Stacey Rosen and Shahla Shapouri
US Department of Agriculture
Washington, DC, USA
Food insecurity is widespread across the globe, and the international community has placed the elimination of famine and hunger on its agenda. Measuring and quantifying food insecurity is a crucial component of making progress towards the improvement of food security. Many estimates of food insecurity depend on statistics that are resource-intensive to obtain. This paper presents one possible approach towards measuring the share of the population that might be affected by food insecurity and to what extent. This approach relies on secondary data sources and thus avoids the high cost of primary data collection. A food security threshold can be calculated as the sum of the cost of a food basket and the cost of other basic necessities, and then compared with available income. The approach is illustrated for nine Latin American countries, all of which have segments of their populations that are considered vulnerable.
From the estimation of the gap between per capita income and the cost of a basket of food as well as other basic necessities, it is possible to determine the number of people who lack the purchasing power to satisfy their basic needs.
In this study, retail prices for several food items for nine lower-income countries in Latin America are used to calculate the cost of two kinds of food baskets: a representative healthy food basket and a low-cost healthy food basket. The representative healthy food basket fulfills basic nutritional guidelines while reflecting the range of foods typically eaten in each country. The low-cost healthy food basket is constructed in a way to satisfy nutritional guidelines at the lowest possible cost.
After the cost of the two food baskets are calculated, assumptions about the cost of other basic necessities are made. The sum of the cost of the food basket and the cost of other necessities can be considered as a food security threshold. The gap between actual incomes and the food security threshold determines the depth of food insecurity.
To estimate the purchase price of the food basket, we distributed 2 170 kilocalories - derived from region-specific caloric standards recommended by FAO - among specific food and nutrient groups, according to several criteria. These criteria included typical country food consumption patterns, FAO/World Health Organization nutritional guidelines for developing countries and standards from various US government agencies. The diets are largely plant-based, and the goal was to have roughly 65 percent of daily kilocalories coming from carbohydrates, 20 percent from fat and 15 percent from protein. Respecting the diets of the countries, one or more commodities were selected to represent each nutrient group. The healthy representative food basket typically included between three and six food items in the carbohydrate group, while the low-cost food basket only included one or two of the least costly food items. Cereals, roots and tubers, bananas and plantains were the food items selected to represent the carbohydrate group; milk, meat or pulses were chosen to represent protein; and vegetable oils represented fat. No attempt was made to analyse the adequacy of micro-nutrients, such as iron or vitamin A, in the diet. However, low-calorie intake is typically closely related to low levels of consumption of a wide range of essential vitamins and minerals, and a more diversified diet is more likely to prevent shortages of micronutrients.
The ratio of available income to food security threshold is a meaningful indicator of food security. A ratio of greater than one indicates that income levels exceed the threshold, and therefore people in this quintile are not vulnerable to food insecurity. Any number less than one alerts us to the danger of food insecurity in this income quintile: the lower the number, the more severe the problem.
To examine the implications of skewed income distribution on food security, we allocated national income across five income groups within each country according to income distribution data from the World Bank. We then compared these per capita income levels of the five quintiles to the food security threshold. On average, incomes in the lowest income group are roughly equal to the food security threshold for the low-cost basket.
Conversely, incomes for the highest income group were about 16 times higher than the threshold level. With respect to individual countries, the income level of all five quintiles in Bolivia, Columbia, the Dominican Republic, Ecuador, El Salvador and Peru exceeded the threshold level to purchase food and basic necessities. This means that less than 20 percent of the population in these five countries are vulnerable to food insecurity. In Guatemala, roughly 20 percent of the population had insufficient purchasing power to afford the necessities. In Honduras and Nicaragua, by far the poorest of the nine countries, about 40 percent of the people fell into this category.
