Recent US work identified “metabolically healthy overweight” and “metabolically at risk normal weight” individuals. Less is known for modernizing countries with recent increased obesity.
Recent US work identified “metabolically healthy overweight” and “metabolically at risk normal weight” individuals. Less is known for modernizing countries with recent increased obesity.
Fasting blood samples, anthropometry and blood pressure from 8,233 adults aged 18-98 in the 2009 nationwide China Health and Nutrition Survey, were used to determine prevalence of overweight (Asian cut point, BMI ≥23 kg/m2) and five risk factors (prediabetes/diabetes (hemoglobin A1c ≥5.7%) inflammation (high-sensitivity C-reactive protein (hsCRP) ≥3 mg/l), prehypertension/hypertension (Systolic blood pressure/diastolic blood pressure≥130/85 mm Hg), high triglycerides (≥150 mg/dl), low high-density lipoprotein cholesterol (<40 (men)/ <50 mg/dl (women)). Sex-stratified, logistic, and multinomial logistic regression models estimated concurrent obesity and cardiometabolic risk, with and without abdominal obesity, adjusting for age, smoking, alcohol consumption, physical activity, urbanicity, and income.
Irrespective of urbanicity, 78.3% of the sample had ≥1 elevated cardiometabolic risk factor (normal weight: 33.2% had ≥1 elevated risk factor; overweight: 5.7% had none). At the age of 18-30 years, 47.4% had no elevated risk factors, which dropped to 6% by the age 70, largely due to age-related increase in hypertension risk (18-30 years: 11%; >70 years: 73%). Abdominal obesity was highly predictive of metabolic risk, irrespective of overweight (e.g., “metabolically at risk overweight” relative to “metabolically healthy normal weight” (men: relative risk ratio (RRR) = 39.06; 95% confidence interval (CI): 23.47, 65.00; women: RRR = 22.26; 95% CI: 17.49, 28.33)).
A large proportion of Chinese adults have metabolic abnormalities. High hypertension risk with age, underlies the low prevalence of metabolically healthy overweight. Screening for cardiometabolic-related outcomes dependent upon overweight will likely miss a large portion of the Chinese at risk population.
In the United States, ∼16.3 million adults are of normal weight but have two or more metabolic abnormalities (1). Understanding heterogeneity in risk by overweight status is important particularly in the metabolically at risk but nonoverweight population as these individuals often elude screening as they are not perceived as high risk. However, little is known about heterogeneity in risk in Asian populations, and particularly in China where rates of obesity-related noncommunicable diseases have recently increased and are now the leading causes of morbidity, disability, and mortality (2,3).
Given that Asians are known to have relatively high abdominal obesity (4) and high prevalence of cardiometabolic risk factors at relatively low BMI (5) and at younger ages (6), understanding how abdominal obesity associates with cardiometabolic risk in the absence of (and in combination with) obesity is important. Although abdominal obesity does not confer the same level of risk for hypertension, diabetes, inflammation, and dyslipidemia, many studies rely on syndrome-based indicators (e.g., metabolic syndrome (7,8) and Framingham risk score (9,10)) and therefore may miss important variation that can be observed by investigating variation in co-occurrence of obesity and cardiometabolic risk across individual risk factors.
Understanding the distribution of cardiometabolic risk by age and across overweight and nonoverweight individuals is essential for determining the impending burden of cardiovascular disease in China, home to more than 1.3 billion people and comprising one-fifth of the world's population. The distribution of cardiometabolic risk factors, particularly among the nonoverweight population is particularly important for prevention as these individuals are likely to be missed in screening efforts. To this end, a national Chinese cohort was examined to determine the prevalence of “metabolically healthy overweight” and “metabolically at risk normal weight” by age. Analyses were structured to test the hypothesis that the association between overweight and five risk factors (prediabetes/diabetes (hemoglobin A1c (HbA1c) ≥5.7%) inflammation (high-sensitivity C-reactive protein (hsCRP) ≥3 mg/l), prehypertension/hypertension (systolic blood pressure (SBP)/diastolic blood pressure (DBP) ≥130/85 mm Hg), high triglycerides (≥150 mg/dl), low high-density lipoprotein cholesterol (HDL-C) (<40 (men)/<50 mg/dl (women)) varies, depending upon the indicator. Further analyses tested independent and interactive effects of overweight and abdominal obesity across prehypertension or hypertension, prediabetes or diabetes, high triglycerides, low HDL-C, and inflammation.
