(if the covariate under investigation is on a continuous scale). Of course,
application of the conditional logistic model is still desirable, at least in the case
of a continuous covariate, because it would provide a measure of association;
the odds ratio.
Example 11.12 Refer to the data for low-birth-weight babies in Example
11.11 (Table 11.14) and suppose that we want to investigate the relationship
between the low-birth-weight problem, our outcome for the study, and the
weight of the mother taken during the last menstrual period. An application of
the simple conditional logistic regression analysis yields the results shown in
Table 11.16. The result indicates that the e¤ect of the mother’s weight is nearly
significant at the 5% levelðp¼ 0 : 0593 Þ. The odds ratio associated with, say, a
10-lb increase in weight is
expð 0 : 2114 Þ¼ 0 : 809If a mother increases her weight about 10 lb, the odds on having a low-birth-
weight baby are reduced by almost 20%.
Note:An SAS program would include these instructions:
INPUT SET CASE MWEIGHT;
DUMMYTIME = 2-CASE;
CARDS;
(data)
PROC PHREG DATA = LOWWEIGHT;
MODEL DUMMYTIME*CASE(0) = MWEIGHT/TIES = DISCRETE;
STRATA SET;
where LOWWEIGHT is the name assigned to the data set, DUMMYTIME is
the name for the makeup time variable defined in the upper part of the pro-
gram, CASE is the case–control status indicator (coded as 1 for a case and 0
for a control), and MWEIGHT is the variable name for the weight of the
mother during the last menstrual period. The matched SET number (1 to 15 in
this example) is used as the stratification factor.
11.7.2 Multiple Regression Analysis
The e¤ect of some factor on a dependent or response variable may be influ-
enced by the presence of other factors through e¤ect modifications (i.e., inter-
TABLE 11.16
Variable Coe‰cient
Standard
Error zStatistic pValueMother’s weight 0.0211 0.0112 1.884 0.0593
418 ANALYSIS OF SURVIVAL DATA