Breast-feeding has been the subject of rapidly growing interest in developing countries because of its important implications not only for the improved health of children, but also for lowering fertility. The suckling infant stimulates the flow of hormones within the mother that delay the return of ovulation. Extended breast-feeding lengthens the period of non-exposure to the risk of conception and thus lengthens the interval between consecutive births, which in turn indirectly reduces fertility (Huffman, 1984; Thapa and Williamson, 1990). Since breast-feeding has no direct impact on fertility, an attempt is made in this article to examine the association between breast-feeding and post-partum amenorrhoea as a proxy for the fertility-inhibiting effect of breast-feeding in Bangladesh.

Bangladesh is considered one of the world's most densely populated countries (836 persons per square km) with an annual population growth rate of 2.2 per cent (ESCAP, 1995). In such a setting, the pressure on the land for agricultural production and the demand for jobs result in large-scale migration to cities and industrial zones. Thus, rapid changes have been occurring in the country's socio-economic and demographic characteristics. Against this backdrop, it is reasonable to expect that the levels and patterns of breast-feeding will have been changing over time. It is in this context that the present study has been undertaken. It should be mentioned that this study is the first one detailing breast-feeding differentials in Bangladesh by the life-table method, using national-level data to monitor the changes taking place over time.

____________

* The authors of this article are Haider Rashid Mannan, Lecturer in Statistics, and M. Nurul Islam, Professor of Statistics, both of the Department of Statistics, Dhaka University, Dhaka, Bangladesh. It is based on the first author's Master of Science thesis which was carried out under the supervision of Professor Islam. They both would like to acknowledge with gratitude the useful suggestions made by Professor S.A. Mallick and Dr. Ataharul Islam of Dhaka University, and Professor M. Kabir of Jahangirnagar University, Bangladesh.

This study utilizes data extracted from the 1989 Bangladesh Fertility Survey (BFS) which was conducted on behalf of the Government of Bangladesh by the National Institute of Population Research and Training (NIPORT). Information was collected from a nationally representative sample of 11,906 ever-married women under 50 years of age. Information on the duration of breast-feeding was collected for ever-married women with at least one live birth within the six years preceding the interview. Questions on breast-feeding duration were related to the last four births of the women. A total of 7,516 women provided information on breast-feeding of their last child: of this number, 198 (2.6 per cent) reported that they never breast-fed their most recently born child, 2,050 (27.3 per cent) said that they had already weaned the child by the date of the interview, 381 (5.1 per cent) breast-fed the child until its death and the remaining 4,887 (65.0 per cent) were still breast-feeding their child.

Retrospective information collected in any survey in most developing countries is affected by memory lapse and a preference for certain numbers. Retrospective data on durations of breast-feeding and amenorrhoea show heaping at multiples of six months. Srinivasan (1980) suggested quotients for measuring digital preference from retrospective data on breast-feeding in both the current open and last closed intervals. These were found to be 0.79 and 0.55, respectively, when the quotient ranges from 0 to 1; the value 0 is used when there is absolutely no digital preference and the value 1 is used when all the frequencies are in multiples of six months. However, in a predominantly Muslim country such as Bangladesh -- especially in rural areas -- it is possible that women continue breast-feeding for 24 months as is a common practice in other Muslim countries (Page and others, 1982; Rabei, 1988). However, current status data on breast-feeding do not show such heaping at multiples of six months.

Women reporting at the last closed interval represent, at least partially, a self-selected group of short-duration breast-feeders and thus add selection bias to the recall bias. Data for the last closed interval under-represent women with a longer-than-average birth interval, because these women have had less chance of having had their second birth. Data for the current open interval do not suffer from this limitation. These biases are minimized in the current status analysis of breast-feeding in the open birth interval (Page and others, 1982).

The methods available for the estimation of breast-feeding patterns differ in their data requirements as well as with regard to the objectives of the study. The methods also differ slightly in their assumptions. Thus, the choice between them will depend in part on the availability and quality of different types of data. In order to evaluate the characteristics of the frequency distribution such as means, median, standard deviation, differences in the survival distributions or the interquartile range, one has to use the life-table method of estimation (Page and others, 1982). If one is interested only in the mean, then one may use the very simple method of prevalence/incidence. But this method does not take full advantage of the data typically available. In contrast, the life-table method permits full utilization of periods of observation of censored and non-censored cases. Moreover, results from the prevalence/incidence method depend on the assumption that the number of births per month has been constant throughout "Z" years preceding the survey when Z is the longest duration of breast-feeding in the population (Mosley and others, 1982). The assumption of a constant flow of births per month may hold true as a broad characteristic, but it is unlikely to be so when a broad group is subdivided into smaller subgroups. The life-table method of analysis, though quite complex, does not suffer from such limitations. Thus, the life-table method is preferred over other analytical techniques such as prevalence/incidence or current status methods (Thapa and Williamson, 1990).

