Randomized Controlled Trial of Social Network Targeting in Honduras

Overview

Social network targeting strategies can be used to improve the delivery and uptake of health interventions. We will enroll approximately 30,000 individuals into a randomized controlled trial of different targeting algorithms in order to explore how social network dynamics affect the uptake, diffusion, and group-level normative reinforcement of key neonatal and infant health behaviors and attitudes in 176 rural villages in the Copan region of Honduras. Our goal is to develop methods by which global health practitioners can exploit face-to-face social network interactions in order to maximize uptake of neonatal and infant health interventions. The villages will be randomly assigned to 16 cells of 11 villages each in a 2 x 8 factorial design of different targeting algorithms.

Honduras has one of the highest neonatal mortality rates in Latin America despite having made significant strides in reducing neonatal, infant, and child mortality in the last several decades. Although many neonatal and infant deaths can be prevented through provision of clinical care services, emerging evidence also suggests that a substantial reduction in neonatal and infant mortality can also be achieved with simple, low-cost interventions within family and community settings. This is particularly important in areas where functional community health facilities are not available. Family and community outreach programs can serve to educate families about beneficial home care practices. In order to accelerate neonatal mortality reduction , there is an urgent need to develop innovative solutions that are not only effective, but also more easily implementable and more readily scalable. An important component of this challenge, which has hitherto not been effectively measured and understood with respect to neonatal mortality, is the "embeddedness" of individuals within social networks. Hence, through a large-scale randomized controlled trial (RCT) in rural Honduras, we will deploy and assess social network targeting algorithms in order to maximize diffusion and adoption of the "Changing behaviors to improve neonatal, child, and maternal health using communication and social networks at the community level intervention (CBNH)". The CBNH intervention is a household-level intervention package that targets health behaviors surrounding neonatal and maternal health, and diarrhea and respiratory illness prevention and management implemented by the Inter-American Development Bank (IADB) and their partners. This RCT is aimed at discerning optimal methods for targeting delivery of the intervention to the population. Specifically we will (1) test what percentage of a community needs to be in a program to achieve social norms change around key neonatal care behaviors, and (2) test whether so-called nominated-friend-targeting, a method that targets individuals who are more highly connected in the network, is more effective than a control strategy. Our 2x8 factorial design will examine how large a subset of the population should be used as a "seed" group in order to maximize the chances of spread of the effect, and the efficiency with which such an intervention might be delivered in the future. To do this, we will assign each of the 176 study villages to either one of the two groups: 1)random assignment (active comparator), where "seed" individuals are chose at random or 2) friend-of-random assignment (experimental), where "seed" individuals are chosen on the basis of being named as a friend of a randomly selected individual. Each of the groups of villages will also be assigned to one of eight treatment percentages (0%, 5%, 10%, 20%, 30% 50%, 75%, 100%), where each represents the percent of targeted households in that village to receive the CBNH health intervention.