Archives of Medical Research 34 (2003) 428–432

BRIEF REPORT
Abdominal Volume Index. An Anthropometry-Based Index
for Estimation of Obesity Is Strongly Related to Impaired
Glucose Tolerance and Type 2 Diabetes Mellitus
Fernando Guerrero-Romeroa,b and Martha Rodrı́guez-Morána,b
a
Unidad Médica de Investigación en Epidemiologı́a Clı́nica, Instituto Mexicano del Seguro Social (IMSS), Durango, Mexico
b
Grupo de Investigación sobre Diabetes y Enfermedades Crónicas, Durango, Mexico
Received for publication October 10, 2002; accepted May 14, 2003 (02/196).

Background. Our objective was to develop an anthropometric-based index (abdominal

volume index, AVI) for estimating overall abdominal volume and to determine its
relationship with presence of impaired glucose tolerance (IGT) and type 2 diabetes
mellitus (DM).
Methods. We conducted a cross-sectional, population-based study between November
1998 and June 2001 among 746 men and non-pregnant women randomly recruited from
Durango City in northern Mexico. AVI was calculated using volume formulas for cylinder
(V ⫽ πr2h) and vertical cone V ⫽ (1/3)πr2h. The formula developed was AVI ⫽ [2 cm
(waist)2 ⫹ 0.7 cm (waist-hip)2]/1,000, which estimates overall abdominal volume between
symphysis of pubis and xiphoid appendix and theoretically includes intra-abdominal fat
and adipose tissue volumes.
Results. Receiver operating curve (ROC) scatter plot showed as best cut-off value of AVI
for estimation of obesity, corresponding to 24.5 liters (L). Logistic regression analysis
adjusted by age and sex showed higher odds ratio between AVI and IGT 1.6 (95%
confidence interval [95% CI] 1.1–9.1, p ⫽ 0.01) as well as between AVI and DM 2.1
(95% CI 1.3–7.9, p ⫽ 0.001) than odds ratio (OR) estimated by other anthropometric
obesity criteria such as waist-to-hip ratio, body mass index, truncated cone, and waist
circumference.
Conclusions. AVI is a reliable and easy-to-calculate anthropometric tool for estimation
of overall abdominal volume that is shown to be strongly related to IGT and DM.
쑖 2003 IMSS. Published by Elsevier Inc.
Key Words: Obesity, Anthropometry, Impaired glucose tolerance, Diabetes.

Introduction have been independently associated with development of

For epidemiologic purposes, estimation of obesity is accom-
plished by extrapolation from anthropometric measures (1).
In this regard, abdominal adiposity pattern is commonly
determined by waist-to-hip ratio (WHR) or waist circumfer-
ence (WC), and overall obesity by body mass index (BMI).
Both overall obesity and unfavorable body fat distribution
Address reprint requests to: Fernando Guerrero-Romero, Siqueiros #225,
esq. Castañeda, 34000 Durango, Dgo., México. Phone: (⫹52) (618) 812-
0997; FAX: (⫹52) (618) 813-2014; E-mail: guerrero_romero@hotmail.com
type 2 diabetes mellitus (DM) (2–6). In this regard, central
obesity estimated by WHR or WC was a better indicator of
risk for DM (7,8) than BMI. However, both WHR and WC
were based on longitudinal measures that were not very
reliable to estimate central obesity in obese subjects who
develop pendulum abdomen, in which line of umbilicus falls
under line of hip.
Because obesity is reaching epidemic proportions, to de-
velop reliable anthropometric indices of obesity that closely
reflect risk of development of glucose metabolic distur-
bances and that are thus useful for clinicians in counseling

