Previous Page Table of Contents Next Page


SECTION III. Supplementary Feeding


ACC/SCN Statement on the Benefits of Preventing Growth Failure in Early Childhood
Which Age Groups Should be Targeted for Supplementary Feeding?
Effects of Supplementary Feeding on the Growth of Children with Infection
Feeding Latin America's Children*
Comments on the Three Previous Papers by George Beaton, Juan Rivera, and Philip Musgrove
Response
Summing-Up
Educational Aspects, Weaning Food Supplements and Targeting in Supplementary Feeding Programmes

ACC/SCN Statement on the Benefits of Preventing Growth Failure in Early Childhood

Growth in young children tends to falter very early in many developing country populations, usually beginning by four to six months and ending by two to three years of age. This growth failure is often pronounced, so that by three years of age the size of the majority of children is outside the normal range expected in a well-nourished, healthy population. Once this faltering has occurred, linear growth tends to resume at a rate similar to that observed for children of that age in healthy populations.

The growth failure which occurs in young children is only one result of the common combination of inadequate child feeding, high rates of infection; and poor child care. It is now realized that there are other very important effects that need to be prevented. Small size indicates other risks, such as reduced nutrient stores, depressed immunocompetence, increased severity of infections, and poor motor and mental development. It is also a marker of risk of long-term functional impairment, including poor educational and intellectual performance.

Among the direct consequences of early growth failure are very short stature and reduced lean body mass in the adult, characteristics which constrain reproductive performance in women and work capacity and productivity in adults engaged in hard physical labour.

Attention to child feeding, the control of infections, and good care results in improvements in child growth and other crucial functions. Specifically, some significant recent evaluations have shown that supplementary feeding programmes, where enough food is delivered to and consumed by young children in need, are effective in:

- preventing growth failure;
- protecting against the negative effects of diarrhoeal diseases on child growth;
- improving educational performance, in later years.

Programmes that integrate interventions designed to attack the multiple causes of growth failure are most effective in improving child growth. These programmes can be viewed as investments in the future, for they lead to adults with a greater capacity for healthy, productive lives. Interventions that prevent growth failure in early childhood, it is now clear, can be expected to have a range of important short- and long-term benefits.


Which Age Groups Should be Targeted for Supplementary Feeding?

George Beaton

University of Toronto

Ten years ago, Beaton and Ghassemi (1982) completed a review of feeding programmes for young children in developing countries. That review was undertaken on behalf of the Subcommittee on Nutrition. In many quarters, that review was interpreted as concluding that the reviewed programmes had little or no effect. In fact it suggested that most research studies "had apparent effect but that few operational feeding programmes (except those that were specifically targeted to individuals with evidence of active very recent malnutrition) had demonstrable effect upon anthropometric outcomes (growth).

In the present paper, the objective is to discuss the desirable age range for targeting supplementary feeding programmes. It seems appropriate to quote one of the major conclusions from Beaton and Ghassemi (1982):

The general impression gained from this review is that food distribution programmes directed toward young children, as now being operated, are rather expensive for the measured benefit. However, the reviewers remain unconvinced that the benefit usually measured, physical growth and development, is either the total benefit to the family and community or the most important benefit. Therefore, it would be deemed unwise to withdraw such food distribution programmes until there has been opportunity to assess their true effects and benefits." [emphasis added]

The general situation does not seem to have changed very much. We continue to attempt to assess the impact of such feeding programmes in terms of anthropometric responses. Habicht and Butz (1985), in a paper addressing the health and nutrition effects of large scale nutrition intervention projects, concluded from sensitivity considerations, that the preferred measure of evaluation was change in length. They specifically concluded that morbidity and mortality were relatively insensitive indicators and that weight change was less sensitive than was length change. This was an excellent paper; but in keeping with the realities of existing measurement capabilities and existing perceptions of purpose, they did not address other possible outcomes.

Table 1 sets forth some of the effects of food distribution programmes that have been mentioned as potentially desirable outcome effects. Note that not all of these are commonly identified as "health" outcomes. These are not presented in any hierarchy of presumed desirability. The list is not even complete. The reasons for listing these now are: i) to provide a framework for addressing the assigned topic; ii) to again emphasize that there are many potential effects of such programmes (that would not be expected to necessarily move together); and (iii) to make the reader aware of the fact that "beauty lies in the eye of the beholder" (different planners may have quite different ideas of the real purpose of such programmes). A subsidiary theme is that which emerged in Beaton and Ghassemi: "if we base programme planning, or evaluation, on one category of effect, is it possible that we will miss the impact in another age group?" There is relevance, then, in asking "Are there differences in the preferred age targeting for these different effects?"

Linear Growth Protection and Promotion

Today, as in 1982, the most commonly perceived purpose of offering complementary feeding programmes in communities that are judged to be in need of such assistance is to promote physical growth. It is generally accepted that if physical growth is protected (maintained at close to genetic potential) through feeding or other interventions, there will be commensurate benefits in terms of other dimensions of human development. In this paper, this assumption is questioned but not directly challenged. As noted in an SCN report on Appropriate Uses of Anthropometric Measures in Children (Beaton et at, 1990), growth failure appears to serve as a valuable marker of situations in which a number of aspects of functional development have been impaired. However, while these different failures may stem from the same cluster of environmental deprivations, it does not follow that they are linked through the same physiological processes and hence that correction of growth failure will necessarily confer the same benefit as addressing the original causal situation - or that the other aspects of functional development might not be influenced without evidence of a growth response.

Table 1. Some Perceived Purposes of Supplementary Feeding Programmes

· Prevent growth failure o Prevent starvation o Treat current malnutrition
· Promote normal development including psychological development
· Control morbidity and mortality
· Provide vehicle for micronutrient supplementation
· Provide program participation incentive and/or income transfer effects and/or food for work


It is the judgement of this author that if improved linear growth is the goal of food distribution, there is an easily described age target of opportunity.

Three lines of argument will be presented to support the personal conclusion that such interventions are most likely to have programmatically significant effects on linear growth between the ages of about 6 to 18 or 24 months, although smaller effects can be seen at older ages, even to 5 years or more. The emphasis is on "programmatically significant" effects, not whether there is any effect. This, of course, goes back to the perception of the purpose of the programme.

The three lines of argument are based on:

i) epidemiologic observation of patterns of growth failure in developing countries.
ii) epidemiologic observation of "natural intervention" in the form of age specific secular trends.
iii) consideration of experience in carefully analyzed operational programmes.
Patterns of Growth Faltering

John Waterlow was probably among the first to give prominence to the fact that linear growth tended to falter very early in many developing country populations. He noted also that once this faltering had occurred, linear growth tended to proceed along expected channels for the now-smaller child. This was the basis of his classification of "stunting" vs "wasting" (e.g. Waterlow, 1976).

The phenomenon has now been extensively documented in many populations. Zerfas and Teller (1990) examined data from DHS/IRD survey databases. Figure 1 presents an example from their analyses. They presented their data in an unusual manner - as the percentage of children without stunting - to make the point that growth faltering/failure started very young. What the figure suggests is that "growth failure had occurred in the study population by one year of age and that thereafter the prevalence of stunting was relatively constant through 5 years of age. In the examination of data from other countries they showed relatively similar patterns although the age range during which the prevalence changed differed somewhat between study groups. It can be concluded from their intercountry comparison that early growth faltering is a frequent but not universal observation in developing country populations (poorer segments of the populations).

A more conventional picture of the same phenomenon is presented in Fig 2, based on data collected in Egypt, Kenya and Mexico as part of the Nutrition CRSP (Calloway et al, 1988; Beaton, 1992). Here results are presented as z-scores. The artefactual bump at 24-30 months is attributable to the known discontinuity in the reference values for length at this age. Nevertheless, one gains the distinct impression that in all three populations the departure from the reference median begins in the 4-6 month period and is most pronounced in the 8-12 month period. School age children (7-9 years) have z scores only slightly different from those at 2 years; the difference is almost non-existent if one takes into account the effect of the discontinuity in the reference data. The impression is that growth faltering, as a group characteristic, begins by 4-6 months and is complete by about 18 months of age in these three groups.

A similar picture is seen in data collected in the Guangzhou area of China in 1985 (unpublished data provided by Ho Ping), shown in Fig 3. Reported mean lengths, grouped by intervals of age, were converted to z scores for this illustration. (This procedure will give rise to errors in the computed z score, the magnitude of the error reflecting the rate of expected growth and the age span of the interval as well as the distribution of actual ages in the interval). Again one sees an early departure from the North American reference population - a departure that follows a similar age pattern. The figure also illustrates the sharp difference in growth between urban and rural population groups. From an unpublished longitudinal data set, the deviation in early growth is not seen in current Hong Kong children. Their lengths in the first year or so were remarkably close to the NCHS reference. This observation is added to suggest that the drop in z score in rural Guangzhou children is likely to have environmental rather than genetic origins. Better evidence for this is presented later.

Since it has been reported by several authors that, judged against existing NCHS standards, exclusively breastfed infants exhibit a departure from reference population norms, Fig 4 presents longitudinal data for breastfed infants followed in Davis, California by Dewey et al (1990) as part of the DARLING study. Since those infants had initial lengths about 0.5 SD above the reference, a line representing birth length has been included. The breastfed infants appear to exhibit linear growth patterns that resemble the early failure of developing country populations (Figures 2 and 3). However, this does not refute the conclusion that early and pronounced growth failure occurs in many developing country populations. If one takes into account the differences in the Y axis scales between Fig 3 and the earlier figures, there is no real comparison between the magnitude of deviations seen in breastfed infants reared under good conditions and those seen in many indigenous populations (where most infants were still fed at the breast but not exclusively, and where the other aspects of the environment were less desirable). However, data such as those form the DARLING study do raise important questions about the age at which true growth faltering begins - a question that has important practical implications for the definition of the age range for targeting complementary feeding.

The first line of argument, then, is simply that failure to grow is an active process during the period 6-18 months and perhaps beyond although the relative magnitude of failure after the first two years is greatly diminished. If this is the period of active failure, it is also the period in which one would expect the greatest absolute impact of dietary or other interventions aimed specifically at growth faltering. Conversely, the perpetuation of the failed growth in developing country situations cannot be interpreted as marking the end of potential responsiveness. It could simply imply that whatever constrains growth continues to operate as long as the setting remains the same.

Evidence from Secular Trends

Populations in many of the developing countries exhibit secular trends in length and weight. In China, such effects have been very strong in recent years. Large surveys were conducted in each of 1975 and 1985, using generally similar sampling designs. It is possible to examine the secular trends in detail. Fig 5A presents published observations (group means by age intervals) from the two "Nine City Surveys" (Zhang et al, 1988). Only data for urban males are presented although similar effects can be seen in urban females and in the rural groups. Again, the early departure from the NCHS reference population is seen in each year. However, it is readily apparent that children in the 1985 survey had grown more, at the same age, than had those examined in 1975 - the secular effect.

If the inferences stated earlier - that the major part of growth failure occurs between 6 and 24 months, and that this is the period when greatest response to an improved environment is to be expected, are correct, one would expect that the strongest secular effects might be seen in this age window.

Fig 5B presents the change in z scores, by age interval, between these two surveys. As expected, the magnitude of change increases after about 6 months, hits a maximum before 2 years and then declines somewhat with increasing age. It must be recognized that these are cross sectional data drawn from a population in which a secular trend is active. When we look at achieved size we are looking at a cumulative history of events occurring in earlier years. The apparent decline of the increment in z score with increasing age can be explained as a function of the secular trend - each succeeding cohort of children (each age group) passed through the period of active failure at a different time, the older the children, the earlier was there exposure to the environment surrounding growth failure and the earlier that exposure, the less favourable was the environment of growth. Secular improvements in that environment would be expected to have had less impact on the growth of older children (Beaton, 1992). The absence of major effect before about 6 months of age again reflects the fact that very young Chinese infants (in both 1975 and 1985) exhibited rates of linear growth that were similar to those seen in the reference North American population.

The inference to be drawn from this picture is that a "natural" intervention, a progressive improvement in childhood environment, appears to have exerted its greatest effect in the 6-24 month period. It would be interesting to analyze secular trends in this manner in other data sets as these progressively become available (Beaton, 1992).

Experience from Intervention Programmes

Retrospectively, one reason that the programmes reviewed by Beaton and Ghassemi may not have had more impressive effects on physical growth may have been that few of these programmes reached children under the age of 2-3 years. Preliminary reports of the Tamil Nadu intervention, targeted toward young children who appear to be faltering in growth (detected through growth monitoring), suggest that it may be having much greater impact on growth.

It should not be a surprise that failing to grow is most apparent in the 6-18 month period - or that response to intervention is more easily seen in this period. This is the period of very rapid growth in populations of industrialized countries. It is also the period when diarrhoea and other infectious diseases become common in many developing countries. The evidence is quite clear that recurring diarrhoea has a major inhibitory effect on young child growth (Tomkins, 1989). Both reduced food intake and intestinal malabsorption seem to play a part in the impact of diarrhoea. It is likely that recurrent infections also have direct impact on the growth process, unrelated to decrease in food intake or nutrient absorption (Beaton et al, 1992).

Lutter et al (1989) provided a very important analysis of data from a major supplementary feeding trial in Bogota, Colombia. Their analyses showed that the difference between supplemented and unsupplemented children in attained height at 36 months was about 3 cm. However, then the comparisons were controlled for the occurrence of diarrhoea (number of days with diarrhoea), the increment rose to 11 cm. The authors interpreted these data to suggest that while supplementary feeding had not affected the incidence of diarrhoea, it had overcome the negative effects of diarrhoea on growth. At the same time, supplementary feeding did not increase achieved height to the levels seen in the NCHS reference population - it had appreciably reduced the deficit, particularly in children exhibiting frequent or prolonged diarrhoea (see Fig 10 presented later).

Lutter et al (1990) also examined the age-specific responsiveness of children in the Bogota study. They reported that the two periods of greatest response coincided with weaning (3-6 months of age) and the peak period of incidence and duration of diarrhoea (9-12 months of age). They suggested that "targeting supplementation programmes to coincide with periods of high nutritional risk should maximize their effectiveness".

In an early study of supplementary feeding, Gopalan et al (1973) were able to show significant linear growth improvements in interventions with children as old as 48-60 months. The reported results are shown in Table 2. The magnitude of effect was greater in young children but effects were seen up to the age of 5 years. Again it is emphasized that in the present discussion, an attempt is made to differentiate between age limits for any biological effect and age windows for programmatically substantial biological effects. Gopalan et al (1973) also reported that linear growth of the supplemented children was not appreciably affected by measles occurring in the study population; the growth rate of the unsupplemented children was depressed by measles. This aspect is in keeping with the recent findings of Lutter et al for diarrhoea, discussed above.

Table 2. Gain In Length During Supplementary Feeding in Hyderabad, India

(Treated - Control)

AGE (years)

Difference in Rate of Height Gain (cm/14 months)

1-2

2.8

2-3

1.7

3-4

1.7

4-5

1.1

Based on Gopalan et al (1973)


Perhaps a bit of perspective would help. Fig 6 presents the length increments of the median lengths in the NCHS reference population. Increments have been estimated for successive three month intervals. It is readily apparent that growth rate is initially very high and falls off quickly with age although, of course, growth continues for many years. Also marked in Fig 6 is the age period in which active growth failure appears to be seen. This is not unexpected. Periods of rapid change are intuitively expected to be periods of vulnerability. Fig 7 displays the Gopalan et al (1973) study results (now expressed as growth rate per 3 months but extending across the 14 month intervals examined by Gopalan et al) superimposed on the growth rates inferred from the reference population. What is very noticeable is that the supplementary feeding appears to have supported growth rates that are comparable to those seen in the reference population and significantly higher than those seen in the control. This was noted also by Gopalan et al (1973). There is no indication that the children "cvergrew" as a catch up. One interpretation of this could be that the "maximum" growth rate might be set by age-related physiological mechanisms. In this perspective it might be concluded that the Gopalan study was extremely effective, restoring growth rate to expected levels for age (Habicht and Butz, 1985). However, one might also suggest that the study illustrates the relative ineffectiveness of supplementation of older children -while preventing further loss of growth potential, there is little evidence of restoration of past losses.

