13 - Causation Flashcards
1
Q
What is the basic definition of cause?
A
- Any factor that produces a change in the severity or frequency of the outcome
- *do NOT need to understand ALL causal factors to prevent or at least control disease
2
Q
What is inductive reasoning?
A
- Process of making generalized inferences about causation based on repeated observations
3
Q
Inductivism and logical fallacies: example with the rooster
A
- Rooster crows just before sun rise
- Therefore, roosters crowing causes the sun to rise
4
Q
Koch’s postulates: limitations
A
- IGNORES environmental factors
- NOT applicable to non-infectious diseases
5
Q
Epidemiology vs. the lab
A
- Can’t always recreate disease in lab
- If wanting to understand complex issues affecting disease in a natural world then need to study the NATURAL WORLD
- *need both natural world study and lab studies
- *most causation discussion are LIMITED to observational research rather than experimental
6
Q
Observational vs. experimental research
A
- Observational: looking for cause
- Experimental: looking for effects
7
Q
Experimental studies
A
- We RANDOMIZE individuals to receive a factor and some to receive nothing
- We know the factor precedes disease and other variables accounted for by randomization
- We contrast outcomes in treatment and control
- Assume EXCHANGEABILITY
8
Q
Observational studies
A
- Estimate outcome differences between individuals that happen to vary in their exposure status
- Matching and restriction where appropriate to minimize differences between groups
- *measure ASSOCIATION between changes in exposure and outcome
9
Q
What are the limits to experimental studies?
A
- Difficult to duplicate realistic dose, exposure pathway or complete set of typical cofactors
- Difficult to carry out experiments that actually resemble “real-world” conditions
10
Q
Observational comparisons: what are you comparing it to?
A
- Ex. compare to current treatment (can’t just have totally untreated animals)
11
Q
Cohort studies: 2 steps
A
- Define groups (cohorts) of animals according to exposure of animals in groups to factors of interest
- Follow groups FORWARD IN TIME to see which animals develop the disease under investigation
12
Q
What do you compare with cohort studies?
A
- Risk in exposed and unexposed groups
- *reported as RELATIVE RISK
- Can look at more than one disease resulting form a specific type of exposure
- **CLOSEST OBSERVATIONAL STUDY WE CAN GET TO RCCT
13
Q
Case-control studies: 2 steps
A
- Define groups of diseased and healthy animals
- Assess whether animals in the 2 groups have differences in past exposure to different risk factors
14
Q
What do you calculate in case-control studies?
A
- ODDS RATIO to indirectly estimate RR provided that incidence of disease is low and cases + controls are truly random samples from the same population
- Good for studying RARE DISEASES
- Can assess more than one exposure in the same study
- *watch for recall bias (did exposure actually come before the disease)
o Hard when there is a long latent period (Ex. cancer)
15
Q
Statistically significance does NOT equal causality
A
- To prove causal association we need to describe a chain of events
o From cause to effect at the molecular level
16
Q
**What is confounding or a confounder?
A
- Effect of an extraneous variable that can wholly or partly account for an apparent association between variables in an investigation
- *can produce a spurious association between study variables, or can mask a real association
17
Q
What are the 3 ‘criteria’s’ that a confounder must be?
A
- Be associated with the response variable
- Be associated with risk factor (exposure or treatment) of interest
- Not be an intervening or intermediate step between the risk factor and response
18
Q
Component model of causation
A
- ALL disease is MULTIFACTORIAL
- Sufficient vs. necessary causes
- Casual mechanism remains constant
- *strength of association between exposure of interest and outcome will VARY
o *depends on distribution of risk factors
19
Q
What is a sufficient cause?
A
- if it inevitably produces an effect
o Virtually ALWAYS comprises a number of COMPONENT CAUSES - Particular disease may be produced by different sufficient causes
20
Q
What is a necessary cause?
A
- If a risk factor is a component of EVERY SUFFICIENT CAUSE
21
Q
What are the components of a sufficient cause?
A
- Factors may present concomitantly or may follow one another in a chain of events
- When there are a number of chains with one or more factors in common then we have a ‘causal web’
22
Q
What is causal complement?
