A Comparison of Methodologies for Assessing Onset of Action in Neuropsychiatric Clinical Trials

• Activity Number: 405
• Type: Contributed
• Date/Time: Thursday, August 12, 2004 : 8:30 AM to 10:20 AM
• Sponsor: Biopharmaceutical Section
• Author(s): Craig H. Mallinckrodt*+ and Michael Detke and Kaiser Christopher and John Watkin and Geert Molenberghs and Raymond J. Carroll
• Companies: Eli Lilly and Company and Eli Lilly and Company and Eli Lilly and Company and Eli Lilly and Company and Limburgs Universitair Centrum and Texas A&M University
• Keywords: mixed-effects models ; survival analyses ; antidepressants ; missing data

Abstract

Antidepressant medications have a delay in onset of effect. Given the importance of onset of action, this is an active area of investigation. This study compared Type I error rates and power of the Kaplan-Meier Product Limit method (KM) to a categorical mixed-effects model repeated measures approach (MMRM-CAT). Four scenarios were simulated by varying mean improvements over time. Within each scenario, three sets of data were created by either leaving the data complete or deleting observations to generate MNAR data according to one of two dropout patterns. All 16 (twice weekly for eight weeks) post-baseline visits were included (frequent assessment), or only weeks 1, 2, 4, 6, and 8 (traditional assessments). Onset was defined as a sustained 20% improvement from baseline. Frequent assessments improved precision compared with traditional assessments in KM more than MMRM-CAT, but the reductions were small in all cases. Both KM and MMRM-CAT provided reasonable control of Type I error. The MMRM-CAT analysis using a traditional visit schedule yielded more power and was biased less by subject dropout than KM with either a traditional or a frequent assessment schedule.