What then about the reanalysis of Bogota data by Lutter et al (1990)? Fig 8 portrays the Bogota data in the same manner as were the data of Gopalan et al (1973) in Fig 7. Again the plot suggests that the intervention was very effective in older infants in preserving growth rates close to age expectations (even in the presence of diarrhoea). It may have fallen a bit short in younger ages even though there was still evidence of significant effect. What is also interesting is that, as would be expected, the absolute difference in growth rates between supplemented and unsupplemented children was greater in the very young than the older. Again it appears that while responses can be generated up to three or more years, these do not appear to represent true catch up as much as what might be called "damage control", or prevention of further losses. The practical implication is that one might expect to see greater impact of food distribution programmes during the period of physiologically normal very rapid growth - a window of opportunity. There is an inference being drawn, without specific test in any of the data sets reviewed here, that if the opportunity is missed, the growth faltering marks a real departure from genetic potential. This then leads us to the very old question - "can true catch up growth occur and if so, till what age?"

For many years, John Waterlow has attempted to collect data that would address this question. In a 1988 article (Waterlow, 1988) he again pointed out that there is a well documented clinical literature from industrialized countries showing that given good clinical care, children recovering from a number of serious illnesses have the potential for catch up growth (recovery of losses incurred during the acute illness?) at least until puberty. It has been documented also that children recovering from severe clinical malnutrition and offered high intakes of energy and protein can exhibit remarkably high rates of linear "catch up" growth. Such children tend to be young (under 5) when treatment was initiated but long term follow up studies suggest that recovery continued for many years.

Dr Waterlow's question remains incompletely answered. In physiological terms linear growth continues until the bone epiphyses close. Theoretically some catch up is possible. Nevertheless, experience has tended to suggest that supplementary feeding has a smaller effect on older children than seen in the very young - without evidence of restoring past deficits. A difficulty of interpretation, of course, is that a small increment in growth rate, sustained over many years might yield major catch up yet go undetected in one or two year studies. Rat models seem to suggest that growth failure induced by severe dietary restriction in early life has lasting effects on growth after return to ad libitum feeding. In those models, the later the insult, the greater the degree of potential recovery. Perhaps there is a parallel phenomenon in the human - perhaps very early failure to grow truly alters the individual's potential to grow while later insults may be more amenable to recovery given a favourable environment. It may be, also, that the nature of the cause of growth failure is a critical determinant of subsequent responsiveness - perhaps this is why acute malnutrition cases or children with other severe illnesses seem to be more responsive to feeding than do children exhibiting the general growth faltering typical of many developing country populations. This author is unaware of evidence that would answer this intriguing question. However, it is suggested that for practical purposes we must assume that real catch-up growth would be an unexpected outcome of supplementary feeding. In offering this assertion a distinction is made between two different perspectives of catch-up growth. The phenomenon of catch up growth ("recovery") in the intervals between intercurrent bouts of infection, undoubtedly occurs in the presence of adequate food intake and is likely to be a pan of the explanation of the results reported by Lutter et al (1989) and Gopalan et al (1973) in terms of the role of supplementary feeding in the avoidance of effects in infectious diseases on linear growth. It is the longer term catch-up to restore failures occurring months or years earlier, that is in question.

FIGURE 1 CHILDREN WITH "ADEQUATE" HEIGHT-FOR-AGE UGANDA

FIGURE 2 OBSERVED ATTAINED LENGTH IN CHILDREN IN THE THREE CRSP STUDY SITES

FIGURE 3 REPORTED LENGTH OF CHILDREN IN THE GUANGZHOU DISTRICT OF CHINA

FIGURE 4 LENGTH FOR AGE Z SCORES OF BREAST FED INFANTS THE DARLING STUDY

FIGURE 5A 1975 AND 1985 CHINA "9 CITY" SURVEYS URBAN MALES A. OBSERVED LENGTH Z SCORES

FIGURE 5B 1975 AND 1985 CHINA "9 CITY" SURVEYS URBAN MALES B. AGE-SPECIFIC SECULAR TREND

FIGURE 6 LENGTH INCREMENTS BY AGE IN BOYS (cm/3 months)

FIGURE 7 RATES OF LINEAR GROWTH BY AGE CM/3 MONTHS

FIGURE 8 RATES OF LINEAR GROWTH BY AGE CM/3 MONTHS

FIGURE 9 A SCHEMATIC PORTRAYAL OF POSSIBLE PATHWAYS OF ENVIRONMENTAL EFFECTS ON CHILD DEVELOPMENT

FIGURE 10 REPORTED LENGTHS OF CHILDREN IN BOGOTA, COLOMBIA

In summary, then, three lines of evidence

a) observation of patterns of "spontaneous" growth faltering in developing countries
b) observation of age specificity in secular trends
c) examination of age-specific responses to supplementary feeding all seem to point to the same conclusion: If one wishes to prevent linear growth failure, the window of maximum opportunity is in the 6-18 month age interval.
A definite closure date cannot be assigned to the targeting of supplementary feeding when linear growth is the desired outcome. However, based on the three lines of evidence reviewed, and in keeping with the suggestion of Lutter et al (1990), the effective target range for substantive impacts on linear growth seems to be 6 months to 2 or perhaps 3 years. It is not suggested that no response can occur after that age, only that the linear growth response in older children may be of such small magnitude that its practical importance would be questioned (was this what led to the conclusions of Beaton and Ghassemi (1982)?).

Supplementary Feeding as an Adjunct to Clinical Treatment

A clearly documented extension of the above arguments is the importance of adequate dietary intake in the treatment of clinical illness and in the immediate post-illness recovery period. Whether or not linear growth is subject to influence, it is absolutely clear that losses in muscle mass accompanying prolonged or severe illness are reversible with adequate diet at any age. by inference, this suggests that food distribution linked to treatment facilities, servicing all ages, remains an important aspect of the integration of nutrition into primary health care. Many have commented upon the utter frustration of the health care worker when faced with a case of acute malnutrition and equipped with nothing but advice to offer, knowing that the family is not able to follow the advice. Providing supplemental food as a pan of treatment of illness in deprived settings yields important benefits not only to the individual under treatment but also to the responsible health worker.

Prevention of Starvation - Emergency Feeding

A situation that needs no argument is the use of food distribution in the face of a major economic, political or other disaster which disrupts the ability of the individual or family to acquire food. While we may argue about relative priorities and cut-off points for screening in such situations, few would argue that in reality, starvation and semi starvation know no age limits. A particular, and very upsetting example of this lesson is presented in the examination of the current situation of refugee feeding presented later in this symposium.

Supplementary Feeding, Work Capacity and Social Function

Physical work requires energy. That energy must ultimately come from food ingested or from progressive catabolism of body tissues. However, the energy needs of occupational activity might be met, in part, by reduction of nonoccupational activities. Thus we have a situation in which there is a physiological rationale for an association among energy intake, work capacity, and social activity. The equilibrium state is undoubtedly influenced by incentives and perceived "wants" -- thus expected effects are intertwined in the fabric of societies. IDECG has been attempting to bring together information in this area. For obvious reasons, research design is difficult. Study planning and implementation should include both physiologists and social scientists since the questions span both domains. Personally, this author has no reservation in concluding that food supplementation in older children and adults (as well as in the young children who are also responsive in terms of growth) in food-constrained environments should be expected to have social effects (Beaton, 1985). This author does not have the qualifications to render judgement on the importance of children playing, or of soccer clubs emerging (documented spontaneous associates of the introduction of supplementary feeding) - he can postulate effects on individual and community development, but cannot prove his rationalisations are valid.

Supplementary Feeding and other Aspects of Human Development

The early part of this paper focused upon linear growth and its age-responsiveness to food distribution. In fact, we probably are not interested in improving linear growth per se. There are few explicit advantages to being big; there are many disadvantageous associates of becoming small, particularly in deprived societies (Beaton, 1989). We most often use achieved size or observed growth rate as an indicator for a number of dimensions of human development. Originally this was based on epidemiologic observations of the association between achieved size/growth retardation, morbidity, mortality, and psychological development. With the INCAP long term follow up of a food intervention in young children, now beginning to appear in publication (Martorell et al, 1992), and reported in this symposium, we have direct evidence of the long term functional significance of a young child feeding intervention.

Short and moderate term effects of supplementary feeding on motor and cognitive development have been reviewed by Ernesto Pollitt (1988, 1991) and many others. There is now convincing evidence that some degree of impaired psychological development is a very frequent associate of early growth failure - and that interventions that ameliorate the growth failure also tend to improve psychological development although it is not easy to be sure what aspect of the intervention exerted the beneficial effects. Thus, for example, food distribution also changes effective income and this may have influences upon the environment of early development. Increased intake by others in the household may influence the care and attention given to the study child. It is not clear that the psychological effects are a part of the same physiological processes that lead to failure to grow. A recent SCN report (Fig 8) attempted to make this distinction (Beaton et al, 1992).

In Fig 8, based on the SCN report, growth failure is a marker of a disadvantaged environment in which a number of processes may operate with a range of developmental effects. There are at least two relevant implications of this portrayal.

i) rectifying growth failure, without modifying the environment (if that were possible) might not result in improvement in all functions. While it is entirely reasonable to target growth failure, we should always recognize that in reality we should be targeting the situation in which growth failure occurs.

ii) If the process of failing to grow and the process of altered psychological development are not physiologically linked, we should not suppose that the age range of peak responsiveness of growth marks the age range of peak responsiveness of psychological development -or for protection of other human functions. It is likely that the same early period is important for both but it is at least conceivable that the period of potential major influence extends longer for psychological benefits than for linear growth.

Pollitt and Gorman (1990), analysing data from the INCAP long term follow up of a young child food supplementation programme suggested that the physical activity response to supplemental feeding may play a role in motor development and that this in turn influences both physical activity and, through interactional behaviours, aspects of psychological development. In these analyses, measures of linear growth were as good predictors of psychological performance in the preschool years as were measures of motor development. However motor test scores at 15 months were stronger than anthropometric measures as predictors of adolescent functional measures of literacy, reading, vocabulary and maximum grade attained in school. These associations suggested that while anthropometric development was indeed an indicator of psychological development, separate pathways of effect were operating; these may have included activity-mediated pathways. Pollitt and Gorman (1990) cautioned that the analyses and interpretations presented were preliminary. Others (e.g. Grantham-McGregor et al, 1990) have challenged such a postulated activity-mediated path of effect. A recent study of total energy expenditure in Gambian and U.K. children (Prentice et al, 1990) has even challenged the notion that physical activity reduction is a pan of the coping response of infants and young children suggested from observations in experimental settings as reviewed by Torun (1990).

Clearly much work remains to be done before we understand the operational influences on early childhood psychological development and hence the expected impact of supplementary feeding. In spite of these limitations to existing knowledge, it is a reasonable supposition that there is a beneficial effect of adequate feeding of the child and household during the early developmental period,

Mention has already been made of the association between growth failure and morbidity. There is abundant evidence that smaller children are at greater risk of morbidity and mortality. There is less evidence that supplementary feeding reduces the incidence of disease episodes. However, there is a growing body of literature to suggest that the response to infectious disease, or the recovery from infectious disease is enhanced in the presence of supplementation (e.g. Gopalan, 1973; Lutter et al, 1989; Tomkins and Watson, 1989). Fig 8 suggests that the immunologic responses of the individual may be altered by dietary inadequacies affecting either susceptibility or response to infection. The literature abounds with documentation that the physical recovery from infection is influenced by the level of feeding. There is no a priori reason to believe that this effect is restricted to very early childhood although, as Lutter et al (1989) pointed out, the effects on growth are more evident in periods when disease is more likely to occur -- and when expected growth rates are high.

School Feeding

As an example of effects of supplemental feeding at later ages, one may point to the continuing uncertainty about benefits of school feeding. There is little evidence suggesting effects on linear growth except perhaps in very dire circumstances. Conversely, as Pollitt (1990) pointed out, there is evidence to suggest that attentiveness and other classroom behaviours impacting on learning are influenced by supplementary feeding in the face of chronic underfeeding. Associations seen in the descriptive data of the Nutrition CRSP studies, particularly in the Kenya data, support the hypothesis that current intake affects learning behaviour.

Pregnancy and Lactation

There are now several studies that document the fact that food distribution programmes directed toward pregnant women can have beneficial effects on intrauterine growth as measured by birth weight (e.g. Adair and Pollitt, 1983; Lechtig et al, 1976; Mora et al, 1978, 1979; Prentice et al, 1983, 1987). The effects may be relatively small in absolute magnitude but they appear to be sufficient (at least in severely constrained settings) to have impact on infant morbidity and mortality. There is less convincing evidence that supplementation of lactating women carries benefit to the nursing infant (except with regard to specific micronutrients such as vitamin A (Stoltzfus et al, 1992)).

The important perspective for food supplementation in pregnancy and lactation may be the mother rather than the infant in utero or at the breast. Women in developing countries face major responsibility for the health and well being of the household. Pregnancy and lactation in the face of inadequate food intake have an unquestionable negative impact on the mother. In turn, this is likely to have negative impact on her subsequent health and function even though that may be difficult to measure and document.

If once we accept this argument then we must also accept the logical argument that we should be concerned about the adequacy of food intake of girls before pregnancy begins. This then extends the potential beneficial age of food distribution from birth through reproduction and child rearing.

Food Distribution as a Vehicle for Micronutrient Supplementation

Food distribution programmes can serve as a vehicle for micronutrient supplementation (National Academy of Sciences, 1982). While this is seldom seen as a primary purpose, it is to be recognized that some of the functional outcomes mentioned in this paper are known to be influenced by micronutrients. To some extent, except in very carefully controlled trials, effects of the distribution programmes may be attributed in pan to improved micronutrient intake or availability. The opposite experience is very real in refugee camps - the development of clinical micronutrient deficiencies on an epidemic scale in mass food distribution programmes -the reappearance of problems that we though had disappeared by the 1950's. We may also see increased attention to the potential role of food distribution as a vehicle for micronutrients given our increased awareness of the importance of vitamin A and iron deficiencies.

Incentive and Other Related Effects of Food Distribution and Feeding

It has long been argued that school attendance is influenced by the provision of meals, whether that be mediated by an incentive effect or by an improved learning environment and hence increased self-satisfaction. There is practical evidence (e.g. Gopaldas et al, 1975) that the nature and perceived worth of food packages distributed at MCH clinics is an incentive for attendance. This was certainly a part of the argument for inclusion of vegetable oil in such packages in CRS programmes in Africa (Father Capone, personal communication, 1981). Obviously, such effects are not limited by age in any physiological sense.

Income Transfer-Mediated Effects of Food Distribution

This topic is not developed with the present paper since it has been well discussed in a National Academy of Sciences (1982) report and is discussed in the present symposium in the connotation of Food for Work. However, the opportunity must be taken to again remind all of us that even though we may think we are targeting food distribution programmes on individuals, we are really directing the food, and its implicit economic value, to "at risk" households in which those individuals live. The "leakage" of distributed food to the household as a whole should not be seen as a loss but rather as an unmeasured potential benefit (Beaton and Ghassemi, 1982; National Academy of Sciences, 1982). In this modality, there is no age limit. One might argue that there is a greater potential advantage in a household in which a younger sib is present or expected than in a household in which all other children are older. The perceived worth of the benefit achieved through these indirect effects depends upon the goals of the programme. All that can be done in this paper is to issue a plea that the potential, but usually unmeasured, benefits not be ignored.

Summary and Conclusions

The moral from these observations, is that we must assume, until established otherwise, that supplementation of underfed individuals may have beneficial functional effects beyond the ages when we can see effects on physical growth. This is not a comment on priorities for targeting. Rather, it is an observation on the topic assigned to this author. He is unable to place an age limit on the potential functional benefits of improving food availability in developing countries.

The linear growth data reviewed early in this presentation strongly suggest that to prevent growth failure, interventions must begin early, certainly within the first year and likely by six months. It is clearly recognized, as many authors have pointed out, that this presents a serious quandary. It appears that even full breastfeeding does not prevent the beginnings of growth faltering, but the introduction of supplementary foods opens the door for infectious disease with its pronounced effects on growth. The operational choice of action must be based upon a careful weighing of the existing patterns of infant feeding in the particular community and the potential benefits and risks associated with whatever action, or non-action, might be chosen. The real programmatic challenge lies in developing strategies and approaches that will promote [exclusive?] breastfeeding for the first four to six months of life -- a current focus of many programmes and activities - but will concurrently promote appropriate complementary feeding from that time onward - not a widely featured goal at present. Future food distribution programmes may have a role in active promotion of appropriate complementary feeding.

It seems unlikely that food supplementation, by itself, can totally prevent growth failure. Even in the Bogota study, Fig 10, where supplementation of the total household began early in pregnancy, and hence where we are looking at the combined effects of supplementing the mother in pregnancy and lactation as well as direct supplementation of the infant, there remains evidence of early growth faltering that was not reversed during the period of the study (to three years). Conversely the analyses by Lutter et al suggest clearly that growth rates can be "normalized" by supplementary feeding after 1-2 years. That is, the supplementary food has an effect on growth but without evidence that earlier deficits are restored. The much earlier study of Gopalan et al suggests the same thing. Their energy supplements appeared to raise mean growth rates to NCHS levels but did not, within the period of that study (to 5 years of age), restore earlier deficits.