0188-4409/03 $–see front matter. Copyright 쑖 2003 IMSS. Published by Elsevier Inc.
d o i : 10 .1 0 1 6/ S0 1 88 - 4 40 9 ( 0 3) 0 0 07 3 - 0
 Abdominal Volume Index
patients regarding promotion of health status is a priority
public health issue. The purposes of this study were to de-
velop an anthropometric-based index for estimation of ab-
dominal volume and to determine its relationship with
presence of impaired glucose tolerance (IGT) and DM.
Methods. With protocol approval by the Mexican Social
Security Institute (IMSS) Research Committee (October
1998) and after obtaining signed patient informed consent,
a cross-sectional, population-based study was conducted be-
tween November 1998 and September 2001. A total of 546
women and 200 men from the city of Durango in northern
Mexico were randomly recruited from a middle-income
neighborhood. Subjects with acute illnesses as well as per-
sons with previous diagnosis of diabetes were not included;
only newly diagnosed subjects with DM were included.
Height (m), weight (kg), waist (cm), and hip (cm) were
measured under fasting conditions with subjects in light
clothing and without shoes. WC was taken as the minimum
circumference at umbilicus level, and hip circumference as
maximum circumference at symphysis of pubis level. BMI
was calculated as weight (in kilograms) divided by height
(in meters) squared, and WHR as WC divided by hip circum-
ference. Furthermore, we calculated volume of truncated
cone, as has been described (9).
Abdominal volume index (AVI) was calculated using
volume formulas for cylinder (V ⫽ πr2h) and vertical cone
V ⫽ (1/3)πr2h (Figure 1, Appendix): the AVI formula devel-
oped is: AVI ⫽ [2 cm (waist)2 ⫹ 0.7 cm (waist–hip)2]/1,000,
in which both waist and hip measurements are in centimeters
(cm). Diagnosis of IGT and DM was established according
to American Diabetes Association criteria (10).
A venous whole blood sample was collected after 8–10
h of fasting and 2-h post 75-g oral glucose load (2 h PG).
Serum glucose was measured by glucose-oxidase method
with intra-assay coefficient of variation (CV) of 2.5%.
Figure 1. Geometric figures that represent obesity patterns in women (A) and men (B). b ⫽ hip circumference at symphysis of pubis level; c ⫽ waist
circumference at umbilicus level; d ⫽ cylinder that results to project waist-on-hip (Figure 1A) or hip-on-waist (Figure 1B); a, a′ ⫽ cylindrical triangle that
results projecting a line between an outside point of the diameter of waist-on-hip circumference; –o–, umbilicus; ▼, pubis. Abdominal volume results of
adding volume of cylinder d to volume of cone a⫹a′.
429
Statistical analysis. Differences between groups were estab-
lished by unpaired Student t test. Pearson analysis was per-
formed to examine correlation between AVIs, as well as
anthropometric diagnosis criteria of obesity and continu-
ous variables. For statistical analysis purposes, all skewed
numerical data were log-transformed.
Sensitivity and specificity of AVI for diagnosis of central
obesity were estimated according to the following formulas:
true positives/(true positives ⫹ false negatives) and true neg-
atives/(true negatives ⫹ false positives), respectively (11).
Optimal AVI value for diagnoses of obesity was established
on receiver operating characteristic (ROC) scatter plot. Pres-
ence of both high WHR (WHR ⱖ0.80 in women and ⱖ0.90
in men) and WC (WC within 4th quartile of distribution)
were considered the gold standard for anthropometric esti-
mation of central obesity.
By calculating odds ratio (OR) using logistic regression
analysis adjusted by age and sex, the relationship between
both IGT and DM and the several anthropometric measure-
ments of obesity was estimated.
Results
Seven hundred forty-six subjects including 546 women
(73.2%) and 200 men (26.8%) with average age of 40.7 ⫾
11.2 years were included. Upper quartile of AVI was 22.5
L, and best AVI value for diagnosis of obesity determined
by ROC scatter plot was 24.5 L, which showed highest
sensitivity (86.5%) and specificity (93%) values (Figure 2).
Thus, subjects (128 women and 48 men) with AVI ⱖ24.5 L
were defined as obese and were allocated as subjects at risk
for IGT and DM. AVI showed high positive correlation with
other anthropometric measurements of obesity such as WC
(r ⫽ 0.757, p ⫽ 0.005), WHR (r ⫽ 0.748, p ⫽ 0.005), BMI
(r ⫽ 0.471, p ⫽ 0.01), and volume estimated using truncated
cone formula (r ⫽ 0.897, p ⫽ 0.0005).
430 Guerrero-Romero and Rodrı́guez-Morán / Archives of Medical Research 34 (2003) 428–432

