Analysis of various subset
variables demonstrated similar efficacy for each variable.
A stepwise regression analysis demonstrated that AE reports
of nausea and vomiting were most likely due to prerandomization
events, particularly opioid consumption and
presence of nausea prior to randomization.