What is also interesting and important is that both the Gopalan and Lutter analyses suggested that the effects of measles in India and diarrhoea in Colombia, on linear growth were essentially prevented even though there was no evidence that the incidence or duration of these disease conditions were affected.

Lutter et al (1989) present the argument that targeting of food distribution should be based on the risk of malnutrition - the peak ages for diarrhoea or other adverse factor effects. The rationale here might be seen as trying to use food distribution to minimize the damage done by other aspects of the environment but the focus is on minimizing adverse effects on growth, not necessarily on other functions.

From that work one would likely conclude that targeted distribution of food should begin early and could be expected to have some growth effect through at least three years of age and perhaps much longer. The absolute magnitude of the beneficial effect appears to decrease with increasing age. One might legitimately ask whether the small effects in older children are worthy of pursuit. The answer undoubtedly lies in one's perception of other beneficial effects that might move with the growth response.

It would be expected that weight as well as length would increase in the situations discussed. However, weight deficits attributable to either illness or simple underfeeding can occur at any age and are likely to be responsive to feeding, perhaps after severe illness is controlled. There is ample evidence (not presented in this paper) to support the assertion that feeding programmes targeted to persons exhibiting evidence of underfeeding will lead to weight responses and probably improved functional health.

In this paper an attempt has also been made to offer a reminder that there are other functions, apparently associated, in part at least, with adequacy of food intake and apparently responsive to food distribution and feeding. These are not necessarily marked by conventional anthropometric indices. This of course presents the programme planner and manager with a double barrelled problem - how to select target populations and how to evaluate the non-anthropometric effect of the programme. Extreme examples of this are the income transfer and incentive effects of food distribution.

The main conclusion that this author comes to was probably apparent to all before the paper was written. There are no definable age bounds for effectiveness of food distribution. Food distribution programmes have multiple effects (see Table 1 for examples), many of which are very poorly understood. Depending upon the effect one wishes to maximize there is likely a preferred target age window. If we continue to evaluate the programmes in terms of anthropometric responses and/or prevention of growth failure, we will have predetermined the target age groups. But we may also be missing important potential benefits for other ages. We must be careful that we do not act like the drunk who was looking for his car keys under a street lamp - because it was too dark to look where he lost them!

POSTSCRIPT: ADEQUATE NUTRITION AS AN UNALIENABLE RIGHT

To a not small degree, 1 am disturbed about making a presentation such as this. It is likely to again lead to long discussions about relative priorities for allocation of scarce resources - arguments which in the end lead to attempts to weigh the relative importance of different dimensions of human development and function. I have the terrible feeling that we are imposing a Western medical perspective that may lead us to miss the important underlying human issues. We cannot, we must not, accept the present condition of the great majority of the world's population as something to be dealt with through application of band aids. If food aid is a major resource at our disposal, we should seek ways of using it toward development of communities and betterment of the human condition, not just as another part of our medicine chest.

I know that several member agencies of the SCN have joined in an effort to raise adequate nutrition to the level of a human right - to make the whole of the UN family recognize that a major reason it exists, and a major reason we come together in meetings such as the SCN, is to recognize that international justice demands that we recognize the rights and needs of all peoples of this world we inhabit - and that we work. jointly, toward a common goal of betterment of the human condition.

In that perspective, debates on the appropriate ages for targeting food distribution seem almost irrelevant.

References Cited

Adair, L.S. & Pollitt, E. (1983) Outcome of Maternal Nutritional Supplementation: A Comprehensive Review of the Bacon Chow study. Am J Clin Nutr, 41, 948-978.

Beaton, G.H. & Ghassemi, H. (1982) Supplementary Feeding Programs for Young Children in Developing Countries. Am J Clin Nutr, 35, supplement, 864-916.

Beaton, G.H. (1989) Small but Healthy? Are We Asking the Right Question? Human Organization, 48, 30-39.

Beaton, G.H. (1992) Nutrition Research in Human Biology: Changing Perspectives and Interpretations. Am J Human Biology, in press.

Beaton, G.H. (1985) The Significance of Adaptation in the Definition of Nutrient Requirements and for Nutrition Policy. In: Nutritional Adaptation in Man. Sir Kenneth Blaxter and J. Waterlow (Eds). John Libbey, London, pp 219-232.

Beaton, G.H. et al (1990). Appropriate Uses of Anthropometric Indices in Children: A Report Based on an SCN Workshop. ACC/SCN State-of-the-art Series. Nutrition Policy Discussion Paper No. 7. ACC/SCN, Geneva.

Calloway, D.H., Murphy, S.P. & Beaton, G.H. (1988) Food Intake and Human Function: A Cross-project Perspective. Report to US AID. University of California, Berkeley.

Dewey, K.G., Heinig, M.S., Nommsen, L.A. & Lonnerdal, B. (1990) Growth patterns of Breast Fed Infants during the First Year of Life: The DARLING Study. In: Breastfeeding, Nutrition, Infection and Infant Growth in Developed and Emerging Countries. S.A. Atkinson, L.A. Hanson and RJ. Chandra (Eds). ARTS Biomedical Publishers, St. John's, Newfoundland, pp 269-282.

Gopalan, C., Swaminathan, M.C., Krishna Kumari, V.K., Hanumantha Rao D. & Vijayaraghavan, K. (1973) Effect of Calorie Supplementation on Growth of Undernourished Children. Am J Clin Nutr, 26, 563-566.

Gopaldas, T., Srinavasan, N., Varadarajan, I., et al. (1975) Project Poshak: An Integrated Health-Nutrition Macro Pilot Study for Preschool Children in Rural and Tribal Madya Pradesh. Vols 1 and 2 CARE INDIA. New Delhi.

Grantham-MacGregor, S., Meeks Gardner, J.M., Walker, S., & Powell, C. (1990) The Relationship between Undernutrition, Activity Levels and Development in Young Children. In: Activity, Energy Expenditure and Energy Requirements of Infants and Children. B. Schürch and N.S. Scrimshaw (Eds). IDECG c/o Nestle Foundation. Lausanne, Switzerland. pp 361-384.

Habicht, J-P. & Butz, W.P. (1985) Measurement of Health and Nutrition Effects of Large-Scale Nutrition Intervention Projects. In: A Methodology to Review Public Health Interventions: Results from Nutrition Supplementation and Water and Sanitation Projects, S.A. Esrey, J-P Habicht and W.P. Butz (eds). Cornell International Nutrition Monograph series No. 15 pp 1-36. Based on a paper previously published in: Evaluating the impact of nutrition and health programs. R.E. Klein et al (Eds). Plenum Publishing Corp.

Lechtig, A., Habicht, J-P., Delgado, H., et al (1975) Effect of Food Supplementation during Pregnancy on Birth Weight. Pediatrics, 56, 508-520.

Lutter, C.K., Mora, J.O., Habicht, J-P., Rasmussen, K.M., Robson, D.S., Sellers, S.G., Super, C.M. & Herrera, M.G. (1989) Nutritional Supplementation: Effects on Child Stunting because of Diarrhoea. Am J Clin Nutr, 50, 1-8.

Lutter, C.K., Mora, J.O., Habicht, J-P., Rasmussen, Robson, D.S. and Herrera, M.G. (1990) Age-specific Responsiveness of Weight and Length to Nutritional Supplementation. Am J Clin Nutr, 51, 359-364.

Martorell, R., Lechtig, A., Yarbrough, C., Delgado, H. & Klein, R.E. (1976) Protein-Calorie Supplementation and Postnatal Physical Growth: A review of Findings from Developing Countries. Arch Latinoamer Nutr, 26, 115-128.

Martorell, R., Rivera, J., Kaplowitz, H., & Pollitt, E. (1992) Long-term Consequences of Growth Retardation during Early Childhood. In: Human Growth: Basic and Clinical Aspects. M. Hernansez and J. Argent (editors). Elsevier Science Publishers, pp 143-149.

Mora, J.O. (1978) Nutritional Supplementation and the Outcome of Pregnancy. III. Perinatal and Neonatal Mortality. Nutr Rept International, 18, 167-175.

Mora, J.O., de Paredes, B., Wagner, M., Suescum, J., Christiansen, M. & Herrera, M.G. (1979) Nutritional Supplementation and the Outcome of Pregnancy. I. Birth Weight. Am J Clin Nutr, 32, 455-462.

National Academy of Sciences, (1982) Nutritional Analysis of Public Law 480 Title II Commodities. National Academy Press, Washington, DC.

Pollitt, E. (1988) Developmental Impact of Nutrition on Pregnancy, Infancy and Childhood: Public Health Issues in the United States. International Rev Res in Mental Retardation, 15, 33-80.

Pollitt, E. (1990) Malnutrition and Infection in the Classroom. UNESCO. Paris.

Pollitt, E. (1991) A Critical Review of Three Decades of Research on the Effects of Chronic Energy Malnutrition on Behavioral Development. In: Chronic Energy Deficiency: Consequences and Related Issues. B. Schürch and N.S. Scrimshaw (Eds). IDECG c/o Nestle Foundation. Lausanne, Switzerland. pp 77-93.

Pollitt, E. & Gorman, K. (1990) Long-Term Developmental Implications of Motor Maturation and Physical Activity in Infancy in a Nutritionally At Risk Population. In: Activity, Energy Expenditure and Energy Requirements of Infants and Children. B. Schürch and N.S. Scrimshaw (Eds). IDECG c/o Nestle Foundation. Lausanne, Switzerland. pp 279-2%.

Prentice, A.M., Cole, TJ., Foord, F.A., Lamb, W.H. & Whitehead, R.G. (1987) Increased Birth Weight after Prenatal Dietary Supplementation of Rural African Women. Am J Clin Nutr, 46, 912-925

Prentice, A.M., et al (1983) Prenatal Supplementation of African Women and Birth Weight. Lancet, 489-492.

Prentice, A.M., Vasquez-Velasquez, L., Davies, P.S.W., Luca, A., & Coward, W.A. Total Energy Expenditure of Free-Living Infants and Children obtained by the Doubly-Labelled Water Method. In: Activity, Energy Expenditure and Energy Requirements of Infants and Children. B. Schürch and N.S. Scrimshaw (Eds). IDECG c/o Nestle Foundation, Lausanne, Switzerland. pp 83-101.

Stoltzfus, R.J., Hakimi, H., Miller, K.W., Rasmussen, K.M., Dawiessah, S., Habicht, J-P. & Dibley, M.J. (1992) High-dose Vitamin A Supplementation of Breastfeeding Indonesian Mothers: Effects on the Vitamin A Status of Mother and Infant. J Nutrition. In press.

Tomkins, A. & Watson, F. (1989) Malnutrition and Infection. ACC/SCN State-of-the-Art Series Nutrition Policy Discussion Paper No. 5. ACC/SCN, Geneva.

Torun, B. (1990) Short- and Long-term Effects of Low or Restricted Energy Intakes on the Activity of Infants and Children. In: Activity, Energy Expenditure and Energy Requirements of Infants and Children. B. Schürch and N.S. Scrimshaw (Eds). IDECG c/o Nestle Foundation. Lausanne, Switzerland. pp 335-360.

Waterlow, J. (1976) Classification and Definition of Protein-Energy Malnutrition. In: Nutrition in Preventive Medicine. G.H. Beaton and J.M. Bengoa (Eds). WHO Monograph Ser no. 62. Annex 5, pp 530-555.

Waterlow, J. (1988) Observations on the Natural History of Stunting. In: Linear Growth Retardation in Less Developed Countries. J.C. Waterlow (Ed). Nestle Nutrition Workshop Series, Vol 14. Nestle Ltd. Vevey/Raven Press Ltd., New York. pp 1-12.

Zerfas, A.J. & Teller, C.H. (1990) Cross-Country Comparisons and Intra-Country Differentials in Nutritional Status and Infant Feeding: Presentation from DHS in 22 Countries and Recommendations. Unpublished report. LTS Corporation. Chevy Chase, MD

Zhang, X. & Huang, Z. (1988) The Second National Growth and Development Survey of Children in China, 1985: Children 0 to 7 years. Ann Hum Biol, 15, 289-305.

Effects of Supplementary Feeding on the Growth of Children with Infection

Juan Rivera

Institute of Nutrition of Central America and Panama

Two opposite views prevail concerning the relative importance of infection and dietary intake on growth. At one extreme there are those who postulate that the principal factor affecting the growth and nutritional status of children is the high frequency and duration of infections, particularly diarrhoeal diseases. At the other extreme are those who place more emphasis on dietary intake. Ample evidence indicates that both factors are causes of growth retardation and that the weight of each specific factor varies in different populations. I will present evidence of the interactive effects of dietary intake and infection or diarrhoea and dietary intake on growth, and argue that both views presented above are incorrect. The effect of inadequate dietary intake on growth depends on the level of diarrhoea, just as the negative effect of diarrhoea on growth depends on the level of energy intake.

Supplementary feeding has been shown to have positive effects on the growth of young children who are mild-moderately malnourished. In a literature review of controlled supplementation trials in malnourished young children of developing countries (Habicht and Butz, 1979; Rivera J. 1988), thirteen reports were found. Nine of these reports also presented evidence of improvements in dietary intake as a result of supplementary feeding. In the nine reports where improvements in dietary intake were demonstrated, effects on growth were shown, although in two of them effects were not shown for certain age groups. Of the four trials where dietary improvement was not shown, three did not show an effect on growth, the fourth did, but only in girls.

These results indicate that dietary improvements in children with mild to moderate malnutrition do result in measurable improvements in growth. In theory, young children's diets can be improved through different types of interventions including agricultural and development projects directed at improving family income, food subsidy programmes, food for work programmes, supplementary feeding programmes for young children, and others. However, such programmes often render little or no impact due to problems encountered in their implementation or in the design of the programme itself. For example, in their well known literature review of over 200 reports of supplementary feeding programmes, Beaton and Ghassemi (1982) conclude that few such programmes have resulted in measurable improvements in child growth. their review focused on actual programmes and not on carefully done supplementation trials. As they report, in some cases effects could not be well measured because the design of the evaluation was flawed, measurement was deficient, and/or the sample size was often inadequate. However, they also report that the amount of food provided was often too small to have an impact. Intended recipients did not always receive the expected benefit due to sharing between and within families. The results of this review indicate the need for improving the operation of supplementary feeding programmes in order to have a greater impact on the nutritional status of young children. To improve the operation of such programmes and to have an impact on the nutrition and growth of young children, it is necessary to identify the characteristics of the programme and of the beneficiaries which maximize the impact of supplementary feeding on nutritional status as assessed by growth. This information can be used for designing and targeting programmes and interventions. The following are some beneficiary-related characteristics that have been associated with the impact of supplementary feeding on the nutritional status of young children, as assessed by different anthropometric indicators: type of malnutrition (wasting vs stunting), degree of malnutrition, time elapsed since the onset of malnutrition, and morbidity, particularly diarrhoeal disease. In addition, the quantity of supplement intake as well as the duration of supplementary feeding are also associated with the size of the impact. Little has been published about the differential effect of the supplement quality on the nutritional status of the beneficiaries.

The rest of this presentation will focus on the evidence of an interaction between supplementary feeding and infection.

The effects of both inadequate dietary intake and diarrhoeal disease have been well established. However, the debate continues concerning the relative importance of each of these factors. Three recent publications provide evidence of interactive effects of dietary intake and diarrhoeal disease on child growth; two of these are results of supplementation trials.

The Longitudinal Study of Bogota, Colombia

Between 1973 and 1980 a longitudinal study on malnutrition and mental development was conducted in Bogota, Colombia. The sample consisted of 456 families belonging to low socioeconomic groups and with a high prevalence of undernutrition, as assessed by weight-forage, in young children. Lutter et al, analyzed information from 241 children who were followed from birth to 36 months of age. The families of these children had been randomly assigned to either a supplemented or an unsupplemented group. Supplements were distributed weekly at a field station and were consumed at home. In order to ensure adequate food intake by the target children, supplements were provided to all family members and were designed to meet a substantial portion of the Recommended Dietary Allowances of the family. At 36 months of age, length, diarrhoeal morbidity, and dietary intake were compared between supplemented and unsupplemented children. Supplemented children consumed 220 and 253 mcal/d more than the unsupplemented group at 18 and 36 months respectively. Differences in the incidence of diarrhoea between the two groups were not statistically significant. Results showed a statistical interaction between diarrhoea and supplement groups. In supplemented children, attained length at 36 months of age was not affected by diarrhoea. In non-supplemented children, however, diarrhoea did have a negative effect on attained growth. In the unsupplemented group, each day with diarrhoea was associated with a reduction of 0.03 cm in attained length at 36 months. The cumulative effect of this deficit in unsupplemented children with the greatest degree of diarrhoea was 5 cm. In the absence of diarrhoea, no difference was found between supplemented groups, indicating that supplementary feeding has a positive effect on growth only in children who have diarrhoea.

