Fig 1.
Delayed-start design under conditions of (a) a disease-modifying treatment effect and (b) a symptomatic treatment effect.
Δ1 represents the initial treatment effect at the end of the placebo-controlled period (T1) and Δ2 represents the treatment effect at the end of the delayed-start period (T2). The hypothetical results for a symptomatic treatment (b) assume that the time to peak treatment effect is equal to the duration of the delayed-start period.
Table 1.
Candidate test procedures for noninferiority test.
Table 2.
Underlying mean ADAS-Cog scores for simulations in the disease-modifying treatment effect scenario (Scenario 1) and the symptomatic treatment effect scenario (Scenario 2).
Table 3.
Type I error rates for candidate test procedures and various conditions of missingness mechanism, discontinuation rate, and data inclusion
Table 4.
Power estimates for candidate test procedures under various conditions of missingness mechanism, discontinuation rate, and data inclusion
Fig 2.
95% confidence intervals for conditional bias.
25: 25% discontinuation rate; 40: 40% discontinuation rate; All: all data from all randomized patients in both study periods; DS: delayed-start period data only; MAR: missing at random; MCAR: missing completely at random; MNAR: missing not at random; S1: scenario 1, disease-modifying treatment effect scenario; S2: scenario 2, symptomatic treatment effect scenario; T1: time 1, end of placebo-controlled period; T2: time 2, end of delayed-start period. Each line segment represents the 95% confidence interval for the conditional bias estimated across various missingness mechanisms, conditions of underlying drug effect, discontinuation rates, and modeling approaches.
Fig 3.
Power curves for underlying treatment scenarios for method 4 (Δ2–0.5 Δ1 > 0) for 25% patient dropout (left) and 40% patient dropout (right).
Fig 4.
Delayed start analysis of ADAS-Cog14 scores in patients mild AD from the EXPEDITION, EXPEDITION2, and EXPEDITION-EXT studies of solanezumab.
Abbreviations: ADAS-Cog14 = 14-item Alzheimer’s Disease Assessment Scale—Cognitive subscale; Q4W = every 4 weeks. Note: All p-values shown for values <.05 and endpoint. Error bars represent standard error. Dashed line indicates end of EXPEDITION and EXPEDITION-2 (feeder studies) and beginning of EXPEDITION-EXT.