Contents - Previous - Next

This is the old United Nations University website. Visit the new site at

Child nutrition and dietary diversity within the family: A view from the Caribbean

Thomas J. Marchione
Department of Anthropology, Case Western Reserve University, Cleveland, Ohio, USA
Visiting Researcher, Institute for Nutrition Research, University of Oslo, Oslo, Norway


Describing the patterns and determinants of intrafamilial food sharing between adults and children is important to both anthropologists and nutritionists, though from different points of departure. Anthropologists have been aware that food distribution may be a mechanism of cultural adaptation in stressful environmental situations. Wolf (19) suggests that in Latin American peasant households a belt-tightening within the household unit is the general response to economic depression. Gross and Underwood (7) demonstrated that among Brazilian plantation workers economic stress due to scarce income for needed food caused children to suffer a disproportionate share of the nutritional shortfall because they were less important to the survival of the household unit than were the working adults. This adaptational view suggests that the intrafamilial distribution pattern is understandable only in terms of the group's survival within the context of a particular physical and socio-economic environment. One might go so far as to suggest that the plasticity of the growing human organism in weight and size has evolved in response to the periodic scarcity which groups have been forced to undergo throughout human history (18). In extreme cases, nutritional deprivation leading to death is obviously a population control mechanism.

Nutritionists and medical people, however, tend to resort to cultural ideation (beliefs and values) to explain why children's diets are poorer than diets of others in the household. Practices have tended to be described as "pre-scientific" or generally "irrational" from a nutritional viewpoint (cf. 11; 4, especially p. 557). The view implied is that ignorance coupled with influences on behaviour from cultural beliefs (taboos) undermines child health and nutrition when household resources are supposedly adequate. An exemplary empirical study using scalogram analysis in a Mexican community demonstrated that at six months of age infant and family dietaries not only were correlated with each other but were correlated more highly with per capita expenditures and income of the household than with any other variable in the analysis (2, p. 129). Nevertheless, the authors concluded that the economic correlations demonstrated resulted from the study's being conducted in a relatively "industrialized" part of Mexico rather than in a more traditional, rural area where more "pre-scientific" views of child diets persist (2, pp. 129-130).

This ideational emphasis has considerable appeal to policymakers because it implies that a change in knowledge or attitude through education within poor households alone will solve malnutrition problems. It does not call for the more difficult task of increasing household food or food-getting resources or redistribution between households. The effort need not be sustained. Once the knowledge is gained, the educational programme can be withdrawn. And lastly, the idea rings true to laymen in terms of their own experience and common sense (3, p. 75).

Recently nutritionists have seen that this limited view of nutrition education must be rethought to account for the intimate relationship between traditional dietary practices and the realities of food scarcity (IUNS, 1980: 25).


Survey work throughout the Caribbean has documented many practices which apparently run counter to western scientific knowledge about child feeding. Considering only materials on the post-weaning period, one finds reports that high-quality protein foods are withheld from children because they are believed to cause worms. Peas are withheld because they are believed to cause flatulence. The pattern which emerges maintains children on high-starch semi-solid paps or porridges well into the second year of life. At the same time, adult men consume a disproportionate share of household animal protein.

Whereas scientific opinion in the Caribbean recommends that children should be fully fed from the whole range of food in the family pot by one year of age, this does not happen in practice (15). In Barbados, only 15 per cent of children were eating out of the family pot at twelve months of age (16, p. 56). Similiar patterns are reported for national samples in Guyana (171 and Jamaica. Further analysis in the Barbados survey showed no association between the age at which solid or semi-solid feeding began and the percentage of standard weight for age for preschool children.


In Jamaica, community nutrition surveys have demonstrated time and again that peak periods of malnutrition are between twelve and twenty-four months (8, p. 654; 5; 1). It appeared logical to assume that this problem was associated with poor feeding habits of the child following weaning-that it was a result not merely of economic pressure but also of practices based on incorrect knowledge or ignorance.

In 1973 I conducted a community survey of households with young children in St. James, a parish of 100,000 people; 15 per cent of the children were under four years of age. The objective of the research was to determine the social epidemiology of young child malnutrition and to establish a baseline to evaluate the impact of a community nutrition education project. The survey included anthropometric and 24-hour dietary recall data on 132 randomly selected one-year-olds. In addition, 24-hour dietary recalls were gathered on the mother or guardian (i.e., the meal-maker) of the target child. (See ref.13 for a discussion of the survey methodology.)

