Associations between Medicare‐Billed Advance Care Planning (ACP) Among Seriously Ill Patients and Their End‐of‐Life Healthcare Utilization

Abstract

Research ObjectiveMedicare began paying clinicians for advance care planning (ACP) conversations in 2016. Research suggests that ACP is associated with less intensive end‐of‐life (EOL) care and greater concordance with patient preferences. Seriously ill (SI) patients may be an important group to engage in ACP discussions. We examined the association of Medicare ACP visits with EOL health care utilization for these patients.Study DesignSI patients were identified prospectively in 2016 using ICD‐10 codes to capture patients with an expected survival of less than two years or fewer and/or substantial suffering. Among SIP decedents in 2017, we conducted patient‐level analyses in which the exposure variable was receipt of first billed ACP (none, any visit more than 1 month before death), and the dependent variables were commonly used measures of EOL intensity (inpatient admission or emergency department (ED) visit or ICU stay within 30 days of death, in‐hospital death, late first hospice within 3 days of death, health care expenditures in last 30 days), adjusting for age, race, sex, and Charlson comorbidity index. Analysis of spending was further stratified by Dartmouth Hospital Referral Region (HRR)—high‐, medium‐, and low‐spending regions.Population StudiedAmong 926 995 SI patients identified in 2016, 105 568 (11.4%) died in 2017; 8252 (7.8% of decedents) had at least one billed ACP visit billed visit prior to 1 month before death.Principal FindingsAfter regression adjustment, SI patients with a first ACP visit more than one month before death experienced significantly less intensive EOL care. They were less likely than SI patients without ACP to be hospitalized within a month of death (OR = 0.82; CI:0.78‐0.86), to have an ED visit (OR = 0.82; CI: 0.78‐0.86), to have an ICU stay (OR = 0.85; CI: 0.80‐0.90), and were less likely to die in hospital (OR = 0.84; CI: 0.79‐0.88). There were no differences in timing of hospice enrollment (OR = 0.96; CI: 0.91‐1.02). Adjusted mean expenditures were $473 higher overall for SI patients with ACP (CI: $6.19‐$941.30). In stratified analyses, expenditures for patients with ACP were higher than patients with ACP in the highest HRRs, lower in the lower HRRs, and not significantly different in the medium HRRs.ConclusionsA strength of this study is that SI patients were identified prospectively and followed until death. ACP was associated with less intensive EOL utilization, although the association with EOL expenditures varied by region. Limitations included our inability to assess if the EOL outcomes were concordant with the discussed goals of care for the patients, and the likelihood that billed Medicare visits underestimate the extent of ACP in real practice or prior to 2016.Implications for Policy or PracticeAlthough ACP may increase the delivery of care aligned with patient preferences, it may not be cost‐saving. Given slow uptake of billed ACP visits, CMS may wish to explore methods to incentivize ACP services, especially among high needs groups. In health systems with limited resources, it may pay to target SI patients for early ACP discussions.Primary Funding SourceNational Institutes of Health.