Table 3.
Multivariate linear mixed effects model of post-transfusion HbA decline
| Univariate General Linear Model | Multivariate Mixed Effects Model | |||
|---|---|---|---|---|
| Parameter estimate (g/dL/week) |
p-value | Parameter estimate (g/dL/week) |
p-value | |
| No splenectomy | 0.107 | <0.0001 | 0.096 | 0.0022 |
| Extended RBC antigen matching | 0.066 | 0.0045 | 0.050 | 0.380 |
| Transfusion characteristics (fraction of units/transfusion) | ||||
| Severe G6PD deficiency | 0.267 | 0.0013 | 0.162 | 0.0238 |
| Moderate or severe G6PD deficiency | 0.102 | 0.034 | ||
| Fy-null | 0.066 | 0.038 | 0.055 | 0.139 |
| S-negative | 0.061 | 0.054 | ||
| U-negative | −0.24 | 0.28 | ||
The decline in HbA (ΔHbA) was represented as the change in HbA (g/dL) from immediate post transfusion to the next scheduled transfusion time point, divided by the number of days between the 2 post-transfusion time points. ΔHbA units were then converted to g/dL/week for this analysis. For the multivariate model, only severe G6PD deficiency (not moderate G6PD deficiency) of donor units was introduced to the model.