Contents - Previous - Next

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

Influences on infant feeding in Dar Es Salaam

Han Bantje
Institute of Resource Assessment, University of Dar es Salaam, Dar es Salaam, Tanzania

Olyvia Yambi
Tanzania Food and Nutrition Centre, Dar es Salaam, Tanzania


This paper summarizes the findings from a study of infant feeding practices in Dar es Salaam, the capital of Tanzania, in 1980. It was part of an international investigation entitled, "Rethinking Infant Nutrition Policies under Changing Socio-Economic Conditions", organized by a task force of the International Union of Nutritional Sciences (IUNS) (1). Its broad aim was to identify negative material constraints on breast-feeding, and to stimulate social and political action to remove them. The comparative results have been published in the final project document (2).

The worldwide shift in infant feeding practice has been a cause for concern for some time. An increase in early weaning and reliance on bottle-feeding with breast" milk substitutes have been reported from many countries. These changes are part of the pervasive social transformation taking place in the Third World, involving the breakdown of traditional cultures, rapid urbanization, more intensive communication between countries and social groups, the demographic transition, and the changing position of women. Rapid social change creates insecurity and a quest for symbols of modernity that makes people vulnerable to the promotion strategies of the producers of milk formulas. Health workers are faced with the distressing and sometimes fatal effects of artificial feeding under poor hygienic conditions. Much energy is spent to convince governments, the milk companies, and the public, of the dangers of artificial feeding. Within the wider context of female activism, such efforts sometimes merge with action for women's rights to breast-feed their babies.

Sociologically, the change of feeding patterns as a result of social transformations raises the issue of the linkage between macro-level processes and individual decision making. Individuals do not react to abstract macro-level processes, but to the concrete micro-level conditions that result from them. The aim of the study was to test the assumption that women's behaviour is, to a large extent, determined by material constraints, such as the availability of money, time, and energy that largely depend on female employment status and the socio-economic level of the household. Furthermore, a shift towards artificial feeding can only be expected to occur if people actually have access to formula milks and are exposed to their promotion. Therefore, a second objective was to study the impact of formula milk promotion on mothers' attitudes and practices.

Fortunately, the shift from breast to bottle has not progressed nearly as far in Tanzania as in many other countries. Government policies restricting the imports of milk formulas, paired with an economic straight-jacket that prohibits almost all food imports, have served to retain adherence to traditional breast-feeding habits. A legally enforced maternity leave period of three months enables women to fully breast-feed for at least that long. Nevertheless, the pressures towards a change in feeding practices may also be observed here.

The study addressed itself mainly to the question: In what way are feeding patterns influenced by household conditions and specific material contraints on women? Practices related to breast-feeding, the initiation of weaning foods, and current infant diets are related to household income and female employment status, as well as to a number of social and cultural variables covering maternal background, living conditions, knowledge and attitudes, and exposure to formula promotion. The impact of government maternity protection legislation is briefly outlined.

The Field Study

Sample households were systematically selected from four urban wards on the basis of a preliminary census. The sample consisted of 219 mother-child pairs, distributed more or less equally over the four sub-groups resulting from subdivisions by income and employment status (table 1). High-income households were defined as those with a total monthly income of over Tshs 2000/-, roughly three times the official minimum wage at the time of the survey. Women were defined as employed when engaged in some form of formal wage earning. The others were termed housewives for want of a more convenient term. Contrary to expectation, so few informally employed women were identified that they could not be considered as a separate sub-group.

TABLE 1. Composition of the Sample

  Low-income High-income Total
Employed 53 55 108
Housewives 77 34 111
Total 130 89 219

Because low-income employed women were few in the area-based sample, their number had to be supplemented by selection from factories to fulfill objective of having four subgroups of comparable size. As a result, the composition of the sample does not reflect the proportions of those subgroups in the population at large.

