Measuring Racial Differences in Obesity Risk Factors in Non-Hispanic Black and White Men Aged 20 Years or Older

Abstract

Obesity prevalence in the United States has increased drastically in the last two decades. Racial differences in obesity have emerged with the increase in obesity, with temporal trends because of individual, socioeconomic, and environmental factors, eating behaviors, lack of exercise, etc., raising questions about understanding the mechanisms driving these racial differences in the prevalence of obesity among non-Hispanic Black (NHB) and non-Hispanic White (NHW) men. Although many studies have measured obesity using body mass index (BMI), little is known about waist circumference (WC). This study examines variations in obesity among NHW and NHB using BMI and WC. We used National Health and Nutrition Examination Surveys (1999–2016) with a sample of 9,000 NHW and 3,913 NHB men aged 20 years or older. To estimate the association between the prevalence of obesity (BMI ≥30) and race, we applied modified Poisson regression; to explore and decompose racial differences, we used Oaxaca–Blinder decomposition (OBD). We found that NHW had higher abdominal obesity (WC ≥102) than NHB, but NHB were more likely to be obese (BMI ≥30) during most years, with some fluctuations. Modified Poisson regression showed that NHB had a higher prevalence of obesity (prevalence ratio [PR]: 1.11, 95% confidence interval [CI] = [1.04, 1.18]) but lower abdominal obesity (PR: 0.845; 95% CI = [0.801, 0.892]) than NHW. OBD showed that age, access to health care, smoking, and drinking contributed to the differences in abdominal obesity. The study identifies a significant increase in obesity among men over the last two decades; generalized obesity (based on BMI) was more problematic for NHB men, but abdominal obesity was more problematic for NHW men.

Introduction

Obesity is a multifactorial, chronic disease that has a broad and comprehensive effect on health (Apovian, 2016). Described as a nationwide pandemic, the obesity prevalence in the United States has increased drastically in the last two decades; between 1999 and 2017, the prevalence of severe obesity increased sharply from 4.7% in 1999 to 9.2% in 2017 Centers for Disease Control and Prevention (CDC; 2022b). In 2020, over 41.9% of American adults ages 20 and over were considered obese, and 9.2% of adults were considered severely obese (Stierman et al., 2021). Additional projections estimate that by 2030, one in two adults will be classified as obese, and one in four will be severely obese (Hales et al., 2020; Petersen, 2019; Ward et al., 2019; Zimmerman, 2011). Similarly, obesity rates among men have increased significantly, from 21.8 % in 1997 to 35.5 % in 2016 (World Data Atlas, 2016). Based on data gathered between 2017 and 2020, approximately 43.1% of Non-Hispanic White (NHW) men and nearly 40.4% of Non-Hispanic Black (NHB) men were classified as obese (body mass index [BMI] ≥30 kg/m²). During this time, 6.8% of NHW and 7.9% of NHB suffered from severe obesity—defined as a BMI ≥40 kg/m² (National Health Statistics Reports [NHSR], 2021).

These disparities are due in part to confluences in risk factors for obesity, such as that between race, education, and health behaviors. Since 1999, racial disparities have widened between NHB and NHW men and suggest the presence of differential effects in structural determinants and discrimination, correlating with an increase in obesity among NHB versus NHW men (Flegal et al., 2016). In 1999, the racial difference in obesity between NHB and NHW men was 0.6%. By 2007, the prevalence rates were 31.9% and 37.2%, respectively, resulting in a 5.3% increase in the differences between NHB and NHW men. They reduced to 3.3% in 2014, and the prevalence of obesity accelerated for NHW in 2016 (Fryar et al., 2018). Although racial inequality between NHB and NHW still exists, there have been some improvements over time that may have affected racial disparities in obesity. For instance, in 2008, the average household income for Black individuals increased by 14.1%. In addition, in 2019, the poverty rate for Blacks was reported as 18.8%, a historical low since the Census Bureau first released poverty statistics in 1959. These reductions in disparities can be attributed to progress by racial groups that have historically faced disadvantages compared with the majority racial group.

The severity of obesity is worsening as the prevalence of obesity increases. BMI measures body fat using height and weight. BMI is used to classify obesity. Class I obesity is defined as having a BMI between 30.0 and 35 kg/m², Class II obesity as having a BMI between 35.0 and 40 kg/m², and Class III obesity as having a BMI of 40.0 or higher kg/m². Class III obesity is sometimes categorized as “severe” obesity (CDC, 2022a). More than 7.7% of adults aged 20 years or older suffer from severe obesity, with a higher percentage among women (9.7) than among men (5.6) (Fryar et al., 2018).

The literature regarding racial disparities and obesity is mixed. Some studies have documented a higher prevalence of obesity among NHBs. Researchers argue that the following are associated with a higher prevalence of obesity among NHB: (a) lower socioeconomic status (SES; Krueger & Reither, 2015; McLaren, 2007), (b) higher income inequality (Zhang & Wang, 2004), (c) higher poverty rates and lower income levels (Wang & Beydoun, 2007), and (d) reduced access to health care (Mylona et al., 2020). Since access to health care also appears to influence racial disparities in obesity, NHB men with limited health care access have a higher prevalence of obesity than their NHW counterparts (Hill et al., 2023). However, other studies have shown that NHW experienced a higher level of obesity than NHB. For example, Ogden et al. (2014) documented that, in 2011 to 2012, the prevalence of obesity among Hispanic men (40.1%) exceeded NHW (32.4%) and NHB (37.1%) for the first time (Ogden et al., 2014). The most recent CDC publication did not report significant differences between NHW and NHB adult men (Stierman et al., 2021). Race is not the only predictor of obesity. Health behaviors such as diet, physical activity (PA), smoking, and alcohol consumption are also critical factors in racial disparities in obesity (Beydoun et al., 2016; Min et al., 2021; Shaikh et al., 2015; Traversy & Chaput, 2015).

