The World Food Summit mandated FAO to measure and monitor progress towards the Summit goal of halving the number of hungry by 2015. The decision to hold a scientific symposium on the measurement of food deprivation and undernutrition was motivated by this objective and the need to review the current status of the widely used methods for measuring hunger. It also aimed to recommend improvements in the methods, which would help FAO to further its work in carrying out this mandate. Since the monitoring of the progress towards the World Food Summit goal involves national and international stakeholders, the Symposium also provided an opportunity for them to present their perspectives.
By promoting dialogue among advocates of various methods, the Symposium served to create a greater appreciation of the strengths and weaknesses of the different methods as well as how the corresponding measures complement each other. The main consensus to emerge at the Symposium was that no single measure can capture all aspects of hunger while at the same time providing policy-makers with relevant and timely information in a cost-effective manner.
The five methods covered in the Symposium were: (1) the FAO method for measuring undernourishment by combining information from food balance sheets and household income and expenditure surveys; (2) measurement of food insecurity using household income and expenditure survey data; (3) measurement of adequacy of dietary intake based on individual intake surveys; (4) measurement of child nutritional status based on anthropometric surveys; and (5) qualitative methods for measuring people's perception of food insecurity and hunger. The first three compare dietary energy availability or intakes with energy requirements, the fourth measures nutritional outcomes, and the fifth measures people's perceptions of hunger.
The FAO method involves the estimation of a distribution function of dietary energy consumption on a per-person basis. The mean of this distribution refers to the usual food consumption level and is estimated by the daily dietary energy supply per capita for a country derived from its food balance sheet (averaged over three years). The variance, as measured by the coefficient of variation, is derived on the basis of food consumption or income data from household income and expenditure surveys. The proportion of undernourished in the total population is defined as that part of the distribution lying below a minimum energy requirement level derived by taking into account the sex and age distribution of the country's population and assuming the minimum acceptable body weight for a given height for all sex-age groups and light activity levels for adults.
Nationally representative household income and expenditure surveys have now been carried out in a large number of countries. These surveys allow each household's average food consumption and energy intake to be calculated, since they typically ask respondents to recall their consumption of food items over a reference period, e.g. the previous week. The proportion of households in a country whose energy intakes fall below a minimum energy level can then be calculated.
The food intake surveys measure actual food intake at the individual level. The modalities for data collection include taking a dietary history, administering a food frequency questionnaire, recording weights of foods consumed, asking respondents to recall what they ate in the previous 24 hours or analysing the chemical and nutrient content of diets. This information can then be compared with dietary energy requirements to arrive at the proportion of the population with deficient energy intakes. However, nationally representative surveys of individual dietary intakes in developing countries are rare and require considerable human and financial resources.
An alternative approach to the measurement of dietary deficiency is to study nutritional outcomes. Undernutrition is diagnosed when individuals' anthropometric measurements in terms of weight and height fall below international reference standards. Poor growth in infants and children as well as underweight in adults may be the consequence of both inadequate food intake and poor absorption of food caused by environmental factors such as infections or inadequate parental care, in the case of children. Nationally representative anthropometric surveys have now been carried out for most developing countries. From these, estimates can be made of the proportion of persons who fall below established cutoffs and who, therefore, are considered to be undernourished.
An important dimension neglected by the methods described above is that hunger is as much a social as a biological problem. People who lack the means to acquire sufficient food may regard themselves as hungry, even if there are no clinically recognizable signs of undernutrition. Furthermore, even if they are not currently hungry, they may have a well-founded fear of future deprivation. Qualitative or "self-assessment" indicators of food insecurity have been developed that attempt to capture these dimensions and are well correlated with other known measures of food insecurity and hunger. Up to now, this work has been mostly confined to developed countries, although efforts are under way to extend the use of qualitative measures of hunger to developing countries as well.
During the Symposium, the discussion revolved around how indicators of food security and undernutrition are used, and what desirable qualities such measures should have to increase their usefulness and reliability. Several criteria were proposed and considered in selecting indicators. Validity and reliability were considered essential, i.e. how well the indicator measures what it claims to measure and whether the measurement is replicable across samples. Also, it is important to know how quickly and frequently the information becomes available to policy-makers and those who work on hunger-related issues, and whether the collection costs in terms of time, equipment and training of personnel are reasonable. In many cases, the benefits of information will outweigh the costs when its collection is justified by a clear link to actions, policies, or programmes that rely on valid and reliable information. When the measures provide additional information on the causes of food insecurity, they can be used for political advocacy and to guide policy changes. There was wide agreement that indicators should be able to indicate trends correctly as well as provide the distribution of hunger at a specific point in time. The comparison of trends is very important across countries but is equally important for monitoring change over time within individual countries.
