Intro to Epidemiology

design that tests if a new intervention works

experimental study

design that tests if an exposure and outcome are associated with each other

analytical observational studies

design that tests the pattern of disease in the population

descriptive observational studies

list the hierarchy of evidence from bottom up

editorials/expert opinion → case reports → case series → cross sectional studies → case control studies → cohort studies → non-randomised controlled trials → systematic reviews/meta analyses

4 reasons for a study result

• true association

• chance

• bias

• confounding factor

define bias

systematic error in design or conduct of study giving a result that is different from the truth

2 sources of bias

• selection bias

• information bias

4 types of selection bias

• self-selection bias

• non-response bias

• healthy worker effect

• survival bias

types of information bias

• recall bias

• interviewer bias

• social desirability bias

• attrition (loss to follow up) bias

define a confounder

a factor associated with both the exposure and outcome that influences the apparent association

3 bradford hill criteria for causal inference - association features

• strength of association (large effect size)

• specificity (outcome has one cause)

• consistence (replicable)

3 bradford hill criteria for causal inference - exposure/outcome features

• temporal relationship (exposure before outcome)

• dose-response relationship

• reversibility (removing exposure reduces outcome)

3 bradford hill criteria for causal inference - other features

• biological plausibility

• analogy (similar known associations)

• coherence