Body Mass Index: Is It The Ideal Universal
Weight For Height Index?
Prakash Shetty
The human body is composed
of two broad compartments: (i) active
tissue mass (lean body mass or fatfree mass) and (ii) the fat mass which
is the principal store of energy in the
body. In steady state, an individual
will maintain his body weight when
the input of food energy is equal to
the output of energy. Under these
conditions, any change in body weight
indicates a change in energy balance
and, most probably, a change in fat
(or energy) stores. Thus body weight
is a proxy for steady state as well as
reflecting the energy stores of the
individual.
The body weight of an individual
is, however, influenced by the stature
of the individual and hence there is a
need for a nutritional status indicator
or index of acceptable body weight
for height. This would enable comparisons of different individuals of
varying heights and provide a measure of their energy status. A 'weight
for height' index suitable for epidemiological purposes should be simple
to calculate from the measurements
made; should be highly correlated
with the body weight and at the same
time be independent of height. This
implies that the numerical value derived should be the same for individuals of standard weight for that height
at different heights. Such a nutritional
index will reflect the relative weight of
the individual or population while also
reflecting the changes in energy stores
of the individual or group.
WEIGHT FOR HEIGHT INDICES
There are two basic types of weight
for height indices commonly used in
assessing nutritional status of individuals or populations:
Relative weight: This is the ratio of an individual's weight to a standard or expected weight for individuals of his/her height. Age may also be
used as an additional standardising
variable. Relative weight expresses
the weight of a given subject as a
percentage of the average weight of
persons of the same sex, age and
height. The standard or average weight
is frequently derived from large popu-
lation samples. However, differences
between standards can be large across
populations and hence relative weight,
despite being a readily interpretable
and an easily usable measure, is of
limited use for international comparisons. It is also subject to error due to
secular changes in heights of populations.
Power-type index: This is derived by the ratio of weight divided by
height raised to the power 'p' where p
has a value, usually between 1.0 and
3.0. There are three possible simple
power type indices combining weight
and height in use. They are weight!
height (W/H), weight!heighF
(W/H2)
and weight!heighP (W/H3) where p is
equal to 1,2 and 3, respectively. However, in order that these three power
indices recommended for use in the
literature at various times are evaluated, it is useful to remind ourselves
that the most desirable quality of the
power index for international use should
be that it is maximally correlated with
weight and is unbiased by height,
that is, poorly or not at all correlated
with stature in all populations. W/H
consistently shows high correlations
with weight but also correlates with
height in adult populations1. W/H2 or
Body Mass Index (BMI) also shows
consistent high correlations2 with weight
and is consistently independent of
height. W/H3 or Ponderal Index, on
the other hand, shows substantially
lower correlations with weight but also
shows negative correlations with height.
A comparative analysis of these three
power indices in several population
groups showed thatW/H had the highest
correlation with height followed by
W/H3 and suggested that W/H2 or BMI
which is highly correlated with body
weight and is relatively unbiased by
height is the power index of choice for
epidemiological
purposes and for international use3.
BODY MASS INDEX (BMI)
It was the Belgian astronomer
Adolphe Quetelet4 who in 1869 observed, 'nous trouverons que les poids,
chez les individus developpes et de
hauteurs diffferentes, sont a peu pres
comme les carres de tailles' which
meant that the body weight of adults
of different heights is more or less
constant to the square of the height.
In 1972, Ancel Keys5 and colleagues
christened this relationship between
body weight of adults and the square
of the height as the Body Mass Index
(BMI) and 'ungraciously' replaced the
term Quetelet's Index6. The choice of
BMI as the likely objective index for
the assessment of nutritional status
of adults was based on the observation that BMI was consistently highly
correlated with body weight (a proxy
for the available energy stored within
the body) and was relatively independent of the height of the individual.
Overweight and obesity is normally assumed to indicate an excess
of body fat. BMI is used as an indicator of choice to diagnose obesity in
adults and Garrow's earlier classification of obesity? has been replaced
following its universal use by the international
criteria developed
and
endorsed
by the World
Health
Organisation
(WHO). Recent WHOB
recommendations include the suggestion that a BMI of between 18.5 -24.9
in adults be considered appropriate
weight for height. A BMI between 2529.9 is indicative of overweight and
possibly a pre-obese state while obesity is diagnosed at a BMI >30.0 (further classified as 30.0-34.9 moderate,
35.0 -39.9 severe and >40.0 very severe
obesity). The need for a simple objective method to assess undernutrition
in adults, a question deliberated at
the first meeting of the International
Dietary Energy Consultancy Group
which met in Guatemala in 1987 resulted in BMI being recommended as
the suitable indicator9. BMI was subsequently accepted by FA010 as a
simple, responsive and useful indicator of nutritional status of adults to
serve as an important and valuable
tool for monitoring nutritional status
of populations3. Thus BMI has now
become a very valuable nutritional
assessment and monitoring tool that
is useful to assess the continuum of
undernutrition,
normal range and
over-nutrition
(that is, the spectrum
from deficiency
to excess) in the
community.