Household interviews ware conducted by Tanzanian university students. They were based on a formal questionnaire covering feeding habits, employment conditions, economic status, living conditions, knowledge and attitudes, health-related behaviour, and how women allocate their time. In addition, in-depth studies were conducted in 28 households over a four-week period to obtain detailed activity and feeding records, as well as to gain some insight into the social background of behaviour and attitudes relevant to infant feeding. The household information was put into perspective by background data obtained from site visits to hospitals and retail outlets, key informant interviews, group discussions, and a number of supporting studies covering demographic aspects, health care, female employment conditions, maternity legislation, and the structure of the formula market (2).

In designing the sample, care was taken to assure similarity of the sub-groups in terms of maternal age, as well as number, sex, and age of the index children. The children were evenly distributed over the 0 to 24-month age range, while a few slightly older children were retained.

A few descriptive characteristics of the sample households have been summarized in table 2. A strong association between income level, employment, and marital status is apparent. Single women were more common in the low income group (23 per cent) than in the high-income group (9 per cent). Among the low-income employed women, 41 per cent were single, while only 5.5 per cent of the high-income employed were unmarried. A strong association was also found between education, employment, and income level. Low-income women very rarely had more than primary education, and often less. High income women often had secondary or university education, and rarely less than complete primary education. In both income groups, education clearly promoted employment opportunities.

The variation in social conditions between the sub-groups is also apparent from the interrelations between house. hold size and dependency ratio. Although high-income households are larger, they have a lower dependency ratio than low-income households. This is because high-income households often accommodate relatives who are working or studying in town, some of whom contribute to the household income. Low-income households can rarely afford to harbour relatives, who in any case would be unlikely to contribute to the household income.

The Effect of Income and Employment Status on Feeding Practices Breast-feeding. The start on breast-feeding was found to be strongly influenced by the conditions at the place of delivery. Of the index children, 9.6 per cent were born at home, 58 per cent in government hospitals, 15 per cent in private hospitals, and 17.4 per cent in smaller health units. Low-income women generally deliver in government hospitals, where rooming-in is standard and artificial feeding discouraged. Ninety per cent of the low-income group said they had started nursing within 24 hours after delivery. High-income women often give birth in private units for which they pay fees, where rooming-in is not general and bottle-feeding is routine. Such women reported a slower initiation of breastfeeding: only 64 per cent had started within 24 hours.

About one-third of the mothers experienced various minor problems with breast-feeding, but only six per cent breastfed for less than three months. Of these, half were Asian women (i.e., women of Indian origin) who are well represented among high-income households, and on the whole are more inclined towards artificial feeding than African mothers are. Only two women in the sample did not breast-feed their babies at all. As most of the children were currently breast-fed at the time of the survey, few data were obtained on the total duration of breastfeeding. From the available information, it appears that low-income women, on average, continue longer than high-income women (table 3). With six months as a cutoff point, the difference is statistically significant. Employment status seems to have a minor effect. Among low-income women there is hardly any difference between the employed and the housewives. Unemployed high-income mothers tend to stop even earlier than the employed. This dispels the notion they employment as such necessarily interferes with the duration of breastfeeding.

TABLE 2. Summary of Household Characteristics

  Low-income High-income
Employed Housewives Employed House wives
Single Women (%) 41.5 10.4 11.3 5.5
Household size 5.9 5.6 6.7 7.2
Dependency ratio 2.3 3.7 1.7 3.2
Education of mother        
< primary level (%) 26.4 70.1 0.0 27.7
secondary level (%) 5.7 1.3 77.3 25.0

TABLE 3. Duration of Breast-Feeding by Income and Employment Groups

Duration (months) Low-Income High-Income Total
Employed Housewives Employed Housewives
0-6.* 14.3% 14.3% 47.6% 64.3% 35.5%
7-12 21.4% 14.3% 30.0% 21.4% 22.8%
13-18 50.0% 52.4% 20.0% 7.1% 31.6%
>18 14.3% 19.0% 3.3% 7.1% 10.1%
N = 14 21 30 14 79

* At six months the difference between income groups is significant: X-square = 10.7; P < 0.01.