In addition to the low-cost food basket, we also made an assessment for a representative healthy food basket, whose costs are obviously higher. Consistent with this is the finding that incomes did not go as far in meeting the food security threshold for this type of food basket. In 1999, the lowest income group in all countries fell short of meeting the food security threshold using the representative healthy food basket approach. This means that at least 20 percent of the population of these nine countries could not afford the healthy food basket and other necessities. Incomes in Peru were the closest to meeting the target level, as the per capita income in the lowest income group was 87 percent of the threshold level. In Bolivia, 40 percent of the population had incomes below the food security threshold. Reflecting the vulnerability of the poorest people in these countries, incomes in this lowest quintile in Honduras and Nicaragua were only 27 percent of the threshold level. In fact, the three lowest income quintiles in these two countries fell short of meeting the threshold level, meaning that roughly 60 percent of the population fell short of the threshold level.
Simon Hales, Tony Blakely,
Charlotte Kieft and Alistair Woodward
University of Otago
Dunedin, New Zealand
Malnutrition is thought to be indirectly responsible for half of all child mortality in poor countries (WHO, 2002). As part of an analysis of the global burden of disease attributable to poverty (Blakely et al., 2002), we analysed the relationship between poverty and anthropometric parameters using data from Demographic and Health Surveys (DHS) (Macro International, 2002). We estimated that children living on less than US$1 per day were two to three times more likely to be malnourished than those living on over US$2 per day. The focus of this paper is on the statistical association of poverty indices and malnutrition in eight regions defined by WHO.
Four rounds of DHS have been conducted in which several thousand households were sampled at intervals in poor countries across Asia, Africa, the Near East, Latin America and the former Soviet Union. These data have been used to generate indices that can be used as a proxy for income (Filmer and Pritchett, 1988). Unlike those analyses, which were carried out at the level of individual countries, we combined household data for 42 countries, taking the most recent survey if the country had been surveyed more than once during 1986-2000. Data from the first round of surveys were recoded to ensure comparability with the subsequent three rounds. We used factor analysis to create a poverty index based on three categorical variables: housing construction material (usually floor material), educational status and availability of electricity. If these variables were missing for a particular country, a substitute variable was used as follows: wall material was substituted for floor material for Pakistan; number of rooms was substituted for floor material in India; possession of a radio was substituted for electricity supply in Burundi, Dominican Republic, Liberia and Tunisia. The variables were recoded with numeric values as follows: electricity: 0 = no, 1 = yes; highest education level: 0 = none, 1 = primary, 2 = secondary, 3 = higher; floor type: 1 = natural, 2 = rudimentary, 3 = finished.
Data were available for age, height and body weight for 160 000 children under five years of age. Within each of the 24 discrete values of the poverty index and for eight study regions as defined by WHO, we calculated average height-for-age and weight-forage Z-scores. Regression models were used to explore the relationship between these parameters and poverty.
The first factor had an Eigenvalue of 0.98. The factor coefficients were 0.277 for educational status, 0.373 for electricity and 0.319 for construction material. There were 24 discrete values of the poverty index, ranging from -0.85 (poorest) to 1.55 (richest).
In weighted linear regression models, childhood malnutrition within each region was well predicted by the poverty index. In a model containing the poverty index plus dummy variables for each region, the height-for-age Z-score decreased by 0.45 (95 percent confidence interval: 0.42-0.50) and the weight-for-age Z-score by 0.39 (0.35-0.42) for a one unit increase in poverty index. Inclusion of interaction terms between WHO regions and the poverty index increased the model fit only slightly (the height-for-age model R2 increased from 0.87 to 0.89, and the weight-for-age model R2 increased from 0.92 to 0.94). The coefficients for these interaction terms were mostly not statistically significant.
We have demonstrated a strong statistical relationship between an index of poverty and anthropometric indicators. This is not an unexpected finding; for example, short stature has previously been shown to be related to poverty rather than to genetic factors (Bustos et al., 2001). The size of the effect demonstrated here is remarkably consistent between WHO regions and quite substantial in public health terms. This finding supports a causal association between poverty and malnutrition. We propose that the distribution of undernutrition by poverty can be predicted based simply on knowledge of the overall prevalence of malnutrition and data on the distribution of poverty. These findings could contribute to assessment and improvement of the nutritional status of vulnerable populations. For example, our poverty index can be calculated for many countries in which direct measurements of anthropometry have not been made.
