Week 5 Flashcards
What is the goal of epidemiology
Identification of causes and preventions for disease
What is strong evidence
- Lowest possible random sampling error
- Based on a strong design free of selection and information biases and under minimal influence of confounding
What dose multivariate world mean
Never a simple exposure outcome model is never a reality relationships are multi variable
What is confounding
Distortion of an actual association due to a mixing of effects between the exposure and an incidental variables
Why dose confounding occur
Because the exposed group and the unexposed group are not exchangeable
They differ by factors other than their exposure status
What is confounding a problem for
A problem for observational studies
How can you reduce confounding
Randomization
What is the criteria for defining a confounding variable
- Casually associated with the outcome (true risk factor)
- Noncausally or causally associated with the exposure
- Not an intermediate in the causal pathway between exposure and outcome
How do you deal with confounding at the design stage
Restriction (study only men or women)
Matching (case and control groups have similar characteristics
Randomization Experimental studies
How do you deal with confounding at the analysis stage
Standardization (make 2 groups the same)
Stratfied anayalysis
Include confounding factors in a multivariate regression model
What is effect modification
Where the exposure disease association changes over values of some third variable
Modifies the effects of the exposure on the outcome not a cause of the exposure and outcome
Don’t adjust just show how the effect modifier is operating
Why is exploring effect modification important
Helps to understand the biology of an effect
We can identify high risk populations
Difference between confounding and effect modification
The stratums RR are different then it is an effect modifier if they are the same its confounding
What do you do if for confounding if your curde rr=adj rr
No major confounding use crude rr
What do you do if Crude rr dosen’t = adj rr
Confounding is present use adj RR
