Development of a Dashboard for End-of-life Care at an Academic Hospital.

6590 Background: Accurate, reproducible, transparent and continuous healthcare utilization measures derived from structured EHR data may facilitate higher value cancer care at the end of life in both community and academic settings. Prior manually abstracted data from Smilow Cancer Hospital at Yale-New Haven showed high rates of chemotherapy and hospital utilization within 30 days of death. However, these data were not actionable due to the challenges with manual collection, physician attribution and date of death. Methods: The Yale Smilow Cancer Hospital and Flatiron Health used a commercially-available obituary source to supplement EHR data for a cohort of patients who received care at Smilow and died in 2016 - 2017. We developed an algorithm to attribute each patient to the correct oncologist based on visit frequency. We then used structured EHR data to measure rates of utilization for the following measures within 30 days of death: inpatient admission, ICU, chemotherapy and immunotherapy. Automated reports with internal benchmarks were generated to summarize resource utilization by individual physician, disease team, and practice site with patient level detail. Prior to wide scale launch across our enterprise, we provided oncologists in our community based practices feedback on use of chemotherapy in the last 30 days of life. Results: We found high rates of utilization at the end of life at both academic and community sites, and were able to identify outliers at the site of care, disease team and physician level. In the group that received quarterly feedback on chemotherapy in the last 30 days of life we saw a 23% improvement (see table). Conclusions: Performance dashboards with patient-level granularity identify performance outliers and opportunities for care improvement interventions. The reusable date of death, physician attribution and dashboard infrastructure allows measurement over time and rapid development of new measures. Efforts are underway to apply patient risk stratification to interpretation of practice variation, in addition to benchmarking against a national patient sample. Chemotherapy in final 30 days. Timeframe Measure Score N 2017 Q1 30.3% (22.3, 39.5) 109 2017 Q2 24.6% (18.2, 32.5) 138 2017 Q3 23.3% (16.2, 32.3) 103