NHBs are significantly more likely than NHWs to live in a geographic area that has limited access to grocery stores and farmers’ markets (Testa & Jackson, 2019). Flegal et al. (2016) also demonstrated a strong correlation between physical inactivity and obesity—especially among NHB men. Although NHB youth display higher PA levels than NHW youth, this trend shifts in adulthood (Belcher et al., 2010). NHB adults exhibit lower levels of non-work PA than NHW (Saffer et al., 2013). Regarding smoking, research has indicated that current smokers typically have lower BMI than non-smokers but higher waist circumference (WC), suggesting a redistribution of body fat (Chiolero et al., 2008). Interestingly, racial differences in the smoking-BMI/WC relationship have been reported, with NHB men showing less of a decrease in BMI with smoking than in NHW men (Frisco et al., 2019).

Evidence on the relationship between alcohol consumption and obesity measures has been mixed, with some studies suggesting a positive correlation; other studies indicate a negative or non-linear association (O’Keefe et al., 2014; Sayon-Orea et al., 2011; Traversy & Chaput, 2015; Yeomans, 2010). Research indicates that NHB men tend to have lower alcohol consumption levels than their NHW counterparts (Jackson et al., 2015). However, NHB men consume more sugar-sweetened beverages than their NHW counterparts (Ford et al., 2016).

The extensive body of literature on obesity disparities has yielded a wealth of insights; however, several gaps in the current literature warrant further investigation (Fryar et al., 2018; Lofton et al., 2023; Stierman et al., 2021). Although the existing literature has shed light on the role of individual behaviors and certain socio-contextual factors in obesity disparities, the studies have primarily examined them in isolation, potentially overlooking the synergistic or counteractive effects they may have on each other and obesity outcomes. According to the WHO model of the social determinants of health, health is influenced by a complex interplay of individual, socio-contextual, and structural factors. These include individual behaviors, social and community networks, living and working conditions, as well as broader social, economic, and political mechanisms (WHO, 2010). In addition, existing research on racial disparities in obesity has often treated race as an isolated variable. Still, this conceptual framework underscores the importance of considering how racial identities intersect with socio-contextual and structural determinants of health to influence obesity outcomes.

Another critical gap in the literature is the limited consideration of different obesity markers: BMI and WC.

Body Mass Index

Research has predominantly utilized BMI as the primary indicator of obesity despite its inherent limitations (Palaniappan et al., 2011). BMI does not distinguish between lean and fat mass and may not accurately represent obesity among different racial groups due to variations in body composition (Flegal et al., 2016; Prentice & Jebb, 2001). Some studies suggest that Black individuals may have a higher muscle mass than Whites with the same BMI, potentially leading to overestimating obesity prevalence among Blacks when using BMI alone (Burkhauser & Cawley, 2008; Wagner & Heyward, 2000). Furthermore, BMI fails to account for the differential health risks associated with visceral adiposity. Several studies have suggested that certain racial groups, including African Americans, may have lower visceral adiposity at the same BMI as their White counterparts (Carroll et al., 2008; Katzmarzyk et al., 2011). This differential distribution of body fat can impact the risk of various health conditions such as Type 2 diabetes, hypertension, and cardiovascular disease. Therefore, relying solely on BMI to assess obesity and associated health risks may lead to underestimating or overestimating these risks among certain racial groups.

Waist Circumference

Conversely, WC has been identified as a valuable measure for evaluating visceral adiposity and associated health risks (Ashwell et al., 2012). Despite its recognized utility, WC is less frequently used in studies examining racial disparities in obesity. WC and BMI should be examined concurrently to yield a more comprehensive understanding of the racial disparities in obesity.

Underrepresentation of men, particularly NHB men, is a significant limitation in obesity research. Most studies have explored these disparities in general populations or have focused disproportionately on women (Thorpe et al., 2015). One of the reasons for this, particularly in terms of Black men and their representation, is that the obesity prevalence is highest (57.9%) among Black women (NHSR, 2021). Although the investigations among women are undoubtedly critical, given their unique health concerns, the relative scarcity of similar research focusing specifically on men signifies a substantial gap in our understanding. The prevalence of obesity in men 20 years and older increased 1.56 times from 27.5% in 1999 to 2000 to 43.0% in 2017 to 2018 (Fryar et al., 2018). At the same time, one study has shown that men—particularly those from NHB populations—tend to exhibit higher obesity rates and related health complications than their NHW counterparts (Thorpe et al., 2015).

Furthermore, men have been found to engage in certain health behaviors, such as smoking and alcohol consumption, at higher rates than women, potentially influencing their obesity risk (Pinkhasov et al., 2010). In addition, men are generally less likely to seek preventive health care and may have different experiences and exposures to socio-contextual and structural determinants of health than women (Courtenay, 2010). The intersection of these health behaviors with race among men remains relatively understudied.

