Mortality prediction of a body shape index versus traditional anthropometric measures in an Iranian population: Tehran Lipid and Glucose Study

Accepted Manuscript Mortality prediction of a body shape index vs. traditional anthropometric measures in an Iranian population: Tehran Lipid and Glucose Study Mahsa Sardarinia, Roya Ansari, Feridoun Azizi, Farzad Hadaegh, Mohammadreza Bozorgmanesh PII:

S0899-9007(16)30075-2

DOI:

10.1016/j.nut.2016.05.004

Reference:

NUT 9773

To appear in:

Nutrition

Received Date: 4 November 2015 Revised Date:

30 April 2016

Accepted Date: 5 May 2016

Please cite this article as: Sardarinia M, Ansari R, Azizi F, Hadaegh F, Bozorgmanesh M, Mortality prediction of a body shape index vs. traditional anthropometric measures in an Iranian population: Tehran Lipid and Glucose Study, Nutrition (2016), doi: 10.1016/j.nut.2016.05.004. This is a PDF file of an unedited manuscript that has been accepted for publication. As a service to our customers we are providing this early version of the manuscript. The manuscript will undergo copyediting, typesetting, and review of the resulting proof before it is published in its final form. Please note that during the production process errors may be discovered which could affect the content, and all legal disclaimers that apply to the journal pertain.

ACCEPTED MANUSCRIPT

Mortality prediction of a body shape index vs. traditional anthropometric measures in an

2

Iranian population: Tehran Lipid and Glucose Study

3

Mahsa Sardarinia1, Roya Ansari1, Feridoun Azizi2, Farzad Hadaegh1, Mohammadreza Bozorgmanesh§1

6 7

Shahid Beheshti University of Medical Sciences, Tehran, Iran

2. Endocrine Research Center, Research Institute for Endocrine Sciences ,Shahid Beheshti University of Medical Sciences, Tehran, Iran

8

§Corresponding author

9

Mohammadreza Bozorgmanesh, MD P.O. Box: 19395-4763

11

Tehran, Islamic Republic of Iran

12

Phone: 98 21 22409301-5

13

Fax: 98 21 22402463

14

E-mail: [email protected]

AC C

EP

TE D

10

15

SC

5

1. Prevention of metabolic disorders research center, Research Institute for Endocrine Sciences,

M AN U

4

RI PT

1

1

ACCEPTED MANUSCRIPT

Abstract

17 18

Back ground

19

A body shape index (ABSI) based on waist circumference (WC) adjusted for height and weight

20

has been shown as to be a risk factor for premature mortality. We hypothesized that ABSI

21

predicts mortality hazard better than other anthropometric measures in an Iranian population.

22

Methods

23

Study population included 9 242 Iranian participants in Tehran, aged ≥30 years, followed for a

24

median 10 years. The risk of mortality was estimated by incorporating ABSI, body mass index

25

(BMI), WC, waist to hip ratio (WHpR), and waist to height ratio (WHtR), one at a time, into

26

multivariate models and also in terms of the effect size, calibration, discrimination, and added

27

predictive ability.

28 29

Results

30

We documented 487 deaths with the annual incidence rate of mortality per 1000 persons being

31

3.9 for women and 8.2 for men. ABSI was associated with all-cause mortality in a curvilinear

32

fashion. ABSI was more strongly associated with all-cause mortality than were BMI, WC and

33

WHtR. Among women, however, WHpR was observed to be a stronger predictor of all-cause

34

mortality than ABSI. Among both men and women, ABSI improved the risk classification based

35

on other anthropometric measures, the only exception being WHpR. None of the anthropometric

36

measures studied could add any value to the predictive ability of the Framingham’s general CVD

37

algorithm.

AC C

EP

TE D

M AN U

SC

RI PT

16

2

ACCEPTED MANUSCRIPT

Conclusion

40

ABSI was the strongest predictor of all-cause mortality among the anthropometric

41

measurements, except WHpR in women. When ABSI was added to the Framingham general

42

cardiovascular disease algorithm, it failed to improve the its predictive ability.

AC C

EP

TE D

M AN U

SC

RI PT

38 39

3

ACCEPTED MANUSCRIPT

Introduction:

45

Obesity is becoming a global epidemic in both children and adults, and is associated with

46

numerous co-morbidities such as cardiovascular diseases, type 2 diabetes, etc.[1].

RI PT

43 44

Magnetic resonance imaging (MRI),computed tomography (CT) and dual-energy X-ray

48

absorption are now considered the gold standard for the evaluation of visceral adipose tissue

49

(VAT) and subcutaneous adipose tissue (SAT) [2, 3]. Since these methods, besides checking

50

laboratory markers, are expensive and complicated to perform, they cannot be recommended in

51

routine clinical practice. Accordingly there is a need for simple techniques that can discriminate

52

regional fat, with regard to predicting metabolic disorders and mortality.

M AN U

SC

47

The famous index for measuring obesity is body mass index (BMI), which, however,

54

does not distinguish between fat distribution, when it seems excess abdominal fat mass warrants

55

closer concern than general fat mass, in terms of affecting mortality [4]. Therefore, apparently,

56

some indices of central obesity like waist circumference (WC), waist to height ratio (WHtR), and

57

waist to hip ratio (WHpR) can outperform BMI. However, there is no agreement on the

58

usefulness of WC, WHtR and WHpR. Some studies indicate that they are strong predictors [5-7]

59

whereas some contradict this [8, 9] or the cut points thereof [10].

EP

AC C

60

TE D

53

Krakaure et al recently developed ABSI based on WC adjusted for height and weight.

61

Body shape, as measured by ABSI, appears to be a substantial risk factor for premature mortality

62

in American population [11]. There are few articles investigating the discrepancies in ABSI

63

predictability for different metabolic disturbances and mortality in various populations which

64

have different findings [12, 13].

4

ACCEPTED MANUSCRIPT

In this study we aimed primarily to examine the ability of ABSI in predicting 10-year all-

66

cause mortality rate and to examine the clinical usefulness of the ABSI, compared to other

67

anthropometric measures using a large sample of Iranian men and women. To do so, we tested

68

the hypothesis that (1) ABSI predicts mortality hazard and (2) it does so better than other

69

anthropometric measures, i.e. ABSI provides more information than do other measures in

70

settings where information on traditional risk factors is not available. Finally, (3) we tested the

71

hypothesis that ABSI improves the predictive performance of the Framingham’s general CVD

72

algorithm.

AC C

EP

TE D

M AN U

SC

RI PT

65

5

ACCEPTED MANUSCRIPT

METHODS

74

Study population

75

A total of 27 340 residents, aged ≥3 years were invited by telephone call, of which 15 010

76

residents participated in first examination cycle and another 3 551 residents were first examined

77

at the second examination cycle. For the current study, among participants aged ≥30 years at

78

their baseline examination (10 801), who attended the follow up study, Exclusion criteria were:

79

1. missing data on any examined covariates (n=565), and 2. Having no follow up (n=987).

80

Eventually, a total of 9 249 eligible subjects (5 075 women and 4 174 men), were followed up.

81

At the time of this study, the median follow up time was 10.1 years.

82

Study design

83

Detailed descriptions of TLGS have been reported elsewhere [14]. In brief, the TLGS is a large

84

scale, long term, community-based prospective study performed on a representative sample of

85

residents of district 13 of Tehran, the capital of Iran. The TLGS has two major components: a

86

cross-sectional prevalence study of non-communicable disease and associated risk factors,

87

implemented between March 1999 and December 2001, and a prospective follow-up study.

