Classical Epidemiology II Flashcards
What are randomised control trials?
Studies in which the efficacy of a health intervention (vaccine, drug, vector control, health education, etc.) is assessed in an experiment-like design.
There is an “intervention” and the investigator (or chance !) controls who receives the intervention
(the “intervention” or ”exposed” group) and who does not receive it (the “control” or “unexposed” group)
It is therefore comparative to a cohort study design.
What are some essential elements of a randomised control trial?
Contemporary controls: if a change in outcome (disease) is seen in response to intervention (exposure, drug etc), we want to be able to say the change was because of the intervention and not that the change was going to happen anyway. E.g. malaria incidence can be related/ influenced to rainfall.
The controls are examples of what would happen in the absence of the “intervention” and provide a baseline against which the intervention is assessed.
Controls CAN receive nothing, they can also receive the “standard” treatment.
Randomisation: chance decides which participants receive intervention and which become “controsl”, not those running the study.
Blindness (optional): The group the participant belongs to is kept secret from the participants, or participants and investigators. Avoids the scenario that either of them would act differently.
Blindness is not always possible.
What are the 3 types of blindness?
Simple-blind:the patient does not know what he receives (intervention or control)
Double-blind: Patient and investigator do not know
Triple-blind: even the analysis is carried out blind. The code is broken only at the very end.
In relation to randomised controlled trials, what is bias?
A decision or assessment in the study is considered “biased” if it is made unevenly between the study groups.
Both the participants and investigators can be biased.
E.g. selection bias: If the most unwell AIDs patients are assigned to the control group in a drug trial, the “exposed” group will likely be seen to do much better and it may be concluded the drug has higher efficacy than it really does. randomisation can prevent this.
How can participant selection control for confounding in a randomised controlled trial?
There are many factors that can lead to an “disease” outcome. In a RCT, we are interested only in the effect of the intervention (vaccine, drug, service) on an outcome and not of the surplus confounding factors.
Between the exposed and control groups, we need an even spread of the confounding factors (e.g age, immune status, weight) for them to be comparable so that as much as possible, the only way they differ is the exposure/ factor of interest.
Investigating whether smoking causes lung cancer, if all the non-smokers are old and all the smokers young, it may be concluded smoking is not as detrimental as it really is. We cannot accurately evaluate the smoking-disease relationship.
