Asia Pacific J Clin Nutr (1992) 1, 131-152

Supplementary feeding in programmes in developing countries: lessons of the eighties

PART 1: FINDINGS OF THE REPORT (click here for Part II)

Hossein Ghassemi

Consultant on international nutrition and health, former Senior Nutrition Advisor, UNICEF, New York, USA.

This extensive report reviews the important lessons learned during the 1980s on supplementary food distribution for the vulnerable groups in developing countries. These lessons may be useful in making such programs a more cost-effective option in narrowing the food/nutrient gap in intake among the program beneficiaries. This report follows a similar report published at the end of the 1970s by the author and George Beaton for UNICEF.

The primary focus in the study has been the food distribution among young children, particularly school children, and also in pregnant and lactating women. The data have been gathered through a comprehensive search of the literature, official reports and documents from several United Nations agencies, aid agencies, national and international institutions. As well as original research papers on theoretical and applied issues, reports on design and evaluation of specific programs in over 20 countries are studied.

Consideration of programme design examines objectives, nutrient/food gap, poverty reduction, malnutrition, mother and child feeding practices, foods, ration sizes, leakages, targeting, coverage, integration of feeding and health care. Potential and measured benefits are considered in the light of the reports published at the end of the 1970s and consequent analysis of work up until the end of the 1980s. Programme costs are documented. In a discussion on context and input, the functional significance of mild and moderate malnutrition is considered, together with diet and physical activity. The author offers some thoughts on future directions and highlights the need for further research.

Introduction

The purpose and scope

This report reviews the important lessons learned during the eighties on supplementary food distribution for the vulnerable groups in developing countries. Those lessons may be useful in making such programmes a more cost-effective option in narrowing the food/nutrient gap in intake among the programme beneficiaries.

In the late seventies, extensive reviews were made on the experiences of the previous decades in feeding the vulnerable groups. One of these reviews was prepared in 1979 by George Beaton and myself’ for the United Nations Children’s Fund (UNICEF) and the subcommittee on Nutrition of the Administrative Coordinating Committee of the United Nations (ACC/SCN).

Now, at the end of another decade, it was judged to be timely and necessary to revisit the field and assess the progress made during the 1980s. Therefore, this study was undertaken at the invitation of the Food Policy and Nutrition Division of the Food and Agriculture Organization of the United Nations (FAO) in Rome.

Our primary focus in this study has been on the food distribution programmes among young children, with selective reference to school children and pregnant and lactating women. The study has given particular attention to the issues of programme design, costs, benefits, and some aspects of implementation.

The database

The database for this study was prepared through a comprehensive search in the literature, gathering of official reports and documents from several United Nations agencies, some bilateral aid agencies, and national and international institutions involved and experienced in design and evaluation of programmes in nutrition and health, plus some original research reports related to the subject of this study. Altogether, our database consisted of the following four categories:

1. Reports on design and/or evaluation of specific programmes in over 20 countries during the eighties;
2. general reviews on this class of programmes and the lessons learned, published in the literature during this decade;
3. original research papers on some theoretical and applied issues related to the subject of this report published in this same decade; and
4. reports to the governments, international agencies and programme sponsors.

The initial documentation produced a fairly large body of information. The final selection was then made through careful screening. The database, in its final form, was limited to those reports offering new documented evidence on the aspects of design, costs and benefits of relatively large-scale programmes, particularly those with innovations in design and management as well as high quality analysis of the programme experiences. Also, pointed efforts were made in including the latest scientific information with strong bearing on the theoretical aspects of programme design, choice of indicators and impact assessment.

This study has intentionally excluded reports on many feeding programmes which have more or less continued with the practice of the previous decade. The body of information examined is by no means exhaustive, but it is sufficiently broad so that the judgements offered are reasonable and in perspective.

Critical issues in programme design

Programme objectives

Food distribution programmes in general have the implicit objective of improving the state of health and nutritional well-being of the beneficiaries. However, the programme documents often do not include clearly stated objectives. The important point is that such programmes are targeted on the undernourished, low income and vulnerable populations, and, hence, their objectives, naturally, include increasing the levels of food intake.

In reality, food distribution programmes are often instrumental in preventing deterioration of the nutritional conditions of the recipients. Otherwise, they concentrate on keeping third degree, and to some extent second degree, malnutrition under control. In the absence of clearly stated objectives, the programme evaluation format and the data from the experimental field trials have provided an indirect basis for some definition of the intended objectives, which often turn out to be improvement in health and nutrition. In this process, weight gain has been the most commonly measured indicator.

No one disagrees with the fact that food distribution is a direct form of assistance targeted on families and individuals who are undernourished. Therefore, the most logical programme objective seems to be an increase in food intake. If that is the case, the proper measures, or at least one proper measure, of impact is the extent to which the food/energy gap is closed or narrowed.

There is some concern among experts and advocates that the priority given to supporting supplementary feeding programmes has been on the decline within the international community. Such trends are seen as being influenced by various reports on the costs and effects of these programmes. In reality, the factors involved have been much broader and more complex.

During the 1970s, many leading figures in international development were quite disappointed with the inadequacies and inefficiencies of the "trickle down" approach in development. As a result, they strongly emphasized the need for a focus on a community-based, participatory approach to development, and also made serious pleas for priority attention to problems of scale and impact. The argument was that the scale of operations and size of coverage in international assistance projects needed to expand. As a result, selection of low-cost technology became a priority issue in programme design. Within the health sector, the primary health care approach became the answer to all these issues. This approach aims at extending universal, primary care at an affordable cost with emphasis on the use of low-cost technology in a participatory fashion².

It was in this climate and context that supplementary feeding programmes found it increasingly difficult to compete for priority support because they were seen as being costly for their effects and hardly lent themselves to a participatory process. Food is a valuable resource and it was the method of its use that was the cause for concern. In some instances, food distribution was seen as conducive to dependency, while the emerging focus was on self-reliance. In essence, these programmes were being seriously challenged for better design and implementation.

In the 1980s, a number of important developments have occurred in the field which will be discussed in depth later. For instance, research data began to show that at least some programmes provide economic benefits to the recipients.³ As a result, donors and programme administrators showed some interest in support of these programmes for income/resource transfer objectives.

In 1979 Beaton and Ghassemi1 proposed two options for improving programmes’ cost-effectiveness. The first was to accept a rather broad programme objective such as poverty reduction among the vulnerable families. This choice would be compatible with a somewhat less rigorous project design and limited implementation capacities. The second was a design appropriately targeted toward reaching the young children and providing them with food and nutrition care. In this approach, the primary focus would be on younger children, and the main purpose would be promotion of proper feeding practices during the weaning period as well as for pregnant and lactating women in the community. Both of these choices have been pursued in the 1980s with different results.

In reality the choice of programme objectives is very much influenced by the relative ease of measurement of outcome indicators. As a result, for two or three decades feeding programmes have looked for weight gain as proof of impact and much less attention has been given to change of intake pattern and intra-household distribution. Evaluation reports rarely present data on a programme’s effect on the existing gap in nutrient intake.

It is interesting to note that the initial evidence of a connection between diet and human health was, to a large extent, the product of our understanding of the nutritive value of foods and human food intake patterns. If we examine the data on nutrition assessment and outcome evaluation in the world today we see that the majority originate from various anthropometric measurements. When Gomez defined his well-known classification of malnutrition in 1956⁴, he could not possibly have imagined that he was putting on record one of the most widely applied techniques of the future. In fact, his approach was related to clinically diagnosed cases of malnutrition, and he did not intend it to be applied in community surveys⁵. The main reasons for such a wide application of the Gomez classification are the case of measurement, the low cost, and the usefulness of the method as a diagnostic tool within the medical/public health progression. If dietary intake assessment techniques while maintaining their reliability, it is quite likely that this indicator will find a wider use in programme design and evaluation. The main point in this discussion is to illustrate the need for more attention to the food intake issues in such programmes, in addition to the measurement techniques. As a part of this process, many programme planners have increasingly turned away from nutritional objectives during the 1980s and have shown interest in poverty reduction objectives. This is to a large extent due to the fact that programmes have met with difficulties showing impressive anthropometric responses.

The final point is that programme objectives, as currently defined, are only a statement of goals rather than objectives. They are highly non-specific and often lack a quantitative statements of purpose and a time frame.

Given the nature of the development over the past decade, supplementary feeding programmes could be designed and evaluated in relation to one or more of the following objectives.

Closing or narrowing the nutrient/food gap among the vulnerables. The review of several programmes in the 197()s has shown that most of them were designed to narrow rather than close the nutrient intake gap¹. Perhaps the WIC programme in the United States is one which has come closest to the level of closing those gaps⁶. In most evaluation reports, increased food intake is at best seen as an intermediate variable in relation to weight gain rather than a final programme outcome. This discussion should not in any way be interpreted as underestimating the usefulness of anthropometry. The point being made is that food intake as an indicator lends itself effectively to policy and programme design, while weight is only a statement of the situation without any specific indication of its determinants and, is therefore, a much less effective and useful tool in planning.

In simple terms, given the first objective, food distribution programmes are to be designed to address the problems of hunger among the undernourished populations. Under this scenario, change of food intake is an indicator of impact and for that purpose we need techniques that are easy to measure, low in cost, and with high reliability.

An important subset of objectives here is the control of seasonal undernutrition. There are two convincing pieces of evidence to justify this option. Prentice and co-workers⁷ reported in 1987 a major increase in birth weight when pregnant women in Uganda were given supplementary foods in the wet season when family food reserves were at their lowest level and the agricultural work at its peak. The same programme had difficulties in showing measurable gains in birth weight over the entire year. Also, Keilman and co-workers⁸ in the Narangwal study observed a strong seasonal effect in weight gain among children. Therefore, it appears that seasonal focus on food distribution can be highly cost effective if administratively feasible.

