Differences in means of continuous variables by treatment
for mother’s baseline characteristics were determined using
analysis of variance; differences in categorical variables by
treatment were analyzed using contingency table analyses and
associated chi-square tests or Fisher’s exact test when expected
counts were low. Differences in children’s baseline characteristics by treatment were analyzed using linear mixed models
for continuous variables and logistic regression analyses in the
context of generalized estimating equations (31) for categorical
variables, to allow adjustment for correlation between siblings.
Baseline characteristics found to be significantly different were
adjusted for in subsequent analyses.