experimental design Flashcards
issues in causal research
- bias
- random error
- confounding
- effect modification
bias
differential error: usually produces deviation/distortion in one direction
random error
less likely to distort/ influence the direction of the relationship. it decreases the probability of finding a real association by reducing the statistical power
confounding
a measure of the effect of exposure on risk is distorted bc of the association of exposure with other factors that influence the outcome
effect modification
differences in the effects of one or more factors according to the level of remaining factors
quantitative designs
- observational: no intervention or experiment
- experimental: manipulation of study factors
observational designs
- cross-sectional
- longitudinal
cross-sectional
one set of observations is collected from every unit in the study, at a certain point in time
longitudinal
2 or more sets of observations are collected for every unit in the study
cohort studies
tracks people forwatd in time from exposure to outcome
effect measure for cohort studies
risk ration
case-control studies
2 source populations: track studies backwards from outcome to exposure
when is it best to use RR
- observational
- prospective
- longitudinal
when is it best to use OR
- observational
- longitudinal
- retrospective
experimental study
a trial that is experiencing the effectiveness and the risks of any medical treatment
advantages of randomized clinical trial
- best evidence study design
- no inclusion bias
- controlling for possible confounders
- comparable groups
disadvantages of randomized clinical trial
- large trials
- long-term follow up
- compliance
- expensive
preclinical trial
models in vitro or in vivo
phase 0
human studies with minimal doses and are designed to accelerate the development of promising drugs
phase I:
first stage of testing in human subjects; how much drug can be administered without causing serious adverse effects: healthy volunteers
-dose escalation
phase II:
the drug is administered to a greater number of individuals, suffering from diseases: establishes the minimum effective dose in humans and the optimal dose regimen and any side effects
phase III:
the drug is administered to a large number of subjects: randomized controlled multicenter studies
Phase IV:
studies are conducted after the marketing of the product under investigation
noncompliance
patients may agree to be randomised but following randomisation do not adhere with the assigned treatment