Poverty reduction among the vulnerable families and communities Many PL-480 Title II projects have been evaluated for both health and income effects. Several have shown considerable economic value. For instance, the economic value of a Title II family ration can be as high as US$ 200 in Africa. In Mauritania, one ration over a year is equivalent to 14% average per capita income and 33% of a rural nomad’s income. Another study by the United States International Developing Agency has shown that participants receive the greatest benefits from income mediated effects whereby families improve the quality of their diet.⁹

An interesting analysis and assessment of drought relief operations in Botswana¹⁰ has shown the practical feasibility of the use of food in addressing serious undernutrition as a short term objective, while at the same time promoting longer term objectives in poverty reduction, and increasing employment opportunities and investment in rural development. The dual objectives of poverty reduction and nutritional improvement present a rather impressive and much more development orientated approach.

Control of moderate and severe malnutrition and improving mother and child feeding practices During the 1980s, Indonesia, Thailand, and the State of Tamil Nadu in India have approached their problems of mother and child malnutrition through community-based programmes designed to increase awareness and education centred on growth monitoring and primary health and nutrition care, with intensive use of community support. In these programmes, food supplements are being used selectively. If properly designed and implemented, these programmes can, in the short term, effectively control moderate and severe forms of malnutrition. In the longer term they can aim at establishing proper feeding practices for the vulnerable, something which is at the root of nutritional problems in those societies.

A very important subset of objectives in this category could be to modify intra-household food distribution patterns in favour of women, young children and girls. Some recent findings in this area further confirm the previous data pointing to the fact that intra-household food distribution has strong economic and social roots and is often biased in favour of adult males in the family. Food distribution programmes with proper design can help in changing such patterns. In this context, the purpose is not to feed one member of the family, but to help families to help themselves with a more equitable and need-orientated sharing pattern.

The Tamil Nadu Integrated Nutrition Project (TNINP) in India,¹¹ supported by the World Bank, represents one of the effective models of the 1980s in design and implementation in this category. (There will be a detailed discussion on this programme later in this report.) The programmes in Indonesia¹² and Thailand¹³ represent sound design, but were less satisfactory in implementation.

A nutrition project in Iringa, Tanzania, supported by UNICEF and the World Health Organization¹⁴ presents another variation of the third category where self-reliance and education are the primary focus in addressing mother and child health and nutrition issues at the community level. This project’s content has a strong focus on primary health care services, growth monitoring, household food security, education, and mobilization.

As a final note, it should be pointed out that food aid supported programmes have not experienced major changes in objectives in explicit terms. However, they have changed in content and design. Increased awareness, on-going challenges to the more traditional patterns of design, and continuing process of review and refinement are likely to result in further modifications in the future.

Figure 1. Points of entry to intervene for child growth: options for selection of programme objectives¹⁵.

In an analysis of child growth in developing countries, Martorell and Habicht15 have shown in a schematic diagram (Fig. 1) the variables through which socioeconomic status influences child growth. This diagram shows alternative choices and objectives. For instance, targeting on change of food resources, food intake, infection control, income and feeding practices is included in this scheme. It is already known that benefits to be derived from different points of entry vary and have different costs.

Foods, ration size and leakages

In the early days, the common mode of food delivery was direct. The beneficiary had to come to a local feeding centre and young children had to be carried to such centres. Naturally, the centre had to be in reasonably close proximity and preferably associated with other community services, particularly health. Such an approach involves considerable costs in administration as well as opportunity costs for recipients, and in many cultures they may mark participants with an undesirable social stigma. This is particularly so if the participants selected are poor. For social and economic reasons, there has been increasing interest in take-home distribution programmes. This alternative in delivery is effective in reducing administrative and opportunity costs and minimizing the risk of social stigma. However, it is less effective for educational purposes and increasing the intake of the intended target.

Figure 2 shows the process of increasing net intake of an intended beneficiary through food distribution. Many studies have shown that critical links in this process include (a) ration size and kind; (b) leakages in the forms of sharing and substitution; (c) participation; and (d) the health status and appetite of the child. Ideally, programme design and a choice of delivery system should overcome unnecessary costs (social and administrative), minimize leakages, lend itself to community participation, and allow for integration with health care services.

Figure 2. Schematic view of distribution process: a framework for programme analysis. SF = supplementary food; TB = target beneficiary.

Table 1. Food intake and energy gap among the vulnerable in the State of Tamil Nadu (India)*.

Age group(years)	Energy intake (kcal)	Gap	Ration size**
12-13	600 (49)	600
24-35	790 (66)	510	370
36-47	960 (80)	440
48-59	1050 (71)	450
Pregnant women	1860 (56)	1450	530
Lactating women	1960 (53)	1740

*From Knudsen, (1981)³.
**Tamil Nadu Supplementary Feeding Programme. Figures in parenthesis indicate percentages of target ration.

Table 2. Food ration size in relation to energy gap among children.

Programme	Age (years)	Baseline intake	Mean energy gap (kcal/d)		Ration size	Energy gap closed (%mean)
		(kcal/d)	1-4 years	3-4 years		1-4 years	3-4 years
Colombia	1-4	978	464	647	305	65	47
Dominican Republic	1-4	877	565	733	337	59	45
Pakistan	1-4	1004	438	557	298	74	53
Tamil Nadu(ongoing)	1-4	811	631	682	273	43	40
Costa Rica	1-4	1033	409	554	591	144	106

From Beaton & Ghassemi (l982)¹.

Table 3. Ration size in selected programmes in supplementary feeding. Values in brackets are percent requirements.

				PL 480 Title II
			WICa USA(1983)	Philippinesb (1982)	Egyptc (1981)	Senegald (1984)	Upper Voltae (1981)
(1) Preschool children		Foods	Formula, milk powder, juice, eggs, beans, cheese	CSM	ICSM/oil	CSM/CM/ sorghum	Dried milk powder, corn meal/soy oil
	Energy (kcal)		898 (64%)	400(29%)	600	750 (70%)	770 (69%)
(2) Pregnant and lactating women			950 (40%)	19% energy requirement 30% protein requirement	Combined with children	750	-
(3) School children		Foods	-	Nutribun	WSB/flour cheese	-	Milk powder/ SFM/rice/soy oil
	Energy (kcal)			300	650		1500 (80%)

^aDwyer (1983); Rush el al (1989)⁴¹.
^bBlumenfeld et al (1982)⁴³.
^cRhoda et al (1981)⁶⁷.
^dEchenberg (1984)³⁵.
^eInternational Science & Technology Institute (1981)⁴⁴.

In this context, reliable data on baseline intake among the vulnerables and design or ration play a major role. Poor data on baseline intake may easily lead to a poor choice or target population and inadequate assessment of programme food needs as well as effectiveness. Looking at a wide range of food distribution programmes, they often provide 300-400 kcal per day for a young child 1-4 years old, while others provide up to 600 kcal (Tables 1 and 4). Most programmes reviewed have been designed to narrow the energy gap by 70% and have actually been effective in doing so by 10-25%.¹ A similar analysis on CARE-supported programmes showed that programme food rations were to narrow the energy gap by 62-83%.¹⁶ It is important to recognize that such programmes have primarily been aimed at narrowing rather than closing the intake gap. Furthermore, the ration size is often determined in relation to the mean baseline intake figures for a wide age range. Therefore, it cannot be effective in narrowing the gap among the recipients whose intake is at significantly different from the mean.

As an example, the energy gap among 1-4 year old children in Tamil Nadu, based on the estimates of mean intake for each age group, was in the range of 440-600 kcal per day (Table 1). Considering variations within each age group, The gap was then in the range of 300-700 kcal. The food distribution programme in this state provides 370 calories daily for 1-4 year olds which is closer to the lower end of the range in intake gap. The same observation is true for pregnant and lactating women in a much larger sense. This programme provides 530 kcal for pregnant and lactating women, and that roughly equals one third of the energy gap. Comparing the mean energy gap for 14-year-olds with that of the 3-4 year-olds shows that food distribution programmes intended to close the energy gap in Colombia, The Dominican Republic and Pakistan, would have to at least double the ration size (Table 2). On the other hand, in Costa Rica the ration covers 144% of the mean gap and 106% of the gap among older children. Actually, Costa Rica provides an excellent illustration in proper determination of ration size, i.e., 50% above the mean energy gap in the relevant age group. In spite of repeated criticisms over prevailing ration size, very little change has been introduced during the 1980s. However, a quick look at some PL 480 Title II supported programmes clearly shows a larger ration size being provided (Table 3). For instance, programmes in Burkina Faso, Egypt, and Senegal provide 600-750 kcal per beneficiary per day. The women, infants and children (WIC) food supplementation programme in the United States provides 898 kcal for young children (64% of requirement) and 950 for pregnant and lactating women (40% daily requirements). In all likelihood, recipients in the WIC programme have a relatively smaller food gap.

Energy costs of pregnancy serve as a good basis for determining ration size for pregnant women. According to WHO and FAO standards, 40 000 additional kcal are needed during pregnancy. Taking this as a basis for calculating the necessary ration size, it averages out as 150 calories per day during pregnancy. However, the assumptions behind this estimate do not always hold true. Sometimes pre-pregnancy diets are deficient and fat reserves are low. Young pregnant women do not reduce their physical activities. Lechtig et al.¹⁷ attempted to establish the level of energy supplementation needed to produce an increment of 200 grams in birth weight. They found that, for equivalent to 10 000 kcal, a birth weight increment in the range of 36-84 g will be achieved. Variation is due to physical activity and substitution.

