Lecture 3 Flashcards
What are the 6 Bradford hill criteria ?
Consistency
Biological plausibility
Dose-response
Temoprality
Strength of the relationship
Reversibility
Consistency
if other studies in diff pops, at diff times, using diff methodologies find the same association it is more probable that the relationship is causal
Biological Plausibility
does the association make biological sense? Does it with with the natural history of the disease?
Dose-response
if increased levels of exposure lead to greater frequencies of the outcome it is more probable that the relationship is causal. Does not have to be cont there can be a threshold
Temporality
if an exposure causes an outcome it must precede the outcome. This is the most important
What is the most important bradford hill criteria?
temporality
Reversibility
if you take away the exposure do you see a reduction in the outcome. THis is not always possible to assess at the indv level
If an individual with lung cancer quits smoking their lung cancer might not go away but if you studied cancer incidence over the last 50 years as smoking rates have declined you would find that lung cancer incidence is also decling
What must you do to determine a relationship is causal
bradford hill criteria
Formulating a research question
SMART questions
specific - nuts and bolts
measurable
answerable- do you have the resources available
relevant-scientific community and to you
time- fit the frame in which you have to answer it
Elements to consider for specific questions
The PI(E)COD method
PI(E)COD
P: pop of interest
I: intervention? (if appropriate)
E: exposure of interest (if app)
C: control or comparison group
O: outcome?
D: Study design?
Ecological variables are properties of ____, _____, or_____
groups, organizations, places
What are the 3 types of measures in ecological studies?
Aggregate
Environmental
Global
Aggregate measures
summaries derived from individuals ina group (proportion of children in a household, median income)
Environmental measures
physical characteristics of a place (pollution level, hours of sunlight)
Global measures
Group level attributes with no distinct individual level analog (pop density, the availability of the healthcare system, the existence of traffic laws) Everyone in the group is subject to the same global measures
Ecological fallacy
associations observed at the aggregate level do not reflect associations at the individual level
Oilers and Flames example of the ecological fallacy
if the edmonton oilers were performing worse than the calgary flames across the NHL season it would be an ecological fallacy that all edmonton oiler players are worse players than all calgary flames players
Atomistic fallacy
The opposite of the ecological fallacy where one assumes that associations found at the individual level also hold true at the group level
3 pros of ecological studies
low cost
sometimes variables are impractical to measure at the individual level
sometimes the effect of interest is at the ecological scale (policy level effects)
4 cons ecological studies
the temptation to extrapolate group effects to the individual (ecological fallacy)
issues in combining data sources
Temporal ambiguity. exposure or outcome first
Migration across groups. If individuals migrate from one group to another over the course of the study
Is there as statistical test for causation?
no
Causal associations
A change in frequency of the exposure causes a change in frequency from the outcome
A necessary cause is ____ for an outcome to occure
required
A sufficient cause
outcome may cause exposure but may not be the only cause