An evaluation of the effects of lithium in the treatment of chronic alcoholism. II. Assessment of the two-period crossover design.

Analysis of self-reported alcohol intake by alcoholic subjects participating in a two-period crossover trial of lithium versus placebo failed to demonstrate a significant effect of lithium in reducing alcohol consumption. An appraisal of the adequacy of this study design to meet the original objective of the trial was undertaken. It was impossible under this study design to simultaneously and unequivocally estimate effects of treatment periods, drug carryover, sequence group and other period by treatment interactions in addiction to the lithium/placebo effects. Presence of one or more of these additional influences on the trial outcome not only complicated the interpretation of results but diminished the power of the trial design to test the major hypothesis of interest, that lithium could reduce alcohol intake. We conclude that the two-period crossover design has serious drawbacks in this clinical trial setting and that alternative strategies need to be considered for future trials.