Background

According to the Joint United Nations Program on HIV/AIDS (UNAIDS), 38.4 million people were living with HIV worldwide in 2023, of which 39.0 million were adults (aged 18 years or older) [1]. Data suggest transgender women (TW) are 34 times more likely to acquire HIV than the general population, while gay, bisexual, and other men who have sex with men are up to 25 times more likely to acquire HIV [2]. The UNAIDS Global AIDS Strategy (2021–2026) seeks to reduce the inequalities in the HIV epidemic through a comprehensive package of prevention services addressed to populations most vulnerable to HIV acquisition [3].

Behavior is one axis for HIV prevention. Hence, health behavior theories have tried to explain how prevention and decision-making behaviors are related and which barriers and facilitators influence people’s conduct when they choose to adopt (or not) HIV prevention strategies [4, 5]. Examples of health behavior theories are the health belief model [6], the protective motivation theory [7], and the theory of planned behavior [8]. All these theories have a common denominator: direct or indirect risk perception is thought to influence behavior. These theories propose that increasing people’s risk perception could reduce their risk behavior, yielding subsequent benefits to physical and mental health [9].

In health sciences, particularly medicine and public health, risk perception is understood as an individual’s subjective assessment of the probability of an undesirable outcome [10], and it has a cognitive and an affective component [4, 11, 12]. The cognitive component refers to the perceived likelihood of harm (i.e., it is the subjective probability of experiencing an adverse outcome given one’s behavior) and the perceived susceptibility to injury (degree of presumed liability arising from one’s conduct) [4]. These two cognitive dimensions have been the most studied components regarding risk perception. The affective component, which encompasses both anticipatory and anticipated risk-related emotions felt during risk evaluation and when facing consequences of risky decisions in the future, frequently takes precedence over a statistical analysis of the risks and benefits when making decisions [13, 14]. For example, contemplating the potential adverse outcomes of a risky activity can evoke negative emotions and stress, reducing individuals’ willingness to engage in perceived risky behaviors. Therefore, understanding how the affective component influences risk perception is crucial for developing effective risk management strategies and promoting health behaviors [15].

It is well-known that the perceived risk for HIV acquisition and sexual HIV exposureFootnote 1 (i.e., one’s sexual behavior) may not be aligned [12, 16,17,18]. Therefore, it is necessary to explore how perceived risk for HIV acquisition and sexual HIV exposure have been assessed to understand their relationship better. This systematic review sought to synthesize the literature on studies evaluating the relationship between perceived HIV risk and sexual HIV exposure among sexual and gender minorities.

Methods

Protocol and registration

This study was registered in the International Database of Systematic Reviews in Health and Social Care (PROSPERO 2021 CRD42021278247), and it is reported according to the Statement of Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) [19].

Eligibility criteria

We included studies in English, Portuguese, or Spanish published from 1981 (when HIV was identified as a worldwide public health outbreak) to July 2023 with the following criteria: (i) participants were adults (≥ 18 years), (ii) sexual and gender minorities of any gender; (iii) cisgender men who have sex with men (cis-MSM) and do not identify as gay or bisexual, (iv) had unknown or negative HIV serostatus, (v) perceived risk for HIV acquisition and sexual HIV exposure were simultaneously assessed, (vi) reported correlation, comparison or association (unadjusted or adjusted) between perceived risk for HIV acquisition and sexual HIV exposure. We excluded reviews, meta-analyses, thesis, dissertations, monographs, conference papers and reports, qualitative studies, or studies that included injection drug users or reported grouped results with other populations different from sexual and gender minorities.

Information sources

We performed a literature search on MEDLINE, IBECS, LILACS, CUMED, LIPECS, medRxiv, LIS (Localizador de Informação em Saúde), Coleciona SUS, BIGG-guias GRADE, PAHO-IRIS, COCHRANE, and SciELO.

Search strategies

The search combined terms derived from five domains: (a) perceived risk for HIV acquisition, (b) sexual HIV exposure, (c) HIV, (d) sexual minority, and (e) gender minority. We used PubMed, Embase, and Lilacs to perform the research. Search keys are available in the Additional file 1. All studies were exported to Zotero software, and duplicates were excluded. The last date we performed the search was July 15th, 2023.

Selection process

Two authors reviewed all abstracts independently, according to the eligibility criteria, and another author reviewed discrepancies to agree on the final list of full-text articles to be reviewed. Authors attempted to reach corresponding authors to request full manuscripts when unavailable. All authors reviewed all articles independently and discussed discrepancies until they agreed.

