Agricultural development and nutrition

Material Information

Agricultural development and nutrition
Series Title:
Agricultural development and nutrition
Pacey, Arnold
Payne, Philip
Place of Publication:
Boulder CO
Hutchinson ;
Westview Press
Publication Date:
Physical Description:
255 p. : ill. ; 23 cm.


Subjects / Keywords:
Food supply -- Developing countries ( lcsh )
Nutrition -- Developing countries ( lcsh )
bibliography ( marcgt )
non-fiction ( marcgt )
Spatial Coverage:


Includes bibliography and indexes.
General Note:
Published by arrangement with the Food and Agriculture Organization of the United Nations and the United Nations Children's Fund.
Statement of Responsibility:
edited by Arnold Pacey and Philip Payne.

Record Information

Source Institution:
University of Florida
Holding Location:
University of Florida
Rights Management:
The University of Florida George A. Smathers Libraries respect the intellectual property rights of others and do not claim any copyright interest in this item. This item may be protected by copyright but is made available here under a claim of fair use (17 U.S.C. §107) for non-profit research and educational purposes. Users of this work have responsibility for determining copyright status prior to reusing, publishing or reproducing this item for purposes other than what is allowed by fair use or other copyright exemptions. Any reuse of this item in excess of fair use or other copyright exemptions requires permission of the copyright holder. The Smathers Libraries would like to learn more about this item and invite individuals or organizations to contact Digital Services ( with any additional information they can provide.
Resource Identifier:
12935194 ( OCLC )
85050544 ( LCCN )
081330265X (US) ( ISBN )

Full Text

Agricultural Development
and Nutrition



Development and


Edited by Arnold Pacey and Philip Payne

London Melbourne Sydney Auckland Johannesburg WESTVIEW PRESS Boulder, Colorado by arrangement with the Food and Agriculture Organization of the United Nations and the United Nations Children's Fund

Hutchinson and Co. (Publishers) Ltd An imprint of the Hutchinson Publishing Group 17-21 Conway Street, London WiP 6JD Hutchinison Publishing Group (Australia) Pty Ltd 16-22 Church Street, Hawthorn, Melbourne, Victoria 3122 Hutchinson Group (NZ) Ltd 32-34 View Road, PO Box 40-086, Glenfield, Auckland 10 Hutchinson Group (SA) (Pty) Ltd PO Box 337, Bergvlei 2012, South Africa First published 1985
Published in 1985 in the United States of America by
Westview Press, Inc.
5500 Central Avenue
Boulder, Colorado 80301
Frederick A. Praeger, President and Publisher � 1985 by FAO and UNICEF The designations employed and the presentation of material in this publication do not imply the expression of any opinion whatsoever on the part of the Food and Agriculture Organization of the United Nations or the United Nations Children's Fund concerning the legal status of any country, territory, city or area or of its authorities, or concerning the delimitation of its frontiers or boundaries.
The designations 'developed' and 'developing' economies are intended for statistical convenience and do not necessarily express a judgement about the stage reached by a particular country or area in the development process. All rights reserved. No part of this publication may be reproduced, stored in a retrieval system, or transmitted in any form or by any means, electronic, mechanical, photocopying or otherwise, without the prior permission of th e copyright oner. Applications for such permission, with a statement of the purpose and exte t of the reproduction, should be addressed to the Director, Publications Div4ison, Food and Agriculture Organization of the United Nations, Via de e Terme di Caracalla, 00100 Rome, Italy. Set in Linotype Times by Saxon Ltd, Derby, England. Printed and bound in Great Britain by Anchor Brendon Ltd, Tiptree, Essex. British Library Cataloguing in Publication Data Agricultural development and nutrition.
1. Food crops-Tropics
I. Pacey, Arnold II. Payne, P$hilip
338.1'9'0913 SB176.T7
Library of Congress Catalog Card Number 85-50544 ISBN (UK) 0091613302 cased
(UK) 0 091613310 paper (US) 0-8133-0265-X cased


Foreword 9
Preface 11

Part One Limits to Measurement 15

1 Food systems and needs 17
Changes have taken place in views of malnutrition
in the last two decades, and these are outlined.
Two examples of new approaches are given - the
current debate about human energy
requirements, and ecological studies of farmers' energy balance. Malnutrition does not arise from
isolated single causes, but from dysfunction in
the 'food system'. Examples are given of the
interaction of health, work and family structure,
in the system
2 Defining malnutrition 37
Even the identification of 'malnutrition' as a
state, is not simple. Physical definitions of
malnutrition can be set lower or higher according
to their intended purpose. These purposes may
range from 'selecting the most malnourished
individuals for treatment' to 'selecting groups to

6 Agricultural Development and Nutrition

receive certain benefits'. The policy implications
of varying classification methods are discussed.
3 Energy andprotein requirements 51
At this point the question'how much food do
people needT is discussed. Man's needs for
energy and nutrients have often been calculated,
with the use of various assumptions. These are
examined, with the question 'What is the nature
of successful and unsuccessful adaptation to
variable and low food intakesT
4 Food system indicators 73
Chapter 1 introduced the general idea of a'food
system', and Chapters 2 and 3 outlined current thinking on how utilisation of food, and human
outcomes in that system may be evaluated. Data available for describing other parts of the system
are now reviewed. Indicators of individuals' nutritional state were discussed in Chapter 2;
here they are put alongside other indicators, and
the use of standard values and cut-off points is

Part Two The Causes of Malnutrition 95

5 Multiple causes in malnutrition 97
Little needs to be added to this chapter's title, by way of summary. Having outlined the 'food
system', it is logical to consider some of the
interacting dysfunctions which lead to
malnutrition. Illness and poverty are important
6 Functional classes and targetedpolicies 120
It is one thing to outline the causes of
malnutrition, and another to direct specific
policies towards the malnourished. This is called
targeting. One way to link explanation and

Contents 7

action is to use an analytical framework which is itself action-oriented. 'Functional classification' is such a framework. It provides descriptions of
population groups which made administrative sense, and which enable agriculture, nutrition
and health policies to be better targeted on
situations which give rise to malnutrition.
Examples of the implication for some
interventions are discussed.

Part Three Food and Nutrition Policy and Agriculture 143

7 Agrarian change andpoverty 145
'Nutrition in agriculture' is not just about
increasing food production. In this chapter we
discuss the contradiction between increasing
production and declining consumption and the
concept of people's entitlement to food. We
review the effect of agricultural development
strategy on the social status, labour
opportunities, and hence the nutrition of
different agrarian classes.
8 Markets andfood availability 163
Markets are an important component of the
'food system', governing some people's access to
food and the returns to others' labour. Market
systems may behave in many different ways but
they normally work to the advantage of those
classes who are least likely to become
malnourished. Policies on food price supply and processing technology tend to favour the urban
rich rather than the rural poor. We give two
important cases where nutritional justifications
for these policies are argued after the event
rather than before. These J ustifications'are
scientifically invalid and socially and
culturally irrelevant.

8 Agricultural Development and Nutrition

9 Nutrition interventions 184
Chapters 7 and 8 have shown that in practice, government policies in agriculture and health
have considerable impact on nutrition - impact
which is often not recognised. On the other
hand, several kinds of government-sponsored
action are regarded characteristically as
nutrition interventions'. Yet under some
circumstances they may have little impact on
malnutrition, as is indicated in this chapter.
10 Professional roles and research 200
What have agricultural and nutritional workers
to contribute to one another' understanding of,
malnutrition, and to programmes which may
reduce its prevalence? They can exchange insights derived from their characteristic,
approach to problem analysis; this can help to
avoid narrow oversimplified'solutions'.
Research can be interactive - and not only among
different classes of researcher but among
researchers and 'subjects', It is important, if we
aim to study and manipulate the 'food system', to
do so as advocates for those who are
disadvantaged within it.

Bibliography 220


During the decade which followed the World Food Conference held in 1974 in Rome, much debate, research and evaluation has focused on the impact of development strategies on the prevalence of malnutrition. Within the UN system the Food and Agriculture Organization (FAO) and the Children's Fund (UNICEF) have been actively involved in promoting new approaches in this field. FAO has been at the forefront of recent efforts to improve the nutritional impact of agriculture and rural development. UNICEF's well-known work guides and supports many types of development that relate to child welfare and services, including the problems of access to food and feeding of children.
There is a unanimity of opinion in all circles that the agriculture sector lies at the centre of any solution to the world's food problems. Yet the sobering thought that has become all too evident in recent years is that food production alone is not the answer. Rapid progress in production and exports of food commodities are found to co-exist with malnutrition even in the same areas of certain countries. Therefore while increased food production is an essential precondition, improved distribution is also necessary to overcome inadequate food consumption and malnutrition. The work that remains to be done is the application of suitable approaches through which agriculture and rural development can prevent the waste of human resources and potential which is caused by malnutrition.


10 Agricultural Development and Nutrition

This text, prepared by the Nutrition Policy Unit of the London School of Hygiene and Tropical Medicine, is an exposed of the many food consumption-related problems which need to be considered alongside agricultural production issues in development. It examines those social, environmental, economic and political factors that determine the degree to which people have access to food and can assimilate its nutrients. After reviewing the present scientific knowledge of energy, nutrient requirements and human growth, the authors concentrate on the many obstacles rural families face in trying to satisfy their basic food needs.
They argue for a balanced agricultural and rural development which would alleviate not only the technical constraints to increased production but also those many social, economic and political impediments that prevent people from having access to the foods that are produced.
This compilation of materials originally prepared for a UNICEF/FAO/Indian Council of Agricultural Research Workshop, is therefore a most important contribution to the thinking that has evolved recently about an interdisciplinary approach to food and nutrition problems. It will stimulate agriculturalists, rural development experts, food and nutrition planners and nutritionists all over the world in devising development strategies that are as concerned about impact as they are about Outputs.
FAO and UNICEF are pleased to have supported the preparation of this manuscript and, while not necessarily endorsing its every argument, are gratified to be associated with a book which provides a modern, and responsible vision of the basic themes and questions which they consider must guide development. The text is intended to stimulate interdisciplinary thought, discussion, research and action at the village, district, national and international level. By questioning the weaknesses of present approaches it cannot help but challenge all of us to search for more effective answers to food and nutrition problems in future development.


Starting in 1971, the Government of India, with the assistance of FAO, UNICEF and UNDP, embarked on a long-term programme to extend and reinforce the subject of human nutrition in higher education. The programme envisaged the teaching of food and nutrition subjects in agricultural, veterinary and home science colleges, and these institutions have been assisted in building up food and nutrition departments to undertake training and research in subjects related to nutrition.
Since 1980, under the title 'Education in Food and Nutrition in Agricultural Universities' (EFNAG) the programme has developed both in direction and content.
The general objective remains the same: to improve the nutrition of rural families and the means of achieving this is through three main aims:
1. to introduce a wider understanding of nutritional issues in the agricultural sector in order to incorporate nutritional considerations in agricultural programmes and hence bring about better nutrition;
2. to promote an understanding of the factors contributing to malnutrition, especially those related to poverty, with special reference to infant malnutrition;
3. to develop a more comprehensive view of training in agriculture and allied subjects that includes elements of food and nutrition education and basic services for children: this broader view should encompass the human aspects of agriculture, such as

12 Agricultural Development and Nutrition

development communication, extension methods, programme planning and food planning.
Of particular concern has been the need to strengthen the capacities of agricultural universities to include a nutrition dimension into the training of post-graduate students, and to lay the foundations of an expanding programme of applied and operational research in areas which link the production of food with the nutrition of rural people.
As an expression of that concern, the Indian Council of Agricultural Research, with the sponsorship of FAO and UNICEF, organised a workshop at Haryana Agricultural Universify in April 1982. The Nutrition Policy Unit of the London $chool of Hygiene and Tropical Medicine was invited to propose a programme of topic areas, to provide background working papers for the review sessions; and to conduct and co-ordinate the proceedings generally.
The objective of the workshop was to review in depth those aspects of the science of food and nutrition which are relevant to agriculture and to rural development, and similarly those aspects of agricultural change which have a direct or indirect impact on the nutritional condition of human populations. With this review as a background, the workshop was then to determine priorities for the inclusion of nutrition topics into programmes of postgraduate training in agriculture and home science, and hence to provide guidelines for future curriculum and research programme development. This book has been based for the main part on material presented to the workshop, and on a record of the discussions which ensued.
It is conventional to preface books of this kind with broad statements about the critical importance of improving the nutritional conditions of human populations, and of accepting that such improvement should be an explicit objective of development. It I's logical to assume that the pace and direction of agricultural change will be a critical factor in achieving this objective: more people consuming a better diet must have implications for food production, and more effective deployment of the means of production should be reflected in more people able to afford to eat adequately.

Preface 13

However, having indicated this fundamental relationship, it has usually been assumed that the objective of improving the nutrition of human populations at the same time as improving the practice of agriculture will be achieved automatically through bringing about an interchange between the two academic disciplines of nutrition and of agriculture. But what exactly should be the nature of that exchange? Should it be factual knowledge? Will malnutrition be reduced if agricultural students in future know that there are 10 amino acids essential for health, or that beans contain more protein than rice, or that vitamin A deficiency can be prevented by eating green vegetables? Or should the exchanges be of concepts, of the nature of the processes, social and economic, as well as physiological, which result in malnourished individuals?
We believe that it is particularly important at this time to take a critical look at how the two disciplines, agriculture and nutrition, should be expected to interact. We hope that this will be especially fruitful in the context of post-graduate activities, since this is traditionally the area in which subjects develop and new concepts are formed. Not the least reason for the timeliness is that despite the widely acknowledged importance of adequate nutrition as a basic component of human welfare, and despite the manifest success of agriculture in supporting the growing populations of the world, malnutrition persists. We have both more food and more destitute and hungry people. How can the connection be made between our growing technical capacity to generate wealth from the soil on the one hand and out understanding of the biological and social needs of man for food on the other? One thing is certain, there are no easy answers, and the reader of this book will search in vain for prescriptions, progress will come only through the realisation that the real problems and solutions lie in the nature of social and political institutions and the human relationships that underlie them., For some people this realisation is painful, and leads either to a rejection of science as irrelevant, or to a retreat into the more comfortable distractions of academic research. We shall be happy if this book helps to strengthen the view that curiosity about the way the world works, and the urge to extend

14 Agricultural Development and Nutrition
knowledge about what things are possible in the world through analysis and criticism can be applied to the solution of the problems that give rise to hunger.
Peter Cutler
Elizabeth Dowler
Barbara Harriss
Philip Payne
Erica Wheeler

Those whose names appear above are the members of the Nutrition Policy Unit of the Department of Human Nutrition at the London School of Hygiene and Tropical Medicine. All of them contributed material to the book. It could not have been written, though, without the help of many other people, in particular N.S. Jodha, Claire Kelly, David Nabarro, Adam Pain, Paul Richards, John Rivers, Young Ok Seo and Anne Thomson, who either work with us, or generously allowed us to use their work. We are also grateful to Madeleine Green and Barbara Kenmir who typed the manuscript.
Above all the book is the result of the integrating perspectives and unstinting labour of Arnold Pacey.
Our grateful thanks go to Margaret Khalakdvia and her staff at Haryana Agricultural University, and to Franciso Coloane of UNICEF for their outstanding hospitality, constant encouragement and careful arrangements which ensured the smooth running of the workshop.
The costs of editing and preparation of the manuscript were borne by FAO and UNICEF. However, neither of these organisations are responsible for any of the views or opinions expressed herein.
Philip Payne


Limits to Measurement

'The people are crying out for bread and we are going to give them statistics.' John Boyd-Orr in 1945 on proposed terms of reference for FA 0 (see Orr 1966, pp. 20, 162)

Changed perspectives
During the last three decades, the application of nutritioilal science to the problems of hunger and malnutrition has passed through a phase of great confidence and hope, followed by one of increasing uncertainty and doubt. Twenty years ago, a book such as this would have discussed well-defined nutritional interventions, aimed at achieving some equally well-defined nutritional objectives. Very little space would have been devoted to questioning the validity of those objectives, or the effectiveness of the programmes themselves. In so far as progress towards reducing malnutrition was acknowledged to be slight, this would have been interpreted as showing the need for more extended programmes, hence for more resources, and especially for more trained people to be deployed.
The connection between nutrition and agriculture would have been presumed to rest on a number of premises: that hunger would be eliminatedif there were an increase in the overall production of food; that malnutrition is often caused by deficiencies of specific nutrients (e.g. protein) and that this can best be countered by emphasis on the production of certain kinds of foods, or by fortification or enrichment of staples; that poor

*This chapter is based mainly on material prepared by Philip Payne: ,,cc Rivel s and Payne (1982) and Payne (1982).

