This is the old United Nations University website. Visit the new site at http://unu.edu
PER PINSTRUP-ANDERSEN
Cornell Food and Nutrition Policy Project, Cornell University,
Ithaca, New York, USA
MARITO GARCIA
International Food Policy Research Institute, Washington, D.C.,
USA
INTRODUCTION
Efforts to target and assess the nutritional effects of food and nutrition programmes suffer from the lack of a unique or commonly agreed-upon indicator of nutrition impact. Anthropometric measures such as weight- and height-for-age are frequently used in evaluations of nutrition or food distribution programmes, while the impact on the food acquisition of calorie-deficient households is more often used as a proxy for the nutritional effects of food and agricultural policies and programmes. The impact on the food consumption of malnourished individuals is used in some cases, but much less frequently because it is more difficult and costly to obtain reliable data on an individual's food consumption than to obtain his or her anthropometric measurements, or data on household food acquisition.
Attempts to establish a direct causal link between individual anthropometric measures and the effects of nutrition programmes have been disappointing because the intermediate steps and relationships influencing or determining the nutrition effects often are ignored. Results of the evaluations of many nutrition programmes have thus been inconclusive. It is quite possible that in many cases the actual impact on nutrition went undetected but could have been detected if the most important intermediate relationships, such as the impact of the programme on household and/or individual food consumption, had been taken into account. Changes in household food acquisition? similarly, may be a poor proxy for the impact on an individual member's nutritional status. This is because the intra-household food distribution patterns are ignored, as is the impact of changes in an individual's food consumption on his or her nutritional status.
This paper addresses the question of whether or not collecting data on individual household members' food consumption is likely to greatly enhance the reliability of efforts to target food and nutrition programmes to households with malnourished individuals, when malnutrition is measured in terms of calorie intake relative to requirements. The paper specifically seeks to answer three related questions:
Answers to these questions should assist in assessing whether the increased cost of collecting data on individual food consumption can be justified as part of efforts to target nutrition programmes more effectively. This paper should contribute to a better understanding of the trade-offs between cost and reliability of the collection of household- versus individual-level food-consumption data.
SAMPLE DESCRIPTION
The analysis presented here is based on data collected as part of a consumer food-price subsidy scheme in three regions of the Philippines' (Garcia and Pinstrup-Andersen, 1987). Data on household-level variables were collected twice during 1983 from a cross-section of 840 households. Anthropometric data on pre-school children were collected monthly from the same households over a 12-month period. Data on the food consumption of individual household members were also obtained twice, but only from a subsample of 134 households (table 1).
The sample was selected from a population of low-income households with a high probability of malnutrition. The overall calorie adequacy for the sample was 70 per cent, as compared to 89 per cent for the Philippines as a whole. About one-third of the sample pre-schoolers had second- or third-degree malnutrition, as compared to 17 per cent for the Philippines as a whole (table 2). Estimated calorie adequacy was low among sample individuals regardless of age, sex, and region. Calorie adequacy was calculated as actual consumption relative to recommended daily allowances. The latter are shown in the Appendix to this chapter. Severe calorie deficiencies were found among pre-schoolers, particularly girls. Other population groups, including adolescent girls and pregnant or lactating women, were also seriously affected (table 3). This table indicates unequal food distribution within the household. Two independent studies, one in rural Laguna (Valenzuela, 1977), and the other in urban Manila (Aligaen and Florencio, 1980), also reveal inequality in food adequacy between adults and children.
Table 1. Sample size by region
Total sample households | Households in subsample with individual consumption |
||||
Region | Round 1a | Round 2 | Round 1 | Round 2 | |
Abra | 240 | 228 | 40 | 38 | |
Antique | 360 | 352 | 60 | 60 | |
South Cotabato | 240 | 220 | 40 | 36 | |
Total | 840 | 800 | 140 | 134 |
a. The first survey round took place during May-June 1983 and the second during September-October 1983.
Table 2. Percentage of sample pre-schoolers falling into second- and third-degree malnutrition on the basis of weight-for-age
Region | Percentage second and third degree |
Abra | 25.5 |
Antique | 34 4 |
South Cotabato | 37.2 |
Total sample | 32.5 |
Philippines (FNRI, 1982) | 17.2 |
Sources: IFPRI/National Nutrition Council Survey,1983; Food and Nutrition Research Institute, Second Nationwide Nutrition Survey of the Philippines (NSTA, Manila, 1982).
