RCT Flashcards
What is Equipoise
Principal investigator is generally uncertain which treatment gives a better outcome. This is called personal equipoise. But we also refer it as clinical equiposise where the clinical community is genuily uncertain which treatment gives out the better result.
What are the characteristics of an RCT
- gold standard for inferring causality - Only study desgin that can investigate whether an exposure causes an outcome.
-
equilizes the baseline characteristics in control and treatment group ; controls for measured and unmeasured confounder. This is achieved through random allocation of participants to treatment vs control arms
3.** Multiple outcomes can be assessed** - Exposure is assigned to either tretment or control
Simple two arm RCT
two arms one gets intervention other does not
Factorial
testing two or more independent interventions. two interventions/treatments should not interact with each other
cross over RCT
similar to simple 2 arm RCT but after sometime the treatment and control arm are swapped. Ensure that swapping is done after some time, there should be no residual effect.
Randomization
- random allocation of participants to control and treatment arm
- helps to ** equalize characterstics at baseline** both measured and unmeasured confounders.
- Group assignment must not depend on patient characteristics or assignment of previous patient
simple randomization
it may not equalize groups
might make groups unequal
subect is randomly allocated to either treatment or control
stratified randomization
creating of sub-groups based on sample characteristics and dividing the sample into these sub-groups. and then randomize them
cluster randomization
naturally occuring groups randomized to treatment or control; risk of contamination ; cluster randomizing schools but the principals know each other and one is in control and treatment and they are thinking why is one in treatment and other in control
allocation concelment
adequate measures to conceal allocation to study group assignments from those responsible for assessing patients for entry in the trial were taken (e.g., central randomization; numbered, opaque, sealed envelopes; sealed envelopes; coded bottles or containers; drugs prepared by the pharmacy).
Protects the randomization process before the subject enters the trial
Selection bias
Blinding
Relevant groups who may/may not have knowledge of treatment assignments:
* Participants
* Investigators/clinicians administering intervention
* Investigators assessing outcomes
* Data analyst(s)
Masking of the treatments after randomization(once trial begins)
Information bias if not followed
per-protocol analysis oversestimation of effect why?
only those who were super compliant will remain in the study hence overestimation
