Association, Causal Inference & Causality Flashcards
What is a cause?
What is an association?
- precursor event/condition/characteristic required for dz or outcome
- relationship b/w exposure/tx & an outcome/dz
What are the 3 types of associations?
- Artifactual (false) Assoc.
- Non-causal Assoc.
- Causal Assoc.
What is an artifactual association?
Not a true assoc.
Arise from biases and confounding
What is a non-causal association?
What 2 ways does it occur?
No direct link b/w exposure and outcome.
Occurs 2 ways
- the dz may cause the exposure (RA causes physical inactivity)
- dz and exposure are both assoc. with a 3rd factor (confounding)
Sufficient cause?
Conditions/events that inevitably cause dz
A cause that precedes a dz & if present the dz will ALWAYS occur
Ex) genetic dz/abnormalities
Necessary Cause?
Cause must be present for dz to occur, but the cause may also be present without the dz occurring.
If present - no guarantee you’ll get the dz, but if you get dx with the dz that factor must be present
Necessary but not sufficient for dz
Ex) Mycobacterium TB
Component cause (risk factor, RF)?
A element/RF, that if present, increased the likelihood/probability, risk or a particular dz
Ex) smoking is RF for lung Ca
Synergism
2 or more component-causes (RFs) both must be present at the same time for increased likelihood of outcome
Ex) gene- & environmental - factors BOTH have to be present at same time for infants to get congenital disorder
** additive — increase MAG of risk
Parallelism
Work in parallel - either/or
2 or more factors, but don’t have to be working together at same time
Ex) infant only get dz if exposed to EITHER gene- or environmental- factors, but wouldn’t get it if exposed to neither
Multiple component-causes (multi-factorial)
Risk factors working in concert to collectively become causes
Process of causal inference determination
How can we pass from this observed association to a verdict of causation?
(An interpretive application process)
By using Hills criteria
What are the 5 components of Hill’s criteria
1) Strength
2) Consistency
3) Temporality
4) Biological Gradient
5) Plausibility
A higher # of Hill’s criteria met, when evaluation association, the more likely it may be causal
Strength
The bigger the size of an assoc (RR/OR/HR) the more powerful
But “a strong assoc is neither necessary or sufficient for causality & weakness of assoc. is neither nec or sufficient for absence of causality”
Consistency
Repeated observation of an association in difference poplns is more powerful
Ex) large # of studies have consistently demonstrated an increased risk in CHD
But may still obstruct the truth
Temporality?
- Proximate cause?
- Distant cause?
Reflects that the cause presences the effect/outcome in time
- Proximate cause: short-term interval
- Distant cause: long-term interval
