Impact of Missing Data on Type 1 Error Rates in Noninferiority Trials

• Activity Number: 670
• Type: Contributed
• Date/Time: Thursday, August 5, 2010 : 10:30 AM to 12:20 PM
• Sponsor: Biopharmaceutical Section
• Author(s): Bongin Yoo*+
• Companies: Bristol-Myers Squibb
• Keywords: ANCOVA ; linear mixed effects model ; non-inferiority trial ; missing data ; pattern-mixture model

Abstract

A simulation study is conducted to investigate the impact of different types of missing data on different statistical analyses; four different linear mixed effects models and analysis of covariance (ANCOVA), in non-inferiority trial settings for the analysis of longitudinal continuous data. Our simulations suggest that two linear mixed effects models using unstructured covariance matrix for within subject correlation with no random effects or first-order autoregressive covariance matrix for within subject correlation with random coefficient effects provide well control of type I error rate when the missing data are MCAR or MAR. ANCOVA using LOCF imputed data set is the worst method in terms of bias and type 1 error rate. Pattern-mixture model does not show much improvement on controlling type 1 error rate under MNAR but is markedly inferior under MAR.