Adaptive trials

Chris Oldmeadow

01/04/2021

What are the problems with a standard RCT?

Design -> wait -> analyse -> learn from mistakes -> repeat
Uncertainty in many design parameters
Ignore what occurs during trial
Inefficient

Driving with your eyes closed

A different paradigm

Pallmann, P., Bedding, A.W., Choodari-Oskooei, B. et al. Adaptive designs in clinical trials: why use them, and how to run and report them. BMC Med 16, 29 (2018)

What is an adaptive trial?

A clinical trial design that allows for prospectively planned modifications to one or more aspects of the design based on accumulating data from subjects in the trial.

Adaptive Design Clinical Trials for Drugs and Biologics Guidance for Industry, Nov 2019.

What can we adapt?

Sample size
Patient population (Adaptive enrichment)
Allocation ratios
Treatment arm selection
Outcome and time-points

Using a GPS to find the best course

When is it appropriate?

Outcomes are available shortly after randomisation
Recruitment is slow

What about type 1 errors?

Traditional Frequentist framework

p-value = probability (under identical replicated experiments) of getting a result more extreme if the null hypothesis were true

α = Type 1 error rate

Probability of asserting an effect if the effect is truly zero
increases with more “looks at the data”, outcomes, groups

Bayesian framework

P = Posterior probability of effect

not conditional on a null hypothesis
not biased by the number of times it is calculated

Example: “pick the winner”

Goal: to select an appropriate treatment dose and confirm the safety and effectiveness
Do this efficiently
randomise subjects to one of N dose arms
carry the “best dose” forward to next phase

Response adaptive randomisation

Continually evaluate the data (Bayesian framework)
Probability each treatment is best
Adapt allocation ratio depending on P_best
Information goes where needed
Decision rules
Drop arms -> accelerates the process

Sub-group enrichment

some sub-groups might perform better for different interventions
A platform trial adapts both the number of interventions (the ratios), and the sub-groups

Advantages of Adaptive Trials

Statistical efficiency

More participants get better interventions

Drop interventions that don’t improve outcomes

Add new interventions when available

Increase/stop recruitment in subgroups

Disadvantages

pre planning adaptive design modifications -> much more front-loading of effort
longer lead times between planning and starting the trial.

Adaptive trials vs adaptive interventions

eg SMART (sequential, multiple assignment, randomised) trials
individual tailored interventions
decisions based at the individuals(or cluster) performance
ignore the collective information (different aims)