Background: To determine the anthropometric indices that would predict type 2 diabetes (T2D) and delineate their optimal cut-points. Methods: In a cohort study, 7017 Iranian adults, aged 20–60 years, free of T2D at baseline were investigated. Using Cox proportional hazard models, hazard ratios (HRs) for incident T2D per 1 SD change in body mass index (BMI), waist circumference (WC), waist to height ratio (WHtR), waist to hip ratio (WHR), and hip circumference (HC) were calculated. The area under the receiver operating characteristics (ROC) curves (AUC) was calculated to compare the discriminative power of anthropometric variables for incident T2D. Cut-points of each index were estimated by the maximum value of Youden’s index and fixing the sensitivity at 75%. Using the derived cut-points, joint effects of BMI and other obesity indices on T2D hazard were assessed. Results: During a median follow-up of 12 years, 354 men, and 490 women developed T2D. In both sexes, 1 SD increase in anthropometric variables showed significant association with incident T2D, except for HC in multivariate adjusted model in men. In both sexes, WHtR had the highest discriminatory power while HC had the lowest. The derived 2 2 cut-points for BMI, WC, WHtR, WHR, and HC were 25.56 kg/m , 89 cm, 0.52, 0.91, and 96 cm in men and 27.12 kg/m , 87 cm, 0.56, 0.83, and 103 cm in women, respectively. Assessing joint effects of BMI and each of the obesity measures in the prediction of incident T2D showed that among both sexes, combined high values of obesity indices increase the specificity for the price of reduced sensitivity and positive predictive value. Conclusions: Our derived cut-points differ between both sexes and are different from other ethnicities. Keywords: BMI, Waist circumference, Waist to height ratio, Obesity, Cohort study, Joint effect Background developing countries such as those in the Middle East Diabetes is the most prevalent metabolic disorder in the and North Africa region (approximately 96.2% increase world . It is a coronary heart disease equivalent [2–4] in prevalence in 20 years) . and had a substantial global burden with 680 One of the major modifiable risk factors of T2D is disability-adjusted life years per 100,000 people in 2010 obesity [6–10]. Despite clear evidence linking obesity to . By 2035 the prevalence of type 2 diabetes (T2D) will various health outcomes including cardiovascular increase by 55% worldwide with an alarming pace in diseases (CVD), obesity has been a puzzling condition for clinicians because it is quite heterogeneous  and the existing evidence regarding the suitable anthropo- * Correspondence: email@example.com 1 metric index to be used as a screening test in each sex is Prevention of Metabolic Disorders Research Center, Research Institute for Endocrine Sciences, Shahid Beheshti University of Medical Sciences, Tehran, controversial. On the one hand, the literature supports Iran the application of central obesity, such as waist to height Full list of author information is available at the end of the article © The Author(s). 2018 Open Access This article is distributed under the terms of the Creative Commons Attribution 4.0 International License (http://creativecommons.org/licenses/by/4.0/), which permits unrestricted use, distribution, and reproduction in any medium, provided you give appropriate credit to the original author(s) and the source, provide a link to the Creative Commons license, and indicate if changes were made. The Creative Commons Public Domain Dedication waiver (http://creativecommons.org/publicdomain/zero/1.0/) applies to the data made available in this article, unless otherwise stated. Zafari et al. BMC Public Health (2018) 18:691 Page 2 of 12 ratio (WHtR) over general obesity indicators in assessing cut-points for the prediction of T2D in each sex, the ana- T2D risk [12, 13]. On the other hand, there are studies lyses were performed only among those who participated that use body mass index (BMI) as their main obesity in the last follow-up phase and those with incident T2D index in predicting T2D [14–16]. Several studies have during the follow-ups (N = 5738, 2419 men, 3319 women) been conducted to obtain the optimum cut-points of an- (Additional file 1). thropometric indices [17–22]. However, these studies were mainly conducted in Asians, European and American Clinical and laboratory measurements Caucasians. Understandably, the results are not necessar- Using a pretested questionnaire, a trained interviewer ily generalizable to other ethnicities . Therefore, collected information regarding demographic data, drug WHO has emphasized the need for prospective studies to history and family history of T2D. Weight was mea- derive and validate ethnic-specific cut-points of body fat sured, with subjects minimally clothed without shoes, composition indices to predict CVD and T2D . using digital scales (Seca 707: range 0–150 kg) and Located in the Middle East, Iran suffers from high inci- recorded to the nearest 1 kg. Using a tape meter, height dence and prevalence of T2D [10, 25]. Yet, data regarding was measured in a standing position without shoes, appropriate anthropometric cut-points in the Iranian while shoulders were in normal alignment; waist circum- population using prospective studies is limited . ference (WC) was measured at the umbilical level and In the current study, we decided to determine the that of the hip (HC) at the widest girth of the hip over sex-specific independent and combined risk of different light clothing, without any pressure to the body surface. anthropometric indices in the prediction of T2D in the Measurements were recorded to the nearest 1 cm. Body cohort of Tehran Lipid and Glucose Study (TLGS) with mass index (BMI) was calculated as weight (Kg) divided over 12 years of follow-up; furthermore, we compared by height squared (m ). Waist to hip (WHR) and waist the discriminatory power of adiposity measures and de- to height ratios (WHtR) were calculated as WC (cm) lineated their optimal cut-points in this population. divided by HC (cm) and height (cm), respectively. Wrist circumference was measured with the anterior wrist Methods surface facing up. The superior border of the tape meter Study design and sample was placed just distal to the prominences of the radial The TLGS, an ongoing prospective population-based and ulnar bones, without any tape pressure over it; study being performed on a representative sample of the values were recorded to the nearest 0.1 cm. Tehran population, aims to determine the prevalence After a 15-min rest in the sitting position, two and incidence of non-communicable diseases and their measurements of systolic and diastolic blood pressure risk factors. Detailed descriptions of the TLGS have (SBP and DBP) were taken on the right arm, using a been reported elsewhere . In brief, one baseline standardized mercury sphygmomanometer (calibrated by (1999–2001) and 4 follow-up examinations at triennial the Iranian Institute of Standards and Industrial intervals have been carried out until January 2015. Researches); the mean of the two measurements was Those who had cancer, end-stage renal disease or considered as the participant’s blood pressure. cirrhosis were excluded from the