Updated prevalence estimates of obesity among adults in the United States, 2005–2018

Abstract

Introduction

Body mass index is an imperfect measure of excess adiposity but is ingrained in both research and clinical practice to define and diagnose adult obesity. The addition of waist circumference measure is recommended to fully capture obesity-related health risks at a population level. The latest reframing of obesity by the Lancet Commission relies on body mass index and waist circumference, as well as organ dysfunction and functional limitations, to differentiate between preclinical and clinical obesity.

Methods

Using data from the National Health and Nutrition Examination Survey (2005–2018), we adhere to the proposed redefinition to examine obesity trends in the adult U.S. population.

Results

In line with previously published estimates of obesity defined solely by body mass index, we show that the prevalence of both preclinical and clinical obesity increased between 2005 and 2018. However, the prevalence of preclinical obesity calculated by BMI plus WC is lower than the estimates derived from using BMI or WC as the sole metrics.

Conclusion

Using the Commission’s added measures yields refined population estimates for two categories—preclinical and clinical obesity. These updated estimates separate preclinical cases—at elevated risk and eligible for prevention—from those meeting clinical-obesity criteria, who require intervention.

Introduction

Since the late 1990 s, the U.S. National Center for Health Statistics (NCHS) has been publishing obesity prevalence estimates, defined as body mass index (BMI) of 30 kg/m² or higher for adults [1]. The Lancet Diabetes & Endocrinology Commission recently developed and published a new definition and diagnostic criteria of preclinical and clinical obesity [2]. This redefinition of obesity incorporates body mass index (BMI) as a measure of overall obesity, along with anthropometric measurements of central obesity, such as waist circumference (WC). The Commission also considers two categories: clinical obesity, characterized by obesity-related organ dysfunction, and preclinical obesity, defined as having excess body fat without impairment of organ function [2]. Reliance on BMI as a sole metric to identify obesity may have led to erroneous prevalence estimates and interpretation of trends, inappropriate conclusions about the efficacy of population-level obesity prevention strategies, and inaccurate projections of obesity-related health risks [2]. In this study, we adhere to the Commission’s new definition of preclinical and clinical obesity status, and use data from the National Health and Nutrition Examination Survey NHANES (2005–2018) to chart obesity trends among adults in the United States [2, 3].

Methods

NHANES is a cross-sectional dataset representative of the non-institutionalized U.S. population [3]. We used the maximum number of 2-year “continuous” cycles, incorporating the greatest possible number of organ dysfunction measures, given the shifting priorities of NHANES across data cycles [3]. The sample was restricted to nonpregnant adults aged 20–74 years old with complete BMI and WC data (72.5% of participants in the pooled data were included). We used NHANES cycle-specific sampling weights, sample strata, and clusters to manage variations in the unequal probabilities of selection and nonresponse [3]. BMI was calculated as weight (kg)/height (m)² [4]. Obesity was defined as a BMI ≥ 30 kg/m² plus WC > 102 cm (men) or WC > 88 cm (women) [4, 5]. The two outcomes are binary indicators representing preclinical or clinical obesity. We defined preclinical obesity as ‘obesity’ [i.e., BMI ≥ 30 kg/m² plus WC > 102 cm (men) or WC > 88 cm (women)] without clear impairment of organ or tissue function. We defined clinical obesity as ‘obesity’ [i.e., BMI ≥ 30 kg/m² plus WC > 102 cm (men) or WC > 88 cm (women)] plus at least one of the following: (1) stroke; (2) heart attack; (3) heart failure; (4) heart disease; (5) osteoarthritis; (6) liver condition; (7) sleep disorder; (8) high blood pressure; (9) functional limitation; (10) low HDL cholesterol; 11) high albumin levels, based on the NHANES “Examination Data”, “Laboratory Data”, and “Medical Condition” sections of the questionnaire data [2, 3]. Individuals with more than one condition were counted for each condition. We adjusted for sex (male, female), drawn from the “Demographics” data.

Results

Between 2005/06 and 2017/2018, preclinical obesity prevalence was lower, ranging from 1.0% to 1.6%, than estimates based on BMI alone (Fig. 1). Over the same period, preclinical obesity prevalence was lower than WC-based estimates, with differences of 17.8% to 19.9%. Preclinical obesity rates declined slightly between 2005/06 (33.6%) and 2007/08 (32.9%). After 2007/08, we observed an increase throughout the remainder of the study period, reaching a peak in 2017/18 (41.2%). Across the study period, clinical obesity prevalence was lower than preclinical obesity, ranging from 6.8% to 10.3%. In Table 1, we show that, over the study period, males had significantly lower preclinical (33.6%, SE = 0.71) and clinical (25.2%, SE = 0.61) obesity prevalence than females (preclinical: 38.6%, SE = 0.63; clinical: 30.3%, SE = 0.64).

Fig. 1.

Historical Trends in Preclinical and Clinical Obesity, Body Mass Index (BMI), and Waist Circumference (WC) for U.S. Adults Aged 20–74: NHANES 2005–2018. Asterisks indicate significant difference evaluated using conditional polynomial regression

Table 1.

