A High-Touch Model of Community-Based Specialist Palliative Care: Latent Class Analysis Identifies Distinct Patient Subgroups

Context Community-based palliative care may support seriously ill homebound patients. Programs vary widely, and few studies have described the heterogeneity of the populations served or service delivery models. Objectives To evaluate a diverse population served by an interdisciplinary model of community-based specialist palliative care and the variation in service delivery over time and identify subgroups with distinct illness burden profiles. Methods A retrospective cohort study evaluated longitudinal electronic health record data from 894 patients served during 2010–2013. Illness burden was defined by measures of performance status (Karnofsky Performance Status scale), symptom distress (Condensed Memorial Symptom Assessment Scale), palliative care needs (Palliative Outcome Scale), and quality of life (Spitzer Quality of Life Index). Service utilization included the frequency of visits received and calls made or received by patients. Latent class analysis identified patient subgroups with distinct illness burden profiles, and mixed-effects modeling was used to evaluate associations between patient characteristics and service utilization. Results The mean age was 72.3 years (SD = 14.0); 56.2% were women; 67.5% were English speaking; and 22.2% were Spanish speaking. Most had congestive heart failure (36.4%) or cancer (30.4%); 98.0% had a Karnofsky Performance Status score of 40–70. Four patient subgroups were identified: very low illness burden (26.2%); low burden (39.5%); moderate burden (13.5%); and high burden (20.8%). The subgroups differed in both baseline characteristics and palliative care service utilization over time. Conclusion The population served by a community-based specialist palliative care program manages patients with different levels of illness burden, which are associated with patient characteristics and service utilization.