Child undernutrition remains a highly prevalent condition in low and middle income countries and a major portion of the global burden of childhood malnutrition is found in South Asia with an estimated 74 million children living with chronic malnutrition (stunted growth) in this region. This burden of malnutrition accounts for approximately 50% of under five child deaths in developing countries. The risk of child death is also related to the severity of malnutrition and progressively increases the more the child's growth deviates from WHO Growth Standard, e.g. children with height-for-age <-3 Z scores from the standard have a four fold increased mortality risk compared to children within one Z score of the growth standard mean. Underweight in children for 18.7% of the global disability-adjusted life years in children less than five years of age. Childhood malnutrition is a pervasive problem in Bangladesh with 43% of children less than five years stunted in 2004 and 38% in urban child populations. An establish approach to promoting appropriate breastfeeding practices is through the use of local peer counsellors to provide information and to support to mothers. A recent study has conducted on pioneering research on this approach in Bangladesh.. The main aim of the study is to use a Cluster Randomized Control Trial (CRCT) to collect high-level evidence of whether peer counselling of women to promote appropriate breastfeeding and complementary feeding can improve feeding practices, child growth and reduce the prevalence of malnutrition in their children.

The investigators will use a community-based Cluster Randomized Control Trial (CRCT) to examine the impact of a peer counselling infant feeding education program starting in the third trimester of pregnancy to one year after delivery, to improve child feeding practices, child growth and reduce the prevalence of malnutrition in their children. This will result in two study groups. The outcome assessments will be made on a cohort of infant-mother dyads measured at baseline and at follow up visits because the investigators expect a likely high correlation between baseline and follow up outcome measures, thus making this approach the most efficient study design. Outcome assessments will be conducted with all the mother-infant pairs recruited in the community clusters in the study, with an expected total of 1950 mother-infant days (975 in each treatment group). The peer counseling education will be offered to eligible pregnant women identified by household surveys over 3 months in each community cluster in the intervention group. Using a similar approach to recruitment, a cohort of mother-infant dyads, who will receive standard maternal and child health care programs, will be identified in the control clusters.

Data will be collected on anthropometry, feeding practices and hygiene and caring practices etc. The investigators will report the results for 2-sided 5% tests for the primary trial outcome. Secondary analyses will examine each outcome variable (stunting, height-for-age, feeding patterns, and mean nutrient intakes) taking account of the repeated measurements within children by using separate mixed models. The investigators will use linear mixed models for continuous outcomes (e.g. height-for-age Z) and generalized linear mixed models for non-continuous outcomes (e.g. logistic mixed models for binary outcomes e.g. percentage exclusively breastfeeding). It is expected that the publications from this research will have substantial impact on child health and will help with the development of public health nutrition policies for children in South Asia and will be widely cited.

1. Experimental design

   a. Study design We will use a community-based Cluster Randomized Control Trial (CRCT) to examine the impact of a peer counselling infant feeding education program starting in the third trimester of pregnancy to one year after delivery, to improve child feeding practices, child growth and reduce the prevalence of malnutrition in their children. This will result in two study groups (see diagram). The outcome assessments will be made on a cohort of infant-mother dyads measured at baseline and at follow up visits because we expect a likely high correlation between baseline and follow up outcome measures, thus making this approach the most efficient study design.

   The trial will be conducted in a total of 50 community clusters (Mahallas - smallest geographic units in Dhaka -see details below in Sampling Scheme), with 25 clusters in each study group. The peer counseling education intervention will be delivered to the mothers, by locally recruited and trained peer counsellors, starting in the third trimester of pregnancy until the child is one year of age. The peer counseling education will be offered to eligible pregnant women identified by household surveys over 3 months in each community cluster in the intervention group. Using a similar approach to recruitment, a cohort of mother-infant dyads, who will receive standard maternal and child health care programs, will be identified in the control clusters.

   Outcome assessments will be conducted with all the mother-infant pairs recruited in the community clusters in the study, with an expected total of 1950 mother-infant dyads (975 in each treatment group). There will be a baseline assessment and further assessments every 3 months from birth until the children are 18 months of age. Two of these assessments (when infant is 15 and 18 months of age) will be made after the education intervention has ceased to assess the sustained impacts on the prevalence of child stunting and infant feeding practices (see evaluation plan for details).

   In this CRCT the interventions will be allocated at a community level, but the outcome assessments will be at the individual level. This is the best comparative design for the proposed interventions, which cannot be delivered to individual mothers in densely populated urban slum communities to avoid contamination of the intervention. A four year study is required to ensure an adequate "dose of the intervention" is delivered to the mothers in the intervention group, and to give time to evaluate the longer-term impact of the interventions.