Mora et al^.18 have shown that women receiving supplementary foods only during the first half of pregnancy did not produce higher birth weight babies as compared to control. Studies in India have shown that providing 200 kcal and 20 g protein daily during the last months of pregnancy with complete rest results in significant increase in birth weight.¹⁶

Habicht et al.¹⁹ have shown that the effects of supplementation will be seen in the most malnourished people. In a review of prenatal supplementation, Stein et al.²⁰ found that the greatest increment in birth weight was in the offspring of women who were undernourished before and during pregnancy. Also, Beaton & Ghassemi¹ have shown that children with the greatest weight deficit at the time of entry into the programme seemed to show the greatest benefit. Data on programmes like the Tamil Nadu Integrated Nutrition Project (TNINP) and others have further confirmed the fact that children with second and third degree malnutrition benefit most.¹¹

Table 4 shows the high extent of sharing and substitution in these programmes. (Sharing of food by non-target family members and reduced share of family food intake in substitution for the ration or the foods purchased in the house for the ration). In our previous estimates, leakages could be anywhere between 30 and 80% of the food distributed.¹

Table 4. Effects of attendance and leakage on ingestion of supplementary foods. Values in brackets are percent of target ration.

Programme	Ration size	Food collected	Food ingested	Food sharing % rations	Net increase intake		Intake increase
	(kcal)	kcal/d (%)		collected	%collected	%ingested	% gap
Colombia	305	153 (50)	83	46	63	116	2l.00
Dominican Republic	337	172 (51)	78	54	14	31	4.00
Pakistan	298	54 (18)	24	56	61	140	7.S0
Costa Rica	591	418 (71)	418	0	55	55	57.00
Tamil Nadu	273	235 (86)	235	0	17	47	18.00

Beaton, & Ghassemi (1982)¹

In 1979 Beaton & Ghassemi emphasized a major data gap on the fate of foods distributed.²¹ Since then very little new data has been reported in this area. Looking at the overall perspective in programme analysis and the nature of the database often used in research and evaluation reports, it is striking how little is said or asked about the consumption pattern of the beneficiaries as influenced by those programmes. Meanwhile, there is a strong tendency to perform an exhaustive analysis on anthropometry information. Some attempts have been made to explain determinants of sharing and substitution, but very limited new information has emerged on the possible reasons for very low net increase in intake. Some investigations have argued that the size of the ration should be increased in order to compensate for leakages.²² There seems to be some support for the argument that food distribution in the household reflects a protective pattern in the poor families and has very little to do with their relative physiological needs.²² In this context, additional food donated to the family is seen as a new resource for the family’s welfare rather than for changing the food intake pattern in favour of a child of a pregnant/lactating woman.^3-9 The often cited work of Mora and co-workers¹⁸ also supports this phenomenon to a large extent. In hope of overcoming the effects of sharing and displacement, they distributed food for all members of the family. The pregnant women only consumed 58% of the supplement (sharing 42%). There was also displacement of another 42% and the net increase in intake from the supplement was 16%. However, it has been shown in this study that those more in need, i.e., with lower intakes, eat more of the supplement. In general, the relationship between baseline intake and net ingestion of supplements seems to be much clearer in very poor households. On the other hand, economically better-off families substitute the donated food towards improving the quality of a family diet. In such cases starchy foods are sold for purchase of more expensive foods.²³

There is substantial evidence that intra-household food distribution has strong socioeconomic roots. There are reports that breadwinners and the economically productive family members are given a much higher preference as compared to the economically dependent and the vulnerables. In an interesting study, Nieves and coworkers¹⁸ explained the "sharing" phenomenon among participants in a food distribution programme in Guatemala. Looking at the social and economic roots of food distribution in the household, they found 38% of families favoured food sharing based on equity, while 18% preferred sharing based on individual needs, and 44% gave preference to those who contribute to the family economy. In this category, preference for adults over children and adult males over everybody else was significant. The most striking point is that none of the respondents gave preference to young children based on their needs. They concluded that education and instruction given through the food distribution programme was neither understood nor followed. In most probability, the rules for family food sharing are so deeply rooted that they are not subject to change as a result of education efforts in their current form.

The most effective options for minimizing sharing known so rare are: (a) choice of delivery system, and (b) kinds of food distributed.²² Naturally, the extent of sharing in on-site feeding programmes is minimal or almost nil. Also, handouts of special food formulas or snacks are less likely to be shared. Special foods for children are not usually perceived as foods for adults. However, the experience in Sri Lanka, reviewed in 1982, showed significant sharing of a special food for children called Triposha.²⁵ Special foods for children usually have a higher energy density, which makes it easier for the children to increase their intake. Low density foods are quite bulky and it is often impossible for younger children to ingest enough to satisfy their energy needs.²⁶ Unfortunately, most of the programmes reviewed have major limitations in reaching the preferred target of the very young.

The economic value of foods has a major influence on programme attendance. In Pakistan, cooking oil has been reported as being of high economic value. Therefore, it serves as a strong incentive for programme attendance. At the same time, oil is the most likely food to be shared.²⁷ Sometimes price fluctuations influence the situation. The milk distribution programme in Chile has been known to be a major factor in improving health and nutritional well-being of women and children for more than two decades. In recent years, price increases and economic difficulties have forced the Government to substitute rice for milk, which apparently has had a major influence on the intake of protein and calcium (R Mardones, personal communication).

During the 1970s, distribution of donated foods was seen as contributing to economic dependence and as a disincentive for local food production. There is considerable truth in the fact that food aid may result in dependency. However, given the relatively small size of food distribution for the vulnerable, the view has been taken that such risks are non-existent.¹ In the 1980s, some additional attempts in promoting the use of local foods were made. Such instances do not by any means reflect a significant shift towards encouraging local foods. But they deserve attention on their own merits. The two specific cases being cited in this respect are the primary health care programme in Thailand¹³ and the new innovative approach of the World Food Programme in Jamaica.²⁸

In Thailand, three aspects of the programme design are rather innovative. First, special foods for child feeding during the weaning period are prepared through village level technology and marketed through village level co-ops. This scheme: (a) Provides nutritionally sound foods for young children; (b) Facilitates reaching the weaning age child; (c) Uses local foods; (d) Provides a source of income and employment for women; (e) Is a self-sufficient, marketing enterprise and does not involve any dependency; and (f) Is well-utilized as an educational means. Second, the programme is strongly community-based and participation and attendance is quite high. Third, it has a targeted approach through growth monitoring and integrated primary health care services for women and children.

In Jamaica, the World Food Programme is exchanging donated food for cash which in turn will support a food stamp scheme through maternal and child health services. In this scheme, the beneficiaries will make food selection from a given list in any retail shop. This approach is more practical in an urban setting. It also, eliminates the costly operation of food transport within the country, encourages the use of local foods, relieves the health services from the burden of distributing foods, and improves the food selection choices for the families. However, it may increase the risk of corruption and decrease the opportunities for interaction between the mother and the health workers. Altogether, the advantages far outweigh the disadvantages. This programme is in its early stages of development and it is too early to expect results.

Figure 3. Schematic presentation of target population and beneficiaries in relation to the coverage and community population. From Timmons et al (1983)²⁹.

Targeting and coverage

Targeting is the process through which the programme beneficiaries are selected. In this process, the programme services are to be more directly and effectively made available to those most in need and more likely to respond. It could also focus on those who can benefit from it.

It is a central element in programme design, which often has considerable effects on programme costs and effects. As Fig. 3 shows, Targeting helps to select a programme’s target from the "at risk" population. This is required because in almost all cases there are not sufficient resources and capacity to serve the entire "at risk" population. Therefore, a large portion of those "at risk" is not the programme target. Also, programme beneficiaries are not entirely the same as the target. In reality, there are often beneficiaries who are neither the target nor at risk. In this situation, programme coverage is the percentage of the programme target population actually served.

In general, targeting is designed to select the programme target: (a) geographically, (b) by family unit, and (c) by individuals. Geographical targeting helps the programme’s service to concentrate in areas where economic and health indicators are relatively poor. At the family level, selection is often based on family income or food expenditure levels. Individual targeting is often based on age, eg, young children, and physiological state, eg, pregnancy and lactation and low weight gain among young children.

Table 5. Types or targeting and errors or exclusion^a.

Projects	Socioeconomic	Age cohort and anthropometry	Anthropometry	Age cohort 7-24 months	Age cohort 7-36 months
Kottar	14.4	17.8	20.1	9.4	20.1
Candelaria	4.7	6.1	6.4	5.3	7.5
Thailand	6.2	10.3	13.2	7.5	6.4
Esperanca	0.0	5.7	6.3	6.7	6.3
Sri Lanka	11.9	16.1	49.7	10.5	10.1

^aErrors of exclusion: the rate at which needy individuals are excluded from a programme due to a flaw in the targeting scheme. Source: Timmons et al. (1983)²⁹.

The most important benefit derived from targeting is the aggregate programme response due to inclusion of a higher proportion of the potential responders from the target population. It also results in substantial cost saving due to exclusion of non-responders or those not in need. Therefore, the most effective targeting strategy is the one which works well in including potential responders. By the same token, it should be effective in minimizing inclusion of non-responders. In doing so one needs to face the curative, preventive dilemma. Targeting, in a curative mode, primarily identifies the individuals already malnourished, while in a preventive sense it attempts to identify individuals most likely to become malnourished. Targeting, especially in the form of growth monitoring, can facilitate contacts between mothers and health workers and serve as an effective tool for education and prevention. Otherwise, socioeconomic methods of targeting are known to be more effective for preventive purposes as compared to anthropometric methods.

Targeting may not be politically popular. The governments may find it undesirable to restrict access to the public services, and the community may perceive targeting as too stigmatizing in the sense that access to the programme service is associated with being poor. Furthermore, they may face technical and budgetary constraints in the face of the realities where they operate. This is a typical example of conflicts between science and politics. Sometimes, targeting costs involve direct monetary costs associated with implementation as well as the costs of extending benefits to those who do not need them as well as those of excluding the needy population. Monetary costs of targeting within an ongoing service are relatively small. Sometimes the challenge is to find politically attractive methods for increasing programme effectiveness.

Table 6. Types of targeting and errors of inclusion^a.