The The China Health and Nutrition Survey (CHNS) collected health data in 228 communities (nine diverse provinces (Guangxi, Guizhou, Heilongjiang, Henan, Hubei, Hunan, Jiangsu, Liaoning, and Shandong)) throughout China from 1989 to 2009 with eight rounds of surveys, with the 2009 survey collecting fasting blood for the first time. Using multistage, random cluster sampling, counties in the nine provinces were stratified by income and weighted sampling was used to randomly select four counties in each province. Villages and small towns within counties and urban and suburban neighborhoods within cities were selected randomly into primary sampling units that were politically and geographically classified based on State Statistical Office definitions. The surveyed provinces represent 56% of the Chinese population. Survey procedures have been described elsewhere (11). The study was approved by the Institutional Review Board at the University of North Carolina at Chapel Hill, the China–Japan Friendship Hospital, Ministry of Health and China, and Institute of Nutrition and Food Safety, China Centers for Disease Control and subjects gave informed consent for participation.
A total of 10,038 adult respondents were surveyed at the 2009 exam, 1,423 did not give blood, 24 were not fasting, and 62 were pregnant, resulting in a total of 8,529 individuals with fasting blood samples. Of these, 237 were missing anthropometry and 60 were missing lab results, resulting in an analytic sample of 8,233 with anthropometry and clinical exam data. There were no statistically significant differences in the total 2009 sample vs. the analytic sample in income, urbanization, BMI, cardiometabolic risk, or overweight, although the analytic sample included more women, was slightly younger, and had slightly larger waist circumference.
Cardiometabolic risk. Following an overnight fast, blood was collected by venipuncture (12 ml). Whole blood was immediately centrifuged and serum glucose testing was completed. The remaining plasma and serum samples were immediately frozen and stored for later laboratory analysis. All samples were processed in a National Central Lab in Beijing (medical laboratory accreditation certificate ISO 15189:2007) using automatic analyzer, with strict quality control.
HbA1c was measured in whole blood with an automated glycohemoglobin analyzer by using high-performance liquid chromatography system (model HLC-723 G7; Tosoh, Tokyo, Japan). HbA1c is used as a marker of prediabetes or diabetes (values ≥5.7%) according to an International Expert Committee (12). We also conducted comparative sensitivity analyses using fasting glucose (Supplementary Tables S2 and S3 and Supplementary Figure S1, online). Plasma triglyceride and HDL-C were measured by using the glycerol-phosphate oxidase method, the polyethylene glycol-modified enzyme HDL-C assay by determiner regents (Kyowa Medex Ltd, Tokyo, Japan) on the Hitachi 7600 automated analyzer (Hitachi, Tokyo, Japan). The International Diabetes Federation (IDF) criteria for adults (13) is used for indicators of dyslipidemia risk (high triglycerides: ≥150 mg/dl; low HDL-C: (male: <40 mg/dl; female: <50 mg/dl) or taking lipid lowering medication. hsCRP, using the immunoturbidimetric method with Denka Seiken, Japan reagents (Hitachi 7600 automated analyzer; Hitachi) is used as a marker of inflammation, with values ≥3 mg/dl indicating high risk (14). Elevated hsCRP is a strong independent risk factor for future cardiovascular events, regardless of the underlying condition (15).
SBP and DBP were measured on the right arm, using mercury sphygmomanometers with appropriate cuff sizes. Measures were collected in triplicate after a 10-min seated rest and the mean of the three measurements used in analyses. Risk was determined as SBP ≥130 mm Hg/DBP ≥85 mm Hg according to the IDF cut point for prehypertension or taking blood pressure medication (16).
