OBJECTIVE

To compare different anthropometric measures in terms of their ability to predict type 2 diabetes and to determine whether predictive ability was modified by ethnicity.

RESEARCH DESIGN AND METHODS

Anthropometry was measured at baseline for 1,073 non-Hispanic white (nHW), African American (AA), and Hispanic (HA) subjects, of whom 146 developed type 2 diabetes after 5.2 years. Logistic regression models were used with areas under the receiver operator characteristic curve (AROCs) comparing the prediction of models.

RESULTS

Waist-to-height ratio (AROC 0.678) was the most predictive measure, followed by BMI (AROC 0.674). Results were similar in nHW and HA subjects, although in AA subjects, central adiposity measures appeared to best predict type 2 diabetes.

CONCLUSIONS

Measures of central and overall adiposity predicted type 2 diabetes to a similar degree, except in AA subjects, for whom results suggested that central measures were more predictive.

Various assessment approaches have been used to characterize the relationship between obesity and type 2 diabetes. Computed tomography and other imaging techniques allow for direct quantification of visceral adipose tissue (VAT) and subcutaneous adipose tissue (SAT). These methods are costly and invasive, however; thus, anthropometric measurements are more commonly utilized. The majority of previous studies have assessed only BMI, waist circumference (WC), and waist-to-hip ratio (WHR), and few data are available from multi-ethnic cohorts (1,,,,6). Our objective, therefore, was to compare a wide range of anthropometric measures in terms of their ability to predict incident type 2 diabetes in three ethnic groups.

The Insulin Resistance Atherosclerosis Study has been previously described (7). The current report was comprised of 1,073 nondiabetic subjects aged 40–69 years at baseline (1992–1994); 56% were women, 40% were non-Hispanic white (nHW), 34% were Hispanic (HA), and 26% were African American (AA).

Glucose tolerance was established using 1999 World Health Organization criteria for a 2-h oral glucose tolerance test. Height, weight, WC, hip circumference (HC), and skinfold thicknesses (triceps and subscapular) were measured following a standardized protocol (7). The sum of the skinfold thicknesses (SumSF) was used to indicate overall obesity, and the ratio of subscapular to triceps (STratio) measures was used to determine the ratio of central to peripheral body fat, with a greater ratio indicating a larger proportion of centrally distributed adipose tissue. The RJL (RJL Systems, Clinton Township, MI) method of bioelectrical impedance (BIA) was used to determine percent body fat calculated according to the Segal formula (8). RJL BIA has previously been shown to have good reliability (interclass correlation ≥0.99) (9). Follow-up examinations occurred an average of 5.2 years later, with 146 cases of incident diabetes diagnosed by oral glucose tolerance test.

Logistic regression analysis was used to evaluate the association between baseline measures of obesity and incident diabetes. All anthropometric variables were log transformed, and odds ratios (ORs) were presented per one standard deviation change. Two models were run for each measure: model A controlled for age, sex, and ethnicity, whereas model B controlled for age, sex, and ethnicity as well as family history of diabetes and systolic blood pressure. These variables are common in current diabetes risk scores and are easily and noninvasively obtained in clinical settings. The area under the receiver operating characteristic (AROC) curve (c-statistic) for each model was calculated and used as the primary criterion upon which to judge a model's discriminative ability (10). Different AROCs were statistically compared using the method of DeLong (11). All analyses were performed using SAS, version 9.1 (SAS Institute, Cary, NC).

Associations between anthropometric measures and risk of diabetes for the overall cohort are presented in Table 1. Waist-to-height ratio (WHtR) was the most predictive of diabetes (AROC 0.678; OR 1.79 [95% CI 1.49–2.15]), followed closely by BMI (AROC 0.674; OR 1.76 [1.47–2.10]). The remainder of the indexes had AROCs ranging from 0.667 to 0.625 with no consistent patterning of central compared with overall measures in predicting diabetes. The only measures that were significantly less predictive than WHtR were SumSF and HC (P < 0.05) (Table 1). The results according to AROC ranking were similar after further adjustment for family history of diabetes and systolic blood pressure (see Table A, available in the online appendix at http://care.diabetesjournals.org/cgi/content/full/dc08-1663/DC1), or for fasting glucose concentration.