The recall included provisions to record food items such as sugar, flour, oil, fish, rice, bread, peas, and mangoes, as many times as they were consumed during the day. Each portion was then classified into one of seven food groups -staples, peas and beans, vegetables, animal products, fruits, fats and oils, and sweets and sugars. In addition, the number of different items consumed were counted to yield a measure of dietary variety. In order to get some measure of intrafamilial food sharing, the number of items in the child's diet was determined as a percentage of the dietary variety of the meal-maker's diet.

Anthropometric measures were converted into weight for age and height for age for each child using the Boston standards (10). I am aware that this may very well underestimate the nutritional deficiency among black children (6)-so one should consider the measure in this analysis an indicator of relative nutritional status.


Social epidemiological analysis of the 1973 data revealed that the dietary variety of the meal-maker was positively correlated with the one-year-old's nutritional status as measured by percentage of standard weight for age and percentage of standard height for age. (Kendal's tau b, 0.15 and 0.20, p < 0.05). Dietary sharing between meal-maker and child was not significantly related to the child's nutritional status. However, greater dietary variety of the meal-maker was associated quite strongly (tau = 0.41) with less dietary sharing between mother and child. As one might expect, a higher income, indicated by higher food expenditures, was significantly correlated to higher dietary variety (tau = 0.20). However,diet sharing was significantly higher in poorer households. Obviously, the poorer the household, the more limited the dietary variety would be and the more likely that the child must partake of greater proportions of family food to get any food at all.

In fact, in factor analyses using orthogonal factors, the dietary measures formed a single variable. When this new variable was considered separately from economic and household social structural factors, no correlation with the anthropometric measures could be found (13; 14).


A further attempt to test the significance of the food-sharing hypothesis was provided by the evaluation of Jamaica's Community Health Aide Programme, which was instituted in 1973. The health aides were charged with the explicit task of improving the child's share of food from the family pot. When sample surveys of children taken in 1973 were compared to those taken in 1975, food-sharing patterns showed no significant change during the 24-month interim period. In fact, although the nutritional status improved significantly from 1973 to 1975 in rural St. James, the food-sharing measure remained constant. The child's dietary variety stayed at about 60 per cent of the mother's (see table 1). Similarly, the overall sharing patterns showed no changes due to the educational input of community health aides. Whether the sample mother had received high contact with the health aide, low contact, or no contact of all, the food-sharing pattern remained the same. In fact, there appeared to be greater increases in sharing where health aides were not in contact with the mother, though the difference was not statistically significant (see table 2). It became clear that the patterns of sharing in Jamaican households were very deeply rooted and appeared to be carefully calculated to maximize family well-being. They were not easily modified by merely supplying new information alone. Change in family economic status was of key significance.

TABLE 1. Dietary Sharing between Meal-Maker and One-Year-Old Child in Rural Areas, 1973 and 1975

Diet Sharing* 1973 % 1975 %
(N = 119) (N = 114)
90% or more 60.9 67.5
Less than 90% 39.1 32.5
Total 100.0 100.0

X not significant at 0.05.
*Child's dietary variety as a percentage of the meal-maker's dietary variety.

In a previous paper (12),1 reported that within the rural part of St. James the greatest nutritional improvements from 1973 to 1975 were seen in the semi-subsistence peasant farming households, and I found the nutritional changes to be largely economic in origin. Grade II and Grade IIl malnutrition had been reduced from 14 per cent in 1973 to 4 per cent in 1975 (see table 3). Focusing on this population alone, I undertook further analysis in an attempt to demonstrate some effect of changing patterns of food sharing between mother and child.

In this analysis I considered my previous measures of food sharing too crude. Consequently, I created new measures of dietary sharing by substracting the food group portion counts of the meal-maker from the child's portion counts. In other words, if the mother had 6 portions of sweets or sugars for the day and the child had 8 portions, the difference would be +2. All of these differences were added for each of the seven food groups to get an overall sharing index as well. Three of these sharing measures were found to be correlated with the child's nutritional status (weight for age). The sharing of animal products and fruit were positively correlated (r = 0.20, p < 0.05;and r = 0.19, p < 0.05). The sharing of sweets and sugars was negatively correlated with nutritional status (r = - 0.19, p < 0.05). The overall sharing index was not correlated to nutritional status.

These results reveal why gross sharing measures may be inadequate to demonstrate nutritional variation. Greater sharing is not always beneficial. Benefits are derived when sharing takes place using high-quality protein such as animal products or using quantities of energy-rich fruits. On the other hand, frequent feeding of small quantities of sugared teas appears positive on the sharing scale but has a negative effect on child growth.