Blakely, T., Hales, S., Kieft, C., Wilson, N. & Woodward, A. 2002. The global distribution of risk factors by poverty: a report to the World Health Organization. Wellington, University of Otago.
Bustos, P., Amigo, H., Munoz, S.R. & Martorell, R. 2001. Growth in indigenous and nonindigenous Chilean schoolchildren from 3 poverty strata. Am. J. Public Health, 91: 1645-1649.
Filmer, D. and Pritchett, L. 1988. Estimating wealth effects without expenditure, data - or tears. Washington, DC, Development Economics Research Group.
Macro International. 2002. http://www.macroint.com
WHO. 2002. World health report. Geneva, World Health Organisation.
Harriet V. Kuhnlein, Suttilak Smitasiri,
Salome Yesudas, Salek Ahmed, Gopa Kothari,
Lalita Bhattacharjee, Li Dan and Zhai Fengying
Montreal, Quebec, Canada
Improving food security, nutrition and health status of indigenous peoples who reside in their homelands in rural areas requires a thorough understanding of the local environment and traditional food system known and used by the culture. Indigenous peoples often use unique food species and methods of processing, and the traditional knowledge required is often as endangered as the species themselves. Further, indigenous peoples can be the economically poorest and most disenfranchised part of many societies for whom attention to health status is a public health necessity, especially when rapid dietary change is in effect or anticipated.
This project, still in progress, has the short-term goal of defining a method for documenting indigenous peoples traditional food resources and the long-term goal of using these local resources for programme planning to ensure adequate diets and good nutritional status. Case studies to test the suitability of the method in Asia are represented by the co-authors and their team members in the following areas:
India: with the Bhil Tribals in Gujarat with guidance from Dr. Gopa Kothari and Dr. Lalita Bhattacharje, Child Eye Care Charitable Trust, Mumbai;
India: with Dalit women farmers in Andhra Pradesh with guidance from Salome Yesudas, Deccan Development Society;
Thailand: with the Karen People of Kan-chanaburi Province with guidance from Dr. Suttlak Smitasiri and team, Mahidol University, Salaya;
Bangladesh: with the Mogh Tribals and Nayakrishi farmers of Coxs Bazaar District with guidance from Dr. Salek Ahmed, Policy Research for Development Alternative (UBINIG), Dhaka;
China: with the Miao National Minority of Sichuan Province with guidance from Dr. Li Dan of the Chinese Center for Disease Control and Prevention, Beijing, and Dr. Zhai Fengying of the Institute of Nutrition and Food Hygiene, Chinese Academy of Preventive Medicine, Beijing.
A workshop held in March 2001, in which the case study leaders participated, developed the draft protocol that includes the participation of six interdisciplinary specialists in each case study. These are: (1) the indigenous local leadership, (2) herbarium and zoological identification specialist, (3) food culture specialist, (4) food analysis laboratory and data specialist, (5) food and dietary database specialist and (6) local agricultural and/or environment protection specialist. The protocol includes five basic steps of collecting qualitative and quantitative data to document the food system of indigenous people and to identify local foods suitable for promotion to improve health. The five steps are: (1) consulting with experts and gathering background data, (2) identifying traditional food lists, (3) seeking the scientific parameters of traditional food, (4) understanding the cultural contexts and dietary evaluation for food use and (5) planning interventions to improve nutrition using local traditional food, if warranted. Focus is being made on macronutrients (protein, carbohydrate, fat), energy and five micronu-trients (vitamin A, iron, zinc, folic acid and vitamin C) in nutrition of women and young children.
As of July 2002, the case studies are still gathering data and testing the method in the respective regions. At this point, we know that there are 90-276 food species documented in the five areas. Table 1 shows the number of species, the number still to be identified and the number requiring analysis for nutrient content. Folic acid and zinc are the most frequently missing nutrient data in the world literature for these species.
TABLE 1. OVERVIEW OF TRADITIONAL FOOD SPECIES IN FIVE CASE STUDIES
Number of species
To be identified
To be analysed
Nutritional status of women and children in the five areas varies and has yet to be completely assessed and reported. Similarly, the feasibility of using traditional food for improving nutritional status and strategies for doing so are still being determined.