On the other hand, the duration of exclusive breastfeeding was found to be influenced by employment status rather than income (table 4). High-income women introduce supplementary foods earlier, but the low-income employed catch up before three months, the end of the legally enforced maternity leave period. At that point the difference between employed and unemployed mothers was statistically significant, while the difference between income groups was barely so.

TABLE 4. Percentage of Children Receiving Supplementary Foods in the First Three Months

  Low-Income High- Income Total
Employed 82.0 83.0 82.5
(50) (47) (97)
Housewives 47.8 69.6 53.3
(67) (23) (90)
Total 62.4 78.6 68.4*
(117) (70) (187)

Statistical significance: between employment groups:
X-square = 17.0; P < 0.001; between income groups: X-square = 4.6; P < 0.05 (Numbers in brackets).

Infant feeding. Infant diets consisted of various combinations of breast-milk, formula milk, and other foods. Maize porridge was by far the most common supplementary food in all age groups. More than half of the children were given maize porridge before the age of six months, and one in four received fruit juice, mainly orange juice. Formula milk was used by slightly over half of the index mothers for children of this age, more often in the high-income group. Fruit juice and glucose were also mentioned more often by high-income women. After six months the children were increasingly fed mixed foods from the family pot, mainly cereals with a sauce of meat, fish, or vegetables.

Almost one-third of all mothers stated that they had used formula milk as a first weaning food. High-income women had done so more often than low-income women, and the employed more often than the housewives. The difference between income groups was small, and that between employment groups only significant among low-income women. It appears that low-income employed women use formula milk as a first weaning food even more often than high-income employed women (table 5).

The use of milk formulas beyond a very early age was found to be significantly more common among the high income group (table 6). The difference between employment groups here was only significant at the low-income level, where unemployed women rarely use formula milk. Income is obviously a major constraint here. Low-income employed women are somewhat dependent on artificial feeding because of their absence from home, but our data indicate that, while in the high-income group the use of formulas increases with age, low-income women often initiate formula feeding but soon abandon it.

The use of feeding bottles was reported by 83 per cent of high-income, and 49 per cent of low-income women. In both groups it was more common with the employed.

TABLE 5. Percentage of Children Receiving Formula Milk as a First Weaning Food

  Low-Income High-Income Total
Employed 48.8 32.6 40.2
(43) (49) (92)
Housewives 15.0 34.5 21.3
(60) (29) (89)
Total 29.1 33.3 30.9.*
(103) (78) (181)

The difference between employment groups is significant in the low-income group: X-square = 12.3; P < 0.001 (Numbers in brackets).

Those who use formula milk invariably also use bottles. However, some of the same women feed other liquid foods (tea, juice) from cups, demonstrating the lack of a real need for the feeding bottle. But the mental association of formula and bottle is strong. Ironically, locally manufactured feeding bottles became generally available around the time of the study, although milk formulas were not.

TABLE 6. Prevalence of Formula Milk in Infant Diets (%)

  Low- Income High-Income Total
Employed 30.2 47.3 38.9
(53) (55) (108)
Housewives 13.0 52.9 25.2
(77) (34) (111)
Total 20.0 49.0 40.0.*
(130) (89) (219)

*Statistical significance: income groups X-square = 4.9; P < 0.01; employment groups X-square" 4.9; P < 0.05 (Numbers in brackets).

Social and Cultural Variables

Of the 50-odd social and cultural variables investigated, many showed no significant associations with infant feeding behaviour. The following is a brief summary of those that were found to be relevant, in particular to the choice for or against formula milk.

Mother's background. No significant associations were found with maternal age, tribal background, religion, place of birth, or length of stay in town. Level of education was associated with the current feeding pattern, but not with the start of weaning. An inverse relationship was found between level of education and breast-feeding: both among housewives and employed women, those with more education were inclined to an earlier termination of breast-feeding (table 7), and were also more likely to use formula milk. This finding has to be seen in the context of the interrelations among education, employment, and income.