The INCAP Longitudinal Study in Guatemala

Between 1969 and 1977 INCAP conducted a longitudinal study of growth and development in four rural villages in Guatemala. Two of the villages were randomly selected to receive a high-energy high-protein supplement (atole) and the other two villages received a low-energy no-protein supplement (fresco). Both beverages had similar amounts of vitamins and minerals. Martorell, Rivera and Lutter examined the relationship between the percentage of time ill with diarrhoea and growth in length, by type of supplement, in children between 3 and 36 months of age. Children supplemented with atole consumed significantly more supplemental energy than those supplemented with fresco (124 Kcal/day). Energy from the home diet did not differ significantly between the two treatment groups. No difference was observed in the percentage of time ill with diarrhoea between supplement groups. In children receiving atole, the percentage of time spent with diarrhoea did not have a statistically significant effect on growth. In children receiving fresco, the percentage of time with diarrhoea was negatively and significantly related to diarrhoea. For this group, the difference in growth between those with the greatest and the least degrees of diarrhoea was 3.7 cm. As opposed to the findings in the Bogota study, in the absence of diarrhoea, growth of children supplemented with atole was 1.5 cm greater than that of children supplemented with fresco.

A Study in Lima, Peru

Brown and colleagues studied the effects of energy intake (including breastfeeding) and diarrhoeal prevalence on the weight gain of poor urban Peruvian infants. Between 6 and 12 months of age the interaction between energy intake and prevalence of diarrhoea on weight gain was found to be statistically significant. Among infants whose usual energy intake was greater than 75% of the recommended intake, no relationship between diarrhoea prevalence and weight gain was found. However, among infants whose usual intake was less than 75% of the recommended intakes, diarrhoeal prevalence was negatively related to weight gain.

The evidence from the three studies presented argues that the effects on growth of supplementary feeding and diarrhoeal disease are interactive. This implies that supplementary feeding is likely to have a greater impact on growth in children with a high incidence of diarrhoea. Therefore, investments in improving dietary intake (like food aid) directed to groups of children with a high prevalence of diarrhoea is not only desirable but is likely to be more cost effective than such investments directed to groups of children with a low prevalence of diarrhoea. In the absence of diarrhoea, supplementary feeding had an effect on growth in Guatemalan but not in Colombian children. This may be due to a larger deficit in the baseline dietary intake and in the nutritional status of Guatemalan children. A possible mechanism explaining the larger effects in children with diarrhoea is a possible increase in appetite immediately after the diarrhoeal episode. To the extent that appropriate foods are available to meet the increased appetite, the negative effects of diarrhoea on energy balance may be offset.

Bibliography

Beaton G.H. & Ghassemi H. (1982) Supplementary Feeding Programmes for Young Children in Developing Countries. Am.J.Clin.Nutr., 35, 864-916.

Brown K.H., Peerson J., Kanashiro H., Lozpz de Romana G, Black R. (1991) The Relationship between Diarrhoeal Prevalence and Growth of Poor Infants Varies with their Age and Usual Energy Intake. FABES J., 5, A 1079 (#4113).

Habicht J.P. & Butz W.P. (1979) Measurement of Health and Nutrition Effects of Large Scale Nutrition Intervention Projects. In: Evaluating the Impact of Nutrition and Health Programmes. R.E. Klein, (ed). New York: Pleanum Publishing Corporation.

Lutter C.K., Mora J.O., Habicht J.P., Rasmussen K.M. Robson D.S., Sellers, S.G. Super C.M., Herrera M.G. (1989) Nutritional Supplementation: Effects on Child Stunting because of Diarrhoea. Amer.J.Clin.Nutr., 50, 1-8.

Martorell R., Rivers J. & Lutter C.K. (1990) Interaction of Diet and Disease in Child Growth. In: Breastfeeding, Nutrition, Infection and Infant Growth in Developed and Emerging Countries. Atkinson S.A, Hanson, L.A., Chandra R.K., eds. ARTS Newfoundland, Canada, 307-21.

Rivera, J. (1988) Effect of Supplementary Feeding upon the Recovery from Mild-to-Moderate Wasting in Children (Ph.D. Thesis). Ithaca, N.Y., Division of Nutritional Sciences, Cornell University, 1988.

Feeding Latin America's Children*

Philip Musgrove

World Bank

*Dr Musgrove's presentation was based on the World Bank publication of this name (Musgrove, 1991); the text here is taken directly from the article published in January 1993 in The World Bank Research Observer, 8, 23-45 - reproduced with the permission of the World Bank.
More than US $1.6 billion is spent annually on 104 programs in nineteen Latin American and Caribbean countries to subsidize or provide food for people supposedly at risk of malnutrition. This amount constitutes only 0.2 percent of these countries' gross national product. If there is no double-counting, these programs reach more than 80 million people, or 21 percent of the population, at a cost of $20 per beneficiary or $4 per capita. Yet some 10 million children are malnourished, which suggests that the expenditures are poorly directed or ineffective. There is little hard evidence that these programs are preventing much malnutrition; even curative results are seldom measured. The effort is too small in some countries with great needs, while other countries have nearly eliminated malnutrition. Where coverage is high, programs - although generally targeted and with sensible criteria - do not always reach the neediest. They may also fail to provide enough food or to combine food with the health care and nutritional education necessary to attack all three root causes of malnutrition: poverty, disease, and ignorance. The evidence, limited mostly to program inputs rather than results, suggests that greater progress against undernourishment is possible even with current spending levels.

It is a commonplace that malnutrition is unacceptably widespread in Latin America and the Caribbean, particularly among preschool-age children (United Nations 1989); that economic growth alone would be very slow to eliminate it; that other means are available to reduce malnutrition more rapidly and surely; that among these means are targeted programs providing some combination of food, nutrient supplements, health care, and education to vulnerable children and their families; and that such programs need not cost an enormous share of a country's resources. These conditions create a need to understand what efforts are under way in the region to reduce malnutrition through such programs, to determine what makes programs more or less effective or efficient, and to improve them where they appear inadequate and extend them where they appear to work as intended.

Answering these questions, which are at the heart of any effort to end malnutrition, is difficult primarily because of poor information. One problem is that information on the prevalence of malnutrition is incomplete and out of date. There are survey data for one country each for 1990 (Paraguay) and 1989 (Brazil) and for two for 1988 (El Salvador and Mexico). But some data are from as long ago as 1978, and those for several countries refer to 1981-82. Many small surveys refer to particular sub-populations, usually defined by geographic region, but it is impossible to tell how well they represent a country's population. And even national samples may be based on 2,000 or fewer children, so that estimates of the prevalence of wasting -low weight for height, the least common form of malnutrition- are derived from just a few dozen underweight children.

A second problem is that almost nothing is known about whether existing programs prevent children from becoming malnourished or restore malnourished children to normal nutritional status. Prevention is intrinsically hard to measure, but curative results are easy to detect; yet even this information is almost universally missing. As a result programs tend to be judged only by measures of input, such as how many people they enroll, how much food they distribute, or how much it costs to reach one beneficiary or to deliver a given amount of calories, protein, or other nutrients.

A third problem is that the typical program provides surprisingly little information even about its inputs, beneficiaries, and costs. Important data are often missing, incomplete, or open to suspicion. For example, an internal World Bank inventory of nutrition programs in 1990 found that program-level data on beneficiaries in Brazil do not correspond well to household-level responses to questions about enrollment or participation. Using measures of inputs to make inferences about outcomes can be misleading, particularly if inputs are poorly measured to start with.

To enlarge the information available on nutrition programs in Latin America, the World Bank produced a detailed inventory of supplementary feeding programs for mothers and children (and some other programs) under way in mid-1990, with as much detailed information as possible about implementation of programs with the highest coverage in each country (Musgrove 1991). Information was obtained on 104 programs in nineteen countries. This article summarizes the findings of that larger study. Because there are so many programs, tables show data for entire countries only; some program-level data are displayed graphically.

Nature, Size, and Consequences of Malnutrition

In an entire population "malnutrition" can include obesity and diet-related risks of cardiovascular disease, diabetes, or other health problems, but for young children, only two conditions matter. One is being too small according to some physical criterion, defined by a reference population of healthy, well-nourished children (Zerfas 1991; Carlson and Wardlaw 1990; Frisancho 1990). Because this condition is associated with inadequate intake or utilization of macronutrients, it is referred to as protein-calorie or protein-energy malnutrition. The other condition encompasses all the specific or micronutrient deficiencies, of which the most important are iron-deficiency anemia and inadequate intake of vitamin A and iodine.

Protein-calorie malnutrition is defined by low weight for age (underweight), low weight for height (wasting), or low height for age (stunting). Stunting indicates failure to grow normally over extended periods, and it is often considered a sign of chronic malnutrition, such as routine failure to eat enough. It can also result from repeated brief interruptions to growth caused by frequent episodes of illness. Wasting is often described as acute malnutrition, because it tends to result from actual loss of weight. A stunted child may have survived various episodes of wasting, with subsequent recovery of weight but not of normal height. Being underweight for age is a common consequence of stunting, but it can, of course, also result from wasting. Underweight is therefore sometimes used as a measure of global malnutrition reflecting unknown proportions of stunting and wasting.

Children who are malnourished during their early years were often born small or prematurely because their mothers were malnourished during pregnancy. If one measures a child's age from conception rather than birth, then low birth weight at normal gestational age is a form of malnourishment. There is thus no clear separation of maternal and infant malnutrition, which justifies the practice of directing nutrition programs to pregnant women as well as to young children.

In the reference population of U.S. children adopted as a standard by the World Health Organization (WHO) and UNICEF (United Nations Children's Fund), the distributions of all three anthropometric indicators are essentially normal. This means that there is no natural limit below which one can be sure a child is malnourished; any cutoff point is arbitrary. It is becoming standard practice to define moderate malnutrition by values lying between two and three standard deviations below the mean, and severe malnutrition by values more than three standard deviations below the mean, with both the mean and the deviation defined by the reference population.

Defining malnutrition relative to a point on a reference distribution has three important consequences. First, even in the reference population, 2.3 percent of children get identified as malnourished (whether they are or not); therefore, the prevalence of malnourishment should be estimated by the excess over this share. Second, if children in the reference population who look malnourished are considered "small but normal" rather than unhealthy, small-but-normal children can similarly be present in any other population. Thus, aside from the question of excess prevalence, there is a problem of identifying individuals. A child whose height or weight falls more than two standard deviations below the mean may not really be malnourished; if the prevalence of malnutrition in the population is low, this error becomes more likely. Conversely, a child not classified as malnourished may in fact have grown more slowly than is normal. These errors matter little when estimating prevalence in large groups-in part because they cancel each other out-but they complicate assessing how well a program is targeted. And third, the status of an individual child is better judged by growth through time than by current weight and height. Every child can be thought of as having a growth path, and failure to grow along that path is a surer sign of trouble than simply falling far below the mean.

The most recent country-level data on the prevalence of malnutrition in eighteen Latin American countries show that some 10-11 million preschool children in these countries are underweight for age (table 1). (There are no recent national data for Argentina.) Even more-perhaps 15 million-are moderately or severely stunted. Wasting is much less common, affecting perhaps 1-2 million children (no estimate is available for Brazil). Comparison across countries is hampered by differences in timing, definition, and accuracy of national estimates. Even so, there is no doubt that the bulk of malnutrition is found in a few countries with very large populations (Brazil and Mexico), with substantial concentrations of poverty (Bolivia, Guatemala, Haiti, and Peru), or both. Prevalence of low weight for age appears to exceed 30 percent in Brazil, Guatemala, and Haiti and to fall below 10 percent only in Chile, Costa Rica, Jamaica, Paraguay, and Uruguay. The prevalence of stunting is estimated to equal or exceed 40 percent in Bolivia, Guatemala, and Haiti, while the prevalence of wasting never surpasses 10 percent and often is less than 1 percent.

Causes of Malnutrition and Interventions to Reduce It

How to combat malnutrition depends on its causes, which fall into three broad classes: poverty, illness, and ignorance. For poverty to be the only cause of malnutrition, it is not enough for the two phenomena to be closely associated. It is also necessary that increases in income be translated into increases in food consumption and reductions in the prevalence of malnutrition; removing the cause should remove the effect. The close association of poverty and malnutrition (World Bank 1979; Gray 1982) has sometimes led to the erroneous conclusion that illness and ignorance are minor contributors and that the only intervention necessary is to give poor families more money. This conclusion ignores a second finding of these same studies, which is a very low tendency to spend additional income on food, particularly on calories.

The evidence that responsiveness of food intake to income changes is very low has in turn been challenged on two fronts. Elasticities are estimated to be much larger among the poor than for the population as a whole (Musgrove 1985; Musgrove 1989, table 22; Senauer 1990), so that poverty appears as an important cause of malnutrition. And since nutritional state does not depend on food consumption alone, higher income can mean better nutrition because of its contribution to the other determinants of nutritional status, such as health and education (Schiff and Valdes 1990). In any case, whether attacking poverty in general is the best way to improve nutrition also depends on how quickly poverty can be reduced.

Illness contributes directly to malnutrition in three important ways. A sick child may eat too little or fail to eat a balanced diet. Illness can lead to loss of nutrients, particularly during diarrhea, or interfere with nutrient absorption. And parasites such as helminths compete with the host for the nutrients in food, causing iron-deficiency anemia through loss of blood. By some estimates sickness is the most important cause of malnutrition and the factor that does the most to explain why some poor children are malnourished and others-even siblings-are not (Shrimpton 1984). Illness is more likely to lead to malnutrition among the poor than among those who are better off, because poor children are likelier to have marginal nutrient reserves at the start of an illness and because poor families can less easily buy food or medical care to promote a child's recovery.

Of the three general causes of malnutrition, ignorance is the most difficult to measure and the most subject to emotional interpretation. To say that children are malnourished because their parents do not know how to care for them sounds like blaming the victims for the problem. It is not true that the poor are malnourished simply because they are ignorant about what they should eat and so spend their limited incomes on the wrong foods. But there are three important ways that more specific kinds of ignorance contribute to malnutrition. The first is that people may know nothing about micronutrients, especially vitamins, so they fail to eat even the cheap and readily available sources. The second is ignorance about disease and its causes and consequences. Valuing good health is not enough; neither is depending solely on one's immune system. Third, people are often specifically ignorant about how to care for very young children: they may undervalue breastfeeding, or believe that food and even water should be withheld from a sick child, or simply not recognize that growing children need to eat more, in relation to their weight, than adults.

If there are three general causes of malnutrition, there are three corresponding interventions to combat it. One is to create or transfer income, in the form of money or goods such as foodstuffs. The second is to prevent or cure the illnesses that contribute to childhood malnutrition. This can include providing health care services, improving sanitation, and distributing micronutrient supplements. Because such supplements may be fortified foodstuffs, there obviously is no clear line between giving away food (basically an income transfer) and giving away a particular nutrient (a health intervention). The third intervention is to change what people know or believe-to educate them about nutrition, child care, and health.