   RANDOMIZATION PROCEDURE:

   b. Sampling scheme The proposed field area for the study is Mirpur, an "auxiliary Thana" in the Dhaka Metropolitan City Corporation with a total population of 5,580,000 (according to the Census 2001, although it may have increased by a further 20% to 30% since then). Mirpur has 16 "Wards" which are the smallest urban administrative unit each of which has a local government authority called a "Ward Council". The average population of a "Ward" is 350,000 people. Below the "Wards" are "Mahallas" which are the lowest urban geographic unit with identifiable boundaries. On average there are five "Mahallas" per "Ward" with an average population of 70,000 people. The unit of randomization for the trial will be Mahallas. In Mirpur there will be a total of 80 Mahallas, which will form the sampling frame for selecting the clusters in the trial. The list of these Mahallas is available from the Dhaka Metropolitan City Corporation.

   The definition of the "crude birth rate (CBR) is the number of births per 1,000 population during a specified time period".[6] The 2004 Bangladesh DHS reported a current CBR for urban Bangladesh of 25.8 per 1000 population for 3 years prior to survey. Thus the average expected number of births for each Mahallas over a 3 month period would be 150, which would be more than the required recruitment number for the trial and ensuring the feasibility of the sampling scheme for the trial.

   c. Recruitment and inclusion/exclusion criteria for clusters

   "Mallahas" on the sampling frame will be eligible for the study if they do not have any other infant feeding interventions currently being implemented either from the government of non-government sectors. Any such programs will be identified by contacting local officials in the Ward Councils in Mirpur. At present there are no infant feeding promotion programs run by the government health services in this area. The Dhaka Metropolitan City Corporation will be contacted in person to obtain their written approval for the study before the "Mahallas" are allocated to a treatment group or trial recruitment of subjects begins.

   d. Assignment of treatments The interventions will be assigned to eligible "Mahallas" using a fixed randomisation scheme with uniform allocation ratio of treatments, and blocks of 5 or 10 to ensure geographic balance across Mirpur area. The random allocation sequence will be generated using SAS software.

   e. Recruitment, inclusion/exclusion criteria and consent of mothers and their families A previously successful approach to recruitment will be used in which trained field assistants will identify women who are in their third trimester of pregnancy by systematic door-to-door surveys across the "Mahallas" over a 3 month period. Pregnant women will be included who are 16 to 35 years with no more than three living children. Women who plan to migrate from the Mirpur area after delivery will be excluded. Also women with documented medical records of heart disease, tuberculosis, gestational diabetes or eclampsia in previous pregnancies will be excluded. Mother-infants pairs will also be excluded where there are multiple births, congenital abnormalities, very low birthweight below 1.5 kg or infant admitted to a neonatal intensive care unit.

   Based on experience with similar projects in Bangladesh[3,8] it is expected that at least 95% of the mothers will consent to participate. It is also likely that up to 30% will be excluded mainly related to their desire to migrate back to their home village after the delivery.

   Sample Size Calculation and Outcome (Primary and Secondary) Variable(s)

   The sample size for the trial was estimated with the following assumptions:

   Sample size calculation for increasing the rates of exclusive breastfeeding in the intervention group

   • Considering the rate of Exclusive Breast Feeding (EBF)43% in the control clusters, Bangladesh Demographic Health Survey (BDHS) 2007 and the expected rate of EBF about 63% i.e. an increase of 20% in the intervention group the required sample size will be 92.4 for each group and using formula; n= P1(100-p1) + P2 (100-p2)/(P2-P1)2 x 7.9 Taking 95 (round up the figure) and multiplying by 4 to reduce the cluster effect the sample size become 380. Calculating a 20% drop out, out migration the required sample size comes to 456 in each group to have a 0.05 significance and with a power of 0.80.

   • Each community-cluster (Mahalla) has an average population of 70,000 and an expected crude birth rate of 4.3 per 1000 total population over 6 months [based on CBR for urban Bangladesh of 25.8/1000 population over 36 months from 2004 BDHS thus giving an average expected number of births of 150 over a 3 month period per Mahalla cluster.
   • Previous research indicates that appropriately 33% of pregnant women will return to their home village following the delivery, thus leaving an expected number of eligible births of 200 over a 6 month period per Mahalla cluster.
   • 39 mother-infant dyads per Mahalla cluster assuming 95% participation but 22% loss to follow-up based on earlier research,from the approximately 200 mother-infant dyads available in each cluster over a 6 month period.
   • 90% power and 5% two-sided alpha
   • Intra-cluster correlation coefficient (ICC) of 0.015 [based on analyses of the child anthropometric measurements from the 2004 Bangladesh DHS survey data for urban child populations.
   • Expected difference in the prevalence of stunting between the treatment groups of 10% (35% in control to 25% in intervention group), which is similar to the change reported in an earlier education intervention for young child feeding in Peru.

