Instability in End-of-Life Goals and Preferences of Patients Who Are Seriously Ill: A Systematic Review.

Importance: Advance care planning (ACP) aims to ensure goal-concordant end-of-life (EOL) care based on patients' prerecorded goals and preferences, assuming these remain stable. However, this assumption has not been systematically examined. Objective: To assess evidence on stability of EOL care goals and preferences over time among patients who are seriously ill, and identify factors associated with changes over time. Evidence Review: PubMed, EMBASE, and Scopus were searched for English-language longitudinal studies published between 2014 and September 2024 assessing outcomes at least twice among adults with a serious illness were included. Studies lacking sufficient data to calculate proportions of patients with unstable goals and preferences during the study period were excluded. Studies were assessed using dual independent screening, single-author data extraction and quality assessment with verification by second author. Study-level regression analyses and narrative synthesis were used to identify factors associated with instability. Findings: In this systematic review of 25 studies examining data from 5711 patients across 7 countries, 10 studies found that at least 50% of patients who are seriously ill changed EOL care goals and preferences over time. Fifteen studies were rated good quality, and 10 as fair quality. Study-level regression of studies rated as good quality using the National Institute of Health Quality Assessment Tool for Observational Studies found a higher number of assessments was significantly associated with greater observed instability (ie, a higher proportion of patients changing their goals or preferences (β = 0.07; 95% CI, 0.02 to 0.12). ACP intervention between assessments, time between assessments, follow-up duration, and patient illness (cancer vs noncancer) were not significantly associated. Narrative synthesis identified that changes in health status, emotional states, and prognostic understanding were key influences on instability. Prior ACP did not stabilize preferences over time. Conclusions and Relevance: In the studies included in this systematic review, EOL care goals and preferences among patients who are seriously ill were unstable and influenced by transient clinical and emotional factors. These shifts often go undetected when preferences were assessed infrequently. These findings suggest that ACP should be reframed-not as a future-focused plan for goal-concordant EOL care, but as an ongoing approach that prepares patients who are seriously ill and their families for real-time decisions in the face of uncertainty.