Comparison of Different Obesity Indices associated with Type 2 Diabetes Mellitus Among Different Sex and Age Groups in China

Background: Obesity is associated with type 2 diabetes mellitus (T2DM), but the obesity index most closely related to type 2 diabetes is still controversial. Therefore, the aim of this study was to compare the associations of ve anthropometric indices (body mass index (BMI), body adiposity index (BAI), waist circumference (WC), waist-to-hip ratio (WHR), and waist-to-height ratio (WHtR)) with Type 2 diabetes mellitus (T2DM) among Chinese adults divided into four groups according to sex and age. Methods: A total of 4,007 adult participants (1,669 males and 2,338 females) were included. Odds ratios (OR) and 95% condence intervals were used with binary logistic regression models to estimate the risk of T2DM for each obesity index. Furthermore, we compared the area under the ROC curve (AUC) of each obesity index for the criterion of hypertension under the inuence of risk factors. Results: For the males, WC had the highest value of OR (3.211 and 1.452) and AUC (0.783 and 0.614) in both age groups. However, WHtR (OR=2.366, AUC=0.771) and BMI (OR=1.596, AUC=0.647) tended to be the best criteria for T2DM among females in the 18-59 and ≥ 60 years age groups, respectively. Conclusions: This study suggests that there is a positive association of obesity-related anthropometric indices with T2DM in different sex and age groups. For males, WC appears to be the best anthropometric index for predicting T2DM. For 18-59 and ≥ 60 age women, the best obesity indices related with T2DM are WHtR and BMI, respectively.


Background
Type 2 diabetes mellitus (T2DM), the most common form of diabetes, accounted for more than 96 percent of cases [1]. In China, the prevalence of type 2 diabetes has risen to 11.6% [2], affecting more than 100 million adults and making it the world's largest number of sufferers [3]. Previous studies have shown that weight gain is an independent risk factor for diabetes, while abdominal obesity is associated with T2DM [4][5][6]. Anthropometric indices used to de ne obesity can help in the identi cation of individuals or groups of people with a variety of speci c health thresholds [7]. The commonly used anthropometric indices of obesity are body mass index (BMI), body adiposity index (BAI), waist circumference (WC), waist-to-hip ratio (WHR), and waist-to-height ratio (WHtR) [8].
Some foreign studies have been conducted to assess the relationship between anthropometric obesity indices and T2DM [9][10]; however, the results of these studies are inconsistent. A cross-sectional study in the population of northern Iran showed that while WHR (AUC = 0.7303 in men, AUC = 0.7529 in women) had an appropriate discriminatory capability for T2DM, BAI and BMI did not. [11] Another hospital-based cross-sectional study conducted among 363 consecutively selected T2DM showed that WHtR (AUC = 0.85) and WC (AUC = 0.79) was the best predictor of metabolic syndrome for women and men, respectively [12]. Similar studies have been done in China, but unfortunately the results are not consistent even in the same country. A study of people over the age of 50 in Jinan showed the best indicator of the relationship between obesity and T2DM is WHtR (OR = 2.572) for men and BMI (OR = 2.764) for women, Page 3/22 respectively [13]. However, another study in Changchun showed that WHtR (AUC = 0.628 in men, AUC = 0.676 in women) is the best index to predict T2DM in Chinese adults [14].
The inconsistency of the optimal indicators obtained in the previous studies may be related to ethnic and regional differences [15][16][17][18]. In addition, previous studies have strati ed studies by age or sex without cross-grouping, and the association between obesity indicators and type 2 diabetes in different sex and age groups remains unclear. Therefore, the aim of this study was to compare the associations of ve anthropometric indices (BMI, BAI, WC, WHR, and WHtR) with T2DM among Chinese adults divided into four groups according to sex and age and to supplement similar studies in different regions of China. The thresholds of these anthropometric indices for obesity were also evaluated as a reasonable predictor of T2DM.