Table 1

Association of baseline anthropometric measures with the 5-year incidence of type 2 diabetes: the Insulin Resistance Atherosclerosis Study, 1992–1994

Full cohort (n = 1073)
Non-Hispanic white (n = 430)
African American (n = 282)
Hispanic (n = 361)
OR (95% CI)AROCOR (95% CI)AROCOR (95% CI)AROCOR (95% CI)AROC
WHtR 1.79 (1.49–2.15) 0.678 2.25 (1.63–3.10) 0.730 1.62 (1.16–2.26) 0.645 1.55 (1.14–2.10) 0.650 
BMI 1.76 (1.47–2.10) 0.674 2.22 (1.63–3.02) 0.734 1.46 (1.04–2.03) 0.616 1.62 (1.20–2.17) 0.658 
WC 1.75 (1.45–2.12) 0.667 2.25 (1.59–3.17) 0.716 1.51 (1.08–2.11) 0.630 1.59 (1.15–2.19) 0.647 
% BF 2.33 (1.74–3.12) 0.662 3.43 (2.06–5.71) 0.728 1.49 (0.88–2.54) 0.597* 2.12 (1.30–3.38) 0.657 
STratio 1.85 (1.44–2.37) 0.659 2.34 (1.51–3.62) 0.729 2.78 (1.65–4.70) 0.714 1.12 (0.75–1.67) 0.564* 
WHR 1.93 (1.49–2.50) 0.641 2.19 (1.39–3.45) 0.670 2.52 (1.57–4.05) 0.691 1.38 (0.90–2.13) 0.582 
SumSF 1.65 (1.33–2.04) 0.630* 2.01 (1.40–2.90) 0.669* 1.28 (0.85–1.93) 0.585* 1.67 (1.18–2.38) 0.638 
HC 1.42 (1.20–1.69) 0.625* 1.75 (1.31–2.23) 0.694* 1.05 (0.74–1.49) 0.567* 1.45 (1.09–1.93) 0.627 
Full cohort (n = 1073)
Non-Hispanic white (n = 430)
African American (n = 282)
Hispanic (n = 361)
OR (95% CI)AROCOR (95% CI)AROCOR (95% CI)AROCOR (95% CI)AROC
WHtR 1.79 (1.49–2.15) 0.678 2.25 (1.63–3.10) 0.730 1.62 (1.16–2.26) 0.645 1.55 (1.14–2.10) 0.650 
BMI 1.76 (1.47–2.10) 0.674 2.22 (1.63–3.02) 0.734 1.46 (1.04–2.03) 0.616 1.62 (1.20–2.17) 0.658 
WC 1.75 (1.45–2.12) 0.667 2.25 (1.59–3.17) 0.716 1.51 (1.08–2.11) 0.630 1.59 (1.15–2.19) 0.647 
% BF 2.33 (1.74–3.12) 0.662 3.43 (2.06–5.71) 0.728 1.49 (0.88–2.54) 0.597* 2.12 (1.30–3.38) 0.657 
STratio 1.85 (1.44–2.37) 0.659 2.34 (1.51–3.62) 0.729 2.78 (1.65–4.70) 0.714 1.12 (0.75–1.67) 0.564* 
WHR 1.93 (1.49–2.50) 0.641 2.19 (1.39–3.45) 0.670 2.52 (1.57–4.05) 0.691 1.38 (0.90–2.13) 0.582 
SumSF 1.65 (1.33–2.04) 0.630* 2.01 (1.40–2.90) 0.669* 1.28 (0.85–1.93) 0.585* 1.67 (1.18–2.38) 0.638 
HC 1.42 (1.20–1.69) 0.625* 1.75 (1.31–2.23) 0.694* 1.05 (0.74–1.49) 0.567* 1.45 (1.09–1.93) 0.627 

Model A adjusted for age, sex, and ethnicity. ORs are estimated by logistic regression analysis and refer to one standard deviation change in the natural log of the variable. Each row is an individual model with adjustments as indicated.

