Background

Cardiovascular diseases (CVDs) are serious circulatory diseases that endanger human health and the main cause of mortality worldwide [1, 2]. The World Health Organization (WHO) reports that about 17.9 million people died from CVDs, accounting for 31% of global deaths [3]. It is established that adults characterized by excessive adipose tissue and ectopic fat stores have a higher risk of CVDs [4, 5]. Long-standing evidence has demonstrated the linkage between obesity and increased risk of hypertension, type-2 diabetes mellitus (T2DM) and dyslipidemia, which are major risk factors for CVDs [6, 7]. However, the definition of obesity remains controversial and discrepancies in definitions may lead to inaccurate assessment for CVD risk factors. Body mass index (BMI) is generally recognized anthropometric index of obesity, and numbers of epidemiological studies have confirmed that BMI can predict CVD risk factors [8]. Nevertheless, the neglect of body fat distribution is considered as a typical limitation of BMI. Waist circumference (WC) is recommended as an effective index for measuring the distribution of visceral fat and closely related to cardiometabolic risk [9, 10]. But the variation of WC thresholds across country, ethnic groups, sex and age remains an important issue to be taken into account [11]. Recently, compared with WC, waist-height ratio (WHtR) can account for differences in height, which is easy to measure and variation slightly, and has been proposed a superior predictor of CVD risk factors [12, 13]. To sum up, with regard to the best anthropometric indicator for CVD risk factors still remains inconsistent.

As we all know, WHO recommend that BMI ≥ 30 kg/m2 was defined as general obesity and WC ≥ 102 cm in male and ≥ 88 cm in female was identified as central obesity [14]. But more evidences have shown that WHO recommended cutoff values may not be able to detect those with CVD risk factors in Asian populations [15, 16]. Compared with Western counterparts, the same level of BMI and WC among Asians may have a higher percentage of body fat due to racial disparity of physical characteristics [17, 18]. It can be seen that the threshold for obesity does not apply to any population. Although some studies have been conducted to assess which anthropometric indicators are more strongly related to CVD risk factors in other provinces in China, controversies still existed [19, 20]. Shanghai is the most populous and economically developed city, the process of urbanization, the degree of aging, socioeconomic characteristics and lifestyle of population are quite different from those of other districts in China. Moreover, few studies assessed the discriminatory power of anthropometric indices for CVD risk factors by sex and age subgroups. It is very necessary to conduct large-scale surveys in our population to explore the optimal cut-off values for CVD risk factors. Thus, the objective of our study is to determine the optimal cut-off points of BMI, WC and WHtR for CVD risk factors (hypertension, T2DM, dyslipidemia and hyperuricemia) and explore the associations between anthropometric measures and CVD risk factors stratified by sex and age among adults in Shanghai, China.

Methods

Study population

From June 2016 to December 2017, we conducted a population-based cross-sectional survey in Songjiang District, Shanghai. A multistage stratified sampling method was used to recruit participants. At stage one, we purposively selected four study areas in Songjiang District stratified by geographic region and economic levels, which included two urban communities (Zhongshan and Xinqiao), one urban-rural mixed region (Sheshan) and one rural district (Maogang). At stage two, nine and eighteen street districts were selected from Xinqiao and Zhongshan randomly. Four neighborhood committees were randomly selected from Sheshan, and sixteen administrative villages were selected from Maogang. At stage three, individuals aged 20–74 years were selected randomly from each of administrative units in stage two. We totally recruited 37,670 permanent participants who had been living in Songjiang district for 5 years or longer. Residents with disabilities, terminal illness or perceptual impairment were excluded. After exclusion of individuals who lacked data in any of the variables, 35,256 residents (a response rate was 93.6%) were left for final analysis.

Data collection

Questionnaire interview

The electronic questionnaire (Additional file 2) consists of questions on demographic characteristics (e.g., sex, age, education level), health-related behaviors (e.g., current smoking and drinking status), and history of self-reported chronic diseases (e.g., hypertension, diabetes). Education was recorded as completed years of schooling and categorized to three categories of ≤6 years (primary school and lower), 7–12 years (middle and high school) and > 12 years (college and above). For smoking, participants were asked two questions: “Have you ever smoked at least one cigarette every day for more than six months (yes, no)?” and “What is your smoking status now (current and non-current)? (for individuals who answered ‘yes’)?” We defined current smoking as having smoked every day for more than six months and currently smoking cigarettes. For alcohol consumption, participants were asked two questions: “Have you consumed alcohol at least 3 times a week for more than 6 months (yes, no)?” and “Have you consumed alcohol during the past year (for individuals who answered ‘yes’)?” In order to ensure quality stringently, all subject information was electronically coded and stored in the database. At each stage of the investigation, two quality controllers were arranged to check data collection, make logical judgment and correct the data with poor quality in time.