TLGS at the base- A blood sample was taken between 7:00 and 9:00 AM line examination . For the current study, 10,727 from all study participants, after 12 to 14 h of overnight participants, aged 20–60 years [8569 people from the fasting. All blood analyses were carried out at the TLGS baseline examination (1999–2001) and 2158 new partici- research laboratory on the day of the blood collection. pants recruited from the second phase (2001–2005)], were For all non-pharmacologically treated diabetic partici- selected. Subjects with prevalent T2D at their baseline pants aged ≥20 years, an oral glucose tolerance test with examination [N = 806, excluding those with known T2D 82.5 g of glucose monohydrate solution [equivalent to using anti-diabetic medications (N = 276), 32 participants 75 g of anhydrous glucose; Cerestar EP, Spain] was had isolated high fasting plasma glucose (FPG), 95 sub- performed; a second blood sample was obtained 2 h jects had high FPG levels while their 2 h-post-challenge after glucose ingestion. Fasting plasma glucose (FPG) plasma glucose (2 h-PCPG) was missing, 202 participants and 2 h-post-challenge plasma glucose (2 h-PCPG) were had isolated high 2 h-PCPG and 201 had combined high measured using an enzymatic colorimetric method with FPG and high 2 h-PCPG], no data on baseline variables glucose oxidase; inter- and intra-assay coefficients of (N = 1342) or not any follow-up data (N = 1562) were variation at baseline and follow-up phases were both less excluded; leaving 7017 participants (2988 men, 4029 than 2.3%. women) to include in the analyses as respondents High-density lipoprotein cholesterol (HDL-C) was (response rate = 7017/(10,727–806) × 100 = 70.7%). Fur- measured after precipitation of the apolipoprotein B thermore, to compare the discriminatory power of containing lipoproteins with phosphotungstic acid. Tri- anthropometric indices and to delineate their optimal glycerides (TG) were assayed using glycerol phosphate Zafari et al. BMC Public Health (2018) 18:691 Page 3 of 12 oxidase. Both inter- and intra-assay coefficients of been shown to be a significant predictor of T2D in the variation were less than 3 and 2.1% for HDL-C and TG, TLGS adult population . respectively. Analyses were performed using Pars Azmon The proportional hazards assumption in the Cox kits (Pars Azmon Inc., Tehran, Iran) and a Selectra 2 models was assessed both graphically and using the auto-analyzer (Vital Scientific, Spankeren, Netherlands). Schoenfeld residual test. All proportionality assumptions Triglyceride to high-density lipoprotein cholesterol ratio were met. Collinearity was checked by estimating the (TG/HDL-C) was calculated by dividing TG to HDL-C. first order correlation coefficients between variables used All samples were analyzed when internal quality control in each model as well as using correlation matrix of co- met the acceptance criteria. efficients in the multivariable adjusted cox model . There was no pair of variables with correlation coeffi- cient of 0.80 or more in models. Receiver operating Definition of terms characteristics (ROC) curves were plotted to compare Subjects who reported a parent or sibling with diabetes the discriminative power of different anthropometric in- were considered as having a positive family history of dices for the prediction of incident T2D. The equality of T2D. Education level was classified in 3 categories: i) the area under the ROC curves (AUCs) of different an- those who had studied less than 6 years, ii) those who thropometric indices was tested using the Stata command had studied for 6–12 years, and iii) those with more than ‘roccomp’ . Using the “R optimal cut-point package”, 12 years of education. In accordance with the definition  cut-points for each variable were estimated by i) the provided by the American Diabetes Association,  maximum value of Youden’s index i.e., sensitivity+specifi- participants were considered to have T2D if they met at city-1,  ii) setting the sensitivity at 75%. least one of these criteria: FPG ≥7.0 mmol/L, or To examine the joint effects of BMI-WC, BMI-WHtR, 2 h-PCPG ≥11.1 mmol/L or taking anti-diabetic medica- BMI-WHR, and BMI-HC on the hazard of T2D in each tion. In addition, participants with missing data on sex, combined variables were created. To compare 4 cat- 2 h-PCPG at follow-up who simultaneously had FPG egories, considering those with normal BMI-WC, levels< 5.05 mmol/L were considered free of T2D . BMI-WHtR, BMI-WHR, and BMI-HC as reference groups, using our cut-points derived from the fixed sen- Statistical methods sitivity at 75%, Cox proportional hazard models with age All statistical analyses were stratified by sex. Continuous as the time scale were used in univariate and adjusted variables were described as mean (standard deviation multivariable models. Model 1 was adjusted for educa- (SD)) or median (interquartile range (IQR)), and categor- tion, family history of T2D, SBP, FPG, and TG/HDL-C. ical variables were summarized as frequency (percent- Model 2 was adjusted for variables included in model 1 age). Baseline characteristics of men and women, as well and wrist circumference. All analyses were done using as respondent and non-respondent groups, were com- Stata (version 12.0) and R (version 3.4.3). pared using independent T-test, Mann-Whitney U test, and Chi-square test whichever indicated. The event date Results for diabetes cases was described as the middle-time be- The study population consisted of 2988 men and 4029 tween the date of follow-up visit at which diabetes was women with the mean (SD) ages of 37.8 (10.2) and 37.3 detected for the first time, and the most recent (10.4) years, respectively. Baseline characteristics of re- follow-up visit preceding the diagnosis; the follow-up spondents and non-respondents (eligible participants time was considered as the difference between the calcu- whose baseline or follow-up data were missing) are lated mid-time date and the date at which the subjects shown in Additional file 2. The only significant differ- entered the study. For the censored subjects, the survival ence between these two groups was that respondents time was the interval between their first and last obser- were 4 years older than their non-respondent counter- vation dates. parts. All other variables were generally the same in both Cox proportional hazard models with age as the time groups. Table 1 depicts baseline characteristics of re- scale  were used to estimate the hazard ratios (HRs) spondent men and women. Men were older and had with 95% confidence intervals (95% CIs) for incident higher levels of education, wrist circumference, SBP, T2D per 1 SD increase in anthropometric indices in uni- FPG and log TG/HDL-C. Regarding anthropometric in- variate (without adjustment for any other variables) and dices, men had higher WC and WHR, whereas, BMI, adjusted multivariable models. In model 1, the hazard HC, and WHtR were higher in women. Frequency of ratio of interest was adjusted for education, family his- baseline consumption of angiotensin-converting enzyme tory of T2D, SBP, FPG, and TG/HDL-C. In model 2, the inhibitors (ACEIs), diuretics, corticosteroids and hazard ratio of interest was adjusted for variables in- lipid-lowering