*AROC P < 0.05 compared with Table's most predictive model. % BF, percent fat from BIA.

There were no significant interactions of sex with anthropometric measures in the prediction of diabetes. Given our a priori interest in ethnic differences in diabetes prediction by anthropometric measurements, logistic models were stratified by ethnicity.

In the nHW and HA populations, BMI was most predictive of diabetes, although there was no clear pattern in central versus overall measures in outcome prediction (Table 1). In contrast, for the AA subgroup, all measures of central adiposity ranked higher in predicting diabetes than overall obesity measures. STratio had the greatest AROC (0.714) and OR (2.78 [95% CI 1.65–4.70]), followed by WHR, WHtR, and WC. HC, percent fat from BIA, and SumSF were significantly less predictive than STratio in model A for this ethnic group (Table 1). Results of analyses in model B were generally similar for each ethnic subgroup (see Table A in the online appendix).

Our main finding was that measures of central adiposity were, in general, not superior to measures of overall obesity in predicting diabetes. In the full cohort, as well as in the nHW and HA subgroups, there was no clear distinction between measures of central versus overall obesity in prediction of diabetes. In contrast, among AA subjects, the results suggested a greater diabetes predictive ability for measures of central adiposity, although there was limited statistical power within this subgroup. Previous studies that have directly measured body fat distribution have shown that AA subjects have less VAT at levels of obesity similar to those of nHW subjects (12). Thus, it is possible that accumulation of greater amounts of VAT has a more detrimental effect on risk of diabetes in AA individuals compared with populations that have higher average levels of VAT.

This is the first paper to examine the predictive ability of a wide range of anthropometric measurements on incident diabetes in three ethnic groups. In previous studies of single or pooled ethnic groups, findings have been inconsistent (1,,,,6), and a recent meta-analysis of 32 studies showed that there was no significant difference between the relative risks of BMI, WC, and WHR for incident diabetes (13).

There are two possible explanations for the nonsuperiority of central versus overall adiposity measures in the prediction of diabetes. First, both VAT and SAT are associated with inflammatory biomarkers and disease risk (14). Second, anthropometric measurements contain a misclassification error in characterizing body fat depots. While WC has been shown to be the simple anthropometric measure that best correlates to VAT, WC actually captures both VAT and abdominal SAT, and it has been reported that WC is more highly correlated to SAT than to VAT (15). Limitations of the present study include a modest sample size for subgroup analysis, a relatively short follow-up period, and skinfold thickness measures from two rather than four sites.

In conclusion, we found no strong evidence that anthropometric measures of central adiposity were more predictive of diabetes than measures of overall obesity in the nHW and HA populations. However, our data suggest that central obesity measures may be useful in predicting diabetes among AA subjects, although additional studies in this population are needed.

The costs of publication of this article were defrayed in part by the payment of page charges. This article must therefore be hereby marked “advertisement” in accordance with 18 U.S.C. Section 1734 solely to indicate this fact.

This study was supported by National Heart, Lung and Blood Institute grants U01-HL47887, U01-HL47889, U01-HL47892, U01-HL47902, DK-29867, and R01 58329 and grant M01-RR-43 from the National Institutes of Health. A.J.G.H. is supported by the Canada Research Chairs Program and the Canadian Diabetes Association.

No potential conflicts of interest relevant to this article were reported.

