The Effects of the Order of Applying Multiple Imputation in Subset Analysis Examining the Association Between Body Mass Index (BMI) and Transrectal Ultrasound Prostate Weight

• Activity Number: 428
• Type: Contributed
• Date/Time: Tuesday, August 11, 2015 : 2:00 PM to 3:50 PM
• Sponsor: Biometrics Section
• Author(s): Irene Helenowski* and Hakan Demirtas and Jennifer A. Doll and Borko D. Jovanovic and Michael J. Gurley and Timothy M. Kuzel
• Companies: Northwestern University and University of Illinois at Chicago and University of Wisconsin - Milwaukee and Northwestern University and Northwestern University and Northwestern University
• Keywords: multiple imputation ; trans rectal ultrasound ; race ; subsetting ; BMI ; prostate cancer

Abstract

Multiple imputation is an approach that affects the evidence of associations among prognostic factors and an outcome in the data set. Subset analysis on the other hand is used to show associations between an outcome and a prognostic factor can be influenced by other factors; subset analysis does this in a manner more comprehensive than multivariate regression. Here we study how combining these two approaches affect our observed association between body mass index (BMI) and prostate weight as measured by transrectal ultrasound from the Northwestern University Specialized Program in Research Excellence (SPORE) database (Grant #: P50 CA 090386) and determine that the order of the procedures affects the inferences for the association.