It is now well established that breast-feeding has important health and contraceptive benefits and for this reason there is much concern about potential shifts in infant feeding regimes in the course of socio-economic development (VanLandingham, Trussell and Grummer-Strawn, 1991). Knowledge of infant feeding patterns in developing countries has been substantially expanded in recent decades as a result of the increased use of sample surveys to monitor the situation. In particular, the World Fertility Survey (WFS) in the 1970s and early 1980s routinely incorporated questions about breast-feeding; its successor in the 1980s and 1990s, the Demographic and Health Surveys (DHS), has continued this tradition and has also commonly added additional questions about other aspects of infant feeding to its core questionnaire (Trussell and others, 1992). One important country for which such information has previously been lacking is Viet Nam since it did not participate in either of these programmes. Fortunately, it is now possible to redress this situation because two recent national surveys, the 1988 Viet Nam Demographic and Health Survey (VNDHS) and the 1994 Viet Nam Inter-censal Demographic Survey (VNICDS), provide the requisite data.

________________

* The authors of this article are Truong Si Anh, Chief of the Population, Labour and Social Affairs Team at the Institute of Economic Research, Ho Chi Minh City; Ngo Thi Thai Hoe, Researcher at the Institute of Hygiene and Public Health, Ho Chi Minh City; Le Huong and Tran Thi Thanh Thuy, Researchers at the Institute of Economic Research, Ho Chi Minh City; and John Knodel, Professor of Sociology at the University of Michigan. This study was funded through UNFPA project VIE/93/P03. The authors would like to acknowledge with thanks the helpful advice and information provided by Hillary Page, Nguyen Thi Thuy, Pham Gia Duc and Nguyen Thi Nhu Ngoc.

The VNICDS was conducted by the General Statistical Office of Viet Nam (GSO) during the period April through June 1994. The nationally representative sample included almost 10,500 ever-married women aged 15-49 in all 53 provinces of the country. The questionnaire also modeled after the DHS but with considerably more detail than that of the 1988 VNDHS. Extensive information was solicited not only on breast-feeding but also on other aspects of infant feeding, including questions about food given concurrently with breast milk as well as the age at which various types of food were first given regularly (see GSO, 1995). The survey also included questions about the return of menstruation following childbirth. Questions on infant feeding and the return of menstruation were asked either about the most recent birth during the previous 10 years or about the three most recent births during this period.

The 1994 VNICDS sample is much larger and far less clustered than that of the 1988 VNDHS. Thus, comparisons between the two surveys can be at best only suggestive of recent trends over time. Comparison of results are nevertheless useful for determining if general patterns are similar and thereby to serve as a basis for judging the overall credibility of both surveys.

Various methods exist for the analysis of breast-feeding durations and other infant feeding practices (Lesthaeghe and Page 1980; Page and others, 1982). In this analysis we rely on two: the "current-status" approach and the "life-table" approach. The current-status approach examines the proportions of children at successive ages who are still being breast-fed (or who have been started on other types of food). By assuming that the cross-section of children at successive ages resembles the experience of a cohort of infants as they age, it is possible to estimate the median duration of breast-feeding (or median age at introduction of other foods). Its main advantage is that there is likely to be minimum response error as to the status of the child since no recall is involved for the respondent to report if the child currently receives breast-milk or a particular type of food. Its main disadvantage is that the proportions currently being breast-fed need to be tabulated for relatively narrow intervals of age and thus may fluctuate because of the small numbers of cases on which they are based. In the present study, current-status results are based on all living children. To smooth irregularities in the pattern of decline with successive monthly ages, five-month, equal-weight moving averages are used.

The life-table approach can provide more detailed information on the duration of infant feeding regimes than the current status approach because it takes advantage of additional information on retrospectively reported ages at which events occur.1 As a result, however, life-table estimates are influenced by inaccuracies in recall of whether the child had ever been breast-fed and of the age at which breast-feeding ceased. The life-table method also ensures that the estimated proportions currently breast-feeding decline monotonically with successive months since birth.