Study design and participants
The participants in this study were adult subjects who were enrolled from the baseline survey of the Chronic Disease Study in 2018, which was a cross-sectional study based on urban community-dwelling residents in the Chongchuan district with an estimated 720,000 permanent residents, Nantong City, Jiangsu Province, China.
In the rst stage of the enrollment, all 10 blocks of the Chongchuan district were selected and one residential community was randomly selected from each of the blocks. In the second stage, systematic sampling was conducted to select 450 households in each residential community. In the third stage, residents who were older than 18 years and had lived in the district for at least one year were selected from each household. Finally, a total of 12,092 people from 4,500 households were recruited to complete the questionnaires and undergo the relevant physical examinations. 4,035 participants (about 1/3) were randomly selected and fasting blood was drawn to measure plasma blood glucose (FPG). Twenty-eight people were excluded because they provided no information about weight, height, WC, or hip circumference or had a history of type diabetes, leaving a total of 4,007 participants for the nal analyses ( Figure.1).

Sample size evaluation
We calculated the minimum sample size required by using the following formula: In this formula, at 5% type I error; p, which is the prevalence of T2DM among Chinese adults, was about 10% [20]; and d, which is the absolute error or precision, p was 10% in our study. In addition, as the multi-stage random sampling method was used in our study. Therefore, the minimum sample size for our study was 3,600. The nal sample size we selected for analysis was large enough compared to the number we calculated.

Data collection
A trained investigator from the Center for Disease Control and Prevention of the Chongchuan district in Nantong and the School of Public Health in Nantong University interviewed each participant using standardized questionnaires and collected information on demographic data (age, sex, education, etc.), lifestyle behaviors (such as smoking, alcohol consumption, etc.), and family disease history (hypertension, diabetes).

Physical examination
Anthropometric indices were measured by trained staff who followed standard procedures. Height was measured using a standard stadiometer; each participant underwent the measurement without shoes. Weight was measured to the nearest 0.1 kg using an electronic scale, with the subjects wearing light indoor clothing. WC was measured to the nearest 0.1 cm at the midpoint between the lowest rib margin and the level of the anterior superior iliac crest using a exible anthropometric tape. HC was measured to the nearest 0.1 cm at the greatest protrusion of the gluteal muscles. Blood plasma glucose measurements required participants to designate a hospital for measurement on an empty stomach the morning of the test.
This study was in accordance with the World Medical Association Declaration of Helsinki-Ethical Principles for Medical Research Involving Human Subjects and was approved by the Ethics Committee of The Third People's Hospital of Nantong City (No: EK2018009; June 18th, 2018). A written informed consent form was obtained from each participant after they were informed of the objectives and bene ts of this study.

De nitions of variables
Patients who had type 2 diabetes mellitus were de ned as those who had been diagnosed in a secondary hospital or who had FPG ≥7.0mmol/L in the fasting blood glucose testing [21]. BMI was calculated as weight (kg) divided by the square of height (m 2 ). According to "the standards of the Chinese guidelines for the prevention and control of overweight and obesity in adults," BMI≥24 kg/m 2 and ≥28 kg/m 2 was regarded as being overweight and obese, respectively [22]. Body adiposity index (BAI) was measured using the formula [hip circumference (cm)/height 1.5 (m)] -18, BAI≥25.5 kg/m 2 & ≥29.0 kg/m 2 for males, ≥30.5 & ≥35.0 for females was regarded as being overweight and obese [23]. WC ≥90 cm for males and ≥80 cm for females were de ned central obesity. WHR was calculated as WC (cm) divided by HC (cm), with values >0.9 for males and >0.8 for females indicating central obesity. WHtR was calculated as WC (cm) divided by height (cm), and values ≥0.5 indicated obesity [22].

Statistical Analyses
All quantitative variables were described as mean and standard deviation (SD) for normal distributions, whereas qualitative data were described as percentages. The student t test and chi-square test were used to estimate the difference between quantitative and qualitative data of different groups, respectively. The following weight variables were estimated in this study: BMI, BAI, WC, WHR, and WHtR. Odds ratios (OR) and 95% con dence intervals (95%CI) were used with binary logistic regression models to estimate the risk of T2DM for each obesity index.
For each index, receiver operating characteristic (ROC) curve analysis was used to identify the value with the Youden index as a predictor of T2DM. P values < 0.05 were considered signi cant. The statistical analyses were performed using SPSS 24.0 for Windows.