1 Food systems and needs*

18 Limits to Measurement
nutrition is often the result simply of ignorance; and hence there is a general need to educate people on how to make proper use of the resources available to them. Infectious disease, though recognised to interact with malnutrition, would be regarded as essentially a separate problem to be dealt with by specifically designed programmes.
There has been a fairly radical change of viewpoint over the last two decades. This has been broadly for three reasons. Firstly, there has been a change in concept: malnutrition, previously regarded as something caused by single physical factors, is now accepted as having multiple causes, many of which are closely linked to the conditions of inequality of resources, or poverty, and of social discrimination. Changes in the system of food production which leave these conditions unchanged will also leave malnutrition unchanged or may even aggravate it. Indeed, we have witnessed some countries becoming net exporters of. food, while sections of their populations, remain inadequately fed, or even experience famine. In addition to this, among young children especially, malnutrition is so intimately related to infectious disease that it makes no practical sense to pursue programmes aimed at improving food consumption without also tackling at least some of the environmental causes of such disease. This may lead the nutritionist to encourage improvements in water supply, sanitation, housing and domestic fuel.
The second reason for a changed perspective is that there has been progress in our understanding of the physiological and biochemical processes underlying malnutrition. We now know that man has the capacity to adapt to a fairly wide range of dietary situations, and that only when that adaptive capacity is stretched beyond its limits does the body fail to maintain its functional capacity,* and malnutrition ensues.
Thirdly, there is a growing awareness that many of the more

*Functional capacity here is taken to mean all aspects of the behaviour of an individual in response to the environment, such as physical and mental activity, response to stress, disease etc. This should not be confused with the term 'functional classification' which is used in later chapters.

Food systems and needs 1.9
conventional types of nutrition programme have not achieved what was hoped of them. Interventions such as the promotion of high-protein crops, food delivery systems aimed at young children, and projects for educating people about the nutritional value of foods, have often been totally ineffective, or effective only as short-term palliatives. Frequently they have also distracted attention from the need to attack more fundamental problems.
It is in thiscontext of doubts and questioning about past ideas that we need to review the relationship between knowledge of nutrition and planning for agriculture. We believe that from a basis of analysis and criticism, we shall be able to begin a synthesis. But the ideas that will carry us forward will be different from those of the past. They will include ideas about food systems, about the epidemiology of malnutrition, and about 'livelihoods'. They will also include suggestions about the use of indicators, and ways of understanding the multiple causes of illness and deprivation. These and other concepts give us a new point of departure, so that we should not expect simply to substitute a new set of nutrition programmes, more potent and more efficient than the old - a set of 'right' answers to substitute for the 'wrong' ones. There are no simple answers, and whatever it is that a nutritional approach to human problems can provide, it is not likely to give us any short-cut solutions or technical fixes. The problems of malnutrition will be overcome as and when we overcome those of poverty, deprivation and disease. But we are all concerned with these, and therefore we could well begin by asking: what is so special about the nutritional viewpoint? Why should we be especially sensitive about nutritional needs rather than economic or social needs?
One clear lesson from the past is that when a particular problem at first sight requires nutritional expertise for its solution, this analysis is not always borne out by closer critical analysis. The skill and integrity of a profession rest at least as much on its willingness to show when it cannot and should not play a key role, as on its ability to demonstrate the effectiveness of its methods when they are relevant. However, there is a general case for concerning ourselves with nutrition and with

20 Limits to M asurement
food as a fundamental aspect of agricultural development, and this rests on two propositions.
The first identifies the nutritional status of a person as both the outcome of the process of acquiring, consuming and utilising food, and one of the critical inputs to that process: the, food a woman or a man cats decides the amount of effort she or he can afford to invest in order to secure food in the future. If we can measure nutritional status, therefore, we have a unique index of the impact upon individuals of the whole system of production, utilisation and exchange.
The other proposition is related to this: it is that, of all the symbols and objects of social exchange, food is arguably the most basic. Co-operation in the acquisition of food, and its sharing among the members of a family marked the beginning of the social evolution of mankind. The extent of an individual's integration with society can be measured by the adequacy and security of his or her ability to produce, control, purchase, borrow or otherwise acquire food (i.e. his 'food entitlement' (Sen, 1981)).*
This book is focused on a range of topics which impinge upon different stages and processes operating within the food 'syste rn' that is to say, the system comprising the production, distribution, consumption and biological utilisation of food. Figure 1. 1 shows, in a much simplified way, a few of the key relationships and processes which will be discussed. It is not intended asa complete definition of the 'system', but more a starting framework within which we may wish to elaborate certain areas. Then, by understanding how people become malnourished in terms of processes within the system, we may be better able to make statements about who such people are, and to describe their relationship with production and the basis of their entitlement to food. Malnutrition, in this context, is a symptom or signal that certain processes are regularly occurring in the lives of people which, if disregarded, will result in the continued generation of sickness and physiological impairment.

*Full references quoted in the text are contained in the Bibliography beginning on p.220.

Food systems and needs 21

Figurel.1 The food system as it is envisaged in this book, in its simplest form

Analysis of the nutritional problems within a society may therefore have implications for agriculture at many different levels. At one level, it may consider the impact of agricultural change on the ability of people to earn an entitlement to food. To produce more, but at a cost many people cannot afford, may be self-defeating. An orientation of technology to production, which neglects problems of consumption, needs to be corrected. At other levels, the analysis points to the need for an integrated approach by planners to the development of land resources for food, for fuel and for cash crops. It underlines the need to improve the domestic environment as well as to plan resource inputs to agriculture. It may also indicate the need to avoid certain directions of change because of the danger of aggravating

22 Limits to Measurement
the risk of malnutrition, or may at least demonstrate the cost of that effect in assessing policy choices.
These, then, are the themes of this book. Our starting point is actually near the bottom of the diagram, which is where we find the more usual focus of food and nutritional science. The traditional emphasis continues in much teaching and research, that is to say on the properties and uses of foods rather than on the nutritional problems of populations. Part of the purpose of this book is to present a reverse perspective - to put the problems of people first and see what questions they prompt about nutrition. To reverse the traditional perspective of nutrition in this way is not, of course, to say that everything done previously was mistaken. What we would argue, though, is that if one constantly approaches a subject from the same point of view, important insights may be missed, and received views may not be adequately questioned. For example, to tackle the question of how much food human beings require we first ask what kind of life those individuals lead, and what physiological functions, therefore, does their food intake have to support?
If the science of nutrition has any central concept, it is surely this notion of nutrient requirements. The initial impetus for the subject came from nineteenth-century attempts to define those requirements, and in many quarters, that approach is still important. The process of revising estimates of energy and protein requirements continues unabated, and the problem of fixing them remains apparently unresolved, What has changed since the early days is that originally pronouncements about nutrient requirements were made by individual scientists and were based on their own research; now, similar pronouncements emanate from committees of experts convened by governments or international agencies and to a large extent reflect the experts' selection of scientific evidence and of theory (Rivers and Payne, 1982).
Despite the authority and influence associated with some of these figures, the concept of a 'nutrient requirement' remains somewhat intangible. Discrepancies between estimates, and the revisions that are made from time to time, do.not seem to reflect differences in technique or theory, but may have more to do with

Food systems and needs 23
political pressures or changing social valuations of the acceptability of particular intake levels. For example, successive estimates of food energy requirements made by the US National Academy of Sciences for a moderately active man with a body weight of 70kg have been as follows (NAS 1943, 1958, etc.):
12.5 MJ (3000 kcal) in 1943; 13.4 MJ (3200 kcal) in 1958; 11.7 MJ (2800 kcal) in 1968; 11.3 MJ (2700 kcal) in 1974.
The 16 per cent fall in estimated requirements since 1958 is not the result of improvements in the process of estimation, nor of any factor such as change in levels of activity, but must be attributed more vaguely to climates of opinion and perhaps concern about obesity.
Nonetheless, the questions 'how much food does a man/ woman/child needT and 'what are the reasons for and consequences of failing to meet that needT remain central to agricultural development. We set out below some of the issues involved in answering them and the problems raised thereby. We use a series of short case studies (from different parts of the world) to illustrate them.

Food energy and agrarian ecology Lack of progress in attempts to define human energy and, nutrient needs illustrates one reason why it is useful to think in terms of the level of nutrition needed to maintain the functional capacity of the body. If we take the traditional view of a nutrient requirement as the minimal amount of nutrient needed to maintain a given physiological state such as 'health', then we are faced with the problem of specifying that state. And although health and well-being are certainly to be valued as social goals, the utility of health as a reference criterion is limited by our inability to define a state of ideal health which an adequate nutrient intake should sustain.
The social nature of the definition of health has been

24 Limits to Measurement

frequently discussed elsewhere (Illich,, 1976; McKeown, 1979), its use as a criterion for fixing energy requirement is particularly complicated by the fact that it seems impossible to identify a healthy population. As one official body states, most of their recommendations on food energy are based on measurements of what actual Populations eat, assuming that these populations are healthy. However, they point out that: 'Many groups of people .living in the industrialized countries are obese. On the other hand some groups living in developing countries are small in stature, light and thin, yet may not be physically less healthy because of their different body size' (FAO/WHO, 1973).
Another approach is to attempt the definition of energy requirement in terms of the maintenance of a specified physiological state, disregarding notions such as 'health' and this leads to the apparently more precise concept of 'nutrient balance', a deceptively simple notion which has been widely applied in animal nutrition (Blaxter, 1967).
Obviously, in the non-pregnant adult animal, nutrient intake and expenditure must match if the body content of the nutrient is not to rise or be depleted. The problem is that balance can be achieved over a range of intakes through adaptations of various kinds. Thus the question 'which level of nutrient balance is preferred?'is inescapable, and can only be answered by referring back to 'health'.
The balance method is thus no more objective than the health criterion. The decision about what level of equilibrium or for children, what rate of growth, and hence what degree of positive balance, will be regarded as a norm remains entirely subjective unless we are prepared to specify a particular set of desired functions and a set of undesirable symptoms we wish to avoid. The decision which is usually made (albeit not always explicitly) is to say that wve prefer the levels and growth rates which are typical of western developed countries, though we are discovering that these are not without disadvantage for health.
It is in order to avoid such ambiguities and difficulties that this book uses 'functional' criteria of the adequacy of food intakes. From this point of view, malnutrition is defined as a state in which the physical function of an individual is impaired to the

Food systems and needs 25
Tablel.1 Energy intakes of New Guinea adults

Village Body weight Energy intake per day

kg MJ kcal
Kaul (coastal location) males 56 8.12 1940
females 47 5.94 1420
Lafa (highland location) males 57 10.54 2520
females 51 8.79 2100
Source: Norgan et al. 1974.

point where she or he can no longer maintain adequate performance in such processes as growth, pregnancy, lactation, physical work, or resisting and recovering from disease. The notion of an adequate level of performance is itself , not a simple one. However, it avoids the rather greater difficulties of talking about adequate levels of health if, in the first instance, we use it to mean the achievement of a sustainable mode of existence. Thus, for example, to avoid malnutrition, members of a farming family must be able to do physical work on their land and crops, sufficient not only to secure their immediate food needs, but to sustain productivity over long periods, and to survive through bad years as well as good. They must in addition be able to adjust total production from their resources to keep pace with the demographic changes (numbers of dependants and number of productive adults) in the family itself.
In taking this approach, we regard dietary energy as the most likely limiting factor. This does not imply that vitamin and other deficiencies do not occur, or are of minor consequence. Rather we are suggesting that the avoidance of energy constraints is a necessary even if not always a sufficient condition for avoiding malnutrition.
In the context of agriculture, the problem of energy deficiency is quite a subtle one, partly because of the wide range of different farming ecologies that can be successfully exploited, and partly because we have to look at equilibrium situations where the working capacity of adult family members is both the outcome of

26 Limits to Measurement

a particular level of food consumption, and the labour input to the production system which generates the food. Viable equilibrium can therefore be maintained over a range of levels of intake and output. As an example, Table 1.1 shows the average food energy consumption levels of men and women in two villages in New Guinea.
The people of both villages had similar and generally good states of health, without any signs of malnutrition. However, members of the coastal community needed to devote only small amounts of labour to their gardens, and despite a plentiful source of fish, rarely bothered to go and catch them. Most of their day was spent in leisure and socialising. The highland community, by contrast, lived and worked on steep mountain slopes, and spent several hours a day in fairly heavy labour. The two situations are both successful from the point of view of health and survival. Food energy needs are different, however, and if malnutrition did occur at some future time because of limited land access, or because of some climatic event, it would do so at di , different levels of food consumption.
Consider now rice -cultivation. It is estimated that dryland rice requires approximately 130 man-days of labour input per hectare in one environment (Bayliss Smith 1981), corresponding to a work input of about 390 MJ'per hectare. It will yield on average about 15,000 MJ net edible yield of rice - about 1000 kg - so the ratio of energy output to input is 38:1. This is not an unusual figure, since according to Leach (1975) energy ratios for preindustrial crops are generally in the range from 13 to 38. When food energy production in a pre-industrial society is at or below the lower end of that range, it is likely that not enough food is being produced to support non-agricultural work such as watercarrying or house-building, and to, support non-working members (old people and children), so malnutrition in the sense of functional failure becomes increasingly probable.

Changes in energy needs over time At this level of analysis, the picture presented is of a simple,

Food systems and needs 27
static equilibrium, with energy flows averaged over time, and no account taken of the effects of either regular seasonal factors, or irregular unpredictable hazards. Seasonal factors may impose two kinds of restriction, in the first place related to time intensity. Certain tasks must be carried out in specific seasonal time periods and this may impose a lower limit on the number of working adults or their equivalents per hectare necessary to secure a given type of crop. Secondly, work intensity may impose limitations bearing in mind that the physical work required for certain tasks may be much greater than average and may only be sustainable by individuals with a high level of physical fitness, i.e. good nutritional status, and freedom from disease or injury.
Further, agricultural seasons are not always the same; there are good years and bad years. Survival in bad years means either the ability to achieve a surplus during good years, with some means to carry this over as stored food, as cash, or as assets which may be collateral on loans, or some other opportunities for converting labour resources to food or cash by alternative employment.
An additional, longer-term result of the passage of time is that the size and demographic structure of the family will change: total food needs increase steadily as successive children are born, but later the children grow up and contribute to the family's labour power. Together, then, the influence of seasonal cycles and family development is such that we need a dynamic view of the whole food system, of the functions which take place within it, and of the environmental and temporal changes which form its context.
A village in Burkina Faso, West Africa, where millet is the principal crop, illustrates one kind of sharply seasonal contrast. In this area, the dry season is a long one; it is followed by a relatively short period when the soils are sufficiently softened by the rain to be workable, and when crops must be sown. During this period, therefore, long hours of labour are necessary. The time intensity and work intensity of the effort people must make are both high. Human energy expenditures are, as a result, comparable to those found only in coal mining in more industrialised economies. By contrast, during the dry season there is little work to be done and

Table 1.2 Energy expenditures offarmers in an Upper Volta village Energy expended per day
mi kcal

Dry season women 9.7 2320
men 10.1 2410

Wet season women 12.1 2890
men 14.4 3460
Sources: Bleiberg et al. (1980); Brun et al. 1981.

energy expenditures are low. The men in particular have a very sedentary lifestyle and their energy expenditure drops sharply (Table 1.2).
This example shows that when we look at energy expenditure and food supply on a single occasion, or indeed, averaged out over a year, we can tell very little about the adequacy of diet, or the likelihood of malnutrition. For these West African villagers, malnutrition might consist of inability to meet peak labour demands because of either insufficientfood available to balance expenditure at that time, or because of body stores insufficiently replenished since the previous working season.
In all probability, people using this kind of production system face problems of both time and work intensity, and their responses are partly behavioural. and social, partly physiological: the periods of peak work output, which may be quite short in time, are sustained partly at the expense of body energy stores. Thus people lose weight, and then regain it during a subsequent period when work output is dramatically reduced, but food intake is maintained at the previous level, or even increased. Figure 1.2 shows these effects as they have been observed in Gambia. This country is also in West Africa, and the seasonal ecology is similar (though less extreme) to that of Burkina Faso (Fox 1953). Periods of peak energy expenditure are associated with harvesting, and with soil preparation and planting.
Obviously, there are physiological limits to the extent of this cyclic process: if too much weight is lost, perhaps because of a

28 Limits to Measurement

Food systems and needs 29

65 kgpounds

60 kg130 /

/ I

55 kg120 J


50kg- 110 /

4 5 k g . 1 0 0 -T. , . . ' t i
July Oct. Mar. June Sept. Mar. June Sept. Dec. Mar.

