20th International AIDS Conference - Melbourne, Australia


THAC0504 - Oral Abstract

Weighing 17 years of evidence: does hormonal contraception increase HIV acquisition risk among Zambian women in discordant couples?

Presented by Kristin Wall (United States).

K. Wall1,2, W. Kilembe3, H.K. Naw2, I. Brill4, B. Vwalika3,5, E. Chomba3,6, B. Johnson7, L. Haddad3,8, A. Tichacek2, S. Allen2

1Emory University, Department of Epidemiology, Atlanta, United States, 2Rwanda Zambia HIV Research Group, Department of Pathology & Laboratory Medicine, Atlanta, United States, 3Rwanda Zambia HIV Research Group, Lusaka, Zambia, 4University of Alabama at Birmingham, Department of Epidemiology, Birmingham, United States, 5University of Zambia, Department of Gynecology and Obstetrics, Lusaka, Zambia, 6Ministry of Community Development, Mother and Child Health, Lusaka, Zambia, 7Emory University, Department of Biostatistics, Atlanta, United States, 8Emory University, Department of Gynecology and Obstetrics, Atlanta, United States

Background: The effect of hormonal contraception (HC) on risk of HIV transmission and acquisition is a debated, priority public health issue. Stakeholders, including USAID, FHI360, and WHO, have called for more rigorous evaluations of these associations. WHO has further urged researchers to assess contraceptive implant use and HIV risk, about which little data exists.
Methods: From 1995-2012, HIV discordant (M+F-) couples identified through couples'' voluntary HIV counseling and testing in Lusaka, Zambia were enrolled and followed longitudinally. Baseline and time-varying demographic, behavioral, and clinical exposures were collected, including self-reported contraceptive use. Women were re-tested quarterly. The association between HC and time to HIV acquisition among women was evaluated by multivariate Cox models. Sensitivity analyses explored varying contraceptive method exposure categorizations and control groups, misclassification of unprotected sex, and mediation by time-varying confounders.
Results: Among 1393 couples, 252 incident infections occurred over 2839 couple-years (8.9 infections/100 couple-years; 95%CI:7.8-10.0). 207 (82%) infections were genetically linked to the study partner. No effect modification was observed by genital ulceration/inflammation or woman''s age; no interaction between genital ulceration/inflammation and contraception was observed. Use of injectables (HR=1.1;95%CI:0.7-1.7), OCPs (HR=1.3;95%CI:0.9-1.9), or implants (HR=1.0;95%CI:0.5-2.2) in the past three months was not associated with HIV acquisition relative to non-HC, controlling for incident pregnancy, family income, presence of sperm on a wet mount, time interval since enrollment, and woman''s: age, language ability, number of sexual partners, alcohol use, and genital ulceration. Use of injectables (HR=1.0;95%CI:0.6-1.5), OCPs (HR=1.2;95%CI:0.8-2.0), or implants (HR=0.9;95%CI:0.3-2.6) in the past three months was not associated with genetically linked HIV acquisition relative to non-HC controlling for the above factors, couples'' unprotected and protected sex, and man´s baseline: number of sexual partners, alcohol use, log viral load, and circumcision status. Sensitivity analyses produced similar results.
Conclusions: We found no association between HC and HIV acquisition risk among women over 17 years of follow-up after thorough consideration of confounding, misclassification, effect measure modification, interaction, and mediation. These findings add to a body of controversial evidence important for discussion at the upcoming WHO Technical Meeting regarding restrictions on HC methods for women with and at-risk of HIV.

Back to the Programme-at-a-Glance