HIV and STIs in Young Adult Networks
Source: NICHD , 5R01HD038210
Active: 07/01/99 - 07/01/04
Investigator: Martina Morris, PI
Mark S. Handcock (University of Washington)
Joseph L. Schafer (Penn State University)
This proposal seeks funding to support analyses of the biomarkers for sexually transmitted infections (STIs) that will be collected in the forthcoming wave (Survey 2000) of the National Longitudinal Study of Adolescent Health (Add Health). The central aim of Survey 2000 is to document the health status of young adults (18-26 years old), and provide data that make it possible to analyze the relationships between social context, behavior, and health outcomes. Specimens will be tested for four STIs in Survey 2000: Chlamydia trachomatis (Ct), Neissera gonorrhea (GC), Trichomonas vaginalis (Tv), and human immunodeficiency virus (HIV). The analyses proposed in this project will focus on three interrelated STI topics: 1) basic descriptive work on STI prevalence; 2) the development of multiple imputation strategies for missing behavioral and biomarker data; 3) the measurement and analysis of epidemiologically relevant aspects of network location and structure; and 4) multivariate analyses of individual STI risk that explicitly integrate measures of network exposure. The prevalence anslyses will provide the first detailed population based estimates of STIs in this age group. Gonorrhea and chlamydia are reportable, and their prevalence is estimated to peak during the age range of the Survey 2000 sample. Trichomoniasis is not reportable, but it is thought to peak during a similar age range. Prevalence of HIV among young adults is not well established. The size of the Survey 2000 sample (N=19,500) will make it possible to estimate prevalence of each of these STIs for detailed race, sex, region, and other subgroups for the first time. All prevalence analyses will implement existing statistical corrections for the sensitivity and specificity of the tests. We estimate that the biomarker collection will have a non-response rate of about 15%. We will therefore develop multiple imputation strategies for the missing data. Comparable strategies have been developed successfully for the NHANES III. There is a wealth of both cross-sectional and longitudinal data in Add Health that can be used to construct an effective imputation scheme. We will produce a public use data set containing the imputed variables, and instructions on the use of existing statistical software for implementing a multiple imputation analysis, available electronically through the Add Health project website. The network and multivariate analyses that comprise the third and fourth foci of this project will take advantage of the unique local network data that will be collected on respondents' sexual partners. STIs travel through networks of sexual partnerships, and Survey 2000 will provide the first nationally representative data on local networks for this highly active age group. Local network data consist simply of respondent reports about their partners, rather than tracing and enrolment of the partners themselves. While relatively simple to collect, these data provide a rich source of information on network exposure, mixing, and concurrent partnerships that can be incorporated into traditional epidemiological models of infection risk. The goal of these analyses will be to identify the relative contribution of individual attributes and network exposure to the risk of infection, and to establish the implications for prevention.