All case studies report that the methods applied up to this point in time have been successful. However, there are several barriers to good data collection evidenced in the case studies. For example, funding delays result in project delays and can interfere with good interdisciplinary collaboration. Participatory techniques and research agreements or informed consent are usually verbal, and the importance of complete and thorough consultation in advance of project initiation is considered to be essential. There are an unexpectedly high number of food items with missing data on scientific identification or analytical data for nutrient contents. Another important methodological issue relates to the requirement for simultaneous work in multiple languages for adequate data collection and interpretation. Regularly, translation is required from the local language to the provincial language, then to the national language and finally to English for report writing. This requirement results in additional personnel and costs. The use of the traditional food system data for incorporation into the analysis of dietary information requires an inexpensive, user-friendly method that can be applied at the grass roots level for which unique local species and local language can be inserted. For example, a successful method was developed in the Bhil case study that used calculation by hand of dietary nutrients.
In conclusion, working with indigenous peoples food systems requires recognition that indigenous peoples universally prefer health promotion activities using their local food. Implementation of the method stressed in all cases that grass roots local community work can be successful when using interdisciplinary and participatory methods. Both qualitative and quantitative data are needed to evaluate the potential of the local food system for improving nutritional status. Understanding the unique local species and varieties of food often requires new identifications, food analysis and dietary analysis methods that are appropriate to the setting in which they are applied.
Lalita Bhattacharjee, Gopa Kothari,
Vidya Ramaswamy, Harriet Kuhnlein and Biplab Nandi
FAO Regional Offi ce for Asia and the Pacific
Understanding the food and dietary patterns of indigenous populations often poses unique challenges. The tribals in India represent a good example of indigenous populations with the vast diversity in their culture, tradition and environment that particularly influences their food systems, food practices and patterns and nutritional status.
A study on 187 Bhil tribal households in Western Gujarat, western India was undertaken to understand their traditional food patterns and dietary profiles. The study group was divided into six categories based on their landholdings. Most of the foods in their diets included those obtained from the jungle and local harvesting.
Ninety-seven commonly consumed foods were identified as part of the Bhil traditional food list. Twenty-nine foods were procured from the jungle, 26 were cultivated and 14 were hunted. The nutritive values of the majority of the foods were calculated using the Indian food composition tables. The nutritive values including the macro- and micronutri- ent content of 16 locally grown vegetables and fruits were not reported in the Indian food composition tables and needed to be analysed, while other foods needed to be identified scientifically before beginning the analysis for nutrient content. The mineral content for calcium, iron, copper, zinc, magnesium and phosphorus of five traditional foods - junglikhand (a type of wild tuber), mokha leaves (tough green leaves), vasarta (a type of mushroom grown on the bamboo tree), teruna leaves (a type of tuber leaves) and doli mahuda seeds (fruit seeds) - were analysed at the National Institute of Nutrition in Hyderabad, India. Only two of these traditional foods, namely, doli mahuda fruit seeds (Bassia latifolia) and teruna leaves (Colocasia/Chlorophytum) have been scientifically identified; identification of the other five is under way. Dried vasarta (mushroom), as it is commonly used, has a protein content of 20 g per 100 g and can serve as an inexpensive and potential source of protein in the Bhil diet.
Data on the community food system were gathered through key informant household interviews, during which information was also collected on the seasonality of use, procurement and cost of production. The community organizers were trained by the professional dietetic study team to undertake community-based dietary data collection. Household and individual dietary intake data were obtained by use of the food record and 24-hour dietary recall methods. Food frequency methods were used to collect information on the frequency of consumption of each food item from a list of food items in reference to a specified period during the previous month or week during the season of study. A semiquantified questionnaire was used that incorporated questions on portion size of each food item using household measures. Individual dietary data were collected for 40 pregnant women, 124 lactating mothers and 15 preschool children. The food consumption data obtained were converted to raw weights, and the nutrient intakes were calculated manually using the Indian Nutritive Values and Food Composition Tables of the Indian Council of Medical Research.