Measurements

Anthropometric measurements were conducted in participants wearing light clothing with no footwear. All the measurements were performed by community licensed physicians according to the standard protocol. BMI was calculated as weight (accurate to 0.1 kg) in kilograms divided by height (accurate to 0.1 cm) in meters squared (kg/m2). WC was measured at the mid-point between the iliac crest and the last rib while the subject was at minimal respiration by a flexible tape. WHtR was calculated as WC (cm) divided by height (cm). Blood pressure was measured on the right arm in a sitting position using a digital sphygmomanometer after five minutes of rest. Three measurements were taken and mean values was recorded.

Laboratory assay

Blood samples were collected into vacuum tubes in the morning after an overnight fast. The specimens were obtained and stored at − 80 °C freezer within less than 6 h and thereafter transported to the Shanghai Dian Diagnostics Co Ltd. Serum total cholesterol (TC), triglycerides (TG), low-density lipoprotein cholesterol (LDL-C) and high-density lipoprotein cholesterol (HDL-C) were measured using enzyme colorimetry (Roche COBAS C501 automatic biochemical analyzer). Fasting plasma glucose (FPG) were measured using Glycokinase method (Roche P800 automatic biochemical analyzer). Glycated hemoglobin (HbA1c) were measured using high pressure liquid chromatography (TOSOH G8, Automatic hemoglobin A1c analyzer). Uric acid (UA) was measured using colorimetry (Roche c702 automatic biochemical analyzer).

Definition of CVD risk factors

Hypertension was defined as systolic blood pressure (SBP) ≥ 140 mmHg and/or a diastolic blood pressure (DBP) ≥ 90 mmHg or self-reported history of hypertension. Diabetes was defined by American Diabetes Association (ADA) criteria: as HbA1c ≥ 6.5% or FPG ≥ 7.0 mmol/L or self-reported history of diabetes [21]. Dyslipidemia was defined as at least one of the following characteristics: high TC (TC ≥ 6.22 mmol/L); high LDL-C (LDL-C ≥ 4.14 mmol/L); low HDL-C (HDL-C < 1.04 mmol/L) or hypertriglyceridemia (TG ≥ 2.26 mmol/L) [22] . Hyperuricemia was defined as uric acid ≥7.0 mg/dl (416.0 μmol/L) in men and ≥ 6.0 mg/dl (357.0 μmol/L) in women [23].

Statistical analysis

Continuous variables are represented by mean (standard deviation (SD)) and categorical variables by percentages. Cut-off value was determined between each anthropometrical index and CVD risk factors by the receiver operating characteristic (ROC) curve, which was quantified by the area under ROC curve (AUC). The Youden’s index (sensitivity+specificity-1) was used to determine the optimal cutoff point of each index. Z test was used to determine differences between two AUCs: \( \mathrm{Z}=\frac{\ {AUC}_A-{AUC}_B}{\sqrt{{S_{EA}}^2+{S_{EB}}^2}} \), (SE is standard error, Z > 1.96, p < 0.05). Multivariate logistic regression analysis was performed to estimate the association between each SD increase of anthropometrical index and CVD risk factors after adjusting for age, educational level, current smoking and drinking by sex. The adjusted odds ratios (ORs) are presented with 95% confidence interval (CIs). All the analyses were conducted using SPSS version 23 software (SPSS Inc., Chicago, IL, USA). Two-tailed p < 0.05 was considered to be statistically significant.

Ethical approval

The Shanghai cohort study protocol was approved by the ethical review committee of School of Public Health, Fudan University (IRB approval number 2016-04-0586). Written informed consent was obtained from all study participants prior to the study.