A very common source of error in the recall of ages at which particular events occur (e.g. weaning) is the tendency of respondents to reply in approximate terms, thus "heaping" them at certain ages. Retrospectively reported ages at weaning are typically concentrated at ages corresponding to half-years even when responses are requested to be in terms of months. Such a pattern of responses is found in both surveys with the most commonly reported age at weaning by far being 12 months (or one year) and sharp peaks evident at durations of 18 months, 24 months, and durations corresponding to successive half years. We have adjusted the life-table results to reduce the inaccuracy that would otherwise be introduced by the severe heaping of responses at particular reported ages at weaning. More specifically, we first calculate the monthly probabilities of weaning using usual life-table procedures. We then calculate five-month, equal-weight moving averages of the observed monthly probabilities of a child being weaned starting at the duration of three months since birth. The smoothed monthly probabilities, rather than the observed ones, serve as the basis for calculating the cumulative proportion weaned at successive monthly ages.

Breast-feeding initiation

Table 1 shows the percentage of children ever breast-fed by selected background characteristics as determined from the 1988 and 1994 surveys. The results clearly indicate that breast-feeding in Viet Nam is practically universal. In addition, almost no change is evident over the decade covered, whether measured by an internal comparison between the two half-decade periods shown for the VNICDS or by a comparison between the VNDHS results and those for the most recent half decade covered by the VNICDS. The persistent universality of breast-feeding over the past 10 years is also consistent with the results from a study recently conducted by the National Institute for Nutrition (Khoi and Giay, 1994:54).

	1988 DHS	1994 VNICDS
	During first 5 year before the survey	During first 5 years before the survey	During second 5 years before the survey
Total sample	98	97	98
Residence
Urban	96	96	98
Rural	98	98	98
Sex of child
Male	98	97	98
Female	98	98	98
Education of mother
No schooling	97	97	99
Grades 1-4		98	98
Grades 5-7	99	97	98
Grades 8-11		98	98
Grades 12+	97	97	98
Occupation of mother
Agricultural	98	98	98
White collar		96	98
Other working	98	96	98
Not working	92	98	98
Wealth status by residence
Urban low	n.a.	96	98
Urban middle	n.a.	97	99
Urban high	n.a.	94	98
Rural low	n.a.	97	98
Rural middle	n.a.	98	98
Rural high	n.a.	98	98

Notes: Results are weighted. The percentage of children ever breast-fed (from the VNICDS) refers to the last three children born within 10 years of the survey; the percentage of children ever breast-fed (from the VNDHS) refers to all the children born within five years of the survey.

n.a.= not available.

In the VNICDS, mothers who indicated they had not breast-fed a particular child were asked the reason for not doing so. The responses are revealing and help to confirm that the norm of breast-feeding in Viet Nam is virtually universal. In two-thirds of the cases, the child either died or was too weak to breast-feed, in 9 per cent of the cases the mother was too ill or too weak herself to breast-feed and in 16 per cent of the cases the mother said she had insufficient breast-milk to feed the child. Thus, in the vast majority of cases of non-breastfed children, the decision not to breast-feed was essentially not a matter of choice but forced by circumstances.

The universality of breast-feeding is underscored by the lack of any pronounced difference in the percentage ever breast-fed according the child's sex, mother's educational level or occupation, or household wealth status.2 Moreover, for none of the categories of women shown is any appreciable decline in breast-feeding initiation evident.

The VNICDS also inquired, with respect to the most recent birth, how long after giving birth the respondent waited before first giving her breast to the child. Since colostrum, the early milk present in the breast during the first few days after childbirth, provides necessary nutrients and immunological protection for newborn infants, health professionals advocate starting breast-feeding immediately, i.e. within the first hour after the child is born (Institute for International Studies in Natural Family Planning, 1990). This practice, which is currently being promoted by Viet Nam's Ministry of Health, is apparently followed only by a minority of Vietnamese women (18 per cent nationwide) according to the results in table 2. Nevertheless, most mothers start to breast-feed the same day that they give birth and only slightly more than one in ten mothers who breast-feed start two or more days after giving birth. Differences in the timing of breast-feeding initiation are minimal only between urban and rural mothers.

Some trend towards early initiation is suggested from a comparison of the three-year period prior to the survey and the previous two years before that. While this could reflect some impact of government efforts to promote this practice, it is significant that the percentage of mothers who started breast-feeding in the first hour is highest when delivery took place in the mother's own home and only about average for cases where the mother gave birth in a government facility.

