Lack of agency and sexual behaviors among married women: a study of population-based HIV impact assessment (PHIA) surveys in Malawi, Tanzania and Zambia

Abstract

# Background Educational campaigns to prevent HIV/AIDS have shown mixed success in Africa. We hypothesized that women’s lack of agency in decision-making and taking discernible actions reduces the beneficial impact of HIV-related knowledge. # Methods We used data from Population-based HIV Impact Assessment (PHIA) surveys in Malawi, Tanzania and Zambia. A subsample of surveys that were completed by married adult women were selected for the HIV knowledge module which included responses for household decision-making questions. We created a binary variable for agency (ability to participate in decision-making about household matters, health, and sex), and scalar variables for HIV-related knowledge and sexual behaviors. We used regression analysis using survey weights with the behaviors as dependent variables. # Results We pooled survey results from 16,822 women (63% from Tanzania, 19% from Malawi and 18% from Zambia). Altogether, 13.5% of women (17% of those 15-24 years old) exhibited poor agency. Those with a higher degree of agency had higher education, were working, and were wealthier. Women lacking agency were significantly less likely to use condoms (4.7% vs. 6.2%, *P*=0.022). Approximately 95% reported having a single partner over past 12 months, while Zambian women with no agency showed significantly lower rate of 92% (odds ratio, OR=0.66, 95% confidence interval, CI=0.45-0.95, *P*=0.027). The rate of HIV testing across the three countries was 83%-92%. In Tanzania, presence of good knowledge and agency together increased the odds of HIV testing by 60% (interaction term). However, among those with good knowledge yet reporting poor agency, the likelihood of HIV testing decreased by 65%, nullifying knowledge impact. We did not observe similar associations in Malawi. Among women with poor knowledge, lack of agency reduced the odds of condom use by 50% compared to those with agency. Alternatively, for women who exhibited good knowledge without agency, the odds of condom use was more than double compared to women with some agency (OR=2.14, 95% CI=1.07-4.27, *P*=0.032)). # Conclusions We detected a moderating effect for agency on knowledge-driven behavior. Results on different behaviors and across individual countries are mixed and suggest caution to derive definitive conclusions. Despite limitations, these findings indicate that policies that increase women’s agency may help anti-HIV programs’ success.