88

Data collection is ongoing, designed to continue for at least 20 years, on a triennial basis. Parallel

89

with cyclic examinations, participants are followed annually for any medical condition by trained

90

nurses via telephone calls. A trained physician collects complementary data during a home visit

91

and a visit to the respective hospital to collect data from the participants’ medical files.

92

Clinical and laboratory measurements

93

Using a pretested questionnaire, a trained interviewer collected information on demographic

94

data, family history of premature CVD, past medical history of CVD, drug history, and smoking

95

status. Details of anthropometric measurements including weight, height, WC and hip

AC C

EP

TE D

M AN U

SC

RI PT

73

6

ACCEPTED MANUSCRIPT

circumference (HC) as well as blood pressure measurement have been reported elsewhere [15].

97

Height was measured in a standing position without shoes, using a tape meter while shoulders

98

were in normal alignment. Waist circumference (WC) was measured at the umbilical level and

99

that of the hip at the maximum level over light clothing, using an upstretched tape meter, without

100

any pressure to body surface and measurements were recorded to the nearest 0.1 cm. Body mass

101

index (kg.m-2) was calculated as weight (kg) divided by square of the height (m2). Waist to hip

102

ratio was calculated as WC (cm) divided by hip circumference (cm) and waist to height ratio

103

(WHtR) was calculated as WC divided by height (cm).

M AN U

SC

RI PT

96

Fasting plasma glucose (FPG) and 2-h post load plasma glucose (2h-PLG) were

105

measured using an enzymatic colorimetric method with glucose oxidase after 12-14 hours

106

overnight fasting. Details of lipid measurement, including total cholesterol (TC) and high density

107

lipoprotein cholesterol (HDL-C) have been reported elsewhere [14].

TE D

104

General CVD algorithms were based on 12 years of follow up of 8491 (4522 women and

109

3969 men) members of the Framingham cohorts, aged 30–74 years, free of symptomatic CVD at

110

baseline examination in 1968–1975 or 1984–1987 to predict CVD outcomes.

111

Outcome measurements

112

Details of cardiovascular outcomes have been published elsewhere [16]. In this ongoing study

113

each TLGS’ participant is first called by telephone and preliminary information is collected by a

114

trained nurse regarding any medical conditions or whether a related event has occurred.

115

Complementary data are then collected by a trained physician during a home visit and a visit to

116

the respective hospital to collect data from the participants’ medical files. In the case of

117

mortality, data are collected from the hospital or the death certificate by an authorized local

AC C

EP

108

7

ACCEPTED MANUSCRIPT

physician. Collected data are evaluated by an outcome committee consisting of a principal

119

investigator, an internist, an endocrinologist, a cardiologist, an epidemiologist, and the physician

120

who collects the outcome data. Other experts are invited for evaluation of non-communicable

121

disorders, as needed. A specific outcome for each event is assigned according to International

122

Statistical Classification of Diseases and Related Health Problems criteria (10th Revision), and

123

the American Heart Association classification for cardiovascular events [14, 17, 18]. CVD is

124

specified as a composite of any CHD (coronary heart disease) events, stroke or cerebrovascular

125

death. All-cause mortality included CVD and non CVD deaths. The most common causes of non

126

CVD mortality included cancer, sepsis, accidents, pneumonia, chronic obstructive pulmonary

127

diseases (COPD), other heart diseases, diabetes complications, hypertension complications,

128

unknown causes and other miscellaneous causes.

129

Definition of terms

130

Following Krakauer et al [11] we defined ABSI as:

TE D

M AN U

SC

RI PT

118

131

మ య

ܹ‫ܥ‬

భ

ቀ‫ ܫܯܤ‬ቁ × ቀℎ݁݅݃ℎ‫ ݐ‬మ ቁ

EP

‫≡ ܫܵܤܣ‬

A previous history of CVD (CVD HX) reflected any prior diagnosis of CVD by a physician.

A current smoker was defined as a person who smokes cigarettes daily or

133

occasionally. Participants using oral hypoglycemic agents or insulin were considered as having

134

diabetes. Diabetes was also ascertained in participants with FPG ≥7.0 mmol.l-1 or 2h-PLPG

135

≥11.1 mmol.l-1[19]. For each participant, the baseline risk of CVD was calculated by re-

136

estimating the Framingham’s “general CVD risk prediction algorithm [20].

AC C

132

137

8

ACCEPTED MANUSCRIPT

138

Statistics analysis We compared the predictive performance of the ABSI with those of the anthropometric

140

variables studied in terms of the effect size (HR), calibration, discrimination, and added

141

predictive ability. Findings on variables are expressed as means (SD) or percentages for

142

continuous- and categorical variables, respectively. We tested for trends across ABSI quintiles

143

using the median in each quintile as a predictor, separately for each sex. Statistical significance

144

in trends was examined by implementing age-adjusted linear and logistic regression model for

145

continuous and binary variable, respectively. The Cox proportional hazard regression model was

146

used to test the significance of trends in incident rates.

M AN U

SC

RI PT

139

In the analysis of all-cause mortality, ABSI, BMI, WC, WHpR and WHtR were assessed

148

using the accelerated failure time method: Weibull survival regression model. Survival time was

149

the time from start of the follow-up period to the date of death (failure). The censoring time of an

150

individual was the time from entry into the study to loss to follow-up or the end of the study,

151

whichever happened first. Censored observation meant the individuals either refused to

152

participate further in the study (lost to follow-up), or continued until the study was ended

153

(administrative censoring). Valid comparison of HRs for different continuous measures requires

154

that the units of both variables be comparable. We, thus, estimated sex-specific unadjusted and

155

multivariate-adjusted HRs, with 95%CIs for mortality events for a one-SD increment in ABSI

156

and each respective anthropometric parameter. Multivariate regression analyses were controlled

157

for confounding bias due to potential confounders i.e. age, systolic blood pressure, prevalent

158

CVD, using antihypertensive drugs, total and HDL cholesterol, diabetes, and smoking [21].

AC C

EP

TE D

147

159

Wald tests of the linear hypotheses concerning the Weibull survival regression models

160

coefficients (paired homogeneity test) were performed to test the null hypotheses that the hazard 9

ACCEPTED MANUSCRIPT

ratios (effect size) for ABSI were equal to those for anthropometric measures. We assessed

162

collinearity of BMI, WC, WHpR, and WHtR, with VAI (Visceral Adiposity Index) using

163

variance inflation factor (VIF). VIFs >10 warrant caution [22]. VIFs were all <10 and therefore

164

collinearity did not appear to be a problem.

RI PT

161

Calibration, as phrased in the reference [23] describes how closely predicted probabilities

166

agree numerically with actual outcomes[24, 25]. A test very similar to the Hosmer-Lemeshow

167

test has been proposed by Nam and D’Agostino. We calculated the Nam-D’Agostino χ2to

168

examine calibration for prediction models [23]. As suggested by D’Agostino and Nam,

169

calibration χ2values greater than 20 (P < 0.01) suggest lack of adequate calibration [23].

M AN U

SC

165

In the survival analysis, Harrell’s C statistic measures the probability that a randomly

171

selected person who developed an event, at the certain specific time has a higher risk score than a

172

randomly selected person who did not develop an event during the same, specific follow-up

173

interval [26, 27].

