Previous Page Table of Contents Next Page

Keynote Paper: Measures of nutritional status from anthropometric survey data

Prakash Shetty
Rome, Italy

Executive summary

FAO is mandated to provide reliable figures of the true extent of the problem of undernutrition to help Member Nations in monitoring trends, determining priorities and evaluating the effectiveness of intervention programmes. In order to do that, there is a need to detect undernutrition in individuals and to assess the severity of the problem in the community. This paper evaluates the use of nutritional anthropometric measures to estimate the numbers of undernourished while highlighting the advantages and limitations of nutritional anthropometric approaches. It addresses issues related to reference values and discusses cut- offs based on the relationship between nutritional anthropometric indices and functional impairment, morbidity and/or other evidence of the consequences of food inadequacy. It also attempts to relate this approach with the other approaches discussed at this Symposium. Nutritional anthropometric measurements, indices and indicators are defined, and the commonly used indicators for the diagnosis of undernutrition throughout the life cycle (i.e. infants, children, adolescents, adults and elderly) are discussed. The validity, reliability and uses of these anthropometric indicators in different situations in the field and communities are also discussed. The paper emphasizes the need to carry out properly sampled, representative surveys to aid this process and highlights the role these nutritional indicators can play in assessing the impact of the developmental process, the effect of nutritional and other interventions, and the consequences of adverse situations such as food emergencies resulting from conflict, natural disasters and economic downturns.

Given the recent controversies related to differing estimates of national and regional numbers of undernourished derived from the FAO food balance method and anthropometric indicators of children and adults, an attempt is made to look at two specific instances where comparative analyses have been carried out. The first is a comparison of data based on these two approaches in nationally representative samples in Brazil. In this case study, correspondence plots and related analyses support the conclusion that both methodological approaches have merit and are likely to provide complementary information. Food intake surveys are more difficult and expensive to carry out on a nationally representative basis at periodic intervals and are plagued with difficulties related to obtaining individual data from household information. Anthropometric data, however, do not necessarily reflect food consumption or energy adequacy per se as they are influenced by other environmental determinants of nutritional status, such as infections. The latter opinion is supported by critical comparative analyses of data derived from these two approaches in several developing countries. These analyses indicate a lack of correlation between the estimates of undernutrition in children and adults when comparing anthropometric data with measures of adequacy of dietary energy supply. It would appear, however, that even though the two approaches reflect different determinants, they do provide complementary information. Should simple, reliable objective anthropometric indicators be used more widely in national surveys, continuity of data collection, projection of trends and long-term forecasts of food needs could be made based on the relationship between these two approaches. Further, anthropometric data in adults are not currently available on a global or regional basis to compile meaningful and representative databases, unlike the data available using the currently well -established FAO methodology, which provides a continuous source of global data, despite its limitations.


FAO has long recognized that reliable figures of the true extent of the problem of undernutrition are essential if Member Nations are to attempt to do away with undernutrition in their midst. Availability of relevant and accurate information will be helpful in monitoring trends, determining priorities and evaluating the effectiveness of intervention programmes. In order to arrive at reliable estimates of the numbers of undernourished individuals and the prevalence of undernutrition in countries worldwide, there is a need to decide on how we detect the existence of undernutrition in an individual and assess the severity of the problem of undernutrition within the community. This is not an easy task, as considerable time and effort can, and has been, spent on deciding the validity of the assumptions made and the methodologies and the diagnostic criteria adopted. These include reference values and cutoffs approved for general use among population groups worldwide, in order to define food needs and to estimate the extent of the problem of food inadequacy in different regions of the world.

One of the approaches long relied upon by FAO is based on the determination of the number of individuals with a high probability of being at risk of food inadequacy in the population. This top -down approach attempts to disaggregate the data based on food balance sheets and the demographic composition of the population to derive per capita intakes, which are then used to determine the likely numbers at risk of food inadequacy. An alternate approach could be a bottom-up approach, which uses individual nutritional status assessment data from nationally representative samples to estimate numbers likely to be at risk of undernutrition. The data generated by these two methods are not expected to be similar but are complementary to each other. The proportions and numbers of individuals considered as "food inadequate "globally or regionally by the former method would not be identical to the proportions and numbers considered to be "undernourished "by the latter method as the two approaches measure different things in quite different ways. This paper focuses on a critical evaluation of the estimation of numbers of undernourished assessed by nutritional anthropometry, a bottom-up approach.

The aim of this paper thus will be to evaluate objectively and critically the assessment of nutritional status in the community and to highlight the advantages and limitations of nutritional anthropometry in the assessment of numbers of undernourished individuals in communities, regions, countries and globally. It will also address issues related to reference values and cutoffs based on the relationship between nutritional anthropometric indices and functional impairment, morbidity and/or other evidence of the consequences of food inadequacy. It will attempt to link the approaches described in the other keynote papers for this Symposium with the nutritional anthropometric approach for inferring the prevalence of food inadequacy in the world. The paper will also emphasize the need to carry out properly sampled, representative surveys to aid this process and will highlight the role of nutritional indicators in assessing the impact of the developmental process, the effect of nutritional and other interventions, and the consequences of adverse situations, such as conflict, natural disasters and economic downturns.

Malnutrition and undernutrition: definitions and the need to assess the extent of the problem

Malnutrition refers to all deviations from adequate nutrition, including undernutrition (and overnutrition) resulting from inadequacy of food (or excess of food) relative to need (respectively). Malnutrition also encompasses specific deficiencies (or excesses) of essential nutrients such as vitamins and minerals. Conditions such as obesity, although not the result of inadequacy of food, also constitute malnutrition. The terms "malnutrition" and "undernutrition" are often used loosely and interchangeably, although a distinction is, and needs to be, made at all times.

"Malnutrition" arises from deficiencies of specific nutrients or from diets based on wrong kinds or proportions of foods. Goitre, scurvy, anaemia and xerophthalmia are forms of malnutrition caused by inadequate intake of iodine, vitamin C, iron and vitamin A respectively.

"Undernutrition" is the outcome of insufficient food of whatever kind caused primarily by an inadequate intake of dietary or food energy, whether or not any specific nutrient deficiency, such as iron deficiency anaemia, is present. Undernutrition is defined as a dietary energy intake below the minimum requirement level to maintain the balance between actual energy intake and acceptable levels of energy expenditure. This must take into account additional needs for growth in children and also for pregnant and lactating women to maintain appropriate weight gain associated with adequate foetal growth in pregnancy and to sustain sufficient milk production during lactation (FAO/WHO/UNU, 1985). This emphasis on dietary energy as a general measurement of food adequacy seems pragmatically justified: increased dietary energy, if derived from normal staple foods, brings with it more protein and other nutrients, while raising intakes of such nutrients without providing more dietary energy is unlikely to be of much benefit to the individual. Thus, in most situations, increased dietary energy is a necessary condition for nutritional improvement, even if it is not always sufficient in itself. However, it is important to recognize that undernutrition, when estimated as an outcome measure, is also the result of other environmental insults, such as infections and poor care, both of which contribute to this process alongside inadequate food intakes.


