Violence motivated by perception of sexual orientation and gender identity: a systematic review

Abstract Objective To assess the prevalence of physical and sexual violence motivated by perception of sexual orientation and gender identity in sexual and gender minorities. Methods We searched nine databases without language restrictions for peer-reviewed and grey literature published from 2000 to April 2016. We included studies with more than 50 participants that measured the prevalence of physical and sexual violence perceived as being motivated by sexual orientation and gender identity or gender expression. We excluded intimate partner violence and self-harm. Due to heterogeneity and the absence of confidence intervals in most studies, we made no meta-analysis. Findings We included 76 articles from 50 countries. These covered 74 studies conducted between 1995 and 2014, including a total of 202 607 sexual and gender minority participants. The quality of data was relatively poor due to a lack of standardized measures and sometimes small and non-randomized samples. In studies where all sexual and gender minorities were analysed as one population, the prevalence of physical and sexual violence ranged from 6% (in a study including 240 people) to 25% (49/196 people) and 5.6% (28/504) to 11.4% (55/484), respectively. For transgender people the prevalence ranged from 11.8% (of a subsample of 34 people) to 68.2% (75/110) and 7.0% (in a study including 255 people) to 49.1% (54/110). Conclusion More data are needed on the prevalence, risk factors and consequences of physical and sexual violence motivated by sexual orientation and gender identity in different geographical and cultural settings. National violence prevention policies and interventions should include sexual and gender minorities.


Introduction
On 17 June 2011, the United Nations (UN) Human Rights Council passed a resolution that expressed grave concern at violence and discrimination against individuals based on their sexual orientation and gender identity. 1 This first-ever UN resolution on sexual orientation and gender identity requested a report by the Office of the High Commissioner for Human Rights. It was published in November 2011 and stated: "Homophobic and transphobic violence has been recorded in all regions. Such violence may be physical (including murder, beatings, kidnappings, rape and sexual assault) or psychological (including threats, coercion and arbitrary deprivations of liberty). These attacks constitute a form of gender-based violence, driven by a desire to punish those seen as defying gender norms." 2 An updated 2014 resolution confirmed these conclusions and culminated in the designation of an UN Independent Expert on sexual orientation and gender identity in September 2016. [3][4][5] Although the UN recognized violence against individuals based on their sexual orientation and gender identity as a form of gender-based violence, we do not know whether such violence is characterized by the same gender dynamics and motivations as gender-based violence against women or if it follows a different path. [6][7][8][9] Violence against individuals based on their sexual orientation is one of the ways in which sexual stigma is expressed. 10 Sexual stigma based on perceived sexual orientation emerges from a society's shared belief system in which homosexuality is denigrated and discredited as invalid relative to heterosexuality. Stigma based on gender identity works along the same lines of a gendered society in which only two gender possibilities, masculine or feminine, are perceived as valid. This stigma is incorporated by a society and enacted by its institutions. 10 In many countries, for example, laws criminalize sexual and gender minorities directly or indirectly on the grounds of morality or promotion of non-traditional values. This can result in physical punishment, death penalty, arbitrary arrest and torture, ill-treatment in health facilities and forced sterilization. [11][12][13] Discriminatory health policies have also resulted in unnecessary gender-conformation operations in intersex babies. 14 Individuals identified as sexual and gender minorities (Box 1) and may internalize the negative attitudes and values of society. This internalized homophobia or transphobia has detrimental effects on their mental health and might result in self-harm or violence among individuals. [15][16][17] Several comprehensive reviews have demonstrated that sexual and gender minorities are more likely to be victims of physical and sexual violence than the general population. [18][19][20][21] However, these did not report whether the victims perceived the violence being against their sexual orientation and gender identity. Our study aimed to review the research evidence on the prevalence of physical and sexual violence motivated by perception of sexual orientation, gender identity or gender expression among sexual and gender minorities. We distinguished this from violence inflicted on a random member Objective To assess the prevalence of physical and sexual violence motivated by perception of sexual orientation and gender identity in sexual and gender minorities. Methods We searched nine databases without language restrictions for peer-reviewed and grey literature published from 2000 to April 2016. We included studies with more than 50 participants that measured the prevalence of physical and sexual violence perceived as being motivated by sexual orientation and gender identity or gender expression. We excluded intimate partner violence and self-harm. Due to heterogeneity and the absence of confidence intervals in most studies, we made no meta-analysis. Findings We included 76 articles from 50 countries. These covered 74 studies conducted between 1995 and 2014, including a total of 202 607 sexual and gender minority participants. The quality of data was relatively poor due to a lack of standardized measures and sometimes small and non-randomized samples. In studies where all sexual and gender minorities were analysed as one population, the prevalence of physical and sexual violence ranged from 6% (in a study including 240 people) to 25% (49/196 people) and 5.6% (28/504) to 11.4% (55/484), respectively. For transgender people the prevalence ranged from 11.8% (of a subsample of 34 people) to 68.2% (75/110) and 7.0% (in a study including 255 people) to 49.1% (54/110). Conclusion More data are needed on the prevalence, risk factors and consequences of physical and sexual violence motivated by sexual orientation and gender identity in different geographical and cultural settings. National violence prevention policies and interventions should include sexual and gender minorities.
of the general population or violence experienced by sexual and gender minorities, but not specifically perceived to be motivated by their sexual orientation or gender identity.

