Assessing methodological differences in estimating exposure to intimate-partner violence against women in Brazil: a multi-source observational analysis.

Background: Intimate partner violence (IPV) is a global public health issue. However, prevalence estimates vary depending on the data source and methodology used. This study aimed to assess methodological differences in estimating the prevalence and age distribution of physical and sexual IPV across multiple Brazilian data sources. Methods: We assessed and analyzed seven Brazilian data sources: five surveys and two administrative datasets. Surveys used either the World Health Organization (WHO) instrument on violence against women and adaptations or an adapted version of the Revised Conflict Tactics Scales. The analysis included only women aged 16 years or older, and race was not examined. Recall periods varied from 12 to 24 months. We estimated prevalence and age distribution of physical and sexual IPV with 95% confidence intervals (CIs). Findings: The seven sources represented national, regional, and local populations. Methodological differences, including questionnaires, interview settings, data collection methods, and recall periods, may explain variation across data sources. Among national-level estimates, physical IPV prevalence ranged from 0.4% (0.3-0.7) to 13.0% (8.3-17.8) across data sources and age strata, with all sources consistently showing higher prevalence and proportions among younger women. For sexual IPV, no age group pattern emerged in prevalence or proportions. Interpretation: This study revealed substantial heterogeneity in estimates of IPV prevalence and age distribution. Without a gold standard for IPV measurement, these findings underscore the need to interpret available estimates in light of the methodological context in which they were generated. Funding: The Gates Foundation.