We will assess peer change agent communication effectiveness through mathematical models that utilize change agent attribute and structural position parameters developed through existing digital communication networks. Peer change agent (PCA)-based HIV prevention programs capitalize on the phenomenon of peer influence and how members in social networks communicate with one another. PCA-based interventions are the most frequently used outreach HIV prevention interventions and have been widely implemented both domestically and internationally. Existing PCA-based interventions, however, have yielded disappointing results in international settings. The central postulate of this proposa is that the change agent can often be more important than the message itself. Oftentimes, the messages promoted by public health officials may be of limited interest to others, even to those at increased HIV risk. In fact, when messages are of limited interest, those at increased HIV risk will tend to focus more on who the change agent is. The selection of change agents based in whole or in part upon their structural position as measured by formal social network characterization is one approach to increase potency of peer influence: it has been successful in business organization-based interventions. It has not, however, been empirically tested in the public health realm. By adopting this methodology, this proposal moves beyond traditional peer outreach models. In addition to utilizing the structural network position of PCAs to enhance the diffusion of innovation, our model also parameterizes important attributes of PCAs, which are determined by primary data collection from a large network of men who have sex with men in South India. These PCA attributes include social status features, leadership and communication behavior, tie qualities with members of their network, and physical attributes. Previously we have developed an innovative approach to social network characterization in this setting that allows for augmented network characterization through the use of cell phone communication networks in high- risk men. We have field survey data collection expertise in India and experience with large scale social network data collection and analyses. We also have experience in diffusion of innovation analyses through social networks and utilizing mathematical models for determining diffusion of disease and/or information through networks of men who have sex with men (MSM). Thus we will model condom communication within the network using attributes generated from participant interviews and social network-generated structural positions. Candidate PCAs will be selected using three classes of algorithms: (1) attribute-based; (2) position- based; and (3) combined attribute- and position-based. Flow of condom information in the network will be explored using multiple stochastic models that address missing data and are parameterized using self- and peer-evaluation data. In future work, we will test how novel bio-behavioral prevention interventions, diffuse through a network facilitated by strategically selected peer change agents.

R21AI098599

2013-03-01

2016-02-29