Please enable JavaScript.
Coggle requires JavaScript to display documents.
L7 Therapeutic trials / Randomized trials :rabbit: (Randomized controlledâĶ
L7
Therapeutic trials
/ Randomized trials :rabbit:
Hierarchy of Evidence & Research designs
Randomized controlled trials (RCT)
Def: Planned experiment designed to assess the
efficacy of an intervention
in human beings by comparing the intervention to a control condition
Allocation to intervention or control is determined
purely by chance (randomization)
:cherries:
Random numbers
Random computerized allocation
:!:
canât randomize on any Systematic characteristics
eg, digits of SS#, attendance at a clinic, etc.
Are a subset of possible experimental designs
Q.
Why randomize
???
Avoid selection bias
â āđāļĄāđāđāļŦāđāļāļ§āļēāļĄāļāļąāļāļāļāļāđāļĢāļēāļāļāļ°
Improves comparability
of Intervention & Control arm populations
Fulfills assumptions underlying tests for Statistical inference
2 āļāļēāļāđāļāļāļĨāļāđāļāđāļēāļāļąāļ To maintain balance between arms one can ->
Individually match
(eg, by age, sex, etc)
Block randomization
(eg. Randomize within blocks of 4)
Q. What does âcontrolledâ trials imply??
âControlledâ implies predefined:
: Eligibility criteria, Specified hypotheses, Primary&Secondary endpoints to address hypotheses, Methods for enrollment&F/U, Rigorous monitoring, Analysis plans & stopping rules
Q.
Why do RCTs
?
Observational & Quasi-experimental designs are subject to
potential bias & Confounding
due to:
Self-selection (lack of comparability)
Observer bias
Secular trends (eg before & after study)
RCT provides
gold standard for proof of concept
:money_with_wings:
Q. Are results of RCTs
always valid
??
If badly design -> could provide Conflicting results (āļāļąāļāđāļĒāđāļāđāļāļāļąāļ§āđāļāļ)
Eg. āļāļāļāđāļāļĒāļāļēāļ/āđāļĄāđāļāļēāļāļĄāļļāđāļ āđāļāđāļĄāļąāļāļāļēāļāđāļĄāđāđāļāđāļĨāļāļĢāļēāļĒāļĨāļ°āđāļāļĩāļĒāļ āļāļēāļĢāļĄāļāđāļāļ§āļāļāļĩāđāļāļāļāđāļĄāđāļāļēāļāļĄāļļāđāļāļāļēāļāļĄāļĩāļāļīāļāļĄāļļāđāļāļĨāļ§āļāļāļĒāļđāđāđāļĨāđāļ§āđāļĨāļĒāđāļĄāđāļāļēāļāļĄāļļāđāļ
RCT design, execution, or analyses can be flawed
Intervention vs Control comparisons are
internally valid
but,
Restrictions
on
participants eligibility
can :arrow_down: external validity
(eg. Specific age or Sex groups omitted)
:!:
Avoid
doing RCT in
Underage (āļĒāļąāļāđāļĄāđāļāļĢāļĢāļĨāļļāļāļīāļāļīāļ āļēāļ§āļ° â 18 yrs old)
Pregnant woman
āļāļļāļāļāļĨāļāļąāļāļāļĢāļēāļĒ, āļāļīāļāļāļāļāļīāļāļēāļāļāļīāļ eg āļāļīāļāđāļāļāļąāđāļāđāļāļ·āļāļ
Phases of Trials
Trial designs
Most trials got 2 arms (Intervention vs Control)
Multiple interventions can also be compared to a Single Control arm
Equivalency
trials
Head-to-head comparison of two or more Rx, without a control group
(eg, Contraceptive trials)
Factorial designs
Eg. Intervention A, Intervention B, InterventionA+B vs Control
Q. When should we do a trial?
Q. When is it
unethical to randomize
??
:deer:
Known
effective Rx
So
cant use placebo
, Need to provide standard care as its alrdy proved whatâs its efffective drugs
Eg. Trials to prevent mother-to-child HIV transmission
:deer:
Personal choice
Canât randomize very different interventions (āļāļēāļĢāļĄāļāđāđāļĄāđ blind)
Eg. Trials of different types of contraceptives â
pill vs IUD
â āđāļŦāđāļāļāļąāļāđāļĨāļĒāļ§āđāļēāļāđāļēāļāļ§āļīāļāļĩāļāļąāļ this is ethically questionable as Women hv right to select a method of their choice
:deer:
Risk of New Rx
likely to exceed risks of existing Rx
Sample size estimates
Most endpoints r measured as events in person time
Need to estimate person years (py) needed to detect a specified date in intervention vs control
Estimate sample size at enrollment & assumes loss to F/U to provide the person years needed
Control groups
If thereâs
no standard of care
-> may receive no Rx (eg placebo)
If thereâs
established standard of care
-> Unethical to withhold this from controls! so
standard care becomes reference control
āļĒāļīāđāļ sample size āļāđāļāļĒ -> āđāļāļāļēāļŠāđāļāļĩāđāļĒāļāđāļāļāļŠāļđāļ (āđāļāļāļāđāļēāļāļŠāļđāļ)
Eligibility & Enrollment