BMI provides for a most useful
epidemiological tool in estimating the
prevalence of obesity and chronic undernutrition and the increased risk of
morbidity and mortality associated
with these nutritional states. However,
it does not account for the wide
1.19
1.33
0.95
0.94
0.86
0
-0.23
--0.12
-0.10
0.18
2.09
1.69
0.88
1.18
1.38
0.93
0.88
0.85
-0.18
-0.01
-0.20
-0.01
2.03
1.65
0.84
0.87
HtHtWt
1.59
0.96
F
0.92
-0.09
-0.02
F0
M
M
F
M
1.95
0.91
M 0.90
P
Benn
Index
0.85
0.84
0.81
F
TABLE
Coefficients
of
and
Index
and
Power Function
andBMI
(p)
Females
with Benn
Body
(n=17,866)
Weights
ofand
FiveHeights
Different
in Ethnic
Males Groups
(n=17,657)
BMI
ces. It is, therefore, important to know
to what extent the correlation with
weight is attenuated or reduced in the
Benn index in order that it is totally
independent of height. Hence serious attenuation of the correlation with
weight would reduce the usefulness
ofW/HP in comparison with otherweightheight indices although unlike the latter the Benn index may be unbiased
by height.
Source: Lee, J., Kolonel, L.N. and Ward-Hinds, M.: Relative merits of weight-correctedfor-height indices. American Journal of Clinical Nutrition 34: 2521-2529,1981.
variation in body composition and body
fat stores and the effect this may have
on the relationship with increased healthrelated risks. BMI does not distinguish·
between weight associated with muscle
mass from weight due to excess body
fat and hence a given BMI may not
correspond to the same degree of
fatness in different population groups.
It is well recognised that for the same
BMI, different populations or ethnic
groups may have quite different amounts
of body fat. Norgan11 showed that rural individuals from India, Ethiopia and
Papua New Guinea had 12, 7 and 1
per cent body fat, respectively, for the
same BMI of 20. More recently, it has
been shown that Polynesians have a
much lower proportion of fat for an
identical BMI compared to Caucasians
from Australia12.
This question is further complicated by the factthat populations seem
to differ in the level of risk of comorbidities of obesity such as cardiovascular disease (CVD) and non-insulin dependent
diabetes mellitus
(NIDDM) associated with the range of
BM!. It is now undisputed that the risk
of these co-morbidities is much higher
among South Asians (Indians, Pakistanis, Bangladeshis) at a lower BMI
range13.14. This finding is further complicated by the variation in the pattern
of accumulation of fat (that is, more
central or abdominal deposition of
fat) in these ethnic groups and its
relationship with known health risks.
Abdominal accumulation
of fat can
vary markedly for the same per cent
total body fat or BMI. It is now well
accepted that waist circumference as
well as waist-hip ratio (WHR) is a good
indicator of abdominal obesity which
increases risk of obesity-related comorbidities. However, populations differ
in their level of risk associated with
abdominal obesity - a good example
being South Asians who show a disproportionate
increase in risk15. All
this has prompted the need to reevaluate the recommended
cut-offs
for obesity and has raised the sceptre
of the need for population specific
cut-offs for obesity16. The eagerly
awaited report of the recent WHO Expert
Consultation on 'Appropriate BM Is for
Asian populations and its implications
for policy and intervention strategies'
held in Singapore in July 2002 will
certainly make helpful recommendations and throw more light on this
topic.
BMI OR BENN INDEX?
In an attempt to provide a unified alternative to the three power indices in their universal applicability,
Benn17proposed the use ofW/HPwhere
the power function p is derived specifically for each population group.
The value of p would be chosen in
such a manner that the index would
be fully independent of height. The
power p has to be specifically derived
for the population to be studied from
any given set of weight-height measurements and would hence be population specific. Benn also showed that
the index is maximally related to body
weight, provided weight is independent of height. However, since weight
is invariably correlated with height,
the Benn index may not in actuality be
as highly correlated with weight as
the other weight-height
power indi-
6
The relative usefulness of the
Benn index has been evaluated by
Lee et a/I in a sample size of 35,523
subjects of both sexes and ages over
18 years in five diverse ethnic groups
(Caucasians, Japanese, Chinese, Filipinos and Hawaiians). They were able
to show that W/H showed the highest
correlation with height, followed by
W/H3and then W/H2 butW/HP remained
minimally
correlated
with height.