1947 1948 1949 1950

Figure 1.2 Fluctuations in 9dult body weight by season in a Gambian village
Source: Fox 1953

30 Limits to Measurement

Table 1.3 Analysis of labour time for different operations of cultivation of HYV and TV paddy in North Arcot District, Tamil Nadu, India

Agricultural Percentage of Additional person- Percentage of
operation person-days spent days for HYV over female labour
in the operation TV per hectare used


Ploughing 18 18 5.9 0 0
Manuring 2 3 1.2 22 26
application 0.5 0.4 0.2 0 3
Pulling seedlings 4 4 2.2 17 21
Transplanting 15 15 6.2 100 98
Weeding 12 12 3.7 100 97
application 0.2 0.1 0.5 0 2
Harvesting and
threshing' 32 28 24.8 57 60
Irrigation2 13 15 1.5 0 1
Others 3 4 3.0 29 55

TOTAL 100 100 41.2 47 47

Labour energy
input per ha 830 710 120
Net edible rice per
ha as food 33,000 24,000
energy output (GJ)
Energy ratio 40:1 34:1

1 The labour required for harvesting and threshing could not be separated in many instances.
2 The work required for irrigation involves switching on pump sets, waiting for electricity, or operating the kavalai.
3 Labour input is estimated roughly as 3MJ per person-day. Source: Based on Table 14.2 in Chinnappa and Silva 1977.

Food systems and needs 31
poor previous season, then peak work output may be impaired, reducing the potential for next year's crop still further. Despite recent research into the nutrition and physiological work, most experimental measurements have sought to measure the steadystate behaviour of individuals and animals. Thus relatively little is known about responses to fluctuations in intake or work output, or about the body's energy and nutrient storage mechanisms. Yet the latter may have a very much larger role than previously realised for key vitamins and minerals as well as for energy (Longhurst and Payne, 1979). Lacking knowledge of these, we cannot assess the risks of malnutrition for people living in highly seasonal environments.
Data on seasonal energy relationships in the context of rice cropping is available from the North Arcot district of Tamil Nadu state (India). Table 1.3 shows a breakdown of the time spent on various stages in the agricultural cycle, both for traditional (TV) and high yielding (HYV) varieties of rice (Chinnappa and Silva, 1977). The change to HYVs in this region increased yields to the point where a family of five people could theoretically meet its basic energy needs by cultivating only 0.5 ha. This estimate is based on a food energy output of 33,000 MJ per hectare annually. The ratio of energy output to input was then 40:1, which is slightly higher than before HYV rice was introduced.
It is clear, then, that a family which adopted HYVs did not require any additional labour to meet its own food requirements. But since yields were greater, these requirements could be me" from less land. The remainder of the land formerly used e d then produce a surplus for sale, but only if more labour was employed. In practice, most families adopting HYVs were those with the larger holdings, and they secured the extra labour by hiring workers from among the smaller cultivators and landless families. That particularly affected the employment of women for tasks occupying short, intensive seasons such as transplanting, weeding and harvesting (Table 1.3), which is significant in relation to a seasonal peak occurring locally in the birth rate (Chambers et al., 1981). At the time when they need to be working in the fields, many mothers have very young infants whom they should be breast-feeding. This can mean that feeds

32 Limits to Measurement

are hurried, or badly spaced, and babies do not get enough (Devkota, 1981; Rickleton, 1981).
Not only do work patterns and birth rates commonly vary markedly with season, but so do infectious diseases. When seasonal peaks of infection coincide with times of food shortage and hard work, the consequences of malnutrition may be compounded. The greatest regular seasonal impact on rural society is often due to infections and parasites (Bradley, 1981), and these have strong interactions with nutrition.
Clearly, disease can be both a cause and a result of poverty. It affects the efficiency of work, and that is one of several reasons why the previous discussion of the minimum energy inputs needed to feed a family from a rice crop is a considerable simplification. In practice, the family's survival would depend on exceeding these levels, so that there is a margin to allow for times when individuals are not working because of illness, as well as to allow for grain lost in store (5-10 per cent) or retained for seed. Furthermore, although the rice could supply all the energy and protein needs of the family, it would not fully supply vitamin and mineral needs. The family would thus have to invest more labour either in farm production of vegetables and other foods, or in paid employment, or in an increased production of rice for sale and exchange.
Very often, the levels of labour input referred to earlier would need to be increased by up to 50 per cent to cover the needs for exchange purposes, and to provide an insurance against poor harvests and crop levels. This insurance investment would depend upon the level of ecological variability of the region. In an extremely variable climate such as that of Burkina Faso, the traditional practice before 1920 was to aim for a level of stocks after harvest equivalent to 2-3 years' grain consumption (George, 1980). However, the ability to make any insurance clearly depends on the labour and land resources available. In South Asia the marginal farm family with land restricted to less than their subsistence production needs will be in a precarious situation with regard to bad years and the landless even more so.

Food systems and needs 33

Family development
Seasonal changes in food supply must be studied in relation to the longer-term cycle of the family's demographic expansion, as children are born and grow up. During the first year or two of marriage, both parents are usually able to contribute in labour or other economic activity, particularly if the timing of pregnancy is such that the women can still work during the time-intensive part of the production cycle. Earning or production capacity will then be high in relation to the number of family members to be supported, and the economic dependency ratio is low.
Things may become more difficult as the number of children increases and if and when they are at school. Later, as the children take on more and more work, the economic burden carried by each working members of the family eases. Food energy needs in a family and the number of workers necessary to support that family are thus constantly changing over time.
Young children require relatively large amounts of dietary energy. A one-year-old child is one-fifth the weight of an adult, but his/her energy intake will be about half the adult level. Somehow the young child has to take in a relatively large amount of food. Much depends on the frequency with which a child is fed and on the amounts of milk she/he is given. Rutishauser (1974), working in Uganda, found that the factors associated with a good energy intake in children under two years old were:
1. breast-feeding continued while other foods were introduced
2. three or more non-breast feeds per day
3. 'good appetite', i.e. absence of infection. It is of critical importance that the mother or whoever cares for a child should have sufficient time in the working day to help that child to feed, and in some cases, to prepare special items for him/her, because even though the family diet may have a good nutrient density, certain nutrient-rich items may be too spicy or hard to handle. The high growth rates and high relative energy needs of young children require frequent concentrated feeds, yet the lifestyle of many poor families makes this extremely difficult to achieve, especially at seasons when work is highly timeintensive (Wheeler, 1982).

We have already noted that adult agricultural workers sometimes lose weight during peak labour seasons and then regain it at times of year when there is more food available and less work This fluctuation may be regarded as a satisfactory adaptation to a changing environment. However, if pregnant and lactating women suffer energy deficits during periods of hard physical ' I work, their offspring will be smaller and more vulnerable (Paul et al., 1979). Evidence that Asian women's energy status does fluctuate seasonally is provided by the observations of Chowdhury et al. (1981) in Bangladesh. There, the women of landless families tend to lose weight during the August-October period" when the highest labour demands for rice cultivation are just over, and the price of rice is highest. Should these women be pregnant or lactating at that season, the risk of malnutrition to both mother and child is increased.
Illness may temporarily alter nutrient requirements. Gastrointestinal diseases often lead to malabsorption of food, and many illnesses lead to tissue breakdown or blood loss to a varying degree. During the acute phase of an illness, nutrient utilisation

Table 1.4 Food energy needs and labour available at three stages in the development of a family

Food energy Adult male Ratio of energy needs labour need to labour: MJ1year equivalent available

Woman and man,
1 child under 2 years 9,500 1.8 5.3
Woman and man,
1 child under 2,
1 child between 2 & 3,
1 child between 4 & 5 14,500 1.8 8.0
Woman and man,
1 child under 5,
1 child between 8 & 10*
I child between 10 & 12*
1 child between 12 & 15 * * 21,500 3.6 6.0

Notes: assumed equivalent to 0.5 of an economically active man;
assumed equivalent to 0.8 of an economically active man.

34 Limits to Measurement

Food systems and needs 35
falls so there is need for some increment during recovery. Clearly, this is a more serious matter for children than for adults; they need nutrients and energy for growth, and also their body nutrient reserves are small. This may be seen particularly in children with acute fevers or gastro-enteritis, who can become wasted and also dehydrated very fast, unless there is a determined effort to keep their fluid and food intake as high as possible (Mata, 1977; Briscoe, 1979).
For a great variety of reasons, then, human nutritional needs vary markedly throughout life. There is a steady fall in necessary intake in proportion to body weight until adulthood, after which the appropriate level of intake depends greatly on work and activity, and on physiological stresses, of which the most important for women is pregnancy.
Human requirements for food are not constant, but vary with the development of family groups as much as for individuals. Table 1.4 presents typical figures for a particular family at three stages in its development, and shows how there may be a critical period when the children are all young. At this stage, energy needs in relation to the labour available to meet them peak sharply. During this period, there is a maximum likelihood of malnutrition occurring, particularly at the most intensive stages of the seasonal cycle. Thus the focus of many nutrition programmes on the problems of young children makes sense not only because of the vulnerability of the children themselves, but also because of the vulnerability of the family group as a whole while the children are young.
To understand the prospects of the family group in more detail, however, we need to look beyond the children and enquire whether the family owns and cultivates its own land, and whether it is able to hire farm workers at especially busy times of year, since that is an important way of dealing with peak energy demands. Families with little or no land of their own will often be in more critical situations. Most difficult of all are the seasonal problems of families with small plots of their own, but who are also partly dependent on paid work. For them, the best opportunities for farm employment are likely to coincide with the heaviest demands of their own crops.

36 Limits to Meastfrement
Any investigation of the problems faced by such people should also take account of agricultural developments which may modify the magnitude and timing of seasonal effects, crop varieties and irrigation.
All our examples so far have been of rural agricultural households, but the principles could be applied to other groups such as urban unskilled labouring families, plantation workers, pastoralists, fishing families, and so on. Investigation of social or economic differences, of the effects of technical change, or seasonal constraints, and the problems of family development, entails classifying families by land ownership and tenure, by dependence on markets, and by major sources of income. This would be an example of what is referred to later (Chapter 6) as a 'functional classification' of the population being studied. Ex mining the prevalence rates for malnutrition season by season for each of the groups identified, enables us to study malnutrition on a comparative basis. Because of the importance of this approach, we shall repeatedly return to such questions as land tenure and cropping patterns. First, though, it is necessary to consider certain other nutritional concepts, and these are the subject for the next two chapters.

2 Defining malnutrition*

Comparative views
Malnutrition, according to the previous chapter, should be understood in terms of failures of bodily functions. The conditions under which it occurs must be appreciated as specific to particular localities, with characteristic agricultural ecologies and work patterns. People adapt to a great variety of dietary and work regimes, and this sometimes involves modifications in body size, levels of physical activity, and metabolic changes. Detecting malnutrition is a matter of detecting degrees of such change which carry unacceptable penalties in terms of hunger, illness, dysfunction and risk of dysfunction. We shall return later to the implications of the word 'unacceptable', and merely note that these ideas may together be characterised as an 'individual adaptability view' of malnutrition.
It is this view upon which the present book is based, and to avoid misunderstanding, we need to compare it with other concepts, of which the most important is the 'fixed genetic potential view'. The basic premise here is that there is an optimal or preferred state of health, fixed for each individual, and determined by his or her genetic potential for growth, resistance to disease, longevity, and so on. It is assumed that everyone

*This chapter is based on material prepared by Elizabeth Dowler and Philip Payne. See Dowler et al., 1982.

38 Limits to Measurement

could and should achieve their full genetic potential and that malnutrition starts as soon as there is any departure from the, preferred state. We cannot measure human genetic potential, however, so it is further assumed that the standards of body size and food intake observed in 'well-fed' and 'healthy' populations approximate to this optimum.
It may be useful to compare these contrasting views with one which gained acceptance two decades ago. Jelliffe (1966, p.8) regarded malnutrition as, 'a pathological state resulting from a relative or absolute deficiency or excess of one or more essential nutrients, this state being clinically manifested or detected only by biochemical, anthropornetric or physiological tests'.
Two points made by this definition should be especially noted. Firstly, it implies that deficiency or excess of nutrients is the major cause of the pathological state referred to, whereas we would now wish to emphasise the role of many other contributory causes (Chapter 4). Secondly, Jelliffe states that malnutrition can be detected only by biochemical, anthropornetric or physiological tests. It cannot be deduced from an individual's level of food intake, nor can it be estimated from the average intake of a population.
In this second respect, the individual adaptability view and Jelliffe's definition are in close agreement, and stand in opposition to the assumption often made that malnutrition can be inferred whenever intakes of nutrients fall below certain fixed levels. Many statistics which purport to represent the number of malnourished people in the world (or in specific countries) actually represent the number with low food intakes. We cited one example in the previous chapter of two societies of apparently healthy populations whose energy intakes were markedly different. We also mentioned the importance of food in cultural and social exchange, and suggested that an individual's ability to obtain food may be a good indication of his integration with society. Thus the number of people whose food intakes fall below a certain level may be an important statistic for measuring poverty or social differentiation (Chapter 6); it may often include some malnourished people and some who are seasonally hungry. But whatever its social significance, such a figure does not measure the prevalence of malnutrition in its scientific sense.

Defining malnutrition 39
The manifestations of malnutrition vary. In what follows, unless otherwise stated, we are referring to protein-energy malnutrition.
In a population with relatively low food intakes, only a minority of children may show clinical symptoms of malnutrition such as oedema, fatty liver, hair and skin changes, or severe muscle wasting, but others may very well be affected especially with regard to their vulnerability to infection. Other clinical signs include diminished subcutaneous fat, and low body weight relative to height. Beaton and Bengoa (1976) and Alleyne et al. (1977) give useful descriptions of these signs and symptoms, and of the recommended methods for weighing children and taking measurements of height (or length of infants), and of arm circumference.
But the question then arises: by what standard should we decide whether body weights are low? If this is one of the populations cited in the previous chapter where it is 'normal' for people to lose weight markedly during certain seasons, are we in a position to make allowance for that? If the average height of the people is unusually small, do we interpret this as an abnormality and call it 'stunting', or might it be an 'adaptation' to environmental conditions? If the people are New Guinea tribesmen whose children are relatively free from illness but whose environment has unusual features, 'adaptation' would seem the appropriate description. But with Indian village children whose growth is checked by episodes of serious illness from which recovery is slowed by low food intake, 'undernutrition' would seem a more relevant description. Between these two extremes there are many uncertain categories. Moreover, we still do not know very much about how to distinguish populations going through'normal' seasonal cycles of weight loss and recovery from those whose body reserves had been diminished by a downward spiral of impoverishment. Thus questions about how we interpret small body size or low body weight are particularly important.
What we really need to know about are the relationships between body size and such functions as immune competence, mental function, physical work capacity, or the likelihood of survival. All these functions are interrelated, but the links

40 Limits to Measurement

between them are not well understood. As Chapter 4 will demonstrate, little is known about the relationship of anthropomnetric data to mortality risk and morbidity, but this suggests that we shall probably never be able to locate fixed points on, say, a weight-for-age scale which will sharply divide individuals at risk from those we regard as normal. To the extent that such cut-off points can be established at all, they will probably only be valid for specified population groups living in particular localities. So, for example, it seems impossible to classify mortality risks for a group of children in Zaire using anthropomnetric criteria which worked well for children in Bangladesh (Kasango Project Team 1983).

Screening and anthropometry
The practical importance of establishing and defining criteria for labelling people as 'malnourished' can be illustrated by discussing the problems of selecting a subgroup of the 'malnourished' for treatment or intervention.
Despite the difficulties of classification, in any practical situation where action is contemplated, there will be a need to have some procedure for comparing risks and for allocating resources. We might in some circumstances consider that risk of death is the most important criterion and confine ourselves to this. We shall find, however, that even in such a relatively simple situation, classifying individuals as 'at risk' will depend not only on the risk relationship' but also on the resources available and the logistics involved.
For example, Chen et al. (1980) have reported the experience of a sample of 2019 children in Bangladesh. Their weights, heights and arm circumference measurements were recorded when they were one year old. After two years, during which only treatment for diarrhoea was available, 112 of them had died.
We can employ the data about these children to show what the effect would be using two different variables, weight-for-age and weight-for-height, as a basis for allocating some more effective treatment. To do this, we imagine two kinds of situation, in both

Defining malnutrition 41

of which it has been decided to use anthropometry for screening the group - that is, as a means of selecting individuals who will be given food rations or provided with some kind of treatment aimed at rehabilitation. We shall assume that resources are limited, and that the treatment, if given, is equally effective in reducing the risk to a level close to the minimum, however severely the recipient is affected. The two situations are:

1. The population to be screened is of fixed and known size, such as an entire refugee camp; our problem is simply to decide who gets treated and who does not.
2. The 2019 children are just a sample of a very much larger group, and there is a continuous queue of applicants for screening.
Table 2.1 shows the numbers of children in various categories

Table 2.1 Numbers of children out of a population of 2019, who would have been identified and treated for malnutrition, using two different indicators'

Number Number of Number of Deaths preidentified as preventable false vented per in need of deaths2 positives treatment treatment treated'

<60% of reference 427 48 379 0.112
<75% of reference 1473 92 1381 0.062
<70% of reference 75 11 64 0.147
<80% of reference 641 39 602 0.061
<90% of reference 1620 92 1528 0.057
Treatment of the
total population 2019 112 1907 0.055

1Data taken from Tables 1 and 2 in Chen et al. 1980, which consists of a study of mortality rates in 2019 Bangladeshi children during a twolear period following a survey of nutritional status. This is the number of children who actually died in the group studied by Chen et al.
3 Number of children who survived.