drugs in men were 0.6, 0.5, 1.5, and 1.4%, cluded in model 1 plus wrist circumference as it has Zafari et al. BMC Public Health (2018) 18:691 Page 4 of 12 Table 1 Baseline characteristics of respondents, Tehran Lipid and Glucose Study (1999–2015) Men (N = 2988) Women (N = 4029) Difference (CI) Age (years) 37.8 (10.2) 37.3 (10.4) 0.5 (0.0;0.1) Family History of T2D; No. (%) 790 (26.4) 1115 (27.7) −1.2% (−3.3;0.8) Education Level; No. (%) < 6 years 471 (15.8) 1127 (28.0) −12.2% (− 14.1; − 10.3) 6–12 years 1888 (63.2) 2402 (59.6) 3.6% (1.2;5.9) > 12 years 629 (21.1) 500 (12.4) 8.6% (7.8; 10.4) Wrist Circumference (cm) 17.6 (1.0) 15.9 (1.0) 1.7 (1.6; 1.7) WC (cm) 88.4 (11.1) 86.0 (12.3) 2.4 (1.8;2.9) Height (cm) 171.1 (6.6) 157.4 (5.8) 13.7 (13.4;14.0) WHtR 0.51 (0.06) 0.54 (0.08) −0.03 (−0.033; −0.026) BMI (kg/m ) 25.7 (4.1) 27.2 (4.9) −1.5 (−1.7; −1.3) HC(cm) 96.8 (7.2) 103.9 (9.5) −7.1 (−7.6; −6.7) WHR 0.91 (0.06) 0.82 (0.07) 0.08 (0.08;0.09) SBP (mmHg) 116.0(14.1) 113.5 (15.6) 2.5 (1.8;3.3) DBP (mmHg) 76.6 (10.3) 75.8 (10.3) 0.8 (0.3;1.3) FPG (mmol/L) 5.00 (0.50) 4.91 (0.51) 0.09 (0.07;1.2) Log TG/HDL-C 0.60 (0.71) 0.26 (0.69) 0.37 (0.34;0.40) For continuous variables, values are presented as mean (SD) and difference (95% CI) was estimated using linear regression models. Categorical variables are presented as frequency (percentage) and difference (95% CI) was estimated by logistic regression CI confidence interval, T2D type 2 diabetes mellitus, WC waist circumference, WHtR waist to height ratio, BMI body mass index, HC hip circumference, WHR waist to hip ratio, SBP systolic blood pressure, DBP diastolic blood pressure, FPG fasting plasma glucose, TG/HDL-C triglyceride to high density lipoprotein cholesterol ratio respectively; the corresponding values for women were sexes, WHtR had the highest AUC (0.69, 95% CI: 0.7, 1.9, 1.9, and 2%, respectively. 0.67–0.72, in men and 0.75, 95% CI: 0.73–0.78, in During a median follow-up (IQR) of 11.9 (4.6) women), whereas HC had the lowest values (0.62, 95% years, 354 new cases of T2D in men and 490 ones in CI: 0.59–0.65, in men and 0.66, 95% CI: 0.64–0.69, in women were detected resulting in an annual crude women). incidence rate (95% CI) of 10.9 (9.8–12.1) and 11.1 Table 3 shows the sex-specific cut-points of different (10.1–12.1) diabetes per 1000 person-years of anthropometric indices for the prediction of incident follow-up in men and women, respectively. Table 2 T2D. In men, fixing sensitivity at 75%, the calculated illustrates sex-specific adjusted HRs (95% CIs) for cut-points for BMI, WC, WHtR, WHR, and HC were incident T2D per 1 SD increase in anthropometric 25.56 kg/m , 89 cm, 0.52, 0.91, and 96 cm while the indices using the univariate and multivariable-adjusted corresponding values for the Youden’s index were Cox proportional hazard models. In men, in line with 26.49 kg/m , 87 cm, 0.54, 0.92, and 96 cm, respectively. the univariate model, model 1 indicated statistically In women, fixing sensitivity at 75%, the calculated significant associations between all anthropometric cut-points for BMI, WC, WHtR, WHR, and HC were measures and T2D incidence with HRs ranging from 27.12 kg/m , 87 cm, 0.56, 0.83, and 103 cm while the 1.25 for HC to 1.37 for BMI. Considering the wrist corresponding values for the Youden’s index were circumference as a surrogate of body frame in model 2, 29.27 kg/m , 91 cm, 0.56, 0.83, and 106 cm, respectively. the hazardous association of all the obesity measures for Table 4, illustrates the joint effects of BMI and other development of T2D was shown except for HC (HR 0.92, obesity measures on the hazard of developing T2D, 95% CI: 0.80–1.06). In women, similar to the univariate based on the cut-points derived from the fixed sensitivity model, model 1 indicated statistically significant hazard- at 75%, in men. As shown in the table, high BMI, ous associations between all anthropometrics and incident whether alone or combined with high other obesity T2D, with HRs ranging from 1.24 for HC to 1.51 for measures, had a significant risk for incident T2D both in WHtR. However, after adjusting for wrist circumference univariate and multivariate models (except for high in model 2, all the HRs decreased but still remained statis- BMI-low HC). Among men population with normal tically significant. BMI, only the presence of high WHR showed significant Fig. 1 shows sex-specific ROC curves and the AUCs risk for incident T2D in the fully adjusted analysis (HR (95% CIs) for different anthropometric indices. In both 1.55, 95% CI: 1.01–2.38). Zafari et al. BMC Public Health (2018) 18:691 Page 5 of 12 Table 2 Associations between anthropometric indices and incident type 2 diabetes, Tehran Lipid and Glucose Study (1999–2015) b c Univariate model Model 1 Model 2 HR (95% CI) P-value HR (95% CI) P-value HR (95% CI) P-value Men BMI (kg/m ) 1.69 (1.54–1.85) < 0.001 1.37 (1.24–1.51) < 0.001 1.48 (1.31–1.68) < 0.001 WC (cm) 1.69 (1.52–1.87) < 0.001 1.35 (1.21–1.50) < 0.001 1.43 (1.26–1.64) < 0.001 WHtR 1.66 (1.51–1.83) < 0.001 1.36 (1.23–1.50) < 0.001 1.39 (1.24–1.55) < 0.001 WHR 1.52 (1.38–1.67) < 0.001 1.27 (1.14–1.41) < 0.001 1.26 (1.13–1.40) < 0.001 HC (cm) 1.52 (1.38–1.67) < 0.001 1.25(1.14–1.37) < 0.001 0.92(0.80–1.06) 0.304 Women BMI (kg/m ) 1.74 (1.60–1.89) < 0.001 1.42 (1.30–1.56) < 0.001 1.41 (1.27–1.57) < 0.001 WC (cm) 1.90 (1.74–2.08) < 0.001 1.48 (1.34–1.64) < 0.001 1.46 (1.30–1.64) < 0.001 WHtR 1.93 (1.76–2.12) < 0.001 1.51 (1.37–1.68) < 0.001 1.47 (1.32–1.64) < 0.001 WHR 1.57 (1.44–1.71) < 0.001 1.30 (1.19–1.42) < 0.001 1.27 (1.16–1.39) < 0.001 HC (cm) 1.46 (1.35–1.58) < 0.001 1.24(1.14–1.35) < 0.001 1.11(1.00–1.24) 0.043 Per 1 SD increase for each index (SD of WC = 11.13, BMI = 4.08, HC = 7.16, WHR = 0.06, WHtR = 0.06 in men and WC = 12.31, BMI = 4.90, HC = 9.51, WHR = 0.07, WHtR = 0.08 in women) Adjusted for age (in time-scale manner), education level, family history of type 2 diabetes, systolic blood pressure, fasting plasma glucose, and triglyceride to high-density lipoprotein cholesterol ratio Adjusted for model 1 variables and wrist circumference HR hazard ratio, CI confidence interval, BMI body mass index, WC waist circumference, WHtR waist to height ratio, WHR waist to hip ratio, HC hip circumference Table 5, illustrates the joint effects of BMI and other power, in both sexes, WHtR performed the best, whereas obesity measures on the hazard of incident T2D, based on HC showed the lowest prediction power. The derived the derived cut-points from the fixed sensitivity at 75%, in cut-points of anthropometric indices for predicting the women. Analyses revealed that high obesity measures development of T2D in our population were generally whether alone or in combination with each other were different for men and women. Investigating the accuracy generally associated with higher risk of developing T2D of combination of BMI and other obesity measures in compared to the reference group. However, there were predicting development of T2D revealed that in both two exceptions: i) in those with high BMI-normal WHtR sexes, the combined high values of general and central whose increased risk was not statistically significant in the obesity indices increase the specificity for the price of multivariate models