Background: Hospitalized patients who are at high-risk for mortality and readmission represent a unique population requiring tailored interventions to improve care outcomes. Advance care planning (ACP) conversations in the inpatient setting can help align care goals with patient preferences, yet their influence on measurable clinical outcomes in this vulnerable population is not well understood.
Methods: We conducted a retrospective cohort analysis of high-risk patients hospitalized on general medicine and oncology services at a tertiary academic medical center between January 2021 and March 2024. High-risk status was determined by combining an in-house developed 12-month predictive mortality algorithm and EPIC’s 30-day unplanned readmission risk score. Patients were high risk if they scored within the top 25th percentile of predicted 12-month mortality and had a >20% likelihood of 30-day readmission. Patients were categorized into two groups based on the presence of a documented advance care planning (ACP) conversation in the electronic medical record (EMR) during hospitalization: the ACP group and the no-ACP group. Patient demographics and clinical outcomes, including 30-day readmission, 30-day ICU hospitalization, hospice referrals, discharge disposition, in-hospital mortality, and time to death were compared between the two groups using univariate analysis.
Results: A total of 6,944 high-risk patients were included, with 1,166 (17%) patients in the ACP group and 5,778 (83%) in the no-ACP group. Patients in the ACP group were more likely to be male (52.4% vs. 44.8%, p< 0.0001) compared to the no-ACP group. There were no differences in non-English language preference or interpreter needs between groups. The ACP group had a longer length of stay during index hospitalization compared to no-ACP group (55.5 vs. 34.2 days, p< 0.0001). Thirty-day readmission rates were higher in the ACP group compared to the no-ACP group (49% vs. 31%, p< 0.0001). However, 30-day ICU readmissions were lower in the ACP group compared to no-ACP group (1% vs 2%, p< 0.0001). Patients in the ACP group showed a trend toward being discharged home more frequently (64% vs. 61%, p=0.09), while no-ACP patients were more likely to be discharged to a skilled nursing facility (26% vs. 22%, p=0.09). The ACP group had fewer days from hospital admission to death compared to the no-ACP group (133 vs. 160, p=0.0183. No significant differences were observed in hospice referral rates or in-hospital mortality between the groups.
Conclusions: In a high-risk population of hospitalized patients, documented ACP conversations were associated with longer length of stay during index hospitalization, higher 30-readmission rates, lower 30-day ICU readmissions and a trend toward discharge home. These findings highlight the complexity of care in high-risk patients and the need for further research to understand the role of ACP conversations in influencing clinical outcomes. We will further evaluate this relationship with analyses adjusting for medical complexity, and social determinants of health.