Projects	Socioeconomic	Age cohort and anthropometry	Anthropometry	Age cohort 7-24 months	Age cohort 7-36 months
Candelaria	81.3	0.0	0.0	82.4	84.1
Esperanza	73.9	0.0	0.0	66.5	71.0
Thailand	53.6	0.0	0.0	51.1	54.2
Sri Lanka	24.2	0.0	0.0	24.6	23.6
Kouar	33	0.0	0.0	30.2	30.2

^aErrors of inclusion: the rate at which individuals without need are included in a programme due to a flaw in the targeting scheme. Source: Timmons et al. (1983)²⁹.

Table 7. Errors in targeting in relation to prevalence of malnutrition.

	Prevalence of	Exclusion errors		Inclusion errors
Project	malnutrition	Geographical	Anthropometry	Geographical	Anthropometry
Kottar (1976)	52.5	0	20.1	33.2	0
Candelaria (1968)	11.7	0	6.4	83.4	0
Thailand (1973)	36.8	0	13.2	55.5	0
Esperanca	16.1	0	6.3	76.6	0
Sri Lanka	42.0	0	49.7	22.0	0

Source: Timmons et al (1983)²⁹.

A very important aspect of targeting strategy is its errors of exclusion and inclusion, inclusion being the rate at which individuals without need are included in the programme, and exclusion being the rate at which needy individuals are excluded from the programme. Both errors are primarily due to flaws in the targeting scheme.

Ideally, targeting should minimize errors arising from admitting those without risk, and excluding those with need. In practice, no such strategy exists. In nutrition programmes the definition of a target unit can be a broad geographical area or as specific as families and individuals. In an interesting study, Timmons and coworkers²⁹ have examined the relative effectiveness of various targeting strategies based on carefully constructed longitudinal data sets from supplementary feeding programme in five countries. In this study the risk population consists of preschool children, and pregnant and lactating women. In effect, nutritional vulnerability is defined as being a function of age and physiological state. The effectiveness in targeting is measured in relation to errors of inclusion and exclusion. They have shown that practically all targeting strategies, no matter how well-implemented, make major errors of exclusion (Table 5). These are children excluded who were shown to become malnourished in a longitudinal follow-up.

Table 6 shows the very high probability of including beneficiaries without need. In the projects reviewed by Timmons, somewhere between half to two-thirds of the children were not needy. This phenomenon explains how the extent of measured benefits at the population level can be diluted.

Table 7 shows that more than 80% of the children included in the project in Candelaria, through socioeconomic or age cohort methods of targeting, were not needy. This is a case where the prevalence of malnutrition was only 11.7 per thousand in 1968, while errors of exclusion in these projects were relatively low. On the other hand, in Sri Lanka and Kottar, with higher prevalences of malnutrition, errors of inclusion are relatively smaller. Therefore, it is generally agreed that targeting is relatively less effective in very poor societies where prevalence of malnutrition is high. Furthermore, targeting for age groups is more effective for preventive modes of intervention, while targeting for anthropometry, especially if carried too far, tends to become more curative in nature. However, it has been shown that regular monitoring of a child’s weight can prevent children from becoming malnourished. Growth monitoring, if properly done, is effective in treating second degree and preventing third degree malnutrition, particularly when combined with selective primary health care services and feeding.

Table 8. Suitable modes of targeting in various socioeconomic settings.

Socioeconomic setting	Prevalence of malnutrition	Outreach and quality of material logistics	Community level infrastructure and organization	Suitable mode of targeting
Developing with large pockets of poverty	+++/+	+++	+++	Anthropometric (individual)
Deprived and improving	++/+++	++	++	Geographical and anthropometric
Severely deprived and slowly improving	+++	+	+	Geographical Age Physiological

+++ Very high. ++ Good/medium. + Poor/low.

Effective implementation of targeting strategies is very much a function of the programme’s organization and staff capacity and motivation. In a well-equipped and staffed programme, marginal costs of targeting are relatively small. However, in settings where logistics, technical and management capacity at the local level are limited, individual targeting is likely to be costly, ineffective and unnecessary. In circumstances such as remote rural areas, geographical targeting is in all likelihood preferred. On the other hand, in situations where relative levels of development in logistics and community level management, operational capacity, and community participation are strong and prevalence of malnutrition varies between classes and regions, then individual targeting is likely to be much more costs-effective (Table 8).

It is also important to note the inverse relation between targeting and coverage. A strict mode of targeting is likely to exclude a large portion of responders (exclusion error), and, therefore, will reduce coverage (percent of target served). As a result, the programme costs will be reduced and, therefore, cost-effectiveness will increase. Conversely, relaxing targeting will result in better coverage, higher errors of inclusion, lower errors of exclusion, higher costs and lower cost-effectiveness.¹

Therefore, the choice on targeting is not an entirely technical issue. It is a programme design choice which will have to be weighed against a number of rather important factors, some of them highly cultural or political, and sometimes becoming a choice between now and the future. It would have been very useful if the relative costs of errors of inclusion and exclusion had been estimated in the Timmons study. Such analysis would have been extremely valuable in understanding the comparative advantages of various targeting strategies, particularly in relation to coverage, impact, and the potentials for reaching very young children, as well as the relative effectiveness for long-term versus short term programme objectives.

Some well known investigators like Gopalan²⁶ have argued against individual targeting in very low income communities in India. He is correct in arguing that growth monitoring is desirable only where local possibilities for education, food supplementation and interaction between mothers and the health workers exist in a satisfactory manner. Gopalan is much more in favour of geographical targeting in poor communities. This is to a great extent in agreement with findings of Timmons and associates.

The Tamil Nadu Integrated Nutrition Project in India¹¹ has proven to be effective in the treatment of third degree malnutrition and has resulted in a sharp reduction in second degree malnutrition; targeting is among the important variables which explain its success. Its design is based on a careful monitoring of growth of children as a basis for selective feeding for 90-120 days and for education. The programme offers a selective package of primary health care, food supplements, and nutrition education.

Another experience of the 1980s, fairly well known for its effective targeting and strong grass roots orientation, is the Iringa Project in Tanzania.¹⁴ In this project, growth monitoring is applied in identification of growth faltering, and is followed by education combined with primary care, promotion of household food security and strong reliance on community participation.

Obviously, targeting strategies are not completely effective for the ideal or optimum programme objectives. However, it is worth remembering that supplementary feeding is by definition a short term approach to the problems of food and nutrition among women and children. By that definition, targeting is then suitable and relevant because it allows for relatively cost-effective control of malnutrition in the short term. Such an approach, combined with a programme focused on poverty reduction, will help to eliminate some of the inefficiencies of targeting.

Integration of feeding and health care

During the 1980s, integration of feeding programmes and health services received considerable attention. This was to a large extent the result of an increasing consensus that simultaneous improvements in dietary intake and the control of infection would have a synergistic effect on child health. A rather well known study in the villages of Punjab, India has made important contributions in this area. This study commonly known as the Narangwal Study, was designed to test a programme’s synergy in impact when the control of infection and improved diet for young children were provided simultaneously. It is the only field trial of its kind in the world thus far.⁸ This study showed a programme synergism for cost-effectiveness when feeding and health care were integrated. Another striking result in this study was the fact that feeding and education by themselves produced significant improvements in height and weight gain as well as reductions in mortality and morbidity.

Table 9. Some aspects of programme management in the 1980s.

Programme (ref)

Date/ duration

Review evaluation

Scale

Services

Community participation

Focus on women

Passive/ active

Staff/ beneficiary

Skills

Method

Community energy

Cost- effectiveness

Narangwal (India) (8)

1969-73

79-82

Pilot

1,2,4,5,10

NA

NA

A

H

H

H

NA

H

ICDS/lndia (26,42)

1975-

87,88,89

National

1,2,4,5,7,9, 11,14,17

A

L

P

M/L

H/L

L

O

L

PHC/Thailand (39)

1980-

-

National

2,3,4,5,6, 7,8,14

A,B,C,D

L

A

M/L

M/L

M

H

M

TNINP/India

1980-87

87,89

State

1,2,3,4,5,6,7, 8,11,12,14

A,D

M

A

H

H

H

M

H

Iringa/Tan-

1982-87

88

District

2,3,4,5,6,7,8, 11,13,15,16

A,B,C,D

M

A

H

H

H

H

H

Community participation: A = awareness; B = role in selection of programme priority; C = financial or time input; D = role in management.
Services: 1 = supplementary foods; 2 = nutrition/health education; 3 = preparation of weaning foods; 4 = distribution of vitamins and minerals; 5 = growth monitoring; 6 = oral rehydration therapy; 7= immunization; 8 = deworming; 9 = rehabilitation of severely malnourished; 10 = control of childhood diseases; 11 = health checkups; 12 = perinatal care; 13 = pregnancy monitoring; 14 = referrals; 15 = environmental sanitation; 16 = household food security; 17 = non-forrnal/preschool education.
NA = not applicable; H = high; M = medium; L = low.

In the meantime, several studies had suggested, that the effects of poverty on child growth operate primarily through poor diet and illness, and it was further suggested that, among common diseases of childhood, diarrhoeal diseases have particularly strong effects.¹⁵ During the 1970s, it had already been shown that the diarrhoeal diseases were a serious child health problem in Latin America, and that the disease occurred more often during the age range 6-18 months when growth faltering was at its peak. For this reason it was termed "weaning diarrhoea".³⁰ The term reflects the fact that diarrhoeal episodes are closely linked to transition of child feeding from breast to family diet. It is well known that contaminants of food and water, as well as unclean utensils and poor personal hygiene, are among the major causal factors.

In the developed world, illness and physical growth are not significantly related. In contrast, in the developing countries, poor physical growth is strongly affected by common diseases of childhood.³¹ There are two reasons for such a difference. First, children in the developed world often experience a much lower disease load; and second, because of their superior nutritional status, They cope much better with the disease stress. The cumulative effects of diarrhoea on young children in poor areas are quite strong. Martorell estimated the total negative effects of diarrhoeal disease among the poor Guatemalan children to be as high as 1.5 kg weight and 3.5 cm height by the age of seven years.³² This is primarily due to dietary restrictions and accelerated catabolic processes which divert nutrients from growth, as well as to rapid intestinal transit and malabsorption.