Data collection process

We collected data using structured Excel spreadsheets. Before data collection, investigators discussed which variables should be collected, considering the main objective of this study.

Data items

Data collected from the selected studies included author(s), year, country, recruitment strategy, study period, sample size, age, gender (self-reported gender identity, regardless of participants’ sex assigned at birth), race/ethnicity, sexual HIV exposure assessment (including recall time), perceived risk for HIV acquisition assessment (including recall time), statistical analyses, and significant findings resulting from comparisons and correlations, as well as unadjusted and adjusted associations estimated with regression models.

Synthesis methods

We divided the selected studies into three groups depending on the analysis performed or the outcome used for multivariable models: (i) correlations or associations studies, (ii) logistic regression or Poisson robust error models using sexual HIV exposure as the outcome, and (iii) logistic regression models using perceived risk for HIV acquisition as the outcome. We also extracted data on other factors associated with perceived risk for HIV acquisition and sexual HIV exposure when available.

Results

Study characteristics

The flow diagram of the selection process is shown in Fig. 1. We identified 40 studies [20,21,22,23,24,25,26,27,28,29,30,31,32,33,34,35,36,37,38,39,40,41,42,43,44,45,46,47,48,49,50,51,52,53,54,55,56,57,58,59] carried out in 18 countries, but two studies were considered as one since they used the same sample but did complementary analysis [43, 56] (final sample = 39). Most studies were carried out in the United States of America and Canada (13; 33.3%), followed by Asia (11; 28.2%), Latin America (7; 17.9%), Europe (5; 12.8%), and Africa (2; 5.1%). One (2.6%) study was conducted on multiple continents [27].

Fig. 1
figure 1

Flow diagram of study selection for the review of perceived risk for HIV acquisition and sexual HIV exposure among sexual and gender minorities

Full size image

All studies were written in English, but one written in Portuguese [26]. Six studies (15.4%) did not specify the study period; the remaining were conducted between 1991 and 2020. Twenty-four studies (61.5%) were carried out before September 2015, when the World Health Organization (WHO) recommended PrEP for all key populations [60]. Recruitment in LGBTQIA + venues and community-based organizations was more frequent (11, 28.2%), followed by online recruitment (10, 25.6%), HIV or STI health clinics (7; 17.9%), respondent-driven sampling (5; 12.8%), and advertisements on magazines (1; 2.6%). Five studies (12.8%) used two or more methods for recruitment. Only one study had a longitudinal design (from 1999 to 2018 – before and after WHO’s PrEP recommendation) [22]; all other studies were cross-sectional.

The sample size range from all studies was from 55 to 16,667. Sixteen studies (41%) reported mean age (from 19.7 to 39 years), 14 (35.9%) reported median age (from 20.4 to 44.5 years), seven (17.9%) defined age groups, and two (5.2%) did not provide information. Most studies included only cis-MSM (29; 73.3%), three only TW (7.7%) [28, 30, 31], five included cis-MSM and TW (12.8%), and two included cis-MSM, TW and non-binary people or other genders (5.2%) [57, 59]. Almost half of the studies (17; 43.6%) enrolled participants from different races or ethnicities, followed by those with participants from only one ethnicity: two (5.1%) with White [54, 55], two (5.1%) with Black [45, 50], two (5.1%) with European [22, 35], and one (2.6%) with Asian [39]. White people were the most prevalent group across the studies (20, 51.3%). Fifteen studies (38.5%) did not provide information about participants’ ethnicity.

Sexual HIV exposure assessment

Sexual HIV exposure assessments across studies varied considerably. Only seven studies (17.9%) used a validated instrument or scale (HIV Incidence Risk Index for MSM [61, 62]), two (5.1%) used study-specific scales, one (2.6%) used a single question, and 29 (74.4%) considered a single or a combination of different sexual HIV exposures from a given list (i.e., condomless receptive or insertive anal sex, inconsistent condom use, sex with a person living with HIV, or previous HIV tests, among other options). Recall time frames across studies were also different: last 12 months (6, 15.4%), six months (20; 51.3%), three or fewer months (8, 20.5%), and other (4, 10.2%). Only one study did not specify the time frame for the recall (2.6%). See columns Sexual HIV exposure for detailed information in Tables 1, 2 and 3.