TABLE 2. Impact of Community Health Aides on Dietary Sharing in Rural Areas

Diet Sharing*

Mother's Contact with Health Aides




1973 % 1975 % 1973 % 1975 % 1973 % 1975 %
(N = 61) (N = 51) (N = 35) (N = 43) (N = 23) (N = 20)
90% or more 67.2 64.7 51.4 65.1 60.9 80.0
Less than 90% 32.8 35.3 48.6 34.9 39.1 20.0
Total 100.0 100.0 100.0 100.0 100.0 100.0

No significant changes within the contact groups at the 0.05 level.
* Child's dietary variety as a percentage of the meal-maker's dietary variety.

TABLE 3. Nutritional Status of Children under Three Years of Age in Peasant Farming Households, 1973 and 1975

Child's Nutritional Status 1973 % 1975 %
(N = 112) (N= 183)
Normal and mild underweight* 85.6 95.7
Malnourished** 14.4 4.3
Total 100.0 100.0

X2 = 9.14, p <0.01.
* Gomez I and normal, or over 75 per cent of standard weight for age.
** Gomez 11 and 111 combined, or 75 per cent of standard weight for age or less.

Using these three measures of sharing (i.e., animal products, fruit, and sweets), an additional analysis was performed to see if sharing of these food groups changed from 1973 to 1975. Table 4 shows that they had not significantly changed. The patterns of sharing animal products, fruit, and sweets and sugars had remained stable, although nutritional status had improved.

TABLE 4. Food-Group Sharing between Meal-Maker and One-Year-Old Child in Peasant Households, 1973 and 1975

Food Group Frequency Differences t-tests
1973 1975
(N = 54) (N = 60)
mean SD mean SD
Animal products 0.94 1.5 0.80 1.5 not significant
Fruits 0.09 0.7 0.15 0.9 not significant
Sweets and sugars 0.87 1.6 0.78 1.2 not significant


In conclusion, this analysis reveals that intrafamilial food sharing appears to affect child nutritional status. However, analysis must focus on specific food groups such as animal products and sweets and sugars. Greater sharing of the former relative to the consumption of the household is beneficial to child nutrition as measured by anthropometry, while greater sharing of the latter is deleterious to nutritional status. Overall, measures of sharing based on food frequency measures may be misleading for this reason.

Nevertheless, sharing patterns do not appear to be readily responsive to nutrition education. In fact, even where known nutritional improvements have taken place due to generally improved economic conditions, changes in sharing patterns could not be found.

I would suggest that intrafamilial food sharing between the weaned child and the mother require additional cross-cultural and intensive observational study. I would especially urge that the food sharing be analysed more in terms of the household's place in the wider social economy of village and nation.

If the patterns are generally as invariant as they seem from this analysis, it would be wise for third-world nutrition policy to focus more on increasing household access to resources than on large-scale "educational" attempts to get a small family pie divided more equitably. Efforts to increase intrafamilial equality should only be emphatically pursued where governments are taking measures to overcome the societal maldistribution both of food and of the means for producing it. To do otherwise is to be blind to the historically created conditions which have pushed families into a cruel contradiction: to survive, they must risk nutritional damage to their young.


1. Alderman, Michael, Owen D. Minott, James Husted, Barry Levy, and Ryan Searle. "A Young-Child Nutrition Programme in Rural Jamaica." The Lancet, 1: 1160-1169 (1973).

2. Arroyo, Pedro, Sara E. Q. de Arroyo, Sara E. Perez Gil, and Adolfo Chavez. "Correlation between Family and Infant Food Habits by Scalogram Analysis." Ecology of Food and Nutrition, 1: 127-130 11972).

3. Berg, Alan. The Nutrition Factor: Its Ho/e in National Development. The Brookings Institution, Washington, D.C., USA, 1973.

4. Davidson, S., R. Passmore, J.F. Brock, and A.S. Truswell. Human Nutrition and Dietetics. 6th ed. Churchill Livingston, Edinburgh, London, and New York, 1975.

5. Desai, Patricia, W.E. Maill, and K. Standard. "A Five Year Study of Infant Growth in Rural Jamaica." The West Indian Medical Journal, 1 8: 21 0-221 (1969).

6. Garn, Stanley, and Diane C. Clark. "Problems in the Nutritional Assessment of Black Individuals." American Journal of Public Health, 66: 262-267 11976)

7. Gross, Daniel R., and Barbara Underwood. "Technological Change and Caloric Costs: Sisal Agriculture in Northeast Brazil. " American Anthropologist, 73: 725-740 11971).