Finally, using the econometric technique such as Blinder–Oaxaca decomposition allows us to decompose the racial gap in health outcomes into how much can be “explained” by racial differences and how much remains “unexplained.” The “explained” portion of the gap identifies the attributability of the difference in mean predicted outcomes between two groups to variations in the levels of observable variables (e.g., SES) as opposed to the differential effects—unexplained—of these variables (e.g., discrimination). Studies have used Oaxaca–Blinder decomposition (OBD) analyses to identify the attributability of obesity risk factors to gender, socioeconomic, and racial differences (Sen, 2014; Taber et al., 2016a, 2016b). However, the Oaxaca–Blinder technique remains underutilized as an empirical method for assessing racial disparities in obesity, especially among Black and White, non-Hispanic adult men (Isong et al., 2018; Kelishadi et al., 2018; Sen, 2014). A few of the above-mentioned studies used this method to analyze racial disparities in obesity based on BMI but not WC. It is crucial to understand how much variance between NHW and NHB could be explained by OBD when it comes to WC.

The present research aims to bridge these gaps in the existing literature and technique by using the most comprehensive dataset between 1999 and 2016. The study sheds light on racial obesity disparities between NHB and NHW men by exploring how demographic, socioeconomic factors, access to health care, and health behaviors correlate with racial differences in BMI and WC. This research adds to the growing body of evidence pointing toward the need for comprehensive, multi-level strategies to combat racial health disparities in the United States.

Method

Study Design and Settings

The present study is a secondary analysis of data collected by the National Health and Nutrition Examination Survey (NHANES), which is publicly available data from the National Center for Health Statistics. NHANES provides national estimates of the health and nutritional status of the U.S. population. We adopted the WHO’s social determinants of health to develop this study (See Appendix 1).

Data Collection Method

Since 1999, NHANES data have been collected every 2 years, with an average response rate of 73.2% between 1999 and 2016 (NHANES, 2018; Zipf et al., 2013). The NHANES combined in-person interviews with standardized physical examinations and laboratory tests. Participants represent counties across the United States and all four regions, including metropolitan and nonmetropolitan areas. NHANES published detail-oriented documents on the sample design, estimation, and analytic guidelines (Akinbami et al., 2022).

Study Population and Sampling

The study population included 10,184 NHW and 4,888 NHB men 20 years and older who had participated in the NHANES between 1999 and 2016—for 15,072 observations over 18 years. After excluding missing observations, we ran an analysis with a sample of 9,000 NHW and 3,913 NHB. The analytical sample included NHW and NHB men 20 years and older—12,913 participants.

Measures

Outcome Variables

The dependent variables of interest were BMI (kg/m²): calculated as the ratio of body weight in kilograms (kg) to height in meters squared (m²)) and WC in centimeters. NHANES reported that:

Trained health technicians collected Body measures data in the Mobile Examination Center (MEC). Standing height was measured using a portal stadiometer. Weight was measured using a portal digital weight scale. Arm length and arm, waist, and calf circumferences were measured using a steel measuring tape. Calf, triceps, and subscapular skinfolds were measured using a skinfold caliper. Arm and calf measurements were made on the right side of the body (NHANES, 2013).

Control Variables

We used various covariates, including demographic and SES, health care access, and health behaviors. For the demographic and SES variables, we included age (years), marital status (1 = married, 0 = otherwise, including divorced, separated, widowed, or never married), educational attainment (less than high school, high school, or general equivalency diploma, more than high school), and the ratio of family income to poverty (<1, 1≤ ratio <2, 2≤ ratio <3, 3≤ ratio <5, and ratio ≥5). The proportion of family income to poverty (PIR) refers to different income categories relative to the poverty threshold. The PIR was calculated by dividing family (or individual) income by the poverty guidelines specific to each survey year. For example, for a family of four with three children younger than 18 years of age with an annual income of US$81,700 and a poverty threshold of US$27,575 (U.S. Census Bureau, 2021), the ratio would be (US$75,000/US$27,575 = 2.96~3). A ratio less than one meant that the income was less than the poverty level.

To control for health care access, we included two sets of variables: a categorical variable for type of health insurance (Medicare, Medicaid, private insurance, governmental health insurance, and any other health insurance) and a binary variable indicating that the respondent had a routine place to go for health care. We controlled for three health behavior variables: smoking (never, former, or current smoker), drinking (never, former, or current drinker), and engaging in vigorous or moderate physical activities (1, if any rigorous activities; 0, if no rigorous activities). These were the self-reported responses by respondents. The NHANES defined rigorous activities as at least 60 minutes of PA per day (NHANES, 2016).

Statistical Analysis

The present study assesses the relationship between obesity and race in NHW and NHB men 20 years and older. To address the study’s aim, we ran three sets of analyses. First, we used descriptive analyses to compare BMI and WC in NHW and NHB. We also compared demographics, SES, access to a health center, health insurance status, and health behavior variables. In addition, using the CDC classification, we compared the obesity categories (0, BMI < 30; 1, 30≤ BMI <35; 2, 35≤ BMI <40; and 3, BMI ≥40) between NHW and NHB (CDC, 2022a). We used independent t-tests and chi-square tests to compare variables between NHW and NHB. Second, to assess the association between obesity and race, we ran two sets of weighted modified Poisson regression analyses, controlling for race, age, marital status, education, the ratio of family income to poverty category, type of health insurance, a routine place to go for health care, smoking, and drinking status, engaging in vigorous or moderate physical activities, and year dummy.