   The sample size required would be 1950 mother-infant pairs (975 in each treatment group) from 50 Mallahas clusters with 39 mother-infant dyads per community cluster recruited over 3 months.

   Facilities Available

   Describe the availability of physical facilities at site of conduction of the study. For clinical and laboratory-based studies, indicate the provision of hospital and other types of adequate patient care and laboratory support services. Identify the laboratory facilities and major equipment that will be required for the study. For field studies, describe the field area including its size, population, and means of communications.

   (a). Manpower (b). Administrative infra-structure (c). Expertise (d). Anthropometric equipment's

   Intervention plan The trial intervention has been selected because its feasibility has been tested in urban populations in Dhaka, and it is likely to be sustainable in the future. The proposed individual peer counseling education will be of sufficient intensity to alter infant and young child feeding practices and to improve the growth of the young children and prevent malnutrition. After one year of preparations for the trial, the interventions will be delivered to the women starting in pregnancy and until their child is one year of age over the second and third years of the study.

   a. Description of intervention The approach to promoting appropriate infant and young child feeding will be through a program of home-based peer counselling by trained, local women from the mothers' community. This approach will reach mothers who deliver at home and will also allow the messages to reach other key family members who may play a role in supporting breastfeeding and influence the foods choices for the infant. The main messages will be directed at encouraging early initiation of breastfeeding, promoting exclusive breastfeeding during the first 6 months of life, promoting appropriate timing of the introduction of complementary feeds, and ensuring an adequate frequency of feeds and diversity of foods used in their preparation.

   i) Selection and training of peer counsellors Women with personal breastfeeding experience, at least 6 years of schooling, residing in the same area, and motivated to work will be selected to become peer counselors. The WHO/UNICEF Breastfeeding Counseling Course adopted to the local language and culture, which has already been validated in a previous study[3] will be used for training of peer counselors. The training will be given for 40-h (4-h daily for 10 days). Counseling skills will be taught mainly by demonstrations and role play and will include: listening to mothers, learning about their difficulties, assessing the position and attachment of babies during breastfeeding, building mother's confidence, giving support and providing relevant information and practical help when required. During the training course antenatal and postpartum counseling will be practiced with pregnant women, mothers with newborns and infants aged 1-12 months in the field site. The counselors will also be taught how to use locally available foods for complementary feeding of infants and young children, and how best to demonstrate these food preparation skills to mothers.

   It is anticipated that each peer counselor will be able to support up to approximately 10 mothers and thus to provide support to the approximately 1000 women receiving the intervention (see section 4.a. Sample Size and Power for details) 100 peer counselors (four in each community cluster) will need to be recruited and trained. The performance of the counselors will be monitored at least four times during the course of the study by the Senior Infant Feeding Counselors.

   ii) Counselling schedules: There will be a schedule of at least 13 visits by the peer counsellors: two before delivery; four during the first month; five monthly visits from age 2 to 6 months; and three monthly visits at age 9 and 12 months. The counsellors will be free to make additional visits if the mother's circumstances require them. The counselling will take place at home to ensure key family members (e.g. mother-in-law and fathers) can also be included in the counselling sessions. The duration of each visit will be from 20 to 40 minutes.

   Antenatal visits: The peer counsellors during the two antenatal contacts will prepare the mothers, and other members of the family who will support her at delivery, about the importance of holding the baby within a few minutes of delivery and how to initiate breastfeeding within one hour of delivery. They will discourage prelacteal feeds and other fluids and foods after lactation has been initiated. They will encourage the mothers to eat more of their usual foods to support enhanced lactation, and to appropriately rest during the third trimester. These meetings will also cover problems with breastfeeding that the mother might encounter and how best to deal with them.

   Visits in the first month of life: The mothers will be contacted four times by the peer counsellors (within 48 hours of delivery, at 5-7 days, at 10-14 days and at 24-28 days). At these visits exclusive breastfeeding will be encouraged and the mother's specific needs addressed. Issues that could be covered include sore nipples, problems with attachment tot eh breast, the baby's position during feeds, family pressure to start other foods and mothers' doubts about the adequacy of their breastmilk. If any of these issues could not be resolved the mothers could be referred to the Senior Infant Feeding Counsellors.