as WHR (OR 1.2, 95% CI 1.7–6.8, p ⫽ 0.04), WC (OR 1.3,

95% CI 1.1–6.5, p ⫽ 0.02), truncated cone (OR 1.4, 95%
CI 1.1–8.4, p ⫽ 0.01), and BMI (OR 1.1, 95% CI 0.9–1.6,
p ⫽ 0.04), although also related to IGT, showed a lower
relationship.
On the other hand, AVI also was the parameter of obesity
that showed strongest relationship to DM (OR 2.1, 95% CI
1.3–7.9, p ⫽ 0.001), whereas WHR (OR 1.6, 95% CI 1.1–
9.1, p ⫽ 0.01), WC (OR 1.9, 95% CI 1.2–7.0, p ⫽ 0.01),
truncated cone (OR 1.9, 95% CI 1.2–9.2, p ⫽ 0.001), and
Figure 2. Receiver operating characteristic (ROC) scatter plot showing BMI (OR 1.2, 95% CI 1.0–3.7, p ⫽ 0.04) showed a lower
sensitivity and specificity of abdominal volume index (AVI) for diagnos- relationship.
ing obesity. Anthropometric gold standard for diagnosis of central obesity
was the combination of both high waist-to-hip ratio (ⱖ0.80 in women and
ⱖ0.90 in men) and high waist circumference (waist circumference within
upper quartile of distribution). AVI best cut-off point for diagnosis of obesity Discussion
was that corresponding to 24.5 L.
Results of this study demonstrated that AVI was a reliable
and easy-to-calculate anthropometric tool for estimation of
obesity that was strongly related to IGT and DM. Reliability
Table 1 shows correlation index between different anthro- and accuracy of the relationship between AVI and presence
pometric measurements of obesity and fasting and 2-h post- of IGT and new diagnosis of type 2 DM was corroborated
load glucose levels. AVI had best correlation index between comparing Pearson correlation and OR values shown by
obesity and glucose levels for both women and men, showing AVI vs. those shown by most common anthropometric
progressive increase in both fasting and 2-h postload glucose obesity criteria. In this regard, AVI had highest r and OR
concentrations by AVI categories (Figure 3). Subjects with values. Among several anthropometric obesity tools evalu-
highest AVI values showed highest fasting and post-load ated, WC, a good anthropometric indicator of visceral ab-
glucose levels. dominal fat (12), showed best correlation with AVI.
On the other hand, obese subjects according to AVI cri- On the other hand, as expected, the volume estimated by
teria exhibited highest blood pressure, weight, WHR, BMI, truncated cone formula also showed high correlation with
WC, and serum glucose levels (Table 2). New diagnosis of AVI. In this regard, although obesity pattern on which we
diabetes was established in 67 (9.0%) subjects and IGT based AVI calculation is similar to truncated cone figure and
was identified in 114 (15.3%) participants. All diabetic that estimation of volume by AVI formula could underesti-
subjects and 89.5% of subjects with IGT were obese mate real abdominal volume, our results showed better corre-
according to AVI criteria. lation of AVI with IGT and DM than that shown by truncated
Logistic regression model used to analyze the relationship cone. This finding could be explained taking into account
between anthropometric measurements of obesity and IGT that obesity pattern in our population was not actually distrib-
showed highest odds ratio (OR) for AVI (OR 1.6, 95% CI uted as truncated cone geometric figure. Thus, probable un-
1.1–9.1, p ⫽ 0.01), whereas other parameters of obesity such derestimation of abdominal volume by AVI formula was
minimal and did not affect estimation of overall volume.
Furthermore, estimation of volume by AVI was simpler
to calculate than volume by truncated cone; in addition, as
Table 1. Pearson’s correlation between anthropometric measurements distribution of fat is gender-dependent predominating on the
of obesity and glucose concentrations lower half of the body in women and on the upper half in
men, estimation of abdominal volume by AVI also possessed
Women (n ⫽ 546) Men (n ⫽ 200) the advantage that it envisioned both patterns. Because AVI
Glucose Glucose
was an estimation of overall abdominal volume that theoreti-
Fasting 2-h Postload Fasting 2-h Postload cally included both intra-abdominal fat and adipose tissue
volume, it could explain the high relationship between AVI
Abdominal 0.497a 0.470a 0.448a 0.387b
volume index and presence of IGT and DM that we documented.
Volume from 0.425a 0.415a 0.398b 0.375b Several potential limitations of this study deserve to be
truncated cone mentioned. First, in development of AVI formula we em-
Waist circumference 0.288b 0.418a 0.245b 0.364b ployed the most approximate and simple geometric figure
Waist-to-hip ratio 0.307b 0.345b 0.259b 0.333b
to calculate volume; for example, the a⫹a′ cone (Figure 1)
Body mass index 0.113c 0.102c 0.103c 0.171c
is not a true cone but is two cylindrical triangles. However,
a
p ⬍0.001; bp ⬍0.05; cp, not significant (NS). differences of volumes between such geometrical figures are
 Abdominal Volume Index 431