A
- The SHARED COMPONENT CAUSES that make up a sufficient cause
23
Q
Interaction among causes
A
- 2 or more component causes acting in the SAME SUFFICIENT CAUSES INTERACT CAUSALLY TO PRODUCE DISEASE
24
Q
What is the objective of epidemiological investigations of cause?
A
- The ID of sufficient causes and their component causes
- *removal of one or more components from a sufficient cause swill then PREVENT disease produced by the sufficient cause
25
Q
What is a web of causation?
A
- Direct and indirect causes representing a chain of actions with indirect causes activating direct causes
26
Q
Relationships are shown using a causal diagram
A
- Direct causes are often the PROXIMAL causes emphasized in therapy
- Indirect causes are where effects of exposure are mediated through one or more intervening variables
27
Q
***What are the Hill’s criteria for causality?
A
- Temporality
- Strength of association
- Biological gradient or dose response
- Coherence or plausibility
- Consistency
- Specificity
- Analogy
- Experimental evidence
28
Q
**Hill’s criteria: time sequence
A
- Cause must ALWAYS PRECEDE effect in time
o *but same factor could occur again after disease in some individuals - *difficult to establish time sequence, especially with surrogate exposure measures
29
Q
Hill’s criteria: strength of association
A
- Strong statistically significant association between factor and disease INCREASES likelihood that the factor is causal
- Assumes that it is less likely that residual confounding could explain the result
30
Q
What does strength of association depend on?
A
- Distribution of other components of the sufficient cause
31
Q
What is an example of an important weak association that has been consider causal?
A
- Environmental tobacco smoke and lung cancer
32
Q
What is an example of a strong association due to confounding?
A
- Birth order and Down’s syndrome
33
Q
Hill’s criteria: biological gradient
A
- Dose-response relationship between a factor and disease INCREASES PLAUSIBILITY of factor being causal
- *exceptions to linear change: threshold
- Most should have a gradient that never changes direction (monotonic)
- *alcohol consumption and death=J-shaped curve
34
Q
Hill’s criteria: consistency
A
- Repeated observations of an association in different populations under different circumstances
- Associations can be causal under unusual circumstances
- *statistical significance should NOT be used to assess consistency
- *example systematic reviews and meta-analysis
35
Q
Hill’s criteria: coherence/plausibility
A
- Compatibility with existing knowledge
- More reasonable to infer that a factor causes a disease if a plausible biological mechanism has been IDed than if such a mechanism is NOT known
36
Q
Hill’s criteria: specificity
A
- A cause can lead to a single effect OR an effect has one cause
- *not necessary, but can be supportive when it can be logically deduced from the causal hypothesis
37
Q
Hill’s criteria: analogy
A
- Too subjective and open
- Possible source of new hypothesis
- Ex. smoking and stomach cancer plausible because smoking has been associated with a number of other cancers
38
Q
Hill’s criteria: experimental evidence
A
- Clinical trials, animal lab experiments OR both
- Uncertainty in extrapolating across species and outside tested dose ranges
39
Q
Hill’s criteria has reservations and exceptions
A
- They are ‘viewpoints’ or ‘perspectives’
- *there is significant individuality in interpreting the same evidence
o Only agreed 68% of the time
40
Q
There must be a MATHEMATICAL ASSOCIATION between exposure and hypothesized effect or outcome
A
- In most cases, outcome (disease) have a monotonic association with the increasing exposure
- *besides temporality, there are NO criteria that are either necessary or sufficient
41
Q
What are some sources of area?
A
- Chance
- Systematic error in design of study or data
- Confounding or mixing effects
- Consistency
- *observed association must NOT be ENTIRELY due to any source of error
42
Q
How can chance be assessed?
A
- Correct application of statistical analysis (ex. p-values or CI)
43
Q
Systematic error in the design of study or data
A
- Sampling bias
- Misclassification of exposure or disease status
44
Q
Confounding or mixing effects
A
- Resulting from a third unaccounted for risk factor associated with BOTH exposure and disease
45
Q
Consistency
A
- Across different study designs and different study populations is SUPPORTIVE
