VALIDATING SELF-REPORTED BODY MASS INDEX FOR SOCIAL DESIRABILITY BIAS IN A NATIONAL SURVEY LINKED WITH ELECTRONIC HEALTH RECORDS
Author(s)
Matthew Brault, MPP1, Kathy Annunziata, BA2, Nikoletta Sternbach, BA3.
1Senior Research Consultant, Oracle Life Sciences, Columbia, MD, USA, 2Oracle Life Science, Matawan, NJ, USA, 3Oracle Life Science, Austin, TX, USA.
1Senior Research Consultant, Oracle Life Sciences, Columbia, MD, USA, 2Oracle Life Science, Matawan, NJ, USA, 3Oracle Life Science, Austin, TX, USA.
OBJECTIVES: Body mass index (BMI) is widely used to estimate obesity prevalence, cardiometabolic risk, and eligibility for weight-management treatments. Studies using BMI calculated from self-reported height and weight may suffer from bias toward socially desirable values. Prior research has suggested using distributional differences between the exam-based and self-reported measures to adjust estimates. This study examines the use of these distribution-adjusted BMI measures and validates self-reported BMI values in the National Health and Wellness Survey (NHWS) against clinical measures of BMI from participants’ electronic health records (EHR).
METHODS: Weighted cumulative distribution of adult BMI from the 2021-2023 National Health and Nutrition Examination Survey (NHANES) were compared against the 2022 US NHWS, a nationally representative online survey of adults age 18 and over. A subset of n=1996 NHWS respondents with available EHR records containing structured values of BMI were compared descriptively against their survey reported values using standard BMI categorizations (Underweight; Normal weight; Overweight; Obese). Regression modeling was used to identify possible predictors of discordant values.
RESULTS: The NHWS cumulative distribution was shifted modestly toward lower BMI values relative to NHANES across much of the distribution. Separation was apparent from lower-to-middle percentiles and persisted through the upper percentiles. In the EHR-linked NHWS analysis, 73% of respondents had concordant categorizations of BMI categories; 10% were miscategorized in higher BMI categories in the NHWS while 17% were miscategorized in a lower BMI category. No obvious patterns of discordance were associated with demographic groups.
CONCLUSIONS: The 2022 US NHWS sample appears to represent a lower BMI distribution than the 2021-2023 NHANES sample. Because most NHWS respondents tended to accurately report their BMI, distribution-based adjustments would correct some error but introduce others. Further research may be needs to better understand who misreports their BMI and why.
METHODS: Weighted cumulative distribution of adult BMI from the 2021-2023 National Health and Nutrition Examination Survey (NHANES) were compared against the 2022 US NHWS, a nationally representative online survey of adults age 18 and over. A subset of n=1996 NHWS respondents with available EHR records containing structured values of BMI were compared descriptively against their survey reported values using standard BMI categorizations (Underweight; Normal weight; Overweight; Obese). Regression modeling was used to identify possible predictors of discordant values.
RESULTS: The NHWS cumulative distribution was shifted modestly toward lower BMI values relative to NHANES across much of the distribution. Separation was apparent from lower-to-middle percentiles and persisted through the upper percentiles. In the EHR-linked NHWS analysis, 73% of respondents had concordant categorizations of BMI categories; 10% were miscategorized in higher BMI categories in the NHWS while 17% were miscategorized in a lower BMI category. No obvious patterns of discordance were associated with demographic groups.
CONCLUSIONS: The 2022 US NHWS sample appears to represent a lower BMI distribution than the 2021-2023 NHANES sample. Because most NHWS respondents tended to accurately report their BMI, distribution-based adjustments would correct some error but introduce others. Further research may be needs to better understand who misreports their BMI and why.
Conference/Value in Health Info
2026-11, ISPOR Europe 2026, Vienna, Austria
Value in Health, Volume 29, Issue 12S
Code
RWD8
Topic
Methodological & Statistical Research, Real World Data & Information Systems
Topic Subcategory
Reproducibility & Replicability
Disease
Diabetes/Endocrine/Metabolic Disorders (including obesity), No Additional Disease & Conditions/Specialized Treatment Areas