This was an observational cross-sectional mixed methods study. UCD was the main approach used. UCD is an iterative approach that involves all end users (and leaders) of a process or product in the entire design process of a solution [12,13]. It involves 5 phases (Figure 1) to understand the design needs, ideate potential solutions, rank order the solutions in relation to the identified design needs, prototype the highest ranked solutions, and then validate the solution prototypes by robustness, acceptability, and feasibility [14]. UCD is ideal for complex systems, with multidisciplinary end users and leaders from multiple organizations needing to interact to achieve a desired outcome [15]. UCD has increasingly been used in health care process and product design. The desirability, viability, and feasibility design thinking methodological orientation was used to analyze the data collected during the UCD process (Figure 2) [16]. The manuscript was written in adherence to the Consolidated Criteria for Reporting Qualitative Research (COREQ) checklist (Checklist 1).

Clinicians and staff from 4 high-level (Level I), 5 low-level (Level II) centers, and 3 nontrauma centers in Illinois participated. Illinois is divided into 11 emergency medical services regions. Each region has at least one Level I trauma center serving as a “resource center” that has quarterly meetings with Trauma Medical Directors and Emergency Medical Services Directors in their region. Patients with severe injury are undertriaged from the field to nontrauma centers 17%‐34% of the time [17-19]. Illinois has 2 tiers of trauma center designation, based on adherence to structural requirements and standardized processes [20]. Illinois Level II trauma center designation requirements mirror those of the American College of Surgeons low-level centers (level III and IV) and Level I trauma center designation requirements mirror those of the American College of Surgeons high-level centers (level I and II) [20]. The terminology non-, low-, and high-level trauma center is used accordingly in this manuscript.

The Northwestern University institutional review board approved the study (STU00213470). Participants were consented before each session and informed of their ability to opt out at any time should they not want to continue with the interview. Data from each session was stored on a password-protected, encrypted server that was accessible only by members of the study team. Participants were not compensated for their participation in this initiative. No identification of individual participants has been made in the manuscript’s figures or supplementary materials.

Four authors JDS, DJ, MX, and AMS conducted the sessions. JDS is a project manager and PhD candidate, with a MPH, DJ, and MX are candidates for a MS in Design. AMS is a trauma surgeon scientist and MS in Health Policy. JDS and AMS had prior experience in conducting mixed methods studies. Two research team members had previously interacted with participants during a previous Failure Modes Effects and Criticality Analysis [21]. The research team’s personal goals, reasons for doing the research, and interests in the topic were delineated in the informed consent script at the start of each virtual session.

The study team initially recruited hospitals by contacting trauma coordinators and surgeons for each hospital listed in the Illinois Department of Public Health Continuous Quality Improvement Committee listserve. They also contacted surgeon champions and surgical clinical reviewers for each trauma center in the Illinois Surgical Quality Improvement Collaborative listserve. Each resource included urban, suburban, and rural high-level, low-level, and nontrauma centers across the state. Snowball and purposive sampling were used to complete the recruitment of participants. Frontline end users who provide direct care to severely injured patients were recruited. These included emergency medicine physicians, trauma surgeons, emergency department (ED) nurses, and ED para-professionals. The study team also recruited transfer end users who participate in key retriage tasks that include transfer coordinators, bed managers, and ED clerical staff. Participant leader end users included trauma medical directors, emergency medical services directors, trauma coordinators, ED nurse clinical coordinators, and quality improvement leaders. Personalized participation invitations were extended via email. An institutional review board approved informed consent was obtained from all participants, initially, and before each of the 5 phases.

Data collection occurred by video-conference sessions, during business hours. Data collection sessions were held between March 2022 and August 2023 (Figure 3).

The study team conducted reflexive thematic analysis beginning with data collected in Phase I and continuing through the completion of Phase V. Transcripts and field notes from each phase were inductively coded by 5 study team members (JDS, DJ, MN, LC, and AC) to develop an initial codebook. Two study team members coded each transcript to promote iterative reconciliation and assure confirmability and dependability throughout the coding process. When reconciliation could not be reached, a third study team member or senior author adjudicated. Codes were organized into a coding tree to identify themes in design requirements and participant preferences for retriage solutions. Potential themes were checked and rechecked throughout the coding process to promote reflexive understanding of emerging trends in the data among team members [28]. After coding and theme generation, the study team evaluated and sorted codes and quotes to generate retriage solutions [29] MAXQDA 2022 (VERBI Software, 2021) was used for data management and analysis [30]. Transcripts were not returned to participants. However, member checking was completed when findings from the previous phase were presented back to participants from high-level, low-level, and nontrauma centers for comment at the beginning of each subsequent phase [31]. Participants corrected the team’s interpretation of findings.

A total of 49 frontline users and leaders participated in 19 sessions across the 5 phases (Table 1). A total of 18 participants attended more than 1 session.

The understand focus group included 7 trauma coordinators, 6 surgeons, 4 emergency physicians, 3 nurses, and 1 bed manager, representing 10 trauma centers. In previous work, we mapped the retriage process between non-, low-, and high-level centers at a high-level trauma center and 2 low-level trauma centers. Low-level and nontrauma centers in Illinois cited obstacles to determining clinical indicators for initiating the retriage process and challenges identifying level I trauma centers with space to accept the patient as 2 key opportunities for improving retriage [6]. High-level trauma centers cited obstacles coordinating patient transportation and efficiently exchanging clinical and radiologic data between sending and receiving centers as intervention targets [21]. These findings provided the basis for the three main design requirements for this study: (1) match patient needs with trauma center resources through timely communication; (2) identify accepting high-level trauma center efficiently; and (3) streamline the transfer of patient data between sending and receiving centers.