   Visits 2 to 6 months of life: The mothers will be contacted monthly by the peer counsellors. Specific problems will be addressed and continued support for exclusive breastfeeding will be provided especially how to deal with family pressures to introduce other foods and concerns about the adequacy of the growth of the infant. From 5 months of age specific messages that include, the importance of complementary feeding, demonstration and preparation of complementary foods will be introduced. Mothers will be given measuring cups and spoons and two types of complementary foods will be demonstrated. The two types of complementary feeds to be promoted are based on prior survey of foods given to young children in urban Dhaka. The first will be "rice-suzi" (powdered rice + soybean oil to make it energy dense), and the second will be "Khichri" (cooked rice + lentil + soybean oil). Mothers will also be discouraged from using bottles for feeding.

   Visits from 6 to 12 months of life: The mothers will be contacted twice at 3 month intervals by the peer counsellors. The mothers will be encouraged to continue breastfeeding and support will be given for an adequate frequency of complementary feeds and an appropriate diversity foods. There will be further demonstrations of the preparation of complementary feeds as needed.

   b. Management of interventions There will be two senior infant feeding counsellors who will train the peer counsellors under supervisor of CI-B. These senior counsellors will provide technical support to the peer counsellors and help them resolve problems they encounter during the implementation of the trial interventions by regular meetings with peer counsellors in the field. . There will be 10 field supervisors who will support the day-to-day logistics and administration of the field activities including support for the senior infant feeding counsellors when they are in the field and the per counsellors.

   To facilitate the overall implementation process, an advisory committee consisting of a local health officers, local government officials, representatives of mothers' groups from the community, representatives of the Bangladesh Breastfeeding Foundation, other relevant NGOs and project staff will help guide the research team and will meet every 3 months as the project progresses.

3. Evaluation plan

1. Outcome assessment i) Key trial outcomes Differences in the percentage of stunted infants (height-for-age <-2 Z) at 6, 12 and 18 months.

   Changes in mean height-for age Z scores from birth till 18 months Changes in the percentage of women exclusively breastfeeding (breast milk and no other foods or milk based liquids) their infants at 3 and 6 months.

   Changes in mean duration of any breastfeeding. Changes in the percentage of women bottle feeding (any liquid or semi-solid food from a bottle with nipple/teat) their infants at 9 and 12 months.

   Changes in percentage of children at 6 & 9 months who receive solid, semi-solid or soft foods.

   Changes in percentage of children consuming foods from >4 food groups at 9, 12, 15 & 18 months.

   Changes in the mean intake of food energy, protein, carbohydrate, fat and selected micronutrients from complementary feeds at 9, 12, 15 and 18 months.

   ii) Measurements

   1. Anthropometry:

      Trained research assistants will collect anthropometric measurements (weight and height) using established methods[9] and will record these measurements on both the research instruments and an infant growth chart for the mother to hold.. These measurements will be standardized before and during the data collection. Anthropometry will be collected soon after birth and at 3 month intervals (3, 6, 9, 12, 15 and 18 months). The 2006 WHO Growth Standard will be used to construct anthropometric indices and standard WHO recommended indicators will be used to assess stunting (Height-for-age <-2 Z), wasting (weight-for-height <-2 Z) and underweight (weight-for-age <-2 Z).

   2. Infant feeding practices:

      Standard questions about infant feeding practices used in the Bangladesh DHS surveys[6] will be used to monitor these patterns at the 3 monthly data collection periods from birth till 18 months of age. These will include questions about current breastfeeding status, current use of other liquids and solid foods, timing of introduction of other liquids or solid foods, use of bottles for feeding and information about who is providing advice about infant feeding amongst family and friends. Ten percent of the interviewer's scheduled visits will be monitored by one of the Senior Research Assistants or Principal Investigators or Investigators. The questionnaires will be checked daily, and if the information is incomplete or not clear, the supervisor will return to the home on the next day to complete the data form. Mother's reported infant feeding practice will be verified by a 4-hour observation period that will take place by unscheduled visits by Senior Research Assistants or Principal Investigators or Investigators.

   3. Infant and maternal morbidity Infant Morbidity Illness histories such as diarrhoea, dysentery (blood and/or mucus), fever and cough, ear infection (purulent discharge) from ears will be obtained at 3, 6 and 12 and 18 months using 2-weeks' recall method. The questions asked will be based on the standard DHS infant morbidity recall questions but expanded to include questions about ear discharge.