Results
The demographic data for the 12,064 eligible participants included in this study are shown in Table 1. The mean age of the participants was 61.78 ± 14.95 years, and at the age of 60, the population was divided into two groups. Naturally, the prevalence of T2DM increased with increasing age. The ratio of male to female participants was 1:0.71, and males had a higher risk of T2DM than females. Regarding education, participants who had primary school or lower education accounted for the highest percentage (44.1 percent). The higher the education level, the lower the risk of T2DM. In addition, smoking status, alcohol drinking status, positive family history of diabetes and hypertension were signi cantly correlated with T2DM. Finally, there was a statistically signi cant difference between participants who had T2DM and those who did not with BMI, WC and WHtR.  Values are presented as mean ± SD. p < 0.05, statistically signi cant.
Considering the confounding effect of age and sex, we divided the participants into four subgroups based both variables. In all four subgroups, people who had T2DM had higher WC than the controls (P < 0.05) and WHR was not signi cantly different between the T2DM group and the controls (P > 0.05).
However, for males, BMI was not signi cantly different between the T2DM group and the controls (P > 0.05), whereas in the female groups, BMI was higher in the T2DM group. Moreover, BAI and WHtR was higher in the T2DM group excluding the male ≥ 60 age group.
The associations between the prevalence of T2DM and the obesity indices are shown in Table 3. In both groups of females, the ndings indicated positive associations between the ve obesity indices but WHR with T2DM even after adjustments for risk-related factors in Model 2.  , the patient may control their weight, we also considered the "time of diagnosis of T2DM" as an adjustment variable. As Table 4 shows, similar to the front results, the AUC of BAI and WC were the largest (0.783 and 0.614) among the ve anthropometric indices in two male groups. In addition, WHtR (AUC = 0.771) and BMI (AUC = 0.647) tended to be the best predictors for T2DM among females in the 18-59 and ≥ 60 years age groups, respectively.