42 Limits to Measurement

of weight-for-age and weight-for-height whose deaths in our theoretical case would be prevented if they were selected for treatment. It will be noticed that instead of quoting the actual weights of the children, we quote them as percentages of a reference weight or standard - in this case the 'Harvard' standard. This represents the average weight of a large sample of apparently healthy children at the same age as the individuals being measured. Recording data as 'percentage of reference' for each individual's age allows older and younger groups to be compared directly, or allows the data for all the children to be amalgamated. (We consider this practice in more detail in Chapter 4.)
In this example, those children who are below a stated percentage level are selected for treatment and the second column in the table shows how many would be involved. Thus if 60 per cent of reference weight-for-age were taken as the critical point and if all children under that were treated, this would entail a total of 427 treatments. Of these, 48 would be given to children otherwise likely to die, so there would be a ratio of one 'genuine' to 8.9 'false' positives being treated. * Obviously, if the critical point on the reference scale were raised from 60 towards 100 per cent, the numbers of treatments that would have to be given out would rise and eventually cover the whole population. The ratio of 'false' to 'genuine' subjects treated would also increase. Supposing the group of 2019 children comprised a complete community and the critical level depended on the number of treatments available. If there were resources for, say, 500 treatments, a point slightly above 60 per cent would give the best allocation, reaching over 50 children who would otherwise die; 1000 treatments would reach 80, and so on.
If weight-for-height were to be used, the situation would be somewhat different. At low critical points, such as 70 per cent reference weight-for-height, the selection is much more specific than by using weight-for-age. It excludes many false positives,

*The notion of a false positive here refers to mortality and not to mortality plus morbidity, where the concept might have to be modified.

Defining malnutrition 43
and gives lower ratios of 'false' to 'genuine' subjects treated; however this specificity is at the expense of also excluding many 'genuine' subjects as well - only 11 deaths would be prevented.
If we again consider what happens when 500 treatments are available, we find that they would have to be allocated on the basis of about 75 per cent of reference weight-for-height, but that only 30 of them would then go to children who would otherwise die. With 1000 treatments, we would need a cut-off point at rather less than 85 per cent, and 60 lives would be saved, and so on. This, of course, assumes that the kind of risk assessed by these two variables, weight-for-age and weight-for-height is the same: that high risk cases identified by both would respond equally well to the same treatment.
The other situation, where we envisage a continuous queue of children being brought for treatment, is more complex. If the rate at which individuals can be treated (the rate of admission for rehabilitation, or the rate at which food supplements can be delivered) is fixed, we might want the critical point for screening and selection to be that which identified individuals for treatment at a matching rate. On the other hand, we would want as low a ratio of treatments to preventable deaths as possible. The second requirement would be satisfied by using 70 per cent of reference weight-for-height as the cut-off point. But then the first requirement might only be realised if the measurement were carried out faster, for example by employing more people to do it, thereby selecting more children for treatment per unit of time.
We have elaborated this example to illustrate that however artificially we constrain the system, we cannot expect to find a unique answer to the question of how to classify the malnourished. It would of course be convenient if a particular percentage of reference weight-for-age could be taken as an unambiguous cut-off point or boundary between 'severe' and ,moderate' malnutrition, or between 'malnutrition' and 'normality'. Some systems for classifying anthropornetric data imply distinctions of this sort, but the example shows that wherever we draw a dividing line, some individuals who are not at risk will be selected for treatment while others who are at risk will not. Thus in practice, the precise percentage of reference weight we take as

44 Limits to Measurement

a dividing line or cut-off point is chosen according to our resources, circumstances and purpose.
It is clear how the resources available, and how the circumstances of a refugee camp or, alternatively, an endless queue of children, may alter procedures. As to purpose, we may compare this example of screening to select children for treatment, with several other purposes for which body weights and heights may be recorded.
For example, when the illness of a single child is being diagnosed, his weight might well be noted, but it is clear from the figures in Table 2.1 that weight-for-age would not signify much by itself. If the child has been attending a clinic regularly, and his weight has been plotted on a growth chart each time, any faltering of growth will be readily recognised. But the clinic staff will also examine the child, looking for evidence of infection as well as for classical signs of malnutrition such as oedema (though these signs are present only in severe cases). Enquiries might also be made about the child's home circumstances, to assess whether food has been in short supply, whether there are particular hygiene problems, or whether the mother is well enough, or has sufficient time, to prepare food for the child. In other words, to form a clear view of an individual's condition, one requires a combination of different kinds of evidence, and as always, diagnosis of an individual condition and needs is by a combination of clinical experience and measurement.
By contrast, where there are large numbers of people to cope with, screening procedures such as we have envisaged being used in a refugee camp must usually be employed. Then, individuals have to be selected for treatment on the basis of only one or two routinely applied tests. In these circumstances, we know that many of those selected will not need treatment, and some urgent cases will be missed. Thus the problem is to design the test so that it selects as well as possible within its limitations.
Yet another purpose which is sometimes served by recording children's body weights and heights is monitoring and surveillance. The aim here is to identify social groups at nutritional risk, or to check whether there is a trend for nutritional status in a population to improve or worsen over time. In surveys of this

Defining malnutrition 45
kind, we again look at the number of individuals whose weights or heights fall below a stated percentage of some reference level. However, we are not now worried by 'false positives' included in the numbers we count, provided that nutritional position of the two is clearly and fairly shown. Thus we may decide to record the proportion of children who are below, say, 85 per cent reference weight-for-age, comparing farmers' children with children in the families of labourers, plantation workers, artisans, and so on. The choice of the 85 per cent reference level would not necessarily reflect any nutritional benchmark, nor would it be determined by the resources available to support some kind of action. It would be chosen because in some social groups, many children are below this level, while in others, most children are above it. If we counted the number of children whose weight relative to age was below 110 per cent of reference, contrasts would disappear because nearly everybody would be in this category.
These examples of different ways in which anthropometric data could be interpreted and used reflect the assumption associated with the 'adaptability view' that an individual's nutritional condition cannot be directly measured. Data on weight, height, arm circumference and so on are therefore indirect or proxy descriptions whose relationship with 'risk of dysfunction' is a complex one. In particular, measurements above which risk is uniformly low, but below which the adaptive capacity of the body has been exceeded.
The 'genetic potential view', by contrast, stresses maximisation of potential for growth, which means that measurements of height and weight are identified much more closely with the individual's nutritional condition. So after allowing for the proportion of the population whose small body sizes are genetically determined, any deficiency in height and weight is regarded as evidence of malnutrition. This leads to more people being counted as malnourished than under most of the other procedures discussed in this chapter.

46 Limits to Measurement
Some policy implications
These differences in the definition and classification of malnutrition are carried over, not surprisingly, into policy recommendations. One approach, appropriate to the conditions in many low-income countries where death rates are high and resources are scarce, is to concentrate on those conditions in which lives are threatened most immediately and directly.
A much more comprehensive approach is sometimes encountered, however, especially among those who take the 'genetic potential' view. One early exponent was John Boyd-Orr, whose concept of adequate nutrition was 'a state of well-being such that no improvement can be effected by a change in diet' (Boyd-Orr, 1936, p.12). Interpreting 'well-being' in terms of a fairly ample and varied food intake, Boyd-Orr made surveys of how much food people in Britain could afford to buy during the economic depression of the early 1930s and estimated that about half the population was below this adequate intake - more than 20 million people (Boyd-Orr, 1936, p.49). When he became the first Director-General of FAO, Boyd-Orr initiated its first World Food Survey, which found that over 50 per cent of the world's population lived in 'calorie-deficient countries' (FAO, 1946). This was regularly interpreted as meaning that half, and by 1950, two-thirds of mankind were malnourished (Poleman, 1981).
A recent instance of estimates derived from the view that every individual should achieve his or her full genetic potential for growth was a statement attributed to Dr C. Gopalan which attracted attention in the western press (Guardian, 1982; Economist, 1983). Of the 23 million babies born in India each year, he said, 4 million die in childhood and 16 million experience malnutrition to a greater or lesser extent, leaving only 3 million to grow completely unscathed into healthy adults.
These viewpoints have validity as general social commentaries about the persistence of levels of nutrition which ought to be regarded as unacceptable. But there are dangers in using such all-embracing views of malnutrition and the high estimates of the numbers of malnourished people to which they lead as a sufficient basis for identifying policy. The kinds of policies proposed are likely to be general and non-discriminatory in

Defining malnutrition 47
character and their effectiveness will be judged on the basis of their impact on the average levels of nutrition of the population. The total benefit may be large, but it will be widely spread, so that even those whose needs are slight will experience some. In addition, of course greatly increased total food production is usually seen as a precondition. The difficulty with this kind of policy, however, is that most benefit is usually felt in the 'middle ranks' of society, and the bottom 5 or 10 per cent may be little affected, and are unlikely to recover the priority that their greater degree of nutritional deprivation should entitle them to. In addition, there is a very real likelihood that the problem of malnutrition comes to be seen as so large and so demanding of resources for increased consumption and welfare as to be impossible of solution within any reasonable time frame without sacrificing other development objectives. To plan welfare measures effectively requires either a more discriminating view of which groups within the population are most in need or the acceptance of the very high costs of interventions which do not discriminate.
A different policy approach is suggested by saying that instead of seeking 'the greatest happiness for the greatest number', one should demand, more modestly, 'the least amount of avoidable suffering for all'. Still other policies would result from suggesting that where some suffering cannot be avoided, as with hunger in times of food shortage, that suffering should be distributed 'as equally as possible'. Karl Popper (1945), whose point this is, goes on to argue that it aids clarity 'if we formulate our demands negatively, i.e. if we demand the elimination of suffering rather than the promotion of happiness' - or health.
The change in emphasis in political action which the minimisation of and equalisation of unavoidable suffering would imply is important. This is because the nature and intensity of nutritional deprivation is highly differentiated as between different groups of people and between various causes of dietary inadequacy. Action either to minimise or to equalise suffering might therefore start by identifying the worst cases. This might then lead to the implementation of measures that are specifically designed to change their circumstances. The point can be illustrated in

48 Limits to Measurement

Table 2.2 Malnutrition in children aged under 5 measured in Rwanda and Kenya

Severe malnutrition Moderate malnutrition (below 60% of (between 60% and 75% reference weight-for- reference weight-for-age, age, or 70% weight- or between 70% and for-height) 80% weight-for-height)

Rwanda, % of children
under 5 9.8 44.9
Kenya, % of children
under 5 1.0 25.0
Source: FAO 1977.

principle, though not in historical fact, by data from anthropometric surveys quoted by FAO (1977), and particularly by figures from two African countries (Table 2.2).
In Rwanda, it would clearly be important to plan policy interventions around the 10 per cent of children who are severely malnourished. In Kenya, policy could be based on a broader definition of malnutrition - we could take 75 per cent weight-f orage as a cut-off point without having a problem which is unmanageable in a given political economy - but we would still need to investigate the 1 per cent of severely malnourished children to check whether they represent concentrations of poverty for which resources for special measures might be needed.
As conditions in a country improve, and if pockets of severe deprivation can be eliminated, cut-off points for defining malnutrition can also be raised. In Britain in the 1930s, Boyd-Orr was right to use a high standard because many people in the nation were already well nourished, and as he pointed out, resources were available which could raise everybody to that level. But to apply the same standards in modern India is to present planners with a task too large for the resources at their disposal; more seriously, it hides the nutritionally urgent problems in the immensity of the less nutritionally urgent.

Defining malnutrition 49
Scientific models
In discussing the interpretation of anthropometric data, Trowbridge (1979) recalls the story of the blind men who encounter an elephant and have to identify it from the shapes they can feel. Each man touches a different part of the elephant and comes to a different conclusion about the nature of the beast. Similarly, weight-for-age, weight-for-height and arm circumference each touch upon 'different aspects of the vague entity called malnutrition'; none gives us a view of the whole condition.
In the next chapter, we explore the adaptability view of malnutrition further by examining the scientific model of food requirements with which it is associated. The elephant analogy is again worth bearing in mind because it is a feature of all scientific models that they represent the phenomena being studied in a selective and deliberately simplified way. Only small parts of any elephant are adequately portrayed. In the physical sciences, the simplifications are obvious when investigators talk about 'ideal gases' or 'weightless bodies', or when they ignore friction. Useful conclusions can be drawn from such models, but when these conclusions are applied to the actual world, care and realism are needed.
The scientific view of malnutrition with which we are concerned simplifies and selects from the overall problem area by taking 'failure of function' as its basic criterion. Partly this is because physiological functions such as growth, or resistance to disease can be measured much more satisfactorily than, say, hunger: but also because we are making a judgement about the seriousness to individuals of the consequences of physiological break-down as compared to hunger, or minor forms of ill health. We must not forget that those value judgements have been made, and are implicit in every application of this (or any other) model.
Some comparable issues arise in connection with the so-called 'human capital' model of welfare policy discussed in Chapter 4. This represents a way of thinking about education and health services as 'investments' in human productive power, and as capable of being evaluated like other investments in cost-benefit terms. Such ideas may be of academic interest to economists, but

50 Limits to Measurement
could lead to very perverse decisions if applied without caution directly to policy. The same goes for biological concepts which regard the human body as a self-regulating machine. Either view might imply, for example, that welfare should be provided at a level which supports the maximum ratio of labour output to cost, regardless of whether the people who form the labour 'resource' experience hunger and distress. The models do not necessarily lead to such conclusions. They can be interpreted this way only if we forget that they are highly simplified views and do not represent the whole human condition.
It must be stressed, therefore, that the adaptability view of malnutrition and related scientific models are not intend ed to discount as unimportant the hunger and ill-health apparently associated with inadequate food intake. What they may do, however, is to encourage us to recognise circumstances where providing more and better food does not solve every problem. In many instances, for example, it will make more sense to improve water supplies or housing, to develop immunisation programmes or find ways of enabling mothers to devote more time to child care. More fundamentally, since the underlying reason why people have to put up with inadequate food and bad health is poverty, to over-state the case concerning malnutrition may distract governments from the politically uncomfortable task of redistributing wealth.

3 Energy and protein


Food energy estimates
Two kinds of statement are commonly made about the amount of food which people need. Firstly, figures are quoted for quantities of nutrients which could safely be recommended for practically all individuals, even though they may be living under a wide variety of situations. When energy is referred to, these figures are based on observed average intakes or expenditures in apparently healthy populations. Figures for other nutrients are for intakes which are adjusted according to the statistical distribution of individual measurements; they are set at two standard deviations above the average. These 'recommended intakes', or 'safe dietary allowances' are thus based on overproviding for the vast majority of people in order to ensure that everybody gets enough. Such a policy would certainly enable each individual to achieve his or her genetic potential for growth, and might be a sensible basis for planning diets for institutions. However, such figures are of little help in assessing the likelihood of malnutrition (Longhurst and Payne, 1979).
Secondly, however, estimates are sometimes quoted for minimum physiological requirements, that is, levels of intake below which there is an increasing probability that some

*This chapter is based mainly on material prepared by Philip Payne. See Payne, 1976, 1982.

52 Limits to Measurement

specified symptom of deficiency will appear. The symptoms concerned are the failures of body function previously mentioned, such as growth in children, reproduction and work in adults, and failure to resist infection. Use of these criteria to define minimum requirements is, of course, consistent with the adaptability view of malnutrition discussed in the previous chapter, except that when figures are quoted, the extent to which requirements may vary with adaptation to different environments is not usually stated.
Most research has been done on western populations in which, for example, the pressure to follow a seasonal work pattern or to continue to work during pregnancy or lactation is relatively slight. Thus while we know something about levels of energy intake which on average are adequate for these populations, we know very little about populations which depend heavily on the seasonal use of stored energy and nutrients, or about populations adapted to a higher level of efficiency through small body size or in other ways.
Overall energy expenditures in the body depend on levels of activity, growth, body composition, diet and metabolic charaC7 teristics. All these are subject to adaptation, but the underlying rate of metabolism necessary for the maintenance of the functions of a resting individual is thought to be more consistent and predictable than most other factors. This is measured as the asal metabolic rate (BMR), and is found to vary with body weight in a fairly predictable way (Kleiber, 1961).
Because of this assumed regularity, and because BMR is clearly of fundamental importance, it was used by an FAO/ WHO committee in 1971 as the starting point for their estimates of minimal energy requirements (FAO/WHO, 1973). However, the values of basal metabolic rate accepted by the committee referred to studies made on American subjects by Talbot (1938), and the measurements relate to resting but wakeful individuals who were comfortable and warm after an overnight fast of 12-15 hours. Metabolic rates below these accepted figures for the resting rate may in fact be recorded during sleep (Benedict, 1915); they may also be recorded in subjects who have adapted to a low food intake over a period of time or who are indisputably undernourished.

