This umbrella review synthesized six systematic reviews and meta-analyses. The six SRMAs together incorporated 80 primary studies and a total of 43,785 women. Investigators reported that 15,147 women across those primary studies had experienced intimate partner violence. The pooled prevalence estimate across the included SRMAs was 35% (95% CI: 28–42%). Heterogeneity across reviews was substantial (I2 = 83.48%). Short and direct: over one-third of women in the pooled data experienced IPV. Longer context: the review team used random-effects models in STATA to generate pooled prevalence estimates and presented forest plots; the source reported that these quantitative syntheses accounted for between-review variability but left large residual heterogeneity.
The umbrella analysis extracted subgroup estimates that were consistently reported in the SRMAs. For pregnant women, the pooled prevalence of IPV was reported as 32% (95% CI: 26–37%). During the COVID-19 pandemic, the pooled prevalence reported across reviews was 33% (95% CI: 25–40%). These subgroup figures approximate the overall estimate, implying that pregnancy and the pandemic period did not substantially lower the measured burden in the pooled data. The source did not provide additional stratified breakdowns (for example by trimester or severity) in the umbrella summary; readers must consult the individual SRMAs for those details.
The review team synthesized estimates for major forms of violence that were consistently reported across SRMAs. The pooled prevalences were: sexual violence 18% (95% CI: 14–22%), physical violence 19% (95% CI: 15–23%), and psychological violence 27% (95% CI: 23–31%). Psychological forms of abuse thus had the highest pooled prevalence in the available evidence. The umbrella review noted variation across primary studies in how specific forms were defined and measured, which contributes to between-study variation in these pooled estimates.
Across the included SRMAs, several predictors showed consistent, statistically significant associations with IPV. The review authors reported associations for the following factors: partners’ educational status and women’s educational status, partner alcohol consumption, reduced women’s decision-making power within relationships, substance use, and income loss during the COVID-19 pandemic. The SRMA-level syntheses indicated that lower educational attainment and partner alcohol use were repeatedly identified as risk factors. The umbrella review did not reproduce all individual effect sizes or adjusted odds ratios for each predictor within the article text; those numerical details are provided in the underlying SRMAs and their supplementary tables.
The authors constructed a citation matrix to quantify overlap of primary studies across reviews. That matrix identified 50 unique primary studies contributing 80 study occurrences, yielding a Corrected Covered Area (CCA) of 12%—classified as high overlap. The overlap increases the risk that particular primary studies influenced pooled estimates disproportionately. The team assessed methodological quality using AMSTAR-2 across the six SRMAs. While the umbrella review reported overall confidence ratings according to AMSTAR-2 guidance, the source text emphasized variability in methodological rigor across reviews. Because fewer than ten SRMAs were included, the authors did not formally test for publication bias with standard statistical methods, citing limited reliability of such tests with small numbers of reviews.
The umbrella review reiterated that intimate partner violence remains a major public health issue in Ethiopia and framed the findings in relation to national policy commitments, including the five-year National Action Plan (2021/22–2025/26). The source recommended strengthening policy implementation and expanding survivor-centered services. Concrete gaps documented in the article include the high heterogeneity of pooled estimates, the high overlap of primary studies (CCA 12%), and the retrospective PROSPERO registration of the umbrella review protocol. The authors also noted that pooled prevalence estimates should be interpreted with caution because of overlap and methodological variation across SRMAs. The source did not report intervention effect estimates or randomized trial data; it focused on prevalence and associated predictors. Remaining research questions identified in the review include more precise, contemporaneous prevalence estimates stratified by region, standardized measurement of specific IPV forms, and higher-quality primary studies to reduce heterogeneity and overlap.
Limitations of the umbrella review, as stated in the source, included the high heterogeneity across SRMAs, potential partial double counting of primary studies because of overlap, and that publication-bias tests were not performed given the small number of included reviews. The source also noted that some SRMAs varied in methodological transparency, which affected their AMSTAR-2 ratings. These methodological caveats frame the pooled numbers and the predictor associations reported above and suggest caution when applying the pooled prevalences to local program planning.