Overweight and abdominal obesity. Height was measured without shoes to the nearest 0.2 cm using a portable SECA stadiometer, and weight was measured without shoes and in light clothing to the nearest 0.1 kg on a calibrated beam scale. Overweight was defined using the recommended Asian BMI cut point of ≥23 kg/m2, with nonoverweight <23 kg/m2 (17). We also conducted comparative sensitivity analyses using the World Health Organization (WHO) BMI ≥25 kg/m2 cut point (18) (Supplementary Tables S4 and S5 and Supplementary Figure S2, online). Abdominal obesity was classified using waist circumference, which was measured with a SECA tape measure midway between the lower rib margin and the iliac crest. The IDF recommends sex and ethnic-specific waist circumference cut points, therefore the Asian cut point for abdominal obesity for adults (waist circumference: men ≥90 cm; women ≥80 cm) was used (13). Abdominal obesity is associated with cardiometabolic risk, independent of BMI (19), particularly in individuals with normal BMI (20,21).
Metabolically healthy and overweight categories. Cardiometabolic risk is defined as having elevated values of one or more of the following five risk factors: inflammation (hsCRP ≥3 mg/l), prehypertension or hypertension (SBP/DBP ≥130/85 mm Hg), high triglycerides (triglycerides: ≥150 mg/dl), low HDL-C: (male: <40 mg/dl; female: <50 mg/dl), and prediabetes or diabetes (HbA1c ≥5.7%). Patterns of risk associated with overweight were used to define the following four categories: “metabolically healthy, nonoverweight” (BMI <23 kg/m2, no risk factors); “metabolically healthy, overweight” (BMI ≥23 kg/m2, no risk factors); “metabolically at risk, nonoverweight” (BMI <23 kg/m2, ≥1 risk factor); and “metabolically at risk, overweight” (BMI ≥23 kg/m2, ≥1 risk factor). Inflammation was considered part of the cardiometabolic risk profile because in developing countries, high rates of obesity occur alongside high rates of infectious disease and malnutrition, resulting in dual burden of obesity-related and pathogen-related inflammation (22) and future cardiovascular disease (23). Abdominal obesity was not considered in the definitions of these phenotypes due to its covariation with BMI and conceptual interest in the association of abdominal obesity with these phenotypes.
Control variables. Age was recorded as the respondent's age on the date of exam, ranging from 18 to 98 years and categorized into the following categories: 18-30, 30-40, 40-50, 50-60, 60-70, and >70 years. Number of cigarettes per day were reported and categorized as currently smoking one or more cigarettes per day. Alcohol was reported as number of alcoholic beverages consumed per week across a variety of beverage types, categorized according to consumption of ≥1 servings of alcohol per day. Physical activity was reported using a detailed 7-day physical activity recall instrument and leisure activities assigned metabolic equivalent values using the compendium of physical activity (24), which has been used in international contexts, including China (25). Metabolic equivalent values were multiplied by the time spent in each activity and categorized according to ≥27 metabolic equivalents per week, equivalent to ≥1 h per week of moderate-intensity, such as brisk walking. Urbanicity was defined using a multidimensional 12 component urbanization index capturing community-level physical, social, cultural, and economic environments and represents the heterogeneity otherwise missed in an urban/rural measure based only on population density (26). Tertiles of the 100 value index represent low, medium, and high urbanicity. Household assets and household income were used to derive income; tertiles represent low, medium, and high income (27).