Methods
Our review followed the Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines. 22 The protocol for this review has not been registered on the PROSPERO register of systematic reviews, but is available on request.
We searched nine bibliographic databases (PubMed®, Embase®, Web of Science, Africa Wide Information, CINAHL, LILACS, Popline, Sociological Abstracts and GenderWatch) for articles published from 1 January 2000 to 28 April 2016. We used a combination of medical subject headings and text words (Box 2), with no language restrictions. These searches were supplemented by a scan of the citations in the articles for studies not found in the search and by consultation with individual experts about their knowledge of other studies.
Studies were eligible for inclusion if they included people belonging to a sexual or gender minority. We included both peer-reviewed and grey literature reporting studies that measured the prevalence of physical and sexual violence perceived as being motivated by sexual orientation, gender identity or gender expression. We excluded intimate partner violence and self-harm. Studies had to be published from 2000 to the search date, refer to data collected after 1995 and include at least 50 participants.
Two researchers screened the identified abstracts. When there was doubt or disagreement about whether an article met the inclusion criteria, the article was taken to the next stage of screening. The researchers then independently assessed the full text of potentially eligible studies. If needed, we contacted the authors of the articles for further information.
After initial screening, we appraised the included studies for quality. The criteria were: sampling method, sample representativeness, description of the population, completeness of the data, description of the methods, reliability of the data, and controls for confounding. We categorized studies as high quality if six to seven criteria were adequate, medium quality if three or five criteria were adequate and low quality if none to two criteria were adequate. None of the studies were excluded based on this quality assessment. We minimized publication bias across studies by including grey literature and consulting with experts.
Two researchers independently extracted details of the studies into a database. The data collected were: country and area; data collection period; study type and sampling method; description of study population; terminology of violence used to elicit responses from participants; time periods of experiences of violence (ever in lifetime, specific dates or time periods); participants' perceptions of motivation for violence; sample size; and number and percentage of respondents affected by different types of violence. The outcome of interest for the review was the prevalence of physical and sexual violence motivated by perception of sexual orientation and gender identity. However, such violence was not the primary outcome in most of the studies.
We made a descriptive summary of the prevalence data in tables and charts. Although the UN resolution 2 included sexual violence within physical violence, most studies reported them separately. Where possible and relevant, we conducted separate descriptive analyses of subgroups of sexual and gender minorities. The results of the studies were highly heterogeneous, due to vari-Box 1. Definitions used in the systematic review of physical and sexual violence motivated by perception of sexual orientation and gender identity ability in the sampling (definition of the population and sampling methods) and the descriptions of violence used to gather data from participants. In view of this heterogeneity and the absence of confidence intervals in most studies (reported in only six), we did not attempt a meta-analysis.

