Epidemiologic study design Flashcards
What are the 3 descriptive studies?
Case reports
Case series
Cross-sectional
What are the 6 Analytic studies?
Ecological Cross sectional Case control Cohort Experimental Meta-analysis
What are the two main kinds of analytic studies?
Experimental
Observational
What is the major difference b/w experimental vs observational analytic studies?
experimental tx or exposure is allocated/asigned to subjects
observational subjects select their own exposure
Validity of experimental studies rests on two important criteria, what are they?
- validity of comparison
2. validity of measurement
What error is it called when there is a lack of comparability b/w study groups?
confounder
What error is it called when collected information is not accurate?
Information bias
How do we best approach a valid comparison during an experimental study?
Randomization is the best way to achieve comparability b/w tx groups
What is random sampling? What is randomization?
Random sampling – randomly choosing a sample
Randomization – randomly allocating groups from the sample to tx or other aspect of the study
What is the key feature of randomization?
One individual is no more or less likely to receive a particular tx than any other individual in the study population
T/F
Randomization is a guarantee for making groups equal
False
It is not guaranteed
What is blinding?
Principle where whoever measures the outcomes in the study subjects be unaware of what the tx assignment was - this way they cannot be influenced (biased) by knowing which group the study subjects are in
What are three advantages of experiments?
- Best way to produce evidence for causal effect
- May be the only practical way to study certain clinical problems
- Can sometimes lead to faster results than observational studies
What are 4 disadvantages to experiments?
- Costly in time and money
- may be ethical problems
- may be lack of variation in participants
- non-compliances and dropouts
What are the 4 main types of observational studies?
What is unique about these observational studies compared to experimental?
- Ecological
- Case-control
- Cohort
- Cross-sectional
Researcher doesn’t control circumstances leading to exposure
When is an ecological study done?
When information is known on entire groups, not on individuals
What are cross sectional studies?
Comparison b/w groups
No follow up!
Study dz and exposure status at one point in time
Cannot measure incidence (risk, rates) of dz – only prevalence
Cannot always know if exposure preceded dz or followed it
What does it mean if a sample is longitudinal?
They are being followed up
What are cohort studies?
Longitudinal study
Group of individuals at risk are followed up
Some of the cohort members are exposed, some are not
Can measure either cumulative incidence (risk) or incidence rates
What is the definition of Cohort?
An assemblage of individuals who are at risk of the dz (outcome) and who share certain characteristics in common
Cohort studies can be ______ or _____________
Prospective or Retrospective
What is the difference b/w prospective and retrospective?
Prospective: assemble cohort at time study begins and follow individuals forward and measure dz occurrence and calculate risks/rates
Retrospective: obtain historic exposure and outcome information on a cohort that was assembled sometime in the past
When are cohort design studies useful?
Disease is common, exposure is rare, or you wish to study multiple outcomes
What are 2 major cohort design issues?
- Expensive
2. Reason an individual leaves a cohort is related at all to the outcome = bias may result
What is this phenomenon called… Reason an individual leaves a cohort is related at all to the outcome = bias may result?
Censoring or lost to follow up
T/F
It is a great idea to do a cohort study on a rare dz
False
What is another kind of cohort study?
Case control
How is a case control study different than a cohort?
Sample them based on their disease status, not their similar exposures
The cases are taken from disease registries, hospital records, vet records, etc.
When do you use a case control study over a cohort study?
When the disease is rare
What does the control in a case control study look like?
Non-diseased individual
What can’t we calculate from a case control study? Why?
risk
risk difference
ratios
The choice of the number of controls is arbitrary
What do investigators calculate from a case control study?
Exposure odds ratio
How do you calculate an exposure odds ratio?
Probability of exposure in the cases / probability of exposure in non-cases
(Cases exposed/cases unxposed) / (non-cases exposed/ non-cases unexposed)
Can exposure odds ratio be used for causal inferences? If no, what can?
No
Disease odds ratio
How do you calculate disease odds ratio?
(cases exposed/non-cases exposed) / (cases unexposed/noncases unexposed)
What are the three main rules for case-control studies?
- control should only be selected if someday, if it were to become a case, it can be eligible to become a case we are studying
- Controls should be selected for reasons that are unrelated to the exposure studied
- Select incident cases, not prevalent ones
What are 5 issues with a case-control study?
- Relatively inexpensive compared to experimental and cohort studies
- If it is performed retrospectively, the duration may be short if we can easily retrieve info
- Ideally suited to examine a large number of population risk factors
- Often no way to verify accuracy (most information is based on what is in records)
- cases may be more likely to recall past exposures than controls (noncases). May result in recall bias (type of information bias)
How do experimental studies measure occurrence?
risk
incidence rate
incidence odds
How do experimental studies measure association?
risk difference/ratio
rate difference/ratio
odds ratio
How do cross sectional studies measure occurence?
prevalence proportion / odds
How do cross sectional studies measure associatoin?
prevelance difference / ratio/ odds ratio
How do cohort studies measure occurence?
risk
incidence rate
incidence odds
same as experimental study
How do cohort studies measure association?
risk difference / ratio
rate difference / ratio
odds ratio
same as experimental study
How do case control studies measure occurence?
exposure odds
How do case control studies measure association?
expoure odds ratio