	Percentage of mothers initiating breast-feeding
	Within 1 hour	Within 12 hours	After 2 or more days
Total sample	18	59	11
Period prior to survey
0-2 years	19	60	11
3-4 years	14	58	12
Residence
Urban	16	59	11
Rural	19	59	11
Place of delivery
Own home	22	59	14
Other home	7	34	23
Commune health centre	16	62	7
District hospital/health center	17	65	8
Provincial hospital	14	57	14
Central hospital	14	52	16
Other	22	64	6

Notes: Results are weighted. Tabulations refer to the mother's most recent birth within five years of the survey and are restricted to children who were ever breast-fed.

n.a.= not available.

Duration of breast-feeding

Figures 1a and 1b show the smoothed current-status estimates of the percentages still being breast-fed and the adjusted life-table results for both the 1988 VNDHS and the 1994 VNICDS. In both surveys, the current-status approach indicates modestly higher percentages of children being breast-fed at ages over 12 months than do the estimates derived through the life-table approach. The modest differences observed in the case of the two Vietnamese surveys is likely attributable to memory error in the reported ages at weaning as it affects results derived from the life-table approach. Although the life-table estimates are substantially improved by the adjustment procedure described above, apparently the procedure does not completely overcome distortions in the estimates attributable to heaping.

Overall the estimates produced by current-status analysis and by the life-table method based on births during the three years prior to each survey show similar patterns. Almost all children are breast-fed through the first six months with both surveys indicating that 80 per cent are breast-fed at least a year. Most children are weaned sometime during their second year of life. As a result, only a small minority are still being breast-fed by their second birthday.

Comparisons between the two surveys with respect to the median durations of breast-feeding for each of two estimation methods, as can also be seen in figures 1a and 1b, suggest little change over time. The current-status estimates indicate a median duration of 16.3 months based on the 1988 VNDHS and 16.9 months based on the 1994 VNICDS. The life-table estimates, based on births occurring within the three years prior to each survey, are 15.3 months according to the 1988 survey and 15.9 months according to the 1994 survey. Thus, both sets of estimates indicate slightly higher figures for the more recent survey. Given the differences in sampling and other aspects of the two surveys, these results are only suggestive, at most, of a recent increase in breast-feeding duration.

Differentials in breast-feeding duration based on the VNICDS and as derived through both the current-status and the life-table approaches are shown in table 3. The median duration of breast-feeding indicate that rural women breast-feed their child longer than urban women. Nevertheless, the median duration for urban women is still over one year according to both types of estimates. Although son preference has been a traditional norm, this apparently has little bearing on the breast-feeding durations of boys and girls, who are breast-fed for about the same amount of time. Although there is not a uniformly inverse association between the mother's education and the duration of breast-feeding, children of mothers with no schooling are breast-fed the longest and children whose mothers are the most educated are breast-fed the shortest. For those children whose mothers are engaged in agriculture, the duration of breast-feeding is longer than for children of mothers in other occupations, especially those with white-collar type jobs. Occupation is of course linked to residence and thus the occupational differentials coincide with the longer breast-feeding in rural than urban areas. Within both rural and urban areas, the duration of breast-feeding is associated with the household wealth level. Despite the small number of cases in the urban categories, the results clearly indicate that children in wealthier households are breast-fed for shorter durations than children in poorer households, both in urban and rural areas.

Table 3 also shows the proportion of children who were breast-fed at least 6, 12 and 18 months according to the adjusted life-table calculations. Differences between the different categories of children shown are quite small with respect to the percentages breast-fed at least six months, and rather modest even with respect to the percentage breast-fed 12 months. Over 90 per cent of children, regardless of the category they were in with respect to the variables examined, were breast-fed at least six months. Moreover, in none of the groupings of children shown are less than 70 per cent of the children breast-fed for less than a year. In contrast, far more pronounced differences are apparent in the percentages breast-fed 18 months for most of the categories shown, except for the sex of the child. Thus, much of the differences reflected in the median durations arise from differences in weaning after the child reaches age one and thus only appear in the probabilities of being breast-fed beyond the first year of life.