For the first set, using BMI, we created a binary variable to identify participants who were obese (if BMI ≥30) as the outcome variable (National Institutes of Health [NIH], 2020). For the second set of analyses, using WC, we created a binary variable to identify participants who were obese (if WC ≥102) as the secondary outcome variable (WHO, 2011).

In our sample, the prevalence of obesity was >10%; therefore, the weighted modified Poisson regression analysis stands as the best-fit model (McNutt et al., 2003; Thorpe et al., 2017; Zou, 2004). We reported the prevalence ratio (PR) and corresponding 95% confidence interval (CI) (McNutt et al., 2003; Thorpe et al., 2017; Zou, 2004). Finally, to decompose the race differences in WC and BMI, we utilized the OBD to measure the contributions of age, SES, health care access, and health behaviors on BMI and WC. We reported the overall difference in adjusted means between the two groups (here, NHB and NHW). Then we noted the contributions of each BMI and WC risk factor (separately) to measure their temporal influences.

We controlled all models for year dummies, age, having access to health care (we included two sets of variables: health insurance and a routine place to go for health care), and health behavior variables by adding smoking status, drinking status, and engaging in vigorous or moderate physical activities.

All analyses were weighted using the NHANES individual-level sampling weights for 1999–2016 (NHANES, 2019). We considered all p-values <0.05 as statistically significant; all tests were two-sided. All statistical procedures were performed using STATA statistical software, version 15.

Ethics

This study was approved by the Institutional Review Board Office of the Johns Hopkins Bloomberg School of Public Health. We used all publicly available data, and the present study does not qualify as human subjects research as defined by DHHS regulations 45 CFR 46.102.

Results

Descriptive Analysis

Table 1 displays the weighted means and standard deviations for all variables used in the models. Overall, the sample age was 47.0 years; the majority of the men in the sample were married (68%), educated with a college degree (61%), and in the middle class or higher, with a poverty-to-income ratio (3.3, SD: 0.35). Eighty-four percent of the study population had health insurance coverage; 25% were current smokers, 82% were current drinkers, and 63% were physically active. About 12% of the population was NHB, and 88% was NHW.

Table 1.

Characteristics of Males 20 Years and Older by Race in the National Health and Nutrition Examination Survey (NHANES) 1999 to 2016

	All (n = 12,913)		NHW (n = 9,000)		NHB (n = 3,913)
	Weighted mean	SD	Weighted mean	SD	Weighted mean	SD	p Value
Dependent variables
Waist circumference (cm)	101.63	(11.75)	102.17	(10.23)	97.87	(20.75)	0.000
Body mass index (BMI)	28.55	(4.41)	28.53	(3.85)	28.63	(7.88)	0.490
Obesity categories
BMI <30 (p = .064)	0.67	(0.39)	0.67	(0.32)	0.65	(0.56)	0.065
BMI: 30–34.9 (p = .019)	0.22	(0.3.4)	0.22	(0.28)	0.20	(0.47)	0.020
BMI: 35–39.9 (p = .0021)	0.07	(0.21)	0.07	(0.17)	0.09	(0.34)	0.002
BMI ≥40 (p = .000)	0.04	(0.16)	0.04	(0.13)	0.06	(0.29)	0.000
% Of the obese population
BMI ≥30	33.2	(35.7)	33.0	(31.8)	35.0	(56.4)	0.065
WC ≥102	46.2	(37.8)	47.5	(33.8)	36.8	(57.0)	0.000
Socioeconomic
Age (years)	46.95	(12.50)	47.45	(11.19)	43.41	(18.43)	0.000
Marital status
Married	0.68	(0.35)	0.70	(0.31)	0.53	(0.59)	0.000
Education level—adults 20
Less than high school	0.14	(0.26)	0.12	(0.22)	0.26	(0.52)	0.000
High school	0.25	(0.33)	0.25	(0.29)	0.28	(0.53)	0.009
College	0.61	(0.37)	0.63	(0.33)	0.46	(0.59)	0.000
Poverty index
The ratio of family income to poverty	3.30	(1.20)	3.42	(1.05)	2.47	(1.87)	0.000
Poverty index categories
Index <1	0.10	(0.23)	0.08	(0.19)	0.22	(0.49)	0.000
1≤ Index <2	0.17	(0.28)	0.16	(0.25)	0.25	(0.51)	0.000
2≤ Index <3	0.15	(0.27)	0.15	(0.24)	0.18	(0.45)	0.000
3≤ Index <5	0.27	(0.34)	0.28	(0.30)	0.21	(0.48)	0.000
Index ≥5	0.31	(0.35)	0.34	(0.32)	0.15	(0.42)	0.000
Race
NHW	0.88	(0.25)	1.00	0.00	NA	0.00
NHB	0.12	(0.25)	NA	0.00	1.00	0.00
Health insurance coverage and health care
Has health insurance coverage	0.84	(0.28)	0.86	(0.24)	0.71	(0.53)	0.000
Type of health insurance
Medicare	0.11	(0.26)	0.13	(0.23)	0.08	(0.32)	0.000
Medicaid	0.04	(0.15)	0.03	(0.11)	0.09	(0.33)	0.000
Private insurance	0.52	(0.41)	0.56	(0.34)	0.44	(0.59)	0.000
Governmental health	0.04	(0.17)	0.04	(0.14)	0.05	(0.26)	0.249
Any other health insurance	0.08	(0.22)	0.09	(0.20)	0.05	(0.26)	0.000
Health insurance—missing	0.00	(0.04)	0.00	(0.03)	0.01	(0.10)	0.000
A routine place to go for healthcare	0.83	(0.29)	0.83	(0.25)	0.80	(0.48)	0.000
Smoking behavior
Never	0.46	(0.38)	0.45	(0.34)	0.51	(0.59)	0.000
Former	0.30	(0.35)	0.31	(0.31)	0.17	(0.45)	0.000
Current	0.25	(0.33)	0.24	(0.29)	0.32	(0.55)	0.000
Drinking behavior
Never	0.07	(0.19)	0.06	(0.17)	0.11	(0.38)	0.000
Former	0.06	(0.19)	0.06	(0.16)	0.10	(0.35)	0.000
Current	0.82	(0.29)	0.84	(0.25)	0.72	(0.53)	0.000
Has vigorous or moderate activities	0.63	(0.37)	0.64	(0.32)	0.57	(0.59)	0.000

