Model of D− donor effect on levels of 4-functional T cells, adjusted for pre- and posttransplantation factors
| . | Estimate . | Robust SE . | Robust z . | P . |
|---|---|---|---|---|
| Intercept | 4.61 | 0.88 | NA | NA |
| Propensity | −0.51 | 0.33 | −1.54 | .12 |
| CMV reactivation | −1.31 | 0.85 | −1.55 | .12 |
| Steroids (> 1.0 mg/kg) | 1.08 | 1.25 | 0.87 | .38 |
| DPT | 0.003 | 0.003 | 0.94 | .35 |
| D− donor | −2.08 | 0.82 | −2.53 | .01 |
| . | Estimate . | Robust SE . | Robust z . | P . |
|---|---|---|---|---|
| Intercept | 4.61 | 0.88 | NA | NA |
| Propensity | −0.51 | 0.33 | −1.54 | .12 |
| CMV reactivation | −1.31 | 0.85 | −1.55 | .12 |
| Steroids (> 1.0 mg/kg) | 1.08 | 1.25 | 0.87 | .38 |
| DPT | 0.003 | 0.003 | 0.94 | .35 |
| D− donor | −2.08 | 0.82 | −2.53 | .01 |
The estimates are additive terms in a linear model predicting the percentage (untransformed) of 4-functional T cells. The estimate column gives the estimated regression coefficient for each term in the model. CMV reactivation, steroids and D− donor were coded as binary (0 or 1) indicator variables, so each of these estimates is a comparison of two groups. The estimated deficit associated with CMV-negative donors is 2.08, which is conservative compared to the crude difference of means (5.44-2.06) for the two donor CMV-status groups. The Robust z column is the estimate divided by the robust SE, a z-score greater than 2 in absolute value being statistically significant. The robust z is referred to a normal distribution, and the tail area multiplied by 2 to yield the 2-sided Wald-test P value. Only the D− donor effect was statistically significant, but other terms were included to adjust for potential confounding. The D− donor effect was stronger and more significant in all models in which non-significant terms were dropped.
NA indicates not applicable.