and ii) in those with normal reduced sensitivity and PPV. BMI-high HC whose HRs were not significantly increased In literature, there is almost unanimous agreement on neither in univariate nor in multivariate models. the association of all general and central obesity measures Table 6 depicts the sensitivity, specificity and positive with incident T2D in both sexes [15, 17, 19, 20, 22, 37–39]. predictive value (PPV) of combinations of high BMI- However, controversy remains as to which index can pre- high WC, high BMI- high WHtR, high BMI- high WHR, dict incident T2D, independent of other obesity variables. and high BMI- high HC using the derived cut-points by While several studies suggest central obesity measures as fixing sensitivity at 75% in men and women. As shown the main obesity indices predicting T2D in men, [38–40]. in the table, all the sensitivities fell to values under 70%, there are investigations highlighting the role of general whereas the reported specificities were increased com- obesity as the optimal index [14, 15]. In line with other pared to the values displayed in Table 3. The overall studies, we found that in both men and women, all an- effect of combining high measures together was a mild thropometric indices are almost similarly associated with reduction in the PPVs. T2D. Moreover, in our study, while wrist circumference acted as a positive risk factor for developing T2D in Discussion women, in men it showed a negative association. Similarly, Investigated in a large Iranian cohort study with 12 years Jahangiri et al. showed the positive association between of follow-up, all anthropometric variables showed sig- wrist circumference and incident T2D in women . nificant association with incident T2D after adjustment With regards to HC, similar to other studies, [41–43] for a wide set of covariates including a surrogate of body we showed the hazardous association of larger HC with frame (wrist circumference) in both sexes, excluding hip development of T2D in both sexes in the multivariate circumference in men. Assessing their discriminatory analysis adjusted for traditional T2D risk factors; the Zafari et al. BMC Public Health (2018) 18:691 Page 6 of 12 Fig. 1 Receiver operating characteristic (ROC) curves, area under the curves (AUCs) and 95% confidence intervals for different anthropometrics by gender. P-values for AUCs comparison in men: BMI-WC: 0.51; BMI-WHR: 0.84; BMI-WHtR: 0.11; BMI-HC: < 0.001; WC-WHR: 0.80; WC-WHtR: 0.002; WC-HC: < 0.001; WHtR-HC: < 0.001; WHR-WHtR: 0.10; WHR-HC: < 0.001. P-values for AUCs comparison in women: BMI-WC: 0.002; BMI-WHR: 0.68; BMI-WHtR: < 0.001; BMI-HC: < 0.001; WC-WHR: 0.001; WC-WHtR: 0.004; WC-HC: < 0.001; WHtR-HC: < 0.001; WHR-WHtR: < 0.001; WHR-HC: 0.003. WHtR, waist to height ratio; WC, waist circumference; WHR, waist to hip ratio; BMI, body mass index association which reached to null after considering body process of developing T2D. Gluteal muscles are one of frame (wrist circumference) in men. Moreover, further the main sites of insulin receptors. Therefore, higher adjustment of HC for BMI, as suggested by many inves- gluteal muscle mass may be indicative reduced risk of tigators, [44, 45] reversed this association only in women insulin resistance which is commonly followed by devel- (HR 0.77, 95% CI: 0.65–0.92 in women, HR 0.82 95% C: opment of T2D. The effect of gluteofemoral fat mass on 0.66–1.02 in men). Obesity is linked to higher risk of metabolism has been investigated in physiological T2D . However, there are studies which has reported studies [51, 52]. A study by Manolopoulos et al  a better metabolic profile in individuals who had more proposed that the fat mass in the gluteofemoral region gluteofemoral mass for a given amount of abdominal fat traps the surfeit serum fatty acids which results in a low [47–49]. The possible explanation for such effect is that serum lipid levels. Furthermore, authors suggested that HC is a presentation of several components namely the secreting adipokines leptin and adiponectin, the gluteo- bone, the gluteal muscles, and the gluteal subcutaneous femoral fat mass might play a protective role in the fat . Each component plays its own role in the development of T2D. Finally, Kuk et al. showed that in Zafari et al. BMC Public Health (2018) 18:691 Page 7 of 12 Table 3 Sex-specific diagnostic test performance of the anthropometric indices for incident type 2 diabetes, TLGS study Cut-point No. of subjects above cut-point No. of Incident T2D Sensitivity (%) Specificity (%) PPV (%) Men (N = 2419) BMI (kg/m ) Youden’s index 26.49 1043 243 67.7 61.4 23.3 Sensitivity = 75% 25.56 1258 269 74.9 52.0 21.4 WC (cm) Youden’s index 87.00 1415 292 81.3 45.5 20.6 Sensitivity = 75% 89.00 1254 265 73.8 52.0 21.1 WHtR Youden’s index 0.54 940 224 62.7 65.0 23.8 Sensitivity = 75% 0.52 1229 262 74.9 51.4 21.3 WHR Youden’s index 0.92 1122 246 71.9 55.5 21.9 Sensitivity = 75% 0.91 1247 267 75.2 50.3 21.4 HC (cm) Youden’s index 96.00 1418 268 74.6 44.2 18.9 Sensitivity = 75% 96.00 1418 268 74.6 44.2 18.9 Women (N = 3319) BMI (kg/m ) Youden’s index 29.27 1080 300 60.4 72.4 27.8 Sensitivity = 75% 27.12 1619 373 75.0 55.8 23.0 WC (cm) Youden’s index 91.00 1166 331 66.6 70.4 28.4 Sensitivity = 75% 87.00 1557 376 75.6 58.2 24.1 WHtR Youden’s index 0.56 1365 366 73.6 64.8 26.8 Sensitivity = 75% 0.56 1414 373 75.0 63.1 26.4 WHR Youden’s index 0.83 1502 358 73.0 59.1 23.8 Sensitivity = 75% 0.83 1589 371 75.0 56.7 23.3 HC (cm) Youden’s index 106.00 1338 302 60.8 63.3 22.6 Sensitivity = 75% 103.00 1833 367 73.8 48.0 20.0 T2D type 2 diabetes mellitus, PPV positive predictive value, BMI body mass index, WC waist circumference, WHtR waist to height ratio, WHR waist to hip ratio, HC hip circumference both sexes, for a given WC, higher HC is associated with long-term follow up studies might be due to different higher gluteofemoral and abdominal subcutaneous fat follow-up times, statistical approaches and covariates. Re- mass and skeletal muscle while being associated with cently, a systematic review and meta-analysis by Ashwell lower visceral fat mass . et al  revealed a stronger association of WHtR with Regarding the discriminatory power of anthropometric T2D rather than BMI in both genders. measures, in both sexes, among all the investigated an- In the current study, we looked for cut-points of thropometric measures, WHtR had the highest prediction different anthropometric measures using Youden’s power, whereas HC had the lowest. Results of the current index, which gives equal weight to both the sensitivity study are in line with several cohort studies [17, 22]. Previ- and specificity. Clinically, however, these may not be ously, in short-term follow-up, we showed the superiority the appropriate cut-points since sensitivity versus of WHtR to BMI in the prediction of T2D in both genders specificity must be weighed against the seriousness of [54, 55]. The differences between our short versus the disease, and the test under evaluation (whether it Zafari et al. BMC Public Health (2018) 18:691 Page 8 of 12 Table 4 Joint-effects of BMI and other central obesity measures on the hazard of type 2 diabetes in men, TLGS study a b Incident T2D /total participants Univariate model Model 1 Model 2 (%) HR (95% CI) P Value HR (95% CI) P-value HR (95% CI) P-value BMI (kg/m )-WC(cm) < 25.56- < 89 70/1302 (5.38) Reference – Reference – Reference – ≥ 25.56- < 89 24/191 (12.57) 2.23(1.40–3.56) 0.001 1.75(1.10–2.80) 0.018 1.83(1.14–2.95) 0.012 < 25.56- ≥ 89 19/184 (10.33) 1.54(0.93–2.57) 0.093 1.04(0.62–1.74) 0.879 1.07(0.64–1.81) 0.778 ≥ 25.56- ≥ 89 241/1311 (18.38) 3.13(2.39–4.09) < 0.001 2.21(1.67–2.91) < 0.001 2.38(1.75–3.25) < 0.001 BMI (kg/m )- WHtR < 25.56- < 0.52 73/1311 (5.57) Reference – Reference – Reference – ≥ 25.56- < 0.52 24/202 (11.88) 2.09(1.32–3.32) 0.002 1.73(1.08–2.75) 0.020 1.81(1.13–2.92) 0.014 < 25.56- ≥ 0.52 16/175 (9.14) 1.18(0.68–2.04) 0.549 0.80(0.46–1.40) 0.444 0.80(0.46–1.39) 0.437 ≥ 25.56- ≥ 0.52 241/1300 (18.54) 2.98(2.28–3.88) < 0.001 2.08(1.58–2.74) < 0.001 2.20(1.64–2.97) < 0.001 BMI (kg/m )- WHR < 25.56- < 0.91 48/1116 (4.3) Reference – Reference – Reference – ≥ 25.56- < 0.91 44/354 (12.43) 2.89(1.92–4.36) < 0.001 2.08(1.37–3.15) 0.001 2.20(1.43–3.38) < 0.001 < 25.56- ≥ 0.91 41/370 (11.08) 2.05(1.34–3.13) 0.001 1.54(1.00–2.36) 0.047 1.55(1.01–2.38) 0.044 ≥ 25.56- ≥ 0.91 221/1148 (11.85) 3.87(2.82–5.31) < 0.001 2.70(1.95–3.74) < 0.001 2.88(2.03–4.08) < 0.001 BMI (kg/m2)- HC < 25.56- < 96 68/1147 (5.93) Reference – Reference – Reference – ≥ 25.56- < 96 23/158 (14.56) 1.98(1.23–3.18) 0.005 1.35(0.83–2.19) 0.224 1.41(0.86–2.30) 0.164 < 25.56- ≥ 96 21/339 (6.19) 1.03(0.63–1.69) 0.881 0.86(0.52–1.41) 0.564 0.92(0.55–1.51) 0.774 ≥ 25.56- ≥ 96 242/1344 (18.01) 2.92(2.23–3.83) < 0.001 2.15(1.63–2.84) < 0.001 2.40(1.75–3.30) < 0.001 Adjusted for age (in time-scale manner), education level, family history of type 2 diabetes, systolic blood pressure, fasting plasma glucose, and triglyceride to high-density lipoprotein cholesterol ratio Adjusted for model 1 variables and wrist circumference HR hazard ratio, CI confidence interval, BMI body mass index, WC waist circumference, WHtR waist to height ratio, WHR waist to hip ratio, HC hip circumference is a simple measurement or an invasive test and its IDF suggested to use WC ≥ 94 cm and ≥ 80 cm in cost). Hence, considering the coronary heart disease Middle-Eastern men and women, respectively . equivalency of T2D among Iranian population, in Comparison between the false positive rates using our line with the Inter99 Study, wedecided not to derived cut-points versus the ones suggested by IDF miss more than 25% of at-risk participants and fixed shows that our derived cut-point of 87 cm in women the sensitivity of all the anthropometric cut-points at performed better (0.42 versus 0.66, respectively), how- about 75%; the cut-points resulted in specificities ever, in men, IDF cut-points showed lower false posi- above 50% in all of the anthropometric measures ex- tive rates (0.47 and 0.30, respectively). But it should cept HC. Accordingly, for BMI, we recommend using be considered that the sensitivity of WC cut-point 25.56 kg/m as a predictive cut-point in men and suggested by IDF decreased to 54.2%; the issue which 27.12 kg/m in women. Data of the national is not appropriate for screening T2D . As a gen- non-communicable disease risk factors surveillance in eral finding in our study, when fixing the sensitivity Iran suggested cut-points close to the current study at 75%, there was not a noticeable difference in speci- 2 2 (24.8 kg/m in men and 26.3 kg/m in women) . ficity and PPV of the cut-points in men but in Assessing the accuracy of the ADA suggested cut-point women, WHtR suggested cut-point had a higher spe- of BMI ≥ 25 kg/m in our population, its sensitivity cificity and PPV compared to other indices. Focusing and specificity were 89.2 and 37.5% in women and 78.8 on WHtR, we suggest the cut-point of 0.52 in men and 46.7% in men, respectively which have noticeably and 0.56 in women, results that somewhat support lower specificities compared to our suggested cut-points. findings of a systematic review by Ashwell et al  Our recommended cut-points for WC are 89 cm in men to “keep your waist to less than half your height”. and 87 cm in women. Studies in different populations sug- However, our derived cut-points showed a lower false gest different WC cut-points for predicting T2D which positive rate both in men (0.46 versus 0.58) and sounds reasonable due to ethnic differences [23, 24]. women (0.38 versus 0.67). Zafari et al. BMC Public Health (2018) 18:691 Page 9 of 12 Table 5 Joint-effects of BMI and other central obesity measures on the hazard of type 2 diabetes in women, TLGS study a b Incident T2D /total participants Univariate model Model 1 Model 2 (%) HR (95% CI) P value HR (95% CI) P-value HR (95% CI) P-value BMI (kg/m )-WC(cm) < 27.12- < 87 77/1776(4.34) Reference – Reference – Reference – ≥ 27.12- < 87 44/389(11.31) 2.29(1.58–3.33) < 0.001 1.89(1.29–2.75) 0.001 1.79(1.22–2.61) 0.003 < 27.12- ≥ 87 46/309(14.86) 2.66(1.83–3.87) < 0.001 2.05(1.41–2.97) < 0.001 1.99(1.36–2.89) < 0.001 ≥ 27.12- ≥ 87 323/1555(20.77) 3.76(2.89–4.89) < 0.001 2.30(1.75–3.02) < 0.001 2.06(1.53–2.77) < 0.001 BMI (kg/m )- WHtR < 27.12- < 0.56 81/1848(4.38) Reference – Reference – Reference – ≥ 27.12- < 0.56 42/463(9.07) 1.86(1.27–2.70) 0.001 1.45(0.99–2.11) 0.053 1.36(0.92–1.99) 0.113 < 27.12- ≥ 0.56 42/237(17.72) 3.13(2.13–4.60) < 0.001 2.33(1.59–3.43) < 0.001 2.32(1.58–3.42) < 0.001 ≥ 27.12- ≥ 0.56 325/1481(21.94) 4.02(3.11–5.21) < 0.001 2.51(1.92–3.29) < 0.001 2.27(1.70–3.03) < 0.001 BMI (kg/m )- WHR < 27.12- < 0.83 46/1464(3.14) Reference – Reference – Reference – ≥ 27.12- < 0.83 85/703(12.09) 3.25(2.25–4.67) < 0.001 2.31(1.59–3.34) < 0.001 2.11(1.44–3.08) < 0.001 < 27.12- ≥ 0.83 77/621(12.40) 3.32(2.29–4.81) < 0.001 2.55(1.75–3.71) < 0.001 2.54(1.75–3.70) < 0.001 ≥ 27.12- ≥ 0.83 282/1241(22.72) 5.67(4.10–7.86) < 0.001 3.33(2.37–4.67) < 0.001 2.99(2.10–4.25) < 0.001 BMI (kg/m2)- HC < 27.12- < 103 92/1627 (5.65) Reference – Reference – Reference – ≥ 27.12- < 103 37/214 (17.29) 2.39(1.63–3.52) < 0.001 1.68(1.13–2.48) 0.009 1.56(1.05–2.31) 0.026 < 27.12- ≥ 103 31/458 (6.77) 1.09(0.72–1.64) 0.669 1.07(0.71–1.61) 0.721 1.02(0.67–1.53) 0.923 ≥ 27.12- ≥ 103 330/1730 (19.08) 2.66(2.10–3.38) < 0.001 1.79(1.40–2.30) < 0.001 1.58(1.20–2.07) 0.001 Adjusted for age (in time-scale manner), education level, family history of type 2 diabetes, systolic blood pressure, fasting plasma glucose, and triglyceride to high-density lipoprotein cholesterol ratio Adjusted for model 1 variables and wrist circumference HR hazard ratio, CI confidence interval, BMI body mass index, WC waist circumference, WHtR waist to height ratio, WHR waist to hip ratio, HC hip circumference Considering the limitation of the available data, differences in the sex-specific relevance of measures [19, 20, 58]the jointanalysesinthe current study of body fat distribution in predicting incident T2D extend the understanding of the combined influence . In women, our findings suggested that different of obesity measures on incident T2D. In line with combinations of BMI-WC, BMI-WHtR, BMI-WHR, our study, Meisinger et al. showed that there are and BMI-HC (except high BMI-low WHtR and low BMI-high HC) were associated with high risk of Table 6 Sex-specific Prediction Accuracy of the Combined High developing T2D. In men, we showed that high BMI either Anthropometric