There are various procedures to account for missing values in clinical trials. Mixed Model and Last Observation Carried Forward (LOCF) analyses are commonly employed methods. We compared the two methods by reanalysing Positive and Negative Syndrome Scale (PANSS) scores from the European First Episode in Schizophrenia Trial (EUFEST). While the predefined Mixed Model analysis detected no differences between PANSS total score changes in first episode patients treated with haloperidol, amisulpride, quetiapine, olanzapine or ziprasidone over the course of a one year treatment trial, a LOCF analysis revealed significant differences in favor of the new generation antipsychotics over haloperidol. Implications for clinical trial design and the interpretation of such studies are discussed.
Copyright © 2012 Elsevier B.V. All rights reserved.