Research Design, Biostatistics, and Literature Evaluation

Dependent Variables

Independent Variables

Test(s)

Continuous

Continuous

Linear regression

Continuous

Categorical

Linear regression

Categorical, ordinal

Continuous

Logistic regression

Categorical, ordinal

Categorical

Logistic regression

Ratios with 95% CI that are used to compare the incidence of categorical outcomes are interpreted in

relation to a reference point (1.0). If the 95% CI surrounding the ratio includes 1.0, the risk, odds, or

hazard of the outcome is equally likely in both groups.

and effect estimates generally better describe outcomes.

Relative risk (RR)

Incidence of outcome in exposed group/incidence of outcome in control group

Relative risk reduction (RRR)

Relative difference in outcome in exposed group/relative difference in outcome in control group

Absolute risk reduction (ARR)

Absolute difference between outcome in exposed and control groups

Odds ratio (OR)

Odds of outcome occurring in group exposed to risk factor or intervention/odds of outcome

occurring in control group

Compares number of events or outcomes between two groups assuming that the incidence remains

constant over a fixed time interval

Hazard ratio (HR)

Hazard rate in exposed group/hazard rate in control group

Compares number of events or outcomes between two groups where the incidence varies over a time

interval

Censoring: Adjusts data so that patients are not included (i.e., “censored”) in the analysis if they did not

experience, or were not observed for, the event

RRR = ARR

(

(

C

C+D

(

(

C

C+D (

(

A

A+B

ARR =

–

A + D

B + C

OR =

(

(

A

A+B

RR = (

(

C

C+D

Outcome

No outcome

Exposed

A

B

Unexposed

C

D

ARR = absolute risk reduction; OR = odds ratio; RR = risk ratio; RRR = relative risk reduction.