TE D

170

Discriminations measures are not sensitive to changes in absolute risk [28]. We, thus,

175

calculated absolute and relative integrated discrimination improvement index (IDI) and cut-

176

point-based and cut-point-free net reclassification improvement index (NRI). IDI and NRI are

177

measures of predictive ability added to an old model by a newer one [28]. Bootstrapping method

178

was implemented in order to obtain bias-corrected 95% CIs.

AC C

179

EP

174

In order to be able to capture a potential nonlinear association of ABSI with all-cause

180

mortality, multivariate restricted cubic splines with 4 knots defined at the 5th, 25th, 75th and

181

95th percentile were used. This method enabled us to flexibly model ABSI while preventing

182

instability and generation of artificial features to some extent [29]. 10

ACCEPTED MANUSCRIPT

The statistical significance level was set at a two-tailed type I error of 0.05. All statistics

184

analyses were performed using STATA version 12 (STATA, College Station, Texas and SAS 9.2

185

(SAS Institute Inc., Cary, NC, USA).

RI PT

183

We hereby certify that all applicable institutional and governmental regulations

187

concerning the ethical use of human volunteers were followed during this research. Informed

188

written consent was obtained from all participants and the ethical committee of Research

189

Institute for Endocrine Sciences approved this study (protocol number: ISRCTN52588395, date

190

of ethical approval: 1998)

M AN U

SC

186

AC C

EP

TE D

191

11

ACCEPTED MANUSCRIPT

Results:

193

We examined the predictability for all-cause mortality of ABSI using data on median 10-year

194

follow up of 9 242 adult participants (5 075 women and 4 174 men) of the TLGS contributing to

195

a total of 84 705 person-year follow up. We documented 487 (women 184) deaths from any

196

cause with the annual incidence rate of mortality per 1000 person being 3.9 (95% CIs 3.3-4.5)

197

for women and 8.2 (7.3-9.2) for men.

SC

199

Participants’ characteristics are shown by baseline ABSI quintiles, separately for men (tables 1) and women (table 2).

M AN U

198

RI PT

192

The risk of mortality was observed to increase with increasing levels of ABSI. The

201

magnitude of unadjusted HR for all-cause mortality of ABSI was higher than those of other

202

anthropometric measures (P values<0.001). The only exception was observed among women,

203

where the magnitude of risks conferred by WHpR exceeded those of ABSI (P<0.001). When we

204

adjusted the regression models for traditional CVD risk factors, none of the anthropometric

205

variables were found to be statistically significantly associated with all-cause mortality; the

206

exceptions were women’s ABSI and WHpR, which resisted all adjustments. Among women,

207

ABSI was observed to be a stronger predictor of all-cause mortality than was WHpR (P=0.008).

208

(Table3).

EP

AC C

209

TE D

200

Among men, Harrell’s C (95% CIs) for the mortality risk were 0.8404 (0.817-0.863) for

210

ABSI, 0.840 (0.817-0.863) for BMI, 0.840 (0.818-0.863) for waist, 0.840 (0.817-0.863) for

211

WHpR, and 0.840 (0.817-0.863) for WHtR.

12

ACCEPTED MANUSCRIPT

Among women, Harrell’s C (95% CIs) for the mortality risk were 0.861 (0.838-0.884)

213

for ABSI, 0.836 (-0.864- 0.884) for BMI, 0.864 (0.842-0.887) for waist, 0.862 (0.839-0.884) for

214

WHpR, and 0.862 (0.840-0.885) for WHtR.

RI PT

212

As shown in Table 4, ABSI was directly compared to BMI, WC, WHpR, and WHtR, the

216

prediction ability of ABSI was generally higher than those of other anthropometric measures,

217

excluding WHpR in both sexes. However, ABSI failed to add to the predictive ability of the

218

Framingham general CVD algorithm. In fact, when ABSI was added to the Framingham general

219

CVD algorithm, the values for NRI and IDI were negative except absolute and relative IDI in

220

women, indicating that the Framingham general CVD algorithm predicted 10-year risk of

221

developing mortality far better than ABSI. In fact, When ABSI or any other anthropometric

222

measures were added to the Framingham general CVD algorithm, IDI and NRI measured were

223

not statistically different from zero (Table 5).

TE D

M AN U

SC

215

As shown in figures 1 and 2, Multivariate restricted cubic splines regression analysis

225

demonstrated that ABSI-mortality dose-response relations without any risk factors’ adjustments

226

were purely u shape with a nadir of 0.08 in both genders. The lowest HRs for mortality rates

227

were between ABSI 0.06 to 0.08 and under 0.06 and above 0.08, this association approaches a

228

parabola. VIFs were all <10 in all analysis and therefore collinearity did not appear to be a

229

problem.

AC C

EP

224

13

ACCEPTED MANUSCRIPT

Discussion:

232

Using data from a community-based prospective study of an adult population of Iranian men and

233

women we demonstrated that the predictability of ABSI for 10-year all-cause mortality is

234

generally higher than other anthropometric measures such as BMI, WC, and WHtR. However

235

ABSI did not have this superiority over WHpR. ABSI, independently of traditional risk factors,

236

can predict mortality, and hence can still be used as a practical criterion to predict adiposity-

237

related health risks in clinical assessments in settings where information on traditional CVD and

238

all cause risk factors is not available. When ABSI was directly compared to the Framingham

239

general CVD algorithm, ABSI failed to confer any additional predictive ability. In fact, the

240

Framingham general CVD algorithm without ABSI predicted 10-year mortality much better than

241

it did with ABSI added.

M AN U

SC

RI PT

230 231

We could not reproduce the observations of Krauker et al where ABSI was found to be

243

associated with mortality in linear fashion[11], we showed that ABSI-mortality dose-response

244

relations were U-shaped. The lowest hazard ratios for mortality were between ABSI 0.06 to 0.08.

245

WHpR is a studied but not as widely accepted measure as it was a decade ago. WHtR can

246

be considered as useful measure currently; however, just WC is widely accepted as a certain

247

measure of fat distribution. Epidemiologic studies have clearly shown that, although WC per se

248

does not help distinguishing subcutaneous from visceral fat mass [30], WC is highly correlated

249

with the presence of metabolic disorders and increased mortality [30-33]. Such risks have been

250

shown to be attributable to the VAT mass, which is thought to be reflected in WC [30, 34]. WC,

251

however, can not predict mortality independent of height [35]. Which could have possibly be

252

due to the fact that taller people of any given relative fat content tend to have bigger WCs [36].

AC C

EP

TE D

242

14

ACCEPTED MANUSCRIPT

To overcome this shortcoming, it has been suggested that WC be divided by height to obtain a

254

more robust predictor for mortality i.e. WHtR [37]. However, some controversial follow-up

255

studies in Denmark and Japan did not support this idea [9, 38].

RI PT

253

It has been shown that physiologic characteristics of subcutaneous fat differ from those of

257

visceral fat in many respects, including insulin-sensitivity, lipolytic activity, and adipocytokines

258

production, which play a fundamental role in the genesis of cardiovascular squeal [39, 40]; these

259

observations motivated researchers to seek better obesity-associated risk markers, which led to

260

the development of a need indices like hypertriglyceridemic waist (HTGW) [41], lipid

261

accumulation product [42] and visceral adiposity index [43].

M AN U

SC

256

The most recently developed obesity index is ABSI [11], an anthropometric measure

263

used to assess relations between WC, BMI and height. ABSI has been shown to be positively

264

correlated with trunk fat mass as estimated from X-ray scans [11].