Focus groups with 10 participants from 3 low-level trauma centers included 2 trauma surgeons, 3 emergency physicians, 3 nurses, 2 trauma coordinators, and 1 health unit coordinator. Participants generated 70 solutions. Candidate solutions were then recategorized into 10 solutions by the study team, based on the 3 key design needs from Phase I (Table 2).

Three potential solutions included a checklist, a bed tracker, and a transfer coordination center. The suggested checklist includes standardized retriage criteria for non-, low-, and high-level trauma centers. The checklist was proposed to help reach consensus about the need for retriage to help match patient needs with trauma center resources (Figure 4).

Bed availability at high-level centers changes rapidly and requires sending centers to call multiple high-level trauma centers before finding an available bed and receiving acceptance of the transfer. Therefore, the bed tracker would need to be updated frequently to provide non- and low-level trauma centers with accurate availability of beds at high-level trauma centers.

A transfer coordination center could solve the need for a streamlined process of finding an accepting high-level center as well as assist in the sharing and transferring patient clinical data among non-, low-level, and high-level trauma centers. The transfer coordination center could reduce the multiple telephone calls to find an accepting high-level center while addressing conflicting opinions on patient needs requiring trauma center resources.

The study team conducted reflexive thematic analysis of transcripts across all 5 phases of the study. The analysis yielded 16 themes that provided the basis for the design requirements and identification of potential solutions to retriage obstacles (Multimedia Appendix 4).

This study found that a centralized, frequently updated bed tracker was the most effective and feasible informatics-based retriage solution. Within Illinois, the feasibility and potential effectiveness of a bed tracker were aided by the existence of hospital resource management data platform accessible to all hospitals across the state. However, the dashboard is underused. A future direction for further development of this solution requires encouraging use of the hospitals' resource management data platform.

Promoting increased uptake of the bed tracker among nontrauma, low-level, and high-level centers can enable Illinois’s trauma system to more effectively use real-time data to streamline the retriage process. In other states, bed trackers serve a crucial function of Medical Operations Coordination Centers, a model that helps trauma centers coordinate information exchanged between hospitals during transfer of severely injured patients [32,33]. States like Arkansas and Oregon use Medical Operations Coordination Centers available to all hospitals that provide frequently updated bed availability data, which has contributed to improved rates of timely, effective retriage in each state [33,34]. Our UCD process produced similar design requirements, adding valuable confirmatory evidence about the requirements for managing information associated with identifying the need for retriaging the resources available for meeting those needs.

Illinois’s trauma system is unique from most other states because it only has Level I, Level II, and nontrauma centers, rather than Levels I-IV, meaning that Illinois trauma centers at each level are uniquely heterogeneous in their capacity to care for severely injured patients [20]. However, our Failure Modes and Effects analysis of low-level or nontrauma centers identified that finding a receiving hospital with available beds was one of the most significant challenges, and there was demand for more effective tools to complete this step [6]. This clarifies the transferability of our study’s findings: applying informatics tools such as bed availability for trauma retriage in different locales (eg, US states) requires considering the heterogeneity in how different locales segment different levels of trauma care needs.

Informatics solutions have become common tools for improving care processes and access in health care contexts that require time-sensitive information exchange. While the retriage process depends on efficient coordination of resources and data exchange between hospitals, there is limited evidence exploring design requirements for informatics solutions in this context. This study is notable for its use of UCD to generate design requirements for an informatics solution that are responsive to retriage obstacles in a complex and heterogeneous trauma system. Informatics solutions that address complex processes in healthcare include approaches to equipment allocation in critical care medicine [35], patient triage during COVID-19 [36], and telehealth to improve rural trauma care access [37]. In addition, the UCD approach has been used to develop solutions for complex health care processes involving large, multidisciplinary teams [38], including hospital rounding [39] and aligning cancer treatment goals and progress between care teams [40]. Our study provides further support for the UCD approach in trauma system design while filling a gap in understanding around elements an informatics solution must have to improve retriage. It also suggests that broader adoption of UCD may enable systems to identify robust, acceptable, and feasible solutions to address complex challenges.

There were several limitations to this study. First, this analysis was prone to sampling bias. Most participants represented centers in a single urban geographic area, hindering the generalizability. However, the study team mitigated this bias by purposefully sampling stakeholders at each trauma center included in two statewide committees, which both include centers from more sparsely populated geographic areas. Second, the UCD approach is prone to detection bias because it relies on participant perceptions of processes or events that include subjective views. We sought to mitigate this bias by engaging a broad range of frontline users and leadership. Third, this study relied on participants to identify barriers in their hospitals' retriage process. This could have introduced recall bias. We mitigated this bias by holding multiple meetings with participants aimed at establishing trust necessary to properly identify design requirements for retriage solutions. Fourth, most interviews and focus groups were conducted virtually due to geographic barriers and COVID-19 restrictions at participating centers. However, the study team was able to leverage the virtual format by using a web-based design interface tool Figma that allowed participants to collectively complete exercises virtually.

This study leveraged UCD to create a centralized, frequently updated bed tracker as the most likely robust, acceptable, and feasible retriage solution. Future directions will focus on refinement of the solution design and implementation.