It is on this basis that the essential components of health care service, in an integrated programme during the 1980s, have consisted of growth monitoring, immunization, oral rehydration, deworming and education on appropriate child feeding in combination with supplementary feeding.

Another frequently cited benefit from integrated programmes is better attendance and utilization of services where "food" serves as a strong incentive. Several reports in the 1980s have further confirmed such effects. For instance, the national primary health care programme in Chile²³ and reports from Food-for-Peace programmes in Haiti, the Philippines and Sri Lanka have further confirmed the effects of food distribution on programme attendance. In Cameroon, a clinic which stopped food distribution, experienced a drastic reduction in the number of attendants.⁹

However, satisfactory programme outcome is not dependent only on the design but also on effective implementation. In this context, there is a major difference between "research studies" and "on-going service programmes", just as there could be between sound design and poor implementation. During the 1970s food distribution was often administered through maternal and child health centres which were more likely to have an urban bias. However, primary health care as the dominant approach of the 1980s has had a major effect in extending the services to the rural areas. As yet, there is really very little data to show the benefits derived from integration. So far, the Narangwal study⁸ the programme in Chile³⁴ and the Iringa Project in Tanzania¹⁴ provide considerable evidence on the benefits of integration. Also, in Sri Lanka²⁵ immunization and deworming were judged to contribute to child growth, while in Senegal³⁵ the infant mortality rates were lower among the recipients as compared to the controls.

Some lessons in implementation

This report does not intend to review problems of implementation, but recognizes some important lessons of the 1980s which merit attention. During the 1980s the international scientific community, in its struggles with nutritional issues, paid more attention to the issues in "how it was to be done" as compared to the previous decade where the almost exclusive concern was on "what was to be done." Six cases are chosen for discussion here. They represent the chronology of advances made in programme development and implementation in these two decades (Table 9).

First, the Narangwal Study in India (1969-73), referred to earlier,⁸ represents one of the well-known experimental models that provided substantial scientific input into programme and policy decisions in the 1980S. The Narangwal Study came at a time when there was substantial debate around the world about whether growth failure among children was really a problem of poor diet or of infection. As indicated earlier, this study had considerable influence of integration of health and nutrition in a programmatic sense.

Second, the Integrated Child Development Services (ICDS) was established by the Government of India in 1975. In a period of 14 years, this programme has expanded to one third of its 5000 development blocks in India and already covers 23% of the target and reaches one half of the poor in the country. The significance of this experience is its tremendous scale. It is probably one of the largest of its kind in the world and works in a country where malnutrition is a major national issue and 98% of the programme’s costs are financed through national resources. ICDS, in its design, clearly reflects the state of scientific thinking and approach of its own time. It is a creation of technocrats supported by academia. It aims for growth and development of children and calls for a typical package of services including child feeding, education, immunization, health check-ups and referrals. Conceptually, it considers collaboration and coordination of three government departments, i.e., Social Welfare, Health, and Education, as the key in its success in implementation, and one or two women with limited training and experience in a village are the primary agents to carry it through.

After 14 years of experience, Gopalan and several other prominent Indian scientists have voiced their assessments of the experience in several reports²⁶ The programme has difficulties in reaching very young children, which it considers a high priority. Also, it has the image of a top down handout and lacks community involvement. In the 1970s, community involvement was not a common factor in programme design, and was not addressed in this programme. In addition, the programme needs to strengthen its focus on women, also not included in its original design. There is a convincing argument that a focus on women is the only assured channel for reaching infants and young children. (This is in addition to the contribution that it will make to the health of women.) Finally, the programme management is passive and needs strengthening and reorganizing. This is to be done particularly through training, improvement in administration and service procedures, supervision, better staff/beneficiary ratio, community support, and incentive for village workers and participants. Gopalan has also emphasized that detailed, careful preparation of the community at the beginning is an important factor which would have made a major difference to its outcome. In essence, what Gopalan has asked for are among the fashionable issues of the 1980s which were rarely given much attention in the 70s.

Table 10. Design and implementation issues and response in the Tamil Nadu Integrated Nutrition Project (TNINP) in 1980.

Design issues	Program Response
1. Missing children in the most vulnerable age	Focus on 6-30 months old children
2. High costs of food/long periods of feeding	90 days feeding to maintain weight gain (90% participated, 65% responded, 15% needed food for 120 days and the rest referred to health services)
3. Risk of dependency	Focus on and incentive for keeping children out of feeding
4. Sharing and substitution	On site feeding/use of special snack with no attraction as adult food.
5. Poor effects on proper child feeding/poor use of local foods	Use of locally produced weaning foods and focus on very young children
6. Poor targeting	Area selection and growth monitoring
7. Poor and irregular attendance	Home visits/use of women’s clubs/encouraging community participation/better staff training /supervision and follow-up.
8. Poor linkage to health care and control of illness	A selective primary health care package: growth monitoring, feeding/education plus diarrhoeal disease control, immunization, deworming and supplements/check-ups and referrals
9. Poor management and supervision	Higher staff/beneficiary ratio; intensive and continuous training/training manuals/monitoring and supervision home visits and community support

*Designed by the World Bank as a model to improve programme efficiency and cost effectiveness (for the State Government of Tamil Nadu in India, 1980-1987).

The third important experience of the 70s developed in Indonesia.¹² It officially began in 1979 and was called the "Family Nutrition Programme". However, the origin of its development in pilot research form goes back to the early 1970s. It emphasized growth monitoring as an educational, integrating and targeting tool for early detection and control of growth faltering. It focused on interaction between mothers and health workers and on the promotion of appropriate feeding of young children. It provided supplementary feeding selectively. The village health worker is a female volunteer who is given very short training. In a so called "rapid expansion" phase the programme expanded rather fast with relatively low input, a narrow spectrum of services and very limited management capacity. In this phase the programme was designed to facilitate expansion and strengthening of the family planning programme in Indonesia. In the consolidation phase, which began in 1985, the programme aims to provide a wider spectrum of integrated health and nutrition services, while improving focus on women and strengthening in management.

Fourth, in 1980 the World Bank assisted the state Government of Tamil Nadu in India in designing and implementing a nutrition project.¹¹ This project, which is often referred to as Tamil Nadu Integrated Nutrition Project (TNINP), was actually designed as a model to improve cost-effectiveness in child nutrition programmes. It represents an excellent example of how programme design and management tools can bring a bearing on a problem under difficult conditions. Table 10 shows the project’s answers to various design and implementation issues which were previously diagnosed in Tamil Nadu’s programme operations. The project targets young children through growth monitoring and provides selective feeding for 90-120 days. It offers a highly selective range of primary health care services in order to balance it against implementation capacity and to avoid a heavy workload. It also encourages community participation and pays strong attention to management issues.

In Table 9, management capacity is characterized by five indicators. These are:

1. the extent to which a programme is active or passive;
2. the staff/beneficiary ratio;
3. the quality and breadth of staff skills;
4. the technical and administrative methods and procedures in a programme
5. the level of community support, called "community energy", which is an indication of the community’s input and sense of ownership.

The most pressing problems in supplementary feeding programmes in the seventies were in targeting, leakage, integration and management. TNINP was actually designed to overcome these problems. Clearly TNlNP’s rating in management is high, but participation in this project only involves awareness and lime input.

In an interesting study on management of supplementary feeding in rural health centres in Egypt, a joint team from Cairo University and Massachusetts Institute of Technology, examined several variables in a detailed survey. In their final judgement how well the programme is implemented at the centre/unit level is a function of two critical factors. First, the extent to which a health centre displays an active approach to outreach and follow-up, and second, the orientation of the physician in charge, most notably in terms of their own job satisfaction and the way they perceive food distribution and judge it for its merit and in the context of the practice of medicine and public health.

The fifth and sixth projects are good examples of primary health care in action. They are in Thailand and the Iringa district in Tanzania.^13,14 The Thailand programme is one of the national programme models which is recognized for its strong focus on participation and sharing with the community the decisions on its content and approach, its focus on reaching the mother and the young child, its reorientation of child feeding practices, its aim of improving women’s functional literacy and income, and its provision of health care for women during pregnancy and lactation. This is a programme administered by the Ministry of Health, supported by the local government and village committees and women’s groups. The significance of this programme is its scale, continuity and internal support. However, like all Government-sponsored programmes it has a relatively lower input per beneficiary on implementation, and is less cost-effective.

The Iringa project, supported by UNICEF and the World Health Organization, has several unique characteristics of its own. It is primarily focused on education, self-reliance and community support as the basis for action. It is a project developed and implemented at this district level under the authority of the district community development office with strong patronage and political support from the Prime Minister’s Office. The primary purpose of the project is to provide substantial training and educational input into the process in addition to service. There is an element of synergy between the service and education. It focuses on control of infection, child care and household food security. Finally, it is a creation of international agencies in collaboration with the Government. Some have called it a nutrition planner’s dream because it has enjoyed the rare continued high level political support, as well as administrative and financial flexibility, and facilities of working at the local level.

In the final analysis, the experiences of the 1980s point to handsome rates of return on investments in support of programmes offering properly designed rations, a focus on narrow but essential services (to avoid loading the system), integration of nutrition and health care, focus on women and child linkages; and emphasis on targeting, community participation and essentials in management as the key ingredients in improving the impact which the programme have on maternal and child nutrition.

It must be added that several disciplines outside nutrition and public health have played a major role in such developments. Among them are micro-economics, public administration, anthropology, and political and behavioural sciences merit special mention.