Energy and protein requirements 53

Table 3.1 The effect of undernutrition on BMR; mean values reported from various studies

Reference and number Days of Mean daily intake Percent Percent of subjects fast during fast change change
MJ kcal in BMR in BMR
per per unit
person body

Benedict (1919)
studying 24 subjects 120 5.85 1400 19 12
Beattie and Herbert
(1947), 11 subjects >100 <7.3 < 1750 26 14
Keys et al. (1950),
with 33 subjects 200 6.55 1570 39 19
Grande (1964)
1 with 13 subjects 20 4.2 1000 17 9
2 with 12 subjects 14 4.2 1000 21 14

Observations of people undertaking partial fasts over long periods suggest that basal metabolic rates in underfed people may be anything between 10 and 30 per cent below the figures regarded as normal (Table 3. 1; compare Kleitman, 1926). These observations would provide a rational explanation for the frequent reports that populations exist in developing countries which habitually consume extremely low food intakes without catastrophic results (Miller and Rivers, 1972; Norgan et al., 1974).
It has been assumed that we can arrive at a figure for the food energy intake required to maintain body weight by working from BMR as a basis: hence the importance of measuring it. There are problems about doing so, however, of which the most serious is that we do not know the efficiency with which energy in food is converted and used by the body. Direct measurements of this efficiency in man are scarce. The FAO/WHO expert committee approached the problem by an ingenious comparative approach. As we have noted, BMR has a definite relationship with body weight that holds for most mammals as well as man. The FAO/WHO committee collated the evidence on food energy intakes at zero energy or zero nitrogen balance in experiments on

54 Limits to Measurement

animals and man; they showed that this too was related to body weight, and that it was consistently about 1.5 times greater than BNIR. Thus they concluded that the minimal food energy cost of weight maintenance is approximately 1.5 x BNIR.
This result could be used to estimate a minimum energy intake required to sustain life, and indeed was so used by FAO in its Fourth World Food Survey. FAO needed to take account of possible adaptations in body maintenance requirements and physiological efficiency which seem to allow some people to survive on less than 1.5 X BNIR. Faced with this problem, FAO took the lower limit of normal variation of BNIR (80 per cent of the mean value) and regarded it as a lower limit of adaptation. Hence they adopted a formula for the minimal energy cost of maintenance as 1.5 x 0.8 x BMR (= 1.2 BNIR). FAO then accepted this value as a 'critical limit' for maintenance requirements which could be used in defining undernutrition.
The conventional method of analysing food energy requirements, according to Wood and Capstick (1928), partitions the energy expenditure of an animal into three separate components: maintenance, growth, and activity. These three components are assumed not to interact, and maintenance, which we have already considered, often represents the biggest part of the total energy demand. With regard to the energy cost of growth, the FAO/WHO committee estimated a requirement of 21KJ (5 kcal) above maintenance for every gramme of tissue gained. (This is in fact a high figure compared with some recorded values (e.g. Ashworth, 1969), but given the errors that must be involved in other parts of the food requirements estimate, accuracy with regard to this component matters little.)
The third component of energy expenditure is activity. Here, of course, the output of people doing manual work can be measured, and a typical figure of 0.735 MJ (175 kcal) per day for adults employed in agriculture is sometimes used as a basis for estimates (Bayliss-Smith, 1981). Bearing in mind the seasonality of agriculture, we may expect average work rates to be less than this. Peak demands may partly be met by drawing on body energy stores, so how much extra food is required to support his labour will depend on the efficiency of body storage as well as on

Energy and protein requirements 55

the conversion efficiency of the muscles. Where people have adapted to a low food intake, that could partly be by increased efficiency, but probably also depends on a reduction in unnecessary exertion. There is evidence for both types of adaptation. Keys et al. (1950) reported that voluntary activity decreased during experimental semi-starvation, but also said that the energy cost of specified tasks was the same per unit of body weight throughout. This differs from Seckler's (1981) finding that overall work efficiency is inversely related to body weight over a limited range. But Keys agrees that the total energy cost of work tends to fall because body weight is reduced. Some physiologists have claimed that 'work capacity' is also impaired by undernutrition, but this is difficult to substantiate; in practice, the maximum rate of work a person can sustain is rarely important in limiting his or her effectiveness as a producer. Actual work rates are more ofte n limited by the nature of the work itself, and by the social factors determining its organisation.
The outcome of all this is to leave maintenance as the dominant aspect of food energy requirements, especially for people surviving on very low intakes. Thus FAG equated their estimate of minimum maintenance requirement with the minimum energy intake required to sustain life. So we end up with two different kinds of estimate for human energy requirements, one an observed average and the other an estimated minimum, the latter being intended to represent a 'critical limit' of adaptation to low food intakes.
To illustrate the difference between these two kinds of requirement, we may consider the energy needs of a 'reference man', aged 25 to 40 and weighing 65 kg. The figures are as follows:
1. recommended intake (FAG/WHO, 1973), equivalent to a predicted average intake for a well-fed population: 12.5 MJ/day (3000 kcal/day);
2. minimum physiological needs, corresponding to a critical limit of adaptation taken as 1.2 BMR (FAG, 1977): 7.5 MJ/day (1800 kcal/day).

56 Limits to Measurement
The corresponding figures for a 55 kg woman of the same age who is neither pregnant nor lactating are 9.2 and 6.3 MJ/day (2200 and 1500 kcal/day).
. Neither figure, of course, can be regarded as a correct statement of the requirements of any specific group of real human beings. The higher figure is now generally thought to over-estimate requirements in most populations. As to the lower figure, this refers to people who are likely to be experiencing hunger, often painfully and for long periods, but whose essential body functions are not impaired. We have noted some of the arbitrary assumptions in its estimation, and have pointed out that to quote a single figure ignores the variations to be expected in different environments. Lipton (1982) stresses that a lower limit of energy intake will not only vary between individuals, gene pools, age-groups, and so on, but that it also depends on the choices people make (or are forced to make) about maintaining work output 'when food is short. We may well end up by questioning whether it can ever make sense to state requirements in this, way, rather than specifying ranges, or stating them relative to local conditions.
It is worth recalling the three apparently healthy populations with sharply contrasting lifestyles which were referred to in Tables 1.1 and 1.2. Body weights in these populations were mostly less than for the reference man and woman quoted above, but the range of average intakes for the groups represented is from 8.12 to 14.4 MJ/day for men, and from 5.94 to 12.1 MJ/day for women, varying with season.

Adaptations and responses
The foregoing discussion left many questions unanswered about the ener gy required for work in populations surviving on very low intakes. Two factors need to be considered: long-term adaptations, particularly in body size, and short-term responses to temporary (or seasonal) food shortages. The latter include physiological responses whereby the conversion efficiency of energy in the body can rise when intakes fall. This may happen,

Energy and protein requirements 57
for example, if food energy is used directly as glucose instead of being converted to fat and released at some later time, and other variations in the pathways for energy use may take place. But behavioural responses to reduced intake are important also, and Lipton (1982, p.36) pointedly asks: 'Suppose a body can, when daily intake falls 10 per cent, maintain work levels and cut requirements via higher conversion efficiency. To what extent will different groups of people choose not to reduce work . so forcing their bodies to "raise conversion efficiency"T Nothing is known about such matters.
With regard to long-term adaptation, one necessary adjustment in conventional thinking concerns the status of individuals who are small and under-weight by international standards, but who show no signs of functional impairment. The nutritional status of such people has often been described as 'mild malnutrition' when populations are screened by height or by weight in relation to age, but it might be better to regard them as 'small but healthy'where no contrary evidence exists. The latter interpretation may seem especially relevant in the light of what has already been said about adaptations to a wide range of seasonal and work environments. Before people are classed as malnourished, therefore, and before their food intakes are deemed insufficient, we ought to examine the environments from which such people have come, and consider the question of adaptation.
For example, traditional diets in Japan have produced many small but undoubtedly fit people. With current trends to more westernised eating habits, many individuals are growing taller and heavier, but there is concern that they may be more vulnerable to the degenerative diseases of later life, and that stamina for work may be less. Certainly there are no objective data to support the idea that western diets and western anthropometric standards represent an optimum.
Under conditions of food scarcity, in particular, there may be distinct advantages in small body size, since nutrient requirements are related to body weight. However, Gopalan (1983) and others have advanced a strong argument against this position. Since large people have more muscle tissue than small people, their work capacity is greater and therefore they can earn more in

58 Limits to Measurement
manual labour than small people. A commonly cited study in support of this position is that of Satyanarayana et al. (1977) showing that among 47 workers engaged in assembling fuses into bundles on a. piecework wage basis, the heavier workers consistently produced more than the lighter workers even in this moderate level of work activity.
Seckler (1981) argues that we should not confuse total production with efficient use of energy in production. Although the heavier workers produced more, they may have done so less efficiently. Two contrary points should be noted, however. Firstly, this job was paid as piecework, which may have led to untypical results. Secondly, it was light industrial work for which the energy requirements of people when they were on the job would be dominated by body maintenance needs rather than muscular energy expenditures. In these latter conditions, it is certainly likely that small people would be more efficient.
In heavy work, there is no evidence of any particular body size favouring high efficiency; muscular energy conversion efficiency is now thought to be the most important rate limiting step and this is not likely to be affected by body size. In studies of road workers made by Tandon et al. (1975), the bigger, heavier men certainly tended to produce a greater work output. But in a comparison of those with the highest outputs and those with the lowest, food energy intakes were found to be almost the same. When local workers (average body weight 45.5 kg) were compared with group of more practised road workers from outside the area (average weight 50.4 kg), it was found that the latter did more work with a lower food energy intake. Their efficiency in this respect was about 50 per cent greater, apparently due mainly to their greater experience and dexterity.
With regard to heavy manual work, therefore, small people do not seem to have any advantage, but it does seem possible that populations of 'small but healthy' people represent other kinds of adaptation with respect to work and survival; especially because for them, maintenance requirements are less. What we need to notice, however, is that small body size is an adaptation which is made during growth, sometimes at the cost of considerable deprivation. By contrast, we have also talked about

Energy and protein requirements 59
short-term modifications of behaviour and metabolism related to seasonal conditions, describing these as 'responses' to low food intakes. They are sustained for a few months at most, and where they entail the depletion of body stores of fat or nutrients, they are clearly not sustainable in the long term.
This distinction between sustained adaptation and short-term response is important if 'critical limits of adaptation' are specified for use in assessing the extent of undernutrition in a population. In most circumstances, we will be enquiring whether people's capacity for short-term response is overstretched, and we will be assuming that any long-term adaptations are fixed. But such adaptations must differ markedly from one region and one social group to another, and the limits of response within that long-term framework must presumably also differ. Thus even if the critical limit of 1.2 BMR is a reasonable best estimate of the lowest level to which existing social groups will be found to have adapted, it should not be regarded as a limit below which no further adaptation is possible.
Some of these ideas challenge the conventional wisdom about the variability of human food requirements. They conflict with the belief that each individual in a population has his or her own specific requirement level for energy or protein, related to genetic potential for growth and health, and in some way fixed for that individual. People are acknowledged to differ slightly, but in a way that is assumed to conform with a normal Gaussian distribution. Standard deviation is then a clear and unambiguous expression of the extent of variations.
By contrast, the arguments put forward in this chapter imply that individuals each have a range of possible responses appropriate to a range of food intakes, and are not characterised by any fixed requirement. Indeed, Sukhatme has argued that the very notion of a fixed requirement is in conflict with most of the evidence collected over the last forty years. The data are much more satisfactorily understood in terms of a self-regulating, homeostatic model (Sukhatme and Margen, 1978).
This is a complex idea, but a simple analogy maybe useful in showing what it involves. Suppose a leaking bucket is being filled at a pump. It is possible to adjust the flow into the bucket so that

60 Limits to Measurement
it exactly matches the rate of leakage. Then equilibrium is maintained, with a steady water level. We can alter the rate of inflow so that equilibrium is reached with the bucket almost full, or with the bucket nearly empty. In the latter case, the rate of leakage is less because of lower pressure, and perhaps because some leaks are above the water line; thus a lower rate of inflow is necessary. Clearly, there is a range of flows over which many different equilibrium states are possible. There is also a critical lower limit of inflow below which the bucket empties, and an upper limit, beyond which it overflows. To some extent, at least, the responses of the human body to variations in food intake are similar.
One implication is that when we observe a range of different food intakes in an apparently healthy population, this not only reflects variability between individuals, but more fundamentally, it reflects a range of possible equilibrium states for each of those people. Thus standard deviation takes on a different significance, since it is now seen to represent the potential for response to changes, rather than random statistical differences. A food intake that is below average by a matter of two standard deviations conventionally represents an intake which is lower than the requirements of all but 2.5 per cent of the population; according to Sukhatme however it could just as reasonably be taken as an estimate of the lower limit of possible adapted responses in that population.
Applied to food energy intakes, this estimate gives a rather low figure for minimum physiological requirements, but not as low as that obtained from FAO's (1977) assumption that the lower limit of adaptation for many populations might be around 1.2 BMR. For example, working on two standard deviations below the average, Sukhatme (1961) calculated the minimum for a 55 kg man as 8.8 MJ or 2100 kcal, which is 22 per cent below the FAO/WHO recommended intake, whereas using 1.2 X BMR gives a value of 40 per cent below.
This example of extremes reinforces the warnings about scientific models that were sounded at the end of Chapter 2, since in this instance, one's choice of model affects conclusions very substantially. In fact little has been done in a systematic way to

Energy and protein requirements 61
test the various models. A major difficulty is that knowledge of adult human requirements is largely based on experiments in which the objective has been to maintain equilibrium. The argument applies to protein as well as to energy intakes, and experiments have been done on apparently healthy people in which levels of intake are found which are just sufficient to maintain the existing level of body nitrogen, However, the assumption then made that nitrogen balance indicates adequate intake is not valid. It is, rather, j ust an indicator that the subject's equilibrium state is not changing; the significance for health of the changes in equilibrium state that do commonly occur is not known.
Even if the adaptability model comes to be accepted as being closer to reality than that currently advocated by FAO and WHO committees, it will still not provide us with a completely general basis for prediction because of other factors which influence variability. Measurements made at a single point in time on individuals of the same age and sex will include at least the following sources of variance:
1. differences in adaptation, including those reflecting the primary expression of genetic factors, and those arising from the effects of diet and environment during growth;
2. differences due to the extent and nature of continuous auto-regulatory responses;
3. a 'noise' component due to random, unrelated fluctuation in the environment, or diet, or due to measurement errors.
If we were to conduct experiments in order to establish the safe limits of adaptive response to different intakes of some nutrients, we would need to distinguish between these sources of variability. Assuming that in general we would find both upper and lower limits to the range, with some risk of dysfunction attached to high levels of intake as well as low levels, the situation would be as shown in the diagram (Figure 3. 1). Instead of a single figure for the recommended intake or safe level for the particular nutrient, we would have a 'safe range', with upper and lower limits bounded by points for which the probability of dysfunction rises above an acceptable level.

62 Limits to Measurement
'Adaptive' models of body weight control The concept of an adaptive, self-regulating system implies some type of negative feedback response by which the body is able in part to cancel out changes in inputs and outputs, and maintain the internal state near its original condition. There will be several such feedback loops, and in the particular case of energy, one even extends outside the body, so that an individual may adapt by modifying his interaction with the environment, and hence change his requirements for energy to bring them into line with

Intake --lo-

Range of individual variation of limits of adaptive response

Figure3.1 Schematic view of the range of adaptive responses likely to
-be encountered with different levels of nutrient intake

Energy and protein requirements 63

intake. Thus the total strategy of adaptive response for energy comprises individual and social changes in behaviour as well as changes in body weight, body composition and . metabolic regulation. It may well be that individuals differ in the relative balance of these components, and that for some, changes in behaviour will be a more important adaptive mechanism than for others.
On a physiological level, it is clear that individuals differ very markedly with regard to energy storage within the body. For the majority, energy is chiefly stored in the form of fat, but for a few people, significant quantities of energy are also stored as protein. We can represent these differences by defining a ratio which represents the proportion of energy stored as protein in the body. The same ratio predicts the energy mobilized from protein when energy intake is less than expenditure. The 'P-value' (Payne and Dugdale, 1977) expresses this component of energy as a ratio of the total energy stored in the form of fat and lean tissues in the body (or the total energy mobilised when intakes fall). Figure 3.2 shows the proportions of energy mobilised from protein in semi-starving men as observed by Keys et al. (1950), and it is evident that these men showed a wide variation in the way they responded to a negative energy balance. On average, 15 per cent of energy was lost from lean tissue stores and 85 per cent from fat. But the average actually spans a tenfold range: some individuals lost only 3 per cent of energy from protein, others as much as 30 per cent.
We do not know the extent to which the propensity to store and utilise protein as opposed to fat is a result of genetic differences, and how much is a result of patterns established during early growth. We do know that under some circumstances, a tendency to use fat as an energy store might be a good survival strategy, whereas in other circumstances it could lead to obesity. The nature of responses to increased energy intakes may be similarly varied. The data of Miller and Mumford (1967) showed considerable variations between individuals in respect of degree of weight gain during overfeeding. This may partly be because the metabolic efficiency of energy storage varies from one person to another and partly because some people are less

64 Limits to Measurement





P value

p =(energy stored as, or mobilised from protein )I
total energy stored in the body, or mobilised/

Figure 3.2 Differences between individual men with respect to the storage of energy as fat or as protein Notes: P-value would be zero for an individual whose body store of energy consisted solely of fat, and 1.0 for an individual storing energy solely as protein. Source: Dugdale and Payne 1977, based on data from the experiments of Keys et al. 1950.