Perhaps the most striking feature of the demographic and socio-economic differentials in breast-feeding duration is how weak they are compared with many other developing countries (Trussell and others, 1992). For all categories of children, the median duration of breast-feeding is well over a year and breast-feeding for at least half a year is almost universal.

	Median duration (months) as estimated from:		Life-table estimates of percentage breast-fed at least:
	Current status	Life table	6 months	12 months	18 months
Total sample	16.9	15.9	95	80	36
Residence
Urban	13.9	13.9	91	72	18
Rural	17.5	16.3	96	82	40
Sex of child
Male	16.9	16.0	94	80	37
Female	16.7	15.5	95	79	34
Education of mother
No schooling	19.2	17.4	93	79	47
Grades 1-4	17.5	15.4	93	77	36
Grades 5-7	16.8	15.4	94	79	34
Grades 8-11	17.3	16.2	96	82	37
Grades 12+	15.4	14.8	94	79	25
Occupation of mother
Agricultural	17.5	16.5	95	81	41
White collar	16.0	15.0	93	78	28
Other working	17.1	14.5	92	74	26
Not working	15.6	14.7	94	78	25
Wealth status by residence
Urban low	14.5	14.3	90	74	20
Urban middle	13.2	13.6	93	70	16
Urban high	13.1	13.8	90	70	19
Rural low	17.9	16.2	94	79	41
Rural middle	17.7	16.5	96	82	40
Rural high	15.8	15.0	95	81	35

Notes: Results are weighted. Current status results refer to all living children born within three years of the survey. Life-table estimates refer to all live births during the same period.

Supplemental food for breast-fed children

Most health specialists not only advocate breast-feeding but also recommend that, during the first four to six months of life, the child be limited exclusively to breast milk. Neither liquids, including plain water, nor solid or mushy food are thought to be beneficial for the infant at least for months (Institute for International Studies in Natural Family Planning, 1990). The recent breast-feeding promotion campaign in Vietn Nam also recommends that exclusive breast-feeding be practised for the first four months and than no solid food be given before six months.

The 1994 VNICDS collected several types of information on food and liquids other than breast milk that were given to children. Mothers who were currently breast-feeding a child were asked whether or not the child was given various types of supplemental foods and liquids during the day or night before the survey. In addition, for each of the last three children born within 10 years of the survey, respondents were asked if the child was ever given supplemental food or liquids and, if so, at what age foods or liquids were first given regularly.

The percentage of currently breast-fed children who were given different types of supplemental foods during the prior day or night are shown in table 4. Since most children are weaned by the age of 24 months, results are shown only for children under two years old. Although the VNICDS questionnaire included a separate question on infant formula and tinned or powdered milk, it is likely that many respondents were unfamiliar with the technical term used to inquire about formula and did not distinguish powdered milk from formula. Thus, in the presentation of results we have combined the responses to these two items. For convenience, we refer to the combined category of tinned/powdered milk and formula as "other milk".

Plain water has no nutritional value and thus in this sense is not truly a supplemental food. Nevertheless, its provision is of concern as a potential source of contaminants (Khan, 1990). The results make clear than it is very common to give plain water to infants at very young ages. Overall, more than two-thirds of breast-fed infants under three months old were given plain water and over 90 per cent of those 3-5 months old had received plain water. Children born to mothers living in urban areas appear to have been given plain water earlier than those born to rural mothers.

	Age of child (in months)
	0-2	3-5	6-11	12-17	18-23	Total
Plain water
Total sample	68	91	95	95	97	91
Urban	(93)	(90)	99	99	(97)	97
Rural	63	91	94	94	97	90
Sugar water
Total sample	16	18	20	25	18	20
Urban	(17)	(18)	18	28	(23)	21
Rural	16	18	20	24	17	20
Juice
Total sample	5	8	20	22	19	17
Urban	(5)	(27)	45	53	(77)	41
Rural	5	6	16	18	14	14
Fresh milk
Total sample	0	1	1	1	0	1
Urban	(0)	(3)	4	3	(0)	3
Rural	0	0	0	1	0	1
Other milk (tinned or powdered milk/formula)
Total sample	14	19	19	16	12	17
Urban	(42)	(36)	43	32	(51)	40
Rural	9	16	15	14	9	14
Other liquid
Total sample	3	5	11	11	14	10
Urban	(1)	(13)	26	30	(38)	22
Rural	3	4	9	9	12	8
Solid/mushy food
Total sample	15	63	90	92	96	78
Urban	(11)	(58)	96	98	(100)	78
Rural	16	64	89	92	95	78
Any liquid/solid food
Total sample	73	94	99	98	99	95
Urban	(93)	(96)	100	100	(100)	98
Rural	70	94	99	98	98	95
Any liquid other than plain water
Total sample	29	36	44	44	37	40
Urban	(51)	(56)	67	69	(79)	64
Rural	25	33	40	41	33	37
Any liquid or food other than plain water
Total sample	37	72	93	96	97	85
Urban	(55)	(73)	99	98	(100)	88
Rural	33	72	92	96	97	84