We conducted statistical analyses using Stata (Release 11.0, Stata, College Station, TX). In descriptive analyses, markers are compared across overweight status (percentages for categorical variables, means for continuous variables). A series of models were used to examine overweight (BMI ≥23 kg/m2) and cardiometabolic risk patterning (≥1 of the following risk factors: inflammation, prehypertension or hypertension, high triglycerides, low HDL-C, and prediabetes or diabetes) and assesses the independent and interactive effects of overweight and abdominal obesity (waist circumference: men ≥90 cm; women ≥80 cm). First, multinomial regression models predicting risk of falling into one of four categories: (i) “metabolically healthy, normal weight”, (ii) “metabolically healthy, overweight”, (iii) “metabolically at risk, overweight,” and (iv) “metabolically at risk, normal weight” (referent), with abdominal obesity as the main exposure were used to test the hypothesis that abdominal obesity would differentially predict membership into one of the four risk categories. Second, a series of five separate logistic regression models for each outcome (inflammation, prehypertension or hypertension, high triglycerides, low HDL-C, and prediabetes or diabetes) were used sequentially to test the hypotheses that: (i) overweight (first main exposure) differentially associates with each cardiometabolic risk factor and (ii) abdominal obesity (second main exposure), controlling for overweight, confers differential risk for each cardiometabolic marker. Third, five separate and sequential logistic regression models for each outcome (inflammation, prehypertension or hypertension, high triglycerides, low HDL-C, and prediabetes or diabetes) with four main exposure variables: (i) overweight with abdominal obesity; (ii) overweight with no abdominal obesity; (iii) nonoverweight with abdominal obesity; and (iv) nonoverweight with no abdominal obesity (referent), were used to test the hypothesis that there is an interactive effect between overweight and abdominal obesity, with overweight and abdominal obese individuals at highest risk. All models were sex-stratified and included the following covariates: age, smoking, alcohol consumption, physical activity, urbanicity, and income. Although model building suggested minimal change in effect of the main exposure with the addition of smoking, alcohol consumption, physical activity, urbanicity, and income, these covariates were retained on conceptual grounds. Two forms of sensitivity analyses were performed: (i) using the BMI ≥25 kg/m2 cut point rather than the more stringent Asian BMI ≥23 kg/m2 cut point and (ii) using impaired fasting glucose (≥100 mg/dl) as opposed to HbA1c (prediabetes or diabetes) as the indicator of diabetes risk. Results from both sensitivity analyses are shown in the Supplementary Material, online.
Although approximately half of the sample was overweight, 78% had elevated values of one or more cardiometabolic risk factors (49% had more than two) (Table 1). The proportion with prehypertension or hypertension (42.8%) and prediabetes or diabetes (38.1%) was particularly high. Rates of isolated hypertension risk (9.9%) and prediabetes or diabetes (7.7%) were roughly twice as high as isolated high triglycerides, low HDL-C, and inflammation, which were around 4%. In addition, mean waist circumference of the metabolically healthy nonoverweight group were on average 16 cm less than that of the metabolically at risk overweight group, 10 cm less than the metabolically healthy overweight, and 3 cm less than the metabolically at risk nonoverweight group.
All cardiometabolic risk factors were higher among the overweight (Table 2). The proportion of overweight and nonoverweight individuals with inflammation, prehypertension or hypertension, and prediabetes or diabetes increased with age, whereas rates of both high triglycerides and low HDL-C were highest in the early adult years for men and HDL-C was relatively stable among women. Relative to the other risk factors, differences in hypertension risk by overweight status were not as large. Mean continuous values for each marker are shown in Supplementary Table S1, online.
A comparatively large proportion of the sample across all age groups was “metabolically at risk normal weight”, notably 31% of the total sample aged 18-30 years (Figure 1). There were very few “metabolically healthy overweight” individuals and the proportion of the sample considered metabolically healthy (regardless of overweight status) declined considerably with age. Findings were similar across low, medium, and high urbanicity.
We used multinomial regression models to test the hypothesis that abdominal obesity differentially predicted membership into one of the four categories of obesity with and without metabolic risk (Table 3). The odds of being in the “metabolically at risk overweight” as compared with the metabolically healthy, nonoverweight group was dramatically elevated with abdominal obesity (men, odds ratio: 39.06, 95% confidence interval: 23.47-65.00; women, odds ratio: 22.26, confidence interval: 17.49-28.33). Abdominal obesity was also predictive of having high metabolic risk among both nonoverweight and overweight individuals.