Economic status and living conditions. Our sample house holds displayed a wide range of economic and living conditions, with concomitant differences in social security and hygienic circumstances. There is no doubt that style of living and the ability to cope with adverse conditions are pervasively influenced by household income, the level of which primarily depends on the type of occupation of the head of the household. Those in the low-income groups were engaged in manual labour, the service sector, and lower level professional jobs, whereas heads of high-income households were in higher and middle level professional, clerical, and managerial jobs. In addition to higher salaries, the latter often had access to various side-benefits such as housing and company transport.

A number of variables related to living conditions of the household correlated with both the start of weaning and current feeding practices. These were expenditure on food, type of house, the number of rooms, the type of water supply, and sanitary facilities. All of these directly depend on income level. Of special interest was the association between the use of formula milk and the presence of a refrigerator. The latter is a status symbol par excellence, but, more importantly, in a hot country the safety of artificial feeding is greatly improved if milk can be stored in a cool place.

Knowledge and attitudes. One of the main concerns of the study was the influence of formula promotion on mothers' perceptions, attitudes, and feeding practices. A distinction was made between direct promotion by means of advertisements, free samples, and spread of information, and indirect or soft informational and attitudinal aspects reflected in opinions about the desirability of certain infant foods and feeding practices.

The promotion of infant formulas has been officially discouraged in Tanzania for a number of years. Moreover, the economic situation of the country has severely restricted the imports of commercial infant foods. No advertisements are heard on the radio or seen in government hospitals, and no free samples are channeled through the government health services.

TABLE 7. Duration of Breast-Feeding by Level of Education(%)

  Primary Education Secondary Education
Employed Housewives Employed Housewives
Duration (Months)        
0-6 30.0 30.0 43.5 75.0
7-12.* 15.0 16.7 34.8 -
13-18 40.0 43.3 21.7 -
> 18 15.0 10.0 - 25.0
N = 20 30 23 4

*The difference between education levels is significant, with 12 months 85 a cut-off point: X-square = 6.0; P < 0.02.

Nevertheless, almost one third of the index mothers reported that they had been directly exposed to advertisements for formulas. The majority of those who did, mention Lactogen, which was also rated the most preferred brand. Lactogen has dominated the market in Tanzania for a long time, and was forcefully promoted in past years.

Advertisements from milk companies are still retained as wail decorations in private hospitals, and free samples are occasionally distributed via that channel. Site visits at health institutions revealed clear differences in administrative practices. In private hospitals there are no rooming-in facilities, and infant formulas are used indiscriminately in the neonatal ward. In government hospitals infants stay with the mother and breast-feeding is on demand. When inevitable, formula milk is fed by cup or tube. Women who delivered in private health units had significantly more often been in contact with direct promotion and specific instructions regarding the use of formula milk than other mothers had.

Thus, it is not surprising that the birthplace of the child appeared to have an impact on feeding practices. Children born in private hospitals received supplementary foods earlier, and much more often these consisted of formula milk. Such children were more often breast-fed for less than six months. However, these differences approached statistical significance only among the high-income unemployed because low-income women very rarely give birth in private hospitals. The high-income employed are induced to early weaning and formula feeding by the demands of their jobs regardless of where they give birth.

Yet, a preference for artificial feeding and early weaning was also expressed by some of the low-income women who had not been directly exposed to advertisements. Such views appear to be a relic of past promotional inputs transmitted by word of mouth and filtering through from the higher to the lower socio-economic stratum. The presence of sizeable Asian and European communities may also play a part. On the other hand, it should be borne in mind that traditional concepts of infant feeding sometimes include notions about early supplementation of breast-milk. No doubt there is a constant flow of ideas and secondary information about such matters in society, but this is difficult to document because of its diffuse, informal nature.

Government Legislation in Support of the Working Mother

Maternity protection legislation in Tanzania provides for 84 days of paid maternity leave once in three years, irrespective of marital status. Our data show near total compliance with this law. In the few cases where employed women had not been granted paid leave it was usually because less than three years had passed since their previous maternity leave. A small number of cases dated back to the time when this leave was restricted to married women. However, our sample included a few cases in which a breach of the law was most likely to occur: in marginally institutionalized occupations such as bar workers and domestic servants.

