Parameterization used for estimation of separate trial and treatment effects
| Trial . | Treatment combination . | Model linear component . |
|---|---|---|
| 1 | IDA and ara-C (IA) | μ + Zβ − (τ1 + βGO-IA)/2 |
| 2 | GO with or without IL-11 | μ + Zβ + (τ1 + βGO-IA)/2 |
| 3 | FAIG | μ + Zβ − τ2/2 |
| 4 | FAIG | μ + Zβ + τ2/2 |
| 5 | FAIG + ATRA | μ + Zβ − τ3/2 |
| 6 | FAIG + ATRA | μ + Zβ + τ3/2 |
| Trial . | Treatment combination . | Model linear component . |
|---|---|---|
| 1 | IDA and ara-C (IA) | μ + Zβ − (τ1 + βGO-IA)/2 |
| 2 | GO with or without IL-11 | μ + Zβ + (τ1 + βGO-IA)/2 |
| 3 | FAIG | μ + Zβ − τ2/2 |
| 4 | FAIG | μ + Zβ + τ2/2 |
| 5 | FAIG + ATRA | μ + Zβ − τ3/2 |
| 6 | FAIG + ATRA | μ + Zβ + τ3/2 |
IDA indicates idarubicin; ara-C, cytosine arabinoside; GO, gemtuzumab ozogamicin; IL-11, interleukin 11; FAIG, fludarabine, IA, and granulocyte colony-stimulating factor; ATRA, all-transretinoic acid; τ1, effect due to trial 2 versus trial 1, the unknown GO-IA trial effect; τ2, effect due to trial 4 versus trial 3, the FAIG trial effect; τ3, effect due to trial 6 versus trial 5, the FAIG + ATRA trial effect; βGO-IA, effect due to GO versus IA, the unknown GO-IA treatment effect; Zβ, the overall effect due to patient prognostic covariates; μ grand mean; and τ1 + βGO-IA, confounded GO-IA trial-treatment effect.