FAO makes a distinction between undernutrition and undernourishment. "Undernourishment" is when food intake is continuously insufficient to meet the dietary energy requirements, while undernutrition is the result or outcome of undernourishment, poor absorption and/or poor biological use of nutrients consumed (FAO, 1999). This distinction may be important when attempting to explain the differences in estimates of numbers of undernourished individuals by the FAO food balance method as compared with the nutritional anthropometric approaches that provide numbers of undernourished, as for instance reported in the WHO Global databases.

"Malnutrition" and "undernutrition" are terms generally used interchangeably to mean more or less the same entity, and they both refer to nutritional situations characteristic of populations belonging to the low-income and poor socio-economic groups of developing countries. In practice, developing-country population groups suffering from malnutrition or undernutrition as defined in this way are likely to be more or less the same. Although it is possible to arrive at the prevalence (percent) and the numbers of individuals within a population manifesting signs of specific nutrient deficiency, for instance anaemia as a result of iron deficiency, when signs of vitamin or mineral deficiencies are observed, they are almost always associated with marginal or low dietary energy intakes.

For the purposes of this paper, the term "undernutrition" is used in the broader sense, referring to any physical condition implying ill health or the inability to maintain adequate growth, appropriate body weight and body composition or to sustain acceptable levels of economically necessary and socially desirable physical activities brought about by an inadequacy in food, both in quantity and in quality. This definition thus includes both undernutrition and specific micronutrient deficiencies.

Assessment of undernutrition in the community

There are two possible ways to assess the adequacy of food and nutrition and to detect the presence of inadequacy in food intake among individuals or population groups: the first measures nutritional intake, and the second assesses nutritional status.

Nutritional intake assessment

Measures of nutritional intake estimate the amount of food a person is eating and can be used to assess adequacy of the quantity of dietary energy (and protein) supply. In simple terms, one can categorize people as being well -nourished or undernourished based on whether their intake of food matches their food energy needs or nutrient requirements. The methodologies that provide such information are those based on national sample surveys or dietary surveys that attempt to measure the food consumption or intake levels of representative individuals within a population, as discussed by Ferro - Luzzi in this series. These methods of ten tend to provide an estimate of the risk of the population or individual to inadequacy of food but do not help to identify actual individuals in the population who are deficient; nor do they help define the degree of severity of the food inadequacy.

Nutritional status assessment

The second option assesses the nutritional status of the individual or a representative sample of individuals within a population by measuring anthropometric, biochemical or physiological (functional) characteristics to determine whether the individual is wellnourished or undernourished. This method makes use of objective, measurable criteria that reflect the changes in anthropometric, biochemical or functional characteristics of the individual as a consequence of inadequate intakes of food for long periods of time, or as a result of seasonal fluctuations in intakes of food or poor absorption and utilization of ingested food. A hierarchical model of the causes of undernutrition emphasizes the importance of repeated infectious episodes and poor care and neglect as determinants of undernutrition, in addition to the lack of adequate food (UNICEF, 1998).

Anthropometry is the most frequently used method to assess the nutritional status of individuals or population groups. Measurements of nutritional anthropometry are based on growth in children and body weight changes in adults.

Nutritional anthropometry

Nutritional anthropometry has been defined as "measurements of the variations of the physical dimensions and the gross composition of the human body at different age levels and degrees of nutrition "(Jelliffe, 1966). Anthropometric measurements are of two types, growth and body composition, and have been widely used for the assessment of the nutritional status of both children and adults. The selection of the ideal single or a combined use of anthropometric indicators depends upon the sensitivity and specificity of the indicator chosen. Sensitive indices are ideally suited for nutritional status assessment in screening or surveillance activities as they are capable of detecting even small changes that occur in nutritional status during conditions of food inadequacy.

Nutritional anthropometry has several advantages (Gibson, 1990), which are summarized in Table 1. There are also, however, several limitations that should be mentioned (Table 2). Nutritional anthropometric techniques are prone to measurement and other types of errors - both systematic and random. These errors arise out of inadequate and improper training of personnel, difficulties in measurement of certain anthropometric characteristics such as skinfolds, and instrumental or technical errors. These errors can be minimized by proper training of personnel to use standardized, validated techniques and by frequent calibration of instruments, thus improving the accuracy and precision of the measurement. WHO (1983) has provided guidelines for standardization of procedures and for the calculation of inter-and intra-observer variations in nutritional anthropometric measurements in children. These limitations to anthropometry should not be overemphasized, however, because all approaches to assess numbers of undernourished, particularly those at the household level, are beset with such problems. In addition to the limitations included in Table 2, anthropometric data may be unrepresentative of the community if care is not taken in the proper selection of the sample. This topic will be covered in more detail later in this paper.

Infants and children

In infants and children under five years of age, assessment of growth has been the single most important measurement that best defines their nutritional status. Disturbances in nutrition as a result of inadequacy of food intake, severe and repeated infections or a combination of both, operating very often as a vicious spiral, invariably affect the growth of a child. These adverse conditions are closely linked to the general standard of living and the population's ability to meet its basic needs for nutritious food, safe water, good housing, acceptable levels of environmental sanitation, and ready and easy access to health care. Assessment of the nutritional status of the child by the use of nutritional anthropometric indicators of growth has thus been used not only to provide information on the nutritional and health status of children but also as an indirect measure of the quality of life of the entire community or population, and thereby as an indicator of the nutritional status and adequacy of food of all members of that community. There are now, however, growing doubts whether estimates of undernutrition based on nutritional anthropometric survey data in children alone (without any information about the adults in the community) necessarily reflect the overall nutritional status and the adequacy of food availability within the entire community (FAO, 1994).


Methods are precise and accurate, provided standardized techniques are used

Procedures use simple, safe and non-invasive techniques

Equipment required is inexpensive, portable and durable, and can be made or purchased locally

Relatively unskilled personnel can perform measurement procedures

Information is generated on past nutritional history

Methods can be used to quantify the degree of undernutrition (or overnutrition) and provide a continuum of assessment from under-to overnutrition

Methods are suitable for large sample sizes such as representative population samples

Methods can be used to monitor and evaluate changes in nutritional status over time, seasons, generations, etc.

Methods can be adopted to develop screening tests in situations such as nutrition emergencies to identify those at high risk

Source: Gibson (1990).


