HIV transmission during oral sex: Simulation assessment of analysis approaches.
Li, Tsai-Chung
1994
Abstract
Oral sex is perceived as safe by some gay male populations and their belief in its safety is reinforced by multivariate analyses which show no effect of oral sex on the transmission of HIV. Consequently some gay male populations engage in more oral intercourse now than in the mid-1980s or before. However, previous analyses of the risk factors of HIV infection could be biased because they assume that the outcome of each study subject is independent of those of other study subjects. In fact, there is strong dependency between the outcomes of different individuals because contacts between individuals are the entities on which risk factors for infection exert their effects. Therefore, it is important to evaluate the performance of the multivariate procedures used in estimating the effects of behaviors on risks of HIV transmission and to develop methods to overcome the inconsistencies between the assumptions made in standard multivariate procedures and the transmission of HIV. The method we developed to help overcome the inconsistencies of standard analyses with transmission dependencies uses derived exposure variables which account for the likelihood of encountering an infected partner. To evaluate the performance of standard methods and of analyses using these new derived variables, we generated data from a discrete event simulation model using GPSS/H language that roughly corresponds to the data of the joint Chicago Multicenter AIDS Cohort Study and Coping and Change Study (MACS/CCS). We applied standard methods to the simulation data where the rear parameter values are known as well as to the MACS/CCS data where they are not known. We found the logistic regression using standard variables to be quite faulty as demonstrated by inconsistencies between the size of the parameter estimates of the logistic regression and the number of transmission instances during oral sex which were documented during the simulations. While 30% of transmission episodes in the simulations were due to oral sex, standard logistic regression analysis showed only weak and inconsistent effects of oral sex. Additionally, the analysis characterized some exposure variables as having protective effects when, in fact, they generated more transmission instances than those having positive effects. The use of exposure variables weighted by the risk of infection in partners resulted in an analysis which reflected the effects of exposure variables much more accurately during the first seven years of the simulation data and somewhat more accurately during the second seven years of simulation. We explain this difference by decreased ability of the derived variables to capture the risk of contagiousness in partners during the second seven years of simulation because of divergence between the overall rate of infection and the rate of highly contagious primary infection. This divergence should have been even greater in the CCS data, and the effectiveness of the derived variables would be further diluted in that data by sample biases and variable definition problems. In fact, the derived variables did not change the effects of exposure variables for the CCS data. Thus we conclude that the weak effects of oral sex variables in the analysis of the CCS data is compatible with a fairly high rate of transmission from oral sex acts which is hidden by inconsistencies between the assumptions of the analytic methods and the forces that generated the data.Other Identifiers
(UMI)AAI9513413
Subjects
Biology, Biostatistics Health Sciences, Public Health
Types
Thesis
Metadata
Show full item recordCollections
Remediation of Harmful Language
The University of Michigan Library aims to describe library materials in a way that respects the people and communities who create, use, and are represented in our collections. Report harmful or offensive language in catalog records, finding aids, or elsewhere in our collections anonymously through our metadata feedback form. More information at Remediation of Harmful Language.
Accessibility
If you are unable to use this file in its current format, please select the Contact Us link and we can modify it to make it more accessible to you.