7R: EBM6 - Study Interpretation

Question 1

systematic error

Answer 1

a flaw in study design, data collection, or data analysis that leads to wrong conclusions
- not impacted by sample size

Question 2

3 types of bias

Answer 2

selection bias: distortion in the effect because of the manner in which subjects are selected or selective losses prior to data collection
information bias: measurement of exposure or disease is systematically inaccurate. Also called misclassification
confounding: when a third variable distorts the exposure-outcome relationship

Question 3

internal vs. external validity

Answer 3

internal: quality of the study and the validity of the findings
external: where did the population come from and can this be generalized?

Question 4

non-differential vs. differential classification of disease

Answer 4

non-differential: when you make the same mistake for exposed and non-exposed
differential: when you don’t make the same mistake for exposed and non-exposed

Question 5

how to avoid selection bias

Answer 5

In study design stage

• develop explicit case definition
• enrolling all cases in a defined time and region
• striving for high participation rates
• taking precautions to ensure representativeness

Question 6

how to avoid information bias

Answer 6

In study design stage

• use memory aids
• validate exposure and disease state
• use standardized data collection forms
• blinding

Question 7

how to avoid confounding

Answer 7

in study design stage
- randomize
- match
in analysis stage
- stratified analysis

Question 8

Define:

a. Medical surveillance bias

b. Berkson’s bias

Answer 8

a. exposure is a medical condition that leads to more frequent/detailed clinical encounters
b. cases and/or controls are selected from the hospital

Question 9

Define:

a. Hawthorne effect
b. association
c. causation

Answer 9

a. people act differently when being observed
b. identifiable relationship between exposure and disease
- no directionality!
c. mechanism leading from exposure to disease (if ____ then ___)

Question 10

8 Causal criteria viewpoints

Answer 10

1. strength of association: strong associations are more likely to be causal
2. consistency: repeatedly observing this in different populations etc.
3. specificity: requires that cause leads to a single effect
4. Temporality: cause precedes effect
5. biologic gradient: increased dose = increased response
6. biologic plausibility: hypothesis matches already known biological facts
7. coherence: hypothesis should not conflict with what is already known about the disease
8. experiment: in the lab or human experiment?