Finding the 'Best' Measurement of Body Weight Status Was Difficult with Missing Data

• Activity Number: 557
• Type: Contributed
• Date/Time: Wednesday, August 3, 2016 : 10:30 AM to 12:20 PM
• Sponsor: Section on Statistical Consulting
• Author(s): Liyun Zhang* and Michele Polfuss and Kathleen J. Sawin and Pippa Simpson
• Companies: Medical College of Wisconsin and University of Wisconsin - Milwaukee and University of Wisconsin - Milwaukee and Medical College of Wisconsin
• Keywords: Spina Bifida ; body fat % ; Bland-Altman ; concordance correlation coefficient ; linear regression ; regression tree

Abstract

Background: DEXA scan is the gold standard for measuring body fat percentage (BF%) and defining body weight status. However it is expensive, time consuming, and exposes the subjects to X-ray. For individuals with Spina Bifida there are a varying number of alternatives depending on the degree of impairment. Assessing which is the best is challenging when there is missing data. Method: Analyses were done on 15 children (4-18 years old), only six of which successfully finished all the procedures. Bland-Altman (BA) plots and concordance correlations were used to evaluate different BMI and BF% calculations, BodPod and BIA based on 4 different height measures, and Skinfold measurements . A linear regression and stepwise backward selection were used to determine which best determined body weight status . A regression tree analysis was used to identify variables to be included in the model. Results: No definitive results could be found. Conclusion: Defining the best in a setting where there are non random missing data is problematic.