End-of-life (EOL) trajectories among older adults dying with lung cancer (LC).

Abstract

e24008 Background: Medicare decedents often receive intensive care during the last month (mth) of life. There is little information on longer (6-mth) EOL trajectories of care intensity. Area hospice use rates may reflect supply and/or patient and physician preferences, and may influence patterns for individual decedents. Methods: Using SEER-Medicare linked registry and claims data, we selected decedents diagnosed with LC between 2008-2013 who survived ≥6-mths and died between 2008-2014. We linked Dartmouth Atlas data on hospital referral region (HRR) % cancer decedents with hospice use. Each mth we assessed claims to identify cancer-directed (CD) care (chemotherapy or radiation) and pharmacologic or other palliative care services (PCS) and assigned decedents to either CD only, PCS only, concurrent CD & PCS, full-mth inpatient (IP) or full-mth hospice. We ordered monthly care intensity from high to low (IP > CD only > concurrent CD & PCS > PCS only > hospice). Using the indicators arrayed by calendar mth, we assigned each decedent to 1 of 6 trajectories: stable (6-mth continuous) hospice, stable PCS only, stable CD only or concurrent CD & PCS, decreasing intensity, increasing intensity, and mixed (multiple directional shifts). Multinomial logistic regression estimated associations between area hospice rates, socio-demographics, and comorbidity with EOL trajectory, controlling for 1st line therapy, and diagnosis stage. Results: Our sample (N = 24,342) was predominantly male (53.7%), age ≥75 years (59.4%), and non-Hispanic white (80.5%); 19.1% lived in HRRs where ≤50% of cancer decedents received any hospice care. Trajectories were 7% stable hospice, 26% stable PCS only, 4% stable CD; 29% decreasing intensity, 9% increasing intensity, and 26% mixed. Relative to stable hospice, higher HRR-level hospice rates were associated with decreasing EOL intensity; higher age, female, and married were associated with increased probability of stable hospice enrollment); Black, non-Hispanic decedents had higher risk of increasing intensity (aRRR: 1.39, 95% confidence interval: 1.09-1.76, p < .01) and mixed patterns. Conclusions: Among older decedents with LC, only 62% had 6-mth EOL trajectories indicating low- (stable hospice or PCS only) or decreasing intensity, but few received persistent CD care. Area hospice use patterns, demographic characteristics and health status were associated with EOL trajectory. Additional research is needed to identify subgroups at risk of high or increasing intensity trajectories, and interventions that may shift trajectories towards lower intensity at EOL.