Potential and measured benefits

The state of knowledge by the end of seventies

The discussion begins with a review of knowledge by the end of the seventies and the gist of it derives from the analysis presented in the UNICEF/SCN report in 1979.¹ This rather critical review of data showed great variation in measured benefits. Reports from field trials and research studies in Colombia, India, and Guatemala, showed that a mean difference in weight gain of 0.5-1.00 kg/year can be achieved among young children through supplementary feeding.^36,37 These studies showed significant changes in height as well. On the other hand several ongoing programmes reviewed in this study rarely showed comparable levels of achievement. Therefore, it was suggested that the outcome of research studies in a way demonstrates the relative magnitude of child growth that can be achieved through food supplementation. The anthropometric response reviewed in this report is based on the data from 43 projects. They varied considerably in their scope, design and evaluation methods. Table 11 shows the quality of the database characterized by the design. The best quality information is derived from controlled longitudinal design (type A) and the second in ranking for quality is controlled cross-sectional design (type B), followed by longitudinal without control (type C) and cross-sectional without control (type D). Clearly some of the reports did not have appropriate experimental design. As Table 11 shows, 23 of the reports presented better quality data, (Types A and B), 12 of which originated from Latin America, 10 from Asia and one from Africa. It is also interesting to note that almost half of the reports were from Latin America and 13 from India alone. Most of the projects (all except four) reported some degree of benefit (Table 12). Altogether 18 projects reported significant weight gain and four of them had also reported significant reduction in mortality and morbidity. Non-significant weight gain occurred in 18 projects. Only one report indicated increased physical activity. It is also important to recognize that the data in half of those projects which reported statistically significant weight gain came from a controlled longitudinal design. However, it is true that the magnitude of weight gain in the ongoing projects - even if significant - was below that achieved in the pilot studies. A few possible explanations were offered for such variations in response. First, in some projects the extent of the energy gap was quite high and the net levels of supplementation were too small. In another scenario, children had a very small weight deficit. These eases may well represent a poor selection of programme target, leading to a disappointing response.

Table 11. Quality of database on measures benefits.

Location/ type of design	A	B	C	D	Unknown	Total
Latin America	7	5	3	1	5	21
Asia	7	3	2	1	3	16
Africa	1	-	5	-	-	6
Total	15	8	10	2	8	43

A = Longitudinal with control. B = Cross-sectional with control. C = Longitudinal without control. D = Cross-sectional without control. Source: Beaton & Gassemi (1982)¹.

Table 12. Measured benefits in various projects^*.

	Number of projects in the regions
Type of benefit	L. America	Asia	Africa	Total
Significant weight gain	6	6	2	14
Significant weight gain/ reduction in morbidity/ mortality	2	2	-	4
Reduced morbidity	1	-	-	1
Improved weight gain (not significant)	9	7	2	18
Improved physical activity	1	-	-	1
Improved cognitive development	1	-	-	1
No effects	1	1	2	4
Total	21	16	6	43

*Summarized from Table 4 in Beaton & Ghassemi (1982)¹.

Other factors having considerable influence on programme outcome were attendance and programme duration. Habicht and coworkers³⁷ have shown that participation of over 50% clearly results in better growth among children. Several analyses showed that participation is a function of:

1. the opportunity costs, such as work time lost for the mother, and the distance to be travelled to reach the health centres
2. programme incentives such as ration size and simultaneous benefits from other services
3. family education and attitude and the perceived programme benefits
4. skills and motivation of the programme staff
5. the extent of community participation and the sense of ownership that the families and communities have for the programme
6. the programme effectiveness in improving the health and wellbeing of the target population.

Nelson and Sahn³⁸ in their review of PL 480 Title II -supported feeding programmes in India outline the major factors leading to a relatively insignificant impact. The review clearly illustrates a typical scenario for poor design and management, leading to a poor outcome. They argue that low coverage (16% of the potential), low levels of supplementation (15%), and high levels of sharing and substitution leads to a small net supplementation increment. When this is placed against the background of frequent infections, fluctuating family food supplies and other variables, it would be hard to expect substantial results.

In the final analysis Beaton and Ghassemi (a) concluded that the available data show relatively small benefits derived from supplementary feeding programmes; (b) the authors reviewed the database against some basic theoretical and practical issues in evaluation; and (c) made a careful note of caution against drawing strong conclusions from what was learned - which happened to be much less than what was needed to be learned. Their report was perhaps much more of a challenge to some of the underlying principles of programme design and evaluation rather than a straight forward evaluative judgement on how these programmes were doing. It is this aspect of the report which has not always been given sufficient attention in its subsequent interpretation within the community of scientists and programme planners.

Figure 4. Some potential pathways of benefits in food distribution systems.

It would be quite useful, at this point, to recapitulate those issues and refresh memories on the nature of the cautionary remarks in this report.

First, it was emphasized that the world of science basically knew about the functional effects of severe malnutrition, and the state of knowledge on the functional significance of mild and moderate malnutrition was quite limited (some progress has been made in the 80s and will be discussed later). As a result there was relatively little knowledge as to which indicators were likely to be sensitive and specific enough to evaluate changes of those functions in relation to food distribution at the population level.

Second, there are several pathways through which the body can benefit from improved food intake. Improved physical growth is only one of those potential benefits. Figure 4 shows some potential benefits to be derived from food distribution programmes. One route leads to improved physical growth and functional development. The second leads to improved functioning of the body system through metabolic processes, and the third is the psychosocial development of the child, which is suggested to be mediated through physical activity. With a relatively low level of net food supplementation among the chronically undernourished, the nature and range of response was not clear. All that was known was that acutely malnourished children during their recovery respond through better growth, increased physical acclivity and improved metabolic function. It was simply not clear which benefits should be expected in individuals or population. Therefore the complex question was what should be measured. Another important point is that physical growth is only one, and not necessarily The most important, of many potential benefits. The fact is that most evaluation studies had chosen physical growth, particularly weight, as an indicator of derived benefits. Only a few programs had been evaluated through changes in others indicators such as mortality or morbidity and only one in relation to increased physical activity.

Third, if weight gain was the only source of benefit in feeding programmes, then the weight increment should, theoretically, account for the net increase in energy intake. Simple calculations show that energy cost of "gain" and maintenance of weight at the end of a year would be about 70 kcal/day or 45-50 kcal at midpoint. This was only 1/2 to 1/3 of net energy intake in most of these programmes. What then happens to the rest of the energy and what were the explanations for the disparities in energy balance? Looking at Fig. 4 a possible explanation is that a pan of the additional energy intake is apportioned towards physical activity which in turn helps the child to play and improve interaction with his surroundings, respond to environmental stimulation, and improve cognition and learning ability. The long term significance of such benefits hardly needs to be emphasized.

Fourth, although the data available did not permit a quantitative estimate of energy costs associated with these processes, They clearly strengthened the argument that not all derived benefits from the programmes were being adequately measured and the unmeasured effects may be more significant in the overall development of the individual and the community than the relatively small improvements in weight gain.

Finally, the discussion certainly implies that the current database on evaluation may well underestimate the real impact and in essence the reader was cautioned on the conclusions to be drawn from the body of evidence on impact. Ambiguities in the definition of objectives, narrow examination of a wide spectrum of expected benefits, wide variations in quality of our database, and, finally, extremely limited knowledge on the functional significance of mild and moderate malnutrition and proper choice of indicators were among the important reasons for caution.

Figure 5. Effects of input services on weight at ages 0-36 months. Experimental groups compared with controls, adjusted for sex, birth order, mother’s age, caste, year, and season of observation._______ Nutrition care (NUT). -- -- -- -- Nutrition and health care (NUTHC). - - - - - - Health care (HC). From Keilman et al (1983)⁸.

Figure 6. Effects of input services on height at ages 0-36 months Experimental groups compared with controls, adjusted for sex, birth order, mother’s age, caste, year, and season of observation From Keilman et al (1983)^8.

Programme experiences during the 1980s

The final report of the Narangwal study was published in 1983.⁸ The study was based on the common knowledge that the leading cause of death, disease and retarded growth and development among children has been the synergies between nutritional deficiencies and common childhood infections Therefore, the question being asked was could there be a similar synergy in programme impact, if the two major causes were addressed together and at the same time? The Narangwal field trial was designed to test the programme synergy It had three groups of study villages Nutrition Care Villages (NUT) received food supplements and nutrition education Nutritional risk and eligibility was determined through growth monitoring. The Health Care Villages (HC) received for common diseases of childhood The third group of villages received a combination of services of the previous groups (NUT HC) and finally there were the control villages (CONT) In the NUTHC group the range of services was equivalent to the sum of the two, ie, NUT and HC, but the level of inputs was intentionally kept comparable to the other two groups. This is one of the few studies in the 1980s which clearly documents the range of impact that can be achieved through a well integrated health/nutrition care programme in poor rural areas. The results showed that children by the age of 36 months in the NUT villages were 600g heavier and 1.3 cm taller (Fig 5 and 6). It also shows that treatment groups were worse off than the control in the first year and managed to catch up during the second and third years. Another important point is the nutrition care (NUT) villages were even a little ahead of the NUTHC villages in growth. This is fairly convincing evidence that child feeding, if properly targeted on the undernourished, will actually result in significant benefits in reduction of mortality and morbidity (Table 13). Another very important finding in this study is that health care by itself is effective in reduction of mortality and control of morbidity but relatively ineffective in promotion and preservation of child growth. It clearly shows the relative disadvantage in many primary health care programmes which focus on control of infection without giving equal attention to the needs of improvements in child feeding and dietary intake In an interesting analysis, the programme effectiveness has been shown for various components in the Narangwal study (Table 14) For instance provision of health care (HC) is almost twice as effective in reducing infant mortality as compared to feeding and nutrition care (NUT), while HC effectiveness in increasing the height of children is only one sixth of that of nutrition care (NUT) It was also shown earlier that NUT villages were about twice as effective in producing weight gain compared to HC villages Finally, NUTHC villages were much more effective in psychomotor development ,among 0-3 year old children Also, illness decreased by 22.2 days among the children in NUTHC group and by 20 days among the HC villages.

Table 13 Effect of nutrition and infectious disease control on child mortality rates in the Narangwal study* (standardized for caste and sex distribution differences).

