Impact of an automatic palliative care consultation trigger on healthcare use in patients with relapsed/refractory acute myeloid leukemia.

224 Background: Patients with relapsed/refractory acute myeloid leukemia (AML) have poor outcomes and high levels of healthcare use at end of life. Palliative care remains underused in hematology and questions remain on how best to integrate palliative care for high-risk patients. We conducted a prospective cohort study evaluating palliative care consultation triggers for patients admitted to a tertiary academic center with advanced AML. Methods: Criteria were developed for all hospitalized patients with hematologic malignancy on the inpatient hematology floors and included: 1) persistent disease after ≥ 2 lines of therapy, 2) length of stay (LOS) > 7 days for symptom management, 3) ECOG performance status > 2, and 4) refractory GVHD ≥ 3 lines of therapy. Patients with relapsed/refractory AML were included if they met criteria 1. A nurse coordinator performed chart review of admitted patients 1-2 times/week from June to December 2020 on the inpatient hematology floor at Smilow Cancer Hospital and contacted the primary team when patients met eligibility. Patient characteristics and healthcare outcomes were compared with patients with AML meeting criteria 1 admitted pre-intervention (June to December 2019) using Fisher t-tests. Results: A total 110 admitted patients were eligible (64 pre-intervention and 46 post-intervention). Baseline patient and disease characteristics were similar, including mean age at admission (60.4 vs 60.9 years, p = 0.848), gender (64% vs 59% male, p = 0.691), prior transplant (56% vs 52%, p = 0.702), and AML risk stratification (67% vs 78% adverse risk, p = 0.283). In the post-intervention group, 61% of eligible patients were screened. Of the screened patients, 54% received a palliative care consult, 18% were declined by the primary team, 14% were marked as not eligible, and 14% did not have consult with reason unspecified. Overall, palliative care consults increased in the post-intervention group (22% vs 43%, p = 0.021). There was a significant increase in advance care planning and/or advanced directive documentation post-intervention (13% vs 28%, p = 0.049). There was no differences in pre- and post-intervention groups in LOS (12.13 vs 12.33 days, p = 0.941), 30-day readmissions (52% vs 39%, p = 0.557), critical/intermediate care escalation (22% vs 13%, p = 0.318) and non-palliative chemotherapy post-discharge (48% vs 39%, p = 0.246). Conclusions: A triggered palliative care referral intervention is feasible and doubled palliative care use in patients with relapsed/refractory AML, a group with high mortality and high healthcare utilization. Our intervention improved documentation of advance care planning. Although there were directional reductions in other healthcare use measures, the differences were not statistically significant, likely from the small sample sizes leading to the study being underpowered.