The relative insensitivity to detect changes in nutritional status following inadequacy of food over short periods of time

The inability to distinguish the effect of specific nutrient deficiencies (e.g. zinc deficiency) that affect growth in children from that due to inadequacy of food in general

The inability to pinpoint the principal causality of undernutrition, as the poor nutritional status may be the result of factors such as repeated insults owing to infections and poor care in children

The relative higher costs and organization required to obtain representative and quality data for the purpose of estimating numbers of undernourished

Height and weight are the most commonly used indicators of the nutritional status of a child. According to a WHO Working Group (1986), appropriate height-for-age of a child reflects linear growth and can measure long- term growth faltering or stunting, while appropriate weight-for-height reflects proper body proportion or the harmony of growth. Weight-for-height is particularly sensitive to acute growth disturbances and is useful to detect the presence of wasting. Weight-for-age represents a convenient synthesis of both linear growth and body proportion and thus can be used for the diagnosis of underweight children. The presence of undernutrition in children is assessed using these three anthropometric parameters (weight-for-age, height-for-age and weight-for-height) and by comparing them with internationally accepted reference standards, i.e. National Center for Health Statistics (NCHS) /WHO international reference population (WHO, 1983). If a child has a low height-for-age, i.e. a Z-score below two standard deviations of the reference population mean (-2 Z- score), such a child is categorized as "stunted ". Similarly, a low weight-for-age is diagnostic of an "underweight" child, while a low weight- for-height is indicative of "wasting ". In a population, the growth retardation prevalence for the under-five-year-old preschool child is estimated by the proportion of weight-for-age, height-for-age and weight-for-height below - 2 Z-score, the accepted cutoff for the diagnosis of undernutrition and hence indicative of an increase in risk of morbidity and mortality (WHO, 1995). The use of nutritional anthropometry also permits stratification of survey results according to age, sex, region, rural /urban or other sociodemographic characteristics of the population, thus providing more information for detecting vulnerable groups and for better understanding the situation. An additional use is the estimate the prevalence of overweight children owing to overnutrition, as measured by the proportion of children with weight-for-height above +2 Z-score).

There is sufficiently good evidence to show that poor growth and smaller size in schoolchildren is associated with impaired development, and a number of studies have demonstrated a relationship between growth status and school performance or intelligence (Grantham - McGregor et al., 1991). However, this cannot be regarded as a simple causal relationship between inadequacy of food, physical growth status and intellectual development. Complex environmental, sociocultural and economic factors affect both growth and development. Although considerable efforts have always been made to obtain the prevalence of childhood undernutrition among infants and under-five -year-olds from representative national samples, there is limited information on the over-five -year-old schoolchild. Anthropometric indicators that are being used in older children to assess nutritional status are the same as those being used for preschool children, i.e. weight-for- age, height-for-age and weight-for-height. However, the same anthropometric indicator provides different information at different ages of children in a community. For instance, a high prevalence of low height-for- age among one -year-old children indicates current nutrition and health problems in a population, i.e. the problems of stunting. In children five years or older, it reflects a past problem in those already stunted but also may indicate that there is active, concurrent stunting among the younger children in the population. Because nutritional anthropometry is a non-specific indicator of multiple past and current processes in older children, proper interpretation requires the use of additional data related to food and diet, socio - economic status, the prevalence of infections, the result of poor care and the presence of adverse environmental circumstances.

In 1993, a WHO Expert Committee drew attention to a number of serious technical and biological problems with the NCHS/WHO growth reference that had been recommended widely for international use since the 1970s. With increasing doubts about the universal applicability of these North American references, WHO has now launched a major enterprise to develop truly international and representative references for growth (WHO, 1999). The WHO global database on malnutrition has adopted certain criteria for inclusion of representative data from surveys world - wide. The criteria for selection of surveys for the database include a clearly defined population-based sampling frame, a probabilistic sampling procedure involving at least 400 children, the use of appropriate equipment and standard measurement techniques as well as the ability to present the data as Z- scores in relation to the NCHS/WHO reference population (de Onis et al., 1993).


Adolescents comprise a significant proportion of the world's population; some estimates put the number of youth at over 30 percent of the world population. The proportion of adolescents within a population group is also rising relative to other age groups, and an overwhelming proportion of young adolescents live in developing countries. Increases in height as well as weight occur during this period. About 25 percent of an individual's attained height is achieved during adolescence as a result of the adolescent growth spurt that marks the end of the growth in height. Variations in adolescent body size and the timing of maturational events are determined genetically in populations whose environment allows full expression of the genotype. Where this is limited by environmental constraints, including nutrition, the observed growth and maturation during adolescence reflect environmental rather than inherited potential. It is now clear that growth differences among groups are also related to nutritional status, socio -economic and other factors.

Growth in adolescence may be limited by prolonged undernutrition, infections and chronic disease. Stunting or short stature in adolescence is not only indicative of past undernutrition during childhood but also may be a cumulative indicator of nutritional status during adolescence. Stunting among adolescents reflects increased health risks, particularly among females who would also tend to have a small pelvis, leading possibly to obstructed labour during childbirth. Gains in weight are also considerable during the adolescent years with increases in both muscle and fat. Girls tend to gain relatively more fat, while boys gain relatively more muscle. Undernutrition in girls during adolescence is characterized by a low weight, which may result in poor pregnancy outcomes, particularly low birth weight. Undernutrition also may limit school achievement and work productivity in later years. There is emerging evidence that stunted individuals are at increased risk of overweight and obesity when food availability increases and lifestyles change (Popkin, Richards and Montiero, 1996).

The diagnostic criterion for defining stunting in adolescents is a height-for-age less than the third percentile of the NCHS/WHO reference data or less than -2 Z-score. Undernutrition or thinness in adolescence is indicated by a body mass index (BMI, weight/height 2) less than the fifth percentile of the NCHS/WHO reference data (WHO, 1995). A BMI greater than the 85th percentile in adolescence is indicative of a risk of overweight.


The lack of a true definition for the assessment of adult undernutrition is due to the difficulty in establishing satisfactory reference standards for normality and in delineating cutoffs to help distinguish between well -nourished and undernourished adults in population groups. A chronic state of undernutrition in the adult has its cost in terms of risk to health and impairment of function, which may include a lowered work capacity, a reduced ability to sustain economically productive work or even socially desirable physical activities and possibly an impaired immune function with a predisposition to repeated infections (FAO, 1994).

BMI is considered to be the most suitable, objective anthropometric indicator of nutritional status of the adult. It was chosen because this anthropometric indicator, derived from measures of weight and height of individuals of both sexes, is consistently and highly correlated with body weight (or energy stores within the body) and is relatively independent of the height of the adult. While a BMI <18.5 is considered as the cutoff for the diagnosis of chronic undernutrition in adults, a series of cut- offs are provided to delineate the degrees of severity of undernutrition (James, Ferro-Luzzi and Waterlow, 1988). The lower limit of normality is based on the BMI of patients with anorexia nervosa and a large sample of healthy, young British soldiers. Concerns that lean but healthy and very active adults may be wrongly categorized or misclassified as undernourished lead initially to the inclusion of energy turnover based on basal metabolic rate as an additional criterion. However, BMI alone is now accepted as the anthropometric indicator of choice for chronic undernutrition in the adult, as the probability of misclassifying nutritional status on the basis of the BMI is considered to be very small. This indicator has similar advantages over weight-for-height in children in that it reflects the degree of severity of undernutrition and also can be used to assess overnutrition in adults, by enabling the classification of over- weight and obese individuals in a population (FAO, 1994).