Study selection
Our literature search yielded 10 601 references, of which 8233 were unique entries. Next, we excluded 8000 articles after screening titles and abstracts. Of the 233 references that potentially met the inclusion criteria, nine could not be retrieved, and of the 224 retrieved texts, 185 were excluded for different reasons (Fig. 1). We added 37 articles and reports after citation tracking and consulting with experts. In total 76 articles were included in the review. 24-100 Seven articles were categorized as low quality, 55 as medium and 14 as high quality (Table 1).

Study characteristics
Of the 76  Thirty-six publications were peerreviewed articles, 38 were study reports, one was a dissertation and one a book chapter.
The 76 articles were based on 74 studies conducted between 1995 and 2014, including a total of 202 607 sexual and gender minorities participants. Sixty-three studies used a convenience sample, four used respondent-driven sampling, four used venue-based or timelocation sampling, one random digit dialling and two used mixed methods ( Twenty-six studies included all sexual and gender minorities, of which eight were exclusively high-school students. Thirteen included homosexual and bisexual participants, of which five focused only on younger participants (maximum age 25 years) and one only on older participants (minimum age 60 years). Thirteen studies included homosexual or bisexual men and of these 8 targeted specific groups: bisexual men (4 studies); Latino men (1 study); homosexual or bisexual men aged < 29 years (1 study); seronegative homosexual or bisexual men aged 15-30 years (1 study); and non-white seropositive homosexual or bisexual men aged 13-24 years (1 study). Homosexual or bisexual women were exclusively sampled in four studies, of which three targeted specific groups: bisexual women (2 studies) and sexual minority women of non-androgynous identity (1 study). One study sampled young people who experienced samesex attraction and another included the same study group together with young people who questioned their gender. One study sampled homosexual or bisexual men and male-to-female transgender people. Fifteen studies were of transgender people, of which five studies were specific groups only: male-tofemale transgender people (3 studies), male-to-female individuals who were sex workers (1 study); and transitioning transgender people (1 study).
The descriptions and definitions of physical and sexual violence motivated by perception of sexual orientation and gender identity that were used to elicit participants' responses varied across studies (Box 3). These included the victim's perception of the motivation of the violence and the types of violence Fifty-seven studies asked about experiences of violence ever in the respondent's lifetime. Six studies specified experiences over certain stages of the lifetime: from 13 years old (1 study); ages 6-10 years, 11-17 years and 18+ years (1 study); 1 year ago, 1-10 years ago and > 10 years ago (1 study); age 12-18 years and 18+ years (2 studies); or childhood versus adulthood (1 study). Other studies asked about experiences over specific time periods: 5 years (1 study); 2 years (1 study); 12 months (21 studies); 6 months (3 studies); or 1 month (2 studies). Ten studies asked about experiences in school: past year in school (7 studies); ever in school (1 study); during high-school years (1 study); and since the beginning of the school year (1 study). Some articles measured violence experienced both over the lifetime and over certain periods.