In life-table analysis, the breast-feeding status of mothers whose children subsequently died and yet are still breast-feeding another child or children are considered as censored cases. Inclusion of data on the duration of breast-feeding, disregarding the survival status of children, may introduce a bias because women whose last child had died during the previous 30 months would obviously be unable to breast-feed the child following its death and thus would have an artificially shortened duration of breast-feeding. For this reason, data on children who died during this interval (30 months preceding the interview) were excluded from our study (Weis, 1993).

If two or more groups in the study can be considered as samples from some large population, then we may want to test the null hypothesis that the survival distributions are the same for the subgroups, thus establishing the statistical significance of the differences. The statistics available in survival analysis are calculated according to the algorithm of Lee and Desu (1972).

The results of the actuarial life-table method (Cutler and Ederer, 1958) are presented in table 1. Studies on a few characteristics of breast-feeding differentials using World Fertility Survey (WFS) and 1976 BFS data were performed by Ferry and Smith (1983) and Ahamed (1986), respectively. Two other studies of limited scope were carried out using data from rural areas of Bangladesh by Huffman and others (1980) and Nessa and others (1987). However, Huffman and his colleagues (1980) reported median values only. The results of the present study are compared with those from previous studies whenever these are available. This comparison will indicate any change in breast-feeding patterns in Bangladesh for different subgroups, although the values of the means may vary owing to the differences in the sources of data used and techniques employed (Ferry and Smith, 1983).

The average duration of breast-feeding in Bangladesh for surviving children is 28.2 months. While Ferry and Smith (1983) found duration to be 28.9 months, Ahamed (1986) using the prevalence/incidence method found it to be 27.3 months and Nessa and others (1987) found it to be 26.4 months for all children. Huffman and others (1980) found the median duration to be 32.0 months for surviving children.

As may be expected, urban women were found to breast-feed for a relatively shorter duration, i.e. 27.1 months, compared with rural women who breast-feed for 28.6 months on average. Ferry and Smith (1983) found these durations to be 26.0 and 29.2 months for urban and rural women, respectively; Ahamed (1986) found that the average durations were 24.0 and 27.6 months, respectively. Among the four administrative divisions of Bangladesh, the average duration of breast-feeding has been lowest in Chittagong division (27.4 months) and highest in Rajshahi division (28.9 months), while it is 28.0 and 28.8 months, respectively, in Dhaka and Khulna divisions.

There is a clear indication that breast-feeding duration is negatively associated with the educational level of the mother. Respondents having no schooling were found to breast-feed on average for 28.8 months, which linearly decreases to 26.2 months for mothers with a higher education. Ferry and Smith (1983) found these durations to be 29.6 and 27.1 months for women having no schooling and schooling, respectively. Ahamed (1986), Huffman and others (1980) and Nessa and others (1987) also found similar inverse associations. Husband's education also demonstrates an inverse relationship with the duration of breast-feeding. The study shows a decrease in the mean duration of breast-feeding from 28.9 months for women having husbands with no schooling to 27.2 months for women having husbands with a higher education.

Mother's age at the birth of the index child exhibits a positive association with the duration of breast-feeding. Only older women (aged 35-49 years) have a different breast-feeding pattern than the others. Ferry and Smith (1983) found that the average durations increased from 26.7 to 31.5 months for mothers aged 15-24 and 35-49 years, respectively; Ahamed (1986) found that the average duration increased from 24.7 months for mothers aged 15-19 years to 32.5 months for those aged 45-49 years. Ferry and Smith (1983), Huffman and others (1980) and Nessa and others (1987) also found similar positive associations between breast-feeding duration and mother's age.

The differential in breast-feeding duration by religion shows that Muslim women breast-feed on the average for 28.1 months while their non-Muslim peers breast-feed for 28.8 months. Huffman and others (1980) also found a similar pattern.