Table 1 Studies with comparison or correlation analysis between sexual HIV exposure and perceived risk for HIV acquisition (n = 10)
Full size table
Table 2 Studies with Simple or Multiple Regression Models to Assess Factors Associated to Sexual HIV Exposure (n = 19)
Full size table
Table 3 Studies with Simple or Multiple Regression Models to Assess Factors Associated to Perceived Risk for HIV Acquisition (n = 10)
Full size table

Perceived risk for HIV acquisition assessment

Twenty studies (51.3%) used a single question with Likert-scale responses, and six (15.4%) used validated instruments or scales (Perceived Risk for HIV Infection Scale [63], HIV Perceived Scale [64], AIDS Health Belief Scale [65], AIDS Prevention Questionnaire [66]), and four (10.2%) study-specific scales. Eight studies (20.5%) did not specify the question but used Likert-scale responses, and one (2.6%) did not provide any information. Most studies (20; 51.3%) did not specify a time frame for perceived risk, but six studies (15.4%) considered the following year, four (10.2%) used the scales’ time frame (i.e., current time), four current perceptions (10.2%), and four studies used other time frames (10.2%). Only one study (2.6%) considered the perception of HIV acquisition across the lifespan [44].

Association between sexual HIV exposure and perceived risk for HIV acquisition

Studies performed different analyses to evaluate and quantify the association between sexual HIV exposure and perceived risk for HIV acquisition. Ten studies (25.6%) performed simple comparison (n = 7/10) or correlation analysis (n = 3/10) (Table 1), and 29 studies (74.4%) performed a multivariate analysis; of these, 19 used sexual HIV exposure (19/29, 65.5%) (Table 2), and ten used perceived risks for HIV acquisition (10/29, 34.5%) (Table 3), as the outcome.

Six out of seven comparison studies were significant between high sexual HIV exposure and high perceived risk for HIV acquisition using Chi-squared test [35, 37, 48], ANOVA [32, 55], and t-test [57] (Table 1). Two [24, 36] out of three correlation studies found a significant positive correlation between high perceived risk for HIV acquisition and high sexual HIV exposure.

Table 2 depicted 19 studies [20, 26, 28, 29, 34, 38,39,40,41,42, 45,46,47, 50, 51, 53, 54, 58, 59] that used sexual HIV exposure as the outcome using univariable [28, 38, 39, 42], or multivariable logistic regression models, including perceived risk for HIV acquisition as a predictor. Three studies [28, 38, 39] with univariable regression analysis found that a high perceived risk for HIV acquisition was associated with lower odds of any sexual HIV exposure, and one did not find any association [42]. Conversely, twelve studies with multivariable logistic models found that those with a high perceived risk for HIV acquisition (reference: none or low) had higher odds of any sexual HIV exposure. Only one study using the same methodology found an inverse association, that is, lower relative risk for sexual HIV exposure (condomless anal sex) (aRR = 0.85 [0.78–0.93]), but only for MSM over 30 years old [45]. One study found the highest odds for this outcome (sexual HIV exposure) among those reporting some (aOR = 10.37 [3.26–33.04]) [40] and high perceived risk for HIV acquisition (aOR = 6.00 [2.31–15.63]) [20]. Two studies did not find an association between perceived risk for HIV acquisition and sexual HIV exposure [50, 53].

Ten studies used perceived risk for HIV acquisition as the outcome (Table 3). Two studies used simple regression models [21, 49], and seven studies used multivariable regression models and found that individuals reporting any sexual HIV exposure had higher odds of high perceived risk for HIV acquisition. Studies using simple regression models found that unprotected anal intercourse (OR = 2.10 [1.61–2.75]) [21] or inconsistent condom use (OR = 1.76 [1.05–2.94]) [49] increased the odds of high perceived risk for HIV acquisition. Studies using multivariable regression analysis also found that any sexual HIV exposure increased the odds of high perceived risk for HIV acquisition, such as having sex partners with unknown HIV status (aOR = 8.9 [2.0-38.5]) [31] for TW and having unprotected receptive anal intercourse with a person living with HIV (aOR = 7.17 [3.26–15.76]) [22] for cis-MSM. Bosga et al. [25] found that having sex with a person living with HIV (β = 0.24, r = 0.22, p < 0.01), high denial of risk for HIV acquisition (β = 0.23, r = 0.22, p < 0.01), and not being religiously active (β= -0.18, r= -0.19, p < 0.05) were associated with perceived risk for HIV acquisition.

Some multivariable studies (n = 21) reported other variables associated with sexual HIV exposure [20, 26, 28, 29, 34, 40, 41, 45,46,47, 50, 51, 54, 58, 59] or perceived risk for HIV acquisition [27, 31, 33, 43, 44, 56] (Table 4). Relevant variables that increased odds for sexual HIV exposure were: younger age (except for one study which found an association with age 25+) [59], low education level, high income, alcohol or substance misuse, never tested for HIV, increased number of sexual partners, transactional sex, previous STIs diagnosis, and history of sexual violence. Variables associated with high perceived risk for HIV acquisition were: younger age, low income or education (except for one study which found an association with high education level) [27], non-White race or ethnicity, alcohol or substance misuse, increased number of sexual partners, transactional sex, previous STIs diagnosis, PrEP eligibility or awareness, and post-exposure prophylaxis (PEP) awareness.