Note. (1) All public insurance, including Medicaid, SCHIP, Military Health Care, Indian Health Service, state-sponsored health plans, government insurance, and single service plans.

(2) We used standard weight status categories associated with BMI published by CDC. https://www.cdc.gov/healthyweight/assessing/bmi/adult_bmi/index.html

(3) We used questions 605, 620, and 650 from the NHANES to create a vigorous or moderate activity. The questions asked if participants were involved in “any vigorous-intensity activity that causes large increases in breathing or heart rate like carrying or lifting heavy loads, digging or construction work for at least 10 minutes continuously? (Q605),” or “moderate-intensity activity that causes small increases in breathing or heart rate such as brisk walking or carrying light loads for at least 10 minutes continuously? (Q620),” or “vigorous-intensity sports, fitness, or recreational activities that cause large increases in breathing or heart rate like running or basketball for at least 10 minutes continuously? (Q650)” (NHANES, 2016).

(4) We ran an independent t-test to compare means between White versus Black men; all means were significantly different at p < .001.

(5) The proportion of family income to poverty refers to different income categories relative to the poverty threshold, with “<1” indicating income below the poverty line, “1≤ ratio <2” representing near-poverty income, “2≤ ratio <3” denoting moderate income, “3≤ ratio <5” reflecting higher income, and “ratio ≥5” indicating significantly higher income levels that are at least five times the poverty threshold. NHW = non-Hispanic White; NHB = non-Hispanic Black; NHANES = National Health and Nutrition Examination Survey; BMI = body mass index; WC = waist circumference; NA = not applicable.

Source: NHANES data, 1999–2016.

Comparing sociodemographic variables by race showed that the NHW population was older (p < .001), with an average age of 47.4 (SD: 11.2), compared with NHB, with an average age of 43.4 (SD: 18.4); 63% of NHW had college degrees compared with 46% in NHB (p < .001). The mean poverty index was 3.42 and 2.47 for NHW and NHB, respectively, depicting higher income for NHW (p < .001). Eighty-six percent of NHW participants had health insurance coverage versus 71% of NHB (p < .001). Eighty-three percent of NHW had a routine place to go for health care compared with 80% of NHB (p < .001). Twenty-four percent of the NHW were current smokers compared with 32% of NHB (p < .001), and 84% of the NHW were current drinkers compared with 72% of NHB (p < .001). Finally, NHW were more likely to be physically active than NHB (p < .001).

Comparing the WC and BMI between NHW and NHB indicated that NHB men had a lower WC of 97.9 than NHW men (97.9 vs. 102.2, p = .000), with no significant differences in BMI (28.53 vs. 28.63, p = .490); however, comparing the prevalence of obesity using BMI (BMI ≥30) and WC (WC ≥102), we found a slightly higher prevalence of obesity in NHB men than in NHW (35% vs. 33%, p = .065) and lower abdominal obesity in NHB than in NHW (36.8% vs. 47.5%, p < .001) (see Table 1 for more details.)

Comparing the Trend of BMI and WC Between NHW and NHB

Figure 1 displays the BMI and WC trends between NHW and NHB men 20 years old and above. In Panel A, the red line plots the BMI distribution in NHW men, as the weighted average BMI increased by 1.4 kg/m² from 25.8 kg/m² in 1999/2000 to 27.2 kg/m² in 2015/2016. The blue-dash line highlights the distribution in NHBs, with a sharper increase (1.7 kg/m²) from 24.8 kg/m² to 26.5 kg/m² in 2016. In Panel B, we compared the weighted average of WC between NHW and NHB. Similar to the BMI, the average WC in NHW (solid red line) increased from 92.7 cm in 1999/2000 to 96.6 cm in 2015/2016, with a 3.9 cm increase. The average WC in NHB (blue-dash solid line) increased from 84.4 cm in 2000 to 89.4 cm in 2016, with a 5 cm increase. Comparing the prevalence of obesity using the BMI ≥30 and WC ≥102 showed that NHB men experienced a higher prevalence of obesity (BMI ≥30) than NHW between 2004 and 2014; the situation reversed in 2016. However, we observed higher abdominal obesity (WC ≥102) in NHW (see Figure 1C and D.)

Figure 1.

