Page 146 - Haematologica August 2018

O.O. Odejide et al.
hospice use over the study period, with rates rising from 28.5% in 2000 to 56.5% by 2013 (Ptrend <0.001); however, there was no significant increase in late enrollment (12.2% in 2000 to 16.3% in 2013, Ptrend =0.19; Figure 2).
Aggressive EOL Care
Slightly over half of the cohort (55.8%) had at least one indicator of aggressive EOL care. Nineteen percent of patients received only one indicator of aggressive medical care, 16.0% received two, and only 0.1% received all six indicators of aggressive EOL care. Univariable associations between patient characteristics and aggressive EOL care are displayed in Table 4. In multivariable logistic regres- sion analysis, year of death was a significant determinant of aggressive EOL care. Specifically, we found significantly
lower odds of experiencing any indicator of aggressive care in more recent years compared to earlier years (Table 5). Patients who were transfusion-dependent (OR 3.40, 95% CI 2.87 – 4.04) or dialysis-dependent (OR 2.32, 95% CI 2.01 – 2.68) had significantly higher odds of having at least one indicator of medically aggressive care. We also found that age, sex, race, marital status, geographic region, comorbidity, and survival were significantly associated with having one or more indicators of medically aggres- sive EOL care (Table 5).
In univariable analysis examining the relationship between hospice use and medically aggressive care at the EOL, we found that 35.7% of patients who enrolled in a hospice experienced aggressive care compared to 74.5% among those who did not enroll. In a separate multivari-
Figure 1. Cohort assembly.
1382
haematologica | 2018; 103(8)