Hi Ron, thanks for making me go back and re-read this paper. The results are presented in a rather confusing way. First they present the proportion of time for which a person took an antipsychotic (which they suggest is more or less equivalent to the notion of ‘compliance’), then they present the average dose people took while taking the antipsychotic, then they present whether or not people had any ‘drug-free periods’ lasting 30 days or more, then whether there was polypharmacy and finally they present cumulative antipsychotic exposure (long-term dose years). Outcomes are SOFAS, remission and CGI (clinical improvement). Having any drug free periods is associated with worse social functioning (SOFAS) scores (just), but so is higher dose and cumulative dose. On the other hand, a higher proportion of drug free time, lower dose, cumulative dose and polypharmacy are all associated with better CGI scores but having or not having drug free periods is not. So it seems that overall, lower compliance is associated with greater improvement and most of the results point in the direction of higher antipsychotic exposure being associated with worse outcomes. They don’t provide data for the outcome of people who discontinued or who used low doses, which would have been more useful. And, as with the other long-term follow-up studies, much of the variation is probably accounted for by variation in severity of the underlying problem.