Comparing BMI and WC Trends in U.S. Adult Men 20 Years Old and Above by Race (1999–2016)

Note. BMI = body mass index; WC = waist circumference.

Figure 2 compares the trends of obesity. Panel A shows the trend of Class I obesity (BMI: 30.0–35 kg/m²). As presented, NHW experienced a higher prevalence of Class I obesity by moving to Panel B (Class II; BMI: 35–40 kg/m²); we see a mixed trend for NHW and NHB. We observe a clear pattern in Class III obesity; NHBs experienced a higher prevalence of severe obesity (BMI >40 kg/m²). Overall, all types of obesity increased for NHWs and NHBs between 1999 and 2016 (see Figure 2).

Figure 2.

Comparing Obesity Class Trends in U.S. Adult Men 20 Years Old and Above by Race (1999–2016)

Estimating the Association Between Obesity in NHW and NHB

Table 2, Column 1, presents the results of the modified Poisson model by using the BMI (BMI ≥30). As presented, NHB had 11.4% times (PR: 1.111, CI = [1.043, 1.183]) more obesity than their NHW counterparts. Column 2 displays the association between race and obesity using the WC outpoint (WC ≥102). Interestingly, using the WC ≥102 to measure abdominal obesity, we observed that NHB experienced lower abdominal obesity (PR: 0.845, CI = [0.801, 0.892]). Considering both models, age, being married, having a higher poverty-to-income ratio, having a routine place to receive health care, being a former drinker, and being a smoker were positively associated with obesity. Having higher education, being under the coverage of Medicare, and engaging in vigorous PA were negatively associated with obesity.

Table 2.

Modified Poisson Regression Analyses to Show the Association Between Race and Obesity in Men 20 Years and Older Across Racial Groups in U.S. Adults, 1999 to 2016

	Column 1: if BMI ≥ 30		Column 2: if WC ≥ 102
	PR	95% CI	PR	95% CI
Race
Non-Hispanics Black (ref. NHW)	1.111**	[1.043, 1.183]	0.845***	[0.801, 0.892]
Age (years)	1.001	[0.998, 1.003]	1.012***	[1.010, 1.014]
Married	1.102**	[1.028, 1.182]	1.087**	[1.028, 1.150]
Education (ref. college)
Less than high school	0.894	[0.798, 1.001]	0.900**	[0.837, 0.968]
More than high school	0.948	[0.872, 1.031]	0.932*	[0.882, 0.986]
The ratio of family income to poverty categories (ref. ≥5)
Index <1	0.968	[0.853, 1.100]	0.936	[0.852, 1.027]
1≤ Index<2	1.08	[0.970, 1.203]	0.997	[0.926, 1.072]
2≤ Index<3	1.093	[0.985, 1.213]	0.991	[0.912, 1.077]
3≤ Index<5	1.157**	[1.059, 1.265]	1.047	[0.978, 1.121]
Type of health insurance (ref. private insurance)
Medicare	0.815***	[0.735, 0.903]	0.878***	[0.819, 0.942]
Medicaid	1.098	[0.925, 1.304]	1.043	[0.922, 1.181]
Governmental health	1.045	[0.915, 1.193]	1.04	[0.941, 1.150]
Any other health insurance	1.057	[0.930, 1.200]	1.022	[0.931, 1.121]
No coverage	0.902*	[0.823, 0.989]	0.953	[0.873, 1.040]
Routine place to go for healthcare	1.313***	[1.192, 1.446]	1.293***	[1.192, 1.402]
Smoking behavior
Former	1.126***	[1.050, 1.206]	1.130***	[1.075, 1.188]
Current	0.779***	[0.708, 0.857]	0.843***	[0.785, 0.905]
Drinking behavior
Former	1.276***	[1.117, 1.456]	1.049	[0.950, 1.159]
Current	1.003	[0.896, 1.124]	0.936	[0.859, 1.019]
Has vigorous or moderate activities	0.773***	[0.721, 0.828]	0.792***	[0.751, 0.835]

Note. Models controlled for a dummy variable for years (1999–2016, nine waves). The proportion of family income to poverty refers to different income categories relative to the poverty threshold, with “<1” indicating income below the poverty line, “1 ≤ratio <2” representing near-poverty income, “2≤ ratio <3” denoting moderate income, “3≤ ratio <5” reflecting higher income, and “ratio ≥5” indicating significantly higher income levels that are at least five times the poverty threshold. BMI = body mass index; WC = waist circumference; PR = prevalence ratio; CI = confidence interval; NHW = non-Hispanic White; NHANES = National Health and Nutrition Examination Survey.

Source: NHANES data, 1999–2016.

^*p < .05. **p < .01. ***p < .001.

Decomposing Differences in BMI

Table 3, Panel A, compares the differences in BMI between NHB and NHW men. As presented, using the average weighted mean of BMI, we did not find any significant difference in BMI between NHB men and NHW. By looking at the control variables, we observed that if NHB increased their levels of drinking behavior as NHW, they would have a higher BMI of 0.156 (p < .001). If NHB increased their PA to the same level as NHW, their BMI would be reduced by −0.099 (p < .001). Controlling by SES, health behaviors, and physical activities, the total unexplained race disparity was 0.253, and the total unexplained was −0.354.

Table 3.