Table 4 Factors associated in multivariable models for sexual HIV exposure or perceived risk for HIV acquisition (n = 21)
Full size table

Discussion

In the present systematic review, we aimed to identify the relationship between the perceived risk of acquiring HIV and sexual HIV exposure. We found evidence of an association between high perceived risk of HIV acquisition and sexual HIV exposure from most studies. Moderate or high perceived risk for HIV acquisition was associated with high sexual HIV exposure, and vice versa. Nevertheless, the definition of high sexual HIV exposure has changed over 40 years since the start of the HIV epidemic. For example, none of the studies did a differential analysis based on the estimated per-act probability of acquiring HIV from sexual exposure (i.e., receptive vs. insertive condomless anal sex) [67], PrEP use status, or the majority considered high sexual HIV exposure as a mixed result between substance use, condomless anal sex, number of sexual partners, etc. A recent study conducted in Brazil found that PrEP moderates the association between high perceived risk of HIV acquisition and sexual HIV exposure, resulting in no significant association between perceived risk of acquiring HIV and sexual HIV exposure among PrEP users [68]. More studies evaluating this association after WHO’s PrEP recommendation are needed [60], as sexual behavior and perceived risk for HIV acquisition among sexual and gender minorities could evolve and change due to the high efficacy of PrEP and treatment as prevention (TasP) in preventing sexual HIV transmission.

Though the studies reported on a wide range of participants, they had few representations of some populations, such as people aged 60 + years. We found no studies with transgender men participants, and only two included non-binary people. Tordoff et al. [69] found that transgender or non-binary people partnered with cisgender people could have worse health outcomes, such as higher self-reported HIV prevalence, history of STI, less HIV testing, or PrEP use than cisgender or transgender people. These findings indicate that more research is needed to understand perceived risk and sexual HIV exposure risk among sexual and gender minorities from low- and middle-income countries and among transgender, non-binary, and gender-diverse persons.

Perceived risk for HIV acquisition was assessed solely from the probability self-perspective. Risk perception is a complex construct, and an accurate assessment should consider various components [70]. These are deliberative component (i.e., evaluating the likelihood of acquiring HIV when having unprotected sex), affective component (i.e., experiencing fear or concern when considering potential health consequences), and experiential component (i.e., “gut-level reactions” due to previous experience with people living with HIV) [13, 71]. Perceived risk for HIV acquisition requires the use of validated and adapted instruments for each context and population (i.e., Perceived Risk for HIV Infection Scale [63], HIV Perceived Scale [64], etc.). Single-item or study-specific scales limit the ability to reflect variability in individual perceptions based on their deliberative, affective, or experiential components, so these assessments likely have insufficient validity to support their results. Therefore, perceived risk for HIV acquisition assessments should include a comprehensive evaluation to reach accuracy and then design better interventions to improve perception among the most vulnerable populations to HIV. Addressing the relationship between perceived risk for HIV acquisition and sexual HIV exposure assessments might include in-depth interviews and specific questions [72], as well as self-administration questionnaires to ensure confidentiality and anonymity [73]. Though we recognize that no single instrument will capture all dimensions of a construct or will apply to all populations, we encourage future research to evaluate currently available validated instruments and their applicability to one’s study. Moreover, for perceived risk, we encourage future studies to broaden risk assessment from its cognitive component to assess its affective, behavioral, and phychosocial components [71, 74].

We also found various definitions and assessments for sexual HIV exposure and perceived risk for HIV acquisition. Only a third of the studies used a validated instrument to assess sexual HIV exposure or perceived risk, and all other studies used one-question or study-specific scales without providing basic validation parameters. This diversity hampers consistency, comparability, replicability, content validity, and an accurate measurement of complex constructs such as sexual behavior or perceived HIV risk while also making the description of behaviors to be addressed in prevention or other interventions more complex [75]. For example, most studies collected sexual HIV exposure or perceived risk for HIV acquisition either with categorical or dichotomous variables or recategorized continuous variables into categories or dichotomous data for analysis. However, dichotomization or categorization could lead to loss of statistical power, reduced ability to detect nonlinear effects, introduce classification biases, and lack of consistency in selected cutoff points [76]. In this sense, studies should use cross-cultural and adapted instruments for their setting and population. This increases the validity and consistency of their findings, as well as the comparability and replicability with other populations. For example, the Perceived Risk of HIV Scale, created in the USA in English [64], was used in Brazil after a proper validation process [57]. Additionally, instruments to assess perceived risk for HIV acquisition should include the affective component of risk perception, which was notably absent in the studies included in this review [71]. One way to minimize recall and social desirability biases would be to include another measure as a point of comparison [77]. An additional measure of sexual exposure, for example, could be obtained with experience sampling or daily diary methods, both of which are designed to capture people’s experiences in real time and, therefore, less prone to recall bias [78]. Additionally, recall periods should be limited to the prior month, including mood assessment, for accurate data [79].