	Study group (1)
Mortality rate	NUT+IDC	NUT	ICD	Control
Perinatal (2)	80	60	65	105
Neonatal (3)	47	49	48	80
Postneonatal (3)	24	50	35	52
Infant mortality (3)	65	94	78	128
1-3 months (4)	11	11	13	18
Total number of live births and stillbirths	548	300	299	724

1 Group codes - NUT nutrition care IDC infectious disease control
2 Per 1000 live and stillbirths
3 Per 1000 live births
4 Per 1000 population of given age
*Keilman, et al (1983)⁸.

One may ask what does the additional weight gain of 500–600g mean and what is its nutritional significance? Table 15 explains the change of malnutrition pattern in relation to such levels of weight gain. This table summarizes the weight gain and change in malnutrition in two of the well known studies of the 1970s, in Bogola and India, which were actually reported in the 1980s. The Bogota study by Mora and co-workers³⁶ shows weight increment by age among children supplemented in a longitudinal study from age 3 to 36 months. This study actually prevented precipitation of third degree malnutrition and managed to reduce the second degree malnutrition (Gomez classification) by one third to one half by the age of 36 months. The Narangwal study⁸ reduced malnutrition (defined arbitrarily as below 70% Harvard standard) by one third Clearly programme impact expressed in terms of change in weight distribution is much more meaningful than simply stating the increment in weight gain resulting from the intervention In this perspective, research studies which reflect better potentials for achievement have actually been effective in elimination of severe malnutrition and in cutting moderate forms by one third to one half and this is consistent across programmes.

Table 14. Relative effectiveness of various service components in Narangwal study.

	Perinatal mortality	Mortality in children 1 year old	Mortality in children 1-3 years old	Morbidity in children 0-3 years old	Growth in children 0-3 years old	Psychomotor scores in children 0-3 years old
Maximum effect observed (a)	(Decrease of 43.3 deaths/ 1000 live & stillbirths)	(Decrease of 59.2 deaths/ 1000 live births)	(Decrease of 7.6 deaths/ 1000 children 1-3 years old)	(Decrease of 22.2 days of illness per child 0-3)	(Increase of 1.3cm in height at 36 months old)	(Increase of 5.2% points by 36 months age)
Index (b)	100	100	100	100	100	100
NUTHC	94	81	70	100	92	100
NUT	100	55	100	-	100	56
HC	54	100	100	94	15	-

- Zero or neglible.
(a) Observed difference from control levels in rates or values in the most effective service package.
(b) For example, if the maximum effect on perinatal mortality is a decrease of 43.3 per 1000, then an index of 94 denotes a decrease of 43.3 per 1000 x 0.94 = 40.7 per 1000.
*From Keilman et al. (]983)⁸.

Table 15. Programme benefits in two field trials in the 1970s.

Age (mo)	Weight gain (g) (experimental minus control)					Change in malnutrition
	Bogotaa			Narangwalb		Bogota*						Narangwal**
						Second degree			Third degree
						Experimental	Control	Difference	Experimental	Control	Difference	Experimental	Control	Difference
6	+197	P<0.05			-800	14	9	-5	3	-	-3	12	10	-2
12	+372	P<0.05			-100	22	12	-10	3	-	-3	27	27	-
18	+423	P<0.05			+200	22	18	-4	-		-	31	20	-11
24	+519	P<0.01			+400	21	7	-14	-	-	-	31	23	-4
36	+476	P<0.05			+600	21	17	-4	-	-	-	-	-	-

*Gomez classification. **Below 70% Harvard standard. ·Mora J. et al. (1981)³². ^bKeilman A. et al (1983)⁸.

Table 16. Measured benefits of supplementary food during pregnancy in controlled studies.

Benefits	USAa	Gambiab			Bogotac		Thailand
	Women/infant and children (WIC)	Wet season	Dry season	All year	Supplemented	13+weeks
Increase in birth weight (g)	+107	+200± 53 p<0.012	13± 58 p<0.059	+124± 42 p<0.05	+63 p<0.05	+90 p<0.025	+263 p<0.025
% change in low birth weight	-4	-18	-8	-11	No LBW
Pregnancy weight gain (g) per month (last trimester)	2 200 p<0.005	1200-1700 p<0.005	300-400 p<0-005		740 ns	920 ns

^aKennedy et al. (1984)⁴⁰.
^bPrentice et al. (1987)⁷.
^cMora et al. (1983)¹⁸.
^dTonisirin et al. (1986) ⁹³.

Table 17. Measured benefits* in some large scale programmes in the 1980s.

Nutritional status	ICDS-Indiaa			Chileb			Iringa (Tanzania)c
(children 0-36 mo)	1976	1985	Difference	1975	1980	Difference	1984	1988	Difference**
Normal	69.5***	72.6	+3.1				46	61	+15
Malnourished
Mild	-	-	-	12	10	-2	44	32	-12
Moderate	19.7	19.8	+0.1	2.4	1.7	-0.7
Severe	6.2	4.3	-1.9	0.7	0.1	-0.6	5	1.6	-3.4

^aGopalan (1988)²⁶.^bGonzales et al (1983)³³.^cWHO/UNICEF (1988)¹⁴ .
*Changes in proportion of malnourished children
**% Children in programme group during the programme operation (growth monitoring)
***%Children in programme group over control or before and after intervention

The impact of supplementary feeding on pregnancy outcomes has been measured in several studies Results of four studies are shown in Table 16. They all show significant increase in birth weight in the range of 60 to 260 g An important observation in the 80s was made by Prentice and co-workers in Gambia .⁷ They have shown an average gain in birthweight of 200g in the wet season against an increase of only 13 g in the dry season The wet season is when the seasonal shortage of food supplies coincides with the peak of agricultural activities and higher demand on women for physical work. This is evidence showing that increased food intake makes a difference when applied to the right target at the right time The same effect is clearly shown with regard to the incidence of low birthweight and pregnancy weight gain Seasonal effects were also reported in the Narangwal study. In this study season had a distinct and statistically significant effect on mean weight of most ages (P<0.001) Weights recorded in the wet (July-September) and hot (April/June) seasons were consistently lower than those recorded in the rest of the year (mild and cold seasons)⁸.

The rate of increase in birth weight in Thailand³⁹ was actually the highest, at least partly because the target population was seriously deprived The data in the American WIC programme⁴⁰ come from a controlled study in the state of Massachusetts. Although a broader evaluation study, it did not show any significant increase in birth weight among the WIC recipients as compared to non-WIC groups.⁴¹

During the seventies data on large scale ongoing programmes were almost non-existent. As indicated earlier, evaluation reports on several large programmes became available in the 1980s. The data from these reports are much more meaningful in terms of achievements under the realities of administrative, technical and financial constraints in developing countries.

Table 17 shows results for Chile, India, and Iringa in Tanzania. The Integrated Child Development Services (ICDS) in India was already described in a previous section³². Tandon⁴² has shown this programme to be effective primarily in the rehabilitation of severely malnourished children. Over a period of 10 years it has helped to reduce severe malnutrition by almost two-thirds, without much effect otherwise The food distribution programme in Chile³³ has a long history It has a national scale coverage and has been shown to be effective in reducing infant mortality drastically in this country, although it is not possible to separate the effect of food distribution from that of health services Furthermore, it has been effective in almost eradicating severe child nutrition and culling cutting malnutrition by one third. Due to economic difficulties in recent years, distribution of milk has been replaced by rice, which has reduced the level of protein intake among women and children considerably The latest data indicate that over the past five years the programme has had a maintenance role in keeping intact the status of health achieved among women and children by 1980 (R Mardones, personal communication).

Finally, the Iringa project in Tanzania¹⁴ which covered a population of 300 000 in Iringa district, showed a significant impact on child malnutrition. Some aspects of this project were discussed earlier. In a period of four years, severe malnutrition has been reduced by two thirds and moderate levels have been reduced by 25%. This is one of the very few experiences of its kind where a change in weight distribution among children has occurred at all levels. Other programmes basically have pushed the lower end of the weight distribution toward the right, with a relatively small shift in the entire weight curve. In Iringa the portion of normal weight gain increased by one third.

As reported by Berg,¹¹ the Tamil Nadu Integrated Nutrition Project was effective in reducing severe malnutrition by almost two thirds and in cutting its moderate forms by one half. Almost similar results were demonstrated by Blumenfeld in evaluation of food distribution/primary health care programmes in the Philippines supported by CARE and CRS.⁴³ The strength of the programme is its high coverage of very young children at the cost of high errors of inclusion. The programme was shown to be least effective for 6-11 month-old children and was most effective for the children with highest levels of malnutrition at the entry. The average programme duration was 18 months.

In Sri Lanka³⁵ a programme of child health including distribution of triposha (a special weaning food for young children made from local ingredients) has shown impressive results in the control of malnutrition. However, it is not possible to separate the triposha effect from the rest of the programme services Given a high rate of sharing and relatively small size of ration, its effects in all probability would be small. Although triposha is a child’s food, surprisingly other family members have shared it. In this programme the rate of malnutrition (below 70% of the standard) decreased by 7.4% among children who stayed in the programme for 13-29 months; 33% of the participating children at the age of three years were below the 70% standard, while the rate among the new entry group was 45%.

A similar programme supported by Catholic Relief Services in Upper Volta⁴⁴ was evaluated in 1981. The results showed improvements among severely malnourished children in the older age. The percentage of participating children below 60% standard at the age of 7-12 months was actually higher than the new entry group By the age of 2-3 years, the rate of severe malnutrition (below 60% standard) was 3% as compared to 10% among the new entry.

A supplementary feeding programme in Senegal³⁵ supported by Catholic Relief Services showed almost no effect when children leaving the programme in 1981-82 were compared to those entering the programme in he same period Botswana¹⁰ has an effective and elaborate nutrition surveillance programme. The surveillance data show that rate of malnutrition varies between 24 and 29% During a severe drought in 1983-84, malnutrition increased to 30% and seasonal variations almost disappeared. A drought relief programme including food distribution for women and children managed to control deterioration of the situation and resulting in a rapid return to the previous levels in a short time. In this operation some districts did much better than others.

