TABLE 4.
FINAL MODELS FROM MULTIVARIABLE PROPORTIONAL HAZARDS REGRESSION ANALYSIS
Variable* | Reference Group* | Parameter Estimate |
Standard Error |
Chi Square |
P* | Hazard Ratio* |
95% CI |
---|---|---|---|---|---|---|---|
A) RBC TRANSFUSION INDEPENDENCE MODEL | |||||||
RBC in 10 Days Prior | None | −0.62 | 0.08 | 58.70 | < .0001 | 0.54 | 0.46, 0.63 |
CD34+ Dose Category | 79.50 | < .0001 | |||||
Medium | Large | −0.22 | 0.09 | 6.50 | 0.01 | 0.80 | 0.68, 0.95 |
Small | Large | −1.15 | 0.13 | 79.40 | < .0001 | 0.32 | 0.25, 0.41 |
ABO Match | 16.30 | 0.001 | |||||
Major mismatch | Identical | −0.46 | 0.12 | 15.60 | < .0001 | 0.63 | 0.50, 0.79 |
Bidirectional mismatch | Identical | 0.02 | 0.15 | 0.00 | 0.91 | 1.02 | 0.76, 1.37 |
Minor mismatch | Identical | −0.14 | 0.14 | 1.00 | 0.31 | 0.87 | 0.67, 1.14 |
B) PLATELET TRANSFUSION INDEPENDENCE MODEL | |||||||
Platelets in 10 Days Prior | None | −0.63 | 0.10 | 40.00 | <.0001 | 0.53 | 0.44, 0.65 |
Diagnostic Category | 18.40 | <.0001 | |||||
Intermediate risk heme | Standard risk heme | −0.10 | 0.13 | 0.60 | 0.44 | 0.90 | 0.70, 1.17 |
High risk heme | Standard risk heme | −0.11 | 0.15 | 0.50 | 0.48 | 0.90 | 0.66, 1.21 |
Benign | Standard risk heme | −0.34 | 0.20 | 2.80 | 0.09 | 0.71 | 0.48, 1.06 |
Solid tumor | Standard risk heme | 0.27 | 0.17 | 2.50 | 0.11 | 1.31 | 0.94, 1.83 |
Regimen Category | 56.30 | < .0001 | |||||
Bone marrow myeloablative | PBSC myeloablative | −0.88 | 0.14 | 37.70 | < .0001 | 0.41 | 0.31, 0.55 |
PBSC nonmyeloablative | PBSC myeloablative | 0.54 | 0.15 | 12.60 | 0.0004 | 1.72 | 1.28, 2.32 |
TLD + PBSC nonmyeloablative | PBSC myeloablative | 0.27 | 0.13 | 4.40 | 0.04 | 1.32 | 1.02, 1.70 |
Time Period | 23.60 | < .0001 | |||||
2001-2005 | 2006-2010 | −0.43 | 0.10 | 20.60 | < .0001 | 0.65 | 0.54, 0.78 |
1993-2000 | 2006-2010 | −0.40 | 0.10 | 14.80 | 0.0001 | 0.67 | 0.55, 0.82 |
Notes and Definitions:
Categories listed in the “Variable” column are compared to the category listed in the “Reference Group” column.
P: The P value for overall significance of each variable in the model is shown in bold. Non-bolded P values are interpreted differently; they only help assess relative effects among different categories of the same variable (smaller suggests a stronger difference).
Hazard Ratio: The hazard ratio characterizes the ratio in instantaneous chance of becoming independent relative to those in the reference group; values greater than 1 reflect faster attainment of transfusion independence.