TE D

262

There are few studies assessing the predictability of ABSI for different metabolic

266

disturbances. Despite the finding that high values of ABSI show correlations with increased

267

surgical complications in patients with gastric cancer [44], it was not found to be independently

268

associated with mortality among patients on renal replacement therapy [13]. Recently, Dhana K

269

showed that among other anthropometric measures, ABSI had a stronger relation with total,

270

cardiovascular and cancer mortality. However, the added predictive value of ABSI in prediction

271

of mortality was limited [45]. In contrast, a study conducted among a European population

272

showed that WC and WHR are stronger predictors for CVD mortality than BMI and ABSI [46].

AC C

EP

265

273

In some studies, no superiority has been found for ABSI over other anthropometric

274

measures in prediction of CVD and metabolic syndrome. We previously reported that ABSI had 15

ACCEPTED MANUSCRIPT

not been found to consistently add to the predictive values of other anthropometric measures in

276

CVD prediction [47]. Another study conducted among Iranian population showed that ABSI is a

277

weak predictor for CVD risks [12]. Another study conducted in a middle-age, older Indonesian

278

population reported that ABSI was less strongly associated with incident hypertension than WC

279

and BMI [48]. In contrast, in a sample of Portuguese adolescents, ABSI explained a greater

280

amount of the variance in blood pressure than did WC and BMI. As such, when examining the

281

effect of weight status on BP, considering use of ABSI alongside BMI would be justified [49].

SC

RI PT

275

In the original paper ABSI was not compared with WHpR [11]. In our study, ABSI was

283

found to outperform other anthropometric measures in terms of mortality prediction in both

284

sexes. However, WHpR remained an exception that predicted mortality much better than ABSI.

285

In WHpR, WC is adjusted for hip circumference. Larger hip may occur in individuals with

286

higher muscular mass including gluteal muscle mass [50]. Furthermore, gluteofemoral fat has

287

been shown to remove free fatty acids from the bloodstream [51]. Women with wider hips

288

(adjusted for weight and WC) have been shown to be less susceptible to hypertension, diabetes,

289

and gallbladder disease, after[52, 53].

EP

TE D

M AN U

282

In the current study, ABSI predicted all-cause mortality independently of traditional CVD

291

risk factors incorporated in the general CVD Framingham study, indicating that there is some

292

ABSI-related mortality that is not mediated by hypertension, diabetes mellitus, or dyslipidemia.

293

However, adding ABSI to the algorithm did not add to its predictive power in terms of integrated

294

decimation or risk classification. By adjusting the effect on mortality of anthropometric measures

295

for traditional CVD risk factors such as hypertension and dyslipidemia or diabetes, we have

296

sought to estimate the mortality rate conferred by obesity independently of indicators of

297

metabolic derangements in multivariate model. Measures of obesity have not contributed to the

AC C

290

16

ACCEPTED MANUSCRIPT

mortality after the effects of metabolic risk factor were taken into account implies that adverse

299

effects of obesity is possibly mediated by metabolic derangement. It is not surprising to observe

300

that obesity that has not led to metabolic derangement does not contribute to mortality.

RI PT

298

The strength of the present study lies in its prospective nature, use of a large population-

302

based cohort of both sexes, accurate and valid data on risk factors at baseline continuous,

303

surveillance of mortality events based on standard criteria, well documented systematic

304

recordings all of the variables required to define abdominal obesity. Some limitations of our

305

study merit mentioning. In this study, the population studied was of Persian ancestry, our results,

306

thus, cannot be readily extrapolated to other populations. As is inherent to any prospective study,

307

levels of risk factors at the baseline examination might have been changed during the follow up

308

period, as such, some degrees of misclassification might have biased estimated hazard ratios

309

towards the null. The numbers of deaths due to different medical conditions were limited. Our

310

study sample, thus, did not have enough statistical power enough to allow investigating cause-

311

specific mortalities. Response rate for subjects is not so high (15010/27340=0.55). This can be

312

regarded as a restriction in extrapolation of results to general population. On the other hand, we

313

were aware that there are linear dependencies among ABSI, BMI, WC, WHpR, WHtR variables;

314

to prevent the overfitting of the multiple fitted models of these dependencies, anthropometric

315

measures were then incorporated into a regression model to test the heterogeneity as suggested in

316

literature. Furthermore, increment of mortality risk in subsequent standard deviations may not be

317

similar, a difficulty which may arise when comparing nonlinear associations, to overcome this

318

limitation, in addition to the hazard ratios, we used other indices to compare the predictive

319

capacity for mortality of different obesity measures.

AC C

EP

TE D

M AN U

SC

301

17

ACCEPTED MANUSCRIPT

In conclusion, using data from a 10-year follow up of a large sample of Iranian men and

321

women, ABSI was observed to be more strongly associated with all-cause mortality than BMI,

322

WC and WHtR. Therefore, it can still be used as a practical criterion to predict adiposity-related

323

health risks in clinical assessments in settings where information on traditional CVD and all

324

cause risk factors are lacking. Among women, however, WHpR was observed to be a stronger

325

predictor of all-cause mortality than ABSI. Among both men and women, ABSI improved the

326

risk classification based on all other anthropometric measures, except for WHpR. None of the

327

anthropometric measures studied were found to add to the predictive ability of Framingham’s

328

general CVD algorithm.

329 330

Acknowledgment:

331

Republic of Iran. We express our thanks to the participants of district-13 of Tehran for their

332

enthusiastic support in this study. The authors also wish to acknowledge Mrs. Niloofar Shiva for

333

critical editing of English grammar and syntax of the manuscript and also Dr. Yadollah Mehrabi

334

for critical observe on statistical analysis.

SC

M AN U

EP

TE D

The study was supported by grant No. 121 from the National Research Council of the Islamic

AC C

335

RI PT

320

18

ACCEPTED MANUSCRIPT

336

References:

337

5.

6.

7. 8.

9. 10.

11. 12.

13. 14.

15.

16.

17.

RI PT

SC

4.

M AN U

3.

TE D

2.

