Using Imputed Body Mass Index to Assess the Relationship Between Obesity and Mortality

Abstract

Listen

Translate article

Introduction: Studies that assess the relationship between obesity and mortality often use self-reported height and weight to calculate obesity. However, there is evidence that this data is biased which can lead to inaccurate obesity reporting. Therefore, the true relationship between obesity and mortality is unclear. The purpose of this study is to explore multiple imputation as a novel method to estimate clinically measured height and weight. The goal is to reduce bias in self-reported height and weight data, which will improve obesity reporting accuracy and allow for a better understanding of the relationship between obesity and mortality. Methods: Multivariate imputation by chained equations was used to impute clinically measured height and weight. Body mass index (BMI) was calculated from the imputed values and then divided into categories (Low, Referent, Overweight, Class I Obesity, Class II-III Obesity). Cox regression was applied to the data and hazard ratios were calculated for each BMI category, then compared to those from clinically measured and self-reported data to determine present bias. Results: In males, the imputed data increased bias between clinically measured and self-reported hazard ratios. In females, the imputed data decreased bias between clinically measured and self-reported hazard ratios in categories Class I Obesity and Class II-III Obesity, but increased bias was found in categories Low and Overweight. Discussion: There is a lack of evidence that imputing clinically measured height and weight minimizes bias when measuring the association between obesity and mortality through hazard ratios. More research is needed to further understand the results.