Page 151 - Haematologica August 2018

End-of-life care for myeloma
dialysis-dependent myeloma patients. This may partially explain our finding that myeloma patients who were dial- ysis-dependent were significantly more likely to enroll late. Unlike transfusions, dialysis itself is unlikely to be palliative.35 Accordingly, rather than incorporating dialysis into hospice care, this group of patients may benefit from bridge programs that provide palliative care services before choosing to discontinue dialysis and transition to hospice. Although the rate of medically aggressive care at the EOL for this myeloma cohort (56%) was substantially lower than a prior analysis that included patients with var- ious types of hematologic cancers (78%),19 it is higher than
Table 5. Multivariable analysis of factors associated with receipt of at entire cohort of myeloma decedents from 2000 to 2013 (n=12,686).
that described for patients with solid malignancies in both single-institution and population-based studies (30- 35%).16,19 This intermediate rate supports the hypothesis that the complex features of myeloma that are similar to solid malignancies (e.g., incurability, pain) may ease the transition from more aggressive medical care toward symptom-directed care as compared to other blood can- cers. In a qualitative study of hematologic oncologists, physicians who focused on myeloma noted that the incur- ability of the disease made it less challenging to transition from disease-directed to symptom-focused therapies near the EOL.36 Moreover, in a population-based study of blood
least one indicator of medically aggressive care at the end of life among
Characteristic
Sex
Age at diagnosis (yrs)
Race
Marital status at diagnosis
College education (census tract quintile)
Median income (census tract quintile)
Region
Time from diagnosis to death
Modified Charlson comorbidity score
Dialysis-dependent
Transfusion-dependent
Year of death
Male
Female
65-69
70-74
Odds Ratio
Ref
0.90
Ref
0.91
0.76
0.55
Ref
1.36
Ref
0.88
Ref
0.86
0.86
0.85
0.78
Ref
0.97
0.85
1.00
1.09
Ref
0.68
0.71
0.73
Ref
0.62
Ref
2.11
Ref
2.32
Ref
3.40
Ref
0.85
0.81
95% Confidence Interval
0.83 – 0.97
0.81 – 1.03 0.68 – 0.86 0.49 – 0.61
1.23 – 1.50 0.81 – 0.95
0.76 – 0.98 0.75 – 0.98 0.73 – 0.99 0.66 – 0.92
0.86 – 1.10 0.74 – 0.98 0.85 – 1.17 0.91 – 1.31
0.60 – 0.77 0.62 – 0.81 0.66 – 0.82
0.58 – 0.68 1.95 – 2.29 2.01 – 2.68 2.87 – 4.04
0.77 – 0.95
0.73 – 0.90
75-79
≥80
White
Nonwhite
Married
Other
1 (lowest)
2
3
4
5 (highest)
1 (lowest)
2
3
4
5 (highest)
Northeast
South
Midwest
West
< 1 year
≥1year
0-1
2+
No
Yes
No
Yes
2000 – 2003
2004 – 2008
2009 – 2013
Only variables with P <0.05 in univariable analysis, specifically sex, age, race, marital status, median income census tract, college education census tract, region, modified Charlson comorbidity score,time from diagnosis to death,dialysis-dependence,transfusion-dependence,and year of death,were included in the multivariable logistic regression model to generate odds ratios. Odds ratio >1 indicate higher odds of receiving at least one indicator of aggressive care at the end of life.
haematologica | 2018; 103(8)
1387