Currently working women breast-feed for a slightly longer duration (29.0 months) as compared with their non-working counterparts who breast-feed for 28.1 months. Ahamed (1986) found that these durations were 28.5 and 28.1 months for working and non-working women, respectively. The reason for this may be that, since most working women in Bangladesh perform physical or manual labour, it is possible that they take their babies with them to their place of work. Again, since the majority of such working women perform physical or manual work, it is reasonable to assume that they are mostly less educated or uneducated.

Parity demonstrates a positive and linear association with duration of breast-feeding. Women with a parity of 1-2 breast-feed for an average of 27.5 months, which rises linearly to 28.6 months for those who were reported to have at least five children. Ferry and Smith (1983) found that the average durations increased from 26.7 to 31.4 months for mothers with parities 1-2 and 5 or more, respectively. Ahamed (1986) and Huffman and his colleagues (1980) also observed such an association, which may be expected, since women of higher parity are also those who are likely to be older; younger women may be expected to have a higher level of education and be more likely to break with traditional behaviour patterns than older and less well educated women.

The differences in breast-feeding patterns according to the occupational status of husbands suggest that the wives of service workers and businessmen breast-feed on average for 27.5 months, wives of production workers breast-feed for 28.6 months, while those of farmers and land-owners breast-feed on average for 29.0 and 27.9 months, respectively.

Current use of contraception shows a very weak positive impact on the duration of breast-feeding. Women currently using contraception breast-feed on average for 28.3 months, whereas those not currently using contraception breast-feed for 28.1 months, so the difference is only marginal.

Possession of household items appears to have a negative impact on the duration of breast-feeding. Respondents possessing specific household items breast-feed on average for a shorter duration (27.2 months) compared with those not possessing such items (29.2 months). This means that mothers belonging to lower socio-economic groups and having a more traditional life-style breast-feed their babies longer than their more affluent counterparts.

Visits by health workers have a positive impact on the duration of breast-feeding. Women are more likely to breast-feed for a longer duration (28.8 months) when health workers have visited them than those who are not visited at all by health workers (28.0 months). Examination of the effect of health decision-making on the duration of breast-feeding shows that women whose husbands make health decisions breast-feed for an average of 28.3 months while women who make health decisions by themselves or jointly with their husbands breast-feed for 28.1 months; however, this difference is only marginal.

BFS 1989 data provide no evidence of a marked differential in breast-feeding duration by sex of the breast-fed child. Male children were found to be breast-fed for an average duration of 28.3 months while female children were breast-fed for 28.1 months, but the difference is only marginal. Ahamed (1986) also found that male children were breast-fed for a longer duration than female children.

Comparison of survival distributions

So far the results of breast-feeding patterns have been discussed only on the basis of the differences in the mean values of certain characteristics. A complementary but statistically more powerful method of comparison is the use of the chi-square test for the survival distributions of the subgroups of a broad category (Lee and Desu, 1972). The differences in the survival distributions for the different subgroups of a variable based on the chi-square test are given in table 1.

Characteristic	Means (in months)
Bangladesh	28.2
Residence
Rural	28.6
Urban	27.1
	p<0.001
Administrative divisions
Chittagong	27.4
Dhaka	28.0
Khulna	28.8
Rajshahi	28.9
	p<0.001
Mother's age at birth of index child
15-24	28.1
25-34	28.1
35-49	29.3
	p<0.01
Parity
1-2	27.5
3	27.7
4	28.1
5+	28.6
	p<0.01
Mother's education
No education	28.8
Lower primary	28.3
Upper primary	27.6
Higher	26.2
	p<0.001
Husband's education
No education	28.9
Lower primary	28.3
Upper primary	27.8
Higher	27.2
	p<0.001
Work status
Working	29.0
Not working	28.1
	p<0.01
Husband's occupation
Sales/service	27.5
Production workers	28.6
Labourer/farmers	29.0
Land-owners	27.9
	p<0.001
Religion
Muslim	28.1
Non-Muslim	28.8
	p>0.05
Sex of the index child
Male	28.3
Female	28.1
	p>0.05
Current use of contraception
Yes	28.3
No	28.1
	p>0.05
Visit of health workers
Never	28.0
Yes	28.8
	p<0.001
Health decision
Respondent alone/jointly	28.1
With husband	28.3
	p>0.05
Household possessions
No	29.2
Yes	27.2
	p<0.05

The results show that the differences between survival distributions are highly significant (p<0.001) for the following variables: place of residence, administrative division, mother's education, husband's education, husband's occupation and visits by health workers. The differences in the variables mother's age, parity, mother's current work status (p<0.01) and household possessions (p<0.05) are also significant.

