Statistical Analysis. Pre-implementation and postimplementation
groups were compared with respect to
potential confounding variables using the two-independentsample
t-test, Wilcoxon rank-sum, or chi-squared tests as
appropriate. We compared nurses’ attitudes toward
smoking cessation counseling pre-implementation and
post-implementation using the paired t-test for the two
decisional balance subscales and the signed rank test for
the self-efficacy and role satisfaction items. We used
logistic regression to compare performance of individual
A’s across the two periods and adjusted for those
covariates that differed significantly between the preintervention
and post-intervention cohorts in bivariate
analysis (p≤0.05). Similarly, we used linear regression to
compare the 5A’s composite score across the two periods.
To check for secular trend, we used linear regression to
analyze the 5A’s composite score as a function of month
of enrollment during the pre-intervention period