Preference-Based Approach to Estimating Treatment Acceptability and Effect in Compliant Patients in a Randomized Controlled Trial

• Activity Number: 627
• Type: Contributed
• Date/Time: Thursday, August 4, 2011 : 8:30 AM to 10:20 AM
• Sponsor: Biopharmaceutical Section
• Author(s): Afisi Ismaila*+ and Stephen Walter
• Companies: McMaster University and McMaster University
• Keywords: Randomized controlled trials ; Intention-to-treat analysis ; Per protocol analysis ; As treated analysis ; Relative Risk ; Preference-based analysis

Abstract

Randomized controlled trials are sometimes plagued by noncompliance in the form of withdrawal from assigned treatment or switching to alternative treatments. We compare the results from a preference-based (PB) analysis [an alternative randomization-based method], an intention-to-treat (ITT) analysis, a per protocol (PP) analysis and an as-treated (AT) analysis in data from a multicenter, blinded randomized trial of 1339 adults evaluating the effect of reamed versus non-reamed intramedullary nailing of tibial shaft fractures on reoperation and/or autodynamization before one year. We estimated the relative risk (RR) by fracture types (open and closed). The RR of comparing reoperation and/or autodynamization outcome rates in reamed versus non-reamed patients were: open fractures [ITT=1.23; PP=1.29; AT=1.30; and PB =1.26]; closed fractures [ITT=0.66; PP=0.67; AT = 0.71 and PB=0.62]; and all fractures types [ITT=0.89; PP=0.92; AT=0.96 and PB = 0.88]. In all results, PB analysis estimates of treatment effects were lower than the estimates from AT and PP. The PB analysis estimate was higher than the ITT estimate in the open fractures group and lower in the closed fracture group.