Results

The basic characteristics and prevalence of CVD risk factors are presented in Table 1. Compared with women, men had a significantly higher mean BMI, WC, SBP, DBP, FPG, Hb1Ac, TG and UA, but lower mean WHtR, TC, LDL-C and HDL-C. Men had higher prevalence of hypertension (56.05% vs. 49.72%), diabetes (15.19% vs. 13.71%), dyslipidemia (34.35% vs.25.74%) and hyperuricemia (16.63% vs.8.86%) than women. Only 0.68 and 0.22% of the women were current drinkers and smokers, respectively.

Table 1 The basic characteristics and prevalence of CVD risk factors in Songjiang district, Shanghai
Full size table

Table 2 shows the correlation between anthropometric index and CVD risk factors. BMI, WC and WHtR were significantly correlated with the CVD risk factors in both sexes (p < 0.001). Compared with BMI and WC, WHtR had the highest correlation with SBP, FPG, Hb1Ac, TC and LDL-C for both men and women.

Table 2 Pearson’s correlation coefficients between anthropometric indices and CVD risk factors among men and women
Full size table

The ROC curves and AUCs of BMI, WC and WHtR for hypertension, diabetes, dyslipidemia, and hyperuricemia by sex are displayed in Fig. 1 and Table 3. Regarding hypertension and diabetes, the AUCs for WHtR were higher than WC among men, and were higher than BMI among women. No significant difference in AUCs was found between three anthropometric indices and dyslipidemia and hyperuricemia in either men or women. The AUCs for anthropometric indices stratified by age groups are summarized in Additional file 1: Table S1. The AUCs between three anthropometric indices and CVD outcomes tended to decrease with age.

Fig. 1
figure 1

ROC curves of the anthropometric indices for hypertension, diabetes, dyslipidemia and hyperuricemia in men and women (a: Men; b: Women)

Full size image
Table 3 AUC for various anthropometric indices and CVD risk factors in men and women
Full size table

Table 4 shows the optimal cutoff points of three anthropometric indices based on ROC analysis for CVD risk factors. The optimal BMI cut-off values varied from 24.39 to 24.93 kg/m2 in men, and 23.09–24.16 kg/m2 in women. The optimal WC cut-off values varied from 83.38 to 85.68 cm in men, and 77.68–80.95 cm in women for all the CVD outcomes. The optimal WHtR cut-off values varied from 0.49 to 0.52 in men, and 0.49 to 0.53 in women. With regard to BMI and WC, the optimal cut-off values were mostly higher for men compared with women. Furthermore, the optimal cutoff values of BMI and WC varied across different age groups, while the difference in WHtR was relatively smaller (Additional file 1: Table S2 and S3). For women, the optimal cutoffs of BMI, WC and WHtR for hypertension and diabetes were higher in older age groups (60–74 years).

Table 4 Optimal cut-off value for anthropometric indices predictive of CVD risk factors
Full size table

Table 5 shows adjusted ORs of CVD risk factors with one SD increase of BMI, WC and WHtR. WHtR had the highest ORs for diabetes, dyslipidemia and hyperuricemia in men, while WC yielded the greatest ORs in women. BMI gave the relatively higher ORs for hypertension in both sexes. Figures 2 and 3 shows the ORs for CVD risk factors according to one SD increase in anthropometric measures categorized across age groups by sex. Results indicated that association between anthropometric indices and CVD risk factors were attenuated with age. WHtR had relatively higher ORs in the group aged 20–44 and 45–59 years except hypertension in women, while among 60–74 years the ORs for BMI were highest in terms of all CVD outcomes except diabetes in women.

Table 5 Odd ratios and 95% CI for CVD risk conditions corresponding to one SD increase in BMI, WC and WHtR
Full size table
Fig. 2
figure 2

Odds ratios (ORs) and 95% CI for CVD risk factors corresponding to one SD increase in BMI, WC and WHtR by age groups in men

Full size image
Fig. 3
figure 3

Odds ratios (ORs) and 95% CI for CVD risk factors corresponding to one SD increase in BMI, WC and WHtR by age groups in women