Prevalence of violence
A total of 57 studies provided data on the lifetime prevalence of any kind of physical violence motivated by perception of sexual orientation and gender identity ( Table 2). Fig. 2 summarizes the data for 51 studies, according to the different populations and the attacker's motivation as perceived by the victim (sexual orientation, gender identity or both). In 14 studies where all sexual and gender minorities were taken together the prevalence ranged from 6% in a study of 240 people 98 to 25.0% (49/196). 26 When transgender people were not included (11 studies) the figures ranged from 3.3% (24/761) 100 to 31.1% (436/1402). 45 In homosexual or bisexual men (29 studies), the prevalence was between 8.5% in a study of 603 people 24 and 34.7% (318/916), 45 although when only bisexual men were included (4 studies), the prevalence was no higher than 22.6% (12/53). 37 A similar tendency was observed in homosexual or bisexual women (21 studies), with a prevalence range from 4.6% in a study including 909 individuals 75 to 25.1% (42/167 people), 30 and a lower prevalence when bisexual women only were included (4 studies). For transgender people prevalence (28 studies) ranged from 11.8% of a sample size of 34 91 to 68.2% (75/110 people). 47 There was no pattern of prevalence for the perceived motivation of the violence (sexual orientation, gender identity or both). The lifetime prevalence of violence in younger aged samples did not seem to be lower ( Table 2).
Seven studies reported data specifically on the lifetime prevalence of being punched, kicked, hit or beaten up. In homosexual or bisexual men and women the lowest value was 10% in a study sampling a total of 1911 people, 51 and the highest value was 15.6% (62/416 people). 42 In studies sampling only men the prevalence peaked at 23.4% (of a total sample of 506). 61 In 10 studies the researchers asked homosexual or bisexual women specifically about having objects thrown at them motivated by homophobia or transphobia. The prevalence ranged from 3.7% (in a study of 909 sexual minority women) 75 up to 35.0% (65/186 lesbian, gay and bisexual people). 62 Among transgender people, values ranged from 7.4% (in a study of 255 people) 29 to 17.4% (70/402). 78 Fig. 3 shows the data from 33 studies reporting lifetime prevalence of any kind of sexual violence motivated by perception of sexual orientation and gender identity. The prevalence ranged from 5.6% (28/504 people) 35 to 11.4% (55/484) for all sexual and gender minority groups (12 studies), 50 and from 2.1% (8/377) 92 to 9.7% (18/186) 62 when only homosexual or bisexual men and women were considered (5 studies). The prevalence in homosexual or bisexual men (17 studies) ranged from 3.7% in a study sampling 3429 people 91 to 16.5% (69/419 people). 30 This was slightly higher than in studies of homosexual or bisexual women (8 studies), where it ranged from 1.0% (1/102 people) 34 to 13.2% (14/106). 50 When bisexual people were disaggregated (10 studies), the prevalence ranged from 2.4% (1/41 people) 34 to 14.7% (24/95). 30 Between 7.0% (in a study of 255 people) 29 and   Barrientos et al., 2008 25 N Six studies reported specifically on rape ( Table 2). Among homosexual or bisexual men and women between 0.3% (6/1740 people) 33 and 10.0% (of 264 people) 88 reported ever being raped due to their sexual orientation or gender identity, with figures for men being higher than those for women. The prevalence of rape for transgender people ranged from 3.3% (in a study sampling 255 people) 29 to 9.9% (25/253 people). 41

