Automated electronic health record frailty assessment for older cardiac patients

BACKGROUND: Frailty is an established risk factor for poorer outcomes in older hospitalised patients, but most measures require additional clinical review. Such formal frailty assessments are rarely undertaken in acute cardiac patients. However, routine electronic health records increasingly record health and functional deficits that together may represent the frailty phenotype. PURPOSE: To report the outcomes of acute cardiac patients in relation to an automated frailty measurement, derived from routine electronic health records. METHODS: This retrospective observational cohort study included consecutive patients aged ≥70 years old who were managed under the specialist care of a consultant cardiologist between April 2016 and August 2020 in three acute hospitals across Edinburgh, Scotland. The Continuous Dynamic Evaluation of Frailty (CODE-f) score was derived from national Care Assurance Standards data that is mandated for older patient hospital care. This includes measures of cognition, functional dependence, nutrition, falls risk, continence, skin health and mobility. A total of 31 data points were included in an unweighted frailty index with scores ranging between 0 (no markers present) and 1 (all present). No CODE-f score was generated if insufficient data was completed (>33% missing). The primary outcome was mortality at 1 year after hospital admission. Secondary outcomes were length of first (index) hospital stay and the number of days spent alive and out of hospital in the year after index admission (“home time”). In a nested cohort study of 318 consecutive patients at one hospital site, the Clinical Frailty Scale (CFS) was determined from manual electronic case note review for comparison with CODE-f scores. RESULTS: A total of 2,406 patients were included (mean age 79±6 years, 60% male). The CODE-f could be generated in 2,158 (90%) patients, with a mean score of 0.20±0.21. The primary outcome occurred in 352 (15%) patients. Those in the highest scoring CODE-f quartile (>0.35) had greater than 3-fold increased risk of death at 1 year compared to patients in the lowest quartile (<0.07), after adjustment for age and sex (27% versus 9%, adjusted odds ratio 3.44, 95% confidence intervals [CI] 2.47 to 4.82, p<0.001). Increasing median length of index hospital stay was observed at CODE-f scores above 0.3 (Figure A). In the highest CODE-f quartile, nearly one third of patients experienced less than 9 months home time in the following year, compared to fewer than 1 in 10 in the lowest two quartiles (Figure B). In the nested cohort study, CODE-f scores were well correlated with the CFS (r=0.50, 95% CI 0.41 to 0.58, p<0.001). CONCLUSION: An automated electronic health record measure can identify frail older adults at risk of poorer recovery and death after acute cardiac illness. This could inform complex treatment decisions and future care planning for this patient group. FUNDING ACKNOWLEDGEMENT: Type of funding sources: Public grant(s) – National budget only. Main funding source(s): Chief Scientist Office