Indices for Incident Type 2 Diabetes, TLGS study in conjunction with high obesity indices or normal ones Sensitivity (%) Specificity (%) PPV (%) (except for normal HC) had a significant association with Men (N = 2419) incidence of T2D in future, whereas those with normal BMI ≥ 25.56- WC ≥ 89 68.1% 59.4% 18.4% BMI but high WHtR, WC or HC did not show a sig- BMI ≥ 25.56- WHtR ≥ 0.52 68.1% 59.8% 18.5% nificant risk for the development of T2D. On the other hand, some studies proposed that BMI, WC, BMI ≥ 25.56- WHR ≥ 0.91 62.4% 64.8% 19.3% and WHR have very similar predictive powers for in- BMI ≥ 25.56- HC ≥ 96 68.4% 58.2% 18.0% cident T2D [19, 20]. This difference may be explained Women (N = 3319) by variation in methodological approaches and the sample BMI ≥ 27.12- WC ≥ 87 65.9% 65.2% 20.8% size. It should be kept in mind that although from a statis- BMI ≥ 27.12- WHtR ≥ 0.56 66.3% 67.3% 21.9% tical point of view some of the anthropometric measures BMI ≥ 27.12- WHR ≥ 0.83 57.6% 72.9% 22.7% have a stronger association with T2D and can predict it better, from a clinical perspective we cannot say that one BMI ≥ 27.12- HC ≥ 103 67.3% 60.4% 19.1% obesity measure overweighs the other. It would be rational BMI body mass index, WC waist circumference, WHtR waist to height ratio, WHR waist to hip ratio, HC hip circumference, PPV positive predictive value to ask physicians to refer those with higher than normal Zafari et al. BMC Public Health (2018) 18:691 Page 10 of 12 obesity indices no matter which index is high. Further- than the traditionally used cut-points, leading to a con- more, it seems that combined high values of general and siderable reduction in false positive rates. Therefore, they central obesity indices increase the specificity for the price might perform better if applied in the Iranian popula- of reduced sensitivity and PPV in predicting incident T2D. tion. Further studies in other Iranian populations are re- Considering limitations, our results cannot be ex- quired to check the external validity of our proposed trapolated to elder populations since participants of cut-points in predicting development of T2D. our study are adults, 20–60 years of age. In addition, excluding participants affected by conditions such as Additional files autoimmune diseases and acute or chronic infections which are known to impair glucose tolerance was not Additional file 1: The selection process of study sample to determine optimal cut-points for prediction of type 2 diabetes. Figure legend: The possible at baseline examination; however, those suf- hatched (those who were free of T2D in the last follow-up examination, fering from cancer, end-stage renal disease or cirrho- despite not participating in one or more follow-ups) and grey (those who sis were excluded. Likewise, we could not collect data developed T2D in each of the follow-up examinations) boxes indicate study sample.T2D, type 2 diabetes mellitus; Lost, lost to follow-up. (PDF 2766 kb) regarding the use of estroprogestins in our partici- Additional file 2: Baseline Characteristics of Respondents and Non- pants, nevertheless, use of other drugs that have the respondents, Tehran Lipid and Glucose Study (1999–2015). Table potential to affect glucose impairment was carefully comparing baseline characteristics of respondents and non-respondents observed. Also, as shown in Table 1, respondent par- in the study. (DOCX 14 kb) ticipants in our study generally had significantly higher levels of risk factors which might lead to over- Abbreviations estimation of T2D incidence. These statistically 2 h-PCPG: 2 h-post-challenge plasma glucose; 95% CIs: 95% confidence intervals; AUC: Area under the curve; BMI: Body mass index; DBP: Diastolic significant differences might be attributed to large blood pressure; FPG: Fasting plasma glucose; HDL-C: High-density lipoprotein study sample size and they are not significant from cholesterol; LMICs: Low- and middle-income countries; ROC: Receiver the clinical point of view (e.g. mean of WHtR = 0.51 operating characteristics; SBP: Systolic blood pressure; SD: Standard deviation; T2D: Type 2 diabetes; TG: Triglycerides; TLGS: Tehran Lipid and Glucose and 0.50 in respondents and non-respondents men, Study; WC: Waist circumference; WHR: Waist to hip; WHtR: Waist to height respectively). Moreover, almost all of the risk factors ratios were adjusted in the multivariable models. Finally, un- fortunately, HbA1C was not measured in TLGS par- Acknowledgments We would like to acknowledge Ms. N Shiva of the Research Institute for ticipants; hence, we could not use HbA1c as a Endocrine Sciences for critical editing of English grammar and syntax of the criterion for defining diabetic patients. manuscript. This article has been extracted from the thesis written by As for the strengths, our study included a large repre- Mrs. N Zafari in the School of Medicine, Shahid Beheshti University of Medical Sciences. (Registration No: 259). sentative sample of Iranian adults in a sex-stratified population with reliable follow-up data. We had a rea- Funding sonable number of events which allowed us to evaluate This study was supported by Grant No.121 from the National Research the long-term effects of anthropometric indices on T2D Council of the Islamic Republic of Iran. incidence. Also, we used both FPG and 2 h-PCPG as in- Availability of data and materials dicators of diabetes status both at baseline and follow-up The datasets used and/or analysed during the current study are available examinations allowing us to have an accurate estimation from the corresponding author on reasonable request. of incident T2D. Authors’ contributions NZ, FH, FA, conceptualized the research question. NZ, ML, MAM, DK, FH, Conclusions planned the methodological aspects.NZ, ML, MAM, DK, FH, PHM, conducted In conclusion, all anthropometric variables showed the statistical analysis.NZ, ML, wrote the original draft.NZ, ML, MAM, DK, FA, significant association with incident T2D considering a FH, reviewed and edited the draft. MAM, DK, FA, FH, supervised the project. DK, FA, FH, administered the project. FA, FH, acquired funding for the wide set of covariates including wrist circumferencein project. All authors read and approved the final manuscript. both sexes, except hip circumference in men. We showed that among all the obesity indices, in both sexes, Ethics approval and consent to participate WHtR performed the best while HC had the lowest pre- All subjects filled a written consent after being informed about the general aspects of the work and the study was approved by the Ethical Committee diction power. Fixing sensitivity at about 75%, not to of Research Institute for Endocrine Sciences. miss more than 25% of at-risk individuals for develop- ment of T2D in long term, our derived cut-points for Competing interests BMI, WC, WHtR, WHR, and HC in the Iranian popula- The authors declare that they have no competing interests. tion were 25.56 kg/m , 89 cm, 0.52, 0.91, and 96 cm in men and 27.12 kg/m , 87 cm, 0.56, 0.83, and 103 in Publisher’sNote women, respectively. Our derived cut-points for BMI Springer Nature remains neutral with regard to jurisdictional claims in and WHtR in both sexes, and WC in women are higher published maps and institutional affiliations. Zafari et al. BMC Public Health (2018) 18:691 Page 11 of 12 Author details different adiposity measures: a 7 year prospective study in 6,923 older men Prevention of Metabolic Disorders Research Center, Research Institute for and women. Diabetologia. 