Notes: Results are weighted. Results in parentheses are based on fewer than 50 weighted births.

Other supplemental foods and liquids not only risk being contaminated but, because they have some nutritional content, also will reduce the child's intake of breast-milk. All other liquids asked about in the VNICDS were given far less commonly than plain water. Least common of all was fresh (cow's) milk which is almost completely absent in Viet Nam as a supplement to breast-feeding. Sugar water was reported to be the most common, although other milk and juice were given almost as commonly, judging by the proportion of all breast-fed children under age two years who received them. However, the ages at which children received these three types of liquids differ. In particular, juice was largely limited to children at least six months old while sugar water and other milk were given in fair proportions even to younger infants. There is relatively little difference in the proportion of urban and rural children who received sugar water. However, both juice and other milk were far more common supplemental liquids in urban than in rural areas.

In Viet Nam young children are traditionally given solid/mushy food, mostly rice products. Although only 15 per cent of breast-fed infants aged 0-2 months were given solid/mushy food, the majority of those 3-5 months old received such food as did 90 per cent or more of older breast-fed children. The results also suggest that solid or mushy food is given earlier to rural than urban infants.

If all liquids and foods are considered, a very substantial proportion (73 per cent) of even the youngest breast-fed infants received at least something in addition to breast milk. Thus, only a minority of infants in the first few months of life can be considered as truly exclusively breast-fed. If plain water is excluded, however, a somewhat different picture emerges with most breast-fed infants under the age of three months having received no liquids or food with nutritional content other than breast milk. By 3-5 months the situation reverses with most receiving some form of supplement to the breast milk. The situation differs for urban and rural children with supplements having nutritional content being far more common for the youngest urban than for the youngest rural infants. The difference between urban and rural children in this respect disappears within a few months after birth. By the time breast-fed children are in the second half of their first year of life, almost all receive some supplemental food having nutritional content.

Based on responses as to whether and when particular categories of food were first regularly given to a child, the life-table approach can be utilized to estimate several features of infant feeding practices. These estimates refer to all children regardless of whether or not they were still being breast-fed at the time the other liquids or food were introduced. The results, presented in table 5, indicate that few infants in their first month of life were given solid or mushy food but that the practice of doing so increased rapidly thereafter. Overall, by four months 50 per cent of children were already given solid/mushy food regularly. Such food is introduced into the diet of rural infants somewhat earlier than urban infants as reflected by the earlier rural median age. By age six months, however, the urban-rural difference has disappeared and by nine months over 90 per cent of children, regardless of residence, receive solid or mushy food. For both urban and rural children, half receive food or liquids other than plain water between ages three and four months. The lack of an urban-rural difference in this respect reflects the earlier introduction of liquids in urban areas (thus compensating for the earlier introduction of solid or mushy food in rural areas).

