The DCE methodology is a rigorous, quantitative, and theoretically grounded approach to elicit preferences. This methodology is based on two theoretical foundations: (1) Lancaster’s consumer theory and (2) random utility theory [18]. The assumption behind Lancaster’s consumer theory is that a person will choose a certain good or service based on an analysis of all its intrinsic characteristics or attributes. The combination of attributes gives rise to utility or value. Random utility theory states that utility (value) is a latent variable that exists but cannot be directly observed. Instead, it is assumed that when an individual chooses a good or service, its utility, that is, the value derived from the combination of all of its attributes, matches or exceeds that of all other alternatives considered. DCEs have been used in numerous pharmacoeconomic, health policy, and consumer preference evaluations in the United States and internationally [19].

The process for developing DCE will follow the guidelines outlined by the International Society for Pharmacoeconomics and Outcomes Research Good Research Practices for Conjoint Analysis Task Force [20].

The study will be conducted in phases, including a systematic literature review, rigorous, theoretically grounded qualitative work, and the design, fielding, and analysis of the DCE. Study participants comprise people of the Lumbee Tribe in southeastern North Carolina. The Lumbee Tribe comprises approximately 55,000 people, making the tribe the largest American Indian tribe in North Carolina and one of the largest tribes in the eastern United States [21]. The Lumbee Tribe has been recently recognized by the federal government but has not yet received federal resources for health care services. Robeson County, where the majority of Lumbee Tribe people live, has the highest prevalence rates of cancers in North Carolina. Previous research has shown that American Indian men statewide in North Carolina have PC incidence rates 72% higher than non-Hispanic White men [5].

First, our team conducted a scoping literature review to identify previous DCEs in PC screening. The goal of the scoping review was to identify which attributes have been previously included in PC screening DCEs and how attributes were operationalized into levels. We searched for articles in PubMed, Embase, Web of Science, and Global Index Medicus databases. We used the Covidence online web application (Melbourne, Australia) to manage the data. Articles were included in our scoping review if: (1) men or males at birth were included, (2) the outcome used marginal rates of substitution (mRS), latent class analysis, multivariable logit regression, or relative attribute importance, or (3) the study used any of the following types of study designs: (a) conjoint analysis, (b) DCE, or (c) best-worst scale. Our initial search identified 1140 studies, of which 273 were duplicates. Interrater reliability in article selection was Cohen κ=0.88. One author (CG) extracted study findings. Appendix A in Multimedia Appendix 1 presents the search strategies that we used to identify the literature. Appendix B in Multimedia Appendix 1 presents the data extraction template. Since this was a scoping review, we did not evaluate the quality of the included articles. Appendix C in Multimedia Appendix 1 presents the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) flow diagram showing the number of articles identified throughout the search process.

We screened 867 article titles and abstracts and ultimately included 5 studies consisting of 3253 participants in the review (Appendix C, Multimedia Appendix 1) [22-26]. There was a median of 6 attributes included in the 5 studies. Table 1 presents the key characteristics of each included study. The most important attribute among all studies was reducing the risk of dying from PC. In the studies that examined an opt-out option (3/5, 60%), which is a realistic choice, most men preferred not to have PC screening in 2 studies [23,26]. In the Charvin et al [23] study, which included an educational video about PSA-based screening, those randomized to the video significantly preferred no screening to getting screened. Howard and colleagues [25] also found that younger men valued preventing PC mortality more than older men, highlighting the need to integrate preference heterogeneity into future studies. Attributes that decreased the likelihood of getting screened for PC included higher risks of needing a biopsy and higher risks of adverse treatment effects. Finally, Pignone and colleagues [26] found that different value clarification methods produced different results on PSA screening decisions; that is, those who received a balance sheet are significantly more likely (P<.001) to choose PC screening (43.7%) than a rating and ranking method (34.2%) or DCE (20.2%). However, there was no statistically significant difference between the different methods (balance sheet, rating and ranking, and DCE) on intention to undergo PC screening. An important finding of the Pignone study is that when men were presented with unlabeled test options, they most often chose the no screening option. Whereas in the labeled item about whether they intended to get screened for PC, the men chose they intended to get screened. These studies collectively report that most men would prefer not to undergo PC screening given the risks of overdiagnosis and overtreatment. However, the studies also collectively suggest that men are willing to accept tradeoffs such as false positives and overdiagnosis if it means their PC is caught at an early stage.

The most common attributes studied in PC DCEs have been the reduction in PC mortality (N=5) as a result of getting screened, overdiagnosis and overtreatment (4/5, 80%) [22-25], and screening sensitivity (3/5, 60%) [22,25,26]. We adapted the checklist from the Hall et al [16] systematic review of cancer screening DCEs to guide the categorization of attributes (Multimedia Appendix 2). Three out of the 5 (60%) studies examined the adverse effects of PC treatment, which are treatment attributes and not screening attributes [22,25,26]. Only 1 (20%) study included attributes solely focused on PC screening. Crucially, there are no studies that included any attributes regarding the visit or provider that have been shown to be important in clinic visits with American Indian patients, highlighting an important literature gap [27,28]. Future work regarding PC screening in American Indian men needs to include culturally important attributes to ensure cultural responsiveness. Importantly, these studies were conducted in the mid-2010s and there have since been notable improvements in reducing the risks associated with biopsies and the use of magnetic resonance imaging to decide whether a biopsy is even needed [29].

