01 Pan American Journal Review of Public Health 02 03 04 05 06 Intimate partner violence in the Americas: a systematic 07 08 review and reanalysis of national prevalence estimates 09 10 11 1 2 1 1 Sarah Bott, Alessandra Guedes, Ana P. Ruiz-Celis, and Jennifer Adams Mendoza 12 13 14 15 Suggested citation Bott S, Guedes A, Ruiz-Celis AP, Mendoza JA. Intimate partner violence in the Americas: a systematic review and reanalysis 16 of national prevalence estimates. Rev Panam Salud Publica. 2019;43:e26. https://doi.org/10.26633/RPSP.2019.26 17 18 19 20 ABSTRACT Objectives. To describe what is known about the national prevalence of intimate partner violence (IPV) against 21 women in the Americas across countries and over time, including the geographic coverage, quality, and com- 22 parability of national data. 23 Methods. This was a systematic review and reanalysis of national, population-based IPV estimates from 24 1998 – 2017 in the Americas. Estimates were reanalyzed for comparability or extracted from reports, including IPV prevalence by type (physical; sexual; physical and/or sexual), timeframe (ever; past year), and perpetrator 25 (any partner in life; current/most recent partner). In countries with 3+ rounds of data, Cochran-Armitage and 26 Pearson chi-square tests were used to assess whether changes over time were significant (P < 0.05). 27 Results. Eligible surveys were found in 24 countries. Women reported ever having experienced physical and/ 28 or sexual IPV at rates that ranged from 14% – 17% of women in Brazil, Panama, and Uruguay to over one-half 29 (58.5%) in Bolivia. Past-year prevalence of physical and/or sexual IPV ranged from 1.1% in Canada to 27.1% 30 in Bolivia. Preliminary evidence suggests a possible decline in reported prevalence of certain types of IPV in 31 eight countries; however, some changes were small, some indicators did not change significantly, and a sig- 32 nificant increase was found in the reported prevalence of past-year physical IPV in the Dominican Republic. 33 Conclusions. IPV against women remains a public health and human rights problem across the Americas; 34 however, the evidence base has gaps, suggesting a need for more comparable, high quality evidence for 35 mobilizing and monitoring violence prevention and response. 36 37 38 Keywords Intimate partner violence; domestic violence; violence against women; surveys and questionnaires; Latin 39 America; Caribbean region; Americas. 40 41 42 43 Violence against women (VAW) has been recognized as In 2015, the United Nations (UN) Member States agreed 44 an important public health and human rights problem, both to work toward eliminating VAW as part of 2030 Sustainable 45 globally (1) and within the Region of the Americas (2). Intimate Development Goals (SDGs) (7). Member States of the Pan 46 partner violence (IPV)—the most common form of VAW—has American Health Organization (PAHO) and the World Health 47 serious consequences for women’s health and wellbeing (3). In Organization (WHO) made similar commitments as part 48 a 12-country analysis from the Region (4), large proportions of of the PAHO 2015 Strategy and Plan of Action on VAW (8) 49 women who experienced IPV reported consequences such as and the WHO 2016 Global Plan of Action on Interpersonal 50 physical injuries, chronic pain, anxiety, depression, and suicidal Violence (9). Countries also agreed to strengthen data collec- 51 thoughts. In most countries, IPV was significantly correlated tion systems and measure SDG Indicator 5.2.1: the proportion 52 with lower age at first union, higher parity, and unintended of ever-partnered women and girls 15+ years of age subjected 53 pregnancy. IPV also has well-documented negative conse- to physical, sexual, or psychological violence by a current or 54 quences for children and the broader society (5, 6). former intimate partner in the previous 12 months. 55 56 1 Independent consultant to the Pan American Health Organization, Washington, 2 Pan American Health Organization, Washington, D.C., United States of 57 D.C., United States of America. * Alessandra Guedes, [email protected] America. 58 59 This is an open access article distributed under the terms of the Creative Commons Attribution-NonCommercial-NoDerivs 3.0 IGO License, which permits use, distribution, and reproduction in any medium, provided the 60 original work is properly cited. No modifications or commercial use of this article are permitted. In any reproduction of this article there should not be any suggestion that PAHO or this article endorse any specific organization or products. The use of the PAHO logo is not permitted. This notice should be preserved along with the article’s original URL. N61 Rev Panam Salud Publica 43, 2019 | www.paho.org/journal | https://doi.org/10.26633/RPSP.2019.26 1 Review Bott et al. • Partner violence in the Americas The number of countries with national IPV prevalence esti- • Explicitly mentioned ‘partners’ in preambles or survey items mates has grown recently (10), but data are not always easy to measuring violence. find, comparable across countries or over time, or