POPH192 - Lecture 21 Flashcards
What type of study is randomised control trial?
analytic and interventional
what does RCT measure?
similar to cohort, but instead of measuring exposure it randomises an intervention
first step of a RCT?
1) identify source population
second step of a RCT?
2) randomly select sample population who don’t have outcome of interest
third step of a RCT?
3) randomise the sample to either the intervention or control groups
fourth step of a RCT?
4) follow up overtime and see who develops the outcome
fifth step of a RCT?
5) calculate measures of association and occurence
What is randomisation?
- aka random allocation
- when participants are randomly allocated into either the control group or intervention group
(not the same as random selection!!)
what is the purpose of randomisation?
- to control for confounding
(this is when another variable like age or sex distorts the relationship between exposure and outcome)
what happens in successful randomisation?
- if done correctly, there should be the same proportion of known and unknown confounders in each group (meaning that groups are comparable)
how can you tell randomisation has been successful?
by looking at the baseline characteristics (age, weight, sex etc) of both groups.
- if successful, they should be similar (don’t have to be identical)
what are the 2 other variants of randomisation?
- cluster
- stratified (block)
what is cluster randomisation?
subgroups are randomised instead of individuals
what is stratified (block) randomisation?
participants are randomised within blocks
what is the first way of protecting randomisation and it’s benefits?
concealment of allocation
what is concealment of allocation?
the person randomising the participants doesn’t know what the next treatment allocation will be
- it is concealed and unpredictable
how can you achieve concealment of allocation?
by a computer generated randomisation code
what does concealment of allocation prevent?
selection bias
- from the other participants or their doctors from selecting the treatment they want
what is the second way of protecting randomisation and it’s benefits?
intention to treat analysis
what is the intention to treat analysis
this is when we analyse as we randomise
- it means you aren’t swapping people between groups and are maintaining the randomisation
what effect does the intention to treat analysis give?
it gives a ‘real world’ effect
- people don’t take intervention perfectly in real life
what is the other type of analysis for RCT?
‘per protocol’ analysis
what is ‘per protcol’ analysis?
- when you analyse participants who fully completed the study protocol
what does the per protocol analysis show/lose?
can show the efficacy of the treatment, but lose the benefits of randomisation
what are the 3 types of bias that can occur in RCT?
- lack of blinding
- loss to follow up
- non-adherance
what can lack of blinding do?
if people involved in the study know whether they were in the intervention or control group, it may influence them
how can studies be blinded? what is the best way to describe this?
studies can be single blinded (participant) or double blinded (participant and researcher), but it is best to be specific on WHO is blinded in the study
are all methods for blinding easy?
- certain methods can be used but sometimes it’s difficult
e. g. if things like physiotherapy or surgery are the intervention
what are the effects of loss to follow up?
if people leave the study, this can cause confounding and bias
- the ones who left could be different to the one’s who stayed
what are the effects on non-adherance?
- participants are sometimes not good at following instructions
- if there is too much non-adherance, it can lead to bias
what are the strengths of RCT?
- gold standard study design to test an intervention and cause and effect associations
- eliminate confounding and bias (if done well)
- calculate incidence, and measures of association
what are the limitations of RCT?
- many exposures cannot be randomised
- need to have clinical equipoise
- expensive
- sometimes participants aren’t representative of general population, so are not generalisable
- not good for rare outcomes
what is clinical equipoise?
genuine uncertainty about benefit/harm of the intervention