Energy and protein requirements 65

efficient when they over eat. It also seems likely that individuals with a propensity for fat storage will gain more weight before increased metabolism balances out the higher intake than will individuals who store metabolically active lean tissue.

Protein requirements
As we know so little about the complex strategy of response and adaptation, and have no reliable indicators to predict how a given individual will respond to changes in intake, either by body weight, composition, or behaviour, the definition of energy requirements is an uncertain business. In respect of protein, we are hardly better off. Obsession with nitrogen balance as a 'criterion' of adequacy has led to a neglect of the study of adaptive changes and mechanisms. Nonetheless, it is well established that such mechanisms exist. Waterlow (1981) has reviewed the evidence for adaptive responses in nitrogen metabolism, and concluded that the body possesses powerful mechanisms for economising on nitrogen. f
For example, Durkin et al. (1981) showed that in most cases, responses to reduced levels of nitrogen intake included the loss of body weight, Nicol and Phillips (1976) showed that Nigerian farmers, habituated to diets relatively low in protein, utilised that protein more efficiently than American subjects, and could maintain balance on about 30 per cent less. It appears that the Nigerians were adapted to low intake.
Considering that little is known about such adaptations, it is not surprising that efforts to define human protein requirements have had a chequered history. 'Kwashiorkor' was described by Williams (1933) as an illness, 'in which some amino acid or protein deficiency cannot be excluded'. The children who developed the disease did so after weaning, when they were fed on starchy porridge rather low in protein, energy and other nutrients, and they were cured by milk. in time, the idea that this syndrome was specifically caused by protein deficiency became generally established.

66 Limits to Measurement

Subsequent work, however, challenged the definition of kwashiorkor as protein deficiency, and has led to a revision of ideas about how much protein young children need. Lower values for their protein requirements have slowly gained acceptance, but energy needs are still estimated to be at around the same level (Table 3.2). No set of symptoms or signs is now generally agreed to be evidence of a specific deficiency of protein in man. This is, of course, not to say that there is no malnutrition in children, nor is it to deny that there is such a thing as kwashiorkor. It is simply that neither clinical studies nor attempts to reproduce the condition in animals provide sufficient justification for identifying the well-known symptoms (oed~ema, hair and skin changes, fatty liver, blood changes) as specific for protein or amino acid deficiency (Waterlow and Payne, 1975).
The results of food consumption measurements made on individual children in a variety of societies seem to confirm this conclusion (Waterlow and Rutishauser, 1974). Where intakes of protein are low relative to estimates of requirements, so also are intakes of energy, and often all other essential nutrients as well. But these deficits of nutrients are often such as would be made

Table 3.2 Estimates of protein and energy requirements at I year of age, corresponding to 'recommended intakes'

Year Per kg body weight Ratio of
Protein' Energy protein energy Source to total energy
g MJ kcal %

1948 3.3 0.42 100 13.2 NAS (1948)2
1957 2.0 0.42 100 8.0 FAG (1957)
1963 2.5 0.42 100 10.0 NAS (1963)2
1965 1.1 0.42 100 4.4 FAG/WHO (1965)
1968 1.8 0.42 100 7.2 NAS (1968)2
1969 1.3 0.47 110 4.7 DHSS (1969)3
1973 1.25 0.44 105 4.8 FAG/WHG (1973)
1974 1.35 0.42 100 5.4 NAS (1974)3
1As protein which is 100% utilised.
2 US source; protein estimates consistently higher than in other sources.
3 UK source.

Energy and protein requirements 67
good automatically if energy needs were satisfied by taking more of the same food. Thus the most significant factor causing child malnutrition would seem to be simply inadequate food intake.
Because of their rapid growth rates, infants need a somewhat higher proportion of protein in their diets than is required by older children and young adults. But we need to remember that protein plays a highly complex role in physiology. It is at once a nutrient in its own right, and a source of energy. When protein levels in diets are high, a large proportion of the amino acids derived from protein is dearninated and oxidised as fuel. This also occurs in situations of overall energy deficit. Thus a child given an adequate protein diet, but at intake levels below his or her energy needs, will not derive full benefit for growth from the protein, but will use it as an energy source.
A further complication is that proteins do not all have the same growth-promoting value. Depending on the balance of their component amino acids, some have higher 'quality' than others. Protein quality indexes (such as 'net protein utilisation', or 'chemical score') are estimates of the percentage of the total protein which can be expected to be utilised for growth and maintenance in a young animal or a child. On this basis, eggs and milk have efficiencies close to 100 per cent. A food or diet with a quality of 70 per cent will contribute more of its amino acids to growth than one of 40 per cent. However, the amino acids which are not used for protein metabolism will not be wasted: rather, they will be diverted for use as an energy source.
Because protein can provide the body with energy in this way, at roughly 17 KJ or 4 kcal from each gramme converted, it is useful and usual to express the protein value of diets in relation to their energy content. The relevant ratio, expressed as a percentage, is given by one of the following formulae, depending on the units employed:

protein-energy ratio = (grammes protein in diet X 0.017 X 100) total megajoules in diet

(grammes protein in diet X 4 X 100)
total kilocalories in diet

68 Limits to Measurement

A further refinement is possible by correcting the energy value of the proteins by a factor which takes quality into account. The result is a ratio of energy derived from fully utilisable protein: the 'net dietary protein-energy ratio' (NDpE%) (Payne, 1972).
Animals are often used to assess the nutritional values of different foods, and for example to compare different strains and varieties of cereal grains as protein sources. Thus the superior growth of rats and pigs on high lysine varieties of maize and rice have often been cited as evidence of their advantages for humans. However, in interpreting such feeding trials it is important to note the major differences that exist in the nutrient needs of different species.
Most young mammals grow very much faster than human infants when growth is measured as weight gain per unit body weight, and need a larger proportion of protein in the diets to support this growth. Figure 3.3 illustrates the point diagramatically. If feeding experiments were carried out with an adult man, a one-year-old child, a new-born baby, and a young pig, using a diet containing 10 per cent of a protein, one of whose amino acids could be varied from zero up to 100 per cent (with perfect balance of all other amino

Positive growth
Baby achieves
maximum growth

I. One-year-old achieves
maximum growth
Adult man maintains
body weight

Weight Adult man
loss loses weight

0 20 40 60 80 100
Protein quality (egg or milk taken as 100)

Figure 3.3 Feeding experiments based on a diet containing 10 per
cent protein of varying quality

Energy and protein requirements 69
acids), then the man would lose weight with a zero ratio, and would show improvement as the level was raised up to about 40 per cent of maximum, and then no further response. Similarly, the one-yearold would continue to benefit from increases up to about 50 per cent, the new-born baby up to 70-80 per cent, but the pig up to 100 per cent.
It should be stressed that this diagram represents the protein requirements for maximum growth in human babies, which is not the same thing as the optimum protein requirement. Children, unlike pigs, are not deliberately fattened for market, so it should not be assumed that the fastest rate of growth is necessarily the best; indeed, we will shortly quote reasons for thinking that it is not. What the diagram makes clear, however, is that the growth of children does not always suffer if protein quality is less than 100. In any practical situation, what has to be assessed is whether the quantity of protein eaten, in conjunction with the balance of amino acids, i.e. its quality, is such that the requirements of the consumer for the most limiting amino acid are met. The balance of evidence from feeding trials seems to .,show that although cereals are limited in protein quality by their content of lysine, threonine and tryptophan, and although this is as true for children as it is for pigs, provided that sufficient cereal is consumed to meet energy and total protein needs, then conventional varieties of most cereals contain sufficient of these amino acids to meet the requirements even of preschool children.
As with energy, so with protein, therefore, the conclusion to emerge from this chapter is that human food requirements are less certainly known than was once thought. This ought not to be regarded as totally discouraging, however, but rather as a challenge to regard nutrition with a more comprehensive understanding of physiological function and environmental constraint. Part of that challenge is to combine these insights with a study of the variability between individuals' capacity to adapt to environmental changes without unacceptable loss of function. Up to now, many nutritionists have tended to overlook this problem area, partly because of the difficulty of studying it experimentally, but partly also because the models conventionally used to describe the interaction of requirements and nutrient intake implicitly assume that under normal

70 Limits to Measurement

conditions, adaptation is a relatively minor part of total variability. If adaptation does occur, the assumption has usually been that the situation must be abnormal, and failing proof to the contrary, the adapted state is regarded as likely to confer some disadvantage.

When a nutritional requirement level is stated, a prediction is being made on the basis of past knowledge about the probability of specific functional disabilities affecting a particular group or individual. Ideally, the requirement level should be presented in a form which indicates this fact. However, it has to be admitted that few satisfactory statements of this kind are possible, e.g. that 'there is a probability of 0.1 � 0.01 that fifteen-year-old non-pregnant females will develop night blindness if they sustain an intake of less than 400 RE (retinal equivalents) of vitamin A per day' (needless to say, we do not know if this is in fact true). Ideally, too, the probability should be established within confidence limits, and the dysfunction to which it refers should be specified and measurable. Other kinds of dysfunction related to vitamin A status may be known to exist, and their probabilities could also be stated for that sample level of intake. Since many dysfunctions have multiple causes, some statement about the context (environment infection, etc.) might also be necessary. Where other departures from health may only be suspected, no true requirement figure can or should be stated with respect to these.
Something which is specifically excluded from this definition is the notion of an 'optimum' state of nutritional health, achievement of which might be the criterion for a requirement level. This is an idealistic notion which seems to be sustained by two kinds of attitude. One is a view which reacts against minimum prescriptions, and would prefer to see emphasis placed on the complete fulfilment of human genetic potential'. This feeling demands respect, and if resources were available - and if the concept of genetic potential could be clarified - policies based on such ideas would deserve support.

Energy and protein requirements 71
The other attitude is that optimum states are somehow inherent in 'natural' systems, and that the right nutritional conditions are those that allow these states to be expressed. This is commonly linked to the idea that natural selection will have maximised 'fitness', and that there should exist a state of nutrition in the individual which will be a reflection of that fitness. However, we have no reason to believe that there is some preferred state of metabolism or physiology in the individual which simultaneously brings to a maximum all qualities that we happen to consider desirable, and so, at the other end of life, is longevity. But there is no evolutionary selection for longevity - only for survival to sexual maturity (Kirkwood, 1977). Nor is there any evidence that children who grow rapidly enjoy long life. Indeed, to the extent that experiments with animals can throw light on the matter, Ross et al. (1976) found that when young rats were allowed to select their own diets, there was considerable individual variation in food energy consumption and growth rate, and that high rates were negatively correlated with lifespan. In this instance, at least, it is clear that energy requirements for young animals could be based either on the criterion of achieving the full genetic potential for early growth, or on subsequent longevity, but not both.
Whether or not this evidence can be paralleled in humans, it serves by analogy to show why any views of 'desirable' or 'optimal' food intakes for human individuals or groups can only be value judgements. If our aim is to be as objective as possible, we have to be content with describing how the probability of failure to sustain function varies with intake. Deciding what levels of probability or risk are unacceptable, brings us back to value statements again. What is quite clear is that where energy and protein requirements are concerned, doing this leads to critical limits considerably below what many people would judge to be desirable.
Yet science is not objective and value-free. The preceding paragraph is itself a value-judgement above the limits of the scientist's responsibilities. Other aspects of the value-system inherent in science are reflected in the simplifications which scientists build into their models, and are to be observed also in uncritical reverence for measurements and statistics. This is often exploited by policy-makers who use the figures so produced both to legitimise

72 Limits to Measurement

past actions and to guide their plans. Enough has been said here to stress that our understanding of individual adaptability with regard to food is extremely sparse. That clearly limits our ability to make quantitative statements about nutrient requirements. Yet over the years, governments and international agencies have produced a stream of reports listing minimum physiological requirements, safe levels, and recommended dietary allowances, which are supposed to have a firm objective basis. In subsequent chapters, we shall describe an approach to the use of quantitative data which we believe can be much more helpful in understanding malnutrition and in informing action to reduce its incidence.

4 Food system indicators*

Variables in food systems
Two kinds of measurement used in monitoring food systems have been discussed in previous pages: anthropometric assessments of nutritional status (Chapter 2), and measurements of food intake related to ideas about requirements (Chapter 3). In considering food supply and utilisation over a whole nation, as well as the nutritional status of its people, there are obviously many other variables which might usefully be measured: foodstocks and production, for example, or people's expenditure on food. Figure 4.1 is developed from the food system diagram presented earlier (Figure 1.1) in order to show whereabouts in the system observations are commonly made or data collected. All the variables concerned are examples of system variables. The values they take at any one moment reflect the state of the system at that moment. Normally, a variety of processes and changes will be continuously under way, and measurements of these variables specify the state of the system in 'frozen' form, rather like photographs taken at particular instants.
It will be noticed that these system variables are of two kinds. Some, like body weight or food stocks, are characteristic of a

*This chapter is based on material prepared by Peter Cutler, Elizabeth Dowler, Philip Payne, Young Ok Seo, Anne Thomson and Erica Wheeler. Fuller versions of some of the material have been published by Dowler et al., (1982) and also by Dowler and Seo (1983).

74 Limits to Measurement

figur 4.1te footsytemofation or ntono sida dfndi comonl measuredivdua Note:alenoes aibe hc esr lw hog h ytm deote varnicalewhcrelccodtoswtitesyem

Food system indicators 75
condition in the system that is liable to change fairly slowly. Others measure flows through the system, and like the flow from a tap, they can fluctuate widely or even be abruptly turned off. One example is cash income, which drops to zero overnight when an individual loses his job. Clearly, these two kinds of variable reflect on the state of the system in different ways. Much confusion surrounding data on energy or protein intakes arises because these latter are flows, yet some people insist on treating the figures as if they represent an established condition of either adequate or inadequate nutrition. The distinction can be understood by referring again to a leaking bucket that is being filled at a pump. Measuring the flow of water from the pump cannot tell us how full the bucket is - we have to observe or measure that directly.
The interpretation of measurements often involves checking them against critical levels of the variables to which they refer. These are levels below (or above) which it is generally agreed there is evidence of an unacceptable degree of immediate or impending distress. It is essentially the use of cut-off points defining such levels which distinguishes the particular variables which we choose to define as indicators, from other system variables.
The levels at which cut-off points are set should ideally depend on the ' purpose for which the indicator is being used. We have seen how this works for anthropometric measurements, where a working definition of a cut-off point for, say, weight-for-age, is defined to fit particular circumstances, and may vary according to whether the purpose is surveillance or screening. With regard to food intakes, the critical points are the requirement figures and estimates of adaptability limits, and which ones we use may again depend on our purpose. If the aim is to identify social groups with the most seriously inadequate diets, we might use an estimate for the lower critical limit of adaptation. If, however, the aim is to buy sufficient food for residents in an institution, we would certainly use a higher requirement figure.
To sum up, then, an indicator is a variable such as food intake or weight-for-age for which critical points have been agreed. A simple example from outside our subject is provided by the gauge which indicates steam pressure in a boiler. Here, the 'system', or set of components, consists of boiler, fire-box and engine. Steam pressure

76 Limits to Measurement

is the system variable which we select to describe the state of that system. In addition to measuring pressure, the steam gauge may also have certain critical points marked on the pressure scale, below which pressure is inadequate, or above which is a danger of explosion. It is these critical points which make steam pressure an indicator of the state of the boiler.
A normal characteristic of an indicator is that it is used to show a number of different aspects of the performance of a system. In this example, the situation is relatively simple because steam pressure is quantitatively related in a known way to other variables such as temperature, and strain in the walls of the boiler. Steam pressure also tells us about the balance between steam output from the boiler and fuel input, but it does not allow us to estimate these flows directly.
In an instance like this, where the relationship between variables is unambiguous and clear, we may use one variable as a 'shorthand' indicator of another without hesitation. In biological and social systems, however, matters are rarely so simple. We may know that, within a particular kind of system, certain variables have usually assumed values which bear consistent mathematical relationships with each other, but we may not know how this happens. In such circumstances, a variable may not have a sufficiently well defined significance to be regarded as a shorthand indicator, but it might be generally accepted as a proxy for a set of system variables, or even for certain aspects of the system which cannot as yet be precisely defined.
Nutritional indicators based on anthropometry are to some extent shorthand indicators, when weight and size are taken as estimates of body nutrient stores, and size at a given age is regarded as a measure of the previous pattern of growth. But in addition, anthropometric indicators are commonly used as proxies for ,nutritional status', an imprecise notion that, as we have seen, covers the outcomes of a wide range of different processes, including the effects of different nutrient deficiencies, and of non-nutritional factors such as infection.
Defining the critical points which are associated with the use of indicators must always involve the expression of value judgements, and attempts to achieve consensus about acceptable levels of risk.

