Epidemiology-Randomized Control Flashcards
In a randomized control trial, what is an advantage and disadvantage of crossover?
Advantage: you can treat each patient as their own control and eliminate confounders. Disadvantage: lasting effects from treatment
How does a randomized controlled trial help eliminate selection bias?
Randomization of who gets put into what group and treatment is assigned.
How does a double blind randomized controlled trial help eliminate measurement bias?
Blinding eliminates measurement bias so people making the observations don’t know who is getting the real drug and who is getting the placebo.
How does a randomized controlled trial help eliminate confounding?
The groups are randomized so unknown confounding factors are equal in each group.
Methods of randomization
Simple, block, stratified and cluster
What type of randomization was done in the study seen below?
Simple randomization. Note that the sample sizes are unequal because randomization does not alway make equal sized groups.
What type of randomization would you do if you want to satisfy a specific ratio of treatment:control in the study?
Block randomization. Example: 2 patients in each block of 8 people are assigned to a specific therapy.
What type of randomization would you do if you want to make certain characteristics (age, sex) equal between two groups?
Stratified randomization. You stratify your population then randomize it.
What type of randomization would you use if you can only administer your intervention to a group of people and not to an individual?
Cluster randomization. This prevents “contamination” of the treatment. (Example shown below of training staff members, you can’t train only one staff member without affecting the others, so you train the whole staff one way at one site and a different way at another site)
4 levels of blinding
Randomizer, patient, physician and outcome assessor
What is the usual comparison group in randomized trials?
Usual care and placebo vs. new treatment
Does the drug work under ideal conditions?
Efficacy trial, usually restricted to compliant patients only
Does the drug work under real-world conditions?
Effectiveness trial, usually analyzed by intention to treat.
If I prescribe the drug, will the patient get better?
Intent to treat analysis, prevents bias caused by loss to follow up
If the patient takes the drug, will he or she get better?
Per protocol analysis
Superiority null hypothesis
No difference between the new drug and placebo. Rejecting the null means that the new drug is better than the old drug.
Equivalence and non-inferiority null hypothesis
New drugs are different than the old drugs. Rejecting the null means that the new drug is pretty much the same as the old drug.
Primary outcomes analyzed in randomized controlled trials?
Time to discontinuation of treatment
Secondary outcomes analyzed in randomized controlled trials?
PANSS (+ and - syndrome scale, measures psychopathology) and CGI (clinical global impressions, measures severity of illness)
How do you compare two statistical proportions?
Chi square test. Relative risk (the ratio between the two).
Number of patients you need to treat in order to prevent an additional bad outcome
ARR = absolute risk reduction
Number needed to treat in order to hurt one more patient.
ARI = adverse risk
Internal validity
Was the study run correctly
External validity
Can the study be applied to my patient population