2010;53(5):890–8. Endocrine Sciences, Shahid Beheshti University of Medical Sciences, Tehran, 18. Huerta JM, Tormo M-J, Chirlaque M-D, Gavrila D, Amiano P, Arriola L, Iran. Non-Communicable Disease Control, School of Population and Global Ardanaz E, Rodríguez L, Sánchez M-J, Mendez M. Risk of type 2 diabetes Health, University of Melbourne, Melbourne, VIC, Australia. Department of according to traditional and emerging anthropometric indices in Spain, a Epidemiology and Biostatistics, School of Public Health, Tehran University of Mediterranean country with high prevalence of obesity: results from a Medical Sciences, Tehran, Iran. Endocrine Research Center, Research Institute large-scale prospective cohort study. BMC Endocr Disord. 2013;13(1):7. for Endocrine Sciences, Shahid Beheshti University of Medical Sciences, 19. Meisinger C, Döring A, Thorand B, Heier M, Löwel H. Body fat distribution Tehran, Iran. and risk of type 2 diabetes in the general population: are there differences between men and women? The MONICA/KORA Augsburg cohort study. Received: 16 July 2017 Accepted: 25 May 2018 Am J Clin Nutr. 2006;84(3):483–9. 20. Wang Y, Rimm EB, Stampfer MJ, Willett WC, Hu FB. Comparison of abdominal adiposity and overall obesity in predicting risk of type 2 diabetes among men. Am J Clin Nutr. 2005;81(3):555–63. 21. Jia Z, Zhou Y, Liu X, Wang Y, Zhao X, Wang Y, Liang W, Wu S. Comparison References of different anthropometric measures as predictors of diabetes incidence in 1. Murray C, Vos T, Lozano R, AlMazroa MA, Memish ZA: Disability-adjusted life a Chinese population. Diabetes Res Clin Pract. 2011;92(2):265–71. years (DALYs) for 291 diseases and injuries in 21 regions, 1990-2010: a 22. Xin Z, Liu C, Niu W-Y, Feng J-P, Zhao L, Ma Y-H, Hua L, Yang J-K. Identifying systematic analysis for the global burden of disease study 2010 obesity indicators which best correlate with type 2 diabetes in a Chinese (vol 380, pg 2197, 2012) Lancet 2013, 381(9867):628–628. population. BMC Public Health. 2012;12(1):732. 2. Haffner SM, Lehto S, Rönnemaa T, Pyörälä K, Laakso M. Mortality from coronary heart disease in subjects with type 2 diabetes and in nondiabetic 23. Razak F, Anand SS, Shannon H, Vuksan V, Davis B, Jacobs R, Teo KK, subjects with and without prior myocardial infarction. N Engl J Med. 1998; McQueen M, Yusuf S. Defining obesity cut points in a multiethnic 339(4):229–34. population. Circulation. 2007;115(16):2111–8. 24. EC WHO. Appropriate body-mass index for Asian populations and its 3. Juutilainen A, Lehto S, Rönnemaa T, Pyörälä K, Laakso M. Type 2 diabetes as implications for policy and intervention strategies. Lancet. 2004;363(9403): a “coronary heart disease equivalent” an 18-year prospective population- based study in Finnish subjects. Diabetes Care. 2005;28(12):2901–7. 25. Esteghamati A, Etemad K, Koohpayehzadeh J, Abbasi M, Meysamie A, 4. Hadaegh F, Fahimfar N, Khalili D, Sheikholeslami F, Azizi F. New and known Noshad S, Asgari F, Mousavizadeh M, Rafei A, Khajeh E. Trends in the type 2 diabetes as coronary heart disease equivalent: results from 7.6 year prevalence of diabetes and impaired fasting glucose in association with follow up in a Middle East population. Cardiovasc Diabetol. 2010;9(1):84. obesity in Iran: 2005–2011. Diabetes Res Clin Pract. 2014;103(2):319–27. 5. IDF Diabetes Atlas. In. Brussels, Belgium: International Diabetes Federation: International Diabetes Federation; 2013. 26. Talaei M, Sadeghi M, Marshall T, Thomas G, Iranipour R, Nazarat N, 6. Merlotti C, Morabito A, Ceriani V, Pontiroli AE. Prevention of type 2 diabetes Sarrafzadegan N. Anthropometric indices predicting incident type 2 in obese at-risk subjects: a systematic review and meta-analysis. Acta diabetes in an Iranian population: the Isfahan cohort study. Diabetes & Diabetol. 2014;51(5):853–63. metabolism. 2013;39(5):424–31. 27. Azizi F, Ghanbarian A, Momenan AA, Hadaegh F, Mirmiran P, Hedayati M, 7. Bell J, Kivimaki M, Hamer M. Metabolically healthy obesity and risk of Mehrabi Y, Zahedi-Asl S. Prevention of non-communicable disease in a incident type 2 diabetes: a meta-analysis of prospective cohort studies. population in nutrition transition: Tehran lipid and glucose study phase II. Obes Rev. 2014;15(6):504–15. Trials. 2009;10(1):5. 8. Cameron AJ, Dunstan DW, Owen N, Zimmet PZ, Barr EL, Tonkin AM, 28. Azizi F, Madjid M, Rahmani M, Emami H, Mirmiran P, Hadjipour R. Tehran Magliano DJ, Murray SG, Wellborn TA, Shaw JE. Health and mortality lipid and glucose study (TLGS): rationale and design. Iranian J Endocrinol consequences of abdominal obesity: evidence from the AusDiab study. and Metab. 2000;2(2):77–86. Med J Aust. 2009;191(4):202–8. 29. Association AD. 2. Classification and diagnosis of diabetes. Diabetes Care. 9. Freemantle N, Holmes J, Hockey A, Kumar S. How strong is the association 2017;40(Supplement 1):S11–24. between abdominal obesity and the incidence of type 2 diabetes? Int J Clin Pract. 2008;62(9):1391–6. 30. Bozorgmanesh M, Hadaegh F, Saadat N, Azizi F. Fasting glucose cutoff 10. Derakhshan A, Sardarinia M, Khalili D, Momenan AA, Azizi F, Hadaegh F. Sex point: where does the risk terminate? Tehran lipid and glucose study. Acta specific incidence rates of type 2 diabetes and its risk factors over 9 years of Diabetol. 2012;49(5):341–8. follow-up: Tehran Lipid and Glucose Study. PloS one. 2014;9(7):e102563. 31. Pencina MJ, Larson MG, D'Agostino RB. Choice of time scale and its effect on significance of predictors in longitudinal studies. Stat Med. 11. Després J-P. Body fat distribution and risk of cardiovascular disease an 2007;26(6):1343–59. update. Circulation. 2012;126(10):1301–13. 32. Jahangiri Noudeh Y, Hadaegh F, Vatankhah N, Momenan AA, Saadat N, 12. Kodama S, Horikawa C, Fujihara K, Heianza Y, Hirasawa R, Yachi Y, Sugawara Khalili D, Azizi F. Wrist circumference as a novel predictor of diabetes and A, Tanaka S, Shimano H, Iida KT. Comparisons of the strength of associations prediabetes: results of cross-sectional and 8.8-year follow-up studies. with future type 2 diabetes risk among anthropometric obesity indicators, J Clin Endocrinol Metab. 2013;98(2):777–84. including waist-to-height ratio: a meta-analysis. Am J Epidemiol. 2012; 176(11):959–69. 33. Verbeke G. Linear mixed models for longitudinal data. In: Linear mixed 13. Ashwell M, Gunn P, Gibson S. Waist-to-height ratio is a better screening tool models in practice: Springer; 1997. p. 63–153. than waist circumference and BMI for adult cardiometabolic risk factors: 34. Cleves MA, Rock L. From the help desk: comparing areas under receiver systematic review and meta-analysis. Obes Rev. 2012;13(3):275–86. operating characteristic curves from two or more probit or logit models. 14. Chei C-L, Iso H, Yamagishi K, Tanigawa T, Cui R, Imano H, Kiyama M, Stata J. 2002;2(3):301–13. Kitamura A, Sato S, Shimamoto T. Body fat distribution and the risk of 35. Lopez-Raton M, Rodrıguez-Alvarez MX, Cadarso-Suárez C, Gude-Sampedro F. hypertension and diabetes among Japanese men and women. Hypertens OptimalCutpoints: an R package for selecting optimal Cutpoints in Res. 2008;31(5):851. diagnostic tests. J Stat Softw. 