In Morocco, food distribution was combined with a fairly well planned and supervised nutrition education.⁴⁵ This initiative was supported by the Catholic Relief Services (CRS) and evaluated in 1980 after being in operation for four years. The programme provided a take-home ration equivalent to 525 kcal and 16 g of protein per day for three members of a family Nutrition education was offered in monthly classes on nutrition, health, sanitation and food demonstrations. The results showed that malnutrition among participating children in food distribution had been reduced by two thirds from 32% under 80% weight-for-age to 10% in four years. Effects of nutrition education were measured through comparing a group of children in "Food and Education" families with their brothers and sisters before the education components were added. The samples controlled for the feeding duration results showed malnutrition being 33% among "fed only" as compared to 11% among "fed and educated" One striking aspect of this analysis is that the local retail value of food distributed was $US 73 per family per year or somewhere between 4-24% of the per capita income. In this programme it is difficult to know the extent to which anthropometric response is mediated through income, food or education, or some combination of the three.

CARITAS and CRS supported a programme of Applied Nutrition Education (ANEP) in the Dominican Republic⁴⁶. It is a community based, grass roots strategy to improve the nutritional status of children in poor rural communities. It involves raising awareness, motivation, and self-reliance through growth monitoring, promotion and education. An evaluation, after three years of operation, showed significant behavioural changes and differences in key health and nutritional practices, such as proper child feeding, management of diarrhoeal diseases, lactation and personal hygiene The prevalence of moderate-to-severe malnutrition was reduced by more than one half in 2-3 years’ participation. It would be interesting to know if education could really be effective in modifying intra-household food distribution patterns in poor households The education effects in the Morocco programme could probably be explained partly by a reduced sharing due to education campaigns However, this issue was not addressed during the programme evaluation.

Programme costs

During the 1970s relatively little cost analysis was performed on supplementary feeding programmes. For instance, out of 43 projects reviewed by Beaton and Ghassemi, only 11 provided some form of cost analysis.¹ The information on cost primarily indicated that most programmes provided a ration of 300-400 kcal at the cost $US of 15-25 (equivalent in 1976). It was also shown that the cost for effective increase of dietary intake by 300 kcal per day would have been closer to $75. The most often cited cost benefit analysis in the 1970s was that of Anderson on five CARE programmes⁴⁷. An interesting piece of information in this analysis was the estimate of costs of closing the energy gap among children in these programmes. That cost was in the range of 47 US cents in Colombia and 12.6 cents in Pakistan per day while the level of the energy gap in these two countries was comparable (Colombia 382, Pakistan 356). This is probably the closest the analysis came to demonstrating the costs of effective improvement in dietary intake of children. Furthermore, it was shown that the costs in research projects were relatively higher than those of ongoing programmes. Also, based on rather limited information, it was shown that management costs varied between 25% to 50% of the total. Such figures are not really comparable because the wage rates and prices vary greatly between countries. However, they help in providing some kind of a norm on costs for management.

During the 1980s cost analysis was given much more attention and is now considered to be an important part of the overall programme analysis. Table 18 shows a summary of cost analysis of several programmes reported during the 1980s The analysis provides much more detail as compared to the previous decade The overall range of cost figures does not show major changes as compared to the 1970s but programme effects are relatively better. For instance, the Tamil Nadu Integrated Nutrition Programme (TNINP) is to a large extent the extension of Narangwal study, which provided the pilot experience The total costs of TNINP are almost 75% of these of Narangwal. This is a case where the intensive costs of research are being reduced in a large scale application. Also, the costs of TNINP and ICDS appear to be similar, but the TNINP is almost three times as effective as ICDS. A more realistic comparison is to be made between TNINP and ICDS in the State of Tamil Nadu. According to Berg,¹¹ annual direct costs of ICDS nutrition centre in the State are around $US 1129 as compared to TNINP’s cost of $US 579. However, this figure does not include costs of health services, training and supervision. ICDS weighs fewer children (43% compared with 60%), but feeds relatively more of them at any given time (100% as against 27% for TNINP). It is also interesting to note that TNINP has reduced the cost of feeding to $4.59 per child, while spending $7.02 for weighing.

The cost analysis on Indonesia by Corina⁴⁸ has some interesting aspects. It shows that the total cost of effective delivery of comprehensive primary health and nutrition care in Indonesia is 37.80 $US per beneficiary per year. This analysis was carried out for UNICEF and the general informed opinion in the Organization id that the figures are too high. The author has also pointed out the limitations of the database for a conclusive costing. However, the method analysis is quite interesting. In Indonesia there are two programmme packages called basic and complete. The basic package provides health care and family planning, and the complete package includes the basic plus selective food supplementation, home gardens and nutrition education (Table 19).

The basic package, which may be considered as child survival piggy backed on family planning, costs about $18 per beneficiary per year. The difference between the basic and complete package is the additional nutrition component and involvement of other implementing agencies. The cost of the nutrition component is $19.65 which is somewhat higher than the basic. The interesting point is that the annual cost of the nutrition project component (selective supplementary feeding/education/ growth monitoring/vitamin supplements and home gardens) comes very close to $25 which is comparable to the cost figures reported for most of the supplementary feeding programmes. However, this approach is much more comprehensive and should have long term health effects on children.

Table 18. Costs of some programmes reported during the eighties.

Country/programme	Target age (years)	Type of food and/or service*	Target ration (kcal/protein)	Days of supplemen tation	Costs/beneficiary/year**
Narangwal/lndia (1982) (Ref 8)	6-60	DSM,O,S,W,/Health Care	400/11	120	Total Nutrition care Health care	23-25 10-14 11-13
The Philippines (1982) (Ref 43)	6-48	CSM/NFDM	400/10	90	Total	22.90-36.40
Morocco (1980) (Ref 45)	0-60	SFF/WSB/S.B O	526/16	365	Total	34,47
Sri Lanka (1982) (Ref 46)	0-60	Triposha (CSM/CSB/Vit)	190/10	365	Total	6,12
ININP/ India (1987) (Ref 11)	6-36	Laddu (Wheat-based snack)		90-120	Total Nutrition and health care Weighing Feeding	21.00 9.41 7.02 4.59
ICDS/INDIA (1987) (Refs 32 and 68)	6-72	SFBW/S,O	300/11 600/22	365	Total	20.00
Indonesia (1983) (Ref 48)	0-36	Selective suppl./ vitamin and mineral suppl./ home gardens/ health care	N/A	N/A	Total Health care/ family planning Nutrition suppl./ home gardens	37.80 18.15 19.65
Iringa, Tanzania (1988) (Ref 14)	0-36	Growth monitoring/ health care/child care household food security	N/A	N/A	Total Start up Expansion On-going	17.00 3.60 5.30 8.05
Domenican Republic (Ref 14) (1988)	0-60	Growth monitoring/ individual and group education	N/A	N/A	Total Local Technical	25.91 23.17

*Code to foods used: DSM = Dried skim milk; WSB = wheat soya blend; NFDM = non-fat dried milk; W = wheat; CSM = corn soy milk; SFF = soya fortified flour; S = sugu; O = oil. **Dollar exchange rate of 1989.

Table 19. Cost analysis of Indonesia family nutrition programme*.

Programme component	Unit cost/year/actual coverage ($US 1988)
	Basic package	Complete package
Growth monitoring	0.08	2.47
Oral rehydration therapy	0.23	0.79
Breast feeding	0.05	0.60
Immunization	1.30	1.74
Family planning	10.07	10.07
Vit. A distribution	0.15
Iron/folate	0.15
Nutrition education	0.30	0.15
Supplementary feeding
(Selective)	-	1.10
Home gardens	-	1.86
Delivery cost	-	6.11
Total	12.60	25.76

*From: Cornia (1984)⁴⁸.

Another important aspect of cost analysis in Indonesia is its estimates of cost sharing. Theoretically, the total costs of these programmes need to be broken down into (a) cost to the government; (b) external assistance; (c) contributions by the community; and (d) the costs to the beneficiaries such opportunity costs (work time lost and transport etc.) In Indonesia, the village volunteers contribute somewhere between 10%-20% of the total programme costs through their time (Table 20). In most of the reports, the costs to the community and beneficiaries are not considered.

The Iringa project in Tanzania (Table 18) is probably among the more cost-effective examples of nutrition programmes of the 1980s. Unfortunately, the contributions of the community have not been reported in this project. The cost figures are derived from a set of data which reflects more accurate accounting practices and is closer to the norms of programme cost estimates in UNICEF and WHO. An interesting aspect of the Iringa report is its breakdown of costs between various phases, i.e. start-up, $3.60, expansion $5.30, and ongoing $8.05. One very important lesson learned from the Iringa project is the critical effect of start-up, the preparatory phase of the project at the cost of $3.60 per beneficiary. In retrospect, the experience has shown that it is a worth while expenditure because a careful and perhaps somewhat costly preparation has proved to be a key factor in the project’s effective outcome. The fact that the project has spent $3.60 per beneficiary during the preparatory phase needs to be noted. Looking at the experience of ICDS, Gopalan²⁶ has pointed out that a careful preparation at the beginning would have made a major difference in its outcome today.

Table 20. Cost sharing in Indonesia family nutrition programme* (US Dollars 1984).

Programme Component	Intended Coverage		Actual Coverage
	Government/ international assistance	Village volunteers	Government/ international assistance	Village volunteers
Basic package	6.10	1.06	15.5	2.65
Complete package	11.74	2.40	31.76	6.00

*From Cornia (1984)⁴⁸.

This study was undertaken at the invitation of the Director of Food Policy and Nutrition Division of the Food and Agriculture Organization of the United Nations (FAO). The views expressed are those of the author and do not reflect the position of FAO or any other UN agency.

PART 2: DISCUSSION AND REFERENCES will appear in the next issue together with Chinese abstract of the report.

Copyright © 1992 [Asia Pacific Journal of Clinical Nutrition]. All rights reserved.
Revised: January 19, 1999 .

to the top