Effect of community-based promotion of exclusive breastfeeding on diarrhoeal illness and growth: a cluster randomised controlled trial

Introduction

The results of two randomised controlled trials1,2 and some observational studies3–10 have shown a protective effect of breastfeeding against diarrhoea compared with other forms of feeding. In developing countries, breastfeeding is common but exclusive breastfeeding is not.1,11,12 No large-scale trial has assessed promotion of exclusive breastfeeding in a population with high breastfeeding rates. In a fairly small trial,1 which was designed to assess this possibility, diarrhoea was reduced in the first 3 months of life in a secondary analysis. To justify large-scale interventions, a reduction in diarrhoea and other illnesses through promotion of exclusive breastfeeding in developing countries during the first 6 months of life needs to be shown.

WHO, in its guidelines,13 recommends exclusive breastfeeding for the first 6 months rather than the first 4–6 months. There are concerns that exclusive breastfeeding for this long might be difficult, however, particularly where maternal malnutrition is common.13 Also, few data exist on physical growth that provide reassurance that exclusive breastfeeding for 6 months does not lead to growth faltering, particularly in low-birthweight infants.14

Two strategies have been successful in the promotion of exclusive breastfeeding: the Baby Friendly Hospital Initiative, which increased the likelihood of exclusive breastfeeding in Belarus, where most births take place in health facilities, 2 and the use of peer counsellors in settings where most babies are delivered at home.1,12 The second approach, based on recruitment of workers dedicated to a single intervention, is unlikely to be sustainable in health systems with few resources.

Our aim was to assess the success, and effects on prevalence of diarrhoea and physical growth, of a community based intervention to promote exclusive breastfeeding until age 6 months and complementary feeding thereafter. Our findings on effects on complementary feeding will be reported separately.

Methods

Protocol

We began our formative research in April, 1998. By means of qualitative research methods, we sought information on community characteristics, children's nutritional status, and feeding practices. We established why infants were not breastfed exclusively and the reasons for use of breastmilk substitutes. Potential channels for the delivery of our intervention within existing services were identified. We observed the routine interactions of different categories of workers with families to see how they could be used for nutritional counselling without affecting their other work. Feeding recommendations were developed, with a standard approach that included assessment of child feeding practices and identification of common feeding problems and locally appropriate ways to solve them.15 We undertook household trials to test acceptability of different recommendations.15 We translated the nutritional recommendations into the local vernacular, and selected the channels for delivery of these messages and the points at which mothers would receive nutrition counselling with the help of representatives from different categories of health workers and the district health authorities.

We undertook a baseline survey of all households with children younger than 2 years and calculated a total score, based on socioeconomic indicators, child mortality, recent morbidity, and the prevalence of wasting and stunting, for each community. We then paired communities with similar scores. To allocate one community of each pair to the intervention group and the other to the control group, the two communities were listed in alphabetical order. A statistician, who was not involved with the rest of the study, generated four single-digit random numbers with a random numbers table; the first listed community in a pair was allocated to the intervention group if the random number was 0–4 and to the second if it was 5–9 (figure).

Click to enlarge image

Figure. Trial profile

*Not at home at time of scheduled visit, and at visits made twice weekly for 1 month thereafter.

Between October, 1999, and June, 2000, we identified all infants born in the study villages through Anganwadi workers and community informants. 70% of neonates were identified within 72 h of birth and 90% within 1 week. Infants were enrolled if they lived locally and parental consent was given. A baseline form, containing details of the child and family characteristics, was completed at the time of enrolment by a study field worker. We used birthweights in the analysis only if obtained within 7 days of birth.

We provided routine services at the control sites; according to national policy, workers are required to advise exclusive breastfeeding for 4–6 months. Our formative research indicated that home visits took place occasionally and, when they did, the focus was on family planning and immunisation. Furthermore, breastfeeding rather than exclusive breastfeeding was normally promoted.

We used the following opportunities for counselling in the intervention communities: traditional birth attendants at birth; local village-based workers (Anganwadi workers), belonging to the Integrated Child Development Service Scheme,16 who weighed children once every 3 months until age 2 years and visited newborn infants once a month at home until age 1 year; auxiliary nurse midwives who run the immunisation clinics; and other health-care providers. One worker in each intervention community was recruited by the local health authority from a local non-government organisation to support community-based activities.

