Citicoline’s effect on cocaine, alcohol, tobacco, and marihuana use plotted as self-reported percent days of use for each week of the treatment and post-treatment period (total days of use divided by total days of follow-up for each study week). Lines represent estimated daily probabilities of cocaine, alcohol, tobacco, and marihuana use as predicted by logistic regression for the treatment period (all substances) and follow-up period (cocaine, alcohol, and tobacco only). Significant or nearly significant differences in changes in the probability of use during the treatment period were observed for alcohol (p=0.02) and marihuana (p=0.051) but not tobacco (p=0.18). Not all participants used all drugs during the study (see text for details).