Palliative Care Social Work Interventions and Co-occurring Outcomes: A Six-Year Descriptive Cohort.

Abstract

Palliative care social workers (PCSWs) contribute across patient, family, and team domains, but routine structured documentation of their interventions and observed outcomes is uncommon. To describe documented PCSW intervention and outcome categories in a structured EHR flowsheet, examine variation across teams and across first-visit versus aggregated documentation, and describe cases with documented staff or team de-escalation. Retrospective descriptive analysis of PCSW flowsheet documentation for 4993 adults consulted by palliative care between November 2019 and December 2025. Categories were developed by consensus among six PCSWs without formal psychometric testing. Patient-level co-occurrence was summarized as odds ratios (ORs) with Benjamini-Hochberg false discovery rate (FDR) correction and post hoc team-stratified and first-visit sensitivity analyses. Among 3287 patients seen by PCSW (65.8%), 18,560 interventions and 8877 outcomes were documented (analytic n = 2399). Of 234 co-occurrence comparisons, 149 of 153 remained significant under FDR (q < 0.05). Co-occurrences most robust to team stratification (I² = 0%) and first-visit restriction included having a discussion with consulting team with change in focus of care (OR 10.66) and avoided invasive interventions (10.35), and family counseling with improved family coping (7.55). Three co-occurrences showed substantial team heterogeneity. Staff de-escalation, documented in 524 patients (21.8%), was unevenly distributed across teams. Structured PCSW flowsheet documentation captured reproducible patterns of intervention and co-occurring outcome at scale. Several patterns were directionally consistent across teams and within first encounters and merit prospective evaluation as candidate quality indicators or staffing benchmarks. Same-clinician same-source recording and the absence of interrater reliability and construct validity testing constrain inference; future work will require multisite psychometric and prospective work.