Table 1. Estimated Prevalence of Malnutrition among Children under 5 Years, by Country

Country

Type of malnutritionb

Year

Global (number)

Global (percent)

Stunting (percent)

Wasting (percent)

Bolivia


1981

609,000

14.5

42.7

0.7

1987c,d

760,200

18.1

-

-

Brazil

1989e

5,024,200

30.7

-

-

Chile

1986

35,000

2.5

9.6

0.5

Colombia


1977-80

684,700

16.7

22.4

4.6

1986f

487,900

11.9

22.7

1.0

Costa Rica


1982

24,000

6.0

7.8

2.0

1987b,c

10,800

2.7

6.4

-

Dominican Republic


1986f

125,000

12.5

20.8

2.3

1987b

58,000

5.8

12.8

1.8

Ecuador

1986

264,000

16.5

34.0

1.7

El Salvador

1988

123,200

15.4

26.8

2.3

Guatemala

1987f

502,500

33.5

57.9

0.4

Haiti

1978

342,000

37.4

39.6

8.9

Honduras

1987

164,800

20.6

33.9

1.9

Jamaica


1985g

43,800

14.6

7.1

5.1

1989h

21,900

7.3

29

2.1

Mexico

1988i

1,584,600

13.9

22.3

6.3

Panama


1980

47,100

15.7

22.0

6.4

1980c

47,400

15.8

25.1

6.1

Paraguay

1990i

25,200

4.2

20.3

0.4

Peru

1984

428,800

13.4

37.8

0.9

Uruguay

1987

20,500

7.4

15.9

-

Venezuela

1981-82

275,400

10.2

6.4

1.3

- Not available

Note: Except as indicated, prevalence is calculated using the reference median minus two standard deviations, or according to the World Health Organization (WHO) definitions of less than 80 percent of the reference median for global malnutrition (weight for age) and wasting (weight for height) and less than 90 percent of the reference median for stunting (height for age). These two definitions coincide approximately for the 0-5 year age group; estimates based on the Gomez definition (less than 75 percent of the reference median) would be lower.

a. Data for Argentina are not included because no recent national estimates were available.
b. Data from WHO (1989), except as indicated: same data reported in Carlson and Wardlaw (1990)
c. Data from PAHO (1990), vol. 1, tables 86-88; sample sizes and age ranges not indicated.
d. Data refer to children brought to health service centers; not a representative sample.
e. Data from INAN (1990); malnutrition defined as less than 75 percent of reference median (Gomez definition).
f. Children aged 6-36 months or 4-36 months, rather than 0-60 months. Estimate of number of malnourished children assumes the same prevalence for the entire population under 5 years of age.
g. Jamaica Ministry of Health (1985).
h. Statistical Institute of Jamaica (1991)
i. World Bank staff estimates. j. Institute for Research Development/Westinghouse (1991).

Source: Musgrove (1991)

Consequences: Malnutrition and Human Capital

Malnutrition may cause no apparent problem because the body compensates for inadequate nutrient intake by growing more slowly or by reducing its physical activity. But compensation works only for low levels of deprivation, and even then an apparently successful compensation can be costly. A child who is anemic or calorie-deficient will be lethargic and will develop and learn more slowly, perhaps retaining a permanent deficit. Malnourished children are also subject to more frequent and more severe infections (Frisancho 1979, ch. 13; Tomkins and Watson 1989), which exacerbate the initial malnutrition, as well as interfering with their early learning and later with their schooling (Pollitt 1990). Malnutrition therefore leads to a waste of resources in both education and health, apart from the damage to children's development. Conversely, children who benefited from an effective program to protect their nutritional status may show the gains in schooling, health, and income many years afterward (Martorell 1991).

Because information on the prevalence of malnutrition is so sketchy and the consequences for health, learning, and productivity are often hard to quantify and are just beginning to be well understood (Behrman 1991), it is not possible to estimate the total economic damage caused by malnutrition in Latin America and the Caribbean. But it is increasingly clear that such damage exists, beyond the pain and suffering associated with more severe nutritional deficiencies, and that the damage is probably substantial, particularly when all the interactions among nutrition, health, and development are taken into account (Selowsky 1981).

Major Findings: A Summary

Some results of the study merit more extensive discussion, but it will be helpful first to summarize the principal findings that emerge from the review of detailed program information.

· There are now a very large number of food programs in Latin America and the Caribbean. Nonetheless, some countries still have few or no programs. Perhaps the most striking cases are Haiti, with a serious problem of childhood malnutrition but only one small maternal and child health program, and the Dominican Republic, which has only a small school feeding program covering 7 percent of the target age group.

· Food transfers per beneficiary vary enormously. Some are undoubtedly too small to have any effect on nutrition or even to alleviate poverty significantly; others may be overgenerous. However, there does not seem to be any standard for the "right" size of transfer.

· The calorie-protein composition of food transfers varies much less than the total amounts. Most programs pay no attention to possible micronutrient deficiencies, but there is no systematic evidence that the composition is incorrect or that the wrong foods are being used. The criteria for selecting foods are generally sensible, at least within the limits set by dependence on foreign donations in many programs.

· Costs per beneficiary per year vary enormously, chiefly because of variation in the amount of food provided. Food costs and total costs are highly correlated: variation in total costs primarily arises not from differences in administrative costs, but from differences in what is spent on food.

· Costs per nutrient unit (calories or protein) bear surprisingly little relation to costs per beneficiary or to the number of beneficiaries. There is no evidence of economies of scale over the large range of program sizes considered.

· If no beneficiaries are double-counted, then coverage of the target population is often very high. When several programs are directed to the same age group or when the target population spills over the intended age limits, coverage can exceed 100 percent. In countries where many people participate in one or more programs, malnutrition is due not simply to a lack of coverage, but to some combination of poor targeting, insufficient food transfers, and failure to attack the other causes of malnutrition.

· Most programs include sensible targeting criteria, but too little is known about the population excluded from programs to tell whether targeting in fact works well. Simple failure to target is generally not the only, and perhaps not the principal, reason for poor effectiveness.

· Unrestricted subsidies are rare, except in school feeding programs, where they may be justified. As the number of programs has increased in the past decade, more have been targeted on maternal and child health.

· Although most maternal and child health programs-and some programs of other types-claim to provide complementary services (health care or health and nutrition education), there is little evidence that these services correspond to needs or are effective.

· Nearly all programs suffer from logistic inadequacies. Programs need adequate storage facilities, vehicles for transport, and systems of supervision to prevent spoilage and theft. Cost-effectiveness does not mean minimizing total cost, and it particularly does not mean minimizing nonfood costs.

· Most programs proclaim reasonable objectives, but only in such vague and qualitative ways that it is difficult to tell whether they are being met. Espousing a variety of goals also leads to treating progress toward any of them as an indication of program success.

· Most programs have never been evaluated, or at least their effect on the beneficiaries has never been measured. The programs that have been evaluated show mixed and generally disappointing results, although a few evaluations are much more favorable.

Type and Number of Beneficiaries

Of the 104 programs, 54 are intended to cover infants aged 5 or younger, 31 involve pregnant or lactating women, 30 are for school children, 23 focus specifically on malnourished children, and 28 are addressed to entire families. In some cases food is regularly distributed to other classes of beneficiaries, such as the elderly or handicapped, or as a payment in kind to volunteers, day-care providers, or other collaborating personnel.

Except when a program is open to the public and does not identify its beneficiaries, it is easy to describe the intended beneficiaries. Much more difficult is finding out how many beneficiaries there really are. Some errors arise for conceptual reasons: individual beneficiaries may participate irregularly, so that the number of people benefiting at a given moment and the number benefiting over an interval are quite different. But often the difficulty of counting beneficiaries arises from a failure to keep accurate records, perhaps aided by political pressures to inflate the numbers.

On the input side of a program that involves giving away or subsidizing food, there are three fundamental quantitative variables: the number of beneficiaries, the amount of food distributed, and the expenditure on the program, all during some interval. There is much other important information, but it is generally qualitative-the mechanisms of beneficiary selection and food transfer, the nature of the food provided, and so on-or, if quantitative, it refers to other program inputs such as immunizations, micronutrient supplements, or educational interventions. From the three basic variables-people, food, and money-ratios for judging the adequacy of a program are commonly created. These ratios are the amount of food distributed per beneficiary, the unit cost of food, and the amount spent per beneficiary.

In principle, any of these ratios can be derived from the other two, since they are tautologically related-a program that spends a lot per kilogram of food will either spend a lot per beneficiary or give each beneficiary rather little food. Differences in cost per beneficiary between two programs can always be "explained" by differences in the amount of food a beneficiary gets or by how expensive that food is to obtain and distribute. Unfortunately, all three variables are subject to errors in measurement, and an erroneous value in just one variable leads to two erroneous ratios. For example, overcounting beneficiaries makes a program appear less costly per person than it really is and "explains" that result by a small food transfer per person.

The number of beneficiaries covered by the 104 programs varies greatly, from fewer than 1,000 individuals in one program in Bolivia to almost 28 million in the Brazilian school lunch program. The average is slightly under 1 million. Eleven programs have more than 3 million beneficiaries, and 18 each have more than 1 million. At the other extreme are 53 programs with fewer than 100,000 beneficiaries each; these are almost all maternal and child health programs.

To estimate the total numbers of individual beneficiaries, individuals in families covered by some of the programs had to categorized and counted. For those purposes it was assumed that 20 percent of the family individuals were mothers, 30 percent were children under 5 years, 30 percent were children of primary school age, and the remaining 20 percent were older children and adults. The total number of pregnant and lactating women could not be estimated because many programs do not report that information. With these adjustments, individual coverage, aggregated for each of the nineteen countries, reaches about 9.8 million adult women, 27.3 million preschool children, 51.9 million school-age children, and 10.3 million other beneficiaries (table 2). Ninety percent of these estimated 99 million beneficiaries are concentrated in six countries: Brazil (58 million), Mexico (almost 15 million), Peru (more than 8 million), and Argentina, Colombia, and Venezuela (more than 3 million each).

Looking at this beneficiary count by country and by type of program-take-home food distribution, direct feeding (including school feeding), or subsidy-shows essentially equal numbers of take-home and direct feeding programs (table 3). Direct feeding programs, however, cover almost twice as many beneficiaries on average-a little under a million people versus half a million for the take-home programs. Subsidy programs are the largest, on average, with nearly 2 million estimated beneficiaries. The size ranges overlap considerably, with the greatest variation found among the take-home programs (figure 1).

Comparing the program totals for preschool children in each country (Argentina and Panama are excluded for lack of data) to the numbers of malnourished children (from table 1) shows a rough proportionality, but a malnourished child's chance of participating in some kind of program still differs enormously according to country (figure 2). Pan of this variation arises, of course, because maternal and child health programs are directed not only at children who are malnourished, but also at children at risk. If the group at risk were adequately covered by programs that effectively prevented malnourishment, there would be more beneficiaries than malnourished children. What is clear is that enrolling large numbers of beneficiaries can still leave many children suffering from malnutrition.

Information on beneficiaries as a percentage of the target population by country and by type of beneficiary is fairly complete for age groups (table 4); much less is known about coverage according to other criteria, such as poverty or the presence of specific risks of malnutrition. Coverage of children varies widely among countries: from 1.9 percent in Haiti to 96 percent in Chile for children under 5 years of age (the average is 50 percent) and from 7 percent in the Dominican Republic to more than 100 percent in Brazil and Uruguay for school-age children (the average is 80 percent). (Coverage above 100 percent reflects two kinds of double-counting: the same child may participate in more than one program, and the primary enrollment rate can exceed 100 percent because repetition keeps some children in school beyond the normal age limit.)

The proportion of the total population covered also varies widely, with one person in every seven receiving some kind of benefit from a food or nutrition program in ten countries (Bolivia, Brazil, Chile, Colombia, Costa Rica, Guatemala, Honduras, Mexico, Peru, and Venezuela). Finally, under the generous assumption that all programs are limited to the poor, or at least concentrated on them, 55 percent of the estimated poor population of eleven countries is reached by some kind of program, with coverage ranging from 5.8 percent in Haiti to 88 percent in Brazil and Uruguay. The higher the coverage, of course, the less plausible is the assumption of perfect targeting.

These estimates suggest three conclusions. First, malnutrition does not persist solely because the programs directed against it are too few and too small. Second, the relatively high coverage is far from universal; in some countries only a small share of a target population appears to participate, and fewer than half of schoolchildren benefit in seven of the countries studied. Third, "coverage" is just head-counting unless more is known about the characteristics of beneficiaries and what a program does for them.

How adequate is this coverage, relative to the magnitude of malnutrition in each country? And how large and how adequate is the expenditure on food and nutrition programs? Before trying to answer these questions, it will help to look at what food the programs actually provide and how much the programs actually cost.

Food Distributed and Program Costs

All the data refer to food "delivered," that is, brought to the point where beneficiaries can eat it. Delivered amounts may not represent additional consumption by the intended beneficiary, however, for three possible reasons. First, the food may be wasted and therefore not be eaten by anyone. Second, the food may be consumed by someone other than the person for whom it was intended. By definition, this problem can arise with take-home programs but not with programs that directly feed the beneficiary. For that reason take-home programs commonly include some effort to persuade the family not to share the food, and in some cases they provide food designed to be palatable only to the intended recipient. Third, even if the beneficiary eats all the food provided, wasting none and sharing none, total intake may increase by much less than the transfer. The family may compensate by feeding that member less of other foods, so as to be "fair" to the other members. This is most likely to occur when the beneficiary is a small child who receives meals away from home.

Table 2. Total Number of Beneficiaries by Country and Type of Beneficiary (Redistributed)

Country

Women

Children under 5 years

Primary school children

Older children and other adults

Total number

Argentina

383,263

968,228

1,827,182

31,787

3,210,460

Bolivia

188,743

482,227

317,764

110,288

1,099,022

Brazil

4,670,713

13,957,298

33,142,040

5,936,408

57,706,459

Chile

116,316

1,342,208

461,041

1,910

1,921,475

Colombia

465,106

1,197,966

1,559,477

50,079

3,272,628

Costa Rica

11,527

64,554

386,135

-

462,216

Dominican Republic

-

107,095

70,000

-

177,095

Ecuador

28,050

172,955

795,939

-

996,944

El Salvador

41,398

110,820

224,804

-

377,022

Guatemala

14,137

254,292

1,239,520

-

1,507,949

Haiti

8,630

16,946

-

-

25,576

Honduras

32,256

140,000

548,578

16,129

736,963

Jamaica

40,000

169,620

95,000

80,000

384,620

Mexico

1,931,149

5,037,724

5,941,721

1,945,149

14,855,743

Panama

-

-

58,860

-

58,860

Paraguay

18,365

21,565

82,500

-

122,430

Peru

1,712,065

2,704,801

2,478,194

1,452,130

8,347,190

Uruguay

7,122

139,564

367,266

406,516

920,468

Venezuela

110,266

448,185

2,323,685

240,400

3,122,536

Total

9,779,106

27,336,048

51,919,706

10,270,796

99,305,656

- Not available.

Note: Beneficiaries covered as families were redistributed as follows: 20 percent to women (mothers), 30 percent to children under 5 years, 30 percent to primary school children, and 20 percent to older children and other adults.

Source: Musgrove (1991).

Table 3. Numbers of Programs and Beneficiaries by Country and Mode of Food Distribution

Country

Take-home

Direct feeding

Food subsidies

Programs

Beneficiaries

Programs

Beneficiaries

Programs

Beneficiaries

Argentina

1

990,521

3

2,219,939

n.a.

n.a.

Bolivia

8

333,006

9

366,016

1

400,000

Brazil

3

16,031,162

1

27,993,257

3

13,682,041

Chile

1

1,354,404

6

567,091

n.a.

n.a.

Colombia

1

29,117

3

2,385,028

2

858,483

Costa Rica

3

58,858

2

403,358

n.a.

n.a.

Dominican Republic

1

89,095

1

88,000

n.a.

n.a.

Ecuador

1

66,045

6

920,899

1

10,000

El Salvador

1

152,218

1

224,804

n.a.

n.a.

Guatemala

3

268,429

1

1,239,520

n.a.

n.a.

Haiti

1

25,576

n.a.

n.a.

n.a.

n.a.

Honduras

2

107,520

2

629,443

n.a.

n.a.

Jamaica

1

4,620

1

80,000

n.a.

n.a.

Mexico

3

610,633

n.a.

n.a.

7

14,245,110

Panama

n.a.

n.a.

1

58,860

n.a.

n.a.

Paraguay

1

21,565

2

100,865

n.a.

n.a.

Peru

3

527,435

4

4,554,895

2

3,264,860

Uruguay

6

638,242

3

263,880

1

18,346

Venezuela

3

168,266

5

2,954,270

n.a.

n.a.