Figure 3. Mean and standard deviation of serum fasting (black circles) and postload (black squares) glucose levels according to abdominal volume index
(AVI) in all participants (n ⫽ 746). Fasting and postload glucose levels showed significant increase from AVI stratum of 20–24.9 L.

minimal, but volume of cone has the advantage that it is Acknowledgments

easier to calculate than volume of cylindrical triangle.
Second, we did not measure subcutaneous adipose tissue or
intra-abdominal fat in a direct manner; thus, we cannot com-
pare AVI to CT abdominal fat volume measurements. In
this regard, taking into account the clearly conveyed aim
of this study, absence of such validation does not affect our
conclusions. Nonetheless, further research will be necessary
to establish correlation between AVI and intra- and abdomi-
nal-subcutaneous adipose tissues. Finally, for developing the
formula we used average value of the line between symphy-
sis of pubis and xiphoid appendix, a line that could exhibit
differences due to ethnicity and height; however, using aver-
age value the formula can be simplified and does not signifi-
cantly affect estimation of abdominal volume. In conclusion,
results of this study showed that AVI is simpler to calculate
and demonstrated a higher relationship to IGT and DM than
other anthropometric measurements of obesity.
Table 2. Anthropometrics and laboratory characteristics according
to abdominal volume index (AVI) quartile distribution
AVI
Quartile of distribution
Uppera 2nd and 3rd
This work was supported by grants from the Sistema de Investiga-
ción Regional Francisco Villa (SIVILLA) (grant #20000402008)
and the Fondo para el Fomento a la Investigación (FOFOI) of the
Mexican Social Security Institute (grant #FP 2001/354).
Appendix
Abdominal volume index (AVI) ⫽ results of adding volumes
of cylinder d plus cone a⫹a′, which represents overall ab-
dominal volume between symphysis of pubis and xiphoid
appendix (Figure 1).
AVI was calculated using volume formulas for cylinder
(V ⫽ πr2h) and vertical cone (V ⫽ (1/3)πr2h). Because ratio
(r) value in the formulas is an unknown measurement, it
was estimated by circumference formula (C ⫽ 2πr); thus,
r ⫽ C/2π.
To calculate volume of cylinder d, we replaced r value
in the corresponding volume formula:
V ⫽ π(C/2π)2h
V ⫽ π(C2/4π2)h
V ⫽ (C2/4π)h
Lower
V ⫽ (0.0796)C2h

