## Glossary of Epidemiology: 8) Cohort Analysis - Measures of Effect

### Deck Info

#### Description

#### Tags

#### Other Decks By This User

- Number needed to treet
- 1/(Po-P1) = -1/RD

- Utility of Risk Risk Difference
- RD can have more clinical/public health meaning since is characterizes absolute risk

- RR vs. RD
- -RR is less dependent upon the underlying population.

-RD provides incidence information.

-No methods to do multivariate RD.

- 2 reasons to use OR
- 1) Logistic regression for fixed (defined) cohort studies.

2) Case-control estimates

- OR can estimate RR if
- D is infrequent in both E+ and E-

min(Po,P1)<0.1 (10%)

- Major problems with risk, incidence, and odds information
- They do not measure the health "impact" of a particular exposure.

- Attributal fraction
- Proportion of the disease occurence that would be eliminated if exposure to the risk factor were prevented

- Attributable fraction: Synonyms
- -Etiologic fraction

-attributable proportion

-attributable risk

-population attributable risk (percent)

- Who is the Attrubutable fraction calculated for?
- 1) E+ group

2) Entire population

- Attributable fraction for the exposed (discription)
- Proportion of cases in the E+ group attributable to E.

Eg. Among NSAID users, what proportion of peptic ulcers are caused by NSAIDS?

- Attributable fraction for the exposed (Eqn)
- (P1-Po)/P1 (risk data)

(I1-Io)/I1 (rate data)

(RR-1)/RR (risk data)

(IRR-1)/IRR (rate data)

(OR-1)/OR (rare disease assumption)

- Attributable fraction for Population (disc)
- Proportion of cases in the population attributable to E.

Eg. For a given patient population, what proportion of peptic ulcers are caused by NSAIDS? (need strength of the association between E and D and prevalence of E+)

- Attributable fraction for Population (eqns)
- (P-Po)/P (risk)

(I-Io)/I (rate)

¶(RR-1)/[1+¶(RR-1)] (risk)

¶(IRR-1)/[1+¶(IRR-1)] (rate)

¶c(OR-1)/OR (Odds)

- Attributable Fraction for population (¶ and ¶c)
- ¶ = N1/N (proportion exposed in cohort for risk data)

¶ = T1/T (Proportion exposed for person-time for rates)

¶c = a/(a+b) (exposure prevalence in D+ for rare D)

- Attributable fraction for exposed (protective exposures) (disc)
- Also called prevented fraction or effictiveness.

Often used in vaccine studies

- Attributable fraction for exposed (protective exposures)(eqn)
- 1)(Po-P1)/P1

1)1-RR

- Marginal cost-effecitiveness
- (Difference in cost)/(Events prevented)

See 8-12