Table 1.
Brief overview of each scenario used in the mock costing exercise
Scenario/non-adaptive design | Adaptive design and features |
---|---|
1. A two-arm parallel-group randomised controlled trial assessing the addition of biomarker-testing to an existing early warning score in the management of patients with suspected sepsis in the emergency department | Group-sequential designa including a single interim analysis with futility stopping after half of patients have had primary outcome observed. |
2. A phase 2b randomised dose-finding study of JAK1 inhibitor for patients with active rheumatoid arthritis | Adaptive dose-finding design that has a single interim analysis after half of patients have primary outcome observed. The dose allocation used in the second stage is set according to an optimal allocation from a three-parameter emax model fitted to stage 1 patient outcomes. |
3. A multi-arm parallel-group phase 3 trial comparing regimens for treating intermediate and high-risk oropharyngeal cancer | Multi-arm multi-stage design with two interim analyses (2 years and 4 years into a 5-year recruitment period) that allows early stopping of experimental arms for lack of benefit. The trial continues to full enrolment unless all experimental arms stop early. |
4. A multi-arm parallel-group trial assessing clinical efficacy and cost-effectiveness of earlier treatment of ovarian hyper-stimulation syndrome (OHSS) | Adaptive umbrella design, allowing early stopping of arms within the two patient subgroups. In the early OHSS subgroup, a MAMS design is used with one interim analysis allowing stopping for lack of benefit; in the late OHSS subgroup, a group-sequential design with early stopping for lack of benefit is used. |
5. Randomised two-arm parallel-group trial of the efficacy of nicotinic acid derivative (NAD) for treatment of fatigue in mitochondrial disease | Sample size re-assessment design that will use blinded estimate of the pooled standard deviation to re-estimate the sample size required. If this is above a specified level, the trial will stop early for futility. |
aGroup-sequential designs are not always considered an example of an adaptive design but were included in the definition within this project as they also involve a pre-specified interim analysis of outcome data