Overall prevalence estimates for U.S. Adults with preclinical and clinical obesity by Sex, 2005–2018; n = 21,044

Adults Aged 20–74 Years Old, % and Standard Error (SE)a
	Male			Female			P value for diffb
Measure
Preclinicalc	33.55	0.71		38.55	0.63		0.00
Clinicald	25.22	0.61		30.29	0.64		0.00
BMI ≥ 30	36.21	0.74		38.70	0.62		0.00
WC > 88 or 102	44.87	0.78		64.68	0.71		0.00

^aData are presented as weighted percentages

^bSignificance evaluated using chi-square

^cPreclinical = BMI ≥ 30 + WC > 88 for females or > 102 for males, without medical conditions

^dClinical = BMI ≥ 30 + WC > 88 for females or > 102 for males, + at least 1 medical condition

In Fig. 2, we display the overall trends in clinical obesity by medical condition subtype. Clinical obesity significantly increased for almost all medical condition categories, except for sleep disorders. The proportion of the population with clinical obesity related to heart failure slightly decreased between 2005/06 (16.3%) and 2007/08 (15.9%). This period was followed by an increase from 2009/10 to 2011/12, a slight decrease in 2013/14 (16.4%), and a steady increase to a high of 20.4% in 2017/18. The rate of clinical obesity due to heart disease increased between 2005/06 (9.4%) and 2009/10 (11.9%), followed by a decline in 2011/12 (9.9%). Then, we observed a steady rise from 2013/14 through the remainder of the study period. The proportion of clinical obesity related to heart attacks was at the lowest point in 2009/10 (4.5%), with a continual increase from 2011/12 to 2017/18. As displayed in Table 2, there are significant sex differences in the prevalence of clinical obesity by medical condition subtype during the 13-year period. Females have significantly higher rates than males in all clinical obesity subcategories, except for liver condition (men are 0.06% higher; p = 0.04) and high blood pressure (men are 0.02% higher; p = 0.00).

Fig. 2.

Historical Trends in Clinical Obesity by Medical Condition Subtype for U.S. Adults with Clinical Obesity Aged 20–74: NHANES 2005–2018. Asterisks indicate significant difference evaluated using conditional polynomial regression

Table 2.

Overall Prevalence Estimates for U.S. Adults with Clinical Obesity by Medical Condition and Sex, 2005–2018; n = 9257

Adults Aged 20–74 Years Old (%) and Standard Error (SE)a
	Male		Female		P value for diffb
Conditions
Heart Failure	14.80	0.53	20.77	0.59	0.00
Heart Disease	9.58	0.40	13.74	0.49	0.01
Heart Attack	4.47	0.25	6.96	0.36	0.04
Stroke	2.02	0.14	2.95	0.25	0.00
Osteoarthritis	0.91	0.11	1.14	0.14	0.00
Liver Condition	0.52	0.09	0.46	0.07	0.04
Sleep Disorder	0.09	0.02	0.19	0.06	0.00
High Blood Pressure	0.06	0.02	0.04	0.02	0.00
Functional Limitations	0.00	0.00	0.01	0.00	0.09
Low HDL	0.00	0.00	0.01	0.00	0.09
High Albumin	0.00	0.00	0.00	0.00	0.10

^aData are presented as weighted percentages

^bSignificance evaluated using chi-square

Discussion

Between 2005 and 2018, preclinical and clinical obesity rates among U.S. adults increased substantially. However, the prevalence of preclinical obesity calculated by BMI plus WC is lower than the estimates derived from using BMI or WC as the sole metrics, which suggest a less pronounced rise in obesity prevalence.[6] For clinical obesity, we observed increases in almost all subtypes throughout the 13-year period.

Limitations

This study has limitations. Balancing the inclusion of as many data cycles as possible with the use of the most comprehensive set of measures was challenging. For example, while the report addresses respiratory dysfunction, questions about breathlessness during physical activity were only included starting in the 1999/00 data cycle and continued until 2017/20. Similarly, whole-body percent body fat measured by Dual-Energy X-ray Absorptiometry (DXA) data were collected from 1999/00 to 2005/06, absent for the 2007/08 and 2009/10, and reinstated from 2011/12 to 2017/18, limiting our ability to incorporate fat mass into our measure of obesity. Other measure constraints involve liver conditions and sleep disorders, as the NHANES questionnaire varied in its specificity about these conditions across different data cycles.

Conclusion

Taken together, the Commission’s guidelines allow for a more nuanced examination of trends in obesity prevalence among persons whose obesity status has, until now, been defined solely by their BMI. The new definitions yield more precise prevalence estimates that distinguish those in the preclinical stage—at risk of progressing to obesity-related disease and eligible for prevention—from those who meet clinical-obesity criteria and require intervention. Knowing the prevalence of both preclinical and clinical obesity improves national surveillance, helps policymakers track the true population burden over time, and refines forecast accuracy for future health and economic costs, guiding long-term resource and policy planning.

Author contributions

AWK: Conceptualization, Methodology, Software, Validation, Formal Analysis, Resources, Data Curation, Writing – Original Draft, Writing – Review & Editing, Visualization, Supervision, Project Administration. DK **:** Conceptualization, Resources, Writing – Original Draft, Writing – Review & Editing, Visualization, Supervision, Project Administration.

Funding

The authors declare no funding.

Data availability

The data are available at: https://wwwn.cdc.gov/nchs/nhanes.

Declarations

Ethics approval and consent to participate

This study uses a secondary dataset and did not need ethics approval.

Consent for publication

This study uses a secondary dataset and did not need consent for publication.

Competing interests

The authors declare no competing interests.