actions). Therefore, to provide a more comprehensive analysis, it is very desir-
able to consider a large number of factors and sort out which ones are most
closely related to the dependent variable. In this section we discuss a multi-
variate method for risk determination. This method, which is multiple logistic
regression analysis, involves a linear combination of the explanatory or inde-
pendent variables; the variables must be quantitative with particular numerical
values for each patient. A covariate or independent variable, such as a patient
characteristic, may be dichotomous, polytomous, or continuous (categorical
factors will be represented by dummy variables). Examples of dichotomous
covariates are gender and presence/absence of certain comorbidity. Polytomous
covariates include race and di¤erent grades of symptoms; these can be covered
by the use ofdummy variables. Continuous covariates include patient age and
blood pressure. In many cases, data transformations (e.g., taking the loga-
rithm) may be desirable to satisfy the linearity assumption.
9.2.1 Logistic Regression Model with Several Covariates
Suppose that we want to considerkcovariates simultaneously; the simple
logistic model of Section 9.1 can easily be generalized and expressed as
pi¼
1
1 þexp½ðb 0 þ
Pk
j¼ 1 bjxjiÞ
i¼ 1 ; 2 ;...;n
or, equivalently,
ln
pi
1 pi
¼b 0 þ
Xk
j¼ 1
bjxji
This leads to the likelihood function
L¼
Yn
i¼ 1
½expðb 0 þ
Pk
i¼ 1 bjxjiÞ
yi
1 þexpðb 0 þ
Pk
j¼ 1 bjxjiÞ
yi¼ 0 ; 1
from which parameters can be estimated iteratively using a computer-packaged
program such as SAS.
Also similar to the univariate case, expðbiÞrepresents one of the following:
- The odds ratio associated with an exposure if Xi is binary (exposed
Xi¼1 versus unexposedXi¼0); or - The odds ratio due to a 1-unit increase ifXiis continuous (Xi¼xþ 1
versusXi¼x).
Afterbb^iand its standard error have been obtained, a 95% confidence interval
for the odds ratio above is given by
326 LOGISTIC REGRESSION