Before the introduction of modern contraceptives, breast-feeding was the main factor determining the interval between pregnancies in developing countries (Kleiman and Senanayake, 1984). In this study an attempt has been made to examine the association between breast-feeding and fertility in the country as a whole and in rural and urban areas. Because breast-feeding has no direct impact on fertility, more emphasis should be given to the effects of the breast-feeding process on the period of post-partum amenorrhoea as a proxy for the fertility-inhibiting effect of breast-feeding. The temporary cessation of ovulation, accompanied by cessation of menstruation after every birth, can be prolonged as a result of breast-feeding (Thapa and Williamson, 1990). Post-partum amenorrhoea lasts for an average of about two months in non-breastfeeding women and increases to roughly 60-70 per cent of the average duration of breast-feeding in a population customarily practising breast-feeding (Leridon, 1977). Thus, breast-feeding affects fertility by prolonging the period of post-partum amenorrhoea, which in turn affects birth intervals. Accordingly, a logical starting point is to study the relationship between breast-feeding and post-partum amenorrhoea. This study is designed to investigate such a relationship, taking into account some of the socio-economic and demographic variables of respondents such as the mother's age, her education, place of residence and parity. It also attempts to evaluate the contraceptive potential of breast-feeding in Bangladesh.

Bivariate analysis

Bivariate analysis (tables 2, 3 and 4) shows that an increase in breast-feeding duration prolongs the length of post-partum amenorrhoea. A longer duration of breast-feeding by women in rural areas leads to longer periods of amenorrhoea compared with those experienced by women in urban areas. The mean duration of amenorrhoea is positively associated with parity, but variations in amenorrhoea by parity are marginal as compared with variations by breast-feeding duration. Urban-rural differentials are more pronounced when breast-feeding durations are controlled. The mean length of post-partum amenorrhoea is positively associated with mother's age while the impact of mother's education on the mean length of amenorrhoea is more pronounced in urban areas than in rural areas.

Parity	Place of residence	Breast-feeding duration (in months)			Overall means
		0-11	12-23	24+
1	Rural	3.6	7.4	11.0	9.0
	Urban	3.9	5.3	9.8	6.6
	Total	3.8	6.6	10.6	8.1
2	Rural	7.6	10.8	12.8	11.8
	Urban	4.9	8.8	11.3	9.3
	Total	5.7	10.1	12.2	10.6
3	Rural	8.2	9.8	13.8	12.3
	Urban	4.9	8.4	10.0	8.3
	Total	6.1	9.2	12.8	10.9
4+	Rural	8.6	10.9	14.8	13.6
	Urban	6.1	10.6	13.0	11.5
	Total	7.5	10.8	14.3	13.0
Overall means	Rural	7.4	10.1	13.8	12.5
	Urban	4.8	8.6	11.7	9.4
	Total	5.6	9.6	13.2	11.4

Mother's age	Residence	Breast-feeding duration (in months)			Overall means
		0.11	12-23	24+
15-24	Rural	4.5	9.0	11.9	10.2
	Urban	3.9	6.4	11.5	7.8
	Total	4.1	8.2	11.7	9.2
25-34	Rural	8.9	10.6	13.5	12.8
	Urban	5.7	9.4	10.8	9.4
	Total	6.8	10.1	12.7	11.4
35-49	Rural	9.2	11.1	15.8	14.7
	Urban	4.5	11.6	14.3	12.7
	Total	7.8	11.2	15.5	14.2

Mother's education		Breast-feeding duration (in months)			Overall means
		0-11	12-23	24+
No schooling	Rural	8.5	10.2	14.5	13.2
	Urban	8.2	10.0	13.7	12.4
	Total	8.4	10.2	14.3	13.0
Schooling	Rural	6.0	9.8	12.5	11.2
	Urban	4.1	7.9	10.0	7.6
	Total	4.5	8.9	11.5	9.5