The messages promoting exclusive breastfeeding were discussed at routinely held monthly meetings convened by the auxiliary nurse midwife with community representatives. The community representatives in turn held neighbourhood meetings once a month to repeat the messages to all individuals involved in the care of children younger than age 2 years. The messages to impart included immediate breastfeeding after birth, feeding only breastmilk for the first 6 months of life, and breastfeeding the infant day and night, at least eight times in 24 h. We also targeted the communications strategy at specific foods and fluids given to non-exclusively breastfed infants, such as water and ghutti—a herbal mixture—to explain their lack of benefit and possible adverse effects. The materials used for communication were posters for doctors' clinics, flip books for workers, a card with feeding recommendations, a counselling guide on solving common breastfeeding difficulties, and a mother-and-child card handed out at antenatal clinics or at the first home visit.

Health and nutrition workers in the intervention communities attended a 3-day course. Half the course was used for hands-on training in counselling individuals or groups of mothers. The training was based on an adaptation of the Integrated Management of Childhood Illnesses Training Manual On Breastfeeding Counselling,17 and included training on communication skills, detection of problems with positioning and attachment to the breast, and resolving breastfeeding difficulties.15,17

In the intervention communities, at each counselling contact, the health worker assessed an infant's feeding practices, identified difficulties, and provided information on the benefits of exclusive breastfeeding. At age 3 and 6 months, mothers and infants were visited at home by a member of the study team to ascertain exposure to different counselling sources, the details of counselling received, and any instances of disease in the infant in the past 3 months. Mothers were asked about diarrhoea in their child in the past 24 h, the past 7 days, and diarrhoeal episodes for which health-care-provider visits were made in the past 3 months. The children's weights, measured with electronic scales accurate to 0·1 kg (SECA, Hamburg, Germany), and lengths, measured with locally manufactured infantometers accurate to 0·1 cm, were obtained at birth, 3 months, and 6 months. Trained nutritionists did 24 h dietary recalls at the 3-month visit. Infants were revisited at age 9 months to ascertain the duration of exclusive breastfeeding and to assess the effect of the complementary feeding intervention. Every effort was made to ensure that individuals assessing growth and morbidity were unaware of group assignment.

All children were measured and weighed twice to ensure minimum between-observer and within-observer variability. The study began only when all the field workers obtained identical readings in both their weight measurements on a child and were in perfect agreement with the supervisor's readings. For length, a difference of 0·5 cm or less between the two readings of a field worker and between the readings of the supervisor and a field worker was judged acceptable. These excercises were repeated every 3 months.

We defined exclusive breastfeeding as maternal milk being the only food source, with no other liquids or food given except medicines, minerals, and vitamins.18 Infants described as predominantly breastfed received breastmilk as the main source of nourishment, but also received water, water-based drinks, fruit juice, herbal mixtures, and vitamins, minerals, or medicines.18 Those described as part breastfed received some breastmilk, and those described as non-breastfed, no breastmilk. Classification into breastfeeding categories reported at the 3-month visit was based on 24 h dietary recall, and the proportion of children fed exclusively on breastmilk during the first 4, 5, and 6 months of life was based on data obtained at the 9-month visit. We defined diarrhoea as the passage of frequent and liquid stools, and also recorded diarrhoea if the mother reported dast (the local term for diarrhoea).

Delivery of the intervention was monitored by the local health authorities at the routine monthly reviews of different activities. The investigators' role was restricted to the measurement of outcomes for assessment of effect. The feedback of the monthly review was given by the authorities to the workers.

Our primary outcomes were exclusive breastfeeding prevalence at 3 months, mean duration of exclusive breastfeeding, 7-day diarrhoea prevalence at age 3 months and 6 months, diarrhoea for which treatment was sought in the 0–3 month and 3–6 month age intervals, and physical growth at age 6 months. Secondary outcomes were physical growth at 3 months, 24 h prevalence of diarrhoeal morbidity, and exclusive breastfeeding rates during months 4, 5, and 6 of life.

Results

In Haryana, the literacy rates are low; 50% of women and 15% of men have never been to school.11 The common occupations for men are agriculture and employment in factories. Water supply is through community hand pumps. Families normally defecate in the fields. Maternal undernutrition rates are high; 26% of married women have a body-mass index of less than 18·5 kg/m2.11 Health care is provided through primary-health centres, each of which serves a population of about 30 000 through two or three medical officers, auxiliary nurse midwives, and other ancillary staff. There are six subcentres (population about 5000) attached to each primary-health centre. Several private practitioners trained in the biomedical or, more often, the indigenous systems of medicine—ie, Ayurveda and homoeopathy—also serve this population. The Anganwadi worker provides preschool education and food supplementation; these services are used mostly by children older than age 3 years.