	Percentage who regularly recieved		Percentage breast-fed exclusive of:
	Solid/mushy food	Food or liquids excluding water	Solid/mushy food	Food or liquids excluding water
Total sample
By age 1 mo.	4.7	14.5	91.8	83.1
By age 2 mo.	14.1	26.1	82.2	71.5
By age 3 mo.	26.8	41.4	70.0	56.5
By age 6 mo.	77.1	84.5	21.7	14.8
By age 9 mo.	93.3	95.2	6.3	4.5
By age 12 mo.	95.8	96.	9 3.9	2.9
Median (in months)	4.0	3.4	3.9	3.3
Urban
By age 1 mo.	0.8	16.3	94.6	82.5
By age 2 mo.	4.4	30.0	90.1	73.0
By age 3 mo.	13.1	42.1	80.9	57.0
By age 6 mo.	78.3	90.4	19.9	9.5
By age 9 mo.	96.5	98.8	3.2	0.9
By age 12 mo.	98.1	99.2	1.7	0.2
Median age (in months)	4.0	3.3	4.2	3.3
Rural
By age 1 mo.	5.4	14.1	91.3	83.2
By age 2 mo.	15.8	26.1	80.8	71.3
By age 3 mo.	29.2	41.2	67.6	56.4
By age 6 mo.	76.8	83.5	22.0	15.8
By age 9 mo.	92.7	94.6	6.8	5.1
By age 12 mo.	95.4	96.5	4.3	3.3
Median age (in months)	3.3	3.4	3.8	3.3

Note: Results are weighted and refer to all births during five- year period prior to the survey.

The duration of breast-feeding exclusive of solid or mushy food and liquids other than plain water is closely related to when these foods are introduced into the diet. As the results show, exclusive breast-feeding is relatively short in Viet Nam. Half of the children were breast-fed exclusive of solid or mushy food for about four months and exclusive of any food or liquid other than plain water for only a little over three months. Although breast-feeding exclusive of solid or mushy food was slightly longer in urban ares, there is no urban-rural difference in the median duration of breast-feeding exclusive of food and liquids.

Breast-feeding, amenorrhoea and contraceptive use

The link between infant feeding practices and the return of menstruation and ovulation following childbirth is well known. Continued breast-feeding tends to prolong the period of post-partum amenorrhoea and with it the return of ovulation. The 1994 ICDS is the first national-level survey in Viet Nam to have collected information on post-partum amenorrhoea. Table 6 presents current-status estimates of the percentages of women still amenorrhoeic. The median duration of amenorrhoea at the national level is almost nine months following childbirth. The steady decline in the percentage of women still amenorrhoeic reflects the increasing proportions of women who experience the return of menstruation as the time since childbirth increases.

The duration of amenorrhoea varies significantly among different groups of mothers in Viet Nam. In general, the differentials in the duration of amenorrhoea parallel those indicated for breast-feeding. Thus, urban women experience more rapid return of menstruation than do rural women. Also, the least educated women are characterized by a longer time to the return of menstruation than the best educated ones, and women with agricultural jobs experience a longer period of amenorrhoea than women not in agriculture.

As figure 2 shows, the return of menstruation is not only closely associated with whether or not a woman breast-feeds but also with whether she supplements breast-feeding with other food. For women not breast-feeding, the return of menstruation is very rapid, with the median duration implied by the current status estimates being only slightly over two months. However, the results also show that women who supplement their breast-feeding with solid or mushy food for their child are more likely to experience the return of menstruation at the various successive months since giving birth than those who do not. This result is consistent with what is known about the physiology of the link between breast-feeding and amenorrhoea. Ovulation (and hence menstruation) tends to be suppressed by the release of prolactin which is stimulated by the intensity and frequency of suckling by the child. When children who breast-feed are also given supplemental food, they tend to suckle less frequently and intensely because their appetite for breast milk is partially met by eating the other food. (McCann and others, 1981).

	Months since childbirth
	0-2	3-5	6-11	12-17	Median
Total sample	93	74	53	25	8.9
Residence
Urban	89	57	38	10	5.6
Rural	94	77	55	28	9.8
Education of mother
Less than 5 years	93	82	59	35	9.0
Grades 5-7	97	71	51	22	8.6
Grades 8-11	96	70	51	26	8.7
Grades 12+	76	59	42	8	6.8
Occupation of mother
Agricultural	94	77	58	30	10.3
Other working	82	60	41	16	7.1
Not working	97	76	46	19	8.0

Notes: Results are weighted and refer to experience with the most recent birth. Medians are determined from five-month, equal-weight moving averages of single-month values. Results in parentheses are based on fewer than 50 weighted cases.

