Introduction to Epidemiological Study Designs Flashcards
What is the only example of a ‘like for like’ sample where we can compare two exact groups?
Identical twin studies.
What is a descriptive study design?
Ecological studies or cross-sectional surveys. Where you study 1 population.
What is an analytical study design?
Case-control studies and cohort studies. Where you study 2 populations and compare them.
What is an ecological study?
A snapshot in time of a group. An exposure is studied like smoking and an outcome like heart disease.
What is categorical data?
Data which is either nominal (numbers or names) or ordinal (in an order).
What is continuous data?
Data which is either an interval (doesn’t have an absolute zero like years) or ratio scale (has an absolute zero like age).
What are some issues with ecological studies?
Definition of characteristics. Measurement variation. Confounding Falsely inferring individual-level association from group-level. Random error.
What is a cross-sectional survey?
Can be a series of surveys on individual level. (Not just a snap-shot).
What are some issues for cross-sectional surveys?
In addition to ecological studies there can be:
A sampling bias so random sampling is preferred.
Responder or participant bias.
Random error.
What is case-control study?
A study that compares two populations. One group vs. a control group. Diseased vs non-diseased. It is always retrospective as a snapshot in time.
What are some issues for case-control studies?
Selection bias like hospital studies. Hospital studies do not reflect on general population.
Controls should reflect study population and controls should be comparable to cases.
Information bias
-Differential misclassification
-Non-differential misclassification
Confounding
Random error.
What kind of ratio is a case-control study? Why?
It is an odds ratio or a prevalence odds ratio.
This is because it doesn’t consider any time so we don’t know how long the groups have been exposed to something.
What is a cohort study?
A concurrent or prospective study which can also be retrospective. It ages with the researcher.
You find an exposed group and an unexposed group and compare them.
For example after 10 years you observe which individuals have fallen ill in a condition, both from exposed and unexposed groups and compare them.
What are some issues with cohort studies?
Like with case-control but also: A loss to follow-up where there is a differential loss and survivor bias. Information bias -differential misclassification -non-diffrential misclassification Confounding Random error
What kind of ratios can be used in a cohort study? Why?
Can use both odds ratio (prevalence) and rate ratio (Incidence).
Because we can either take a snapshot in time or let time be a factor for a rate ratio.