Trends in Obesity Prevalence by Race and Hispanic Origin—1999-2000 to 2017-2018

Abstract

Public health surveillance of obesity describes changes over time and identifies subgroups at risk of obesity-related consequences.We analyzed trends, stratified by race and Hispanic origin, in high weight for length, obesity, and severe obesity in the US from 1999 to 2018.Methods | Data from the National Health and Nutrition Examination Survey (NHANES), a cross-sectional survey of the civilian noninstitutionalized US population, were used.Participants older than 17 years provided consent; for those younger than 18 years, parental permission was obtained, and documented assent was obtained for 7 through 17 years.NHANES was approved by the National Center for Health Statistics (NCHS) Research Ethics Review Board.The response rate was 76.0% during 1999-2000 and declined to 48.8% during 2017-2018.1 During the physical examination, standardized measurements of weight, length, and height were conducted.Among children younger than 24 months, high weight for length was defined as weight for length at or above the 97.7th percentile of the World Health Organization growth standards.2 Among children and adolescents between 2 and 19 years, obesity was defined as body mass index (BMI) for age at or above the 95th percentile, and severe obesity was defined as BMI for age 120% of the 95th percentile or higher on the CDC growth charts.3 Adult obesity and severe obesity were BMI of 30 or higher and 40 or higher, respectively.Linear and quadratic trends were evaluated in regression models that treated 2-year cycles as continuous.Joinpoint software was used to assess changes in slopes.Differences in slope between race and Hispanic-origin groups were tested, and Satterthwaite-adjusted F statistic P values were reported.Trends were stratified by age (children and adolescents) and sex (adults) because of significant 3-way interactions.Although trends using 2-year cycles are more efficient than using 4-year pooled estimates, 4 the 2-year estimates for youth were not reliable, so pooled 4-year estimates were shown.Examination sample weights were used to adjust for oversampling, nonresponse, and noncoverage.Prevalence estimates were evaluated using NCHS presentation standards.5 Estimates for adults were age adjusted to the projected 2000 census population.A 2-sided P < .05determined statistical significance.SAS version 9.4 (SAS Institute Inc) and SUDAAN version 11.0 (RTI International) were used for analyses.