In general, the analysis of food consumption showed that rice was the most commonly consumed cereal among all income groups, followed by ragi (millet), more popular among the lower income groups. Of the traditional food list, 18 types of meat, poultry and fish were commonly consumed because of their easy availability, followed by 13 different types of pulses and legumes and 10 types of green leafy vegetables. Fish, an integral part of the diet, was consumed by 27 percent of households, and animal foods (meat and poultry) were consumed by 19 percent. Because of their easy availability from the nearby jungle, fruits were consumed in large quantities but showed a seasonal consumption pattern. Eight kinds of typical fruits were collected from the jungle, while mango and papaya were grown by some of the households.
The use of traditional foods was found to vary with season. Out of 78 traditional foods, 23 including six varieties of fish and five types of green leafy vegetables were typically consumed during the monsoons. Sixteen traditional foods were consumed during the summer, but very few were taken during the winter. Mango in particular was a favourite among the families during the summer, as well as green leafy vegetables like cowpea leaves and meat such as deer. Typical types of wild roots and tubers were gathered from the jungle to satisfy the food needs of households, and these were commonly eaten during the monsoon and winter season. Twenty foods were consumed throughout the entire year.
An analysis from the 24-hour dietary recalls showed the inter-relationship between the top ten foods rich in carotene, iron and vitamin C taken from the traditional food list and the consumption pattern by the subjects. It was noted that rich sources of carotene such as drumstick and fenugreek leaves were eaten by only 1 percent of the study group, while other fairly good sources such as fruits and other vegetables were consumed by 3-6 percent. Among the dietary sources of iron, green leafy vegetables were consumed by 0-2 percent of the households, while fish and crab were consumed by 3-6 percent. Among the vitamin C-rich foods, seasonal fruits had a 1-2 percent consumption pattern, while tomatoes, a fair source of vitamin C, showed the highest consumption of 6 percent.
Micronutrient-rich foods were consumed on average once or twice a week by children 1-6 years of age and once to three times a week by pregnant and lactating women. The dietary energy and protein intakes of pregnant women and preschool children largely failed to meet the recommended daily allowances.
Within the context of the Bhils traditional food system, there exists a high potential to promote the usage of a variety of micronutri-ent-rich foods in the diet. This would help to increase dietary diversity and would serve to meet their micronutrient requirements. Community-based, integrated nutrition interventions facilitated by social mobilization strategies emerge as potentially promising ventures to be undertaken.
Kate Ogden, Sylvie Montembault, Caroline Wilkinson
and Mija Tesse Ververs
Food Security Department, Nutrition Department, Action Contre la Faim
The relevance of a spatial and integrated analysis to understand the underlying causes of malnutrition was presented by Action Contre la Faim (ACF) with a focus on a case study in an emergency context: that of the Kailahun district in the Eastern Province of Sierra Leone.
ACF has been operational in Sierra Leone for a decade in the ever-changing context of that country. One of the key issues during this time has been the continued movements of population fleeing the terror of war. April and May 2001 saw a large influx of people returning to their homes in the district of Kailahun, either from their displacement in Sierra Leone or from their period as refugees in the neighbouring countries of Liberia and Guinea. Coupled with this was an influx of refugees seeking refuge from the continued insecurity in Liberia. ACF attempted to understand more fully the humanitarian situation of these different population groups including, as well, those who had been residents in the area throughout the war.
Background to the project
The project had, as its base, a number of assumptions, which were as follows:
the increasing volume of information in increasingly complex contexts of intervention;
the limiting aspects of existing tools;
the increasing importance of external communication and quality information-backed positioning.
The project is in a phase of continued development, but when completed, the final structure for the integrated geographic information system (GIS) will comprise the water and sanitation database that includes water and sanitation works, the database of the nutrition department including beneficiary tables from the nutritional care centres, a food aid database with information from beneficiary registers, the food security database including food-security-specific contextual tables and market price tables and finally contextual tables common to all domains.
Manipulation of these data and transfer into a map format (MapInfo) allows a spatial vision of the raw data.