Multivariate analysis

The bivariate analysis in our study establishes a relationship between breast-feeding and post-partum amenorrhoea after controlling for one or two variables. So it is important to examine the association between breast-feeding and post-partum amenorrhoea while statistically controlling for the effects of all other variables that might influence the post-partum amenorrhoea period. This relationship can be expressed by using multiple regression analysis, taking the post-partum amenorrhoea period as a dependent variable and duration of breast-feeding as the independent variable along with place of residence, mother's education, mother's age, parity, work status of the mother and ever use of contraception as other independent variables. In order to obtain an accurate picture of the examined relationship between breast-feeding and post-partum amenorrhoea, the regression analysis is carried out only for those who stopped breast-feeding. The analysis has been performed over both the current open and last closed birth intervals of post-partum amenorrhoea to determine whether breast-feeding is the principal determinant of post-partum amenorrhoea in both intervals. The closed amenorrhoea interval reflects the complete experience of amenorrhoeic mothers, whereas the current open interval does not give the complete experience of amenorrhoeic mothers owing to the exclusion of censored cases from the analysis. Table 5 shows the values of the coefficient of determination (R2) and partial regression coefficient (B) in the two models. The variables are included in the models by a step-wise regression method.

The results show that breast-feeding has a positive significant effect on amenorrhoea and is the principal determinant of amenorrhoea in both the current open and last closed birth intervals. However, the effect of breast-feeding on the post-partum amenorrhoea period is stronger in the last closed interval than in the current open interval. The breast-feeding duration explains 18.9 per cent of the total variation in post-partum amenorrhoea, which is 86.3 per cent of the total variation explained by the seven variables: breast-feeding, place of residence, mother's education, mother's age, parity, mother's work status and ever use of contraception in the last closed interval. Breast-feeding explains 9.8 per cent of the total variation in the post-partum amenorrhoea period in the current open interval, which is 71.5 per cent of the total variation explained by all the variables, excluding parity. The partial regression coefficient (B) for breast-feeding indicates that, on average, one month of breast-feeding adds about 0.36 month to the period of post-partum amenorrhoea in the last closed interval, whereas this value is about 0.2 month in the current open interval. The other variables such as place of residence, mother's education, mother's work status, mother's age and ever use of contraception were found to be significant in both models, although to a much lesser extent than breast-feeding. Parity was found to have a positive significant effect on the length of amenorrhoea only in the last closed interval; it was found to be insignificant in the other model.

Variables	Current open interval		Last closed interval
	R2	B	R2	B
Duration of breast-feeding	0.098	0.199*	0.189	0.358*
Mother's age	0.018	0.155*	0.007	0.157*
Education of mother
(No schooling)	-	-	-	-
Schooling	0.012	-1.467*	0.005	-1.207*
Ever use of contraception
(No)	-	-	-	-
Yes	0.004	-1.199*	0.013	-1.374*
Residence
(Rural)	-	-	-	-
Urban	0.003	-1.248*	0.002	-1.044*
Work status of respondents
(Never worked)	-	-	-	-
Ever worked	0.002	1.273**	0.002	0.975*
Parity	-	-	0.001	0.204*

Note: The reference category is in parentheses.

* P < 0.01

** P < 0.05

Background

We have studied the effect of breast-feeding on post-partum amenorrhoea as a proxy for the fertility-inhibiting effect of breast-feeding. In the absence of spontaneous intrauterine mortality, amenorrhoea is one of the most important phenomena that occur during the birth interval (Bongaarts and Potter, 1983). Thus, it would be most pertinent to study the extent to which breast-feeding delays pregnancy as a result of post-partum amenorrhoea since breast-feeding is known to prevent women from becoming pregnant under certain circumstances. Although neither the contraceptive effects (for the mother) nor the health benefits (for the infant) of breast-feeding can continue for an indefinite period post-partum, breast-feeding does function as a nearly perfect contraceptive under two conditions: (a) when a mother is fully or nearly fully breast-feeding and (b) when a mother remains amenorrhoeic (ignoring any bleeding or "spotting" during the first two post-partum months).

If these two conditions are fulfilled, breast-feeding provides highly effective contraceptive protection for the first six months post-partum. Thereafter, the contraceptive effect decreases, although for the majority of women, the contraceptive benefits do not end abruptly with the return of menses (Kennedy and others, 1989).

