We will define the cohort by conducting a census of pregnant women living in the catchment areas of 309 subcentres in Tamil Nadu. We will then randomly select 25 women per subcentre for inclusion in the study. 2 The first round of the cohort study (the pre-birth “baseline”) will be conducted in two waves over a 4-month period. We will also survey the mother or mother-in-law of each sampled woman in order to paint a rich picture of household dynamics and decision-making, health practices, and beliefs before children are born; in households where the mother has other children, we will collect detailed information about their health and nutritional status. In addition, we will collect data on the pregnant mother’s upper arm circumference as a measure of her prenatal health. (The division of the grandmother sample between the mothers and mothers-in-law of sampled women will be determined in the initial study design phase based on information about the expected proximity of households to the natal families of both the husband and wife.)
Immediately following the conclusion of the baseline survey, we will conduct the first randomized controlled trial, which will focus on exclusive breastfeeding (see the next section for more detail).
The research team will then conduct a second wave of data collection from the full cohort of mothers and infants approximately eight months after the baseline, when most sampled children are between the ages of 16 months. This wave will collect data on beliefs about breastfeeding, self-reported measures of breastfeeding, and household decision-making power (especially vis-à-vis mothers-in-law). Data will also be collected on infant health, including reported diarrheal episodes, respiratory illness episodes, and anthropometric data on weight and height; this data will allow us to measure stunting and wasting. We will also track any infant deaths, though the sample size suggests that these will be limited in number, and re-survey the sample of grandmothers to understand whether their knowledge and attitudes around infant feeding practices and infant health change following the birth of a grandchild, and to evaluate whether intrahousehold decision-making has changed in the postnatal period. For any infant deaths observed, we will conduct a verbal autopsy with the household to attempt to ascertain cause of death. Finally, we will collect data on services delivered and hours worked by the VHN and anganwadi workers.
Although this proposal only covers the first two waves of data collection, the long-run objective of this project is to continue following the cohort of children as they age. For this reason, we will also collect identifying information for the households that will allow us to track the infants in the future. This will include Aadhaar numbers, if available; birth registration identifiers; and full contact information for the family.
First Randomized Controlled Trial
The first randomized controlled trial falling under the umbrella of this project will focus on improving health during the first six months of life by promoting exclusive breastfeeding. Since new mothers often turn to their own mothers and mothers-in-law for advice on infant care, we believe a two-pronged approach to changing breastfeeding behavior and enhancing infant health could be especially effective:
First, address information gaps among mothers themselves. Here, we build upon existing research demonstrating that behavioural change counseling delivered via community-based health workers or women’s groups can increase breastfeeding rates and improve child health (Bhandari et al. 2003, Tripathy et al. 2010). An important limitation of these studies, however, is that they both examined intensive, multi-faceted interventions. A guiding principle of our project is that the interventions evaluated should be simple and scalable from the perspective of the Government of Tamil Nadu (GoTN). Second, engage grandmothers in supporting their daughters and daughters-in-law to exclusively breastfeed. Again, we believe that scalability is key in order to minimize the cost of including grandmothers in breastfeeding education. Accordingly, we are interested in evaluating the effectiveness of including both mothers and grandmothers in behavioural change sessions conducted by VHNs and anganwadi workers.
Together with the Government of Tamil Nadu, we have already pilot-tested a behavioral-change curriculum that follows the principles above (see the appendix for more details on the pilot). The proposed full-scale experiment is designed to answer the following key policy questions:
1. Can improved breastfeeding counseling increase rates of exclusive breastfeeding among new mothers, and does engaging grandmothers in the education process increase adherence to exclusive breastfeeding? 2. Do the two enhanced education policies (mothers alone, mothers and grandmothers) lead to measurable health benefits for young children: particularly, a reduction in the incidence of illnesses in infancy, and increases in height-for-age and weight-for-age? 3. From the perspective of GoTN, which policy is the most cost-effective way of improving breastfeeding behavior and infant health?
Accordingly, we seek to evaluate two versions of our enhanced exclusive breastfeeding education program. The interventions will follow the same structure and use the same materials as employed in the pilot, with minor modifications. In the first treatment arm, breastfeeding education will only target mothers; in the second treatment arm, both mothers and grandmothers will be targeted. In both arms, the primary objectives of the enhanced breastfeeding curriculum can be described as follows.
1. Close knowledge gaps among women: The educational sessions are designed to target major information gaps we identified among mothers and grandmothers during the pilot. The sessions will emphasize the benefits of exclusive breastfeeding and aid mothers and their families in addressing the common dilemmas faced by expectant and new mothers. Given that mothers may not have previously had access to such a forum, the opportunity to discuss challenges in infant health may have a positive impact on infant thriving.
2. Reinforce key messages with visual aids: Health workers will utilize a series of four posters specifically designed to support the group sessions. These posters, designed with behavioural change strategies in mind, convey the main messages around exclusive breastfeeding and reinforce the enhanced curriculum. The posters are meant to be used during the group sessions and to be displayed at the angwanwadi center at all times to provide women with ongoing reminders to exclusively breastfeed. 3. Provide specialised training for health workers: VHNs and anganwadi workers will attend a 3 day training session (developed during the pilot) specially designed to prepare them to conduct the group meetings. The training will include a refresher on the primary GoTN breastfeeding guidelines, along with extensive instruction on behavioural change communication and how to conduct a group session, as well as supervised practice.
Our calculations suggest that in order to attain sufficient statistical power, a full-scale evaluation would need to be implemented in 300 sub-centres in Tamil Nadu – thus, the cohort study can comfortably accommodate this evaluation. The sub-centres enrolled in the evaluation (and affiliated VHNs and anganwadi workers) would be randomly assigned to one of three evaluation arms:
1. Status quo policies (control arm) – VHNs and anganwadi workers in these sub-centres would receive no new training and conduct their duties as usual throughout the study period. 2. Enhanced education for mothers – VHNs and anganwadi workers in these sub-centres would be trained on the enhanced breastfeeding curriculum and would be instructed to administer the curriculum in regular group meetings in all villages falling under their purview. These health workers would not be instructed to include grandmothers in the process. 3. Enhanced education for mothers and grandmothers – VHNs and anganwadi workers in these sub-centres would be trained on the enhanced breastfeeding curriculum and would be instructed to administer the curriculum to both mothers and grandmothers. The treatment arms will be slightly larger than the control arm (38% of sub-centres will be assigned to each treatment arm, and the remaining 24% to the control arm) in order to give us more statistical power to detect differences between the “mothers only” and “mothers plus grandmothers” interventions. VHNs and anganwadi workers in treatment arms will be trained on the new curriculum. Given the large number of health workers included, we anticipate training health workers in multiple batches over four weeks; each individual training session will include no more than 30 individuals. Following the conclusion of the training, health workers will begin to conduct weekly family-based behavioural change sessions in the villages they serve. The research team will conduct random spot-checks to monitor the implementation of the intervention, assessing whether a given meeting took place and evaluating the quality of the meeting using instruments developed as part of the pilot.