The DCE survey will involve respondents making iterative choices from sets of hypothetical scenarios that are characterized by preference-relevant features identified in phase 2. By systematically varying the attribute levels by means of an experimental design, participants must make tradeoffs between more- and less-preferred characteristics. The patterns of choices are then analyzed to quantify tradeoffs, as well as identify the relative importance of each feature, and predict the likelihood of PC screening uptake.

The study aims to identify the key considerations that American Indian men make when deciding about whether to get screened for PC. The choice context presented to the participants in the DCE will be framed as follows:

We hypothesize that American Indian men evaluate several screening attributes, such as mortality benefit, unnecessary biopsies, and treatment, as well as contextual factors, such as experiences with PC and the manner in which information is presented, when considering PC screening [25].

DCE choice tasks will resemble the task shown in Table 2. The sample task includes attribute levels identified in the literature review and several attribute levels hypothesized a priori to influence the choices of American Indian men; the selection of attributes and attribute levels in the actual DCE will be guided by the findings from phase 2. Columns represent screening options. Rows describe the attributes identified in phase 2. Cells represent feasible levels for each attribute. Each choice task involves the comparison of 2 hypothetical PC screening tests, followed by an “opt-out” question, that is, whether the participant would get tested in the preferred scenario, which reflects a realistic choice based on current USPSTF PC screening recommendations [10]. This opt-out option provides a reference point that allows for estimates of the potential uptake of PC screening. Prior to the choice tasks, participants will be introduced to each attribute, presented with ceteris paribus conditions (eg, the participant has insurance that covers the test), and will complete a training and comprehension task.

This study has been approved by the IRB of Wake Forest University School of Medicine (IRB00084519). We will seek approval from the Lumbee Tribe’s IRB for the DCE phase as well as official approval to start data collection from the Lumbee Tribe. One of the team members of this study is a member of the Lumbee Tribe IRB and will serve as a liaison between the study team and the IRB. The Lumbee Tribe IRB was set up after our study began. All phases of the research study that involve human participants will be conducted in accordance with the Declaration of Helsinki and the Belmont Report. All identities of the participants’ survey and IDI data will be blinded to the research team to ensure anonymity. In phases 2 and 3, Lumbee men will provide informed consent to participate in the study. We will also use Indigenous Data Sovereignty Principles and acknowledge that the Lumbee Tribe is the ultimate authority on how the research is conducted [41]. We also acknowledge that the Lumbee Tribe is an equal partner in this project. All surveys will be confidential and will not contain any identifying information. Regarding the qualitative formative work, all participant references to specific people and places will be replaced with “xx” to ensure anonymity. Data will be transferred through the use of the Health Insurance Portability and Accountability Act–compliant Box cloud software (Box Inc). Participants will receive $50 remuneration for participating in the formative qualitative work. Participants will receive $30 for completing the DCE survey.

This study will be the first to use DCE methods to inform the development of preference-concordant screening strategies aimed at reducing the PC mortality disparity among American Indian men in the United States. We chose to focus on the Lumbee Tribe due to multiple factors. First, American Indians are an underrepresented group in PC research. Second, the Lumbee Tribe is located primarily in rural southeastern North Carolina; rural areas are disproportionately impacted by cancer mortality [5]. Although rural areas in the United States have lower cancer incidence than urban areas, cancer mortality is higher in rural areas [42]. Third, prior work has identified that American Indian men in North Carolina are at higher risk for dying because of PC than other racial and ethnic groups, possibly due to seeking care when their PC is at a more advanced stage. Therefore, new approaches to PC screening are needed. Fourth, this is the first study to tailor a DCE on cancer screening using culturally responsive approaches, and it is being driven by members of the Lumbee Tribe who are cancer prevention researchers (RB, TL). This DCE is the first of its kind and will advance what is known about PC screening beliefs and practices of American Indian men. Estimates from our study will inform provider-patient communication and provider behaviors in PC screening consultations with American Indian men. The findings of this study can be used to design preference-concordant screening strategies and pragmatic trials that compare the effectiveness of different options and incentives to increase PC screening.

Prior to submitting the findings of phases 2 and 3, we will inform the Lumbee Tribe of our findings and ask for permission to submit the report to a peer-reviewed journal, in accordance with Indigenous Data Sovereignty Principles [41]. We acknowledge that the Lumbee Tribe has the ultimate authority over where and how research findings that relate to the Lumbee Tribe are disseminated and respect that principle.

Prior DCEs investigating PC screening have found that, among the attributes studied, decreasing one’s chances of dying because of PC is the most important attribute. We hypothesize that the results from this study will be consistent with that principal finding. Our study, however, will provide a culturally responsive approach to PC screening among American Indian men. The information from our study will inform future interventions with providers and American Indian men to increase PC screening.

The strength of our study is that it is being developed in collaboration with and under the approval of the Lumbee Tribe of North Carolina, one of the largest tribes in the eastern United States. The area in which members of the Lumbee Tribe reside is also rural, providing one of the first estimates of, which we are aware of, PC screening preferences in this population.

Like all studies, our DCE has limitations. First, indigenous tribes are incredibly diverse, and the findings from one tribe are not expected to apply to other tribes. Second, the low expected sample size precludes any possibility of investigating preference heterogeneity and latent class analysis. Additionally, while our attributes and approach have been guided by formative qualitative interviews, there may be other aspects that inform preferences of American Indian men in PC screening that were not captured and thus excluded from our DCE.