In view of this effect, in countries where modern contraceptive use is limited, breast-feeding has been established as a major mechanism in achieving birth intervals of up to 30 months, accounting for an average of five fewer births per woman than would have occurred in the absence of breast-feeding (Family Health International, 1989). However, in the context of Bangladesh, breast-feeding actually inhibits an average of 6.5 births per woman (Huq and Cleland, 1990).

The work previously done by Weis (1993) has been reconfirmed independently by the authors. The results of this study are presented here because the implications are important for policy purposes. It may be observed that in Bangladesh the adoption of contraception is quite low, i.e. ranging from a low of 10 per cent at three months post-partum to about 22 per cent at nine months, with the percentage then rising very steadily to about 40 per cent at 36 months post-partum. At three months post-partum, 9 and 15 per cent of rural and urban women use contraception, respectively. At nine months this figure increases to 20 per cent for rural women and 40 per cent for urban women. For rural women, use of contraception increases very slowly to about 25 per cent at 36 months, while for urban women it rises to 60 per cent at 36 months post-partum. The study further shows that for non-amenorrhoeic women the first subsequent pregnancy occurs only at three months post-partum while for amenorrhoeic women it occurs at 12 months post-partum. Thus, lactational amenorrhoea offers breast-feeding women a significant degree of natural protection against pregnancy for the first 12 months post-partum.

For women who are currently using contraception and are amenorrhoeic, exposure to pregnancy starts from 12 months post-partum; the percentage of pregnancy assumes a slightly fluctuating but low value between 0 and 5 per cent up to 36 months post-partum. However, for mothers who are not currently using contraception and are not amenorrhoeic, the first subsequent pregnancy occurs after only three months, steadily increasing to a value of 35 per cent at 36 months post-partum. Therefore, the timing of initiation of contraception should be revised since there would be no point in offering contraception to amenorrhoeic women fully breast-feeding their babies during the first 12 months post-partum, since lactational amenorrhoea provides natural protection against pregnancy during this time. However after the period of 12 months post-partum, contraceptive use should be initiated as women are highly susceptible to pregnancy from that time on.

This study shows that breast-feeding is virtually universal (97.4 per cent) and homogeneously prolonged in Bangladesh. The average duration of breast-feeding in Bangladesh for surviving children was found to be 28.2 months, whereas it was 28.9 and 27.3 months for all children according to WFS and 1976 BFS data, respectively.

It was observed that rural women have a higher duration of breast-feeding than urban women. Also, older women are likely to breast-feed for a longer duration than younger women. Women in Rajshahi and Khulna divisions have comparatively longer breast-feeding durations than women in the two other divisions, while women in Chittagong division have the shortest breast-feeding duration. Women with post-primary education breast-feed quite markedly less than other women, while mother's exposure to education up to but not beyond the primary level has little impact on breast-feeding duration. Higher parity women have a longer duration of breast-feeding than those with lower parity. Women belonging to the lower socio-economic groups and having a more traditional life-style, breast-feed longer than their more affluent counterparts. The sex of the index child does not result in any marked differential in breast-feeding duration. Working women have a slightly longer breast-feeding duration than their non-working counterparts. Women are likely to breast-feed for a longer duration when they are visited by health workers; however, the differential by current use of contraception is marginal. Women whose husbands make decisions on family health breast-feed longer than those who make health decision by themselves alone or jointly with their husbands, but the difference is only marginal. Wives of labourers and farmers breast-feed for a longer duration than those of other professions, while wives of salesmen and those working in services breast-feed for shorter durations than the others.

Our examination of the effect of breast-feeding on post-partum amenorrhoea as a proxy for the fertility-inhibiting effect of breast-feeding has identified breast-feeding as the principal determinant of post-partum amenorrhoea. Breast-feeding duration is positively associated with the length of amenorrhoea and thus contributes to a reduction in fecundability. Lactational amenorrhoea offers breast-feeding women an excellent natural protection against pregnancy for up to 12 months post-partum. Therefore, frequent breast-feeding for up to 12 months post-partum should subsequently be followed by the use of modern contraceptives in order to derive the maximum benefits from breast-feeding as a family planning method, confirming the conclusion by Weis (1993).

