Patterns of Referral to Embedded Palliative Care and Impact of Timing on End-of-Life Outcomes.

CONTEXT: Early palliative care referral has demonstrated benefits in quality of life and end-of-life outcomes, yet results have been inconsistent across cancer types. Previous guidelines recommended time-based referrals, but optimal timing may vary by individual patient factors.

OBJECTIVES

Investigate factors associated with the timing of referral to palliative care and the impact of timing of referral and the number of appointments on end-of-life outcomes.

METHODS

We conducted a retrospective analysis of 779 patients with metastatic solid tumors referred to outpatient palliative care at two cancer center sites (2021-2023). We examined referral patterns and analyzed associations between timing, patient characteristics, and end-of-life outcomes. For bivariate analysis, "early" was defined as referral within three months of diagnosis.

RESULTS

Overall, 38.9% of patients received early referrals, with significant variation by cancer type (P < 0.001): patients with head and neck cancer had the highest early referral rate (53.3%) while those with breast cancer had the lowest (9.6%). Patients with Medicaid insurance were more likely to receive early referral (42.8% vs. 35.2%, P = 0.04). Among 321 patients who died during follow-up, earlier referral relative to death was associated with increased hospice enrollment (OR 1.016 per month, P = 0.01), reduced end-of-life chemotherapy (OR 0.964 per month, P = 0.002), lower hospital death rates (OR 0.988 per month, P = 0.04), and decreased ICU utilization (IRR 0.962, P = 0.02). The timing of referral proved more influential than visit intensity across all end-of-life outcomes.

CONCLUSION

Palliative care referral timing varies significantly by cancer type and insurance status. Earlier referral relative to death influenced end-of-life outcomes more than visit intensity, supporting stepped models, which have found noninferior outcomes to early palliative care despite fewer visits.