BMI is thus a simple but objective anthropometric indicator of the nutritional status of the adult population and is closely related to food consumption and the prevalence of inadequacy of food in the community. Data on BMI are relatively easy to collect and inexpensive to analyse. Collection of data on heights and weights of adults from which BMI is easily derived can be readily incorporated into regional and national surveys being conducted. BMI can be used for the purpose of nutritional surveillance and for monitoring the effectiveness of intervention programmes, and it also allows for interregional and intercountry comparisons over seasons, years or decades (FAO, 1994).


Adults 60 years of age and older represent the fastest-growing segment of the population throughout the world. Decline in height with age is well documented in the elderly, and reduction in weight also occurs with increasing age, although the pattern of change in weight is quite different from that of height and varies with the sex of the individual. The use of anthropometry is relatively recent in the elderly, and the anthropometric index of choice is the BMI, as in the case of non-elderly adults. Thus, height, weight and BMI are good indicators of nutritional status and the risk of morbidity and mortality in the elderly population. Height can be difficult to measure in the elderly as a result of increasing spinal curvature with age; there are no guidelines regarding the degree of spinal curvature that would invalidate the measurement of height. It can be estimated from knee height or from arm span, although WHO (1995) recommends knee height as being the more satisfactory of the two. The estimated height can then be used to derive BMI, using the recommended cutoff points of <18.5 for under- weight and >25 for overweight, the same as those used for non-elderly adults.

Comparative analysis of adult anthropometry, food consumption and socio- economic variables: case study of nutritional surveys in Brazil

A comparative analysis of the numbers of individuals with food inadequacy estimated on the basis of the FAO DES methodology and the numbers of undernourished adults estimated by anthropometric indices such as BMI is useful for two important reasons. The estimates of numbers with inadequate food intakes, both globally and regionally, have hitherto been based on the FAO methodology. This method uses the per capita DES data from food balance sheets in combination with information relating to the distribution of food consumption among households obtained from national household surveys. Even if simple, reliable objective anthropometric indicators of food adequacy were incorporated and used more widely in national surveys, a projection of trends and long -term forecasts of food needs could be made only if the relationship between these two approaches were firmly established. Anthropometric data in adults are not currently available on a global or regional basis in order to compile meaningful and representative databases, unlike the data available using the currently well -established FAO methodology. This necessitates a critical and comparative analysis of nutritional anthropometry with food consumption and socio - economic variables that are collected in national surveys. National surveys conducted in Brazil provide an excellent opportunity to assess the relationship of adult BMI to a variety of social, economic and nutritional factors.

Data from two nationally representative nutritional surveys conducted in Brazil were used to compare two different methodological approaches for assessing numbers of undernourished individuals (de Vasconcellos, 1994). The two methods compared were: (1) anthropometric measures of adults, adolescents and children, and (2) nutritional adequacy ratios based on food consumption surveys and estimates of requirements. The two nationwide surveys conducted by the National Institute of Statistics and Geography of Brazil (IBGE) were: (1) the National Food Consumption and Household Expenditure survey (ENDEF) of 1974 -1975 and (2) the National Health and Nutrition Survey -Pesquisa Nacional de Saude e Nutricao (PNSN) -conducted in 1989. The relevant methodology and important findings of this comparative analysis are summarized below.

The ENDEF household survey of 1974 -1975 used a non-weighted probabilistic sample of around 55 000 households with approximately 230 000 individuals, and was designed to represent 22 geographical areas in Brazil. The food consumption data collected from each household were very comprehensive. Over seven consecutive survey days, foods purchased, consumed and wasted, and food stocks of several items in the household were weighed. The weekly food consumption of the family was normalized for the individuals' attendance at meals, taking into account differences in age and sex of individuals within the household. An "equivalent adult "scale was thus developed, based on the mean energy requirement of each sex and age group, which was then used to derive the per-adult energy intake of the household, calculated by dividing the family/household intake of energy by the total number of "adult equivalents "in the household. Since the survey was conducted in 1974 -1975, prior to the FAO/ WHO/UNU (1985) energy recommendations, the analysis was limited to the estimation of minimum energy requirements and therefore to the minimum energy adequacy ratio. The minimum energy requirement used for the ENDEF data bank corresponded to 1.5 times the basal metabolic rate, which by coincidence is nearly identical to the factors of 1.56 for males and 1.55 for females used for calculating the minimum energy requirements in the FAO analysis for the Sixth World Food Survey (FAO, 1995).

A detailed statistical analysis of the ENDEF database showed that curves of BMI classes and minimum energy adequacy classes went in the same direction, from a low BMI and low adequacy to a high BMI and high adequacy. The family energy adequacy classes were related closely to the individuals' BMI classes. Thus, the probability of a low BMI individual being in a family/household of low energy adequacy was greater than being in a family with a high energy adequacy. The correspondence plot of the relationship between per adult energy intake and BMI class showed a good correspondence despite problems related to comparing a family-level variable with an individual one. The two curves went in the same direction from a low energy intake and low BMI to a high energy intake and high BMI. Individuals with a low BMI tended to be in those families /households with low energy intakes, while obese individuals were more frequent in households with high energy intakes. Energy intakes were expressed as per-adult equivalent intakes of energy, and results were weighted by the sample weight, therefore considered to be representative of the whole Brazilian adult population (de Vasconcellos, 1994). Similarly good correspondence relationships were established for a range of socio -economic variables such as family total expenditure and BMI categories.

The distribution of the minimum energy adequacy ratio observed in the ENDEF survey of 1974-1975 was compared with the prevalence of undernutrition based on anthropometry measurements from same survey. The prevalence of adult undernutrition based on BMI was lower than that based on the energy adequacy ratio, while the prevalence of obesity as measured by anthropometry was higher. There were several assumptions made during the analysis of the ENDEF 1974 -1975 survey data that may contribute to this discrepancy. It is important to emphasize that one source of data was at the family/ household level data, while the other was individual. Overall, the national data, summarized in Table 3, most probably mask large regional differences in the prevalence of the three nutritional states (undernourished, normal and obese) based on anthropometric and energy adequacy ratios. However, the correspondence plots and other analyses support the conclusion that both methodological approaches have merit and are likely to provide complementary information. Food consumption surveys are more difficult and expensive to carry out on a nationally representative basis at periodic intervals and are plagued with difficulties related to obtaining individual data from household -level information. Anthropometric data, however, do not necessarily reflect food consumption or energy adequacy per se, as they are also influenced by environmental determinants of nutritional status such as infections. However, the usefulness of nutritional anthropometry stems from its close correlation with the multiple dimensions of individual health and development and their socio -economic and environmental determinants. The data based on both methodological approaches thus provide complementary information that can be used to estimate the prevalence and numbers of undernourished children and adults. In addition, these data may be used to assess the presence of overnutrition in individuals and the community, now recognized as an equally important problem particularly among countries in transition (Monteiro, Conde and Popkin, 2002). There is an urgent need for similar critical comparative analytical studies to examine this issue further.