Table 3. Estimated calorie adequacy (in percentage of RDA) of various sex- and age-groups of sample by regiona
Population group | Abra | Antique | South
Cotabato |
|
Fathers | 75.5 | 83.9 | 86.8 | |
Pregnant women | 70.1 | 63.3 | 46.4 | |
Lactating women | 67.1 | 71.1 | 62.9 | |
Adults | ||||
Male | 75.8 | 81.3 | 83.7 | |
Female | 77.6 | 81.9 | 71.5 | |
Adolescents | ||||
Male | 62.2 | 65.0 | 57.2 | |
Female | 53.6 | 64.9 | 50.4 | |
Schoolers | ||||
Male | 56.4 | 72.4 | 66.3 | |
Female | 56.8 | 66.6 | 62.7 | |
Pre-schoolers | ||||
Male | 57.2 | 69.7 | 63.7 | |
Female | 49.5 | 62.3 | 61.8 | |
Average, household | 65.7 | 73.6 | 70.1 | |
Overall average | 70.4 | |||
The Philippines (FNRI) | 88.6 |
a. See Appendix to chapter for RDA for calories for Filipinos Sources IFPRI/National Nutrition Council Survey,1983; Food and Nutrition Research Institute Second Nationwide Nutrition Survey of the Philippines (NSTA, Manila, 1982).
HOUSEHOLD ADEQUACY AS A PROXY FOR ADEQUACY AMONG HIGH-RISK INDIVIDUALS
The efficiency of estimates of household-level calorie adequacy as a proxy for the calorie adequacy of high-risk household members, as shown in tables 4 and 5, will be discussed in an attempt to answer question 1 above. Reliance on household calorie adequacy as an indicator of the degree of calorie adequacy of pre-schoolers introduces large errors. Only 39 per cent of the households with pre-schoolers consuming less than one-half of their calorie requirements would be captured by a programme targeted on households below 50 per cent calorie adequacy (24 per cent of all households). A programme targeted on households consuming 60 per cent of requirements or less would leave out 61 per cent of the pre-schoolers who consume 60 per cent or less, including 29 per cent who consume less than 50 per cent of their requirements (table 4).
Table 4. Relationship between calorie adequacy of households and those of pre-schoolers within the households
Calorie adequacy of pre-schoolers (%)a | Household calorie adequacy (%)b | |||||||
0-50 | 51-60 | 61-80 | 81-100 | Above 100 | Total | |||
0-50 | Number of households | 43 | 17 | 42 | 7 | 2 | 111 | |
Percentage of sample | 15.69 | 6.20 | 15.33 | 2.55 | 0.73 | 40.51 | ||
Row percentage | 38.74 | 15.32 | 37.84 | 6.31 | 1.80 | |||
Column percentage | 64.18 | 34.69 | 40.00 | 16.28 | 20.00 | |||
1-60 | Number of households | 14 | 10 | 18 | 1 | 1 | 44 | |
Percentage of sample | 5.11 | 3.65 | 6.57 | 0.36 | 0.36 | 16.06 | ||
Row percentage | 31.82 | 22.73 | 40.91 | 2.27 | 2.27 | |||
Column percentage | 20.89 | 20.41 | 17.14 | 2.33 | 10.00 | |||
61-80 | Number of households | 8 | 17 | 31 | 16 | 1 | 73 | |
Percentage of sample | 2.92 | 6.20 | 11.31 | 5.84 | 0.36 | 26.64 | ||
Row percentage | 10.96 | 23.28 | 42.46 | 21.92 | 1.37 | |||
Column percentage | 11.94 | 34.69 | 29.52 | 37.21 | 10.00 | |||
81-100 | Number of households | 2 | 4 | 11 | 12 | 3 | 32 | |
Percentage of sample | 0.73 | 1.46 | 4.02 | 4.38 | 1.09 | 11.68 | ||
Row percentage | 6.25 | 12.50 | 34.37 | 37.50 | 9.37 | |||
Column percentage | 2.98 | 8.16 | 10.48 | 27.91 | 30.00 | |||
Above 100 | Number of households | 0 | 1 | 3 | 7 | 3 | 14 | |
Percentage of sample | 0.00 | 0.36 | 1.09 | 2.55 | 1.09 | 5.11 | ||
Row percentage | 0.00 | 7.14 | 21.43 | 50.00 | 21.43 | |||
Column percentage | 0.00 | 2.05 | 2.86 | 16.28 | 30.00 | |||
Total | Number of households | 67 | 49 | 105 | 43 | 43 | 274 | |
Percentage of sample | 24.45 | 17.88 | 38.32 | 15.69 | 3.65 | 100.00 |
a. Calorie consumption by household members
between 12 and 84 months of age divided by the sum of the RDAs
for these household members and multiplied by 100.
b. Total household calorie consumption divided by the sum of the
RDAs for all household members and multiplied by 100.