We used cardiometabolic risk factor-specific logistic regression models to examine the association between overweight and each risk factor, with and without control for abdominal obesity (Set A, models 1 and 2). One set of models included overweight exposure only (Table 4, column 2), a second set of models included both overweight and abdominal obesity as exposure variables (Table 4, column 3). As expected, overweight independently increased the risk of each cardiometabolic marker in men and women, with variation across indicators. Independent of overweight status, abdominal obesity conferred risk for all cardiometabolic indicators.
A third set of cardiometabolic risk factor-specific logistic regression models was used to examine the interactive association between overweight and abdominal obesity with each risk factor (Set B models): (i) overweight with abdominal obesity, (ii) overweight with no abdominal obesity, and (iii) nonoverweight with abdominal obesity relative to (iv) nonoverweight with no abdominal obesity. Overweight individuals with abdominal obesity were at comparatively high cardiometabolic risk, with variation across prehypertension or hypertension, prediabetes or diabetes, inflammation and dyslipidemia (Table 4, column 6). In the majority of cases, there was close to or above a threefold higher risk with the combination of overweight and abdominal obesity relative to nonoverweight and no abdominal obesity.
Results were remarkably similar when impaired fasting glucose (≥100 mg/dl) as opposed to HbA1c (prediabetes or diabetes) was used as the indicator of diabetes risk (Supplementary Tables S2 and S3 and Supplementary Figure S1, online). Further, results were similar when overweight was defined using BMI ≥25 kg/m2 cut point (18) (Supplementary Tables S4 and S5, online), but with a considerably larger portion of the population considered “metabolically at risk nonoverweight” (Supplementary Figure S2, online). Relative to the BMI ≥23 kg/m2 cut point, using the BMI ≥25 kg/m2 cut point suggests that abdominal obesity has the same (albeit stronger) positive association with the metabolically at risk overweight.
Over three-quarters of this national Chinese sample had elevated levels of at least one cardiometabolic risk factor. Even in young adults, only half were “metabolically healthy” at 18-30 years of age, but this decreased to 6% in the 60-70-year-olds. Findings were similar across low, medium, and high urbanicity, indicating high prevalence of metabolic abnormalities even in the rural Chinese population, according to the IDF criteria for metabolic risk. Just over 5% of the overweight adults had no risk factors. Among the nonoverweight adults, 33.2% had at least one risk factor. Rates of isolated high triglycerides, low HDL-C, and inflammation were relatively low. The predominant risk factor was hypertension risk, which increased dramatically with age, underlying the decline with age in the proportion of adults considered metabolically healthy. Although we cannot rule out the fact that the high rates of hypertension risk might be a result of the normal increase in blood pressure with age, the high prevalence of metabolic abnormalities among the 18-30-year-old group, particularly in low HDL-C and high triglycerides, is notable. Our findings highlight the importance of screening in the general population, given the large portion of nonoverweight individuals with cardiometabolic risk. This should be at the forefront in China's efforts to prevent and control noncommunicable diseases.
Previous research suggests that hypertension risk is particularly high in China and is found at low BMI levels in Asians relative to other subpopulations (5,28,29). A recent paper on diabetes in China documented high diabetes prevalence (30), although that sample does not represent poor, less urbanized areas, as well as the CHNS. Although HDL-C tends to be relatively low in Asians (31), rates of low HDL-C were not extraordinary in these Chinese adults. Inflammation, a risk factor for type 2 diabetes and cardiovascular disease (32-34), is associated with insulin resistance and cardiometabolic risk in the absence of obesity (35). The observed findings suggest that inflammation increases with age across all adults, which is common in developing countries where high rates of obesity occur alongside high rates of infectious disease and malnutrition, resulting in dual burden of obesity-related and pathogen-related inflammation (22). Elevated hsCRP is a strong independent risk factor for future cardiovascular events, regardless of the underlying condition (15). For example, risk may be present even if the underlying etiology relates to periodontal disease, which is an independent risk factor for coronary heart disease (36). Despite the sex difference in abdominal obesity (largely due to the lower threshold for abdominal obesity in women (80 cm) vs. men (90 cm), inflammation rates were similar. The combined association between overweight and abdominal obesity with cardiometabolic risk factors suggests a multiplicative effect on cardiometabolic risk, with risk that is magnified in overweight individuals, even after controlling for smoking, alcohol consumption, total physical activity, urbanicity, and income.