Lavie CJ, Milani RV, Ventura HO: Obesity and Cardiovascular DiseaseRisk Factor, Paradox, and Impact of Weight Loss. J Am Coll Cardiol 2009, 53:1925-1932. Luu Y, Lublinsky S, Ozcivici E, Capilla E, Pessin J, Rubin C, Judex S: < i> In vivo quantification of subcutaneous and visceral adiposity by micro-computed tomography in a small animal model. Med Eng Phys 2009, 31:34-41. Bertin E, Ruiz J-C, Mourot J, Peiniau P, Portha B: Evaluation of dual-energy X-ray absorptiometry for body-composition assessment in rats. J. Nutr 1998, 128:1550-1554. Kalmijn S, Curb J, Rodriguez B, Yano K, Abbott R: The association of body weight and anthropometry with mortality in elderly men: the Honolulu Heart Program. Int J Obes Relat Metab Disord: journal of the International Association for the Study of Obesity 1999, 23:395-402. Esmaillzadeh A, Mirmiran P, Azizi F: Waist-to-hip ratio is a better screening measure for cardiovascular risk factors than other anthropometric indicators in Tehranian adult men. Int J Obes 2004, 28:1325-1332. Reis JP, Araneta MR, Wingard DL, Macera CA, Lindsay SP, Marshall SJ: Overall obesity and abdominal adiposity as predictors of mortality in US white and black adults. Annals of epidemiology 2009, 19:134-142. Burton RF: Waist circumference as an indicator of adiposity and the relevance of body height. Med Hypotheses 2010, 75:115-119. 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, 74:315-321. Han T, McNeill G, Seidell J, Lean M: Predicting intra-abdominal fatness from anthropometric measures: the influence of stature. Int J Obes 1997, 21:587-593. Hadaegh F, Zabetian A, Sarbakhsh P, Khalili D, James W, Azizi F: Appropriate cutoff values of anthropometric variables to predict cardiovascular outcomes: 7.6 years follow-up in an Iranian population. Int J Obes 2009, 33:1437-1445. Krakauer NY, Krakauer JC: A new body shape index predicts mortality hazard independently of body mass index. PLoS One 2012, 7:e39504. Haghighatdoost F, Sarrafzadegan N, Mohammadifard N, Asgary S, Boshtam M, Azadbakht L: Assessing body shape index as a risk predictor for cardiovascular diseases and metabolic syndrome among Iranian adults. Nutrition 2014, 30:636-644. Afsar B, Elsurer R, Kirkpantur A: Body shape index and mortality in hemodialysis patients. Nutrition 2013, 29:1214-1218. Azizi F, Ghanbarian A, Momenan AA, Hadaegh F, Mirmiran P, Hedayati M, Mehrabi Y, Zahedi-Asl S: Prevention of non-communicable disease in a population in nutrition transition: Tehran Lipid and Glucose Study phase II. Trials 2009, 10:5. Freedman DS, Kahn HS, Mei Z, Grummer-Strawn LM, Dietz WH, Srinivasan SR, Berenson GS: Relation of body mass index and waist-to-height ratio to cardiovascular disease risk factors in children and adolescents: the Bogalusa Heart Study. Am J Clin Nutr 2007, 86:33-40. Hadaegh F, Harati H, Ghanbarian A, Azizi F: Association of total cholesterol versus other serum lipid parameters with the short-term prediction of cardiovascular outcomes: Tehran Lipid and Glucose Study. Eur J Cardiovasc Prev Rehabil 2006, 13:571-577. Gibbons RJ, Abrams J, Chatterjee K, Daley J, Deedwania PC, Douglas JS, Ferguson TB, Fihn SD, Fraker TD, Gardin JM: ACC/AHA 2002 guideline update for the management of patients with

EP

1.

AC C

338 339 340 341 342 343 344 345 346 347 348 349 350 351 352 353 354 355 356 357 358 359 360 361 362 363 364 365 366 367 368 369 370 371 372 373 374 375 376 377 378 379 380 381

19

ACCEPTED MANUSCRIPT

22.

23. 24. 25. 26.

27.

28. 29. 30.

31.

32.

RI PT

SC

21.

M AN U

20.

TE D

19.

EP

18.

chronic stable angina—summary articlea report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines (Committee on the Management of Patients With Chronic Stable Angina). J Am Coll Cardiol 2003, 41:159-168. Braunwald E, Antman EM, Beasley JW, Califf RM, Cheitlin MD, Hochman JS, Jones RH, Kereiakes D, Kupersmith J, Levin TN: ACC/AHA Guideline Update for the Management of Patients With Unstable Angina and Non–ST-Segment Elevation Myocardial Infarction—2002: Summary Article A Report of the American College of Cardiology/American Heart Association Task Force on Practice Guidelines (Committee on the Management of Patients With Unstable Angina). Circulation 2002, 106:1893-1900. Genuth S, Alberti K, Bennett P, Buse J, Defronzo R, Kahn R, Kitzmiller J, Knowler W, Lebovitz H, Lernmark A: Follow-up report on the diagnosis of diabetes mellitus. Diabetes Care 2003, 26:3160. D’Agostino RB, Vasan RS, Pencina MJ, Wolf PA, Cobain M, Massaro JM, Kannel WB: General cardiovascular risk profile for use in primary care the Framingham Heart Study. Circulation 2008, 117:743-753. Bozorgmanesh M, Hadaegh F, Azizi F: Predictive accuracy of the ‘Framingham’s general CVD algorithm’in a Middle Eastern population: Tehran Lipid and Glucose Study. Int J Clin Pract 2011, 65:264-273. Abbate LM, Stevens J, Schwartz TA, Renner JB, Helmick CG, Jordan JM: Anthropometric measures, body composition, body fat distribution, and knee osteoarthritis in women. Obesity 2006, 14:1274-1281. D’Agostino R, Nam B-H: Evaluation of the performance of survival analysis models: discrimination and calibration measures. Handbook of statistics 2004, 23:1-26. Hosmer D, Lemeshow S: Applied Logistic Regression. New York, NY: A Wiley-Interscience Publication. John Wiley & Sons Inc; 2000. Hosmer Jr DW, Lemeshow S, May S: Applied survival analysis: regression modeling of time to event data. Wiley. com; 2011. Hlatky MA, Greenland P, Arnett DK, Ballantyne CM, Criqui MH, Elkind MS, Go AS, Harrell FE, Hong Y, Howard BV: Criteria for Evaluation of Novel Markers of Cardiovascular Risk A Scientific Statement From the American Heart Association. Circulation 2009, 119:2408-2416. Harrell F, Lee KL, Mark DB: Tutorial in biostatistics multivariable prognostic models: issues in developing models, evaluating assumptions and adequacy, and measuring and reducing errors. Stat. Med 1996, 15:361-387. Pencina MJ, D'Agostino RB, Vasan RS: Evaluating the added predictive ability of a new marker: from area under the ROC curve to reclassification and beyond. Stat. Med 2008, 27:157-172. Royston P, Sauerbrei W: Multivariable modeling with cubic regression splines: a principled approach. Stata Journal 2007, 7:45-70. Pouliot M-C, Després J-P, Lemieux S, Moorjani S, Bouchard C, Tremblay A, Nadeau A, Lupien PJ: Waist circumference and abdominal sagittal diameter: best simple anthropometric indexes of abdominal visceral adipose tissue accumulation and related cardiovascular risk in men and women. Am J Cardiol 1994, 73:460-468. Schneider HJ, Glaesmer H, Klotsche J, Böhler S, Lehnert H, Zeiher AM, März W, Pittrow D, Stalla GK, Wittchen HU: Accuracy of anthropometric indicators of obesity to predict cardiovascular risk. J Clin Endocrinol Metab 2007, 92:589-594. Grinker J, Tucker K, Vokonas P, Rush D: Changes in patterns of fatness in adult men in relation to serum indices of cardiovascular risk: the Normative Aging Study. Int J Obes 2000, 24:13691378.

AC C

382 383 384 385 386 387 388 389 390 391 392 393 394 395 396 397 398 399 400 401 402 403 404 405 406 407 408 409 410 411 412 413 414 415 416 417 418 419 420 421 422 423 424 425 426 427 428

20

ACCEPTED MANUSCRIPT

38.

39. 40.

41.

42.

43.

44.

45.

46.

47.

48.

RI PT

37.

SC

36.

M AN U

35.

TE D

34.