Comparison of the estimated numbers of undernourished by DES and by nutritional anthropometric approaches

The lack of apparent agreement in the estimated numbers of undernourished in different countries based on food energy inadequacy (i.e. the DES method) and estimates based on anthropometry (Svedberg, 2000) highlights the urgent need for additional comparative analytical studies, similar to that described in the Brazilian case study, in order to examine this issue critically. There is need for a more detailed investigation of the relationship between the different approaches and their respective estimates. Such a critical examination will ideally provide useful insights into the relative merits of the various approaches, create opportunities to both improve and refine these methods and enable better interpretation of the differences observed in these estimates. A study by Nubé (2001) attempted to do precisely that using a wide range of datasets based on both approaches from a wide range of countries worldwide, and is summarized below.

The analysis undertaken by Nubé (2001) used two main sources of data -national level DES published by FAO and anthropometric data of adults and children made available from various Demographic and Health Surveys (DHS) in a large number of developing countries, together with anthropometric data of children supplemented from the WHO Global database. The results of this analysis showed that the relationships between estimates derived from the two approaches for 23 countries, e.g. DES and indicators of poor nutrition (underweight in preschool children), were poorly correlated. When the data were analysed at the regional level, the relationships between similar levels of energy inadequacy did not reflect similar patterns of childhood undernutrition; for example, for similar levels of energy inadequacy, the prevalence of a low BMI (<18.5) in adult women was much more marked in countries in the Asian region than in countries in Sub -Saharan Africa. It was further shown that DES was poorly correlated with both numbers of adult women with low BMI and numbers of underweight children. The author argues that the lack of consonance between the estimates based on these two approaches could be attributable to two possible causes. First, there is a fundamental difference between countries or regions with regard to factors that affect the relationship between food consumption and anthropometric outcomes both in children and adults. Second, the poor quality of the data themselves may be responsible for these apparently incoherent findings. It is probable that both of these causes operate to a greater or lesser extent to explain the discrepancies, but above all, this analysis further strengthens the view that the two approaches measure entirely different processes and outcomes, and therefore it should not be entirely surprising that they do not agree with each other.


Percentage distribution










Based on Energy Adequacy Ratio







Nutritional status of entire householdb







Nutritional status of adults in householdc







a Source: de Vasconcelles (1994).
b Nutritional status based on weight-for-height of children and BMI cutof fs of adults.
c Nutritional status of adults only using BMI cutoffs.

Issues related to sampling and representativeness of surveys to collect anthropometric data

In order for data obtained from anthropometric surveys at the community, regional, state or national level to be representative, attention must be paid not only to the quality and reliability of the data collected but also to the sampling frame. If the sample size is adequate, a survey can produce truly representative data with the required degree of precision. As it is often impossible to carry out simple random sampling of the survey population in developing countries, a multistage sampling procedure such as cluster or stratified sampling may be more feasible, with care to take design effects into account for estimating the population variance. The sample size is determined by the proportion of undernourished in the population and the desired precision of the estimate; an additional determinant of sample size is the availability of financial resources. In fact, costs associated with adequate sampling may be considerable and thus a limiting factor to the use of anthropometric surveys on a large scale.

The multidimensional causality and consequences of undernutrition

The causes of undernutrition in a population are multidimensional. The determinants of malnutrition include both food and non-food related factors that often interact to form a complex web of biological, socio -economic, cultural and environmental deprivations in developing countries. Childhood malnutrition rates are known to respond over the short term to variations in food supply owing to seasonal and climatic changes, and are associated with chronic food deficiencies largely resulting from social, political and economic factors. Although establishing a relationship between these variables and the indicators of childhood undernutrition does not necessarily imply causality, they do demonstrate that in addition to food availability, many social, cultural, health and environmental factors influence the prevalence of undernutrition. Undernutrition is thus more than a problem of inadequate food supply. Although people suffering from inadequacy of food are generally poor, not all the poor are undernourished. Based on studies in China, even in food secure households some members may be undernourished while others may be overweight, suggesting that with the process of nutritional transition with economic development, both features may be seen within the same household (Doak et al., 2002). Income fluctuations, seasonal disparities in food availability and demand for high levels of physical activity, as well as proximity and access to marketing facilities, may singly or jointly influence nutrition. For example, the transition from subsistence farming to commercial agriculture and cash crops may help improve nutrition in the long run; however, it may also have neutral or even negative impacts over the short term unless it is accompanied by improvements in access to health services, environmental sanitation and other social investments (von Braun and Kennedy, 1994). Rapid urbanization and rural-to-urban migration may lead to nutritional deprivation among segments of society. Cultural attitudes reflected in food preferences and food preparation practices as well as women's time constraints including child care can influence the nutrition of the most vulnerable, i.e. the women and children. Poor housing and over- crowding, poor sanitation and lack of access to protected water supply, through their links with infectious diseases and infestations, are potent environmental factors that influence biological food utilization and nutrition. Inadequate access to food, health care, safe water, a clean environment and educational opportunities are in turn determined by the economic and institutional structures as well as the political or ideological superstructures within a society. Nutritional anthropometric approaches are constrained by their inability to capture the multidimensional nature of the problem of hunger and poverty but may reflect the situation of poverty in the community better than other approaches that only assess food availability.

The poor nutritional status of populations affects physical growth, intelligence, behaviour and learning abilities of children and adolescents. It impacts on their physical and work performance and has been linked to impaired economic work productivity during adulthood. Inadequate nutrition predisposes the person to infections and contributes to the negative downward spiral of undernutrition and infection (Tomkins and Watson, 1989). Good nutritional status, however, promotes optimal growth and development of children and adolescents. It contributes to better physiological work performance, enhances adult economic productivity, increases levels of socially desirable activities and promotes better maternal birth outcomes (ACC/SCN, 2002). Good nutrition of a population manifested in the nutritional status of the individual in the community contributes to an upward positive spiral and reflects the improvement in the resources and human capital of society. The use of nutritional anthropometry to assess the nutritional status of population groups can contribute to evaluating both the adequacy of food in a country and the development of human resources and capital. As countries industrialize and develop economically, they face problems associated with full access to food, both in quantity and in quality. Unlimited food access and changes in lifestyle lead to an increased incidence of obesity and the consequent risk of chronic non-communicable diseases such as diabetes mellitus, cardiovascular disease and cancer. Assessment of nutritional status using anthropometric methods will also provide information on the risk of chronic disease in population groups where the availability of food is improving.