Food system indicators 77
With regard to the boiler, for example, the owner of the plant may see advantages in terms of efficiency in running it at a high steam pressure, and provided he can afford the insurance premium, he may be happy to accept an increased risk of explosion. The men who stoke the boiler, however, may see no advantage in the higher pressure, but will be aware of a greater risk of injury or death. Their trade union might seek a much more stringent standard of safety than the management wishes to accept.
This illustrates a question which was raised in previous chapters but has not so far been answered. When we define malnutrition in terms of unacceptable levels of dysfunction, who decides what is unacceptable? The scientist cannot do this alone, despite the fact that he or she possesses some of the necessary information. With regard to the boiler, for example, an engineer might be called in to advise, but is in no position to judge what insurance premium the firm can afford, nor what level of risk the stokers (and their families) are willing to accept. Thus the engineer has to explore with the management, or perhaps with the trade union, what would be the consequences for them of a higher steam pressure and a greater risk. In other words, valuations of risk and of what is acceptable or otherwise should arise from a process of dialogue, either explicit or indirect, in which the scientist or technician is only one of the parties involved.
In nutrition, as in other fields, the role of the scientist is defined partly by his or her expertise in using a range of specialised techniques for observation, measurement and prediction. However, in no sense is his or her responsibility limited to this. Besides the work of measurement there are questions of social valuation, and whereas the former is a purely technical activity, the latter entails dialogue in some form with other sectors of society - with planners, no doubt, with those who fund research, and also, we might hope, with the ultimate consumers of the food. One of the differences between measuring a variable and using an indicator is that the former is a more or less objective technical activity while the latter always involves the use of critical points which express social valuations.

78 Limits to Measurement

Indicators commonly used
We have already commented on the use of anthropornetric indicators in the food system. We will now look at two other sets of indicators commonly used: food supply and consumption, and expenditure on food items.

Food supply and consumption
Among the more important variables that reflect on the operation of food systems are data on food supplies collected at national level, which are commonly expressed as per capita protein or energy availability. Such data are based on national food balance sheets (FBS) compiled regularly for many countries in a manner that takes account of food production, imports, exports and increases or decreases in food stocks. Grain used for seed or animal feed is subtracted, and so are estimated losses. (For detailed accounts of methodology, see FAO, 1980; Poleman, 1981.)
Despite several decades of effort, errors in these estimates are still known to be significant, and some authors suggest that for low income countries especially, they tend towards the understatement of food available (Poleman, 1981). Figures expressed as per caput supply introduce additional errors wherever population data are inaccurate. Moreover, as the food system diagram emphasises (Figure 4.1), FBS data represent flows through the system at a point which is rather remote from the individual eating a meal. Thus the figures do not represent human consumption, nor do they represent food distribution over time or over population. With good reason, the FAO cautions against misunderstanding their basis (FAO, 1980).
The usual way of estimating actual food intakes is through household consumption surveys (HCS, Figure 4. 1). These employ a range of methods including weighed intake surveys (larder stocks or plate/pot consumption), diary recording and recall interviews (Marr, 1971; Burk and Pao, 1976, 1980).
Among the difficulties involved in using this method are problems of sampling and of seasonality in food supply or usage.

Food system indicators 79

A further problem is that precise methods of weighing meals or ingredients tend to be intrusive, and may induce behavioural change and untypical intakes. Variation in eating patterns from day to day compound the problem, and household-based measurements usually omit food eaten outside the home, although in some instances, an estimate of food and/or nutrients in this category is included when results are published.
Do these two sources of data on food give the same information? Since FBS and HCS measurements of per capita food flows are made at different points in the system, we would not expect them to yield precisely the same results. Furthermore, FBS estimates cover 356 days; while HCS measurements typically cover 3-7 day periods in each household. Yet the differences which do occur have proved difficult to explain. FBS figures are always greater by about 1 MJ per capita in less developed countries, and by 3-4 MJ per capita in developed regions.
These differences may arise partly because alcohol and casually gathered fruit and nuts are not always included in the surveys as 'food'. Such mismatching cannot account for the discrepancies fully, however. Much emphasis is placed by some authorities on underestimation of wastage in the home as a source of error in HCS data in developed countries, but studies in Britain have failed to demonstrate such underestimation (Dowler, 1977; Wenlock et al., 1977, 1980).
In Japan, a nationwide survey of household food consumption based on seven-day weighed intakes has been carried out annually for over thirty years, and an interesting comparison can be made between HCS and FBS figures over the whole of that period (Seo, 1981). Figure 4.2 shows the results graphically, plotted against per caput gross national product (GNP) on a log scale. Average energy intake as measured by HCS remains reasonably constant over the period considered. However, per caput energy supply as judged by FBS figures rose rapidly as GNP increased. Comparison with cross-sectional data from other countries shows that not only is the FBS estimate always higher than the household estimate, but the discrepancy increases as the economy of a country, and thus its food system, becomes more extensive and complex, with more opportunities

kcal 3,0002,800-


GNP per caput ($ US log scale)

260 300 ' 50'6 0' 7 '0 1,000
1951 1958 1965

2,000 I ' 4,000 1'6,000 ' 10,000
1972 1974

Figure 4.2 Longitudinal comparison -Japan (energy and GNP/caput: energy supply data (food balance sheets) and energy intake data (household consumption surveys)) Source: Seo 1981.

-13 MJ

- (Food balance sheets)

. (FAO country requirement) (Hose hold consumption surveys)

2,600 2,400-10 2,200


Food system indicators 81

for loss or waste within the system as a whole, before reaching the household level.
Given the difficulties associated with these food supply data, and the uncertainty in estimates of 'requirements', efforts to use FBS figures to identify countries with an 'energy deficit' are bound to run into difficulties. Yet official bodies continue to describe global and national energy deficits in this way. The US Department of Agriculture produced two World Food Budgets relating to the 1960s (USDA, 1961, 1964) which concluded that almost the entire population of the developing world lived in 'diet deficient' countries. The World Bank (1976), using data from the 1960s, saw a problem of roughly the same magnitude involving about 1200 million people, or some 60-70 per cent of the total population of the countries represented. FAO has tried to modify FBS to estimate intakes by imposing an apparent distribution on the mean figure; this process leads to an estimated total number of people affected by 'protein energy malnutrition' of just over 400 million (FAO, 1974, 1977).
When FBS data are used to produce figures of this latter sort, what has happened is that the system variable measured by food balance sheets has been turned into an indicator by the application of a cut-off point. This latter is either a minimum level of food consumption or a recommended intake adjusted for such characteristics of the local population as age structure or average body weight (as in FAO's 'country requirement' figures).

Expenditure on food items
Since purchases of food intervene between the points in the food system where FBS and HCS measurements are taken (Figure 4. 1), it might seem that a further way of exploring the operation of this part of the system might be to enquire into family food expenditures. As usual, there are several difficulties to be overcome. Spending on food by low income families as quoted to survey teams is quite often greater than the reported income of the same families; the former is over-estimated and/or the latter is under-reported (Scott and Mathew, 1983). In addition, there is little agreement about how data on family income should be

82 Limits to Measurement
used. Again we are turning a system variable into an indicator by defining some particular low income as a cut-off point, which then becomes a 'poverty line'. But as we saw with the indicators previously discussed, unless the cut-off point is carefully defined in relation to some clear, practical purpose, its use may be misleading.
With indicators derived from anthropometry, we have seen cut-off points can be set at levels at which specified risks of physiological dysfunction begin to rise steeply. An attempt to introduce similar 'objective' criteria into the definition of a poverty line has been made by Lipton (1982). He notes that the poorest people in a wide range of countries commonly spend an irreducible 20 per cent of their income on non-food essentials. He further takes account of the tendency of food requirements figures to over-estimate food needs, especially in relation to 'small but healthy' people living in warm climates, and with Sukhatme's (1961) work in mind, he suggests that it might be more realistic to work with 80 per cent of the conventional requirements figures. This leads him to a 'double 80' rule, according to which people are judged to be 'ultra poor' if 80 per cent of their income is insufficient to purchase 80 per cent of conventionally defined food requirements.
Lipton points out that people living just above this double-80 standard may be to some extent undernourished, and probably experience hunger quite frequently. He calls them 'moderately poor' on the grounds that the risk of physiological impairment .they face is likely to be much less than for the 'ultra-poor'. He further estimates that 'the ultra-poor. form 10-20 per cent of populations' in most low-income countries, and for India finds a startling contrast between urban and rural populations. Calculating on the basis of monthly expenditure per person (MEP), he comments that in 1972-3: '41% of India's rural households - but under 2% of urban - fell into MEP groups averaging food/outlay ratios above 80%' (Lipton, 1982, p.45).
In urban areas, more money is probably needed for non-food items such as fuel, but this does not fully account for the rural/urban difference.

Food system indicators 83

Social valuations
Discussion of poverty, even more than of undernutrition, brings us back to the question of what is acceptable or otherwise to a particular nation or social group. Among several ways in which this question might he tackled, one is for the investigator himself to lay down minimum expenditures on food, clothing, fuel, housing and so on for families of different size, and to consider that those with incomes less than this are living in poverty. This is what Rowntree (1901) did in his pioneer work on poverty in Britain. In using this method, the investigator makes his own valuation of what should be regarded as unacceptable, in the light of what he knows of social norms and physiological needs for food and warmth.
Table 4. 1 summarises a wide range of estimates concerning the extent of poverty in India at three dates for which information from the National Sample Survey (NSS) enables such calculations to be made. The estimates differ considerably, and have been the subject of great and continuing controversy. However, it is important to note that they are not all attempting to measure the same thing. Sukhatme (1981a,b) deals specifically with malnutrition; but Rao (1981), Dandekar (1981) and others include undernutrition within their definitions of poverty.
The table also illustrates some of the many different energy requirements figures that have been used in arriving at poverty lines. These figures are for a standard 'consumer unit', that is, a moderately active 'reference man' weighing 55 kg. Family food needs are estimated from this by counting women and children as fractional consumer units at various levels.
A rather diffe rent way of establishing criteria for the estimation of poverty is to observe how the poor express their own valuations through their behaviour- when purchasing food. Boyd-Orr used this approach in his surveys of food, health and income in the UK during the 1930s (Boyd-Orr, 1937). Lipton's double-80 criterion is based on the empirical observation that in several countries, including India, almost no group of poor people allocates significantly more than 80 per cent of household expenditure to food. The 80 per cent level seems to represent a

84 Limits to Measurement

Table 4.1 Estimates of the extent ofpoverty in India based on National Sample Survey data on consumer expenditure (compare Kumar 1973)

energy intake
used in Cut-off level Percentage of
calculating adopted as the population poverty lines poverty line, or to below the cutAuthors and dates based on define under- off level
of NSSfigures used income nutrition defined
MJ kcal Rural Urban

Bardhan (1970a, b)
1960-1 1967-8
Dandekar and Rath (1971); Dandekar (1981)

1967-8 1971-2
Ojha (1970)
1960-1 1967-8 Rao (1981)
Sukhatme (1981a,b)

FAO (1977)
Lipton (1982)


Income of 11.3 2700 Rs 15 11.3 2700 a

Income of
9.4 2250 Rs 15 (rural)
Rs 22.5 (urban)
9.4 2250 a 9.6 2300 a Income of
9.4C 2250' Rs 18-21
9.4C 2250' a

Income of
9.6' 2300c Rs 21-24 Food intake
of 8.8 MJ
Food intake
of 9.6 MJ
Food intake
= 1.2 x BMR
80% of spending
on food
Breakdown of
Engels's law

a denotes that the same poverty line was used as for 1960-1, but with adjustment for inflation; b refers to percentage below stated energy intake, not below the income poverty line; c based on average intake; d refers to rural and urban populations.

38 53

50 40 50 46.5 d

52 70

20b 25b 17b 21b

41 2 10-30 5-20

Food system indicators 85
critical condition below which people behave as if they have no real options left: they cannot easily economise further on non-food items, nor can they manage with less food. If their income per caput falls further, forcing more cuts in spending, food and non-food items are reduced in the same proportion.
Another way of looking at this is in terms of Engel's law. When comparison is made between better-off and worse-off groups as judged by income per family member, it is found that a progressively larger percentage of family expenditure is used to buy food as one goes down the income scale. The relationship is stated by Engel's law to be a linear one, as Figure 4.3 illustrates for a particular instance. The graph also shows that below the point where 80 per cent of spending is devoted to food, actual expenditures diverge from the sloping line, and the law ceases to be obeyed. Indeed, the point where Engel's law ceases to be applicable might itself be used to define an 'ultra-poverty' line (V.B. Rao, 1981), although this cannot be done with much precision. Using the NSS data for India, Lipton (1982, p.45) concluded that 'for the poorest 10-30 per cent in rural areas and 5-20 per cent in towns, Engel's law does not apply'.
The larger proportion of rural people identified by Lipton as living in ultra-poverty is again significant, especially in view of the higher mortality rates (including infant mortality) recorded in rural areas. Yet according to HCS data, rural food intakes are much the same as urban populations. Noting this, Lipton suggests that for rural people, food energy requirements are greater. There is more heavy work to be done, not only in agriculture but in carrying water and other tasks, and there are more pregnancies. As we observed in Chapter 1, malnutrition occurs at different levels of intake in different environments. By contrast, if a single, fixed food energy requirement is assumed for everybody, poverty and undernutrition seem more prevalent in urban India.
It will be apparent that two sorts of data are used in these various attempts to define criteria for the nutritional aspect of poverty: observations of people's behaviour, and conclusions drawn from scientific models of energy needs. The way these data are combined depends on a variety of assumptions and

90 700

o JJ600 60-011 "o50-C.

ED 0- m

_20- T .o

0 20 40 60 801O0 190 140 160 180
Monthly expenditure per person (Rupees) Figure 4.3 Monthly consumer expenditure per person (MEP) in rural areas of Maharashtra, October 1972-September 1973

Notes: Each vertical bar represents one MEP class. The modal class in the survey, with 1051 households, was that for Rs. 34-43 per person. There were only 74 households spending less than Rs. 13 per person, and only 25 in the Rs. 150-200 class.
Source: NSS data, twenty-seventh round 1972-3, Sarvekshana, January 1979, quoted by Lipton 1982, table 6.

Food system indicators 87
valuations. Conventional nutritional requirement figures, based on observing healthy populations whose intakes are not constrained by poverty, provide one type of behavioural evidence. These data are then frequently used in conjunction with some form of 'genetic potential' model, and in relation to Table 4. 1, lead to the estimates which place 50 per cent and more of the Indian population below various poverty lines.
By contrast, Sukhatme (1981ab) uses a version of the individual adaptability model, with minimum intake fixed and no allowance made for differences in adaptation between rural and urban groups and sub-groups. The proportion of the population below the poverty line is then given as between 20 per cent and 25 per cent (Table 4.1). The same approach was used by FAO (1977): a critical limit of response or adaptation, estimated at 1.2 x BMR (see Chapter 3 above). The authors themselves applied this level to the 1971-2 data from India and estimated that around 17 per cent of rural people should be regarded as suffering from undernutrition at this level. The estimate is very close to Sukhatme's figure.
Lipton (1982) relies much more than the other authors on observed behaviour that is symptomatic of critically low levels of food intake, and by implication gives less weight to any of the scientific models. Assuming that the real extent of adaptability in individuals is reflected in their behaviour, Lipton's results may well come closest to reflecting real limits of adaptability or response in the populations he considers. This makes his conclusion about rural/urban differentials doubly important.
Observing people's food purchasing behaviour falls a long way short of the 'dialogue' between scientists, planners, and consumers of food we earlier envisaged (p.77). But taking account of behavioural evidence is an important way of at least beginning an exploration of social valuations regarding food needs; it points to some of the more extreme limits of what is acceptable in a given society.

