Clinical Trials Flashcards
experimental studies
investigator intervenes with the course of nature and assigns exposure (treatment) to subjects
what is the goal of clinical trials
quantify the effect of an intervention
GOLD STANDARD for establishing causality because effect is relative to a comparison group
control group
a comparison group to determine whether exposure to a factor caused the outcome
provides an estimate for disease occurrence had the factor under study not been present
what is the goal of the control group
limit the differences between the treatment and control groups so that the only interfering cause of differences would be the assigned treatment
inclusion and exclusion criteria
criteria by which subjects must meet in order to be enrolled
limits variability between groups but only PARTIALLY satisfies comparability
randomization
avoiding differentiation assignment of treatment by randomly assigning
goal: equally distribute confounders between groups and reduce investigator bias
does randomization always guarantee comparability between groups
NO - small sample sizes have a higher chance of confounders despite randomization
what qualifies as a confounder
something that is causal to BOTH the exposure and the outcome
ex:
- disease: lung cancer
- exposure: smoking
- confounder: heart disease
subjects with heart disease should not be enrolled in a study of the effect of smoking on lung cancer because smoking also causes heart disease, so it is more likely that the heart disease subject has been a smoker
confounding
mixing of effects; attributing an effect to one factor when in reality it is due to another
what can help decreased confounding
large sample sizes - distributes confounders equally among groups
masking/blinding
blinding the trial staff from knowing treatment status of subjects to ensure valid comparison
single blind study
either the study subject or the investigator is unaware of treatment group
double blind study
both the study subjects and the investigator are unaware of group allocation
triple blind study
the study subject, investigator, and scientist analyzing the data are unaware of group allocation
blocking
randomly chosen blocks used to assign treatments
ex. if block size is 4, then for every 4 subjects enrolled, 2 will be group A and 2 will be group B but by variable permutations
blocking by time
the variable in enrollment is based on the TIME the patient was enrolled
ex. every week there will be 4 subjects enrolled - 2A and 2B so that if unexpected factors cause the study to end early, there will be equal numbers of As and Bs
stratification
dividing the trial into many smaller trials in order to block on different risk factors (ex. breed, sex)
increases study efficiency by narrowing the confidence intervals
occurs BEFORE randomization
multicenter trials
trials conducted in more than 1 location with the same protocol followed in each location
information gets processed by a single center
requires a very large sample size
advantages of clinical trials
- best way to produce evidence for causal effect
- may be the only practical way to study certain clinical problems
- can sometimes be faster than observational studies
disadvantages of clinical trials
- can be costly and time consuming
- not all projects can be experimentally studied
- may be ethical problems in doing the experiment
- outcome may be too rare to study
- narrower scope than observational studies