8. Gurney, J. Michael, Helen Fox, and John Neill. "A Rapid Survey to Assess the Nutrition of Jamaican Infants and Young Children in 1970." Transactions of the Royal Society of Medicine and Hygiene, 66: 653-662 11972).

9. IUNS, Committee 10/V. "Rethinking Food and Nutrition Education under Changing Socio-economic Conditions." Food and Nutrition Bulletin, 2 12): 23-28 (1980).

10. Jelliffe, D.B. The Assessment of the Nutritional Status of the Community. World Health Organization, Geneva, 1966.

11. Jelliffe, D.B. Child Nutrition in Developing Countries. Agency for International Development, Washington, D.C., USA, 1969.

12. Marchione, Thomas. "Food and Nutrition in Self-Reliant National Development." Medical Anthropology, 1: 57-79 (1 977).

13. Marchione, Thomas. "Factors Associated with Malnutrition in the Children of Western Jamaica." In N. Jerome, R. Kandel, and G. Pelto, eds., Nutritional Anthropology, pp. 223-273. Redgrave, New York, 1980.

14. Marchione, Thomas, end Fred Pryor. "The Dynamics of Malnutrition in Jamaica." In L.S. Greene and F.E. Johnson, eds., Social and Biological Predictors of Nutritional Status, pp. 201-222. Academic Press, New York, 1980.

15. Pan American Health Organization. Guidelines to Young Child Feeding In the Contemporary Caribbean. Scientific Publication No. 217. PAHO/WHO, Washington, D.C., USA, 1970.

16. Pan American Health Organization. The National Food and Nutrition Survey of Barbados. Scientific Publication No. 237. PAHO/WHO, Washington, D.C., USA, 1972.

17. Pan American Health Organization. The National Food and Nutrition Survey of Guyana. Scientific Publication No. 323. PAHO/WHO, Washington, D.C., USA, 1976.

18. Stini, William. "Evolutionary Implications of Changing Nutritional Patterns in a Human Population." American Anthropologist 73: 1019-1030(1971).

19. Wolf, Eric R. "Types of Latin American Peasantry." American Anthropologist, 57: 452-471 11955).


Food and Nutrition Bulletin, vol. 3, no. 1
( January 1981), p. 20.

In the report of the third meeting of the ACC/SCN Consultative Group on Maternal and Young Child Nutrition, the title and footnote to the first section should have read:


Conclusions and Recommendations*
* These conclusions and recommendations will be included in a review of this subject being prepared by Dr. Barbara Underwood and Dr. Yngve Hofvander.

Supplementary feeding and nutrition of the young child

G.H. Beaton
Department of Nutrition and Food Science, Faculty of Medicine, University of Toronto, Toronto, Canada

H. Ghassemi
School of Public Health, University of Tehran, Tehran, Iran Nutrition Consultant, UNICEF

The writers were given the responsibility of reviewing the experiences of past food distribution programmes for young children and offering a judgement of their benefits, side effects, and costs and, where appropriate, indicating areas in which improvement might be effected. Reports of more than 200 projects were reviewed. About half of these provided quantitative or qualitative information about particular food distribution (take-home or supervised feeding) programmes. (Full listings of the sources reviewed are given in the complete report.)

Much of the available quantitative data is drawn from research or pilot projects. This creates a bias in that, in general, such projects were more effective than ongoing programmes for a variety of reasons. The general impression gained from this review is that food distribution programmes directed toward young children, as now operated, are rather expensive for the measured benefit. However, the reviewers remained unconvinced that the benefit usually measured-physical growth and development-is either the total benefit to the family and community or even the most important benefit. Therefore, it is deemed unwise to withdraw such food distribution programmes until there has been opportunity to assess their full effects and benefits.

"Leakage" of food between its distribution and the net increase of intake of the target recipient has been measured in terms of sharing with others and displacement of food that would otherwise have been consumed. In supervised feeding programmes, the former does not occur; in take-home distribution programmes, sharing may account for 30-60 per cent of the food distributed. Displacement of food was generally greater in supervised feeding programmes than in take-home programmes. Overall, the net increase in intake by the target recipient was 45-70 per cent of the food distributed, with one programme showing a net effect of only 10-15 per cent. Since the food distribution (participation rate) ranged from 25 to 80 per cent of the intended level of distribution, the programmes providing detailed information suggested that a relatively small part of the apparent food (energy) gap was being filled.

