Increasing Access to Genetic Testing in Underserved Patients Using a Multilingual Conversational Agent

ABSTRACT Background. The volume of patients who meet national criteria for germline genetic testing based on a cancer diagnosis alone, regardless of family history (i.e. ovarian, pancreatic, advanced prostate, etc.) is rapidly growing. Germline genetic test results can inform oncology and surgical treatment decisions, as well as early detection

and prevention for the family. However, traditional pre- and post-test genetic counseling approaches may not be sufficient to meet the growing need. Additionally, traditional pre-test counseling may be a barrier, leading to decreased uptake, especially in those with advanced disease. Patients cared for in settings with limited or no

genetic services, including low-literacy, non-English speaking, and rural community patients already face disparities in access. As such, these patients remain underrepresented in clinical and research cohorts, and innovative strategies to optimize genetic counseling approaches are understudied. Relational Agents (RAs)

are an effective means of automating health education and counseling, as well as overcoming literacy barriers in the use of information technologies. RAs, animated computer characters simulate face-to-face conversation between a patient and a provider using verbal and nonverbal conversational behavior. Overall

Goals. This study will develop an English and Spanish RA to communicate personalized pre-test genetic education to a cohort of cancer patients who meet established cancer-based genetic testing criteria across two diverse clinical settings (including a low-resource, urban, safety-net hospital and a university medical center that

serves a significant rural population). We hypothesize that the use of an RA will increase the proportion of patients who receive genetic test results within 90 days of initiating cancer care, compared to usual care. Aim 1 is to develop an English- and Spanish-language RA using a patient-driven approach. Aim 2 is to conduct a

multisite randomized controlled trial of the RA to deliver pre-test education versus usual care in English- and Spanish-speaking patients to compare the proportion in each arm who receive genetic test results in 90 days at Los Angeles County+University of Southern California Medical Center and University of Rochester Medical

Center. Aim 3 is to understand the implementation context and identify facilitators and barriers to utilizing the RA in these clinical settings. Impact. Our deliverable will be multilingual patient-facing RA to deliver pre-test genetic education. This RA will be developed and evaluated in two distinct patient populations and clinic

workflows, can be rapidly updated as practice evolves, and will ultimately be available on the internet for clinics and patients to utilize. If successful, this would be a novel, effective, and scalable means of providing genetics education that could improve patient decisional preparedness, knowledge, and satisfaction, ultimately leading to

increased access for patient populations who are traditionally underserved in genetics. Understanding the implementation context and identifying facilitators and barriers to integrating a RA will increase sustainability and generalizability.