Nutritional anthropometric indicators provide a reflection of the nutritional status of the community and hence complement the information obtained by other approaches. Data on adult BMI are now increasingly available, consequent to the initiative taken by the FAO to consider BMI as an alternative and complimentary indicator of nutritional status alongside the standard methodology adopted by FAO to assess energy adequacy. Documents from the the World Food Summit, held in 1996, state that where it is possible to differentiate between the effects of sanitation, health and care, and those of household food security, indicators of nutritional status can provide the most direct ways of assessing the status of food security at the household level (emphasis mine). This is indeed the challenge. Anthropometric indicators are useful, as they provide a simple and practical way of describing the problem in the community.

They are possibly the best general proxy for constraints, such as dietary inadequacies, infections and other environmental risks, on the well -being of the poorest. Anthropometric indicators are strong and easily obtainable predictors at the individual and population level of subsequent morbidity, functional impairment and mortality, i.e. the consequences of poverty and hunger. They are reliable indicators for measuring the success or failure of interventions at the micro level and for measuring the impact of macro level changes. Nutritional anthropometric assessment has a wide range of additional uses (Gibson, 1990), summarized in Table 4. Measures of chronic undernutrition in the community, particularly among vulnerable children and women, are included in international development goals and should be included in the evaluation of development interventions. A recent workshop in Fiuggi, Italy sponsored by the International Fund for Agricultural Development made the following important recommendation: "Monitoring through anthropometric indicators (stunting, wasting and underweight) can be adopted as an affordable and acceptable means to assess the impact of interventions and programs "(FAO, 2001).

There is thus an increasing recognition of the role of nutritional anthropometric indicators in the work of UN agencies, Bilaterals and the non-governmental organizations, but also by Member Governments.

Projecting future trends and forecasting the numbers of undernourished are important functions of organizations such as FAO and can be achieved in different ways. Any approach at projection is likely to be uncertain; it is therefore essential that all assumptions be explicitly stated. Projections may be made on the basis of historical trends or by using more complex econometric modelling techniques. Given the uncertainty of all available approaches, the use of simple, not too complex, methods has its advantages. Attempts at forecasting the numbers of undernourished have been made only recently. Even simple, rather superficial approaches at projection based on anthropometry have been found to be largely comparable with those using more detailed and time -consuming methodologies. The main point, however, is that these methods measure different aspects of undernourishment, and together, they provide complementary information for a more comprehensive understanding of the situation.


Screen populations to help identify individuals at risk of undernutrition (or overnutrition)

Identify the degree and severity of undernutrition, i.e. mild, moderate or severe

Identify vulnerable groups within a community based on age, gender and other categories

Provide information on previous long-term nutritional stress

Evaluate changes in nutritional status of populations /communities over time

Assess secular trends in nutritional status of populations /communities, i.e. from one generation to next

Compare the nutritional status of different population groups and nations

Assess the impact of nutritional intervention programmes or social, economic, political or structural changes in society on the nutritional status of the community

Assess nutritional status of an individual, community or population groups

Source: Gibson (1990).


Administrative Committee on Coordination/ Standing Committee on Nutrition. 2002. Ending malnutrition by 2020: an agenda for change in the millennium. Food Nutr. Bull., 21: 3 -88.

von Braun, J. & Kennedy, E. 1994. Conclusions for agricultural commercialization policy. In J. von Braun & E. Kennedy, eds. Agricultural commercialization, economic development and nutrition. Baltimore, MD, Johns Hopkins University Press.

Doak, C., Adair, L., Bentley, M., Fengying, Z. & Popkin, B. 2002. The underweight /overweight household: an exploration of household socio -demographic and dietary factors in China. Public Health Nutr., 5: 215 -230.

FAO. 1994. Body mass index: a measure of chronic energy deficiency in adults, by P.S. Shetty & W.P.T. James. Food & Nutrition Paper No. 56. Rome.

FAO. 1995. Sixth world food survey. Rome.

FAO. 1999. The state of food insecurity in the world. Rome (also available at

FAO. 2001. The state of food insecurity in the world. Rome (also available at

FAO/WHO/UNU. 1985. Energy and protein requirements. Report of a joint FAO/WHO/UN ad hoc expert consultation. WHO Technical Report Series, No. 724. Geneva.

Gibson, R.S. 1990. UK, Principles of nutritional assessment. Oxford, UK, Oxford University Press.

Grantham-McGregor, S.M., Powell, C.A., Walker, S.P. & Himes, J.H. 1991. Nutritional supplementation, psychosocial stimulation, and mental development of stunted children: the Jamaican Study. Lancet, 338(8758): 1 -5.

James, W.P., Ferro-Luzzi, A., & Waterlow, J.C. 1988. Definition of chronic energy deficiency in adults. Report of a working party of the International Dietary Energy Consultative Group. Eur. J. Clin. Nutr., 42: 969 -981.

Jelliffe, D.B. 1966. The assessment of the nutritional status of the community. WHO Monograph No. 53. Geneva.

Monteiro, C.A., Conde, W.L. & Popkin, B.M. 2002. Is obesity replacing or adding to undernutrition? Evidence from different social classes in Brazil. Public Health Nutr., 5(1A): 105 -112.

Nubé, M. 2001. Confronting dietary energy supply with anthropometry in the assessment of undernutrition prevalence at the level of countries. World Dev., 29: 1275 -1289.

de Onis, M., Monteiro, C., Akre, J. & Glugston, G. 1993. The worldwide magnitude of protein -energy malnutrition: an overview from the WHO Global Database on Child Growth. Bull. World Health Organ., 71: 703 -712.

Popkin, B.M., Richards, M.K. & Montiero, C.A. 1996. Stunting is associated with overweight in children of four nations that are undergoing the nutrition transition. J. Nutr., 126: 3009 -3016.

Svedberg, P. 2000. Poverty and undernutrition: theory, measurement, and policy. Oxford, UK, Oxford University Press.

Tomkins, A. & Watson, F. 1989. Malnutrition and infection: a review. ACC/SCN Nutrition Policy Discussion Paper No. 5, pp. 1 -136. Geneva.

UNICEF. 1998. The state of the world's children 1998. Oxford, Oxford University Press.

de Vasconcellos, M.T. 1994. Body mass index: its relationship with food consumption and socio -economic variables in Brazil. Eur. J. Clin. Nutr., 48 (Suppl. 3): S115 -S123.

WHO. 1983. Measuring change in nutritional status. Guidelines for assessing the nutritional impact of supplementary feeding programmes for vulnerable groups. Geneva.

WHO. 1995. Physical status: the use and interpretation of anthropometry. WHO Technical Report No. 854. Geneva.

WHO. 1999. Nutrition for health and development. Geneva.

WHO Working Group. 1986. Use and interpretation of anthropometric indicators of nutritional status. Bull. World Heath Org., 4: 924 -941.

Discussion opener - anthropometric survey methods

Peter Svedberg
University of Stockholm
Stockholm, Sweden

In this short discussion opener, I would like to elaborate on one of the statements in Dr Shetty's paper that particularly interests me. He correctly states that the FAO and anthropometric methods for assessing nutritional deficiency in a population "measure different things "and should not be expected to come up with "similar results ". I fully agree, but we need to go one step further and find the main explanations for why and how the two methods are "complementary ".

