New infection transmission models for populations with heterogeneous contact patterns will be used to 1) estimate HIV transmission probabilities, 2) define the data needs of proposed AIDS related studies, and 3) analyze epidemic patterns and intervention effects. This combined statistical and causal modeling approach differs from other transmission modeling efforts in that it focuses on the pattern of infection flow through a population rather than on just overall infection levels. Modeling the contact patterns between subgroups of individuals with different characteristics is central to this effort. Two determinants of sexual contact patterns, needle contact patterns, or combined patterns are modeled: 1) the social setting of partnership formation, 2) selectively in the choice of different partner types. These contact models will be used within other models to estimate transmission probabilities during specific sex acts with individuals in different stages of infection. Three estimation procedures will be used: a maximum likelihood estimation procedure which uses a probability model for data on individuals, a weighted least squares fitting procedure which uses differential equation models of the aggregate data, and a hybrid of these. Data on homosexual populations from the Coping and Change Study (CCS) of the Chicago MACS will be used. Heterosexual transmission could also be analyzed. All three procedures overcome the deficiencies of standard multivariate statistical procedures by a) providing a causally meaningful model form with interpretable parameters, b) defining dependencies among the outcomes in different study subjects instead of assuming independence, and c) permitting the estimation of transmission probabilities from data which specify characteristics of a contract but not necessarily the infection status of the person contacted. The models will be used to test the sensitivity of transmission parameter estimates to assumptions about contact patterns and thus specify critical information needed from prospective HIV transmission studies. Besides statistical estimation, mathematical analysis and simulation of epidemics will be carried out. The population studied will include transmission through homosexual acts, needles, and/or heterosexual acts. These analyses will: 1) assess which specific sexual behavior and contact questions should be included in proposed national surveys in order to achieve scientific understanding or make public policy decisions, 2) combine the information to be collected in various surveys on risk behaviors with information on transmission and natural history of infection so that the effect of behaviors and contact patterns on transmission can be assessed, and 3) assess the potential effects of a variety of infection control strategies.
| Jacquez, J A; Simon, C P (1993) The stochastic SI model with recruitment and deaths. I. Comparison with the closed SIS model. Math Biosci 117:77-125 |
| Jacquez, J A; O'Neill, P (1991) Reproduction numbers and thresholds in stochastic epidemic models. I. Homogeneous populations. Math Biosci 107:161-86 |