Exclusive breastfeeding and diarrhoea morbidity rates obtained from the baseline cross-sectional survey were similar in the intervention and control communities. In this survey, exclusive breastfeeding rates in infants aged 4–6 months were 5% (78) and 8% (74) in the intervention and control communities, respectively. Similarly, 13% (195) of infants aged younger than 6 months in the intervention group had diarrhoea in the previous 7 days compared with 15% (196) in the control areas.

The figure 1 shows the trial profile of the randomised trial. 1115 infants born between October, 1999, and June, 2000, were identified within the eight communities (four made up the intervention group and four the control group). Both groups had similar baseline socioeconomic characteristics (table 1), though more mothers in the intervention communities than in the control villages worked outside the home. The baseline characteristics of dropouts and participants within the intervention and control communities were also similar.

Click to view table

Table 1. Baseline characteristics

By age 3 months, more infants and mothers in the intervention than in the control communities had been visited at home by Anganwadi workers (254 [53%] vs 110 [27%], p<0·0001), had attended weighing sessions (202 [42%] vs 22 [5%], p<0·0001) and immunisation sessions (414 [86%] vs 330 [80%], p=0·185), had visited a primary-health centre (80 [17%] vs 28 [7%], p<0·0001), and had met with an auxiliary nurse midwife at the monthly meeting (71 [15%] vs none, p<0·0001). Our intervention had a small effect on the proportion of infants delivered by traditional birth attendants (384 [70%] vs 305 [66%], p=0·050). In intervention communities, mothers were more frequently counselled by health and nutrition workers than they were in control villages. The proportion of health worker to mother interactions in which mothers spontaneously recalled being counselled on exclusive breastfeeding was 49% (218) versus 1% (three) in the immunisation sessions (p<0·0001), 61% (155) versus 2% (two) for home visits (p<0·0001), and 61% (123) versus none at the weighing sessions (p<0·0001). This analysis was based on the most recent reported contact to achieve maximum recall. A third of mothers were counselled on immediate breastfeeding just after delivery by traditional birth attendants in the intervention group, compared with less than 1% (six; p<0·0001) in the control communities.

Between the 3-month and 6-month visit, infants and mothers in the intervention group compared with those in the control communities had one or more of the following contacts in the past 3 months: home visits (247 [53%] vs 95 [23%], p<0·0001), and attendance at weighing sessions (228 [49%] vs 19 [5%], p<0·0001), immunisation sessions (391 [84%] vs 322 [78%], p=0·037), primary health centres (100 [21%] vs 52 [13%], p=0·0006), and at meetings held by the auxiliary nurse midwives (110 [24%] vs two [0·4%], p<0·0001). The proportion of interactions in which breastfeeding counselling was reported was 45% (110) versus nil (p<0·0001) at home visits, 50% (113) versus nil (p<0·0001) at weighing sessions, and 45% (185) versus 1·2% (four; p<0·0001) at immunisation sessions. Physicians at primary health centres rarely counselled women.

Information obtained at the home visit when infants were aged 3 months showed that prelacteal feeds of honey, tea, and diluted milk were fed to just under a third of neonates in the intervention group compared with three-quarters in the control communities, and many more children were exclusively breastfed in the intervention group (table 2). The frequency of breastfeeding was overall significantly higher in the intervention group (p=0·001). There was also a 50% reduction in babies receiving animal milk in addition to breast milk in this group (p=0·01). A similar effect was seen in a subgroup of low-birthweight (birthweight ≤2500 g) infants. The exclusive breastfeeding rates in the low-birthweight group at 3 months were 79% (125) versus 40% (49; p<0·0001) in the intervention and control communities, respectively.

Click to view table

Table 2. Effect of intervention on reported breastfeeding practices at age 3 months

The positive effect of the intervention on exclusive breastfeeding was also seen during the fourth, fifth, and sixth months of life (table 3). The mean age up to which children were exclusively breastfed was higher in the intervention than in the control communities (p<0·0001). In the low-birthweight subgroup also, higher exclusive breastfeeding rates in the intervention group than in the control group continued even in the 6th month of life (41% [62] vs 4% [5]; p<0·0001).

Click to view table

Table 3. Effect on intervention on reported breastfeeding practices at age 4, 5, and 6 months

At the 3-month and 6-month visits, fewer mothers in the intervention areas than in control areas reported infants with diarrhoea in the previous 7 days. The proportion of infants who were taken to a health-care provider for the treatment of diarrhoea in the past 3 months and in the past 6 months was also significantly lower in the intervention group than in the control group (table 4).

