Alternative Designs Flashcards
What are the advantages of factorial design trials?
Efficiency in testing the effect of two different interventions in one trial
Interaction between treatments and effect of combined therapy
Sample size (assuming no interaction between interventions)
List some disadvantages of factorial design trials
Lack of power to detect interactions
Practical limitations in consenting participants for multiple interventions
Compliance
Biased main effects in the presence of interaction
What is a suitable control for a non-inferiority trial?
An active treatment whose benefits have been established
Why is a standard significance test a poor choice for non-inferiority studies?
Encourages small trials
What is the most important variable when it comes to determination of sample size in non-inferiority trials?
Delta (the smallest difference, which if excluded in a favourable direction, would indicate non-inferiority)
What points need to be considered before undertaking a non-inferiority trial?
Both treatments effective or neither
Provision of standard treatment
Primary response and duration of treatment/follow-up
True non-inferiority or poorly conducted trial
Intention-to-treat vs. per-protocol analysis
What type of analysis is more suitable for non-inferiority trials?
Per protocol - only analysing individuals who comply with protocol perameters
If ITT and PP analyses agree - this shows strong support for non-inferiority
What common design error is important to consider in the design of cluster trials?
Correlation that likely exists between individuals within the same cluster
How does cluster correlation affects trial design?
Randomisation - only randomise a few clusters (risk of substantial imbalance)
Sample size - degree of correlation, randomisation etc can affect sample size
Informed consent - use of cluster guardians e.g. leader of a community/headteacher at a school
Why is a washout period important in the design of crossover trials?
he reason for a washout period is to ensure there are no remaining effects of the first treatment carried over into the treatment phase for the alternative treatment. Note that it is not always possible to include such a period particularly in situations where it is considered unethical to withhold treatment, e.g., in an asthma trial of two steroid inhalers.
What is the typical scenario that lends itself to a crossover design?
Short term effects of chronic disease course e.g. relief of symptoms
What are the two particular effects which need to be checked during crossover trials
Period effects - A period effect is when the particular outcome or measurement of interest changes with time irrespective of any treatment effect. For example, participants may, on average, improve (or deteriorate) between the first period and second period.
Carry-over effects - A carry-over effect is when the impact of treatment depends on the period in which the treatments are given, i.e. there is an interaction between period and treatment. For example, the benefit of some treatment over another may be greater in period 1 than period 2.
How might a carry-over effect occur?
- Inadequate washout
- Psychological carry-over affecting patient judgement
- Underlying condition may improve/deteriorate irrespective of interventions
Which of the two effects is more significant period or carry-over?
Carry-over
What are the advantages of crossover trials?
Particiapants will get both treatments - no feelings of missing out
Within person comparisons need a smaller sample size than parallel trial
