Table 2: Multivariate logistic regression models.

Estimate (SE)Adjusted odds ratio (95% CI)Pr (> )

(A) Cross-sectional dataset (
Dependent variable = current
smoking at baseline ( )
 Intercept.83 (.56).13
 Self-efficacy−1.75 (.68).17* (.05, .66).01
 Narcotic pain medication use.46 (.22)2.50 (1.08, 5.81).033
 Emotional responsiveness−.62 (.24).29 (.11, .72).008
 Age−.56 (.23).57* (.36, .89).014
c-statistic = .76

(B) Prospective dataset ( )
Outcome = current smoking at 36
weeks after baseline ( )
 Intercept−1.02 (.74).17
 At least 12 years education
−.56 (.26).33 (.12, .90).03
 Emotional responsiveness−.57 (.24).32 (.13, .82).018
 Categorical body mass index.54 (.29)1.72 (.98, 3.03).058
c-statistic = .71

with missing educational attainment data; *odds of smoking for each 1 SD increase from median score.
with missing self-efficacy or use of narcotic pain medication data.