"As the global momentum for investing in nutrition ramps up, there is an urgent need for demonstrated large-scale solutions for improving the most fundamental of nutrition actions, i.e., home-based behaviors to improve the quality of diets. This study offers compelling evidence that such interventions can be implemented at scale to deliver impact on what remains a substantial global challenge - improving children's diets."

Noting that the inadequacy of complementary feeding (CF) in Bangladesh and other similar contexts remains a substantial and somewhat intractable challenge, the Alive & Thrive (A&T) initiative, carried out in Bangladesh, Ethiopia, and Vietnam, used intensified interpersonal counseling (IPC), mass media (MM), and community mobilisation (CM) in an effort to foster optimal child growth and development. This paper reports on findings from a cluster-randomised impact evaluation, which sought to compare the impact of 2 A&T intervention packages on CF practices and anthropometric outcomes. In this way, the study seeks to contribute to the literature on improving CF practices through a proof-of-concept rigorous evaluation of a set of interventions delivered at scale.

BRAC, a large non-governmental organisation (NGO), delivered standard and intensified IPC and CM in 50 rural subdistricts in Bangladesh through its existing countrywide essential health care programme. For standard nutrition counseling, BRAC frontline workers (called Shasthya Kormi) and volunteers (called Shasthya Sebika, or SS) conducted routine home visits and provided information on infant and young child feeding (IYCF) practices. In intensive areas, a new cadre of nutrition-focused frontline workers, the Pushti Kormi (PK), together with the SS, conducted multiple age-targeted IYCF-focused counseling visits to households with pregnant women and mothers of children 2 years of age or younger, coached mothers as they tried out the practices, and engaged other family members to support the behaviours. The MM component, implemented in both intensive and nonintensive areas, consisted of the national broadcast of 7 television spots that were designed to reach mothers, family members, health workers, and local doctors with messages on various aspects of IYCF; 3 of the spots focused on CF. In intensive areas that had low electricity and limited access to television, supplemental activities were conducted to air the television spots and other IYCF films produced by the project through local video screenings. In intensive areas, CM included sensitisation of community leaders to IYCF, and community theatre shows focused on IYCF. In nonintensive areas, CM was less structured and covered general health care topics such as family planning, pregnancy registration, and antenatal care, and did not include IYCF-related information.

Thus, A&T used 3 different platforms, i.e., IPC, CM, and MM, to deliver interventions. The intensive group received all 3 interventions; the nonintensive group received standard IPC and less-intensive MM and CM.

The programme was delivered over 4 years, reaching nearly 2 million Bangladeshi families. The MM intervention was delivered nationwide via national television channels; the IPC and CM interventions in the intensive areas first were implemented in 50 of 493 rural subdistricts, and later integrated with BRAC's other health service platforms. During the intervention period, A&T facilitated the training of more than 75,000 frontline workers and health providers across the country. An estimated 1.7 million mothers of children 2 years of age or younger in 50 subdistricts were accessed by IPC by mid-2014. Furthermore, the programme engaged with the national government's programme by forming strategic partnerships with the Institute of Public Health Nutrition under the Ministry of Health and Family Welfare.

A cluster-randomised, nonblinded impact evaluation design was used to compare the impact of the 2 A&T intervention packages. A cross-sectional household survey was conducted at baseline (2010) and exactly 4 years later (2014) in the same communities in households with children 0-47.9 months of age [n=~600 and 1,090 children 6-23.9 months and 24-47.9 months/group, respectively, at baseline and n=~500 and 1,100 children of the same age, respectively, at endline]. The researchers derived difference-in-difference impact estimates (DDEs), adjusting for geographic clustering, infant age, sex, differences in baseline characteristics, and differential change in characteristics over time.

Groups were similar at baseline. CF improvements were significantly greater in the intensive than in the nonintensive group [DDEs: 16.3, 14.7, 22.0, and 24.6 percentage points (pp) for 4 CF practices: minimum dietary diversity, minimum meal frequency, minimum acceptable diet, and consumption of iron-rich foods, respectively]. In the intensive group, CF practices were high: 50.4% for minimum acceptable diet, 63.8% for minimum diet diversity, 75.1% for minimum meal frequency, and 78.5% for consumption of iron-rich foods. Timely introduction of foods improved. Significant, nondifferential stunting declines occurred in intensive (6.2 pp) and nonintensive (5.2 pp) groups in children 24-47.9 months.

The researchers observed a strong dose–response association between exposures to more than 1 platform and improved knowledge and feeding practices (see Table 5). For all 4 CF practices, exposure to MM alone was not significantly associated with improved practices compared with no exposure. Exposure to IPC alone was significantly associated with a 2- to 3-fold higher odds of improved CF practices, with the exception of the achievement of minimum dietary diversity. Exposure to IPC + MM was associated with a 1.7-3.5-fold greater odds of improved CF practices. Exposure to the greatest number of intervention platforms, i.e., to IPC + MM + CM, was associated with increased odds of improved CF practices ranging from 2.8-5.9-fold greater odds for different CF practices compared with no exposure. There was no similar discernible pattern of exposure to combinations of programme interventions with stunting or height-for-age z score (HAZ), although exposure to MM or IPC alone was associated with lower odds of stunting.

In short, the study showed that "[a] program providing intensified IPC, MM, and CM (the A&T intensive intervention) at scale had a substantial and significant impact on several CF practices in comparison with changes observed with a less intensive behavior change intervention in Bangladesh. Although improvements in child growth were observed in both groups and for all age groups over time, the DDEs for linear growth and stunting at 24-47.9 mo were not statistically significant; hence, we cannot attribute improvements to the A&T intensified interventions."

The researchers say (footnote numbers removed): "Lessons learned for programming to improve CF practices at scale are worth noting in the context of this program evaluation. First, the behavior change interventions tested were developed through formative research for program design; their roll-out and scale up were done through monitoring of quality and coverage; the intervention delivery also included several supervision and management approaches that supported implementation....The most critical lessons for scaling up relate to an explicit focus on specific behavior change goals; ensuring adequate investments in intervention design; using data to make decisions about coverage and intervention quality; and ensuring the availability of adequate, stable, and flexible financing for delivery. At the same time, the evaluation results suggest that programs that aim to improve CF practices in food-insecure and poor environments should carefully consider complementing the behavior change interventions with complementary interventions that can help address financial and resource constraints to adoption of optimal practices. Programs that aim to achieve an impact on child nutritional status will also need to consider how to address the other determinants of child growth, such as maternal nutrition, sanitation, and poverty. This could be done by incorporating other tailored interventions to specifically address other constraints."

In short, with a cluster-randomised evaluation design, this study demonstrates that multiplatform behaviour change interventions reach households with both scale and intensity, and that this approach improves several CF practices.