Randomized Clinical Trials Flashcards
Understand characteristics of randomized controlled trials (RCTs): randomization, control, and blinding/masking
the independent variable is randomly assigned (can be done by simple, block, stratified, or cluster randomization), the control is the group without treatment or independent variable altered, blinding/masking is that the administrators don’t know who’s in what group
Differentiate among objectives of different RCTs: superiority, equivalence and non- inferiority
equivalence is that drug a is similar to drug b
non-inferiority hopes to prove drug a is not worse (ie equivlalent or better) than drug b
superiority hopes to prove superiority of a treatment to another form of treatment
Identify effectiveness and efficacy trials, and determine when each is appropriate
efficacy is if drug works under ideal conditions, effectiveness is under real world conditions
Distinguish between intent-to-treat and per-protocol analyses, and determine when each is appropriate
intent-to-treat is based on effectiveness and prescribing drug, all patients are analyzed in the data whether or not they completed treatment
per-protocol is if patient takes drug and is based on efficacy, only analyzes patients who complied with treatment
Describe alternative RCT designs: parallel vs. crossover, simple randomization vs. cluster or stratified randomization
parallel is when subjects are assigned to only one treatment whereas in crossover patients do all the treatments in time periods but are randomly assigned what order
simple randomization is purely random assignment, doesn’t guarantee even numbers of participants in groups
stratified randomization divides all patients by categories (like gender and age) then randomly splits those categories evenly
cluster randomization randomly assigns treatments to a group of people (like when you can only treat an entire village by adding something to water supply, randomly chose which village gets it)
Interpret validity, precision and generalizability of RCTs
validity: internal (were the study results true) or general (can study results accurately predict the whole population’s response)
precision is if the same results will be consistently obtained
generalizability is the relation of internal validity to external validity
Calculate and interpret the “Number needed to treat”
number needed to treat is number of patients needed to treat in order to prevent one bad outcome, ideally is one
NNT=1/(absolute risk reduction)=1/(proportion of placebo same or worse minus proportion of treatment same or worse)
number needed to harm is vice versa
Apply each of these concepts to the critical appraisal of a randomized controlled trial