2014;61(8):1–36. 15. Tulloch-Reid MK, Williams DE, Looker HC, Hanson RL, Knowler WC. Do 36. Perkins NJ, Schisterman EF. The inconsistency of “optimal” cutpoints measures of body fat distribution provide information on the risk of type 2 obtained using two criteria based on the receiver operating characteristic diabetes in addition to measures of general obesity? Comparison of curve. Am J Epidemiol. 2006;163(7):670–5. anthropometric predictors of type 2 diabetes in pima Indians. Diabetes 37. Vazquez G, Duval S, Jacobs DR, Silventoinen K. Comparison of body mass Care. 2003;26(9):2556–61. index, waist circumference, and waist/hip ratio in predicting incident 16. Glümer C, Carstensen B, Sandbæk A, Lauritzen T, Jørgensen T, Borch- diabetes: a meta-analysis. Epidemiol Rev. 2007;29(1):115–28. Johnsen K. A Danish diabetes risk score for targeted screening the Inter99 38. Schulze MB, Heidemann C, Schienkiewitz A, Bergmann MM, Hoffmann K, study. Diabetes Care. 2004;27(3):727–33. Boeing H. Comparison of anthropometric characteristics in predicting the 17. Wannamethee S, Papacosta O, Whincup P, Carson C, Thomas M, Lawlor D, incidence of type 2 diabetes in the EPIC-Potsdam study. Diabetes Care. Ebrahim S, Sattar N. Assessing prediction of diabetes in older adults using 2006;29(8):1921–3. Zafari et al. BMC Public Health (2018) 18:691 Page 12 of 12 39. Borné Y, Nilsson PM, Melander O, Hedblad B, Engström G. Multiple anthropometric measures in relation to incidence of diabetes: a Swedish population-based cohort study. The European J Public Health. 2015:ckv044. 40. Wei M, Gaskill SP, Haffner SM, Stern MP. Waist circumference as the best predictor of noninsulin dependent diabetes mellitus (NIDDM) compared to body mass index, waist/hip ratio and other anthropometric measurements in Mexican Americans—a 7-year prospective study. Obes Res. 1997;5(1):16–23. 41. Parker ED, Pereira MA, Stevens J, Folsom AR. Association of hip circumference with incident diabetes and coronary heart disease: the atherosclerosis risk in communities study. Am J Epidemiol. 2009;169(7):837–47. 42. Conway B, Xiang Y-B, Villegas R, Zhang X, Li H, Wu X, Yang G, Gao Y-T, Zhang W, Shu X-O. Hip circumference and the risk of type 2 diabetes in middle-aged and elderly men and women: the shanghai women and shanghai men’s health studies. Ann Epidemiol. 2011;21(5):358–66. 43. Janghorbani M, Amini M. Associations of hip circumference and height with incidence of type 2 diabetes: the Isfahan diabetes prevention study. Acta Diabetol. 2012;49(1):107–14. 44. Seidell JC, Pérusse L, Després J-P, Bouchard C. Waist and hip circumferences have independent and opposite effects on cardiovascular disease risk factors: the Quebec family study. Am J Clin Nutr. 2001;74(3):315–21. 45. Snijder MB, Dekker JM, Visser M, Yudkin JS, Stehouwer CD, Bouter LM, Heine RJ, Nijpels G, Seidell JC. Larger thigh and hip circumferences are associated with better glucose tolerance: the Hoorn study. Obesity. 2003;11(1):104–11. 46. Chan M: Obesity and Diabetes: the Slow-motion Disaster Keynote Address at the 47th Meeting of the National Academy of Medicine In.; 2016. 47. Snijder MB, Dekker JM, Visser M, Bouter LM, Stehouwer CD, Yudkin JS, Heine RJ, Nijpels G, Seidell JC. Trunk fat and leg fat have independent and opposite associations with fasting and postload glucose levels: the Hoorn study. Diabetes Care. 2004;27(2):372–7. 48. Janghorbani M, Momeni F, Dehghani M. Hip circumference, height and risk of type 2 diabetes: systematic review and meta-analysis. Obes Rev. 2012;13(12):1172–81. 49. Cameron A, Magliano D, Söderberg S. A systematic review of the impact of including both waist and hip circumference in risk models for cardiovascular diseases, diabetes and mortality. Obes Rev. 2013;14(1):86–94. 50. Seidell J, Han T, Feskens EJ, Lean M. Narrow hips and broad waist circumferences independently contribute to increased risk of non-insulin- dependent diabetes mellitus. J Intern Med. 1997;242(5):401–6. 51. Arner P. Differences in lipolysis between human subcutaneous and omental adipose tissues. Ann Med. 1995;27(4):435–8. 52. Manolopoulos K, Karpe F, Frayn K. Gluteofemoral body fat as a determinant of metabolic health. Int J Obes. 2010;34(6):949. 53. Kuk JL, Janiszewski PM, Ross R. Body mass index and hip and thigh circumferences are negatively associated with visceral adipose tissue after control for waist circumference. Am J Clin Nutr. 2007;85(6):1540–4. 54. Hadaegh F, Shafiee G, Azizi F. Anthropometric predictors of incident type 2 diabetes mellitus in Iranian women. Annals of Saudi medicine. 2009;29(3):194. 55. Hadaegh F, Zabetian A, Harati H, Azizi F. Waist/height ratio as a better predictor of type 2 diabetes compared to body mass index in Tehranian adult men–a 3.6-year prospective study. 2006;114(6):310–5. 56. Shabnam A-A, Homa K, Reza M, Bagher L, Hossein FM, Hamidreza A. Cut-off points of waist circumference and body mass index for detecting diabetes, hypercholesterolemia and hypertension according to National non- Communicable Disease Risk Factors Surveillance in Iran. Arch Med Sci. 2012;8(4):614–21. 57. Alberti K, Eckel RH, Grundy SM, Zimmet PZ, Cleeman JI, Donato KA, Fruchart J-C, James WPT, Loria CM, Smith SC. Harmonizing the metabolic syndrome: a joint interim statement of the international diabetes federation task force on epidemiology and prevention; national heart, lung, and blood institute; American heart association; world heart federation; international atherosclerosis society; and international association for the study of obesity. Circulation. 2009;120(16):1640–5. 58. Folsom AR, Kushi LH, Anderson KE, Mink PJ, Olson JE, Hong C-P, Sellers TA, Lazovich D, Prineas RJ. Associations of general and abdominal obesity with multiple health outcomes in older women: the Iowa Women's health study. Arch Intern Med. 2000;160(14):2117–28.
BMC Public Health – Springer Journals
Published: Jun 5, 2018
It’s your single place to instantly
discover and read the research
that matters to you.
Enjoy affordable access to
over 18 million articles from more than
15,000 peer-reviewed journals.
All for just $49/month
Query the DeepDyve database, plus search all of PubMed and Google Scholar seamlessly
Save any article or search result from DeepDyve, PubMed, and Google Scholar... all in one place.
Get unlimited, online access to over 18 million full-text articles from more than 15,000 scientific journals.
Read from thousands of the leading scholarly journals from SpringerNature, Elsevier, Wiley-Blackwell, Oxford University Press and more.
All the latest content is available, no embargo periods.
“Hi guys, I cannot tell you how much I love this resource. Incredible. I really believe you've hit the nail on the head with this site in regards to solving the research-purchase issue.”Daniel C.
“Whoa! It’s like Spotify but for academic articles.”@Phil_Robichaud
“I must say, @deepdyve is a fabulous solution to the independent researcher's problem of #access to #information.”@deepthiw
“My last article couldn't be possible without the platform @deepdyve that makes journal papers cheaper.”@JoseServera