These findings hold the following implications for policy purposes:

o The Government and policy makers should take appropriate measures that will help to preserve the practice of breast-feeding where it is currently common, and encourage and facilitate breast-feeding where the practice is declining.

o The family planning programme should identify mothers who are highly susceptible to pregnancy, i.e. women who have started to menstruate after they have given birth but who are not using a modern contraceptive method, and promote modern contraception among them.

o The Government should embark upon a balanced programme of family planning that simultaneously promotes the practice of breast-feeding and the use of modern contraception.

o Programmes advertising the various benefits of breast-feeding, particularly those related to improved infant health and lowered susceptibility to pregnancy, should be undertaken. Likewise information, education and communication (IEC) programmes using the mass media such as radio and television should be used to increase awareness of such beneficial effects. A ban should be imposed on advertising infant formula and related breast-milk substitutes through the mass media.

o The marketing of breast-milk substitutes should be controlled in order to encourage and promote the practice of breast-feeding, and a code of conduct for substantially restricting the marketing of breast-milk substitutes should be formulated through government and private-sector consultations.

Ahamed, M.M. (1986). "Breastfeeding in Bangladesh", Journal of Biosocial Science, 18(4):425-434.

Bongaarts, J. and R.G. Potter (1983). Fertility, Biology and Behavior: An Analysis of the Proximate Determinants (New York/London: Academic Press).

Cutler, S.J. and F. Ederer (1958). "Maximum utilization of the life table method in analyzing survival data", Journal of Chronic Diseases, 8:699.

ESCAP (1995). "1995 ESCAP Population Data Sheet" (Bangkok: Economic and Social Commission for Asia and the Pacific).

Family Health International (1989). "Experts say breastfeeding is an appropriate alternative method of family planning", Progress Newsletter, 9:4.

Ferry, B. and D.P. Smith (1983). "Breastfeeding differentials", World Fertility Survey Comparative Studies - Cross National Summaries, No. 23 (Voorburg, Netherlands: International Statistical Institute).

Huffman, S.L., A.K.M.A. Chowdhury, J. Chakraborty and N.K. Simpson (1980). "Breast-feeding patterns in rural Bangladesh", American Journal of Clinical Nutrition, 33(1):144-154.

Huffman, S.L. (1984). "Determinants of breastfeeding in developing countries: overview and policy implications", Studies in Family Planning, 15(4):170-183.

Huq, M.N. and J. Cleland (1990). Bangladesh Fertility Survey 1989: Main Report (Dhaka, Bangladesh: National Institute of Population Research & Training [NIPORT]).

Kennedy, K.I., R. Rivera and A.S. McNeilly (1989). "Consensus statement on the use of breastfeeding as a family planning method, Bellagio, Italy, August 1988", Contraception, 39(5):477-496.

Kleiman, R.L. and P. Senanayake (1984). Breast-feeding: Fertility and Contraception (London: International Planned Parenthood Federation), IPPF Medical Publications.

Lee, E. and M. Desu (1972). "A computer program for comparing k-samples with right-censored data", Computer Programs in Biomedicine, 2:315.

Leridon, H. (1977). Human Fertility: The Basic Components, translated into English by Judith F. Helzner (Chicago: University of Chicago Press).

Mosley, W.H., L.H. Werner and S. Becker (1982). "The dynamics of birth spacing and marital fertility in Kenya", World Fertility Survey Scientific Reports No. 30, (Voorburg, Netherlands: International Statistical Institute).

Nessa, F., S. Rahman and A.K.M.S. Ahmed (1987). "Breast-feeding pattern in Bangladesh and its effects on lactational amenorrhoea", Bangladesh Journal of Nutrition, 1:15.

Page, H.J., R.J. Lesthaeghe and I.H. Shah (1982). "Illustrative analysis: breastfeeding in Pakistan", World Fertility Survey Scientific Reports No. 37:17-21, (Voorburg, Netherlands: International Statistical Institute).

Rabei, A. (1988). "Breastfeeding aspects and its impact on fertility in Egypt, 1984", In: Studies on African and Asian Demography: CDC Annual Seminar, 1987 (Cairo: Cairo Demographic Centre, 1988) CDC, Research Monograph Series No. 17:3-38.

Srinivasan, K. (1980). "Birth interval analysis in fertility surveys", World Fertility Survey Scientific Reports No. 7:19, (Voorburg, Netherlands: International Statistical Institute).

Thapa, S. and N.E. Williamson (1990). "Breast-feeding in Asia: an overview", Asia-Pacific Population Journal, 5(1):7-25.

Weis, Peter (1993). "The contraceptive potential of breastfeeding in Bangladesh", Studies in Family Planning, 24(2):100-108.

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