Applications of Generalized Linear Mixed Models in sexual networks to understand dynamics of sexually transmitted infections (STIs)
by
Khangelani Zuma
University of Waikato

Population mobility has been identified as one of the many contributing factors in the spread of STIs. Oscillatory labour migration is the most common type of migration in South Africa. Like any other disease, the spread of STIs follow human mobility. Transmission of STIs is possibly clustered within some sexually active groups due to common sexual behaviour. Ignoring clustering in the analyses of sexual network data can produce misleading results. In this talk, we apply Generalized Linear Mixed Model to estimate the dynamics of STI in polygamous sexual network cohorts of migrant men, non-migrant men and their sexual partners. In the model, we take into account unobserved random effects of sexual networks. We use EM-algorithm (Dempster, Laird and Rubin, 1972) to estimate these unobservable random effects of sexual networks. Intractable integrals are approximated with Gaussian-Hermite quadratures. The results show an importance of considering heterogeneity between sexual networks in the model. Thus, any interventions aimed at reducing the spread of HIV and other STIs should be aimed at sexual networks rather than focussing exclusively on individuals.

Date received: May 13, 2002