Discussion
Our review found a high prevalence of physical and sexual violence motivated by perception of sexual orientation and gender identity experienced by sexual and gender minorities, particularly among transgender people. These values suggest that such violence accounts for a large part of all the violence encountered by sexual and gender minorities. Nevertheless, it remains to be researched whether such violence explains the higher prevalence of violence against sexual and gender minorities in comparison with the rest of the population. The higher prevalence in transgender people might be partly explained by a higher risk of being involved in sex work. 101 Violence motivated by perception of sexual orientation and gender identity might not be confined to a minority population. Recent research identified distinct populations on the sexual orientation continuum who identify as mostly heterosexual with a small degree of same-sex sexual or romantic attraction, including occasionally having sexual relations with someone from the same sex. 102 Although we found no publications on this population, earlier research has shown they were 1.47 times more likely than heterosexuals to report experiences of childhood victimization by adults. This elevated proportion is similar to those found among homo-sexual or bisexual men and women compared to heterosexuals, which might be explained by gender non-conformity in childhood. 103 Moreover, people who do not belong to a sexual or gender minority, have also reported being victims of violence motivated by perception of sexual orientation and gender identity. 104 A review of systematic reviews showed that sexual and gender minorities are highly burdened by human immunodeficiency virus infection, sexually transmitted infections, sexually transmitted infection-related cancers, mental health conditions and violent experiences. 105 We suggest further research into the associations of violence motivated by perception of sexual orientation and gender identity with adverse health and social outcomes, including criminalization. This includes the effect of what has been termed syndemic vulnerability 106 or the synergistic interaction between health conditions, exacerbated under circumstances of structural and political adversity.
Notes: Y indicates that the study met the criterion adequately; N that the study did not. We categorized studies as high quality if six to seven criteria were adequate, medium quality if three or five criteria were adequate and low quality if none to two criteria were adequate. If we want to eradicate violence motivated by perception of sexual orientation and gender identity, we must identify the mechanisms and motivations of such violence. The perpetrators are often male and although violence is not necessarily a part of men's dominant position in society (hegemonic masculinity), the two are often linked. In many parts of the world, women are perceived as inferior and therefore both femininity and homosexuality are denigrated and discredited. 107 Physical or sexual force and threats are ways to achieve control, including punishment of perceived acts of resistance to or transgression of gender norms and behaviours. 108 Although same-sex attraction and gender nonconformity can negatively affect the personal relations of individuals with their peers, 109 some authors believe that sexual and gender minorities are mainly attacked because they defy gender stereotypes. 87 This has prompted calls for the elimination of the dichotomist gender characterization. 87

Box 3. Terminology used in studies included in the systematic review of physical and sexual violence motivated by perception of sexual orientation and gender identity
A wide range of descriptions and definitions of violence were used to elicit responses in the included studies:

Motivation for the violence
The victim's perception of the motivation of the violence was variously defined as: "because you're lesbian/gay/bisexual (or someone thought you were)", "because somebody thought or knew you were gay?", "because of/based on/attributed to (perceived) sexual orientation", "on the grounds of homosexuality", "related to MSM-status, " "because of/based on sexuality", "because of your sexual identity (or sexual preferences)", "experienced lesbophobic situation", "because someone knew or presumed you are attracted to men?", "based on sexual orientation and gender identity", "on the basis of gender issues", "for being gay or being perceived as effeminate", "related to their sexual orientation, how they express their gender", "due to gender stereotype transgression", "due to being trans(gender)", "(thought it was) because of gender identity (or gender presentation)", "because you're trans or because of your gender expression", "for being transgender or effeminate", "because you're lesbian/gay/bisexual/transgender", "transgender status, gender identity or expression", "because of the status as a transgender person", "because of transgender identity or background", "due to being queer", "an incident that you felt was homophobic (or transphobic)", "an anti-lesbian/gay incident", "heterosexist violence and harassment because of sexuality or gender identity".

Types of physical violence
Different terms for physical violence were used in different articles. Some used "physical violence", others combined "physical" with "attack", "assault", "victimization", "abuse", "aggression", "mistreatment" or "injury". One article used a longer definition: "the intentional use of physical force with the potential for causing death, disability, injury, or harm; some examples: scratching, pushing, shoving, throwing, grabbing, biting, choking, shaking, slapping, punching". Another used the term "criminal victimization", including specific incidents of physical violence: "experience of a crime against their person (hit, beaten, physically attacked, sexually assaulted).... ". Similarly, most articles included specific violent acts or a combination of actions: "thrown some object", "hit", "knocked down", "injured with some weapon", "punched", "kicked", "beaten", "hurt with a knife, gun, bat, or some other weapon", "assault/robbery with violence", "assault with a weapon, assault without a weapon", "grievous bodily harm", "attempted murder" and "violent assault". One article used an extensive scale to measure physical violence. Physical violence was surveyed making use of 11 items, ranging from "an object was thrown at me", "I was being pushed or pulled", "someone hit me with his or her hand" to "someone tried to strangle or suffocate me".