W/HP was the most invariant of height
(r = -0.02) followed by W/H2 (r =0.10);
hence invariance with height is a positive
feature of W/HP as compared to W/H'
or other weight-height indices. W/H'
was highly correlated with weight in
all the ethnic-sex groups, but was
relatively unbiased by height only in
certain groups. There was a distinct
sex difference in their study with W/H'
being generally less height-biased for
males than for females. The Benn index, on the other hand, was consistently unbiased by height and was
also highly correlated with weight in
all the ethnic-sex specific groups and
remained virtually uncorrelated with
height in the pooled data from diverse
populations. An assessment of relative weight, BMI and Benn index based
on the analysis of data on 10,606 men
and women aged 18-74 years obtained
during NHANES II survey has reportedH
that the value for the exponent p fOI
women among the different age range~
varied from 1.16 to 1.62 while that fOI
men were fairly similar and rangec
from 1.69to 1.86. TheTablesummarise~
the analysis carried out by Flegal tc
show that the power function 'p' varies with the ethnic group and with se)c
and also with age (not shown in Table).
Flegal's analysis showed that relativE
weight was still superior to both anc
had a distinct advantage in that rela·
tive weight was perfectly invariant wittheight while the power type index variec
with height in a curvilinear fashion.
If the Benn index was intendec
as a practical tool then its iimitatiom
should also be recognised. Mechani
cal application of the formula will nol
automatically provide the best weightheight index in all situations. Since
W/HP has a non-linear, that is, curvilinear association with height, which
becomes more pronounced the greater
the range of heights, then this argues
against the proposed use of this index in a mixture of heterogeneous
populations that vary considerably in
height. The original choice of W/HP
was made on the criterion that an
ideal weight-height index should be
independent of height. This criterion
is based on the assumption that relative adiposity or fat mass is independent of height, an assumption that
may not always be valid. Earlier studies have shown a better correlation of
skin-fold thickness with BMI rather
than Benn index19. Weight-height indices are only approximate indicators
of relative weight or optimal or acceptable weight for height and of body
energy stores. Practical considerations
such as ease of computation is also
an important criterion. BMI is more
readily computable than W/HP. Benn
recommended use of either relative
weight or a power-type index with p
'rounded to the nearest whole number to make computation easy'.
From an operational point of view,
it may be concluded that the closest
integer of 'p' is 2 and hence it may be
concluded that W/H2 is possibly the
best index of the type W/HP in populations and thus the most useful in practice.
However, given that we are beginning
to question the appropriateness
of
BMI and considering the need to reevaluate the recommended
cut-offs
based on variation of risk among different population groups, this may
indeed be an opportunity to evaluate
once again whether the use of a population specific power index may actually help us assess the risk of obesity
and its co-morbidities better than we
have been able to do so far.
The author
Nutrition
Division
is Senior
Officer
in the Food
and
of FAO.
References
1. Lee, J., Kolonel, L.N and Ward-Hinds, M.: (1981)
Relative merits of weight-corrected-for-height
indices. American Journal of Clinical Nutrition 34: 25212529,1981.
2. Khosla, T. and Lowe, C.R.: Indices of obesity
derived from body weight and height. British JournalofPreventive&
Social Medicine 21: 122-128,1967.
3. Shelly, P.S. and James, W.P.T.: Body Mass
Index: A measure of chronic energy deficiency in
adults. FAO Food & Nutrition Paper 56. Rome:
Food & Agriculture Organisation, 1994.
4. Quetelet, A.: (1869) Physique sociale ou essay
sur Ie development des facultes de I'homme Bruxelles:
C. Muquardt, 1869
5. Keys, A., Fidanza, F., Karvoven, M.J., Kimura,
N. and Taylor, H.C.: Indices of relative weight and
obesity. Journal of Chronic Diseases.25: 329-343,1972.
6. Garrow, J.S.: Obesity and related diseases.
London: Churchill Livingstone. 329pp, 1988.
7. Garrow, J.S.: Treat obesity seriously: A clinical
manual. London: Churchill Livingstone. 246pp, 1981.
8. WHO (2000) Obesity: Preventing and managing the global epidemic. WHO Technical Report
series 894. Geneva: World Health Organisation.
