Lecture 13 (Causality) Flashcards
Importance of the “Web odf Causation”:
- lists out all of the possible causes of a disease/adverse outcome
- some are modifiable, some are not. Modify the modifiable causes, and decrease rates of disease/adverse outcomes.
- YOU DO NOT HAVE TO UNDERSTAND ALL THE CAUSAL MECHANISMS TO INTERVENE AND PREVENT
Sufficient definition:
- the cause will lead to the effect regardless of other factors
Necessary definition:
- the effect will not occur without both the cause and other factors
If you need other cofactors in addition to the cause to manifest disease, the cause if classified as:
neccesary
Koch’s Postulates:
- microorganism in diseased, not in the healthy
- microorganism can be isolated
- isolated microorganism can cause disease in healthy host
- microorganism can be re-isolated from newly infected healthy host
The 8 Bradford Hill Guidlines:
- temporality
- strength association
- dose response
- reversibility
- consistency
- biological plausibility
- specificity of association
- analogy
What is the only required criteria for causality?
temporality
- exposure to cause precedes disease
Temporality:
- exposure to cause precedes disease
ONLY REQUIRED CRITERIA FOR CAUSALITY
Strength of Association:
- magnitude of association between exposure and disease
Dose-response:
- disease rate increases as exposure rate increases
- can be linear or U-shaped correlation
Threshold:
- important for dose-response
- lowest dose at which response (disease) occurs
Reversibility:
- remove exposure, disease rate decreases
- may not always be true if the pathogenic process is irreversible
Consistency:
- results across different study types and populations are similar
Biological plausibility:
- mechanism of how cause initiates disease must be plausible
Specificity:
- exposure/cause is associated with one specific disease outcome
- watch for exposures that cause multiple outcomes
Analogy:
- is exposure/outcome relationship similar to other known causal relationships
What are the two weakest Bradford-Hill criteria regarding cause?
- reversibility
- pathogenic process may not be reversible
- specificity
- cause may have more than one adverse outcome
Ecological study design:
- analyze groups rather than individuals
- i.e. compare average smoking levels with average lung cancer rates in a county/state
- stat: correlation
- good at generating hypotheses
What is the problem with ecological studies?
- you cannot determine if the individuals who had the exposure developed the adverse outcome
- YOU JUST HAVE RATES
What is ecologic fallacy?
- attribute overall group characteristics to individual members of that group
Meta-analysis:
- a study of studies - what is the overall result?
- take the results of all studies and combine
Pooled analysis:
- combine data from a lot of studies, re-analyze
- increases power and precision
- good with similar study protocols
What are the two forms of quantitative review?
- meta-analysis
- pooled-analysis
Is disease randomly distributed?
- No.
- factors that determine disease are knowable and modifiable.
MODIFICATION WILL LEAD TO PREVENTION OR CONTROL