Table 1.
Typical model parameter estimates with SDs, and random effect variances with SDs
| Structural model | |||||
|---|---|---|---|---|---|
| Parameter | Estimate | SD | Ω | SD | |
| λ0 | Proportion of theoretical thymic output13 (cells/day) | 0.216 | 0.0711 | 1.57 | 0.55 |
| d 0 | Proportion of expected loss (/day) | 0.477 | 0.0959 | 1.62 | 0.386 |
| p 0 | Proportion of expected proliferation (/day) | 0.207 | 0.0239 | 0.251 | 0.0960 |
| X 0 | Initial concentration of T cells (cells/μL) | 168 | 21.5 | 1.31 | 0.206 |
| λh | Time to recovery in thymic output (days) | 133 | 20.3 | 1.27 | 0.247 |
| λr | Rate of recovery in thymic output | 9.66 | 1.36 | 1.22 | 0.431 |
| σ | Variance of the residual error | 0.219 | 0.0167 | — | — |
| Covariate model | ||||
|---|---|---|---|---|
| Parameter | Covariate | Effect size | SD | P value |
| X 0 | Alemtuzumab | ‐0.842 | 0.029 | <0.001 |
| X 0 | ATG | ‐0.939 | 0.052 | <0.001 |
| X 0 | Acute GvHD | 0.283 | 0.196 | <0.001 |
| λ0 | Leukemia | 1.32 | 0.442 | <0.001 |
| p 0 | No conditioning | ‐0.844 | 0.025 | <0.001 |
ATG, antithymocyte globulin; GvHD, graft‐vs.‐host disease.
Parameter estimates and the random effect variances (Ωs) were estimated from the model‐building dataset. The SDs for both the parameter means and for the variances of the random effects were found through 200 bootstrap samples using PsN version 3.5.3.44 The significant categorical covariates were included through multiplication of the parameter by (1+Effectsize), testing the null hypothesis that the effect size is zero.