The law also ensures the right of working mothers to take two half hour nursing breaks a day to feed their babies. This provision cannot easily be put into effect because creche facilities are rare at work-places, and the distance from home is often far. Day care centres care for older children, and only women with access to private transport are able to take advantage of the short breaks. Of 94 respondents, only 11 said they had a chance to breast-feed during office hours.


In this summary article, only a few aspects of the study could be presented. The detailed analysis of variables was not an end in itself, but instrumental to a better understanding of the factors influencing infant feeding practices, to provide guidelines for government policy making. Although less advanced than elsewhere, the trend towards artificial feeding is also present in Dar es Salaam, but could perhaps be curbed by appropriate action.

Because of the lack of a major industrial focus, urbanization in Tanzania has started late. Currently about 10 per cent of the population live in urban centres. Even Dar as Salaam retains a distinctly rural touch. Women often return to their home villages for delivery and spend most of their maternity leave there. Alternatively, relatives may come from the rural area to assist the mother after childbirth. All this encourages the persistence of traditional patterns of child care and feeding. Nevertheless, the dependence on commercial infant foods is already very real in certain sectors of the population. In a group discussion, employed women emphatically expressed the hardships they faced because of the scarcity of baby foods. They have to go to great lengths and pay exorbitant prices to obtain any at ail. In view of this, generalized warnings about the dangers of bottle-feeding are obviously ineffective. As long as formula milks are scarce, their use will be limited by economic factors. In the long run, a well considered national policy will be needed to avoid the adverse side effects of artificial feeding.

The importance of our study lies not so much in the documentation of the negative effects of formula milk promotion on infant diets, which was more clearly evident in the parallel studies in Sao Paulo and Colombo, but in the more refined picture it presents of the conditions surrounding women's lives in a Third World town. Economic and social conditions make it impossible to transfer the Western ideals of female employment and unhampered opportunity to breast-feed without adapting them to the local situation. Women can only be persuaded to change their practices if they perceive this as desirable for themselves or their children, and possible within the context of their lives. The prevailing conditions in Dar es Salaam make the combination of employment and motherhood strenuous, and hardly possible without recourse to some form of non-human milk feeding. Cow's milk being in very short supply, mothers avidly turn to whatever; other milk products are available. It does not help to dismiss such solutions imposed by circumstances, as one member of the project team suggested. There are very real contradictions between the goals of Western activists and the needs of Third World women facing the hard realities of life. Whoever seriously wants to improve those realities cannot avoid facing this contradiction.

In trying to document the current situation and associated factors, our study concentrated on the respective effects of income and female employment status. The findings indicated that overall socio-economic level, epitomized by total household income, is the dominant factor influencing material conditions and feeding practices. Therefore, conditions surrounding the mother-child dyed cannot be isolated from those facing the household and social group. The material and social support mobilized on the basis of social networks was found to be an important source of security, depending to a large extent on where one stands in the social and economic continuum.

Women in Third World societies function as individuals to a lesser degree than in the West. Their behaviour is very largely conditioned by the social groups to which they belong, and from which they have little freedom to withdraw. We came across many instances where women drew on family and neighbourhood groups for support in the struggle for survival. Perhaps the most important factor is the quality of marital relationships. Marital instability is a frequent cause of social and financial insecurity, leading to an emotional instability that may result in erratic child care. Conversely, the social position of the husband in many cases gives the mother access to resources she would not have had otherwise.

Social status was also found to influence the choice of the health care facility for delivery. The practices followed in the maternity wards are one of the strongest influences on the choice of feeding methods. An association was found between the feeding pattern and first food supplements with subsequent feeding. Those who give supplements early also stop breast-feeding early. Those who give formula milk tend to continue doing so. It appears that the start of food supplementation (at what age and with what foods) is the most crucial decision in the whole feeding process.