Given the clear health benefits of breast-feeding for children, concern about declines in the practice in numerous developing countries during the last several decades has been widespread among health professionals (McCann and others, 1981; Millman, 1986; Trussell and others, 1992). Thus, the results of the present study indicating that breast-feeding is virtually universal in Viet Nam, that the average duration of breast-feeding is well over a year, and that there appears to be no indication of a decline during the last decade in either breast-feeding initiation or duration are quite welcome. Moreover, the nearly universal initiation of breast-feeding and reasonably long durations, i.e. in excess of a year on average, are characteristic of a wide spectrum of socio-economic groupings.

Viet Nam is not alone in having a tradition of universal and long breast-feeding. Indeed most Asian countries, at least until very recently, followed similar practices (see March 1990 issue [vol.5, No. 1] of this Journal, which contains nine articles on breast-feeding in various countries and areas of Asia). There are several likely reasons why in Viet Nam there has so far been no sign of the erosion of breast-feeding as appears to have occurred in some other national settings.

First, during most of the last two decades, Viet Nam has been relatively closed to international commercial and other influences outside the Socialist Block. This relative isolation conditioned social and economic change in a rather different way than was taking place elsewhere in the non-socialist developing world and this phenomenon held important implications for consumer tastes and attitudes. Undoubtedly of particular relevance was the absence of commercial activities by multinational corporations marketing and promoting breast-milk substitutes, especially infant formula. During this same period, domestic production of infant formula was very limited and not promoted commercially. Only in the last few years have multinational companies come on the scene with active marketing of formula and other potential substitute infant food products. Thus, consumption of such products has so far been quite low.

Second, most of the population live in the countryside and engage in agricultural occupations. The rural life-style they follow has been conducive so far to maintaining the traditionally universal initiation and long duration of breast-feeding.

A third and related factor is the prevalent low standard of living. This acts as an important barrier to the consumption of non-breast-milk products because most families are unable to afford to buy commercially made breast-milk substitutes.

Government policies and programmes have also been supportive of extended breast-feeding and may be contributing to its continued prevalence. These may be of interest to other developing countries when formulating policies and implementing programmes for this purpose. For example, various programmes of Viet Nam's Ministry of Health have included information both for staff and the public intended to promote a better understanding of the value of breast-feeding and appropriate timing for the introduction of supplementary foods. Public "mass" organizations such as the Women's Union, Youth Union, and Voluntary Health Workers have also participated in these programmes. Another government policy that probably helps to sustain breast-feeding, especially among the urban non-agricultural population, are labour regulations regarding maternal leave from work. Since 1985, the duration of maternal leave in the government sector was extended from 10 weeks to 4-6 months, depending on the particular working conditions. The explicit purpose of this regulation is to facilitate full breast-feeding for at least the first few months of a child's life. Many state enterprises have crèches located on the premises or nearby to facilitate breast-feeding after the mothers return to work. In addition, those mothers who return to work are entitled to take off one hour daily with no penalty to their pay until the child reaches the age of 12 months; this time can be taken off at any time during the work day.

There are also some aspects of infant feeding in Viet Nam that are less than ideal. Health professionals typically recommend that women start to breast-feed their child during the first hour following childbirth. However, only a minority of mothers apparently follow this practice. Interestingly, our analysis indicates that the delivery of a baby at a government health facility is not associated with an increased chance of early breast-feeding initiation. Supplemental foods and liquids, including plain water, are introduced to infants by many mothers before the time considered by health professionals to be ideal. Government efforts to promote early breast-feeding and to encourage the appropriate timing of the introduction of supplementary food apparently have yet to have much impact on changing less-than-ideal traditional practices.

There are reasons to suspect that the social and economic change now under way in Viet Nam, if not effectively counteracted, may weaken breast-feeding. Free market reforms (Doi Moi) implemented from 1986 and a considerable opening of the country to broader international commerce are likely to exert negative pressures on breast-feeding. Infant formula and powdered milk, including major brands from multinational companies as well as a domestic brand, are far more available than previously. Furthermore, even though commercial promotion and advertisements for food substitutes through the mass media and in hospitals and health care centres are officially banned, there is anecdotal evidence that formula companies evade the regulations. For example, in Ho Chi Minh City, company agents obtain lists of new mothers from ward health stations and contact the mothers directly, providing free samples or selling formula or milk products initially at below market prices. The official ban against selling formula and similar products in hospitals is also not vigorously enforced. Even when it is, shops right outside the hospital gate openly sell these products that, in the big cities at least, are becoming a fashionable gift for visitors to bring to new mothers.