Self-evaluation of sexual behavior could be biased by recall and social desirability, mainly if an interviewer assesses it. Recall bias in retrospective studies is frequent due to participants’ ability to accurately remember and report past events [80], especially when obtaining information on health-related behaviors [81]. Also, the influence of mood at the time of any assessment can distort how people remember past events, so cognitive biases may cause them to reflect on events more positively or negatively than how they occurred. For example, the influence of prior expectations on the interpretation of past events and the difficulty in recalling specific events compared to more general events [82]. Additionally, the reviewed studies encompassed different timeframes for participants to identify, increasing the inaccuracy in recognizing behavior as more time passes [81]. Moreover, when studies assess sexual behavior, there are challenges associated with social desirability bias. These challenges become particularly important considering the social stigma attached to certain sexual practices, such as same-sex intercourse, anal sex, recreational drug use during sexual encounters, or having multiple sexual partners [72]. Additionally, social desirability bias could make individuals portray themselves positively to others, resulting in denial or underestimating behaviors or traits they perceive as socially undesirable or stigmatizing [73]. For example, high internalized homo or transphobia, resulting from long-term social rejection, are associated with mental health problems, low use of HIV preventive methods, or more frequent sexual HIV exposure [83, 84].

Finally, when it comes to social or behavioral research (e.g., sexual behavior or perceived risk for HIV acquisition), researchers should consider the influence of intersectionality on the decisions or behavior of specific populations [85]. Sexual and gender minorities, who are at increased vulnerability of acquiring HIV, build their identity and shape their perception and behavior under the influence of social and structural obstacles from very early ages as the result of their ethnicity, socio-economic status, stigma, discrimination, lack of access to education or work, among others [86, 87]. Therefore, the perceived risk for HIV acquisition or sexual HIV exposure among these minorities is not only a result of individual responsibility but also of those social identities that place them in subordination or vulnerability (e.g., being under socio-economic vulnerability, self-identified as trans women or Black/Latinx race/ethnicity) compared to other privileged social identities (e.g., being White, cisgender men, heterosexual, and highly educated) [88]. In this sense, 38.5% of the reviewed studies did not report participants’ ethnicity and how the perceived risk for HIV acquisition or sexual HIV exposure might be different according to their race or ethnicity. Assessment of perceived risk or sexual HIV exposure among sexual and gender minorities could be constructed under an intersectionality framework to achieve accurate outcomes. This reinforces the importance of the collection of race/ethnicity data on future studies.

Our review has limitations:1) most of the studies included only cis-MSM, predominantly white, and from high-income countries, which limits the generalizability to other sexual and gender minorities in low or middle-income countries; 2) most of the studies were cross-sectional, so they could not capture the temporal dynamics of the relationship and the causality between perceived risk for HIV acquisition and sexual HIV exposure; 3) as mentioned before, most of studies asked past sexual behavior which could be biased by recall and social desirability; 4) we did not collect information on intersectional aspects associated with perceived risk for HIV acquisition or sexual HIV exposure; and 5) we did not include qualitative research studies, which could have provided additional insights and understanding of the reasons and reasoning behind perceived risk and sexual HIV exposure. These limitations highlight the complexity of the relationship between perceived risk for HIV acquisition and sexual HIV exposure, suggesting that a combination of individual, social, and contextual factors can influence risk perception and sexual HIV exposure.

Conclusions

We found evidence of an association between high perceived risk of HIV acquisition and sexual HIV exposure. Our results may be useful for the development of prevention and education strategies to address known risk behaviors and underlying factors affecting risk perception. Having an adequate perception of risk aligned with behavior is crucial for prevention, informed decision-making, access to education and resources, reduce stigma and discrimination (by gaining an accurate understanding of how the virus is transmitted and practicing safe sex, stereotypes, and misconceptions can be avoided), and promote self-care and personal well-being.