The measured benefit of most programmes was anthropometric change or difference from a control population. Not all programmes examined had appropriate controls. Most claimed beneficial effects. For some major ongoing programmes there was no demonstrable increase in anthropometric indices for the programmes as a whole; postive effects were reported in some distribution centres and not in others. Close scrutiny of the results of the total experience suggests that anthropometric improvement was surprisingly small. In part, this may be explained by the relatively low levels of average net supplementation. Children with the greatest apparent weight deficit at entry into a programme tended to show the greatest response to supplementary feeding. Although apparent effectiveness, and cost effectiveness, might be improved by selecting only children who would be expected to respond (selection by indices for which response will be measured), it has been demonstrated that this would exclude many children who could have responded and thereby would reduce effective coverage of the at-risk population.

A few studies reported morbidity and mortality data. When food distribution programmes significantly reduced the prevalence of severe malnutrition, there appeared to be an accompanying reduction of morbidity or mortality from infectious diseases. The effect of preventing "less-than-severe" malnutrition is not clear. There would seem to be a synergistic effect of combined nutrition and health-care programmes. Data from one study suggest that recurrent illness, and accompanying anorexia, may limit voluntary food intake.

One study demonstrated that a response of young children to additional food intake was increased activity. This was reported also for adults. The other studies reviewed did not examine this possible outcome. It is noted that increased voluntary activity in children (play) may affect cognitive development.

The data suggest that the observed growth response accounts for only a small part of the net increase in energy intake. The "missing energy" may be producing unmeasured responses in the children (e.g., physical activity, de-adaptation of the basal metabolic rate, body composition changes). Some of these might have greater significance than growth per se.

The reviewers were concerned that leakage of food has been seen and measured only as an undesirable source of inefficiency of food distribution programmes. There appears to have been little or no attempt to trace the effects of this food, which may account for 30-80 per cent of the food distributed. It cannot be assumed that there is no effect or that the effect in the family and community is not beneficial. If nothing else, it represents an increase in effective income/buying power. It is recommended that detailed investigations be undertaken to establish the effect of supplementary food on the individual and on the family and the community. Until such research has been undertaken, the real scope of effects and benefits of supplementary feeding programmes cannot be known.

Total costs (food plus administrative costs) of operational programmes intended to provide 300-400 kcal/day were about US$15-25 per enrollee per year (1976 dollar equivalent!. In take-home programmes about 70 per cent of this was cost of food. In supervised feeding programmes the total costs, and administrative costs in particular, were somewhat higher than in take-home programmes. These costs are expressed in terms of nominal enrolment. If participation rates approximated 100 per cent, costs would approximately double. Costs in research and pilot programmes were generally higher, reflecting the increased numbers and training of workers. These programmes were also more effective. This may suggest that administrative costs have been kept too low in operational programmes -that additional or better prepared personnel might increase effectiveness.

There was little information about the educational impact of food distribution programmes. In those programmes that had an educational activity, it was usually directed toward appropriate usage and targeting of the distributed foods. No clear judgement can be made about the benefit or otherwise.

The present scale of food distribution programmes, although accounting for substantial amounts of money (or food equivalents), is probably much too small to have major impact on total communities or countries. If programmes are to expand to the point that they can exert real impact, it is essential that their true objectives be defined and that the programmes be designed and implemented accordingly. The objective might be to use such programmes as an instrument for redistribution of effective income/demand and for community development (with the community as the "target") or as a specific form of supplementation targeted toward high-risk individuals. The design would be quite different for these two goals.

The reviewers addressed, in general terms, the question of food aid and have concluded that for many countries, and for many years to come, food aid must continue and must be increased if nutritional conditions are to be improved. The question then becomes how best to use food aid rather than whether it should be accepted. Appropriate food distribution programmes may be a desirable way of using food aid as a force for income redistribution. Because of the small scale of existing programmes, this proposal cannot be tested from experience.

It must be noted that in many, if not all, of the communities that might be selected for food distribution programmes, populations are now in equilibrium with their unfavourable environment, including chronic under-feeding, as the result of social and other adaptations. A food distribution programme may disrupt these adaptations as well as effecting an improvement in overall health. If such programmes are abruptly withdrawn, the adaptations may not be quickly re-established. Concern must therefore be expressed about the need for reasonable assurance of continuity before food distribution programmes-whether experimental or operational - are initiated. The programmes reviewed do not provide data on the effect of withdrawal. However, the literature provides ample evidence that programmes have been withdrawn after a year or two of operation.

This report may raise more questions than it answers. Certainly it challenges many of the assumptions that have been widely held. For this reason, it is strongly recommended that additional, intensive studies of the effects of food distribution programmes be undertaken. Since such programmes are likely to expand in response to both political pressure and general scientific opinion that if there is widespread undernutrition feeding must be good, there would seem to be some urgency in addressing the questions raised in the present report.

Contents - Previous - Next