Skip to main content

Table 2 Issues we faced while designing CHARIOT and suggestions on how to deal with them, based on what we have learnt and wished we knew before starting the trial design

From: Challenges in implementing model-based phase I designs in a grant-funded clinical trials unit

Area Issue What to do
Software All trial details must be programmed Use existing software to get as close as possible
Funding A great deal of statistical input is required before the grant is awarded − This issue is more pronounced for the first few model-based trials that a clinical trials unit undertakes. Once the process is well-known, code will be in place and templates will have been designed for text in the protocol and statistical analysis plan
− Consider stating the design without much detail in the grant application. Leave some aspects of the design to be refined within the grant, during trial set-up
Data management Trial requires fast (e.g., within 24 hours) and accurate data entry − Have required data collated on a single case report form
− Accept data during teleconferences (as is done in 3 + 3 designs) and run the model during the teleconference
− Classify all dose-limiting toxicities as serious adverse events to ensure prompt reporting
Statistical knowledge − Trial statisticians need to check they have made good decisions
− Enhance trial statistician learning
– MAST (UK-based)
– Consult colleagues with experience in such designs through a registered clinical trials unit network
  1. MAST Methodology Advisory Service for Trials