P1684: Automated Metrics From the Electronic Health Record Enable Evaluation of Palliative Care in Patients With Acute Myeloid Leukaemia: A Data Science Approach

Abstract

Background:Patients with acute myeloid leukaemia (AML) who receive early specialist palliative care (SPC) input report improved quality of life. However, rates of SPC referral for haematology patients have been found to be lower than for other types of malignancy. New therapies for AML allow for treatment of patients with relapsed disease and those who might previously have been judged unfit for treatment. Treating this patient group creates significant resource implications for the utilization of SPC services and this burden is likely to increase further in the future Aims:We aimed to audit the care provided at our centre to patients with AML and high-risk MDS (HR-MDS) against palliative care related quality metrics to highlight unmet palliative care needs. Methods:We leveraged routinely collected data from the electronic heath record system (EHRS) at a large tertiary haematooncology centre to describe the clinical course of all adult patients with newly diagnosed AML and HR-MDS between Apr 2019 and Oct 2022. The cohort was identified using the departmental database of AML patients. We excluded patients <18 years of age, patients with APML and those managed solely in other hospitals after their initial diagnosis. A data pipeline was constructed to extract data for each patient from the EHRS. Patient mortality data were from national mortality records. All data were analysed in R studio 3.1.4. Results:278 AML or HR-MDS patients were diagnosed and treated at UCLH during the study period. 131/278 (47%) patients had died as of 13/2/2023. 115/131 (88%) patients received chemotherapy for AML before death (40 high intensity induction regimen, 54 venetoclax containing regimen, 21 low intensity regimen). 34/131 (26%) patients received chemotherapy within the last 28 days of life (range 2-28). Total days spent hospitalized from diagnosis to death was higher for those treated with intensive induction regimens (mean 104 days) than venetoclax containing regimens (mean 62 days) (two sample t test, p= 0.001). The median number of days patients spent hospitalised in the last 90 days of life was 26 days (range 2-90 days).103/278 (37%) patients were formally referred for inpatient SPC review during an inpatient admission. Of these, 64 died (62% of referrals died, 48% of all those who died were referred). 21 patients were first referred to the palliative care team within the last week of life and 6 patients were referred within 24 hours of death. 12 patients were referred to the palliative care team within 30 days of diagnosis. The majority 37/67 (55%) of patients who died whilst an inpatient at UCLH had an inpatient referral to SPC. Initial treatment regimen did not affect rates of SPC referral for those who died (Fisher's exact test, OR 1.18, p = 0.8).