Source: IFPRI/National Nutrition Council Survey in Abra, Antique, and South Cotabato, the Philippines. 1983.
Table 5. Relationship between calorie adequacy of households and those of pregnant and lactating women within the households
Calorie adequacy of pregnant and lactating women (%)a | Household calorie adequacy (%)b | ||||||
0-50 | 51-60 | 61-80 | 81-100 | Above 100 | Total | ||
0-50 | Number of households | 17 | 2 | 2 | 1 | 0 | 22 |
Percentage of sample | 11.72 | 1.38 | 1.38 | 0.69 | 0.00 | 15.22 | |
Row percentage | 77.30 | 9.10 | 9.10 | 4.50 | 0.00 | ||
Column percentage | 39.50 | 6.90 | 3.60 | 5 60 | 0.00 | ||
51-60 | Number of households | 10 | 10 | 16 | 0 | 0 | 36 |
Percentage of sample | 6.90 | 6.90 | 11.03 | 0.00 | 0.00 | 24.80 | |
Row percentage | 27.80 | 27.80 | 44.40 | 0.00 | 0.00 | ||
Column percentage | 23.30 | 34.50 | 29.10 | 0.00 | 0.00 | ||
61-80 | Number of households | 12 | 14 | 23 | 4 | 0 | 53 |
Percentage of sample | 8.27 | 9.66 | 15.86 | 2.76 | 0.00 | 36.60 | |
Row percentage | 22.60 | 26.40 | 43.40 | 7.50 | 0.00 | ||
Column percentage | 27.90 | 48.30 | 41.80 | 22.20 | 0.00 | ||
81-100 | Number of households | 3 | 2 | 10 | 8 | 0 | 23 |
Percentage of sample | 2.07 | 1.38 | 6.90 | 5.52 | 0.00 | 15.80 | |
Row percentage | 13.00 | 8.70 | 43.50 | 34.80 | 0.00 | ||
Column percentage | 7.00 | 6.90 | 18.20 | 44.40 | 0.00 | ||
Above 100 | Number of households | 1 | 1 | 4 | 5 | 0 | 11 |
Percentage of sample | 0.69 | 0.69 | 2.76 | 3.45 | 0.00 | 7.60 | |
Row percentage | 9.10 | 9.10 | 36.40 | 45.50 | 0.00 | ||
Column percentage | 2.30 | 3.40 | 7.30 | 27.80 | 0.00 | ||
Total | Number of households | 43 | 29 | 55 | 18 | 0 | 145 |
Percentage of sample | 29.60 | 20.00 | 37.90 | 12.40 | 0.00 | 100.00 |
a. Calorie consumption by pregnant and lactating household members divided by their RDAs and multiplied by 100.
b. Total household calorie consumption divided by the sum of the RDAs for all household members and multiplied by 100.
Source: IFPRI/National Nutrition Council Survey in Abra, Antique, and South Cotabato, the Philippines, 1983.
Household-level calorie adequacy is a more efficient indicator of the calorie adequacy level of pregnant and lactating women (table 5). Thus, a programme targeting households with atrisk women and calorie adequacy levels of 50 per cent or less would capture 77 per cent of the pregnant and/or lactating women whose calorie adequacy levels are 50 per cent or lower.
The problem of excluding such a large number of target households through a targeting approach based on household-level calorie adequacy is compounded by the inclusion of a correspondingly large number of non-target households. Thus, 28 per cent of the households benefiting from a xprogramme targeting households with preschoolers and with a household-level calorie adequacy of 60 per cent or lower do not include any pre-schoolers whose calorie adequacy is this low. Similarly, in more than half of the households with pregnant and/or lactating women which would be included on the basis of household-level criteria, these women's calorie adequacy would be actually higher than 60 per cent.