Weight (kg) 87.1 ⫾ 15.2 75.3 ⫾ 10.3 62.9 ⫾ 9.8 h Value is the line between xiphoid appendix and symphysis
Abdominal volume 26.7 ⫾ 3.8 18.9 ⫾ 2.1 13.4 ⫾ 2.0 of pubis and C is the waist circumference (Figure 1). In our
index, liters (L)
population, average h value for men and women is 26 cm
Waist-to-hip ratio 1.0 ⫾ 0.1 0.9 ⫾ 0.1 0.8 ⫾ 0.1
Waist (cm) 115.2 ⫾ 7.9 96.9 ⫾ 5.6 80.9 ⫾ 6.5 (26.5 ⫾ 4.5 cm and 24.8 ⫾ 3.3 cm for men and women,
Body mass index (kg/m2) 33.9 ⫾ 6.4 28.5 ⫾ 3.1 24.8 ⫾ 3.5 respectively). Thus: V ⫽ (0.0796) (26 cm) (waist)2; V ⫽ 2
Systolic blood 122.9 ⫾ 17 114.2 ⫾ 20 109.7 ⫾ 13.5 cm (waist)2.
pressure, mmHg On the other hand, volume of cone a⫹a′ was estimated as:
Diastolic blood 77.1 ⫾ 9 72.7 ⫾ 10 69.8 ⫾ 10
pressure, mmHg V ⫽ (1/3)πr2h
Fasting glucose, mmol/L 6.0 ⫾ 1.2 5.5 ⫾ 1.1 5.2 ⫾ 1.0 V ⫽ (1/3)π(C/2π)2h
2-h Post-load 7.6 ⫾ 2.0 6.6 ⫾ 1.9 6.2 ⫾ 1.6
V ⫽ (1.0472)(C2/4π2)h
glucose, mmoL/L
V ⫽ 1.0472(C2/39.4786) h
Values are mean ⫾ standard deviation (SD); aCut-off point of AVI ⫽ 24.5 L. V ⫽ 0.0265C2h
432 Guerrero-Romero and Rodrı́guez-Morán / Archives of Medical Research 34 (2003) 428–432
Circumference (C value in the formula) of this hypotheti-
cal cone is the result of subtracting waist from hip circumfer-
ence, while h value is the line between xiphoid appendix
and symphysis of pubis. Thus, the formula can be written
as follows:
V ⫽ 0.0265 (26 cm) (waist-hip)2
V ⫽ 0.7 cm (waist-hip)2
Adding volume of cylinder d to cone a⫹a′, developed
formula is:
AVI ⫽ 2 cm (waist)2 ⫹ 0.7 cm (waist-hip)2, in which
both waist and hip are expressed in cm.
For transforming to liters, it is necessary to divide the
result by 1,000; thus, the final formula is:
AVI ⫽ [2 cm (waist)2 ⫹ 0.7 cm (waist-hip)2]/1,000
In the formula, values 2 and 0.7 are constant.
References
1. Valdéz R, González-Villalpando C, Mitchell BD, Haffner SM, Stern
MP. Differential impact of obesity in related populations. Obes Res
1995;3(Suppl 2):223–232.
2. Björntorp P. The associations between obesity, adipose tissue distribu-
tion and disease. Acta Med Scand (Suppl) 1988;723:121–134.
3. Ramachandran A, Snehalatha C, Satyavani K, Sivasankari S, Vijay V.
Cosegregation of obesity with familial aggregation of type 2 diabetes
mellitus. Diabetes Obes Metab 2000;2:149–154.
4. Pan XR, Yang WY, Li GW, Liu J. Prevalence of diabetes, and its
risk factors in China, 1994. National Diabetes Prevention and Control
Cooperative Group. Diabetes Care 1997;20:1664–1669.
5. Chan JM, Rimm EB, Colditz GA, Stampfer MJ, Willett WC. Obesity,
fat distribution, and weight gain as risk factors for clinical diabetes in
men. Diabetes Care 1994;17:961–969.
6. Haffner SM, Stern MP, Hazuda HP, Pugh J, Patterson JK. Do upper-
body and centralized adiposity measure different aspects of regional
body-fat distribution? Relationship to non-insulin-dependent diabetes
mellitus, lipids, and lipoproteins. Diabetes 1987;36:43–51.
7. Haffner SM, Mitchell BD, Stern MP, Hazuda HP, Patterson JK. Public
health significance of upper body adiposity for non-insulin dependent
diabetes mellitus in Mexican Americans. Int J Obes Relat Metab Disord
1992;16:177–184.
8. Schmidt MI, Duncan BB, Canani LH, Karohl C, Chambless L. Associa-
tion of waist-hip ratio with diabetes mellitus. Strength and possible
modifiers. Diabetes Care 1992;15:912–914.
9. Valdéz R, Seidell JC, Ahn YI, Weiss KM. A new index of abdominal
adiposity as an indicator of risk for cardiovascular disease. A cross-
population study. Int J Obes Relat Metab Disord 1993;17:77–82.
10. Report of the Expert Committee on the Diagnosis and Classification
of Diabetes Mellitus. Diabetes Care 1999;21(Suppl 1):5–20.
11. Kramer MS, editor. Clinical epidemiology and biostatistics. A primer
for investigators and decision-makers. Berlin, Germany: Springer-
Verlag;1988. pp. 201–210.
12. Keller C, Chintapalli K, Lancaster J. Correlation of anthropometry with
CT in Mexican-American women. Res Nurs Health 1999;22:145–153.