Thus, there appears to be a typical sequence from a hesitant start of breast-feeding to early supplementation, and early termination of breast-feeding. However, this is not a simple causal chain, but a complex syndrome. Both the start of weaning and current feeding practices have highly significant associations with income level, while the start of supplementary feeding is also associated with employment. Income and employment conditions influence the choice of the health care facility for delivery, the initiation to weaning, and the subsequent feeding pattern.

An inverse relationship was found between breast-feeding and bottle feeding. Bottle-feeders tend to use formulas as the first supplement, and do so at an earlier age. Among families with formally employed mothers, infants who were given formula milk as a first supplement all received less than three months of exclusive breast-feeding. There is also a tendency towards shorter total duration of breastfeeding.

No evidence was found that a mother's employment as such was conducive to shorter breast-feeding, although it encourages earlier supplementation. The maternity protection legislation is relatively progressive. Extension of the paid maternity level has been proposed by some, but serious questions may be raised about its feasibility in view of the present state of the economy. It may even be counterproductive by motivating employers to hire male workers instead of females. It should also be realized that formally employed women constitute only a minute section of the population. No system of formally recognized maternity leave can be envisaged for the mass of rural women, who usually do more physically exhausting work than the urban employed.

As nursing breaks during work hours tend to remain ineffective, a better alternative would be to allow nursing mothers a shorter work day, enabling them to return home earlier. Creches at workplaces seem an attractive solution, but are beset with practical problems. An experiment carried out at the Tanzania Food and Nutrition Centre showed that women are reluctant to bring very young babies. The main reasons are the hazards and inconvenience of travelling to the workplace by public transport with a small baby, and mothers' reluctance to leave their babies under the care of attendants over whom they have no control.

In view of the public demand for infant formulas, and the real need for them in a few cases, a total ban appears neither feasible nor desirable. Those who have become dependent on formulas will continue to procure them by any means. Government policy should aim to maintain the still high rate of breast-feeding by creating the best possible conditions for nursing women, and reducing as much as possible the negative effects of formula feeding There is no easy, clear-cut prescription for doing this.

It is felt that the official attitude against formula promotion, and the restriction of imports of baby foods, are beneficial and should be maintained. There are inconsistencies at the policy level, however, because preparations to establish a formula milk factory in Tanzania itself have reached an advanced stage. The local production of feeding bottles is also a sore point, especially since it began at the very time when awareness of the dangers of bottle feeding was growing. Since the time of our survey, private medical practice has been officially abandoned in Tanzania, although it continues in institutional disguise. It should be possible to put some pressure on private maternity clinic personnel by at least making them aware of the implications of their management of neonatal feeding.

The importance of encouraging breast-feeding should be stressed in the curricula of medical training schools at all levels, to make the health care delivery system instrumental in the promotion of breast-feeding. Generally, more attention should be paid to the role of nutrition in the aetiology of diseases of both women and children.

A secondary finding in our study was that the knowledge about home preparation of weaning foods leaves much to be desired. Improvement of both home-made and commercially produced weaning foods in necessary, as well as education on proper feeding practices. The Tanzania Food and Nutrition Centre has meanwhile undertaken further work on this matter. Thus, there are a number of practical actions possible that would serve the goal of reinforcing or restoring adequate infant feeding practices without involving large financial commitments. The current economic and ideological climate in Tanzania provides a favourable matrix for such actions.


The authors gratefully acknowledge the contribution of the whole project team to the Tanzania country study, in particular those of the methodology coordinator, Dr. Thomas Marchione. However, the views expressed in this paper are not necessarily shared by all team members.


1. T.J. Marchione and E. Helsing (Eds.), Rethinking Infant Nutrition Policies Under Changing Socio-Economic Conditions. Interim report of IUNS-Sponsored Task Force. IUNS Publication No. C36 (1980), Xerox, 107 pp.

2. T.J. Marchione and E. Helsing (Eds.), Results and Policy Implications of the Cross-National Investigations: Rethinking Infant Nutrition Policies Under Changing Socio-Economic Conditions. Project report of IUNS Sponsored Task Force. IUNS Publication No. C41 (1982), 97 pp.


Contents - Previous - Next