Full size image

Discussion

This study evaluated the optimal cut-off values and discriminatory ability of BMI, WC and WHtR for CVD risk factors among adults in Shanghai. Our results indicated that BMI cutoff values were comparable with those of the recommended cutoff by Working Group on Obesity in China (WGOC), but WC cut-off values were relatively lower. Results from previous studies demonstrated that WC threshold of about 80–85 cm for men and 75–80 cm for women [9, 24], which were similar to our results. One study of deriving the optimal WC threshold for cardiometabolic risk in sub-Saharan Africa found that the optimal cutoff at increased cardiometabolic risk was lower (81.2 cm) than current recommend guidelines in men [25]. Several prospective studies have proposed that WHtR cutoff value of 0.50 could be applied to different populations [26]. Our study supported this view that the cutoff of WHtR was approximately 0.51 for CVD risk factors. One cardiovascular study among Filipino-American women revealed the best cutoff point of WHtR was 0.50 and only 1% of patients with hypertension and 0% of patients with diabetes cannot be detected [27]. The cutoffs of WHtR were 0.52–0.53 for identifying several CVD risk factors in both sexes in Jinan, China [28]. Another study suggested that the optimal threshold of WHtR was 0.48–0.51 for men and 0.47–0.50 for women for predicting cardiometabolic risk in Taiwan [29]. The slight variations of WHtR made it likely to be applied to public health practice as a simple screening tool. The reason for small changes of threshold of WHtR was probably differences in sample size, age group or statistical methods.

In this study, WHtR tended toward the best performance for discriminating hypertension and diabetes, which was in agreement with other studies [30, 31]. However, the differences in AUCs between anthropometric indicators and dyslipidemia and hyperuricemia were small and 95%CIs were overlapping. Hence, BMI, WC, and WHtR may be comparable in their ability to identify these two CVD risk factors. Apart from this, due to the units of each anthropometric indicator were different, it is inappropriate to directly compare OR values. Therefore, we estimated the ORs in terms of each SD increase of the anthropometric measure, and found that WHtR or WC may have better association except for hypertension based on point estimates of associations in both sexes. Previous epidemiology studies have shown that central obesity indicators such as WC and WHtR are more closely associated with CVD risk factors than BMI, which was consistent with our results [32, 33]. This may be because ectopic fat deposition triggers pathological metabolic reactions, which increased the risk of metabolic diseases [34]. It is noteworthy that in our study, the AUCs decreased with age, indicating that the ability to recognize CVD risk factors gradually weakened with age. The association was strongest between anthropometric indices and CVD risk factors among 20–44 age group regardless of sex. A recent study on the association between obesity and hypertension among Chinese elderly provided evidence that anthropometric indicators varied with age [35]. Other studies have also demonstrated this similar age-specific difference [36, 37], possibly because that older people tend to lose muscle mass and have more concentrated fat distribution with age [38]. Thus, targeted weight management will benefit greatly among young population. Additionally, the present study also demonstrated among aged 60–74 years the ORs of BMI were highest in terms of all CVD outcomes except for diabetes in women, but WHtR had relatively higher ORs for people under 60 years except women with hypertension. The potential mechanism is unclear, thus more researches are needed to explore the potential age difference for elucidating the role of age in the pathogenesis of obesity and CVD risk factors.

Strengths of this study include a large population-based sample size and extensive CVD risk factor data available for study participants. This study is the first representative sample containing most age groups to determine the optimal cut-off values for CVD risk factors among Shanghai adults. The present study also has some notable limitations. Firstly, cross-sectional study cannot estimate the causality between anthropometric measures and CVD risk factors. In order to confirm the results of this study, further prospective cohort studies are needed. In addition, as our study was conducted only in Shanghai, it limits the generalizability of our results to other regions in China. Secondly, diet, stress, treatment and genetics are known confounders for anthropometric measures of obesity and CVD risk factors, due to unavailable of these data, we did not consider these variables in the analysis. Finally, we could not account for the relationship between more measures of obesity (hip circumference, waist-to-hip ratio, body adiposity index along with others) and CVD risk factors because of limited data.

Conclusions

In summary, our data suggested the optimal cut-off of WHtR was approximately 0.51 for CVD risk factors and it tended to perform best in discriminating hypertension and diabetes in both sexes in Shanghai. The optimal cutoff values of BMI and WC varied greatly for CVD risk factors across different sex and age groups, while the difference in WHtR was relatively smaller. The discriminating power of anthropometric indictors and associations with CVD risk factors were attenuated with age. Thus, weight management and control should be strengthened at a young age for primary prevention of cardiovascular diseases.