9. James, W.P.T., Ferro-Luzzi, A. and Waterlow,
J.C.: Definition of chronic energy deficiency in
adults. European Journal of Clinical Nutrition 42:
969-981,1988.
10. FAO: The state of the food insecurity in the
world 2000. Rome: Food & Agriculture Organisation,
2000.
11. Norgan, N.G.: Body Mass Index and body energy stores in developing countries. European Journal
of Clinical Nutrition 36C: 79-84,1990.
12. Swinburn, B.A., Craig, P.L., Daniel, R., Dent,
D.P., and Strauss, B.J.: Body composition differences between Polynesians and Caucasians assessed by bioelectrical impedance. International
Journal
of Obesity
& Related
Metabolic
Disorders
20:889-894,1996.
13. Reddy, K.S.: The emerging epidemiology of
cardiovascular disease in India. In: Diet, nutrition
and chronic disease: An Asian Perspective (Editors: Shelly & Gopalan). London: Smith-Gordon.
P50-54, 1998
14. Yajnik, C.S.: Diabetes in Indians: Small at birth
or big as adults or both? In: Diet, Nutrition and
Chronic
Disease:
An Asian Perspective (Editors:
Shelly & Gopalan). London: Smith-Gordon. p4346, 1998.
15. McKeigue, P.M.: Metabolic consequences of
obesity and body fat pallern: lessons from migrant
studies. In: The Origin and Consequences
of Obesity (Editors: Chadwick & Cardew) Chichester:
Wiley.54-67, 1996.
FOUNDATION
NEWS
Study Circle Lectures
• Dr Ashish Bose (Demographer and
Honorary Professor at the Institute of
Economic Growth) on 'National Human Development Report 2001' on
July 19, 2002
• Dr N. Kochupillai (Professor, Department of Endocrinology, All India
Institute of Medical Sciences) on 'Maternal nutrition and foetal imprinting'
on August 27, 2002
Dr Sarath Gopalan, Honorary
Director of the Centre of Research for
Nutrition Support Systems has been
appointed as the Indian Editor of the
international journal, Nutrition, published from the USA. He will take over
this responsibility
with effect from
January 1, 2003.
Foundation
Day
The Annual Foundation Day of
the Nutrition Foundation of India will
be celebrated on November 25. Dr
N.K. Ganguly, Director General, Indian Council of Medical Research,
will give the C. Ramachandran Memorial Lecture on that occasion.
NUTRITION
NEWS
18. Flegal, K.N: Ratio of predicted weight as an
alternative to a power type weight-height index
(Benn index). American Journal of Clinical Nutrition
51: 540-547,1990.
• Eighth Annual Conference of the
Indian Society for Parenteral and
Enteral Nutrition (ISPEN) will be held
from December 6-8,2002, in Hyderabad
at Hotel Viceroy.
The Scientific
Programme will include actual clinical case capsules for interactive discussions, controversies in clinical nutrition in addition to guest lectures,
symposia and free papers. The first
date of registration is November 15,
2002 and the deadline for submission
of abstracts is November 10, 2002.
19. Frisancho, A.R and Flegal, K.N.: Relative merits
of old and new indices of body mass with reference
to skin-fold thickness. American Journal of Clinical
Nutrition 36: 697-699,1982.
• PENSA 2003: .India will be hosting
the IX Annual Conference
of the
Parenteral and Enteral Nutrition Society of Asia (PENSA) at Goa from
16. WHO/IASO/IOTF The Asia-Pacific perspective:
Redefining obesity and its treatment. WHO Regional Office Western Pacific Region pp 56, 2000.
17. Benn, R.T.: Some mathematical properties of
weight for height indices used as measures of
adiposity. British Journal of Preventive & Social
Medicine,
25: 42-50,1971.
November 6-8, 2003. The other countries which are members of PENSA
(which have registered PEN Societies
in their respective countries) are Thailand, Malaysia, Taiwan, Korea, Indonesia, Japan and the Philippines. The
Scientific Programme of the Conference would comprise issues pertaining to hospital-based
nutrition support. India will be hosting the PENSA
Meeting for the first time.
• The annual meeting of the European Society for Parenteral and Enteral Nutrition (ESPEN 2002) was held
from August 31- September 4,2002 at
Glasgow, Scotland. It was attended
by over 2,000 delegates. The Scientific Programme of the Congress included a diverse range of topics such
as 'Nutrition in space' and 'Geo-gravity'. The current status regarding potential benefit of some immunonutrients
was discussed and the available scientific evidence was acknowledged.
•
IX Asian Congress of Nutrition
Arrangements for the Congress
which is to take place between February 23-27,2003 at Hotel Ashok, New
Delhi are proceeding satisfactorily.
