Timing Prompts for Advance Care Planning Discussions and Advance Directive Completion.

Abstract

Advance care planning (ACP) is recommended to ensure that medical care for patients with serious illness aligns with their goals; however, effective pragmatic strategies have been elusive. To compare the association of an appointment-based ACP behavioral intervention delivered with the patient just before a primary care visit vs a non-appointment-based intervention. Retrospective cohort study of patients from a cluster randomized trial of ACP interventions conducted October 2019 to June 2022 (data analysis June to November 2025). This included 50 primary care clinics across 3 academic health systems. Participants were primary care patients with serious illness 18 years and older without an advance directive (AD) or Physician Order for Life Sustaining Treatment (POLST) in the electronic health record (EHR). Three different appointment-based automated ACP interventions delivered to patients 7 to 21 days before a primary care visit. Eligible patients who did not receive an appointment-based intervention received a non-appointment-based intervention after 6 months. Posthoc analysis of AD or POLST in the EHR at 12 and 24 months and documented patient-clinician ACP discussions by 24 months. Generalized estimating equation logistic regression models were used to compute adjusted differences. Among the original study sample of 8707 patients, there were 5810 patients with no AD or POLST in the EHR at baseline (mean age 71 [15] years; 3017 [51.9%] male). Of these, 5435 (93.5%) received an ACP intervention by 24 months: 2842 patients (52.3%) received at least 1 appointment-based intervention, and 2593 (47.7%) received only non-appointment-based interventions. After 24 months, 475 patients (16.7%) receiving any appointment-based intervention had an AD or POLST in the EHR compared with 254 (9.8%) receiving only non-appointment-based interventions (adjusted difference, 6.5 percentage points [pp]; 95% CI, 4.9-8.2%. Among patients receiving appointment-based interventions, 1143 (40.2%) had ACP discussion documentation in the EHR compared with 714 (27.5%) of patients receiving only non-appointment-based interventions (adjusted difference, 12.2 pp; 95% CI, 8.7-15.6). Appointment-based interventions were associated with more AD or POLST and documented ACP discussions than non-appointment-based interventions across all 3 ACP intervention groups. In this cohort of primary care patients with serious illness, an ACP intervention designed to engage patients and promote ACP discussions was associated with greater uptake if delivered before an office visit.