Lecture 21 - Randomised Control trial

Question 1

Randomised Control trial

Answer 1

Analytic, Intervention study

Question 2

What is randomisation

Answer 2

It is random allocation
- Equal chance for each participant to be in either group
- If enough people randomised, should have same proportions of confounder in each group
- This applies to know and unknown confounders
- Successful randomisation means confounding is an unlikely reason

Question 3

How do we tell if randomisation worked?

Answer 3

• Baseline characteristic: age, sex. these should be similar
• Variants:
  1. cluster randomisation: where subgroups are randomised instead of individuals
  2. Stratified randomisation: where participants are randomised within blocks e.g. 20 people each group

Question 4

Protecting randomisation, key strength of RCTs is randomisation

Answer 4

• Large number
• Concealment of allocation
• Intention-to-treat analysis

Question 5

Large number

Answer 5

Randomisation more likely to lead to balanced group with large number of participants

Question 6

Concealment of allocation

Answer 6

• Allocation concealment is different concept is blinding
• It means that the person randomising the participants does not know what the next treatment allocation will be so they don’t interfere in the study

Question 7

Intention to treat analysis

Answer 7

• This is when we analyse as we randomise: means we aren’t swapping people between groups (maintaining randomisation)
• other type of analysis, ‘per protocol analysis’- we analyse participants who fully complied to the study protocol, shows efficacy of the treatment, but we lose the benefits of randomisation

Question 8

RCT

Answer 8

similar to cohort study, but increase of measuring exposure, we randomise an intervention

Question 9

RCT step by step

Answer 9

1. Identify a source population
2. Randomly select the sample population who don’t have the outcome of interest
3. Randomise the sample to either intervention or control groups
4. Follow up over time and see who develops outcome
5. Calculate your measures of association and occurrence

Question 10

Measure of occurrence and association

Answer 10

• Incidence
• RR and RD

Question 11

Bias in RCT

Answer 11

Three types of bias
1. Lack of binding: if people involved in the study know whether they were in the intervention or control group, this may influence them.
2. Loss of follow up: people leaving the study which can lead to confounding and bias
3. Non-adherence: Participants are sometimes not good at following instruction which can also lead bias.

Question 12

Strength of RCT

Answer 12

• Standard study design to test an intervention and cause effect association
• Eliminate confounding and bia
• Calculate incidence and MOA

Question 13

Limitation of RCT

Answer 13

• Many exposures cannot be randomised , and need to have clinical equipoise
• Expensive
• Sometimes participants aren’t representative of general population, therefore are not generalised
• Not good for rare outcomes