Total

43

21,476,712

51

45,050,125

17

32,478,840

n.a. Not applicable
Source: Musgrove (1991).
Table 4. Program Coverage as a Percentage of the Total Population, by Country and Type of Beneficiary

Country

Population under 5

Primary school population

Total population

Poor population

Total (millions)

Covered

Total (millions)

Covered

Total (millions)

Covered

Total (millions)

Covered

Millions

Percent

Millions

Percent

Millions

Percent

Millions

Percent

Argentina

3.2

0.968

30.3

3.2

1.827

57.1

31.9

3.211

10.1

-

-

-

Bolivia

1.2

0.482

40.2

1.2

0.318

26.5

7.1

1.099

15.5

-

-

-

Brazil

18.7

13.957

74.6

27.0

33.142

122.8a

147.3

57.706

39.2

31.8

27.900b

87.7

Chile

1.4

1.342

95.9

2.6

0.461

17.7

13.0

1.921

14.8

4.9

1.308

26.7

Colombia

4.1

1.198

29.2

3.3

1.559

47.2

32.3

3.199

9.9

-

-

-

Costa Rica

0.4

0.065

16.3

0.4

0.386

96.5

2.7

0.462

17.1

-

-

-

Dominican Republic

1.0

0.107

10.7

0.1

0.070

7.0

7.0

0.177

2.5

-

-

-

Ecuador

1.6

0.173

10.8

1.6

0.796

49.8

10.3

0.997

9.7

4.4

0.997

22.7

El Salvador

0.8

0.111

13.9

0.8

0.225

28.1

5.1

0.377

7.4

2.7

0.377

14.0

Guatemala

1.5

0.254

16.9

1.5

1.240

82.7

8.9

1.508

16.9

2.4

1.508

62.8

Haiti

0.9

0.017

1.9

0.9

-

-

6.4

0.026

0.4

4.4

0.260

5.9

Honduras

0.8

0.140

17.5

0.8

0.549

68.6

5.0

0.737

14.7

3.6

0.737

20.5

Jamaica

0.3

0.170

56.7

0.3

0.095

31.7

2.4

0.385

16.0

1.0

0.085

8.5

Mexico

11.4

5.038

44.2

11.4

5.942

52.1

84.6

14.856

17.6

-

-

-

Panama

0.3

-

-

0.3

0.059

19.7

2.4

0.059

2.5

0.6

0.059

9.8

Paraguay

0.6

0.022

3.7

0.6

0.083

13.8

4.2

0.122

2.9

1.4

0.122

8.7

Peru

3.2

2.705

84.5

3.2

2.478

77.4

21.2

8.347

39.4

-

-

-

Uruguay

0.3

0.140

46.5

0.3

0.367

122.3a

3.1

0.920

29.7

0.6

0.532

88.7

Venezuela

2.7

0.448

16.6

4.6

2.324

50.5

19.2

3.123

16.3

11.3

2.994

26.5

Total

54.4

27.337

50.3

65.0

51.921

79.9

414.1

99.232

24.0

69.1

36.879

53.4c

- Not available.

a. Coverage can exceed 100 percent, reflecting participation In more than one program and/or children who remain in school beyond the normal age limit (5-9)

b. Assuming one-half of all subsidy beneficiaries and one-fourth of all school lunch beneficiaries are poor.

c. Percentage of the estimated population under the poverty line in the countries with data on poverty; the real value of the poverty line may vary among countries.

Source: Musgrove (1991).

Figure 1. Number of Beneficiaries by Type of Program (logarithmic scale)

Source: Musgrove (1991).

Figure 2. Number of Malnourished Children and Beneficiaries under 5 Years in Seventeen Latin American Countries (logarithmic scale)

Note: Data for Argentina and Panama were not available.
Source: Musgrove (1991).

Figure 3. Planned Daily Supply of Calories and Protein per Individual Beneficiary for All Programs

Source: Musgrove (1991).
Data on the daily supply of calories and protein per individual beneficiary show planned rations ranging from 75 to 1,936 calories and from 2.5 to 50.8 grams of protein, with a mean of 616 calories and 20.6 grams of protein. Because many programs operate for less than a full year, annualized figures are lower (494 calories and 16.3 grams of protein). Displaying this information for all kinds of programs together (figure 3) demonstrates how much programs vary in the amount of food provided and in the protein-calorie balance, which shows considerable dispersion around a trend of approximately 30 calories per gram of protein. Actual supplies (not shown in figure 3) tend to fall below those planned-generally, the greater the planned supply, the larger the shortfall. Unexpectedly, the actual supply shows less dispersion in the ratio of calories to protein; calories supplied fall short of planned amounts more often than protein supplied does, despite the higher cost of protein-rich foods.

Programs to combat malnutrition should be judged by the cost of obtaining some nutrition-related result or some other outcome, such as improved health or learning. In the absence of information on outcomes, data on costs are usually related to the number of beneficiaries and the volume or nutritional value of food distributed. Estimates of program costs per beneficiary are subject to errors in either the numerator or the denominator; beyond that, they are poor proxies for program efficiency because high spending per beneficiary may be associated with low cost per case of malnutrition prevented or cured.

A comparison of beneficiary numbers and costs per beneficiary shows no apparent relation, in particular no evidence of economies of scale across programs. At very small size, high unit costs might be expected because of some minimum administrative expenditure, but the evidence is that all the programs studied are large enough for their costs to be unaffected by size or else that errors in the data hide any relation. Estimates of food cost and total cost per beneficiary show clearly that food costs are an almost constant share of total costs and that the variation that does occur is independent of cost per beneficiary (figure 4; for the many programs that provided only one of these estimates, the observations are presented along the corresponding axis of the figure). If the cost per calorie transferred were constant, then the caloric content of the ration and the food cost per beneficiary would be proportional. A comparison of the data shows only a rough proportionality (figure 5). More food per beneficiary means more cost, but the cost per calorie is far from uniform, in part because of variation in the composition of the ration.

To assess the overall adequacy of program effort by country, table 5 pulls together data on country population and gross national product (GNP), estimates of the prevalence of malnutrition, and cost data and beneficiary numbers for the eighty-three programs for which cost estimates are available. More than one-fifth of a total population of 414 million in the nineteen countries appear as beneficiaries, so if programs were well targeted and transferred enough food and other services to make a difference, the coverage attained should be sufficient to reduce malnutrition to very low levels. Of course, the share of the population benefiting is far from uniform: it reaches more than one-third in Brazil but falls below one-tenth in the Dominican Republic, Ecuador, El Salvador, Haiti, Panama, and Paraguay.

The total cost of these programs exceeds $1.6 billion, which is just under $20 per beneficiary, or about $4 per capita. 1 Costs per beneficiary of $10 or less are found only in Guatemala, Mexico, and Venezuela. Relatively high expenditures occur in Chile, Costa Rica, Jamaica, and Peru. On average, expenditures on these programs represent one-fifth of 1 percent of GNP, which is not a trivial sum but is still far from being an unbearable burden. If these expenditures were concentrated on the nutritionally neediest one-tenth of the population, that would mean a transfer on the order of $40 per beneficiary per year.

Comparing the number of "potential" beneficiaries to the number of malnourished children provides some notion of how well the programs in a country are responding to its nutrition problems (last two columns of table 5). To estimate the number of potential beneficiaries that could be covered for the same total cost, total program expenditure is divided by a constant cost of $35 per beneficiary per year. This $35 unit cost is equal to that of a large, relatively successful maternal and child health program in Chile, the Programa Nacional de Alimentacion Complementaria, or National Supplementary Food Program (Castaneda 1985; Torche 1985). It is also comparable to unit expenditure in several of the countries studied, including some of the poorer countries such as Ecuador and Honduras, which suggests that it is affordable. At that unit cost the $1.6 billion spent on these programs could adequately cover 47 million beneficiaries, or about 56 percent of present total coverage and 12 percent of total population.

Figure 4. Food Cost and Total Cost per Beneficiary per Year (logarithmic scale)

Source: Musgrove (1991).

Figure 5. Number of Calories and Food Cost per Beneficiary for All Programs

Source: Musgrove (1991).
Forty-seven million beneficiaries is only a little more than the current coverage, but it is nearly five times the number of malnourished children in eighteen of these countries (data for Argentina are unavailable, see last column of table 5). Of course, programs should not be limited to children who are already malnourished. For every malnourished child, several others are at risk of being malnourished and should be covered if a program is to have any preventive effect; mothers of children who are malnourished or at risk should also be covered to protect their own health and to ensure greater benefit to their children. If the resources now being spent were concentrated on currently malnourished children and on about four other needy beneficiaries for each such child, it seems plausible that malnutrition could be largely eliminated from the region.

In fact, malnutrition has not been eliminated: some 10 million children in the countries studied are malnourished. One reason is that the relation of potential beneficiaries to actual needs varies enormously among countries. At one extreme is Chile, where malnutrition has been reduced so far that current spending could cover seventy times as many beneficiaries as there are malnourished children in the country. At the other extreme are countries where current spending simply would not be enough. This is the case in Bolivia and Guatemala, which could only provide for some three-fourths of children in immediate need. Colombia, the Dominican Republic, and Mexico spend enough to reach all potential beneficiaries who are currently malnourished but almost surely not enough to cover all the population at risk.

Ineffective use of resources within countries is the other reason why the substantial level of expenditure reported here has not succeeded in eliminating malnutrition. This is a consequence of diluting resources over too many beneficiaries, spending resources on food transfers with little or no health or educational component, directing resources to adults and families who may be poor but who are often not at risk of malnutrition, and wasting some part even of resources that are well targeted. Brazil provides an example: its expenditure of nearly $1 billion could, at $35 per beneficiary, cover 26 million people, or 5.2 times the number estimated to be malnourished. Yet the prevalence of malnutrition in the country is estimated at 30.7 percent, almost as high as in much poorer countries such as Guatemala and Haiti.

Program Effectiveness

The information on programs does not describe their outcomes, so it does not provide answers about what works and what does not. Such answers can come only from evaluation, and much is already known from a variety of programs in various countries (Anderson and others 1981; Beaton and Ghassemi 1982; Feachem 1983; Mora, King, and Teller 1990; Musgrove 1989) and from the specific experience of the World Bank in promoting and financing nutrition projects or nutrition components of projects (Berg 1987).

To prevent or cure protein-calorie malnutrition, programs need to satisfy a short list of crucial requirements:

· Programs need to target the population according to the risks of malnutrition, not necessarily waiting for nutritional damage to occur, and to be more concerned with errors of exclusion than of inclusion.

· Programs must be stable through time, because the status and needs of beneficiaries vary over time, and a program that cannot deliver food or services when needed is of no use. Programs should off set-not exacerbate-the instability in the lives of their beneficiaries.

· Programs should be able to detect problems or possible sources of problems-growth faltering, illness, pregnancy, failure in school-quickly, and then respond to them rapidly.

· Programs need to respond with the right mixture of interventions, from food distribution to health care and education, to head off malnutrition or correct it once it begins, with relatively small transfers of food and with no lasting damage to the child.

Table 5. Program Cost and Actual and Potential Coverage by Country

Country

Total population (millions)

Population covered

Program cost

GNPa (millions of dollars)

GNP spent on programs (percent)

Potential beneficiariesb'

Millions

Percent

Total (millions of dollars)

Per beneficiary (dollars)

Per capita (dollars)

Millions

Ratio to number of malnourished children

Argentinac

31.9

0.3

0.8

8.9

35.02

0.28

53,070

0.02

0.26

-

Bolivia

7.1

1.0

14.7

22.0

21.07

3.09

4,520

0.49

0.63

0.8

Brazil

147.3

53.7

36.5

907.5

16.90

6.16

319,150

0.28

25.93

5.2

Chile

13.0

1.8

14.0

86.9

47.76

6.69

25,250

0.34

2.48

71.0

Colombia

32.3

3.2

9.9

44.4

13.87

1.37

39,410

0.11

1.27

1.85

Costa Rica

2.7

0.4

15.0

18.0

43.93

6.67

5,220

0.35

0.51

21.3-47.7

Dominican Republic

7.0

0.2

2.5

4.9

28.82

0.70

6,650

0.07

0.14

1.1-2.4

Ecuador

10.3

1.0

9.6

32.6

32.92

3.16

10,380

0.31

0.93

3.5

El Salvador

5.1

0.4

7.4

11.9

31.29

2.33

5,860

0.20

0.34

2.8

Guatemala

8.9

1.5

16.9

11.4

7.54

1.28

8,150

0.14

0.33

0.7

Haitic

6.4

0.0

0.4

0.7

27.69

0.11

2,370

0.03

0.02

0.1

Honduras

5.0

0.7

14.8

22.7

30.72

4.55

4,320

0.53

0.65

3.9

Jamaica

2.4

0.4

16.0

18.8

49.55

7.66

3,880

0.49

0.54

12.3

Mexico

84.6

9.9

11.7

102.6

10.33

1.21

200,730

0.05

2.93

1.8

Panamac

2.4

0.1

2.5

0.1

0.86

0.02

4,550

0.00

0.00

0.0

Paraguay

4.2

0.1

2.9

2.5

20.50

0.59

4,130

0.06

0.07

4.0

Peru

21.2

4.8

22.5

274.0

57.45

12.93

28,610

0.96

7.83

18.3

Uruguay

3.1

0.8

25.0

22.4

28.71

7.22

7,170

0.31

0.64

31.2

Venezuela

19.2

3.1

18.8

38.1

12.20

1.98

48,830

0.09

1.09

3.9

Total

414.1

83.4

20.9

1,630.4

19.62

3.9

782,250

0.21

46.59

4.5d

- Not available

Note: Program cost and beneficiary data cover eighty-three programs, including nearly all the largest ones, for which costs estimates, either in total or for food alone, are available.

a. Data from World Bank (1991), referring to 1989
b. Assuming a cost of $35 per beneficiary
c. Data from only one program
d. Excluding Argentina
Source: Musgrove (1991)

Programs that meet these requirements still do not guarantee results. To know whether a program works, there is no substitute for measuring its impact on the intended beneficiaries (Sahn, Lockwood, and Scrimshaw 1984). As noted, most of the programs reviewed here have never been evaluated. One reason for this neglect is that evaluations take money and time-to anticipate and avoid methodological problems, to take account of possible long-term effects, and to track individual beneficiaries rather than just looking at changes in totals or averages.

Evaluation is resisted for two other reasons as well, one humanitarian, the other bureaucratic. If a program distributes food to people who are mostly poor, often hungry, and occasionally malnourished, it seems obvious that it must do some good-and churlish or unethical to question its benefits. But while the direction of the program's effect may be correct, its magnitude may be inconsequential. Or the program may be ineffective because it supplies only one of the needed interventions, not recognizing the potential value of micronutrient supplements, education, or health efforts. The bureaucratic obstacle is one of attitude: in many implementing agencies, operating a program is considered a task to be performed rather than a result to be accomplished. These agencies are seldom required to show results and may be satisfied only to push food through a clogged, bent, and leaky pipeline.

But more surprising and distressing than the paucity of evaluation is the finding of several evaluations that programs have had little or no effect on malnutrition. Three factors deserve particular attention. One is the failure to measure any preventive benefits, even though these may figure among the program's objectives. Another is that rations are often too small to make much difference, unless they are exclusively provided to the needy beneficiary and accompanied by other appropriate interventions. Finally, failure may easily result from the simple irregularity of operation that plagues many programs. Overall, however, the main problem is not scarcity of resources; enough is being spent to reduce malnutrition substantially.

Notes

Philip Musgrove is on the staff of the Latin America and the Caribbean Technical Department of the World Bank.

1. Dollars ($) are U.S. dollars throughout.

References

The word "processed" describes informally reproduced works that may not be commonly available through libraries.

Anderson, M. A., Austin, J.E., Wray, J.D., & Zeitlin, M.F. (1981) Supplementary Feeding. In: J.E. Austin and M.F. Zeitlin (eds.) Nutritional Intervention in Developing Countries. Published for the Harvard Institute for International Development. Cambridge, Mass, Oelgeshlager, Gunn and Hain.

Beaton, G.H. & Ghassemi, H. (1982). Supplementary Feeding Programs for Young Children in Developing Countries. American Journal of Clinical Nutrition, 23, 707-15.

Behrman, J.R. (1991). Nutrition, Health, and Development. In: George Psacharopoulos (ed.). Essays on Poverty, Equity, and Growth. Published for the World Bank. London: Pergamon Press.

Berg, A. (1987). Malnutrition: What Can Be Done? Lessons from World Bank Experience. Baltimore, Md.: Johns Hopkins University Press.

Carlson, B.A. & Wardlaw, T.M. (1990). A Global, Regional, and Country Assessment of Child Malnutrition. UNICEF Staff Working Paper 7, New York.

Castaneda, T. (1985). Determinants del descenso de la mortalidad infantil en Chile: 1975-1982. Cuadernos de Economia, 22(66), 195-214.

Feachem, R.G. (1983). Interventions for the Control of Diarrheal Diseases among Young Children: Supplementary Feeding Programmes. Bulletin of the World Health Organization, 61(6), 967-79.

Frisancho, A.R. (1979). Human Adaptation: A Functional Interpretation. St. Louis, Mo.: C.V. Mosby.

Frisancho, A.R. (1990). Anthropometric Standards for the Assessment of Growth and Nutritional Status. Ann Arbor: University of Michigan Press.

Gray, C.W. (1982). Food Consumption Parameters for Brazil and Their Application to Food Policy. Research Report 32. Washington, D.C.: International Food Policy Research Institute.

INAN (Instituto Nacional de Alimentacao e Nutricao). (1990) Pesquisa Nacional sobre Saude e Nutricao: Resultados Preliminares. Brasilia.

Institute for Research Development/Westinghouse. (1991). Demographic and Health Survey for Paraguay. Columbia, Md.

Jamaica, Ministry of Health. (1985) Title unknown (a report on nutrition and health status of children, by Deanna Ashley and Kristin Fox). Kingston.