Besides increased commercial availability and promotion of breast-milk substitutes, other social changes may contribute to a decline in breast-feeding. Living in urban areas, working outside of agriculture, having higher levels of education, and being wealthier are all associated with shorter duration of breast-feeding. Given that socio-economic change in Viet Nam is operating to increase the relative share of women in each of these categories, such compositional changes would serve to modestly reduce the average breast-feeding duration in the future. Moreover, economic growth is likely to lead to improved living standards which in turn will make store-bought formula and powdered milk more affordable to an increasing number of mothers. The fact that many of the infant formula products are from foreign companies adds to their image as a desirable good and to their appeal to consumers with increased disposable income.

It is probably true that the long duration of breast-feeding is currently still largely an outcome of a long tradition in the infant feeding practices of Vietnamese mothers. As a normative practice, its persistence probably reflects traditional thinking more than conscious decisions based on knowledge of the benefits of breast-feeding. Thus, these practices may be quite vulnerable to change as Viet Nam follows the path to rapid socio-economic development.

The Ministry of Health already recognizes many of the potential threats to breast-feeding and has in 1994 launched a new national-level programme designed to promote favourable practices, including the issuance of regulations to control the production, circulation and advertisement of food substitutes.3 Fortunately these regulations and programmes have been issued while the traditional pattern of universal and prolonged breast-feeding still persists. Continued monitoring of the situation and sustained efforts to promote healthful infant feeding practices in a coordinated programme by government and non-governmental organizations and health-care institutions are likely to be required to maintain and improve the currently favourable situation.

1. In the present analysis, children who died but were breast-fed until the time of death are included in the life-table calculations but are treated as "censored events" rather than as "terminal events" for which the age at weaning is known (see Norusis, 1992).

2. The wealth index was constructed based on a combination of information on selected household possessions and quality of the house.

3. Directive of the Ministry of Health 8257/PC, Hanoi, 18 December 1992: decision of the Government of Viet Nam, No. 307/TTg, Hanoi, 10 June 1994.

General Statistics Office of Viet Nam (1995). Viet Nam Intercensal Demographic Survey 1994: Major Findings (Hanoi: Statistical Publishing House).

Institute for International Studies in Natural Family Planning (1990). Guidelines for Breastfeeding in Family Planning and Child Survival (Washington, D.C.: Georgetown University).

Khan, M.E. (1990). "Breast-feeding and weaning practices in India", Asia-Pacific Population Journal 5(1):71-88.

Khoi, Ha Huy and Tu Giay (1994). Malnutrition and Community Health in Vietnam (Hanoi: Medical Publishing House).

Lesthaeghe, R. and H.J. Page (1980). "The postpartum non-susceptible period: development and application of Aodel Schedules", Population Studies 34(1):143-169.

McCann, M.F., L.S. Liskin, P.T. Piotrow, W. Rinehard and G. Fox (1981). "Breastfeeding, fertility and family planning", Population Reports, J-24, Baltimore, Maryland: Population Information Program, Johns Hopkins University.

Millman, Sara (1986). "Trends in breastfeeding in a dozen developing countries", International Family Planning Perspectives, 12(3):91-95.

Norusis, Marija (1992). SPSS/PC+ Advanced Statistics Version 5.0, Chicago: SPSS.

Page, H.J., R. Lesthaeghe and I.H. Shah (1982). "Illustrative analysis: breastfeeding in Pakistan", WFS Scientific Reports, No. 37 (Voorburg: International Statistical Institute).

Trussell, James, Laurence Grummer-Strawn, German Rodriguez and Mark VanLandingham (1992). "Trends and differentials in breastfeeding behavior: evidence from WFS and DHS", Population Studies 46(2):285-307.

VanLandingham, Mark, James Trussell and Laurence Grummer-Strawn (1991). "Contraceptive and health benefits of breastfeeding: a review of the recent evidence", International Family Planning Perspectives 17(4):131-136.

For further information on this material please contact: loftus.unescap@un.org