Table 1 Major tasks required to run a clinical trial and how adaptive designs may affect them
Major tasks | How might an adaptive design affect the task? | Potential resource implications |
|---|---|---|
Development of trial design, protocol and trial materials including SOPs | More scenarios to plan, possibly involving pre-trial simulation studies, and more milestones | Additional statistical and trial management staff resource |
Regulatory, ethical and governance applications | Increased complexity in communicating the design in applications, greater chance revisions may be required, more complex contracting | Additional trial management, statistical, and administrative staff resource |
Database set-up and maintenance | Case report forms and database may change during the trial due to adaptations; more complexity and thought needed based on the scenario planning during trial development to enable adaptations to be managed efficiently; more testing required | Additional data management, programmer, statistical and trial management staff resource; higher fees for outsourced services |
Randomisation system set up and maintenance | Randomisation method may be bespoke and not implemented in standard systems; randomisation systems may need updating during the trial; more complexity and thought needed based on the scenario planning during trial development to enable adaptations to be managed efficiently; more testing required | Additional data management, programmer, statistical and trial management staff resource; higher fees for outsourced services; cost/time of making changes to randomisation systems if trial changes |
Site set-up (and securing service support costs and excess treatment costs) | Contracts may need to reflect variability in expectations of recruitment periods, breaks in recruitment, expectations on data entry and cleaning to enable robust decisions based on timely cleaned and locked data. There may also be variability in excess treatment costs required due to change in dose or sample size. More frequent site training generally required | Additional trial management resource |
Data queries and cleaning | Requires more time and ongoing review for cleaning | Additional data management staff and statistical resource |
Interim analyses and data monitoring | May require more monitoring at centres or centrally in addition to data cleaning prior to data lock; requires interim statistical analysis plan; requires time for interim statistical analyses | Additional data management, statistical and trial management staff resource |
Statistical analysis plan | Requires rigorous upfront development with scenario planning; may involve running extensive simulations to ascertain the design’s operating characteristics | Additional statistical staff resource |
Statistical analysis | May require additional staff to protect core team from knowing accumulating, comparative results; may involve more advanced statistical methods (e.g. for point and interval estimation, multiplicity adjustment or Bayesian analysis) and additional programming | Additional statistical staff resource |
Close-down | Timing will depend on outcomes of interim analyses | May require more buffer room to allow for uncertainty, especially for smaller institutions |