Lecture 29: Association and causation Flashcards
If we find an association, does that mean
the exposure causes the outcome?
No it can be very complex and need multiple exposures to develop into something. Other factors at play etc…
What is a cause?
“an event, condition or characteristic [or a combination of these factors] that plays an essential role in producing an occurrence of the disease”
What is the causal pie model? What is component cause?
A range of exposures represented in a pie
Together, these exposures are sufficient to cause the outcome - So we call the whole pie a sufficient cause of the outcome
Each exposure is a component of the sufficient cause - So we call each of the exposures a component cause
Understand how to calculate causal pies and eliminating what exposure would prevent the most disease (SLIDES 19 - 28)
IMPORTANTTSTST
What is a necessary cause?
A component cause which is necessary for the disease to
occur.
It must be part of every sufficient cause.
Exposure A is a necessary cause in this example (LOOK AT SLIDE 28)
Is causation and exposures simple? When considering an outcome?
Complexity
* Reality is often complex
* Different combinations of exposures may lead to the same outcome
* The cause of a disease may not just be the last thing that
preceded it
“The presence of a valid statistical association in no way implies causality.”
Judgement is required
What are the guidelines to determine if something is causal?
“…‘aids to thought’ ….not absolute requirements to be met…”
Consider each of the guidelines and then make a judgement based on the totality of evidence
BEST CD’S
1. Biological plausibility
2. Experimental evidence
3. Specificity
4. Temporal sequencing
5. Consistency
6. Dose-response relationship
7. Strength of association
Consider them all and then make a judgement based on the totality of evidence
“Is there any other way of explaining the set of facts before us, is there any other answer equally, or more likely than cause and effect?
What is Biological plausibility of BEST CD’s?
Is there a plausible mechanism for the association?
What is Experimental evidence of BEST CD’s?
ONLY IN RCT’s so will be NO for critical appraisal (probably)
Is there evidence from human RCTs or animal experiments?
However, animal studies may not apply to humans
What is Specificity of BEST CD’s?
Is the exposure specifically associated with a particular outcome but not others?
However, it is quite common for exposures to be related to many outcomes (and vice versa)
Example:1
Inhaled corticosteroid use is not the only exposure associated with hospitalisation for diverticular disease, so the association is not specific
What is temporal sequencing of BEST CD’s?
Exposure or outcome first?
Example:1
Yes there is evidence of the temporal sequence between the exposure and the outcome as the exposure (inhaled corticosteroid use) comes before the outcome (hospitalisation for diverticular disease)
What is Consistency of BEST CD’s?
Are the findings consistent with findings from other studies?
However, there can be a number of reasons why
studies might have different findings
Example:1
There was no consistency demonstrated with other
studies as only studies about oral corticosteroid use
were specifically discussed whereas we are interested in inhaled corticosteroids.
What is Dose-response relationship of BEST CD’s?
Does the risk of the outcome change with increasing or
decreasing amounts of the exposure?
However not all relationships are linear
Example:1
There is a dose-response relationship shown between inhaled corticosteroid use and hospitalisation for diverticular disease. Increasing duration of use of inhaled corticosteroids was associated with increased risk of being hospitalised for diverticular disease.
What is Strength of association of BEST CD’S?
The stronger the association, the less likely it is to be due to confounding or bias
However, this is not always the case e.g. if confounding/bias is really strong
The different levels are:
RR or OR > 2 = moderately strong
RR or OR > 5 = strong
RR or OR < 2 = not very strong
Example:1
The multivariable RR for inhaled corticosteroid use and hospitalisation for diverticular disease is 1.71 (1.36, 2.14). This would not be considered a strong association.