Martorell, R. (1991). Childhood Nutrition and Functional Competence in Guatemalan Adults. Paper presented at the International Symposium on Nutrition Research and Policy Issues in Mexico, Central America, and the Caribbean. Cosponsored by the Pew Charitable Trusts and the Pew Nutrition Program. January 7-9, San Jose, Costa Rica. Processed.

Mora, J.O., King, J.M. & Teller, C.H. (1990). The Effectiveness of Maternal and Child Health (MCH) Supplementary Feeding Programs: An Analysis of Performance in the 1980s and Potential Role in the 1990s. Washington, D.C.: Agency for International Development, Bureau for Peace and Voluntary Assistance, Office of Program, Policy and Management.

Musgrove, P. (1985) Household Food Consumption in the Dominican Republic: Effects of Income, Price, and Family Size. Economic Development and Cultural Change, 34 (October), 83-101.

Musgrove, P. (1989). Fighting Malnutrition. An Evaluation of Brazilian Food and Nutrition Programs. World Bank Discussion Paper 60. Washington, D.C.

Musgrove, P. (1991). Feeding Latin America's Children: An Analytical Study of Food Programs. World Bank, Latin America and the Caribbean Technical Department, Regional Studies Program Report 11. Washington, D.C. Processed.

PAHO (Pan American Health Organization). (1990) Health Conditions in the Americas, 1990. Washington, D.C.

Pollitt, E. (1990). Malnutrition and Infection in the Classroom. Paris: UNESCO.

Sahn, D.E., Lockwood, R., & Scrimshaw, N.S. (eds.) (1984). Methods for the Evaluation of the impact of Food and Nutrition Programmes. Food and Nutrition Bulletin, Supplement 8. Geneva: United Nations University.

Schiff, M. & Valdes, A. (1990). Nutrition: Alternative Definitions and Policy Implications. Economic Development and Cultural Change, 38(January), 281-92.

Selowsky, M. (1981). Nutrition, Health, and Education: The Economic Significance of Complementarities at Early Age. Journal of Development Economics, 9, 331-46.

Senauer, B. (1990). Household Behavior and Nutrition in Developing Countries. Food Policy, 15 (October), 408-17.

Shrimpton, R. (1984). The Economy of Childhood Malnutrition: Analysis of the Evidence for Relationship between Socio-Economic Variables and Nutritional Status, with Special Emphasis on Latin America, and in Particular Brazil. Brasilia: UNICEF.

Statistical Institute of Jamaica. (1991). A Report on the Survey of Living Conditions in Jamaica. November 1989. Kingston.

Tomkins, Andres, & Watson, F. (1989). Malnutrition and Infection: A Review. ACC/SCN State-of-the-Art Series, Nutrition Policy Discussion Paper 5. Geneva: United Nations Administrative Committee on Coordination, Subcommittee on Nutrition.

Torche, A. (1985). Una evaluacion economica del Programa Nacional de Alimentacion Complementaria (PNAC). Cuadernos de Economia, 22 (66), 175-93.

United Nations. (1989). Update on the Nutrition Situation: Recent Trends in Nutrition in 33 Countries. A report compiled from information available to ACC/SCN. Geneva: United Nations Administrative Committee on Coordination, Subcommittee on Nutrition.

WHO (World Health Organization). (1989). Global Nutritional Status: Anthropometric Indicators. Geneva.

World Bank. (1979). Brazil: Human Resources Special Report. Washington, D.C.

Zerfas, Alfred. (1991). Choice of Nutritional Status Indicators for Young Children in Public Health Programs. World Bank, Latin America and the Caribbean Technical Department, Regional Studies Program Report 8. Washington, D.C. Processed.

Comments on the Three Previous Papers by George Beaton, Juan Rivera, and Philip Musgrove


Reynaldo Martorell
Eileen Kennedy

Reynaldo Martorell

Woodruff Professor of International Nutrition Center for International Health, Emory University

I agree with Professor Beaton that the conclusions of Beaton and Ghassemi (1982) have been misinterpreted in some quarters. There are many reasons why larger effects of food distribution programs on growth were not found. For example, too little food was given in some programmes, and often, too little was actually consumed by target children. Dr. Juan Rivera referred to weak research designs, and in this session we have heard that many studies included older children in whom an effect on growth would not be expected. Professor Beaton summarized the literature available in the late 1970s (Beaton and Ghassemi, 1982). Since then a number of carefully designed research studies have been carried out - including the INCAP and the Bogota studies summarized by Dr. Rivera - which show a greater impact of supplementation on physical growth. Also important is the finding that there is a protective effect of food supplementation against the negative effects of diarrhoea on growth. Moreover, these recent studies have taken place in populations that were not severely deficient in terms of food, and thus an even greater impact would be expected if severely stressed populations were studied. One should no longer feel doubtful about the potential impact of supplementation programmes on growth.

I agree with Professor Beaton that growth failure is a phenomenon of the first few years of life. However, I would be a bit more conservative and define the age range of growth failure as extending from 6 to 36 months of age, rather than to just 18 or 24 months. While it is true that growth retardation is severest in the very young, the process does continue for longer in some societies.

There are many reasons why growth failure occurs in early childhood. Professor Beaton showed us the growth velocity curve by age, and children in this age group (6-36 months) grow very quickly, having very high energy and nutrient requirements. Children at this age and younger are highly susceptible to gastrointestinal and respiratory infections - in fact the incidence of diarrhoeal disease is greatest at 6-24 months of age in most settings. In addition, young children are less able to express their needs and are vulnerable to poor care, whilst older children are more likely to be able to fend for themselves. In the INCAP study that Dr. Rivera referred to, we find two interesting things. First, growth rates after about three years of age were similar in this Guatemalan population to values found in well nourished populations. Secondly, supplement had no effect on growth rates after three years of age; thus, the effect of the supplement was concentrated in what Professor Beaton described as "the window of opportunity".

As many at this symposium have pointed out, decisions as to which population groups should be targeted for supplementation depend on the intended results. If the aim is to provide an income transfer then surely the target population must be poor families. If we want to improve school attendance and such aspects as attention span, then it is the school child in poor areas. If we want to improve economic activity, say of sugar cane workers, then it is the working adult population we must target. However, if our aim is nutritional then I would argue that children in the first three years of life must be at the top of our priority list. Protein energy malnutrition and wasting are most common in this age group, as are diarrhoeal diseases, and even though certain micronutrient deficiencies such as vitamin A more commonly manifest themselves in older children, the biggest drain on tissue reserves occurs in early childhood. We know very little about physical activity, but if this is most constrained in early childhood along with physical growth, improving energy intakes would have tremendous implications for psychological development. I remain convinced that the massive growth failure that occurs in early childhood and at no other period in life is a marker, as Professor Beaton said, of functional impairment in a number of other domains. I also believe that in preventing this massive growth failure through diet and health interventions, we will prevent most of these associated effects.

Professor Beaton referred to the INCAP follow-up study (1987-88), and I would like to say a few words about it. This study was carried out in collaboration with INCAP, specifically with Dr. Juan Rivera, and involved the follow-up study of adolescents and adults who had participated as young children in the INCAP Longitudinal Study (1969-77). Two approaches have been used to look at the relationship between early nutritional status and later functioning. First, analyses were carried out which used the nutritional intervention design described by Dr. Rivera (i.e., Atole vs. Fresco supplements). And secondly, the total nutritional experience, defined in terms of the degree of stunting in early life were related to later outcomes. Both approaches pursue the same objective -- to document that improved nutrition in early childhood results in better functioning in the adult. In one analysis, we divided children who were three years of age into three categories of length-for-age; severely stunted, moderately stunted, and not stunted. We defined as severely stunted those who were three standard deviations or less below the mean. Moderately stunted children were those who were -2.9 to -2.0 sds and those not stunted were above -2sds. Figure 1 (Martorell, 1993) shows the percentage of children below three standard deviations by calendar year in Atole (high energy, high protein supplement) communities and in Fresco (low energy, no protein supplement) communities. In 1969 the prevalence of severely stunted children was about 45% -an incredible degree of growth retardation. In 1976-77, when the study ended, the prevalence was more or less the same in Fresco children, but had declined to about 20% in Atole children. Therefore, it appears that the nutrition intervention was effective in decreasing the prevalence of severe stunting. Work soon to be published in a special issue of The Journal of Nutrition shows that improvements are also seen later in life. Adolescents and young adults who were exposed to the Atole during the first three years of life were taller and had greater fat-free masses, particularly women, than those who received Fresco. Work capacity was significantly improved but only in males. Interestingly, intellectual performance was more affected in these subjects in adolescence and adulthood than during the preschool years. Thus, improving the diets of young children in Guatemala had an immediate payoff in terms of reduced prevalences of stunting and also resulted in enhanced physical and intellectual status in adulthood.

The follow-up study has also permitted us to examine whether catch-up growth occurs subsequent to the phase of growth failure in early childhood. We find that growth from three years of age to follow-up is the same regardless of the degree of stunting at three years of age; in other words, there is no catch-up growth. A second finding is that growth from three years of age to adulthood, say to 18 years of age, is just a few centimeters less than that seen in the United States, and in fact, is a little more than that seen in Mexican Americans growing up in the US (Martorell et al., 1992; Martorell et al., 1990). This clearly reinforces the characterization of the first few years of life as the "age of growth failure".

Figure 1 Changes over time In percent of 3-year-old children with severe growth failure: sexes combined +

+ Length values 3 S.D. or more below the reference median.
(Reproduced with permission from Martorell, in press)
I would like to discuss briefly some of the consequences of stunting using our follow-up study results. There are several cut-off points used in obstetric work in developing countries as indicators of obstetric risk, and one of them is stature less than 149 cm. In women 18 years or older, those who were stunted at three years have a 65% prevalence of stature less than 149 cm, whilst those who were not stunted at three years of age have a prevalence of around 4% (Martorell et al., 1992). This effect is also seen if one looks at lean body mass which is an important determinant of work capacity. We have begun to study the relationship between stunting and intellectual achievement through analyses which first control for such things as maternal education and family wealth. These answer the question "is there a lingering relationship between length at three years of age and intellectual achievement?" Preliminary analyses indicate that intellectual achievement was constrained in subjects who were stunted as young children. Intelligence, numeracy, literacy and school attainment were all lower in children who were stunted. One interpretation of these results is that the conditions which gave rise to marked growth retardation also affected cognitive development adversely. Growth failure becomes, in this view, not a cause, but a marker of the syndrome of developmental impairment of early childhood.

To sum up, first, early childhood is when massive growth failure occurs. There is now quite a lot of evidence to show this. Secondly, nutritional supplementation in early childhood reduces the degree of growth failure and improves physical status and intellectual performance in the adult. And thirdly, stunting in early childhood has direct consequences such as short maternal stature and reduced lean body mass which are important concerns for obstetric risk and work performance respectively. Finally, stunting also has indirect functional consequences, such as poorer intelligence and limited school achievement. Therefore, the prevention of massive growth failure in early childhood should be a high priority, and I think it should be at the top of our list when we consider nutritional benefits as the outcome we desire from supplementation programmes.

References

Beaton GH, Ghassemi H. Supplementary feeding programs for young children in developing countries. American Journal of Clinical Nutrition 35(Supplement):, 1982.

Martorell R. Overview of long-term nutrition intervention studies carried out in Guatemala (1968-1988). Food and Nutrition Bulletin, in press.

Martorell R, Rivera J, Kaplowitz H, Pollitt E. Long-term consequences of growth retardation during early childhood. Proceedings of the Vlth International Congress of Auxology, Madrid, Spain, 15-19 September 1991, Madrid, Spain. IN: Hernández M and Argente J (eds.) Human Growth: Basic and Clinical Aspects. Elsevier Science Publishers B.V., 1992, pp. 143-149.

Martorell R, Rivera J, Kaplowitz H. Consequences of stunting in early childhood for adult body size in rural Guatemala. Paper presented at the symposium "Long Term Consequences of Nutrition in Infancy and Childhood", XIX International Congress of Pediatrics, Paris, July 1989. Annales Nestle, 48:85-92, 1990.

Eileen Kennedy

International Food Policy Research Institute

One of the major conclusions from Beaton and Ghassemi (1982) was that "...food distribution programmes directed towards young children, as now being operated, are rather expensive for the measured benefit..." Professor Beaton discussed the "measured benefit" issue -highlighting the fact that there is probably a whole range of unmeasured and unobserved benefits which result from supplementary feeding programmes.

[I would like to highlight the issue of "programme-as now being operated". As an illustration, the examples of programmes Professor Beaton chose to give were well controlled research studies and not typical of ongoing programmes - and this is an issue which also arises in the work of Dr Musgrove.]

I would not have been quite so negative as Dr Musgrove in his paper. The work represents a very useful starting point in finding out exactly what is going on in Latin America and the Caribbean as far as these types of programmes are concerned. It was interesting to note that half the programmes included in this 104 programme assessment begain the 1985-1990 period - after the Beaton and Ghassemi work had been done.

Dr Musgrove is clear that there is little information on effectiveness of these programmes as described by outcome measures, and that programmes "tend to be judged only be measures of input" - but I feel that, on the other hand, well run programmes, or programmes with appropriate mixes of inputs, are more likely than less well run programmes to meet their goal.

One comment in particular in the report intrigued me: "There is no great mystery as to how to design and conduct a programme so that it actually reduces malnutrition. What is needed is to apply these lessons universally." Let me discuss further what I think some of these lessons are. Firstly, that beneficiaries in these programmes need enough food at the right times. But what is enough food? You mention a low in Mexico of 140 calories and a high of slightly over 1000 in Costa Rica. But what was unclear to me outside the context of cost - which is clearly connected with calories - is exactly how one decides where to set the level of supplementation to be given. This is especially important when the second lesson is taken in to account - that one needs an integrated package of resources.

Dr Musgrove talks about poverty, illness and ignorance being the three underlying causes of malnutrition. To alleviate poverty, food is given; for illness, health care is provided; and to combat ignorance, nutrition education is given. Dr Musgrove then argues that in most of these programmes too much has been spent on food relative to health care and nutrition education. How do you balance the appropriate mix of inputs? In addition, in his report, Dr Musgrove has indicated that nutrition education in the region does not seem to have been particularly effective. However, it is not that nutrition education services cannot be effective, but that the way they have generally been operated, they have not been.

Before leaving the issue of level of supplementation, it is useful to consider supplementation not only in terms of absolute level of calories, but also to ask about recommendations on levels of micronutrients. Additionally, the CRSP data has indicated that there appear to be some foods, at least in children's diets, that above and beyond the level of caloric intake have a beneficial effect on growth. We have re-analyzed some of the IFPRI data from Southwestern Kenya, and have similarly found that, holding total calories in the child's diet constant, there is a beneficial effect of increasing the percentage of animal protein in the diet. Thus, in considering supplementation packages, we should not only be thinking about absolute levels, but also about quality issues, including micronutrients.

From the knowledge gained from the programme implementation assessment reported here, I would like to ask Dr Musgrove what he would do differently. It is not clear to me how a programme planner would use the information here to plan a more effective form of intervention.

This question comes at a particularly critical time because of current reassessment of some issues related to the US supplementation schemes - the largest one of which is the WIC Programme. This is basically a combination of food, health care, and nutrition education. Food takes the major part of the money going into the programme, and people are now asking what it is about this combination of inputs that is supposedly having an impact.

From a look at a compilation of studies, at least on the prenatal component, it is fairly clear that participation in the WIC programme is associated with improved neonatal outcome. However, when you begin to unpack this black box of food, health care and nutrition education, you find that of the 900 calories a day that are delivered, only about 110 are incremental in the prenatal diet. Can we, then, give less food without sacrificing the efficiency and effectiveness of the programme?

I would like to push Dr Musgrove to think about more than just the broad issue of supplementation, or the other lessons that come out of these 104 programmes that give us a better understanding of where to go when planning the future direction of these very substantial programmes.

I would like to mention briefly three other issues. Firstly, concerning the illness-food intake debate which came out in the papers by Dr Beaton and Dr Rivera. Clearly, diarrhoea has a major inhibitory effect on young child growth. One of the conclusions of Dr Rivera was that "supplementary feeding is likely to have more of an effect on children with a high incidence of diarrhoea, possibly through increase in appetite after diarrhoea." I would like to mention that, after repeated episodes of diarrhoea, anorexia may, in fact, be more of a problem than increasing appetite. Some work has been carried out recently by the Dunn Lab group in the Gambia, where they have begun to look at the pathophysiology of the different types of diarrhoea, and whether catch-up growth can occur or not. One of their conclusions has been that in the absence of intestinal injury, catch-up growth in preschool-age children can be rapid. However, if there is damage to the small intestinal mucosa, rapid catch-up growth or any growth may not take place.