The FAO method is capable of providing timely information at a low cost, since it can be compiled readily from food balance sheet data collected routinely in almost all developing countries. Therefore, the information derived is useful for decision-making at the global and national level and also for identifying trends. Although the FAO method can be applied to subnational population groups using household income and expenditure survey data, it cannot provide estimates disaggregated by sex or age categories. There is a great need for subnational information, as food insecurity is increasingly concentrated in particular regions or groups within countries and is becoming less a matter of aggregate food availability. An advantage of household surveys is that they can provide this type of information, but they also require more resources and time, particularly if these surveys are nationally representative. In general, it is difficult to conduct high-quality surveys regularly on a large-scale at intervals of less than three to five years. However, this may not be necessary. There was widespread support at the Symposium for the proposition that most indicators do not need to be measured every year.
In brief, there may be an inverse relationship between data quality and usefulness, and the cost and ease of collection. For example, very few nationally representative individual food intake surveys have been carried out in developing countries for reasons of cost, expediency and logistics, despite the fact that these surveys provide the most precise data based on rigorous studies of validity and reliability, and much is known about their error structure. The same is not true of the other measures. With respect to the FAO method, questions remain about the accuracy of the underlying food balance sheet data, arising from flawed production and trade data in countries with relaxed borders, as well as of the food composition tables used for determining the energy content of foods. There are problems as well with the other measures discussed in the Symposium that may limit their ability to provide accurate and complete information on food deprivation and undernutrition.
As the survey-based measures in one way or another require inferences from the frequency distribution or the percentile distribution of the variable of interest, a related concern is whether such inferences based on the available sample survey data are valid or reliable. The keynote paper on the FAO method reminded participants that sampling designs are usually chosen to give an accurate estimate of the mean and not necessarily the frequency distribution of the variable. Hence, the issue of making inferences on the frequency distribution or the percentile distribution from sample survey data needs to be addressed.
It was established during the Symposium that no individual measure suffices to capture all aspects of food insecurity. A suggestion was made, widely endorsed by participants, that a suite of indicators was needed to cover the different dimensions of food security: availability, access, utilization and stability of access. Thus, the FAO method would measure availability, the surveys of household expenditures and individual food intake would measure access, the anthropometric method would measure utilization, and the qualitative method would measure stability of access or vulnerability to food insecurity.
Also endorsed during the discussions was the need to focus on trends rather than magnitudes, as the ranking of countries by prevalence of food insecurity at a single point in time poses some difficult problems. The following may contribute to the inaccurate ranking of countries: (1) food balance sheets fail to take account of the importance of roots and tubers because their production cannot be measured as accurately as for cereals; (2) household surveys often fail to take an accurate account of food eaten outside the home; (3) activity levels are largely unknown for assessing dietary energy requirements at the individual or household level; and (4) anthropometric measures for children require that ages be reported correctly - this may be problematic in traditional societies. Not only does a focus on trends allow these problems to be sidestepped, but more importantly, it is often the trend that interests policy-makers and not so much the absolute value.
An additional advantage to the "suite" approach is that trends in indicators can be used to triangulate on likely causes of hunger and provide guidance to policy-makers. Thus, if in some country, the FAO measure shows a marked improvement over a time period, while the anthropometric measure shows little or none, policy-makers might use this information to focus on sanitation and public health interventions rather than on increasing food supply. However, in advocating the use of a suite of indicators, it is also necessary to keep in mind some potential pitfalls. For example, national stakeholders expressed concerns over excessive demands for information by international agencies, the collection of which often uses up scarce national resources without a commensurate return from the use of the information. Many felt that it might be preferable to help countries make better use of the information they already have, to eliminate duplication or to fill in gaps, a course adopted by the Food Insecurity and Vulnerability Information and Mapping System (FIVIMS). Information has to be demanded by the countries themselves, and capacity-building at the country level for generating and utilizing information is crucial for changing the demand for, and hence the quality of, generated information.
In conclusion, the FAO mandate to monitor progress in hunger reduction cannot be fulfilled without accurate, reliable and timely methods that measure the prevalence of hunger, food insecurity and vulnerability, and that also highlight changes over time. What was discussed in the three days of the Symposium is just the beginning of a partnership with the participants and other experts towards the improvement of current methods and the development of new methods that will contribute towards fulfilling this very important mandate.