The type of information collected for the project, more specifically for this case study, falls into the following categories:
context/programme static data: collected at the village level and chiefdom level;
context dynamic data: collected at the village level and chiefdom level;
water and sanitation dynamic data: collected at the household level, village level and chiefdom level;
nutrition/health dynamic data: collected at the household level, village level, health structures level (and methodology agreed upon) and chiefdom level;
food security dynamic data: collected at the household level, village level and chief-dom level.
The original data were collected through a variety of sources and triangulated before being entered into the database. Key informants, focus groups and individual households contributed to the information collection, which, where necessary, was cross-checked at the national level. This information collection was rapid, sometimes over just a few days, owing to the emergency context and the need for a rapid reaction. Nutrition surveys carried out by ACF provided the nutrition rate data, while the crop and water information was obtained through field assessments and from ongoing fieldwork conducted by the food security and water and sanitation departments, respectively.
The information presented in the oral contribution to the Symposium was kept simple. In Kailahun, in the ACF chiefdoms of operation, among much more detailed information, the following were collected and analysed:
the population by sectors: refugees, returnees and hosts (residents);
the agricultural production by major crop: upland rice, swamp rice and groundnuts;
the agricultural production per family and per crop;
the malnutrition rate;
the water supply coverage.
Classically, this information is analysed using the nutritional causal analysis as a framework to understand the immediate, underlying and basic causes of malnutrition, and to make programming recommendations as appropriate for tackling one, two or all of the causes. The GIS system also allows ACF to exploit fully this analytical framework that is at the base of all its assessments, through entry of the information into the database and subsequent mapping of the integrated information collected. In the case of Kailahun, this system allowed ACF to see clearly chiefdom differences in production and malnutrition that related significantly to the concentration of population sectors in that chiefdom. More precisely, the map showed that in one chiefdom of Kailahun, there was a high proportion of returnees with limited crop production, high numbers of people sharing the water points and relatively high levels of both severe and moderate malnutrition, revealing that the causes of malnutrition were not only food-related but also linked to water access, and that returnees were more affected in that area. As mentioned earlier, this is very much simplified; far more complex analysis can be, and is, undertaken with this system, providing links between the information that otherwise would be more complex to analyse.
Programming outcomes are entirely dependent on the analysis of data collected in detailed assessments. In the case of Kailahun, the interventions in those areas revealing the highest needs shown clearly on the maps included:
rehabilitation of water wells,
nutritional feeding centres, and
recommendations on food aid and on seeds and tools inputs.
Further work on, and prospects for, integrated spatial analysis in ACF
Development of the GIS database is itself dynamic, and future tasks includes:
Work on methodology by the technical department on standard information qualification for systematization of the above information collection. While the databases for water and sanitation and the nutrition information have been finalized, work is still to be carried out on standardization of food security information, this being very much qualitative- and context-related. To this end, an expatriate has been appointed, combining work in the field with that in headquarters to complete this complex but crucial step.
Training of staff.
Development of the integrated Access database.
Test of the defined rapid assessment methodology and new information collection system.
While ACF is not the only organization operating geographic information systems (others collecting information for early warning systems include FIVIMS, FEWS, Global Information and Early Warning System, etc.), what differentiates the work of ACF is that the information collection is done at the micro (household) level or village scale and is pro-gramme-related. It is therefore much more precise, collected often where ACF is the only actor and, through wider assessments, can be directed to potential areas of intervention for ACF. This tool is also a complement to information collected and analysed on a macro scale by the global early warning systems as a result of this specificity.
Internally, the integration of information from all the technical departments within ACF (nutrition, food security, water and sanitation, and medical) using this system allows not only a comparison in time and space within an area, region or country, but also a between-country comparison. More importantly, it gives ACF the ability to understand the underlying causes of malnutrition and to obtain more precision in programming to address these underlying causes. It helps to channel information collection to those immediate factors that are useful for ACF in a programming aspect, helping the organization to find the best way to be reactive: identifying needs and targeting the most vulnerable populations with pertinent programme design to address these needs in a better way and thus providing quality humanitarian responses.
 Case study as part of
FAO Project on Documenting Traditional Food Systems of Indigenous Peoples in