The results were analyzed using repeated measures analysis
of variance adjusted for missing data, post-hoc contrasts, and
trends within treatments according to Kirk.6 If more than
50% of assessments were missing for any period, the data from
that period were not analyzed.