1.
Meisinger
C
,
Doring
A
,
Thorn
B
,
Heier
M
,
Hannelore
L
:
Body fat distribution and risk of type 2 diabetes in the general population: Are there differences between men and women?
The MONICA/KORA Augsburg Cohort Study Am J Clin Nutr
2006
; 
84
:
483
489
2.
Wang
Y
,
Rimm
EB
,
Stampfer
MJ
,
Willet
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
:
555
563
3.
Tulloch-Reid
MK
,
Williams
DE
,
Looker
HC
,
Hanson
RL
,
Knowler
WC
:
Do measures of body fat distribution provide information on the risk of type 2 diabetes in addition to measures of general obesity?
Comparison of anthropometric predictors of type 2 diabetes in Pima Indians Diabetes Care
2003
; 
26
:
2556
2561
4.
Warne
D
,
Charles
M
,
Hanson
R
,
Jacobsson
L
,
Mccance
D
,
Knowler
W
,
Pettitt
D
:
Comparison of body size measurements as predictors of NIDDM in Pima Indians
.
Diabetes Care
1995
; 
18
:
435
439
5.
Sargeant
LA
,
Bennett
FI
,
Forrester
TE
,
Cooper
RS
,
Wilks
RJ
:
Predicting incident diabetes in Jamaica: the role of anthropometry
.
Obes Res
2002
; 
10
:
792
798
6.
Stevens
J
,
Couper
D
,
Pankow
J
,
Folsom
AR
,
Duncan
BB
,
Nietro
FJ
,
Jones
D
,
Tyroler
HA
:
Sensitivity and specificity of anthropometrics for the prediction of diabetes in a biracial cohort
.
Obes Res
2001
; 
9
:
696
705
7.
Wagenknecht
LE
,
Mayer
EJ
,
Rewers
M
,
Haffner
S
,
Selby
J
,
Borok
GM
,
Henkin
L
,
Howard
G
,
Savage
PJ
,
Saad
MF
:
The Insulin Resistance Atherosclerosis Study (IRAS): Objectives, design and recruitment results
.
Ann Epidemiol
1995
; 
5
:
464
472
8.
Segal
KR
,
Van Loan
M
,
Fitzgerald
PI
,
Hodgdon
JA
,
Van Itallie
TB
:
Lean body mass estimation by bioelectrical impedance analysis: a four-site cross validation study
.
Am J Clin Nutr
1988
; 
47
:
1
14
9.
Fornetti
WC
,
Pivarnik
JM
,
Foley
JM
,
Fiechtner
JJ
:
Reliability and validity of body composition measures in female athletes
.
J Appl Physiol
87
(
3
):
1114
22
,
1999
9
.
10.
Pepe
MS
,
Janes
H
,
Longton
G
,
Leisenring
W
,
Newcomb
P
:
Limitations of the Odds Ratio in gauging the performance of a diagnostic, prognostic, or screening marker
.
Am J Epidemiol
2004
; 
159
:
882
890
11.
DeLong
ER
,
DeLong
DM
,
Clarke-Pearson
DL
:
Comparing the areas under two ore more correlated receiver operating characteristic curves: A nonparametric approach
.
Biometrics
1988
; 
44
:
837
845
12.
Hoffman
DJ
,
Wang
Z
,
Gallagher
D
,
Heymsfield
S
:
Comparison of visceral adipose tissue mass in adults African Americans and Whites
.
Obes Res
2005
; 
13
:
66
74
13.
Vazquez
G
,
Duval
S
,
Jacobs
DR
 Jr
,
Silventoinen
K
:
Comparison of body mass index, waist circumference, and waist/hip ratio in predicting incident diabetes: A meta-analysis
.
Epidemiol Rev
2007
; 
29
:
115
128
14.
Pou
KM
,
Massaro
JM
,
Hoffman
U
,
Vasan
RS
,
Maurovich-Horvat
P
,
Larson
MG
,
Keaney
JF
 Jr
,
Meigs
JB
,
Lipinska
I
,
Kathiresan
S
,
Murabito
JM
,
O'Donnell
CJ
,
Benjamin
EJ
,
Fox
CS
:
Visceral and subcutaneous adipose tissue volumes are cross-sectionally related to markers of inflammation and oxidative stress: The Framingham Heart Study
.
Circulation
2007
; 
116
:
1234
1241
15.
Fox
CS
,
Massaro
JM
,
Hoffman
U
,
Pou
KM
,
Maurovich-Horvat
P
,
Liu
CY
,
Vasan
RS
,
Murabito
JM
,
Meigs
JB
,
Cupples
LA
,
D'Agostino
RB
 Sr
,
O'Donnell
CJ
:
Abdominal visceral and subcutaneous adipose tissue compartments: Association with metabolic risk factors in the Framingham Heart Study
.
Circulation
2007
; 
116
:
39
48

Supplementary data