Behavioral clustering in adolescents with STDs

点击翻译按钮获取中文摘要

DESCRIPTION (provided by applicant): Though the risk for STDs among adolescents is well described, knowledge of their sexual, drug-using and social networks is less well developed. In addition, the role of behavioral and psychosocial factors in transmission of STDs and blood borne infections (BBIs) has been considered primarily on the individual level, and rarely as group phenomena. In this study, we will explore whether the networks of young persons demonstrate assortative mixing with regard to behaviors and psychosocial characteristics as a function of infection with STDs and BBIs. Our specific aims are (1.1) to compare the egocentric networks of infected and non-infected boys and girls with regard to specific behaviors, psychosocial characteristics, and types of associations; (1.2) to compare the sociograms (connected subgraphs) created by joining index respondents who have ties in common; (1.3) to compare the changes in egocentric networks and sociograms among adolescent boys and adolescent girls over time; (1.4) to estimate the relative importance of behavioral/psychosocial factors vs. network factors in the acquisition of STDs/BBIs; and (2.0) to examine the extent to which assortative or disassortative mixing with regard to personal behaviors and psychosocial factors in groupings of adolescents is associated with network structure and with the occurrence of STDs/BBIs. We propose enrolling 20 boys and 20 girls, age 15 years (10 each infected and uninfected), interviewing and testing them for STDs/BBIs six times over a 3 year period, and, at every other interview, eliciting contacts according to a fixed chain link design (total N=440 at every other interview). The resulting clusters and groupings will provide the basis for testing hypotheses that relate to the specific aims, primary among which are that network structure will exert an independent influence on the probability of infection, and that "infected clusters" (i.e. those in which the index respondent is infected) will demonstrate a greater degree of assortative mixing with regard to behavioral and psychosocial variables than uninfected clusters. For this purpose, we will take advantage of the fact that the intraclass correlation coefficient and the design effect derived from it measure intracluster homogeneity and are thus a direct measure of assortative mixing within clusters. Demonstration of such phenomena will be of use in designing interventions that incorporate network information into their formulation.