Secondly, for a long time, equity issues were used as the basis for justifying many supplementation schemes, and it is right that children should be allowed to reach their growth potential. I think equally important now is the emerging literature that supplementation schemes may not only facilitate equity, but also efficiency. The labour nutrition literature, some of which is coming out of the economic literature, is concluding that better nourished individuals are more productive as adults. Recent work by IFPRI, from Kenya, where the same women have been tracked since 1984, has found that women with better nutritional status, in this case as measured by a higher mean BMI, were more active in the later (1986-87) period, both in terms of home production and farming. Professor Beaton mentioned women in one of his closing arguments as important in the overall survival of the household. I think it is very critical to now look at the links between women's nutritional status - not just mothers - and their roles in health and nutritional activities, and how improving women's nutrition affects or does not affect this responsibility.

Finally, I would like to talk about indicators. We probably have limited ourselves to too narrow a range of indicators for evaluating the full range of impacts of supplementary feeding (and other) programmes. But, given that many of the outcomes are very difficult to measure, what do we do if we are trying to develop a capability at local level for monitoring and evaluating supplementation schemes? Can we try to find a range of indicators which are "friendly" - and such that there is no need for an evaluation of a specific programme every ten years, but evaluations that can be done locally in a participatory way. Is this, in fact, totally in contrast to looking at the full range of impacts that may be taking place?

Response


Philip Musgrove

Philip Musgrove

I would like to start by replying to a comment from Mr Jonsson; even though it was not directed primarily at what I was saying, it leads into a couple of other comments.

He said he was surprised that anybody would expect there to be much connection between nutritional status and food distribution programmes and I entirely agree; there are times when it feels like anything you see is probably random and if you just closed your eyes it would go away. It is worth thinking a little more about this. One reason for it, of course, is the difference between giving away food, and feeding or making sure that somebody eats it. There are problems of intra-houshold distribution, problems of quality, problems of sanitation.

There is another issue related to the whole series of questions Eileen Kennedy was putting: how do we measure growth response? When we measure a child, we are measuring accumulated failure to grow; the integral of a long process, affected by all sorts of things all along the way. This starts with genetic factors and ends up with what the child ate yesterday. Yet most measurements (unless we do very careful longitudinal studies) are just moments in that process. There really is not any good reason, for example, to discover that food intake last week and nutritional status of children as measured last week in a cross-section survey should have very much to do with each other, unless we know that people who are poor are uniformly poor all their lives and their children are born poor and nothing ever changes. Introduce a little bit of fluctuation in the income stream and it is enough to pull in so much noise that any real relation is going to be very hard to find. That is why it is encouraging when people do really good longitudinal studies. It also means that there probably ought to be an outright ban (I don't know - something where if you break it you can't publish anywhere for five years) on studies that simply report associations on a cross-sectional basis.

Mr Jonsson also remarked that, in his view, any kind of programme that we are talking about ought to be judged on its nutritional outcome and should not be allowed to get away with claiming spin-offs and making comparisons to other projects. I think that's pushing a little too far. If we decide that the objectives are nutritional we ought to demand that these are met - that is true. But there are two problems. One is that there are other legitimate objectives, depending on what the programme is and who you are trying to reach with it: for example, in the case of school feeding. The other problem is that the people who defend these programmes politically do not always do so on nutritional grounds -they are just as likely to defend them on the basis of income distribution. In fact that's often the last resort. If you take a look at a long list of objectives that somebody would state for his programme and it was possible (with data) to go knocking them off one by one - say, well you say that the kids will grow faster but here's the data and they don't; and you say that they are going to stay in school longer, here's the data and they don't - when you get them all the way back in a corner, he will throw up his hands and say "but we re-distributed income!" and he's got you there. This is a real serious problem - and it's the scoundrel's last resort. How do we value this? What is worth to us - anything? If nothing then we should not allow this, but if it is worth something, we need a way of judging it compared to the other things that we are trying to accomplish. One of the advantages of claiming a long list of objectives is that you always have this escape. I think we have a problem of not having any idea how to value one objective versus another; we have something much worse than apples and oranges here. We have no notion in general, of whether it is worth more for a one year old to be the right size at age three or whether it is worth more for a seven year old to stay in school for another year. If somebody says "I'll give you what is costs to produce either of these results, but you only have on eof them" I think we have no idea what to do next. This takes me back to the difficult questions that Eileen Kennedy was asking. She is really confronting us with the fact that we have a multi-input production function - it takes in food, it takes in health care, it takes in child care and stimulation, it takes in knowledge; and ideally, produces a multi-output function, in which we have a child who is not only normal size, but also has been properly reared, and is going to school at the right age and is going to stay there a while.

The problem is this: turn it around, start at age ten and look back. At age ten, I want a healthy, normal size child that has already been in school for three or four years. Now, what is the best distribution of resources between that child's conception and its tenth birthday, to get there. This is what I think we do not know. All requests for blueprints come down to trying to solve what is just mathematically a fiendish problem. I have tried it - I have tried to write the equations and I do not have a clue how to get started; if anybody does, I hope they publish it soon.

I think that what you can get out of this kind of study that I have done is to say we know something about what is wrong with the starting point - that is, in most programmes (but not all) too much emphasis on food, partly because sometimes the food is, or appears to be free. There is too little emphasis on the other inputs, and this tells you the direction to push in - move away from the present balance towards more health care and more education. But it does not tell us where to stop. It does not fix the optimal mixture of inputs, which is why we do not really have a solution or a blueprint. It only says the present mixture on average is unbalanced. This in turn carries another message - try to figure out what the problem is before specifying the solution.

I tried to classify the problem as a mixture of poverty, illness, and ignorance.

If I can beg your indulgence for a moment, this leads me to want to tell you my favourite science and religion joke because it illustrates perfectly this problem of what do you tell people and how do you make it stick so it does some good? I think the story is the paradigm of this problem. It involves some missionaries who started converting people in some remote place, and converted them so successfully that the recent converts do not want to leave the mission - they want to hang around and sing hymns all day. And so they decide that they need the help of some of these people in feeding the others, and so they pick a particularly bright, very thorough new convert and ask him if he would like to be a cook and help in the kitchen, and he ways he would. So the missionaries take him through the kitchen and insist repeatedly and violently to him on the urgency or always washing his hands and always boiling the water. Of course, he looks at them like they are crazy, and they say "Jacob, you can't see them and I can't see them, but there in that water are things that can make us sick and can kill us," and he turns round and says "I thought we Christians didn't believe in that stuff anymore." Well, that's the problem. How do we get across a message that would be believed and acted on and not either insult or drive people off, or give them what they do not need, or waste money - a lot of which has been spent with not enough results.

Summing-Up


John Mason

John Mason

UN ACC Sub-Committee on Nutrition

One of the major conclusions from the first paper by George Beaton is that we need to accept that there are multiple benefits from supplementary feeding. Growth is a marker of progress, but not the only important outcome. The overall benefits concern immunity and tissue integrity, physical activity and its contribution to psychological development, and growth. This has been discussed further in the SCN's paper on "Appropriate Uses of Anthropometric Indices in Children" (ACC/SCN Nutrition Policy Discussion Paper No. 7, December 1990).

The second conclusion is that the top priority, the age-group that has to be reached, is the six months to maybe 18 months, maybe 24 months, maybe 36 months old children. Two good reasons were brought out. One is for growth itself, but growth is mainly a marker. The most compelling argument is from the long-term effects which the studies in Guatemala have established: you have to reach the six to 18 month old group of children to protect their long-term development, including educational achievement. Beyond this the question of age targeting is a bit fuzzier.

Where there are very limited possibilities of targeting it may be one targets the poorest communities or households, irrespective of age. However, it seems unsatisfactory to only say it depends upon the objectives, because that does not provide guidance as to whether, for example, limited resources should be used for school age children or preschool children. Even if we cannot give a formula, we should try to lay out some of the considerations. How do you compare food as a magnet for bringing children to school and helping their educability with food for, say, the four to five year old children, which will still to some degree help their growth? Moreover, this focus on age-targeting does not give sufficient attention to supplementary feeding for women (in this context particularly for pregnant women). The issue has not been fully resolved as to the relative benefit of feeding the pregnant woman, hence increasing birth weight and contributing to survival and development of the child.

The issue also arose that we have considerably more information now concerning benefits of supplementary feeding than when the original paper was published by Beaton and Ghassemi (1982). It is now quite clear that if you can deliver food to children who need it they will benefit in a series of ways, including growth (as a marker), psychological development, preventing morbidity, preventing mortality. As usual, it comes down to getting enough of the right food to the right people at the right time. The SCN might consider making this very explicit in a statement1.

1 See "ACC/SCN Statement on the Benefits of Preventing Growth Failure in Early Childhood" at the beginning of this section (p.36).
Another look at the benefits of supplementary feeding came from the very specific study reported by Juan Rivera. This concerned the interaction with diarrhoea. One can summarize the findings in various ways. One is to say that if the underlying calorie consumption is low, then a dietary energy supplement can be shown both to counteract the effect of diarrhoea and also to have an effect on growth when the children are well. When, on the other hand, the underlying calorie availability is somewhat higher, there is still a benefit in terms of preventing the effects of diarrhoea on growth, but no effect on well children, who are growing adequately anyway. So there is an unequivocal yes to the question: "Is supplementation worthwhile without infectious disease control?" Yes, it does have an effect. This does not mean this is necessarily the desirable approach, and Dr Rivera brought out that one needs the conventional but important mixture of interventions, including breastfeeding promotion and primary health care, but that there is a clear role for food supplementation.

I think that a number of us were quite surprised at the extent of the programmes in Latin America, as reviewed by Dr Musgrove. These have a high coverage and a fairly considerable expenditure per head. One conclusion was that there is scope for more targeting, both to ensure that the needy groups are fully covered, and to focus expenditure on these. It could be hypothesized that much of the coverage is not very effective because it is not intense enough. A further point concerns the need perhaps for increasing the non-food expenditure. This relates to another point of food programmes versus feeding programmes: simply delivering food may not have all the intended effects. These require the non-food components i.e. education, health care, and so on.

Finally, to re-emphasize an earlier point, in giving guidance on when and where supplementary feeding is appropriate, we should not constantly throw the ball back, and say that it depends what your objective is. Supplementary feeding and other nutrition interventions have been promoted over the years and we should not back off from that. We should be more confident in saying what the benefits are: growth is a marker, and there are many other benefits which are very important, both in the short and long-term. We should not be timid in laying these out: in sum, better health, better individual development, happier people, higher education attainment: better "human capital".

References

ACC/SCN. (1990) Appropriate Uses of Anthropometric Indices in Children. ACC/SCN State-of-the-Art Series. Nutrition Policy Discussion Peper No. 7. ACC/SCN, Geneva.

Beaton, G.H. & Ghassemi, H. (1982) Supplementary Feeding Programs for Young Children in Developing Countries. American Journal of Clinical Nutrition, 23, 707-715.

Educational Aspects, Weaning Food Supplements and Targeting in Supplementary Feeding Programmes

Ken Bailey

World Health Organization

With your permission I would like to express some points of view, based on experience of reviewing more than 200 WFP projects over a period of 28 years as a WHO Regional Officer in Africa and the Western Pacific. These aspects seem to me important and have yet been expressed at this meeting. Especially, the educational aspect and impact of WFP projects.

In the Regions and countries where I have worked, and I believe in most countries, except at times of stress such as war and drought, shortage of food at the household level is not the basic problem, at least in rural areas. In some cases there are seasonal food shortages. But otherwise the main problems in young child feeding are other factors which come to the fore mainly when the child is from 6 to 18 months old. Such problems include: how to find time to feed the young child more than two or three times a day; use of porridge which is too dilute and consequently leads to undernutrition and eventually marasmus; the fact that this porridge is often consumed without other local foods to provide necessary fat, protein or micronutrients.

Food aid in whatever type of project will usually be in the form of cereals, some beans or a cereal/legume mix like CSB, oil and perhaps some animal protein component. So much depends on the type of communication and education that goes along with these commodities; and usually according to all project reports I ever read, this aspect is painfully weak. The presence of beans may convince the mother that this is a good thing for the young child; or very likely, she may consider the infant unable to digest them, so she may sell them or use them for the older members of the family; and at the same time she is likely to stop growing them or buying them for herself, since she is getting them for free. When food aid stops, what will happen? Sometimes, worse, she may get wheat or rice, instead of the sorghum or maize she is used to, and so start off on a whole new dietary pattern, for the infant or even for the family.

Surely one of our main jobs in nutrition is to help that mother to cope better with the foods they grow or have available in the community. I am well aware that WFP makes every effort to provide the cereal staple to which the mother is accustomed; but often, especially in Africa, that commodity is not available to WFP, especially when it is normally a tuber. So the risk of changing dietary patterns, even radically, is there.

How many school feeding projects are based on provision of wheat flour for making a small loaf of bread for the children, when wheat is not grown in the country and there is little opportunity for this to be a creative educational experience, since it does not blend well with locally available foods.

So my first suggestion is that, just as every project is reviewed from the point of view of its impact on women, its educational impact, positive or negative, should be critically examined; and if possible, positive educational messages spelled out clearly. Otherwise we risk seeing continuing distortions of local food habits, loss of prestige of local foods, and other anti-educational effects.

An example of the latter is when one tries to encourage mothers in MCH centres or at village level to develop good local recipes for use in the weaning period. It has been my misfortune to work in several communities for self-help efforts along these lines, for instance with beans and vegetables grown individually or collectively by village women, only to have the activity overturned by the provision of food aid (but not by WFP), in situations where there was no basic shortage of food. Thus, the food aid became an impediment to the better use of local foods.

One way in which this difficulty can be overcome is through making suitable weaning food mixtures locally, using appropriate technology in the form of simple equipment or existing machinery such as corn mills. In a few countries, WFP is using local weaning food plants to produce mixtures of local beans and cereals, such as Weanimix in Ghana, Misala in Burundi and Likuni Pala in Malawi. WFP purchases these products locally and thus the possibility exists at least, to show the people the proportions used, and how to make these mixtures themselves using local cornmills. Thus it can at least have some positive educational value - if the effort is made to point this out to the people participating in the programme. We have with us here today a distinguished specialist in the field of this kind of technology, Mr Dijkhuizen, who has developed appropriate technology for use at various levels, e.g., in villages and at district level. A heavy input in the forming of training for local women's groups and community leaders for handling such projects is needed. UNICEF and World Bank as well as WFP and the Dutch cooperation have supported such approaches in a few countries and I think such approaches would be valid and feasible in nearly every African country at least, and probably much more widely in other Regions as well. It is one way which can lead to self-reliance in the communities and countries concerned.

So my second suggestion is that organizations responsible for food aid programmes take seriously the possibility of producing these convenience supplementary foods locally, in front of the people's eyes, using inexpensive technology, instead of temporarily providing "wonder-foods" as handouts.

This may sound like a counsel of perfection but the fact that it is working successfully and expanding in several countries gives me confidence to say that we should not accept indefinitely, with equanimity, the continued distribution of this sort of cereal-legume mixture from industrialized countries.

Thirdly, my estimate is that food aid is often used in countries and even for vulnerable group feeding, where there is no real shortage of food, or for families or individuals for whom there is no shortage. Considering the high costs of delivery to remote places, is it cost-effective? We need to be more selective in targeting food aid (at least the supplementary feeding type) to those communities and individuals where it is really needed. This includes the aspect of timing. As food shortages are seasonal in most countries, it would be ideal if food aid could be provided to needy families in rural areas exclusively during the pre-harvest season. While it has been said that the organizers of food aid are not able to provide supplementary foods seasonally, I feel it would be possible if commitment existed.

The basic point is thus that we should exercise much more imagination and sensitivity to the socio-cultural situation when designing food aid programmes, especially those of the supplementary feeding type. We should put much more effort into minimizing negative educational impacts and developing positive sustainable approaches. This point was hinted at by a couple of speakers but not, in my view, sufficiently stressed, and without some modifications in our approaches, achievements may in the long term be negative or few, rather than positive and sustainable. There are many other aspects like dependence which could be further elaborated but time does not permit. I did not find these aspects reflected at all in the WB/WFP Agenda for the 1990s on Food Aid in Africa, which otherwise I found realistic and admirable.

These remarks are not in any way to belittle the positive value of food aid projects especially of food for work, community development type as well as for emergency feeding, nor the excellent work done in planning and executing them. These three suggestions are put forward not as criticism but as a challenge to more imaginative thinking in the organization of food aid programmes.


Previous Page Top of Page Next Page