Epidemiology: 1) Study Designs
Terms
undefined, object
copy deck
- Classes of Mistakes
-
External Validity
Internal Validity
Execution - External Validity
- Narrow research question not Clinically relevant. Inappropriate population or outcome
- 6 issues of Internal Validity
-
Fails to address narrow question.
Selection bias
Information bias
Misclassification
Confounding
Power
Surrogate Endpoint - Execution
-
missing data
inaccurate assays
data entry error - Synonyms: Exposure
-
Etiologic factor
Treatment, stimulus
Risk factor
Independent Variable
Prognostic Variable - Synonyms: Disease
-
health status
response variable
endpoint
dependent variable
outcome - Primary Study designs
-
Clinical Trial
Cohort
Case Control -
Clinical Trial:
Risk
Risk Ratio
Risk Difference -
Risk (E+): p1=a/N1
(E-): p0=b/N0
Risk Ratio: p1/p0
Risk Difference: p1-p0 - Strengths of RCT
-
Random
Strong Inferential Validity - Limits of RCT
-
Impractical:
rare disease
long follow-up
difficult to manipulate
Unethical
Impossible -
Cohort:
Risk
Risk Ratio
Odds Ratio
Risk Difference -
Risk (E+): p1=a/N1
(E-): p0=b/N0
Risk Ratio: p1/p0
Odds Ratio: (ad)/(bc)
Risk Difference: p1-p0 -
Cohort (variable follow-up)
Incidence Rate
Rate Ratio
Rate Difference -
Incidence rate (E+) = I1 = a/T1
Incidence rate (E-) = I0= a/T0
Rate Ratio: IRR = I1/I0
Rate Difference: IRD = I1-I0 - Strengths of Cohort Study
-
"Natural expriment"
Easy to understand - Limits of cohort study
-
Other differences betwen E+, E-
Impractical:
rare disease
long follow-up - Strengths of Case Control
-
Logistically Effecient
Good for rare disease
Efficient for sporadic exposures - Limits to case-control
-
More susceptible to bias
Difficult to understand
Unless nested: no disease rate
cannot always determine E after D
Inefficient for infrequent exposures
Differences between E+ and E- - Types of Ecological Studies
-
Geographic/Static exposure
Temporal or changing exposure - Limits to static ecologic studies
-
Temporal precedence
Variation in other factors
confounding
Ecological Fallacy - Ecological studies best when:
-
E is dominant risk factor for D
E is manipulated externally (change in policy) or other temporal variation - Secondary designs
-
partial cohort: E+ only
partial case-study: D+ only
Cross sectional - Why use E+ only
-
Very rare D and information needed quickly. Can be augmented with external controls.
Often use pre/post design - good and bad about D+ only?
-
limit of no D- group.
good if need info quick. - Limits of cross-sectional study
-
temporality:
E is altered by D+
Survival post D+ is affected by E - Inception Cohort
-
all individuals assembled at a given
point based on some factor, e.g. where they live or work