The data at hand reveal that with the two methods, different and "conflicting "proportions of the population are categorized as "undernourished "and underweight (or stunted or wasted), respectively, in different parts of the world. The most challenging observation is perhaps that the FAO finds the prevalence of undernourishment to be the highest in the Sub -Saharan African countries, while all anthropometric indicators identify South Asia to be the continent with by far the highest incidence of underweight and stunting, for both adult women and young children of both sexes.

We all know that both methods are subject to measurement errors and biases, which could be the chief reason for the "conflicting" results. In this short note, however, I will outline a theory for how to reconcile on factual grounds the diverse estimates derived with the two methods, and will propose a simple empirical test of this theory.

Let us start by agreeing on important features that are common to both estimation methods. They both rely on (1) cutoff points to delineate the "undernourished "and the "anthropometrically failed", respectively, and (2) both these cutoff points are based on similar norms for peoples' minimum acceptable body weights (Table 1).

The FAO method estimates the share of the population that does not have access to the calories required to maintain the norm body weights (and some light physical activity). WHO and DHS estimate the shares of adult women and young children that actually have body weights below the norms.


Components in the cutoff points



Body weight


BMIb = 18.5

BMI = 18.5


Median in reference

-2 SD of median in reference

Physical activity


0.56 × BMRc

Not considered


Median in reference

Not considered

a Demographic and Health Survey.
b Body mass index.
c Basal metabolic rate.

Considering (1) that the cutoff body weights are quite similar and (2) that BMR for maintaining health -consistent body weight accounts for about two -thirds of the total calorie requirement in FAO's cutoff, one would expect an association across countries between the two sets of estimates. This is not what we see in the data, however. There is hardly any correspondence across countries between the prevalence of undernourishment as estimated by the FAO and of underweight as estimated by anthropometry. This holds for adult women as well as for young children, the two categories for which data on underweight exist. In tackling the question of whether a "rationale "exists for the lack of a bivariate association between undernourishment and underweight, we should first recall a few non-debatable facts:

(1) Human energy (calorie) expenditures can be for four uses only, represented by the following variables:

(a) BMR (basal metabolism for body functions that is proportional to body weight);

(b) PAL (physical activity, including work);

(c) WASTE (energy wasted through illnesses, such as fevers, intestinal parasites and diarrhoea);

(d) body weight (BW) change.

(2) The energy contained in ingested food cannot disappear, what "goes into" a person (i) must "go out" in some combination of these four uses (the First Law of Thermodynamics). The following identity must always hold:

EXPENDi º BMRi + PALi + WASTEi + DBWi. (1)

In the short term, DBWi»0. We then have that calorie expenditure is equal to calorie intake, which defines energy balance. This further means that:

BMRi = INTAKEi - PALi - WASTEi. (2)

In going from this indisputable theoretical relationship for individuals to the formulation of a testable model of the relationship between the prevalence of underweight and "undernourishment" across countries, a number of assumptions have to be made:

(1) There is a strong inverse relationship between the prevalence of underweight (UND) and per capita expenditures on BMR across countries.

(2) The anthropometric estimates of underweight for adult women and young children are representative for the entire population in each country.

(3) There is a strong inverse correlation between the prevalence of undernourishment (POU) and "intake" ((dietary energy supply, DES) (the adjusted R 2 between these two variables across countries is actually 0.97).

(4) PAL and WASTE can be measured on a per capita basis in the respective countries.

(5) The simultaneity between UND and the explanatory variables can be handled.

We can then test the following regression model:

UNDj = a + bPOUj + dPALj + gWASTEj + e. (3)

The hypothesis is hence that the share of the population underweight in a country (j) is a function of the prevalence of undernourishment (POU), the average energy intensity of work activities and the frequency of illness (waste of calories) in the country. The theory leads us to expect that â, ä and ã should all be positive. If so, and if the t-statistics indicate significance, we would have a tentative empirical explanation for the lack of a bivariate association between UND and POU (owing to omitted variable bias).

Finally, what proxy variables are likely to be good candidates for PAL and WASTE? For physical work intensity, we can think of the share of the labour force in agriculture (given the degree of mechanization), proportion of (female) adults in the labour force, estimated hours worked per person and year, urbanization, etc. WASTE can be proxied with outcome indicators such as DALYs (disability adjusted life years) or age -specific mortality, or by input indicators such as provision of health services, doctors /nurses per 1 000 inhabitants, vaccination rates and also (female) education.

Discussion opener - anthropometric survey methods

Stephan Klasen
University of Munich
Munich, Germany


The paper by Shetty provides a good overview of the issues regarding the use of anthropometric survey data for the measurement of undernutrition. The paper argues that such anthropometric indicators, particularly the body mass index (BMI) for adults and measures of stunting, wasting and underweight for children, should be used for an assessment of undernutrition in developing countries. But there are a number of issues that are insufficiently covered and warrant a discussion which will follow.

FAO and anthropometric- based methods to measure Undernutrition

The paper argues that the two approaches should be seen as complementary. It is stated that they are unlikely to yield identical results as the former is an input (top -down) measure, and the latter is output-oriented (bottom- up). The reality, however, is much more problematic, as the two methods generate highly contradictory results, particularly regarding the regional intensity of undernutrition in the world. While the FAO method identifies Sub -Saharan Africa and the Caribbean (!) as the regions most affected by undernourishment, the anthropometric assessments find undernutrition to be worst in South Asia, while child mortality indicators show yet another regional pattern (see Figure 1). Also, the two methods do not agree on trends over time overall or within the various regions. This is unlikely to be entirely (or even mostly) related to the difference between inputs and outputs. Thus, they cannot complement each other meaningfully; different inherent biases in the methods (with the biases being much larger in the FAO method; see, for example, Svedberg, 2000) lead to these contradictory results.

The advantages of anthropometry

While one -shot anthropometric surveys suffer from issues of reference standards, age structure of the sample, sample selection, individual heterogeneity, etc., the most reliable way to measure undernutrition appears to be repeat measurements of individuals, particularly children. Growth faltering and failure to gain weight are the most reliable indicators of nutritional problems in children, and weight loss among adults can point similarly to nutritional problems (WHO, 1995). This is not mentioned in the paper but might be the most promising avenue to use anthropometry for assessments of undernutrition. A second critical advantage of the use of anthropometric measures (compared with particularly the top -down approach of the FAO method) is that it allows the identification of particularly affected people (rather than just a population share) and a careful analysis of the determinants of undernutrition. Using anthropometric data and socio -economic information, as available, for example, in the Demographic and Health Surveys, allows a careful assessment of the socio -economic, demographic and also spatial determinants of undernutrition that can be of critical importance for the formulation and monitoring of appropriate policies (see, for example, Kandala, Fahrmeir and Klasen, 2002.


