Abstract
100 Background: Colorectal cancer disparities loom large for underserved communities of color in the US where barriers to screening uptake can contribute to late-stage diagnosis and poor outcomes. Despite active outreach by skilled patient navigators (PN) at a NYC cancer center serving an ethnically minoritized and disadvantaged population, 59% (1,925/3,276) of patients either cancelled or did not show for their colonoscopy appointments in 2022. While PN re-engagement efforts led to 410 (21%) completing colonoscopy, 1,500 patients did not undergo potentially life-saving colon cancer screening that year. With the advent of conversational Artificial Intelligence (AI)-driven applications within health care offering a potential extension to a stretched workforce, our cancer center examined the use of an AI-based virtual patient navigator, MyEleanor, as part of a colorectal cancer screening quality improvement (QI) project. Methods: This QI project employed MyEleanor between Apr-Dec 2023 to target re-engagement of 2,400 patients nonadherent with colonoscopy appointments in 2022-2023. In place of human PNs, MyEleanor (a) called patients to discuss rescheduling, (b) assessed barriers to uptake, c) offered live transfers to clinical staff to reschedule, and d) provided procedure prep reminder calls. Evaluable outcomes included: (a) engagement with MyEleanor via identity confirmation, (b) live transfers accepted (actionable), (c) colonoscopy completion rate, and (d) patient volume, with (d) barriers to care, and (e) predictors of actionable engagement examined. Results: Over 8 months, 57% (1,368/2,400) of patients engaged with MyEleanor, with 58% (789) of this group, or 33% overall, accepting the live transfer. The completion rate for patients who did not show for initial appointment nearly doubled from 10% to 19% from 2022 to 2023 (pre to post-MyEleanor). Overall patient volume increased by 36%. Patients who engaged were a Mean age of 56.66 (41-79 yrs), female (66%), Hispanic (41%), Black (33%), English (73%) or Spanish (25%) speaking, and partnered (37%). Nearly one third reported at least two barriers to screening; top barriers included lack of perceived need (19%), time (18%), and MD encouragement (16%), medical mistrust (14%), and concerns about findings (13%) and cost (12%). Greater number of barriers predicted actionable engagement, (F(1366) = 354, p < 0.001), with Spanish-dominant patients and those declining to identify their race reporting nearly twice the number of barriers, F(1366) = 138.98 and F(1366) = 5.17, p < 0.001, respectively). Conclusions: This project demonstrates high potential of an AI patient navigator in helping to overcome patient attrition that can lead to colon cancer disparities while improving patient volume. Next phase of the project will examine impact on patient prep adherence, staff burden, and revenue.
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have