published in full (4). Databases of the UN Minimum Set of Gender Indi- Eligible surveys collected data from January 1998 – cators (11) and the SDGs (12) have begun compiling national December 2017 and published findings by 15 July 2018. The estimates, but these come primarily from Demographic and timeframe was expanded after work began, so database Health Surveys (DHS) and are often limited to IPV in the past searches were updated in July 2018. Peer-reviewed journal 12 months, due to the formulation of SDG Indicator 5.2.1. More- publication was not required because national survey find- over, published estimates are often constructed in diverse ways ings do not always reach journals in a timely manner, if at all. regarding age, partnership status, and forms of violence (10). Urban-only studies were included to allow wider geographic As a result, researchers and policy makers may lack access to coverage. Crime Victim Surveys (14) were excluded because comparable IPV estimates, even when data exist. they ask about violence by any perpetrator without explicitly This study aims to describe what is known about the national mentioning partners—an approach known to underestimate prevalence of IPV against women in the Americas across coun- prevalence (15). However, to ensure adequate geographic cov- tries and over time, including geographic coverage, quality, erage, surveys that explicitly mentioned partners in preambles and comparability of data. A systematic review was carried or survey questions were considered eligible, even if they asked out along with a comparative reanalysis of national, popula- about violence by ‘family members’ or ‘any man.’ If published tion-based, IPV prevalence estimates from PAHO Member reports provided inadequate information about methods or States. In addition, changes over time were analyzed in coun- operational definitions, the information was sought directly tries with 3+ rounds of comparable data collection. To conclude, from the authors/researchers. In four cases (16 – 19), the attempt recommendations for improving measurement and dissemina- to get more detail was not successful, so the surveys were tion are presented. excluded. MATERIALS AND METHODS Most recent IPV estimates Per PRISMA guidelines (Figure 1), a systematic search for For the most recent eligible survey in each country, a sec- nationally representative, population-based surveys with IPV ondary analysis of IPV prevalence was carried out by type data from PAHO Member States was carried out in duplicate (physical; sexual; or physical and/or sexual); timeframe (ever; (by SB and AR) using terms such as ‘intimate partner violence,’ or past year); and perpetrator (any partner in life; or current/ ‘violence against women,’ ‘domestic violence,’ ‘spouse abuse,’ most recent partner—‘current’ for women with a partner and ‘prevalence,’ ‘national survey,’ and country names. The search ‘most recent’ for those separated, divorced, or widowed). was performed on SciELO (Latin American and Caribbean Cen- Emotional/psychological IPV was not reanalyzed given the ter on Health Sciences Information, São Paulo, Brazil), LILACS enormous diversity of measures across surveys in the Region (Latin American and Caribbean Center on Health Sciences and the lack of international consensus on definitions (3). Information, PAHO/WHO, São Paulo, Brazil), PubMed Central When datasets were open-access, estimates with confi- (U.S. National Library of Medicine, Bethesda, Maryland, United dence intervals (CIs) were reanalyzed for comparability (by States), and Google Scholar (Google Inc., Mountain View, Cali- JM or AR) using SPSS Statistics for Windows, Version 20 (IBM fornia, United States); the databases of UN Women (13), SDGs Corp., Armonk, New York, United States), SAS 9.1 (SAS Insti- (12), the Global Health Data Exchange (GHDx), Reproductive tute Inc., SAS Campus Drive, Cary, North Carolina, United Health Surveys (RHS), DHS, and websites of national institutes States), or Stata Statistical Software®/MP14 (StataCorp LP, of statistics (or similar agencies) in each country. Bibliographies College Station, Texas, United States). Sample weights were of global and regional reviews were manually searched, and applied to adjust for sampling design and non-response differ- more than 100 researchers and government officials throughout entials when available. Analyses were reviewed by all authors the Region were contacted. After screening 1 046 records (once and shared with original research teams, who often provided duplicates were removed), 133 records were selected for full technical assistance. When microdata were unavailable or not text review. Eligibility was independently assessed by at least feasible to reanalyze, the original researchers were contacted two authors (SB, AR, JM); differences of opinion were resolved to request estimates reanalyzed for comparability. Otherwise, by consensus among all authors. estimates were extracted from published reports in duplicate by A priori inclusion criteria were: at least two authors (SB, AR, JM) and confirmed with country teams when possible.