Discussion
In this study, we compared the associations of ve anthropometric indices for T2DM among Chinese adults divided into four groups according to sex and age. Using survey data from a cross-sectional study of chronic diseases in Nantong city, east of China, we found strong evidence that there is a positive correlation between obesity-related measures and T2DM; a nding which was consistent with those of previous studies in China [15][16]. Previous studies have found that leptin level in obese people is higher than that in normal people, while adiponectin level is opposite. Leptin can activate renal sympathetic nerve and increase arterial pressure. Adiponectin can effectively protect cardiovascular system, which is negatively correlated with insulin resistance [24][25]. Obesity can lead to islet dysfunction and decrease of insulin secretion, meanwhile, adipocytes are insensitive to insulin, which leads to an increase of blood sugar [26].
The results of this study also show that, compared with other obesity indicators, WHtR had a greater impact on the risk of T2DM in the general population, which was consistent with previous research results. Previous studies have shown that height has an important in uence on diabetes, while WC and WHR have not considered the in uence of height [27][28]. In order to overcome this defect, most researchers used WHtR to study the relationship between overweight, obesity and chronic diseases [29][30]. WHtR is easy to calculate, and has a signi cant relationship with waist circumference without gender difference. Taking ≥ 0.5 as the cut-in point can well predict the risk of diabetes [31]. A study in China showed that waist-height ratio has no effect on impaired fasting blood glucose, but it is related to diabetes, which can be used as an important screening index for the elderly [32]. In addition, central and systemic obesity indices showed different correlations in different sex and age groups.
For males, WC had the highest OR and the largest AUC before after adjusting for confounding variables in both 18-59 and ≥ 60 age groups. The results of a meta-analysis of 16 cohort studies involving different ethnic groups in Asia showed that for every additional SD in BMI and WC, the risk of diabetes was 52% and 54%, respectively. WC was more closely related to diabetes [33]. A survey in China found that for every standard deviation increase in BMI and WC, the risk of diabetes increased by 53% and 64%, respectively. WC had a greater impact on the risk of diabetes [34]. The conclusions of these studies were the same as those in the present study, which might be related to the fact that insulin resistance is an important pathogenic factor for diabetes [35]. Studies have con rmed that insulin antagonism is more obvious in central obesity, and WC is recognized as an important indicator for measuring central obesity, which can better re ect insulin resistance [36]. Therefore, it was recommended that the waist circumference can be reduced through appropriate physical exercise and other means on the basis of weight control, thereby reducing the risk of type 2 diabetes.
However, among the females, WHtR and BMI tended to be the best predictors for T2DM in the 18-59 and ≥ 60 age groups, respectively. The reason for the sex difference in the relationship between obesity indicators and T2DM may be related to the sex difference in body fat distribution. Fat was mainly in the viscera in men and subcutaneous in women, which can provide evidence for gender differences. In thin Asian men with less subcutaneous fat, WC may be a better indicator of visceral obesity than BMI. Among thin Asian women, the impact of subcutaneous fat was greater than that of WC, and BMI may be more suitable for the indicator of overall fat and abdominal fat accumulation than WC [37][38][39]. On the other hand, compared with women, men were more stressed at work and slept less. Working pressure can increase the epinephrine secretion level of the body, and then lead to more fat into the abdomen [40]; Sleep deprivation was closely related to cardiovascular risk factors such as in ammatory markers in the blood circulation, and the resulting in ammatory response may be one of the mechanisms leading to metabolic diseases such as abdominal obesity and T2DM [41]. However, in recent years, Chinese women's rich diet nutrition, reduced exercise and increasing pressure at work have led to the increase of central obesity in young and middle-aged women [42]. This may explain why obesity indicators most closely associated with T2DM are inconsistent across age groups.
BAI was also included in this study as a new obesity index proposed in 2011, which was believed to be able to assess percent body fat more objectively based on hip circumference and height [43]. We found that BAI was positively associated with T2DM except in male ≥ 60 age group. The positive association of T2DM with BAI was not stronger than that observed in other generally used obesity indices. Consistently, the capacity of BAI in predicting T2DM was also found lower than BMI, WC and WHtR, which recognized as the most frequently used obesity indices. This was inconsistent with some American research results that BAI was superior to BMI [44][45], possibly because BAI was proposed by the research on hip circumference and height of African Americans and Mexican Americans [43], while the characteristics of height and hip circumference of Chinese Han people are quite different from them.
We also noticed that there was no positive association between WHR and T2DM in all sex and age groups. Therefore, using WHR as a weight predictor for T2DM was not recommended. This nding was similar to that of a study in Changchun. In that study, the cut-off value for WHR was not suitable for predicting T2DM compared with those of BMI and WHtR [19].
One of strengths in the present study was that we randomly selected a community-based population with a broad age range; this increases the generalizability of our results to local populations. Further, we compared the predictive values of ve anthropometric indices (BMI, BAI, WC, WHR, and WHtR) to achieve accurate evaluation of T2DM prediction by obesity indicators for different age and sex groups. We believe that the ndings of the present study may provide a reference for the selection of the appropriate anthropometric obesity indices for estimating the risk of T2DM in people of different sexes and ages.
Several limitations of the present study should be considered. Firstly, as a cross-sectional study, the causality between obesity indices and T2DM cannot be properly proven. Longitudinal follow-up studies need be conducted to further verify the results of the present study. Secondly, participants who reported T2DM by themselves may alter their lifestyle habits such as diet and exercise to skew their anthropometric indices. Although this was considered and adjusted for in the model and ROC curve analyses, information migration was unavoidable. Finally, only residents from a single area were recruited for this study. Therefore, the conclusions of this study may have some limitations in extrapolation due to the diet and lifestyle differences of different regions.

Conclusion
This study suggests that there is a positive association of obesity-related anthropometric indices with T2DM in different sex and age groups. For males, WC appears to be the best anthropometric index for predicting T2DM. For 18-59 and ≥ 60 age women, the best obesity indices related with T2DM are WHtR and BMI, respectively.
Declarations Figure 1 The sample diagram of the study object

Supplementary Files
This is a list of supplementary les associated with this preprint. Click to download.