Oaxaca Decomposition Estimates to Show the Impact of SES and Health Behaviors on Waist and Body Mass Index in NHW, NHB, Men 20 Years Old and Above, NHANES: 1999–2016

	Model 3 (n = 12,913)
	Coefficient	95% CI
Panel A: BMI
Differential
NH White	28.533***	[28.380, 28.687]
NH Black	28.635***	[28.373, 28.896]
Difference 1	−0.102	[−0.398, 0.194]
Explained by the difference in measured risk factors distribution (endowment)
Year 2	−0.015	[−0.056, 0.025]
Age 3	0.042	[−0.016, 0.100]
SES 4	0.157*	[0.022, 0.292]
Health access 5	0.011	[−0.080, 0.102]
Smoking and drinking 6	0.156***	[0.083, 0.230]
Physical activity 7	−0.099***	[−0.142, –0.055]
Total explained race disparity	0.253**	[0.092, 0.414]
Total unexplained race disparity	−0.354*	[−0.662, –0.047]
Panel B: waist
Differential
NH White	102.170***	[101.748, 102.592]
NH Black	97.868***	[97.191, 98.545]
Difference 1	4.302***	[3.528, 5.076]
Explained by the difference in measured risk factors distribution (endowment)
Year 2	−0.044	[−0.140, 0.052]
Age 3	0.785***	[0.569, 1.002]
SES 4	0.335*	[0.012, 0.658]
Health access 5	−0.048	[−0.292, 0.197]
Smoking and drinking 6	0.342***	[0.176, 0.508]
Physical activity 7	−0.303***	[−0.428, –0.177]
Total explained race disparity	1.069***	[0.615, 1.522]
Total unexplained race disparity	3.233***	[2.437, 4.030]

Note. CI = confidence interval; BMI = body mass index; SES = socioeconomic status.

Source: NHANES data, 1999–2016.

¹Difference equals the sum of the total explained, and total unexplained equals 0.

²Year: reports summed contributions for a dummy variable for years (1999–2016, nine waves).

³Age: reports summed contributions for age.

⁴SES reports summed contributions for marital status, education, and poverty.

⁵Healthcare access: reports summed contributions for Medicare, Medicaid, private insurance, governmental health, any other health insurance, and a routine place to go for health care.

⁶Smoking and drinking: reports summed contributions for smoking and drinking behaviors

⁷Physical activity; reports contribution for having vigorous or moderate physical activities.

^*p < .05. **p < .01. ***p < .001.

Decomposing Differences in WC

Table 3, Panel B, displays the results of the OBD model for WC. As reported, NHB men had lower WC by 4.3 cm compared with NHW men. The most significant contributors to explain the black-white differences in WC were age (0.785 cm; p < .001), SES (0.335 cm; p < .001), smoking and drinking behavior (0.342; p < .001), physical activities (−0.303; p < .001), college education, higher family income, and access to health insurance. Notably, certain predictor variables such as “marital status” actually detract—that is, literally subtract—from the “explained gap.”

After balancing each other, the contributions of these factors could explain 1.069 cm of the differences between NHW and NHB. The total unexplained race disparity was 3.233 cm (p < .001).

Discussion

In this study, we explored the association between BMI and WC in NHB and NHW men, using a large, nationally representative data set (n = 12,913). We examined the association between obesity and race using modified Poisson regression analyses. We used the OBD to decompose racial differences in BMI and WC. The obesity prevalence was measured using the CDC and the WHO guidelines. Men with a BMI and WC greater than 30 and 102 cm were considered obese.

The results of modified Poisson regression showed that NHB had a higher PR of being obese using BMI >30 (PR: 1.111, 95% CI = [1.043, 1.183]) but had lower abdominal obesity measured by WC ≥102 with a PR of 0.845 (95% CI = [0.801, 0.892]). Notably, NHB had a higher obesity prevalence on average over the 17 years. Our finding confirms the literature, which suggests that BMI leads to a potential overestimation in obesity prevalence due to differences in body fat and muscle mass, about how BMI is inaccurate and leads to overestimations in NHB compared with NHW (Burkhauser & Cawley, 2008; Wagner & Heyward, 2000). However, we observed higher abdominal obesity (WC ≥102) in NHW, which has been highlighted by other studies (Katzmarzyk et al., 2010, 2011). It suggests that WC has an advantage in measuring abdominal obesity in NHW. (Sun et al., 2022). To control and measure obesity in the U.S. population, combining two obesity metrics (BMI and WC), using the econometrics technique, such as OBD, can help policymakers identify high-risk populations to reduce obesity-connected disease and all-cause mortality.

Using OBD, our results showed no significant differences in BMI between NHW and NHB, which aligns with the most recent CDC report (Stierman et al., 2021); however, there were differences for WC. Our findings showed that NHBs are more likely to have lower WC than NHWs even when they are the same age and have the same SES. Our results support the literature that confirms the link between health behaviors and obesity. We found that if NHBs adopted drinking behaviors similar to NHW men, they would experience an increase in WC. Given the relationship between PA and obesity prevention, if NHB level their PA to NHW, they would experience a reduction in WC. (Wolfenden et al., 2020; Zare et al., 2021).

We found different results using WC and BMI to measure obesity prevalence among NHB and NHW men; some differences may be due to measurement bias. BMI is commonly used as a metric to define obesity; however, it fails to account for abdominal adiposity (Ross et al., 2020). BMI cannot distinguish between lean body mass, body fat, or bone density. This limitation contributes to an overestimation of obesity among racial minorities due to racial differences in body composition (CDC, 2009). Studies have demonstrated that White men have a higher abdominal visceral adipose tissue mass than African American men (Katzmarzyk et al., 2010, 2011). Clinicians capture a more comprehensive cardiometabolic risk factor of obesity when supplementing WC with BMI (CDC, 2009).