Problems of cut-off points: a poverty lines Amartya Sen (1974) has commented that the poor of India'may

88 Limits to Measurement

not be accustomed to receiving much help', but they are 'beginning to get used to being counted'. Knowing that precisely X per cent of India's or the world's population falls below a certain poverty line does not necessarily help alleviate that poverty. Thus the number of 'ultra-poor' people identified by Lipton matters less than the way he is also able to point out concentrations of rural poverty in certain places - on inferior land in Bihar, Assam, Orissa and West Bengal, for example, or among the low-caste and tribal landless in western India. These latter are the kind of facts on which some sort of action might be based. We shall take up these ideas again later.
Scott and Mathew (1983) suggest in a study of poverty in Kerala that the variables relating to poverty, 'should be expressed initially in terms of a distribution', without reference to any particular poverty line, because the choice of a cut-off point is always relative to the socio-economic context. They go on to argue that, 'it may be more helpful to administrators or local village committees whose job it is to provide relief or ameliorate conditions, to decide on their own critical points . in the light of . resources and needs, and with full knowledge, rather than the half knowledge provided by a poverty line'.
This is an approach to be commended wherever it can be adopted, because once cut-off points have been defined and found to be useful, they are often taken over and used out of context by other investigators. Poverty lines based on income, for instance, are not necessarily accurate indicators of nutritional risk. Rigidity in the use of an indicator implies that no dialogue is taking place about what cut-off points might best represent 'unacceptable' conditions. Furthermore, in a country in which development policies are achieving at least some of their goals, we would expect the borderline between acceptable and unacceptable risk to be raised continuously. Poverty lines should then be adjusted from year to year. Actually the poverty line used by the Indian Planning Commission is based on 1961-2 real prices. And another example of what ought not to happen is provided by Woolf (1946) and Walker and Church (1978): Rowntree's 'absolute minimum for survival' poverty line continued to be used in Britain for a period of over forty years and

Food system indicators 89
indeed is still the basis of state social security and welfare intervention. One danger with Lipton's double-80 criterion is that its evident usefulness and apparent objectivity may give it a similar permanence.

Problems of cut off points: b anthropometry
Similar considerations apply to the cut-off points used in connection with anthropomnetric data. Gomez et al. (1956) used hospital mortality results to define a cut-off point at 60 per cent of standard weight-for-age, and described children below this level as suffering from 'severe' or 'third degree' malnutrition. This cut-off point and these terms have been used by many subsequent authors, and have become part of the vocabulary of nutrition; they are summarised in Table 4.2 along with similar classifications used with other anthropometric indicators.
The use of Harvard standards for heights and weights was mentioned in Chapter 2. The standards were originally derived from measurements of several thousand American children, and are presented as sets of age-standardised centiles. These should be regarded as probability statements. The 90th centile for any measurement is a set of numerical values below which we would expect to find 90 per cent of a sample, with only 10 per cent above. It thus represents a condition well above average, and in a set of data representing body weights, the 90th gentile would correspond to about 115 per cent of standard weight-for-age.
At the other end of the scale, the 1st centile might be used as a cut-off point in weight-for-age to identify children who are at risk in some way. There is a probability of only 0.01 that a healthy individual would have this weight, so if we found that all the individuals in a sample were on or below the 1st centile, we would have good reason to think that not all of them were 'well nourished'. In the Harvard distribution, body weights of individuals in the 1st centile are near 60 per cent of those of individuals in the 50th centile - that is. they are about 60 per cent of 'standard'. And as we have said, 60 per cent weight-for-age is very widely taken as the cut-off point identifying individual's who

90 Limits to Measurement

Table 4.2 Panel showing examples of Harvard Standards for anthropometric data and illustrating some classifications commonly used.

(a) Harvard 50th centile weights for young children of both sexes at representative ages.
Age/years 0 0.5 1 2 3 4 5
Weight/kg 3.4 7.4 9.9 12.4 14.5 16.5 18.4

(b) Interpretation of weights below Harvard Standard 50th centile Weight as % of Gomez
Harvard standard classification
> 90 normal includes individuals in the
f 25th centile and above
75-90 . undernutrition
(1st degree)

60-75 (2nd degree)

<60 severe f includes only individuals
(3rd degree) lin the 1st centile

have a low probability of being 'healthy', and therefore a high probability of 'severe malnutrition' (Table 4.2).
These cut-off points are given because they are so widely used, but we also need to recall that their significance is always relative to our circumstances, resources and purpose (Chapter 2). In screening a particular population, for example, we might have resources to treat all those on the 10th centile and below. This means that we would use 85 per cent weight-for-age as a cut-off, regardless of how it relates to the Gomez classification.
It is sometimes possible to attach greater significance to a cut-off point if it defines conditions where some particular risk begins to increase steeply. For example, in the group of children studied by Chen et al. (1980) in Bangladesh mentioned in Chapter 2, the numbers of deaths occurring in the two years after anthropometric data were recorded has the distribution relative

Food system indicators 91

Harvard 50th centile heights or lengths for young children of both sexes at representative ages.
Age/years 0 0.5 1 2 3 4 5
Length/cm 50.4 65.8 74.7 87.1 96.0 103.3 109.0

The Waterlow classification (Waterlow 1972, 1976) combines two indicators, weight-for-height, and height-for-age. Children of low weight-for-height are referred to as 'wasted'; those of low height-forage are termed 'stunted'.
More than 80% Less than 80%
Harvard Standard Harvard Standard Height-for-age
More than 90% classed as classed as WASTED
Harvard Standard NORMAL

Less than 90% classed as highest risk category:

to weight-for-age which is shown in Figure 4.4. There is an 'elbow' in the curve where mortality risk begins to rise rapidly, and this break in the curve is located at values of weight-for-age around 60-65 per cent of Harvard standard. Bairagi (1982) quotes similar data from a group of Indian children studied by Kielman and McCord (1978), and discusses statistical considerations to locate the 'best cut-off point' more precisely on the elbow of the curve. As Figure 4.4 shows, the point is not the same for the two communities. Moreover, the Narangwal study also showed that the shape of the curve depended on the age of children measured: risk of death for a given nutritional status decreased with age. The possibilities of using anthropometric measurements as indicators and the statistical, rather than value issues involved are discussed further by Habicht etal. (1982), and

92 Limits to Measurement


801 *---*- Bangladesh
0--0 Narangwal
o \



20-------55 65 75 85
Percentage of Harvard standard weight for age

Figure 4.4 Variation of death rate vs weight-for-age for two groups of children: Bangladesh (children aged 12-24 months), Narangwal, India (children aged 0-36 months) Notes: The vertical lines at 63 and 73 per cent weight-for-age represent Bairagi's view of the 'best cut-off points' for nutritional monitoring. Sources: Bangladesh, Chen et al. 1980; Narangwal, Kielmann and McCord 1978; diagram after Bairagi 1982.
also by Trowbridge and Staehling (1980) and Eusebio and Nube (1981).
Another question that arises is whether standards derived in one country can be used to evaluate individuals in another. Some studies indicate that there are small differences in growth pattern between Africans, Asians and Caucasians, although these only appear after 5 years of age (e.g. Hiernaux, 1964; Jelliffe, 1966; Habicht et al., 1974). We may ask whether, even if all human groups have basically the same genetic growth potential, national standards should not still be used because they take account of environmental and economic differences.
Much discussion has been given to these questions, but to answer them we need to come back to a basic point. The actual value of weight-for-age at which children's mortality rates increase sharply can be called X kg. On the Harvard standard, X is 60 per cent weight-for-age. Any standards could be used to

Food system indicators 93
define 'malnutrition'; cut-off points may be chosen according to measured risks, or to resources available for allocating treatments, or to suit some other purpose. If such procedures are consistently followed, it does not matter very much what standard values are used, as long as they are derived from a large sample, with acceptable methods of measurement.
In any case, as we saw in Chapter 2, there is no unambiguous way of fixing points on the scale so that 'high risk' are sharply divided from 'normal' individuals. For any fixed point, there will be some healthy individuals who fall below it, so in a practical situation where action is contemplated, there will be some healthy people who are treated as if they are sick, or at risk of death.
Although it makes sense to use the 'elbow' of the risk curve as a cut-off point, we should adopt this procedure with caution, for two reasons. Firstly, to apply a statistical test in order to establish the best cut-off point may pre-empt the dialogue we have mentioned as necessary in making social valuations. Risks are not zero for children who are better than 60-65 per cent weight-for-age, and we cannot be sure that the 'elbow' of the curve coincides with the view of these risks taken either by planners, or by parents of the children concerned. It is also often the case, however, that those whose children fall into the extreme, worst category are excluded from any intervention or any dialogue at all. Where there is an option of spending more money in order to provide treatments (or food) for extra children, those who provide the funds might also have a view - they might wish the treatments to be given even though the children who benefit are in low-risk category.
Secondly, we should note that a cut-off point defined by reference to mortality risk in one part of the world may not be applicable to morbidity risks in the same population groups, nor applicable at all in any useful way to children in a different region or continent (Kasongo Project Team, 1983). Investigating the incidence and prevalence of diarrhoea among children in Nigeria, Tomkins (1981) found that the freqUency of diarrhoea was no greater for children of low weight-for-age than for others, but that such children took significantly longer to recover from each

94 Limits to Measurement

attack. Weight-for-height proved to be more effective as an indicator of the risk of an attack. Among children who were below 80 per cent standard weight-for-height, episodes of diarrhoea were 1.47 times more likely than among those above this level. Chen et al. (1981c) found similar results. Reddy et al. (1976). and Cunningham-Rundles (1982) among others, have looked at the association between resistance to infection and nutritional status.
In these circumstances there are no clear-cut answers about which indicator to use, and with what cut-off points. Parents might be worried about the high incidence of some particular illness, while educationists point out that slow recovery from another disease is holding back children's learning. Different indicators and fixed points would be needed to respond to these different concerns. Once again, therefore, several points of view must be considered before social valuations of risk can be fully appreciated. And as in other instances we have considered, the deciding factors about cut-off points must include dialogue concerning the relative importance of risks of different aspects of malnutrition, and a clear idea about the purpose for which the indicator is being used and the resources available.


The Causes of


'It. places the emphasis on man, and endeavours to study him in, and in relation to, his environment. economic, nutritional, occupational.' John A. Ryle, on epidemiology as an aspect of 'social medicine' (see Ryle, 1948, p.1H)

5 Multiple causes in


A focus on people
The direction in which previous chapters have pointed is towards the need for a comparative understanding of malnutrition as it occurs in different sectors of society, and as it is affected by geographic, environmental and disease factors. We approached this conclusion by a negative route, noting the limitations of more conventional approaches. Measurements of average food consumption do not tell us about social patterns of underconsumption; requirements figures do not in practice tell us about which food intakes are likely to lead to malnutrition. Taking a more constructive approach, we may now say that our goal in studying the social distribution of malnutrition is to understand how people become malnourished in terms of processes within society. We may then hope to find out who these people are and what policies or changes in society might help them (Joy and Payne, 1975; Payne, 1982).
Two major themes in any such study must be the distribution of malnutrition in the population, discussed below in Chapter 6, and causality, dealt with here. An example of limited but pioneer

*This chapter is based on material prepared by Peter Cutler, David Nabarro and Philip Payne; see especially Cutler (1982b) and Nabarro (1982). We are also indebted to a teaching text on epidemiology by Jones et al. (1982) for one way of presenting the concept of multiple causes.

98 The Causes of Malnutrition
work on both aspects was carried out in the UK in the 1930s, under the banner of 'social medicine'. In one study, begun in 1932, James Spence studied malnutrition among children in the northern British city of Newcastle-upon-Tyne. His work 'amongst the poorest classes' led him to think that three factors were interacting: infectious disease, the effects of poor housing, and inadequate diet. He described the symptoms he saw in the children as 'apparent malnutrition' to express his view that inadequate food was not always the prime cause (Spence, 1960).
The relative importance of infection, environment and food in causing the problems with which we are concerned has been a continuing subject for debate. For example, Gopalan (1983) accepts that environmental factors often influence the growth of children in ways that lead to malnutrition, but argues that: 'There is no way by which environment can influence growth and development except by conditioning and modifying the availability of essential nutrients. at the cellular level.' Thus retarded growth: 'is a reflection of under-nutrition and of nothing else' (Gopalan's italics). However, poor housing can have an effect by inducing 'stress situations, like exposure to cold or infections which condition the net availability of nutrients at the cellular level'.
In his work among children in Newcastle, Spence recognised the same issue. The aim should be to study, 'how disease and social circumstances are related', and to work at a level that 'places the emphasis on man, and endeavours to study him in, and in relation to his environment . economic, nutritional, occupational' (Spence, 1960; quoting Ryle, 1948).
Our argument is that we chiefly require better understanding of these factors at a level accessible to policy interventions such as improvements in housing or more ample food supplies. Indeed, we might need to know which of these options is likely to give more benefit to a particular social group, and at what cost.
This is not an easy set of issues to pursue, however. The environmental circumstances of people's lives are very varied, and a causal analysis of nutritional problems could rapidly become an excessively cumbersome analysis of the human condition (Payne, 1976). One strategy adopted in the past to

Multiple causes in malnutrition 99

avoid such complexities was to identify those few symptoms or combinations of symptoms for which simple nutritional explanations do seem to work. During the period from 1930 to 1950, nutritionists following this procedure were successful in identifying several vitamin deficiencies, such as pellagra (niacin deficiency) among impoverished people living mainly on maize in the United States. They also identified thiamin deficiency among polished-rice eaters in Japan, and inadequate vitamin A in the diets of low-income groups in Newfoundland. These investigations were all followed by fortification projects, and over subsequent years, deficiency symptoms in the populations declined. A careful historical analysis shows that in all these situations, however, the decline was as much associated with rapid social and economic improvements as it was with the nutritional intervention: indeed- in all cases the decline was already well under way before the programmes were started. However, these interventions established a pattern for public health nutrition in the 1950s, and it was against this background of apparent success that the disease of small children called kwashiorkor gradually came to be accepted as due to protein deficiency alone. We saw in Chapter 3 that more recent research has largely disposed of this view.
Some diseases which seem explicable in terms of nutrient deficiencies are distressingly widespread. The documentation for India's sixth five-year plan quotes lack of vitamin A in diets as the cause of widespread xerophthalmia, leading to blindness in the most severe cases (Government of India, 1981). Nutritional anaemia and goitre are often cited as the two other most prevalent deficiency diseases affecting the poor in many developing countries (FAO, 1977).
Nothing that is said in this book should be taken as minimising the significance of these forms of malnutrition. Our argument is simply that the model of single dietary deficiency causes used to understand them is of limited application. It is based on an epidemiological approach, certainly, but one which lacks adequate means of analysing the multiple causes of disease. For example, according to FAO (1977), nutritional anaemia affects at least 20 per cent of children in developing countries and even

100 The Causes of Malnutrition
more adult women; among men the prevalence could be around 10 per cent. This does not necessarily mean their diets are deficient in iron however. One study of labourers employed on road works in north India, for example, found that although there was no correlation between work output and energy intake, reduced levels of output observed for some labourers were significantly correlated with anaemia (Tandon et al., 1975). Surprisingly, perhaps, the diet eaten by these men included 'generous' helpings of lentils, and supplied more than three times the recommended daily allowance of iron. As with many other people eating largely vegetarian diets, the problem seemed to be one of malabsorption of iron rather than inadequate intake. In many social groups there is also a heavy burden of parasitic disease due mainly to poor sanitation, and this gives rise to iron loss. Thus even a superficial account of one type of anaemia leads us to notice other causes apart from deficient food intakes. (For example, see M. C. Latham et al., 1983.
With other illnesses that are conventionally attributed to poor nutrition, causal explanations become even more complex. Referring again to kwashiorkor in young children, it is becoming increasingly clear that this is not a problem of low food intake alone (Whitehead, 1977; Lunn et al., 1979; Martorell et a!., 1980). Modern research has come to focus on diarrhoeal infections as a precipitating factor, and there is a danger that those who think in terms of single causes will now rather easily fall into the habit of speaking about infection as 'the cause' of malnutrition. However, the precise mechanisms by which diarrhoea leads to this type of malnutrition are uncertain. Also while some investigators (Rutishauser, 1974; Mata et al., 1977; Tomkins 1983) suggest that anorexia is the main cause, others stress that, during diarrhoea, there is malabsorption of nutrients due to abnormality in gut flora and function (Gracey et al., 1977; Rowland, 1981); yet others (Tomkins et al., 1983) have shown that the fever associated with infections raises the body's requirement for dietary energy.
In examining the nutritional problems of young children, we first have to remember that their energy needs are proportionately higher than those of adults. When children are breast-fed, mothers produce adequate amounts of milk to

Full Text
xml record header identifier 2009-03-11setSpec [UFDC_OAI_SET]metadata oai_dc:dc xmlns:oai_dc http:www.openarchives.orgOAI2.0oai_dc xmlns:dc http:purl.orgdcelements1.1 xmlns:xsi http:www.w3.org2001XMLSchema-instance xsi:schemaLocation http:www.openarchives.orgOAI2.0oai_dc.xsd dc:title Agricultural development and nutrition dc:creator Pacey, ArnoldPayne, Philipdc:subject Food supply -- Developing countries ( lcsh )Nutrition -- Developing countries ( lcsh )dc:description b Bibliography Includes bibliography and indexes.Statement of Responsibility edited by Arnold Pacey and Philip Payne.Published by arrangement with the Food and Agriculture Organization of the United Nations and the United Nations Children's Fund.dc:publisher Hutchinson ;HutchinsonWestview Pressdc:type Bookdc:format 255 p. : ill. ; 23 cm.dc:identifier (OCLC) 85050544 (LCCN)081330265X (US) (ISBN)dc:source University of Floridadc:language Englishdc:coverage India