The Prime Minister of India, who
is the Patron of the Congress, has
kindly agreed to inaugurate the Congress. A comprehensive
scientific
programme consisting of 6 Plenary
Lectures, 6 Plenary Sessions, 30 Symposia, 10 Free Communication
Sessions and daily Poster Sessions is
being organ;sed.
The Plenary Sessions
dress the following areas:
• Changing
nutrition scene in Asia,
• Newer technologies
ing food production
century,
• Diet and genes:
nutrigenomics, and
• Traditional
of Asia.
will ad-
for augmentin the 21 st
nutrigenetics/
health promoting foods
In addition there will be the following two innovative programmes.
• Meet the Professors session with
the theme 'Nutrition Vision for the
Future' led by Director-General, ICMR,
Dr N.K. Ganguly
and Professor
M.K. Gabr. The participants include
Prof J.C. Waterlow, Dr J.E. Dutra
Oleivera, Prof Z.A. Bhutta and Prof
M.A.G. Tuazon.
• Science-Industry
Interactive Session with the theme 'Role of Industry
in the Advancement of Nutrition Science' led by Director-General, CSIR,
Dr R.A. Mashelkar and Dr Florentino
Solon with other participants being
Mr Poddar and Mr R.C. Venkateish to
be followed by general discussion.
The Congress has aroused wide
interest among scientists of all Asian
countries as also of Europe, America
and Africa. The Organising Committee of Congress is keen that a large
number
of delegates,
especially
from Asian countries, should participate. To encourage participation of
foreign delegates
the Organising
Commetee has decided the following
arrangement:
Special'concession
rate of $250
on Registration Fee will be available
for groups participating
from each
country in numbers of Six or more
sponsored by the official nutrition society
of the country or any other recognized institution. For student delegates
participating in groups of Six or more
the Registration Fee will be $150 each.
The students have to furnish a letter in
this regard from a recognised university or institution. The details regarding the name, designation of the delegates in the group and total sum of
Registration Fee of all the delegates
in the group should be sent in advance before October 15, 2002 to the
Congress Secretariat.
In response to the request made
by Congress Organisers, the authorities of Hotel Ashok (the Congress venue)
and Hotel Samrat (adjoining the Congress venue) have reduced the room
tariff for the period of Congress to
US$ 84 per day for Hotel Ashok, the
Congress venue and US$ 64 per day
for the adjoining Hotel Sam rat.
The details of the Congress are
also available
on the website:
www.acn2003india.net.
The United Nations System
Standing Committee on Nutrition
(SCN) 30th Session, including a oneday Symposium on 'Mainstreaming
Nutrition to Improve Development
Outcomes' will take place in Chennai,
India, hosted by Professor
M.S.
Swaminathan, March 3-7, 2003.
Registration
and programme
details will be available on the SCN
website: www.unsystem.org/scnfrom
mid-October 2002.
DR THAMMU ACHAYA
Dr Thammu
Achaya,
former President of the Nutrition Society of India, passed
away on September 5 and India lost a distinguished scientist. Dr Achaya started his
scientific career as a specialist in oils and fats and had
made important contributions
in this area. In later years he
distinguished himself as a versatile and erudite scholar in
the broad field of Food Science. He was the author of
widely-read
bestsellers
on
the history of Indian foods and
food practices. These valuable
and well-researched publications could adorn any good
library.
Dr Achaya was elected
President of the Nutrition Society of India; his contributions
during his term of office, towards strengthening the society, were widely appreciated.
He was a genuine friend of the
Nutrition Foundation of India
and was greatly interested in
its programmes. He had contributed several papers to the
Bulletin of the Nutr~tion Foundation of India.
Dr Achaya's interests extended far beyond science into
the fields of music, arts and
culture. He was a connoisseur of both Carnatic and
Western classical music. He
was a highly 'civilised' human
being, always soft spoken and
courteous; and was no seeker
of power or publicity. It is to
be hoped that the example
he has set will be emulated.
Nutrition scientists of India
will gratefully remember his
contributions.
The legacy he
has left behind will be deeply
cherished.
C.G.
The Nutrition
Foundation
of
India is grateful to FAO and
WHO for matching grants
towards the cost of this
Edited by Dr (Mrs) Hema Sarath Gopalan for the Nutrition Foundation of India, C-13 Qutab Institutional Area, New Delhi 110016.
e-mail: [email protected]
Designed and produced
by Media Workshop
India Pvt Ltd. e-mail: [email protected]
publication
website:
www.nutritionfoundationofindia.or(