Topic 3: Trial Design Flashcards
What is a Superiority Trial
Demonstrates an improvement in clinical outcome compared to the control group
What is a Non-inferiority Trial
Demonstrate the treatment is not inferior to the control
What analysis population does a superiority trial typically use
Intention to Treat
Why should you do both intention to treat and per-protocol
As a kind of sensitivity analysis, compare results and conclusions
What bias can intention to treat analysis population cause
Bias towards no treatment effect, potentially leading to a false conclusion of non-inferiority
What bias can per-protocol cause
Allows for exclusions of patients, so potentially leads to biased estimate of treatment effect.
Three types of trial design
Parallel, Crossover and Factorial
What is a crossover trial
Patients are randomised to a sequence of treatments and act as their own control.
What is a positive of crossover trials
Since patients are acting as their own control, requires fewer patients
What is a washover period and what does it avoid
A period of time between treatments where the participant receives no treatment. Avoids carryover effect, where the effect of treatment A impacts the effect of treatment B.
What is the duration of the washout period
The time it takes to return the participant to the original disease state, needs to be at the same baseline values as at the start of period 1. Depends on how long the drug remains active in the body.
What does the AB/BA crossover design mean
Patients are randomised to recieve either A first then B, or B first then A.
What is a factorial Design
Where combinations of interventions are tested
What is a 2x2 factorial design
patients are randomised to one of four combinations of two distinct interventions. Each treatment has two doses or levels
What is the objective of the factorial design
determine which combination is the optimum treatment. Testing combinations of interventions efficiently, needs less patients.
When is a factorial design appropriate
When treatments can be given in combination
What does a factorial design allow
Interactions to be detected. Can power for interactions.
How is the sample size for a factorial design determined
Determine sample size when there is no interaction then increase sample size to detect an interaction if needed. Do separate sample size calculations for each intervention, then choose the larger of these as the sample size.
What do we test in an equivalence trial
Whether the treatment is neither worse nor better.
In an equivalence trial, when do we conclude equivalence
When the confidence intervals lies entirely between some range delta and -delta, since the alternative hypothesis is that the treatment difference is bigger than -delta and less than delta.
What is the null hypothesis in an equivalence trial
The treatment is less than -delta, or more than delta, so its essentially outside of the range [-delta , delta].
When would we conclude non-inferiority
When the lower end of the confidence interval is greater than the non-inferiority margin.
What is the null hypothesis in a non-inferiority trial
The treatment difference is less than or equal to -delta
What is the alternative hypothesis in a non-inferiority trial
The treatment difference is greater than - Delta.
