Intimate partner violence and antenatal care utilization predictors in Uganda: an analysis applying Andersen's behavioral model of healthcare utilization.

BACKGROUND: Optimal utilization of antenatal care (ANC) services improves positive pregnancy experiences and birth outcomes. However, paucity of evidence exists on which factors should be targeted to increase ANC utilization among women experiencing intimate partner violence (IPV) in Uganda. OBJECTIVE: To determine the independent association between IPV exposure and ANC utilization as well as the predictors of ANC utilization informed by Andersen's Behavioral Model of Healthcare Utilization. METHODS: We analyzed 2016 Uganda Demographic and Health Survey data that included a sample of 1,768 women with children aged 12 to 18 months and responded to both ANC utilization and IPV items. Our outcome was ANC utilization, a count variable assessed as the number of ANC visits in the last 12 months preceding the survey. The key independent variable was exposure to any IPV form defined as self-report of having experienced physical, sexual and/or emotional IPV. Covariates were grouped into predisposing (age, formal education, religion, problem paying treatment costs), enabling (women's autonomy, mass media exposure), need (unintended pregnancy, parity, history of pregnancy termination), and healthcare system/environmental factors (rural/urban residence, spatial accessibility to health facility). Poisson regression models tested the independent association between IPV and ANC utilization, and the predictors of ANC utilization after controlling for potential confounders. RESULTS: Mean number of ANC visits (ANC utilization) was 3.71 visits with standard deviation (SD) of ± 1.5 respectively. Overall, 60.8% of our sample reported experiencing any form of IPV. Any IPV exposure was associated with lower number of ANC visits (3.64, SD ± 1.41) when compared to women without IPV exposure (3.82, SD ± 1.64) at p = 0.013. In the adjusted models, any IPV exposure was negatively associated with ANC utilization when compared to women with no IPV exposure after controlling for enabling factors (Coef. -0.03; 95%CI -0.06,-0.01), and healthcare system/environmental factors (Coef. -0.06; 95%CI -0.11,-0.04). Predictors of ANC utilization were higher education (Coef. 0.27; 95%CI 0.15,0.39) compared with no education, high autonomy (Coef. 0.12; 95%CI 0.02,0.23) compared to low autonomy, and partial media exposure (Coef. 0.06; 95%CI 0.01,0.12) compared to low media exposure. CONCLUSION: Addressing enabling and healthcare system/environmental factors may increase ANC utilization among Ugandan women experiencing IPV. Prevention and response interventions for IPV should include strategies to increase girls' higher education completion rates, improve women's financial autonomy, and mass media exposure to improve ANC utilization in similar populations in Uganda.