCT with a conventional control group (no testing), 2) RCT with a standard intended-effect control group (ie, control gets testing but not the results), 3) modified intended-effect RCT emphasizing disclosure safeguards to ensure the unknowability of control group test results to participants and researchers, and 4) modified intended-effect RCT with time-delayed testing on control group blood samples after trial completion. Outcomes were study interest (5-point scale, with higher values indicating more interest) and intention to participate (yes/no). Analyses of variance, chi-square tests, and regression-based analyses examined the main effects and potential mediating roles of the perceived net benefit of participating in the study and trust in study researchers. Results: Intention to participate was similar across all conditions (range: 14%–21%). Study interest was higher in the conventional control ( M = 3.73) vs the 3 intended-effect designs ( P < 0.05), which did not differ ( M s = 3.36–3.45). The negative influence of intended-effect designs on study interest was partially mediated by perceived net benefit. Conclusion: Intended-effect RCT designs lower interest in MCD trial participation, in part by reducing prospective participants’ perceptions of benefiting from the trial. Strategies to reduce ambiguity do not mitigate this effect.