Lifestyle is highly correlated with obesity. Unhealthy behaviors such as smoking, drinking, and sedentary practices increase the risk of developing obesity (Campbell & Baker, 2019; Pabayo et al., 2018; Zare et al., 2021). Research has shown that alcohol intake is associated with a greater risk of obesity because alcohol increases abdominal fat (Ryu et al., 2010; Traversy & Chaput, 2015). Our study builds on this literature by providing a potential mechanism driving the racial disparities in obesity. First, we identified that NHW men tend to consume more alcohol than NHB men. In our sample, 84% of NHW self-identified as current drinkers compared with 72% of NHB. Second, our study found that NHB men have higher BMI than NHW, but NHB men have less abdominal obesity and smaller WC. Thus, it is likely that the racial differences in obesity may be partly due to lifestyle factors contributing to abdominal obesity. Furthermore, it confirms the importance of using WC and BMI to measure obesity. Simply examining WC or BMI in isolation would not allow us to see the complete picture of obesity.

SES is another predictor of obesity. Minority populations are disproportionately at risk of obesity due to socio-structural factors, including unequal access to quality education, good-paying jobs, quality food, and health care services (Casey et al., 2017; Manduca, 2018; National Academies of Sciences, Engineering, Medicine, 2017; Salazar et al., 2019). NHBs are more likely than NHWs to live in food deserts, resulting in diminished access to healthy foods and fresh produce (Beaulac et al., 2009; Gordon et al., 2011).

Although social epidemiology has presented compelling arguments to suggest that obesity is strongly determined by socio-structural factors—as opposed to simply differences in individual health behaviors and genetics—most public health interventions for obesity rely on promoting individual-level clinical interventions and lifestyle adjustments without considering the social basis. In addition to SES factors, food culture is another critical indicator; it influences favoriting specific foods, eating patterns, and behaviors (Williams et al., 2012). Eating foods that contain high fat and sugar contributes to weight gain and obesity in Black families (Qobadi & Payton, 2017).

Limitations

A few aspects of this study need explanation. First, we could not determine the potential influence of neighborhood effects on differences in BMI and WC. For this analysis, we used publicly available data, and the datasets did not provide any geographical variables to control for geographical disparities. It would have been beneficial to control neighborhood data in the study since geographic locales are important determinants of access to opportunities that facilitate socioeconomic attainment and health.

There is a limitation to note with OBD. For example, OBD combines the effects of a few variables. We recognize that some readers may want to see the specific contribution of each variable. OBD also uses the counterfactual exercise to aggregate decomposition, which is a limitation (Słoczyński, 2020).

Finally, the use of BMI and WC as definitions of obesity have limitations. BMI does not distinguish between muscle, bone mass, and excess fat. BMI alone is not an adequate biomarker of abdominal obesity as it falls short of accurately capturing cardiometabolic risk (Ross et al., 2020). The fundamental limitation of WC appears to be its inability to distinguish between visceral and subcutaneous fat (Gurunathan & Myles, 2016). In addition, body composition changes with age, sex, and race, and there is not enough normative data to define obesity for each group (Gurunathan & Myles, 2016). Finally, studies have reported that food culture and unhealthy eating behaviors play a role in developing obesity, but we were not able to control in this study (Qobadi & Payton, 2017; Williams et al., 2012).

Strengths

Despite its methodological limitations, the present study has several strengths. We significantly contribute to the literature on obesity and race. Our study is among the first to examine obesity, specifically in NHB and NHW men. We utilized data spanning 18 years, from 1999 to 2016, from the NHANES. It is a nationally representative and generalizable dataset that uses standardized and calibrated measures for determining BMI and WC.

In addition, we employed modified OBD models controlled for age, marital status, educational attainment, the ratio of family income to poverty, health coverage, and health behaviors, including smoking, drinking, and PA. This allowed us to decompose the differences between NHB and NHW men and their differential effects. Our findings ultimately support the conclusion that disparities in obesity risk factors vary across racial groups and that addressing the consequent health inequities requires targeted interventions on both the structural, socio-contextual, and individual levels to ameliorate the disparities related to SES, education, and health care access, as well as the health behaviors including smoking, drinking, and PA.

Conclusion

We studied race differences in WC and BMI between NHB and NHW men. The modified Poisson regression showed that NHB had a higher PR of being obsessed measured by BMI ≥30 than their NHW counterparts but lower abdominal obesity measured by WC ≥102. The results of OBD in WC showed that age, access to health care, smoking, and drinking contributed to the differences in WC. The results of OBD showed that if the smoking and drinking behaviors among NHB were level with NHW, then NHB would have a higher BMI of 0.164 and that if NHB achieved the same PA level as NHW, their BMI would be reduced by −0.093. Our findings support addressing health inequities in obesity between NHB and NHW men through targeted interventions at both the structural, socio-contextual, and individual levels by improving health and lifestyle behaviors, reducing socio-contextual barriers, such as access to adequate food and employment resources, and delivering health programs that are accessible, affordable, and culturally adaptable to specific individuals or groups.

Appendix 1.

WHO conceptual framework for social determinants of health.

Source: WHO (2011).