78.5, 56.2% female, 18.9% black). The ICG-R mean score was 44.5 (SD 9.5), and varied from 11 to 55 (lower score indicating worse grief). A multivariate linear regression found the hospice primary caregiver's perception that the patient's wishes were not followed was associated with a 4.7 point lower grief score. Other measures associated with worse grief included concerns with coordination (4.1 points lower, 95% CI -7.0, -1.2), lack of spiritual support (3.7 points lower, 95% CI -6.9, -0.42), and no prior experience with hospice (2.7 points lower, 95% CI -4.6, -0.74). Persons with a formal advance directive had improved grief score by 2.3 points (95% CI, 0.2, 4.4).Conclusions. Primary caregivers' prior experience with hospice and their perceptions of concerns with the quality of care are associated with their level of grief 6 months after the death of the patient.Implications for Research, Policy, or Practice. This cross-sectional study suggests that efforts to improve hospice quality may lead to diminished grief. ObjectivesUnderstand the complex relationships among palliative care, depression, and survival. Report the results of a secondary data analysis of 2 RCTs. Original Research Background. We previously demonstrated improved depression and survival in advanced cancer patients participating in two palliative care RCTs. ENABLE II (EII; n¼322) compared intervention versus usual care and ENABLE III (EIII; n¼207) compared immediate versus delayed intervention. The interventions were similar (eg, in-person PC consultation, weekly phone sessions facilitated by a nurse coach, and monthly follow-up calls), except in EIII there was a delayed intervention group (beginning 12 weeks after enrollment) and a caregiver intervention. The Center for Epidemiologic Studies-Depression (CES-D) was collected at baseline and approximately every 12 weeks until death or study completion. Research Objectives. To determine whether baseline depression moderates the effect of the intervention on survival in the combined RCTs' sample (n¼529; intervention n¼368; usual care n¼161). Methods. A Cox proportional hazard analysis wasconducted with (a) intervention (as a time-varying covariate), (b) baseline CES-D scores, and (c) their interaction, entered simultaneously.Results. There was a significant effect of the interaction (intervention x CES-D) on mortality risk (p¼.035), indicating a moderating role of depression. To clarify the nature of this interaction, we classified patients as depressed (baseline CES-D>16) or not and conducted a separate Cox analysis within each depression group that included intervention as the sole predictor variable. Among depressed patients, receiving the intervention was associated with lower mortality risk (HR ¼ 0.65, CI: 0.44-0.95, p¼029), but this relationship was not significant among nondepressed patients (HR ¼ 0.89, CI: 0.65-1.21, p¼45).Conclusions. The ENABLE intervention effect of reduced mortality risk was moderated by baseline depression such that the magnitude of the intervention effect increased as ...