Okosun I, Tedders S, Choi S, Dever G: Abdominal adiposity values associated with established body mass indexes in white, black and hispanic Americans. A study from the Third National Health and Nutrition Examination Survey. Int J Obes Relat Metab Disord: journal of the International Association for the Study of Obesity 2000, 24:1279. Clasey JL, Bouchard C, Teates CD, Riblett JE, Thorner MO, Hartman ML, Weltman A: The Use of Anthropometric and Dual-Energy X-ray Absorptiometry (DXA) Measures to Estimate Total Abdominal and Abdominal Visceral Fat in Men and Women. Obesity Research 1999, 7:256-264. Burton R: Waist circumference as an indicator of adiposity and the relevance of body height. Medical hypotheses 2010, 75:115-119. Bacha F, Saad R, Gungor N, Janosky J, Arslanian SA: Obesity, regional fat distribution, and syndrome X in obese black versus white adolescents: race differential in diabetogenic and atherogenic risk factors. J Clin Endocrinol Metab 2003, 88:2534-2540. Ashwell M, Gunn P, Gibson S: Waist-to-height ratio is a better screening tool than waist circumference and BMI for adult cardiometabolic risk factors: systematic review and metaanalysis. Obes Rev 2012, 13:275-286. Nakamura K, Nanri H, Hara M, Higaki Y, Imaizumi T, Taguchi N, Sakamoto T, Horita M, Shinchi K, Tanaka K: Optimal cutoff values of waist circumference and the discriminatory performance of other anthropometric indices to detect the clustering of cardiovascular risk factors for metabolic syndrome in Japanese men and women. Environ Health Prev Med 2011, 16:52-60. Kershaw EE, Flier JS: Adipose tissue as an endocrine organ. J Clin Endocrinol Metab 2004, 89:2548-2556. Kelley DE, Thaete FL, Troost F, Huwe T, Goodpaster BH: Subdivisions of subcutaneous abdominal adipose tissue and insulin resistance. Am J Physiol Endocrinol Metab 2000, 278:E941-E948. Tankó LB, Bagger YZ, Qin G, Alexandersen P, Larsen PJ, Christiansen C: Enlarged waist combined with elevated triglycerides is a strong predictor of accelerated atherogenesis and related cardiovascular mortality in postmenopausal women. Circulation 2005, 111:1883-1890. Kahn HS: The" lipid accumulation product" performs better than the body mass index for recognizing cardiovascular risk: a population-based comparison. BMC Cardiovasc Disord 2005, 5:26. Amato MC, Giordano C, Galia M, Criscimanna A, Vitabile S, Midiri M, Galluzzo A: Visceral Adiposity Index A reliable indicator of visceral fat function associated with cardiometabolic risk. Diabetes Care 2010, 33:920-922. Eom BW, Joo J, Yoon HM, Ryu KW, Kim YW, Lee JH: A body shape index has a good correlation with postoperative complications in gastric cancer surgery. Ann Surg Oncol 2014, 21:11151122. Dhana K, Kavousi M, Ikram MA, Tiemeier HW, Hofman A, Franco OH: Body shape index in comparison with other anthropometric measures in prediction of total and cause-specific mortality. J Epidemiol Community Health 2015. Song X, Jousilahti P, Stehouwer CD, Soderberg S, Onat A, Laatikainen T, Yudkin JS, Dankner R, Morris R, Tuomilehto J, Qiao Q: Comparison of various surrogate obesity indicators as predictors of cardiovascular mortality in four European populations. Eur J Clin Nutr 2013, 67:1298-1302. Bozorgmanesh M, Sardarinia M, Hajsheikholeslami F, Azizi F, Hadaegh F: CVD-predictive performances of "a body shape index" versus simple anthropometric measures: Tehran lipid and glucose study. Eur J Nutr 2015. Cheung YB: “A Body Shape Index” in Middle-Age and Older Indonesian Population: Scaling Exponents and Association with Incident Hypertension. PloS one 2014, 9:e85421.

EP

33.

AC C

429 430 431 432 433 434 435 436 437 438 439 440 441 442 443 444 445 446 447 448 449 450 451 452 453 454 455 456 457 458 459 460 461 462 463 464 465 466 467 468 469 470 471 472 473 474 475 476

21

ACCEPTED MANUSCRIPT

49.

50. 51. 52. 53.

Duncan MJ, Mota J, Vale S, Santos MP, Ribeiro JC: Associations between body mass index, waist circumference and body shape index with resting blood pressure in Portuguese adolescents. Ann Hum Biol 2013, 40:163-167. Snijder M, Van Dam R, Visser M, Seidell J: What aspects of body fat are particularly hazardous and how do we measure them? Int. J. Epidemiol 2006, 35:83-92. Manolopoulos K, Karpe F, Frayn K: Gluteofemoral body fat as a determinant of metabolic health. Int J Obes 2010, 34:949-959. HARTZ AJ, RUPLEY DC, RIMM AA: The association of girth measurements with disease in 32,856 women. Am J Epidemiol 1984, 119:71-80. Lissner L, Björkelund C, Heitmann BL, Seidell JC, Bengtsson C: Larger hip circumference independently predicts health and longevity in a Swedish female cohort. Obes Res 2001, 9:644-646.

RI PT

477 478 479 480 481 482 483 484 485 486 487 488

AC C

EP

TE D

M AN U

SC

489

22

ACCEPTED MANUSCRIPT

(0.046-0.075)

(0.075-0.078)

(0.078-0.080)

Q4 N=839

Q5 N=835

(0.080-0.082)

(0.082-0.113)

RI PT

Table 1. Basal characteristics of participants across ABSI quintiles, among men. Q1 Q2 Q3 N=825 N=835 N=840

P for trend*

AC C

EP

TE D

M AN U

SC

ABSI range Age (years) 41.52 (10.96) 44.64 (11.89) 48.98 (13.00) 52.46 (12.77) 59.49 (12.87) <0.001 SBP (mm Hg) 115.02 (15.04) 118.63 (17.28) 121.32 (18.51) 125.53 (21.00) 129.84 (21.39) -1 0.077 TC (mmol.l ) 5.18 (1.10) 5.35(1.07) 5.32 (1.06) 5.48 (1.09) 5.42 (1.10) -1 <0.001 HDL-C (mmol.l ) 1.02 (0.27) 0.96 (0.24) 0.96 (0.23) 0.96 (0.24) 0.95 (0.23) <0.001 Weight (kg) 72.06 (12.07) 75.88 (13.12) 75.76(13.19) 76.24 (12.23) 75.24 (12.03) <0.001 Height (m) 1.69 (0.06) 1.69 (0.06) 1.69 (0.06) 1.68 (0.06) 1.68 (0.07) -2 <0.001 BMI (kg.m ) 25.04 (3.99) 26.35 (4.10) 26.35(4.00) 26.83 (3.80) 26.55 (3.66) <0.001 Waist (cm) 0.81(0.09) 0.88 (0.09) 0.91 (0.09) 0.94 (0.09) 0 .98 (0.09) <0.001 Hip circumference (cm) 0.94 (0.06) 0.97 (0.07) 0.97 (0.06) 0.97 (0.06) 0.97 (0.06) <0.001 WHpR 85.97(5.11) 91.14 (4.53) 93.65 (4.57) 96.56 (4.62) 100.80 (5.30) <0.001 WHtR 48.16(5.62) 52.34 (5.57) 53.84 (5.48) 56.19 (5.49) 58.43 (5.46) 0.017 Anti-hypertensive drug(yes or no) 37 (0.19) 88 (0.28) 111 (0.31) 145 (0.35) 202 (0.40) 0.880 Smoking(yes or no) 338 (0.47) 298 (0.45) 275 (0.44) 276 (0.44) 240 (0.42) <0.001 Diabetes (yes or no) 69 (0.25) 112 (0.31) 157 (0.36) 189 (0.39) 286 (0.45) 0.179 CVD Hx (yes or no) 48 (0.21) 73 (0.07) 98 (0.29) 103 (0.30) 150 (0.35) All-cause mortality, n 39 (4.73) 40 (4.79) 48 (5.71) 60 (7.15) 116 (13.89) Per 1000 person-year 4.8 (3.5-6.6) 5.08 (3.7-6.9) 6.4 (4.8-8.5) 8.3 (6.4-10.7) 18.0 (15.0-21.6) <0.001 Data are presented as either mean (SD) for continuous variable or frequency (%) for categorically distributed variables. Q stands for Quantile. ABSI a body shape index; BMI, body mass index; CHD, coronary heart disease; CVD, cardiovascular disease; DBP, diastolic blood pressure; FPG, fasting plasma glucose; HDL-C, high-density lipoprotein cholesterol; MAP, mean arterial pressure; PCPG, 2-hour postchallenge plasma glucose; SBP, systolic blood pressure; TC, total cholesterol; TGs, triglycerides; WHpR, waist-to-hip ratio; WHtR, waist-to-height ratio, CVD Hx, previous history of CVD . * The statistical significance of trends across ABSI quintiles was tested by using the median in each quartile as a predictor in general linear models and logistic regression model for continuously- and binary-distributed variables, respectively incorporating age. The Log-Rank test and Cox test were used to examine the significance of trends in incident rates and survivor functions.

