A rules-based algorithm to identify patients who would benefit from re-addressing advanced care planning.

Abstract

e21510 Background: Advance care planning (ACP) should be initiated early and readdressed often for cancer patients. We hypothesize that a rules-based algorithm based on occurrence of high-risk events can predict decreased overall survival, and can be used to target patients who would benefit from readdressing ACP. Methods: We performed a retrospective analysis of 221 patients receiving palliative chemotherapy with a diagnosis of leukemia, cholangiocarcinoma, esophageal, gastric, pancreatic, lung or urothelial cancer at the University of Chicago Medicine. Patients were included if they had an index outpatient oncology visit from April 1, 2015 through June 30, 2015. Starting at the date of index visit, we examined a three-month window for a “high-risk event,” defined as: 1. change in chemotherapy 2. emergency department visit 3. hospitalization. Patients were followed from index visit until date of death or last clinical encounter as of January 31, 2017. Each “high-risk event” was treated as a time-varying covariate in a Cox proportional hazards regression model to calculate a hazard ratio of death compared to those without an event. Results: Sixty-six percent of patients (146/221) experienced at least one high-risk event over the 3 month time frame. A change in chemotherapy regimen, an ED visit, and a hospitalization occurred in 53% (118/221), 10% (22/221) and 26% (57/221) of patients respectively. The hazard ratio of death for patients with at least one high-risk event when compared to those without was 1.86 (95% CI: 1.26-2.74, p = 0.002), when adjusted for age, gender, and race. Inpatient admission had the highest hazard of death among the high-risk events (HR 2.52: 95% CI: 1.69-3.76, p < 0.001). Conclusions: The rules-based algorithm identified patients with a greater risk of death. Implementation of this algorithm in the electronic medical record can identify patients with increased urgency to readdress goals of care.