      Diarrhoea will be defined as an episode of the passage of three or more loose or watery stools within 24 hours. Presence of blood with stool will be defined as invasive diarrhea. When a single episode of diarrhea lasts for more than two weeks, it will be called persistent diarrhoea. Acute respiratory illnesses will be define as an episode of cough with reported fast and rapid breathing or difficulty breathing with or without fever.

      Maternal morbidity A simple recall method will be used to assess morbidity in the mothers. They will be asked to recall, over the month preceding the interview, any days they were ill enough to have their usual daily activities restricted. If they have experienced restricted day illness they will be asked about key symptoms (cough, diarrhoea, urinary discomfort and fever). A medical history will be collected at baseline to capture information about maternal chronic illnesses as well.

      The total number of restricted days and the number of restricted days with fever will be used in analyses of their relationship with feeding practises. If maternal morbidity is equally distributed across the treatment groups we would not expect that this factor would bias the trial results.

   4. Dietary intake: Trained interviewers will collect 24-hr dietary recalls using standard methods starting at 6 months of age till 18 months. All the foods consumed in the 24 hours prior to interview will be recorded and portion sizes measured in local utensils. Recipes used to prepare foods will be recorded including the amounts of raw food used and the preparation methods. The foods will be coded by the research assistants and nutrient calculated in special purpose programs using the Bangladesh Food Composition Tables. Both nutrient intakes and food groups consumed will be analyzed and presented. Duplicate 24 hour recalls will be taken in a 10% subsample of respondents to assess within person variability and to allow adjustment of the prevalence of low intake nutrients.

   5. Focus group discussion:

   A Focus Group Discussion (FGD) is an unique method of qualitative research that tries to find the views of people in a group through an interactive way. FGD will be conducted with the mothers to learn about their perception and practices on child feeding and dietary pattern. FGD will be focus on -

   1. Perception on colostrum, Timely Initiation of Breast Feeding (TIBF), Exclusive Breastfeeding (EBF), continuation of BF, and complementary feeding.
   2. Constraints to feeding colostrum, EBF, continuation of BF, and complementary feeding.
   3. Perception, causes, possible consequences of malnutrition from mothers that they experience in their day to day life.
   4. The perceptions on food security of children like complementary feeding practice and pattern and adequacy of breast feeding.
   5. Perception on disease, services taken during illness.
   6. Beliefs about appropriate types of complementary feeding

   The FGDs will be conducted with mothers in 6 groups from the intervention areas and 6 groups from the control areas from randomly selected communities. Each FGD will consist of 6 to 8 participants, and will be held at a location and time, which is convenient for the group. The FGD will be conducted over 2 hours for each group. A trained Field Research Officer will facilitate the session. The group facilitator will be assisted by research personnel, who will record all answers in several broad categories. She will also record the session using cassette player for later transcription and translation. The qualitative answers will then be transformed into quantitative expression. The data from the focus groups will be analyzed to extract essential issues and terms.

   A statistical program ATLAS ti will be used for qualitative data analysis which will use 1) Free listing. 2) File sorting and 3) Rating. Dr. Sabrina Rashid an anthropologist has been included in the study team (as consultant) who has expertise in qualitative data analysis.

   iii) Schedule of data collection

   1. A pilot study will be conducted in the first year of the study in a single cluster to test the recruitment methods, the implementation of the intervention and the evaluation instruments and field methods before commencing the trial.
   2. Assessment of trial outcomes: A baseline survey at enrollment will be used to collect basic socio-demographic information about the family and the maternal characteristics. The details of the birth and pregnancy will be recorded shortly after delivery. The trial outcomes will be assessed by a series of measurements taken at 3 monthly intervals from birth until 18 months of age. These include anthropometry, recording of dietary patterns and dietary intake.

2. Process assessment is needed to examine the fidelity, dose, reach and intensity of interventions:

i) Participation of mothers in peer counselling sessions

ii) Self report adherence to peer counselling guidelines by the peer counsellors

iii) Unscheduled observations of the peer counsellors at home visits

iv) Focus groups & open ended questions with mothers & peer counsellors about the program This data will be collected during the implementation of the interventions, and by focus groups at the end of the intervention year.](streamdown:incomplete-link)

Inclusion Criteria:

• Pregnant women will be included who are 16 to 35 years with no more than three living children.

Exclusion Criteria:

• Women who plan to migrate from the Mirpur area after delivery will be excluded.
• Women with documented medical records of heart disease, tuberculosis, gestational diabetes or eclampsia in previous pregnancies will be excluded.