6 - Cause and Effect Flashcards
What is a risk factor?
Anything that increases the likelihood of disease
Criteria for Causal Association in Biology
(HILL)
- Strength of the association
- Consistency of the association
- Specificity of the association
- Time sequence
- Biological gradient
- Plausibility
- Coherence
- Experiment
- Analogy
1 Strength of Association
Quantitative strength (relative risk > 2.0) (magnitude)
2 Consistency of the Association
- Different study designs
- Different groups (within reason)
- Different places (reproducibility)
3 Specificity of the Association
Is there a specific disease being caused?
4 Time Sequence
Exposure should occur before disease
5 Biological Gradient
Association between dose and effect (symptoms)
6 Plausibility
Depends on bio knowledge of the age
* Mechanism may not be understood BUT If strength great enough ➔ investigate new biological process
7 Coherence with Existing Information
Causal relationship should have coherence with known facts about the disease
8 Experiment
Can changes in putative causal variables increase / reduce rate of disease in question?
9 Analogy
Examples of previous scenarios can be put forward to support hypothesis
Possible types of association
- None
- Artefactual (Chance or Bias)
- Indirect
- Casual
No association
Two variables are independent of one another
* retain null hypothesis
* did the test have sufficient power?
Artefactual association (Chance)
Chance is hopefully eliminated by stat. testing
* Must control family-wise error rate for multiple testing
Artefactual association (Bias)
Systemic variation
* interview or recall bias
* design fault
* significance is irrelevant if bias present
Indirect Association
A variable confounds a relationship if it is associated both with the exposure and the disease
Some confounding variables
May exaggerate, reduce or mask association
* Age
* Education
* SES
How can we deal with confounding factors?
- Randomisation (similar comp. in both groups)
- Matching (choosing participants based on similarity between groups)
- Stat. methods (Stratification and ANCOVA)
ANCOVA
Type of ANOVA
* controls for presence of continuous var. (age)
* Use age as covariate
