Randomised Studies Flashcards
Safeguards
Blinding (masking)
Avoiding biases
Avoiding confounding
What is an intervention study?
Preventative or therapeutic measures.
The investigator attempts to start with two or more groups of individuals who are similar in all respects other than the fact that one group is assigned to receive an intervention whilst the other isn’t.
Choice of comparison treatment
Placebo
Or
The current best treatment.
A careful consideration is that participants should not be deprived of effective interventions that they would receive in the absence of a trial.
Why should the control group always receive something?
- The potential for bias due to lower compliance among the comparison group who receive nothing from the trial.
- Unethical for them to not gain anything from the trial.
Principles of equipoise
General principle in which intervention studies are only justified when there is genuine uncertainty as to the relative benefits of the two interventions.
Sample size
Should be big enough to create statistically significant difference between the two groups.
Several other factors also need to be taken into account such as compliance.
Sampling
The aim is to obtain two groups which are similar in all respects other than the introduction of the preventative or therapeutic measure under study. Unless this is done, the underlying difference between the study groups may be responsible for any differences in outcome.
To check the similarity if he groups baseline data is useful. It should include all variables which are known or thought to affect the outcomes of interest.
The bigger the sample the better chances of distributing both known and unknown risk factors evenly in an RCT.
How do you get around selection bias?
Aim: remove the element of human choice.
Best way random allocation. Allocation should not be known in advance.
Other way: stratified random allocation.
Participants are separated into different strata based on their risk factors.
Other way is systematic allocation: determining allocation in advance. However there could still be a bias there because of the structure of the allocation (male, female) and room for human bias.
What may a lack of repeatability result from?
Intra-observer
Inter-observer
Blinding
Bias can arise both from the participant and from the observer. Blinding can help prevent differential misclassification. In this way neither the participants or the observers know which group the participant is in. That is when the study is double-blind. When it is single-blind only the participant or the observer are blind.
Name some forms of biases from participants?
Non-compliance.
Almost invariably, non compilers are different from compliers in terms of their risk of adverse health outcomes. They tend to have a higher risk.
Follow up.
Same for follow up.
What are the different toes of randomised control trials?
Clinical trials : participants are patients or people with an apparent disease.
And
Field trials : participants are not patients but members of the general population.
Or
Efficacy: trials designed to estimate the maximum potential benefit to be derived from the intervention in ideal circumstances, including as close as a 100% compliance. Such ideal interventions are very rarely achieved in a real health programme. The benefits derived from a real life programme are known as the programmes effectiveness.
Or
Cluster randomised trials
In these trials groups of individuals are randomised, with all individuals within a group receiving the same intervention. This design is most appropriate when there is a major risk of contamination between the trial arms, for logistics reasons. However need to consider this when establishing sample size and carrying out Analysis.
Design effect: people in one cluster will probably be similar.
And
Community randomised trials when intervention target whole community.
Multi factorial trials
Two or more interventions are compared.
Cross over trials
Each participant acts as its own control by receiving at least two different treatments at different times during the trial.
Order through which they receive the treatment should be randomly allocation.
Need a wash out period between each intervention to avoid spill over effects.
Advantage: two groups are almost perfectly similar.
Non inferiority / equivalence trials
In some studies the aim is to test whether an intervention is at least as good as a standard treatment.