Click to view table

Table 4. Effect of intervention on diarrhoea morbidity during the first 6 months of life

There were no significant differences between groups in mean weights and lengths, and the proportion with weight-for-height and height-for-age z scores of 2 or less, in all study children and in the low birthweight subgroup at age 3 months (data not shown) and 6 months (table 5). The birthweights and lengths and maternal heights in the low birthweight subgroup did not differ between the intervention and control communities.

Click to view table

Table 5. Weights and lengths of all children and low birthweight children at age 6 months

Discussion

Our findings indicate that promotion of exclusive breastfeeding until age 6 months in a developing country setting through existing primary-health-care services is feasible, does not lead to growth faltering, and reduces the risk of diarrhoea. Additionally, educational intervention greatly improved the rates of exclusive breastfeeding, as previously indicated by the results of two community-based trials, which assessed the use of peer counsellors, and several hospital and clinic based programmes.1,2,12,21–23 Our findings are, however, especially important since behaviour change was achieved with an approach that is feasible on a large scale and is sustainable, because it was implemented through the routine health and nutrition services.

Our data resolve some concerns about the recommendation that infants be exclusively breastfed until 6 months of age. Furthermore, our findings show that rates of exclusive breastfeeding can be significantly increased for the initial 4 months of life as well as during the 4–6 month window in a setting where maternal malnutrition and low birthweight are common.

Ours is a large study that convincingly shows that exclusive breastfeeding promotion in a setting with nearly universal breastfeeding reduces diarrhoea morbidity in the first 3 months of life as well as in the 4–6 month period. Results of some previous studies12,14,24 have shown no adverse effect of exclusive breastfeeding on weight and length gain between age 4 and 6 months, but the effect on rates of wasting and stunting at 6 months was unknown. We did not find any increase in wasting and stunting in the intervention group despite a 5–10-fold increase in exclusive breastfeeding rates between 4 months and 6 months of age. However, we cannot be sure that these results would be maintained if universal exclusive breastfeeding up to 6 months was achieved.

Features that could have contributed to the success of this intervention were the many channels that facilitated contact with the target group soon after birth and throughout the first 6 months of life. Other than primary mothers, home visits by community workers were especially useful in reaching families, since they have an important effect on infant feeding practices. Doctors, despite enthusiasm during training, counselled only infrequently; incorporation of breastfeeding counselling into the medical curriculum could improve this situation. The clarity of messages, especially about the duration of exclusive breastfeeding—ie, by replacing 4–6 months with 6 months—was important, with a large proportion of the infants in the intervention group starting complementary feeding at exactly 6 months of age.

Some limitations of this study need to be noted. Although designed as an effectiveness trial in which the intervention was delivered and monitored by the existing system, the association with the study team could have affected the motivation and performance of health workers. The exclusive breastfeeding rates for the 4th, 5th, and 6th month were based on interviews at 9 months of age. The bias on account of longer recall seems not to be significant, however, since the exclusive breastfeeding rates for the first 3 months based on the interviews done when infants were aged 3 months and at age 9 months were similar. The exclusive breastfeeding rates are reported and not based on observations. We attempted to keep to a minimum reporting bias by use of a separate team for assessment of outcomes; this team did not take part in the intervention delivery and was unaware of the hypothesis being tested. Among mothers in the intervention communities, however, the adoption of recommended behaviour could still have been over reported to some degree. In fact, over-reporting could explain in part why increases in exclusive breastfeeding rates were larger than the reduction in diarrhoea prevalence. Additionally, greater contact with workers in the intervention sites than in the control sites might have promoted adoption of unrelated primary-care interventions that have a bearing on diarrhoea prevalence or other health outcomes. Furthermore, we did not assess the effect of our intervention on iron and zinc status at 6 months of age, an issue that needs to be addressed. Finally, our findings can only be generalised to developing countries with a similar socioeconomic and cultural milieu to that we studied.

Contributors

All investigators contributed substantially to the design and undertaking of the study, its analysis, and to the writing of the paper. Additionally, N Bhandari, S Mazumdar, and M K Bhan were responsible for the daily implementation of the study. R Bahl, J Martines, and R E Black provided technical help during all stages of the project.

The Infant Feeding Study Group—Department of Paediatrics, All India Institute of Medical Sciences, New Delhi, India (Sunita Taneja, Brinda Nayyar, Vandana Suri, Poonam Khokhar, Tivendra Kumar); District Faridabad, Government of Haryana, India (R C Agarwal, S K Sharma).

Conflict of interest statement

None declared.