Most cardiometabolic screening is aimed at individuals above a given threshold of body mass, despite the fact that elevated cardiometabolic risk is common in the general population (37). The observed high risk among nonoverweight Chinese, and the increase in risk with age, implies that screening for cardiometabolic risk should occur across all adults, regardless of weight status. Further, the high prevalence of prehypertension and hypertension, prediabetes or diabetes, high triglycerides, low HDL-C, and inflammation risk even below the BMI 23 kg/m2 threshold supports a lower BMI cut point for the Chinese population (4,17).
China has had a history of undernutrition followed by the most rapid increase in obesity and related diseases worldwide, with differential rates across rural and urban areas (2,38). In the last decade, changes in diet and activity in China have occurred more rapidly than previously recorded for any country (2), including increased intake of edible oil, fried foods, animal source foods, and snacking, as well as declining occupational, domestic, and travel activity expenditure and increasing TV time (25,39). The observed findings fit with the trend toward less healthier lifestyles, and indicate elevated cardiometabolic risk even in the rural areas of China.
There are a few limitations to this analysis. This cross-sectional analysis does not examine trends in risk or determinants of cardiometabolic risk. The main objective of this study was to determine the prevalence of “metabolically healthy overweight” and “metabolically at risk nonoverweight” in Chinese adults and to examine patterns of risk by age and gender. As such, the analytic strategy was designed to adjust for key covariates, rather than undertaking causal modeling of these relationships. The use of conventional, anthropometric, and cardiometabolic cut points do not capture differences in cardiometabolic risk that exist on the continuum of BMI and waist circumference. Likewise, cut points do not capture the complex process of cardiometabolic risk. However, cut points are needed for clinical guidance and there is rationale for use of Asian-specific BMI cut points based on higher risk at lower BMI (4,17,40). Finally, BMI does not discriminate between lean and adipose tissue and waist circumference is an indicator of both the subcutaneous and visceral adipose tissue compartments, therefore we cannot distinguish the two.
In summary, findings suggest a high prevalence of cardiometabolic risk in China that increases with age. The majority of the sample is considered metabolically at risk, with similar patterns of risk across urbanicity. Rates of risk increase dramatically with age and with overweight and abdominal obesity. The combined association between overweight and abdominal obesity with cardiometabolic risk factors, even after controlling for smoking, alcohol consumption, total physical activity, urbanicity, and income, suggest extraordinary burden of future cardiovascular disease with population increases in weight and abdominal obesity – a major clinical concern. Screening for cardiometabolic-related outcomes dependent upon overweight will likely miss a large portion of the Chinese at-risk population and has implications for other Asian populations.
We thank Ms. Frances Dancy, BS and Ms Erica Brody, MPH, UNC Carolina Population Center for their helpful administrative assistance, Jim Terry, BA, Phil Bardsley PhD, and Dan Blanchette BA for programming and technical support. P.G.L. and L.S.A. contributed to study design, P.G.L. and L.S.A. contributed to data analysis, and all authors contributed to writing of the manuscript. P.G.L. had full access to all the data in the study and takes responsibility for the integrity of the data and the accuracy of the data analysis. This work was supported by NIH: NIDDK (R21DK089306) and NHLBI (R01-HL108427). NIH had no role in the design and conduct of the study; collection, management, analysis, and interpretation of the data; and preparation, review, or approval of the manuscript. This research uses data from the China Health and Nutrition Survey, funded by NIH: NICHD (R01-HD30880), although no direct support was received from grant for this analysis. Dr. Meigs is supported by National Institute for Diabetes and Digestive and Kidney Diseases (NIDDK) K24DK080140.