Anthropometry of children and adults

The Shetty paper discusses anthropometric indicators for children and adults. While there is consensus that the indicators among children yield meaningful and important information about nutritional status (although there remain some controversies), the use of a single BMI cutoff is much more controversial as a healthy BMI is likely to vary with age, sex, pregnancy/lactation, ethnicity, climate and other factors (WHO 1995). Just taking a BMI of <18.5 as the cutoff for everyone everywhere is therefore likely to be far too crude. It is unclear what cutoffs are recommended for adolescents, for whom the BMI clearly is not an appropriate indicator. Also, it is unclear how one can make a correspondence between low BMI of adults and wasting or stunting among children. This would be necessary, however, if one wanted to arrive at global figures of undernourished people (adults as well as children) using anthropometry.

The problem of type-II error (sensitivity)

Cutoff-based anthropometric assessments suffer from the problem that many undernourished people who may be only 1 -2 standard deviations below the reference median will be wrongly identified as adequately nourished, as the cutoffs are usually set at <2 standard deviations of the reference median. This problem cannot be solved easily, as a more generous cutoff will then increase the Type 1 error and erroneously identify well - nourished people as undernourished. This trade -off is a critical issue to consider also for BMI -type measures.

The appropriate method for the appropriate purpose

The choice of indicators to measure undernutrition largely depends on the purpose at hand. If the interest is to combat undernutrition in clinical practice, clearly anthropometry will provide one of the best tools to assist and monitor interventions for identifying undernutrition in individuals. If the purpose is to understand the determinants of undernutrition for designing appropriate economic and health policies, anthropometry can also be very useful, particularly in connection with the many available micro datasets that include socio -economic and anthropometric information. If the purpose at hand is to understand food intake behaviour and its likely consequences, a combination of food expenditure surveys and anthropometric indicators is particularly useful. If the purpose is to assess the magnitude of undernutrition in the world and compare its relative prevalence, anthropometric assessments have their usefulness but clearly also their limits. The limits relate to the low reliability and comparability of anthropometric information of adults and thus a need to exclude adults and simply concentrate on children. Other limitations are associated with the currently used problematic reference standard of undernutrition for children (Klasen, 1999), and problems of timeliness, reliability, sampling, etc. This is not to say that alternative methods are superior for that purpose; they are not, as they suffer from different and generally even larger problems and biases. But given that, with anthropometry, we can identify undernourished individuals, understand the causes of their undernutrition and design and monitor appropriate policies, the question of counting how many there are and where they are, for which we do not presently have an acceptable method, seems much less important.


Kandala, N.B., Fahrmeir, L. & Klasen, S. 2002. Geo -additive models of undernutrition in three Sub -Saharan African countries. Munich, Faculty of Economics, University of Munich (mimeo).

Klasen, S. 1999. Malnourished, but surviving in South Asia, better nourished but dying young in Africa: what can explain this puzzle? SFB Working Paper No. 214.

Svedberg, P. 2000. Poverty and undernutrition. Oxford, Oxford University Press.

WHO. 1995. Physical status: the use and interpretation of anthropometry. WHO Technical Report Series No. 854. Geneva.

Discussion group report - anthropometric survey methods

Gina Kennedy
Rome, Italy

The speakers opened the discussion by reviewing some of the advantages and weaknesses of using anthropometry to estimate the prevalence of undernutrition and by comparing this approach to the FAO method. One of the major strengths of anthropometry mentioned was that it is an outcome measure and therefore is well suited for monitoring and evaluating interventions. Anthropometry can be used also to track individual status. For example, in growth -monitoring programmes, an individual child's weight can be monitored over time in order to track positive, negative or stagnant trends in weight gain. This can be a powerful tool in a community setting and has been used within the framework of many community nutrition programmes. Another important advantage of anthropometry mentioned was that the measurements are often carried out in the context of larger household surveys that collect data on many aspects related to outcome, such as health status, household income, literacy rates and access to clean water. Some of the weaknesses that were pointed out included the lack of internationally accepted indicators for children six to 18 years of age and the scarcity of data on BMI, particularly for men.

Both speakers highlighted that the FAO method and anthropometry do not show geographical concordance. For example, the FAO method finds that Sub -Saharan Africa has the largest number of undernourished, while South Asia has the highest prevalence of underweight children. A method was presented that attempted to reconcile conflicting evidence seen arising from cross-country comparisons of the various indicators. It was proposed that a model factoring in both physical activity levels and health status of populations could be used to provide an empirical explanation for the lack of association between prevalence of underweight and undernourished. In order to apply the model, several factors would need to be taken into consideration, such as valid variables for measuring physical activity level and health status of populations and their distribution across countries.

Most participants were not concerned over the lack of concordance between methods, as they measure different things. Anthropometry is an outcome measure encompassing various factors including food, health status and general care patterns, while the measurement of undernourishment is a gross calculation based on per capita food availability. Many participants stressed the need to concentrate on trends more than on levels or absolute numbers. Country trends are particularly useful for determining the rate and slope of progress or regress. However, numbers were seen as powerful advocacy tools that can be used for political motivation. Numbers were also seen as useful to calculate the cost of interventions. For example, numbers are useful for determining the cost of supplying vaccines to at-risk populations. Lastly, the group stressed that any improvements to the methods should be both affordable and replicable. The idea that indicators do not need to be measured every year was widely supported.

The group summarized the uses of anthropometric indicators for children, adolescents and adults. Participants agreed that the use of anthropometry among children under five years of age has reached a level of international consensus. The availability of these data can contribute to the assessment of vulnerability in populations. However, there is a lack of information, and there are no accepted indicators for the adolescent ages. The most important factor related to the difficulties of developing appropriate indicators for this age group is the effect of puberty, which has varied ages of onset and differs in intensity and duration from one individual to the next. Anecdotal evidence from the group also indicated that this age group is particularly difficult to capture at home and that non- compliance is a factor in lack of progress towards developing valid indicators. The group stressed that the use of BMI as an anthropometric indicator for adults is relatively new compared with the indicators used to assess child growth and therefore needs more time to develop. While there is evidence linking a low BMI to increased morbidity, mortality, decreased work productivity and low birth weight in offspring, increased efforts are needed to establish these relationships.

Several recommendations were made during the discussion regarding the future role of anthropometric indicators. It was suggested that anthropometric data always be presented together with confidence intervals and information on the distribution (mean, Z-score and standard deviation). Nationally representative data should continue to be collected on children through surveys such as the Multiple Indicator Cluster Surveys and Demographic Health Surveys or similar national initiatives, with intensified efforts for countries where data are scarce. The information necessary to calculate anthropometric indicators (weight, height, age and gender) should always be collected. For adolescents, efforts should be intensified to develop appropriate indicators. For adults, weight and height data necessary to calculate BMI should always be collected for both men and women during surveys. It was felt by the group that the international community needs to reach a consensus on issues such as appropriate age groupings and BMI cutoff points, as has been accomplished for child anthropometric indicators.

Previous Page Top of Page Next Page