ACCEPTED MANUSCRIPT

Table 2. Basal characteristics of participants across ABSI quintiles, among women. Q1 Q2 Q3 N=1012 N=1009 N=1017 (0.042-0.072)

(0.072-0.076)

(0.076-0.079)

Q4 N=1027

Q5 N=1010

(0.079-0.083)

(0.083-0.113)

P for trend*

AC C

EP

TE D

M AN U

SC

RI PT

ABSI range Age (years) 40.28(9.10) 43.06 (10.16) 47.01 (11.28) 50.94(11.82) 57.48(11.92) SBP (mm Hg) 114.21(16.01) 117.65(18.05) 120.35(19.48) 125.30(21.13) 131.45(23.55) <0.001 -1 TC (mmol.l ) 5.28(1.08) 5.46(1.11) 5.63(1.18) 5.90(1.29) 6.09(1.35) 0.001 -1 HDL-C (mmol.l ) 1.19(0.28) 1.14(0.29) 1.13(0.28) 1.13(0.27) 1.14 (0.28) <0.001 Weight (kg) 68.53(11.90) 69.99(12.64) 71.01(12.68) 69.74(12.23) 67.34(11.27) 0.013 Height (m) 1.56 (0.05) 1.56 (0.06) 1.55(0.05) 1.55 (0.05) 1.54(0.06) <0.001 -2 BMI (kg.m ) 27.96(4.79) 28.62(4.97) 29.25(5.10) 29.01(4.60) 28.14(4.28) <0.001 Waist (cm) 0.80(0.09) 0.87 (0.10) 0.92 (0.10) 0.95 (0.10) 1.00 (0.10) <0.001 Hip circumference (cm) 1.04(0.09) 1.05(0.09) 1.06(0.09) 1.05(0.09) 1.03(0.09) <0.001 WHpR 76.44(5.02) 82.04 (4.17) 86.12(4.43) 90.18(4.64) 96.19(5.56) <0.001 WHtR 51.35(6.42) 55.72(6.76) 59.05(7.09) 61.49(6.60) 64.57(6.59) <0.001 Anti-hypertensive drug (yes or no) 122 (0.32) 140 (0.34) 193 (0.39) 264 (0.44) 311 (0.46) 0.851 Smoking (yes or no) 50 (0.21) 39 (0.19) 53 (0.22) 33 (0.17) 28 (0.16) 0.076 Diabetes (yes or no) 63 (0.24) 96 (0.29) 151 (0.35) 226 (0.41) 333 (0.47) <0.001 CVD Hx(yes or no) 100 (0.30) 81 (0.27) 111 (0.31) 142 (0.34) 161 (0.36) 0.359 All-cause mortality, n 7 (0.69) 26 (2.5) 18 (1.77) 39 (3.80) 94 (9.31) Per 1000 person-year 0.7 (0.3-1.4) 2.6 (1.8-3.9) 1.8 (1.2-3.0) 4.1 (3.0-5.7) 10.3 (8.4-12.6) <0.001 Data are presented as either mean (SD) for continuous variable or frequency (%) for categorically distributed variables. Q stands for Quantile. ABSI a body shape index; BMI, body mass index; CHD, coronary heart disease; CVD, cardiovascular disease; DBP, diastolic blood pressure; FPG, fasting plasma glucose; HDL-C, high-density lipoprotein cholesterol; MAP, mean arterial pressure; PCPG, 2-hour postchallenge plasma glucose; SBP, systolic blood pressure; TC, total cholesterol; TGs, triglycerides; WHpR, waist-to-hip ratio; WHtR, waistto-height ratio; CVD Hx, previous history of CVD. * The statistical significance of trends across ABSI quintiles was tested by using the median in each quartile as a predictor in general linear models and logistic regression model for continuously- and binary-distributed variables, respectively incorporating age. The LogRank test and Cox test were used to examine the significance of trends in incident rates and survivor functions.

ACCEPTED MANUSCRIPT

Multivariate-adjusted b Men

M AN U

SC

RI PT

Table 3. Hazard ratios for mortality of ABSI vs. BMI, WC, WHpR, and WHtR. HR (95%CIs) P value Univariate Men ABSI 1.89(1.67-2.14) BMI <0.001 1.68 (1.32-2.05) Waist 1.11(0.99-1.26) <0.001 WHpR 1.53 (1.35-1.74) <0.001 WHtR 1.34 (1.17-1.54) <0.001 Women ABSI 2.00 (1.80-2.21) BMI 1.00 (0.86-1.15) <0.001 Waist 1.48 (1.30-170) <0.001 WHpR 2.28 (1.97-2.64) <0.001 WHtR 1.71 (1.49-1.96) <0.001

TE D

ABSI BMI Waist WHpR WHtR

0.94 (0.81-1.09) 0.90 (0.77-1.05) 0.86 (0.78-1.02) 0.99 (0.97-1.00) 0.90 (0.78-1.05)

0.257 0.792 0.326 0.203

Women

AC C

EP

ABSI 1.28 (1.10-1.47) BMI 0.91 (0.78-1.06) 0.204 Waist 1.02 (0.87-1.19) 0.114 WHpR 1.02 (1.00-1.04) 0.008 WHtR 1.66 (0.91-1.24) 0.420 ABSI a body shape index; BMI, body mass index; CVD, cardiovascular disease; HR, hazard ratio; WC, waist circumference; WHpR, waist-to-hip ratio; WHtR, waist-to-height ratio. a. P values were derived from Wald tests of the linear hypotheses concerning the Weibull regression models coefficients (paired homogeneity test). As such, we tested the null hypotheses that the hazard ratios (effect size) for ABSI were equal to those for WHpR, WHtR, or BMI (comparison of HRs between ABSI and other anthropometric measures). b. Adjusted for the effects of age, systolic blood pressure, anti-hypertensive medication use, total and high-density lipoprotein cholesterol, diabetes, smoking and CVD prevalent.