Types of sexual violence
Several different terms were used in different articles to define sexual violence: "sexual violence", "sexual assault", "rape", "sexual aggression", "sexual victimization", "sexual abuse" and "other sexual violence". In some publications definitions for sexual violence or similar concepts were applied: "ever been forced to engage in unwanted sexual activity", "any sexual act that is perpetrated against someone's will; some examples: completed non-consensual sex act, an attempted non-consensual sex act, abusive sexual contact and non-contact sexual abuse", "sexual aggression: sexually molested and/or forced to have sexual relations with penetration" and "sexual victimization: ever been sexually abused or raped". Physical and sexual violence Karel Blondeel et al.
The quality of our data was relatively poor due to a lack of standardized measures and sometimes small and non-randomized samples. The evidence base needs to be strengthened. More and better research on the prevalence and adverse outcomes of violence motivated by perception of sexual orientation and gender identity is needed across many different geographical and cultural settings (especially outside the USA) and different socioeconomic and age groups. Community organizations should be empowered to add scientific value to their existing efforts to map such violence. A consensus is needed on definitions and measures of violence motivated by perception of sexual orientation and gender identity and how to operationalize them to allow for comparisons across studies.
Some limitations of this review are that most studies used a non-probability sample, mostly a convenience sample, and provided little information on the representativeness of the sample, the potential impact of non-participation, or the study power. The reliability and comparability of studies were limited, as it was not possible to compare between countries, regions or cultural backgrounds. The studies relied on the participants' self-reports to determine whether they had been a victim of violence and whether that violence was motivated by their sexual orientation and gender identity. Without increased understanding of respondents' narratives about violence and its motives, research in this field will be vulnerable to criticism. 110 Despite these limitations, our review shows that high proportions of sexual and gender minorities experienced physical and sexual violence, motivated by perception of sexual orientation and gender identity, which might have an effect on their health and well-being. National violence prevention policies and interventions should include such violence, integrating it into national health surveys and health promotion efforts and improve data collection and reporting of incidents. ■ Competing interests: None declared.

Physical and sexual violence
Karel Blondeel et al.

Violencia motivada por la percepción de la orientación sexual y la identidad de género: una revisión sistemática
Objetivo Evaluar la prevalencia de la violencia física y sexual motivada por la percepción de la orientación sexual y la identidad de género en las minorías sexuales y de género. Métodos Se realizaron búsquedas en nueve bases de datos sin restricciones de idioma en busca de literatura gris y examinada por expertos publicada desde 2000 hasta abril de 2016. Incluimos estudios con más de 50 participantes que midieron la prevalencia de la violencia física y sexual percibida como motivada por la orientación sexual y la identidad o la expresión de género. Excluimos la violencia en la pareja íntima y la autolesión. Debido a la heterogeneidad y la ausencia de intervalos de confianza en la mayoría de los estudios, no realizamos ningún metanálisis. Resultados Se incluyeron 76 artículos de 50 países. Estos cubrieron 74 estudios realizados entre 1995 y 2014, incluyendo un total de 202.607 participantes de minorías sexuales y de género. La calidad de los datos fue relativamente baja debido a la falta de medidas estandarizadas y, en ocasiones, muestras pequeñas y no aleatorizadas. En los estudios en los que todas las minorías sexuales y de género se analizaron como una única población, la prevalencia de la violencia física y sexual varió del 6% (en un estudio que incluía a 240 personas) al 25% (49/196   We only report violence perceived by the victim to be based on sexual orientation or gender identity/expression. Specific descriptions and definitions of physical and sexual violence that were used to elicit participants' responses varied across studies (Box 3). c Number of cases are not reported in all articles, notably for respondent-driven sampling where different weights are given to different participants. d Data from a presentation of the study. e Data provided by the author. f Data from the technical report of the study.