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

Table 4. Added predictive ability conferred by ABSI to different anthropometric measures. Men Women Predictive PPredictive P95% CIs 95% CIs index value index value a, c BMI alone vs. ABSI Absolute IDI (%) 0.2995 0.0239 0.0359 <0.001 0.0093 0.0033 0.0153 0.002 Relative IDI (%) 80.4754 67.63 93.31 <0.001 0.0913 0.0346 0.1481 0.002 Cut-point-based NRI (%) 0.3423 0.1973 0.4873 <0.001 0.0410 -0.0079 0.0899 0.100 Cut-point-free NRI (%) 0.4443 0.2724 0.6162 <0.001 0.2160 0.0029 0.4290 0.047 c Waist circumference alone vs. ABSI Absolute IDI (%) 0.0294 0.0206 0.0382 <0.001 0.0337 0.0218 0.0456 <0.001 Relative IDI (%) 34.5435 25.12 43.96 <0.001 4.1283 2.4584 5.7982 <0.001 Cut-point-based NRI (%) 0.3517 0.2377 0.4657 <0.001 0.3482 0.2045 0.4920 <0.001 Cut-point-free NRI (%) 0.4902 0.3266 0.6538 <0.001 0.6926 0.5346 0.8506 <0.001 c WHpR alone vs. ABSI b b b Absolute IDI (%) -0.0006 -0.0021 0.0008 0.399 0.0053 -0.0035 0.0143 0.237 b b b Relative IDI (%) -0.0031 -0.0103 0.0041 0.400 0.1477 -0.8137 0.4092 0.268 b Cut-point-based NRI (%) 0.0113 -0.0088 0.0314 0.270 0.0966 0.0145 0.1787 0.021 Cut-point-free NRI (%) 0.0458 -0.0762 b 0.1679 0.462 0.1912 0.0269 0.3554 0.022 c WHtR alone vs. ABSI Absolute IDI (%) 0.0242 0.0170 0.0314 <0.001 0.0260 0.0123 0.0396 <0.001 Relative IDI (%) 3.9658 2.9112 5.020 <0.001 1.6325 0.8219 2.4431 <0.001 Cut-point-based NRI (%) 0.2714 0.1817 0.3611 <0.001 0.2652 0.1632 0.3673 <0.001 Cut-point-free NRI (%) 0.3756 0.2668 0.4843 <0.001 0.5937 0.4895 0.6978 <0.001 ABSI a body shape index; BMI, body mass index; CVD, cardiovascular disease; IDI, integrated discrimination improvement index; NRI, net reclassification improvement index; WHpR, waist-to-hip ratio; WHtR, waist-to-height ratio. a. BMI were not associated with all-cause mortality among women. b. Negative signs indicate less predictive ability for ABSI as compared to the general CVD algorithm, WHtR, WHpR, or BMI. c. Calculated in accordance with D'Agostino, R.B., Sr., et al., General Cardiovascular Risk Profile for Use in Primary Care: The Framingham Heart Study. Circulation, 2008. 117(6): p. 743-753. For cut-point based NRI, the cut-points were set at 0.03and 0.07 of estimated risk for women and 0.07 and 0.14 of estimated risk for men

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

Table 5. Added predictive ability conferred by different anthropometric measures as compared to the Framingham’s general CVD algorithm. Men Women PPPredictive Predictive 95% CIs 95% CIs index index value value b BMI+ General CVD risk Absolute IDI (%) 0.0017 -0.0015 0.0050 0.296 0.0019 0.0006 0.0032 0.003 Relative IDI (%) 0.0085 -0.0069 0.0241 0.280 0.0151 0.0042 0.0260 0.007 c Cut-point-based NRI (%) -0.0211 -0.0436 0.0012 0.065 0.1711 -0.0222 0.0564 0.394 Cut-point-free NRI (%) -0.0607 -0.1889 0.0673 0.352 -0.0304 -0.2019 0.1409 0.727 b Waist circumference+ General CVD risk Absolute IDI (%) 0.0007 -0.0022 0.0037 0.629 -0.0001 -0.0005 0.0003 0.688 Relative IDI (%) 0.0035 -0.0110 0.0181 0.631 -0.0007 -0.0045 0.0030 0.694 c -0.0168 -0.0511 0.0174 0.335 -0.0014 -0.0026 Cut-point-based NRI (%) 0.020 0.0002 Cut-point-free NRI (%) -0.0496 -0.1355 0.0362 0.257 0.1412 -0.1018 0.3844 0.255 b WHpR+ General CVD risk Absolute IDI (%) -0.0003 -0.0019 0.0013 0.696 0.0041 0.0001 0.0081 0.043 Relative IDI (%) -0.0015 -0.0094 0.0062 0.690 0.0321 -0.0008 0.0652 0.056 c Cut-point-based NRI (%) -0.0132 -0.0301 0.0036 0.123 0.0004 -0.0479 0.0489 0.984 Cut-point-free NRI (%) -0.0885 -0.2567 0.0796 0.302 0.2375 0.0646 0.4105 0.007 b WHtR+ General CVD risk Absolute IDI (%) 0.0004 -0.0026 0.0035 0.758 0.0000 -0.0012 0.0012 0.955 Relative IDI (%) 0.0023 -0.0124 0.0172 0.754 0.0002 -0.0095 0.0101 0.956 c Cut-point-based NRI (%) -0.0171 -0.0365 0.0022 0.083 0.0189 -0.0086 0.0465 0.178 Cut-point-free NRI (%) -0.1080 -0.2200 0.0038 0.058 0.1044 -0.0358 0.2448 0.145 b ABSI+ General CVD risk Absolute IDI (%) -0.0008 -0.0022 0.0005 0.236 0.0085 0.0030 0.0141 0.003 Relative IDI (%) -0.0041 -0.0111 0.0028 0.267 0.0666 0.0237 0.1094 0.002 c Cut-point-based NRI (%) 0.0017 -0.0105 0.0071 0.707 0.0213 -0.0378 0.0805 0.479 Cut-point-free NRI (%) 0.0651 -0.0257 0.1559 0.160 0.1941 -0.0446 0.4328 0.111 ABSI a body shape index; BMI, body mass index; CVD, cardiovascular disease; IDI, integrated discrimination improvement index; NRI, net reclassification improvement index; WHpR, waist-to-hip ratio; WHtR, waist-to-height ratio. a. Negative signs indicate less predictive ability for ABSI as compared to the general CVD algorithm, WHtR, WHpR, or BMI. b. Calculated in accordance with D'Agostino, R.B., Sr., et al., General Cardiovascular Risk Profile for Use in Primary Care: The Framingham Heart Study. Circulation, 2008. 117(6): p. 743-753. c. For cut-point based NRI, the cut-points were set at 0.03 and 0.07 of estimated risk for women and 0.07 and 0.14 of estimated risk for men.

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

Figure 1: Multivariate restricted cubic splines of ABSI-mortality dose-response curve in men.

ACCEPTED MANUSCRIPT

AC C

EP

TE D

M AN U

SC

RI PT

Figure 2: Multivariate restricted cubic splines of ABSI-mortality dose-response curve in women.

ACCEPTED MANUSCRIPT

Age ≥30 years (n=10 801)

M AN U

SC

With missing data on any examined covariates (n=565)

RI PT

Flowchart: Selection of study participants; Teheran Lipid and Glucose Study (1999–2010).

TE D

With no any follow up (n=987)

AC C

EP

9 249 followed till March 2010 (4 174 men and 5 075 women)

ACCEPTED MANUSCRIPT

Highlights: •

A body shape index (ABSI) is based on waist circumference (WC) adjusted for height and weight.

•

RI PT

ABSI appeared to be a substantial risk factor for premature mortality in an American population. •

ABSI was the strongest predictor of all-cause mortality among the anthropometric measurements, except women’s WHpR in an Iranian population.

SC

•

AC C

EP

TE D

M AN U

ABSI failed to improve the predictive ability of Framingham general cardiovascular disease algorithm.