This article was preregistered in the International Prospective Register of Systematic Reviews (PROSPERO, http://www.cdr.york.ac.uk/prospero, CRD42019123815). The work was reported in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) (23) and its protocols (24). The PRISMA checklist is reported in Supplementary Table 1 of the Supplementary Materials.

Comprehensive searches were conducted in the PubMed, Web of Science, PsycINFO, and Google Scholar databases to identify relevant articles related to awareness and use of PEP among MSM. The search strings were intersections of disease-, population-, PEP-, and outcome-related terms (Supplementary Table 2 of Supplementary Materials). Additional searches were conducted in the Cochrane Library. To identify additional relevant citations, the reference lists of published reviews and journal articles were also screened. All searches were limited to English peer-reviewed journal articles. The original bibliographic searches for PsycINFO and PubMed were conducted on June 15, 2021. The original search for Web of Science, Google Scholar, and Cochrane Library were conducted on June 16, 2021. The updated searches for PsycINFO, PubMed, and Web of Science were conducted on November 19, 2021, while for Google Scholar and Cochrane Library on November 20, 2021.

Studies were considered eligible if they met all of the following inclusion criteria: (1) conducted primary data collection; (2) were studies related to PEP use for HIV prevention; (3) reported data about gay and bisexual men, regardless of age, country, and HIV status; (4) reported history, awareness, or intention of PEP use among MSM; and (5) appeared in English-language, peer-reviewed journal articles. Studies that reported data about MSM and other populations, such as transgender people or sex workers, were included, but only data related to MSM were considered and abstracted. Studies were identified as ineligible if they were (1) case reports; (2) longitudinal studies or trend studies; (3) unpublished book chapters, theses, dissertations, or articles; and (4) non-original research, secondary reports, review articles, or theoretical framework articles.

EndNote (version 6) was adopted throughout the process of study selection. First, titles and abstracts were preliminarily screened to exclude irrelevant studies. Second, full-text versions of selected papers were assessed independently to ensure that all inclusion criteria were met. Disagreement in the process of study selection was resolved by discussion with a third reviewer. The two researchers used standardized Microsoft Excel spreadsheets to extract the following information: authors, year of publication, country of study, study design, settings, study populations, outcomes, and factors. Interview quotes relating to awareness or use of PEP were also collected to aid conceptual understanding of qualitative findings.

The primary outcomes of interest were measures of PEP awareness and previous use or potential usage intention (Supplementary Table 3 in Supplementary Materials). All available data were pooled and synthesized using a combination of single-rate meta-analysis and a narrative synthesis approach. For quantitative studies, a random effects model was used to quantitatively summarize the pooled event rates (ERs) of PEP awareness and previous PEP use. Random effects meta-regression was used to assess the relationship between year of data collection and PEP awareness or use. Subgroup analysis was performed to determine the potential influence of the following covariate: level of national economic development.

A synthesis of factors determining PEP awareness and use was performed and classified into individual, social, or structural layers based on a socioecological model (25, 26). Qualitative data were drawn to provide context for the quantitative findings as recommended (27, 28) by identifying participant perspectives about factors that might affect awareness and previous use or potential intention of use of PEP among MSM. Thematic analysis was used to abstract, sort, compare, and categorize data to construct a set of emerging descriptive themes from relevant quotes. Themes were then used to construct a conceptual framework of awareness and previous use or potential usage intention of PEP in the individual, social, and structural domains.

Two independent reviewers used the Cochrane Collaboration tool to assess the risk of different types of biases, including recruitment, sampling, attrition and non-response, social desirability, and researcher (29). The potential risk of impact of biases on the robustness of the included findings was also discussed (see Supplementary Table 4 in Supplementary Materials).

Of a total of 3,004 titles, 402 were further assessed for inclusion, and 20 were included in the final review (Figure 1). The initial screening excluded studies that did not specifically focus on HIV, PEP, and duplicates, leaving a total of 402 citations. Subsequently, 382 citations were excluded after screening the abstracts and full papers. Several papers focusing on other population groups (e.g., heterosexual, sex workers, health care workers, and transgender populations) were identified through the search and were excluded (n = 129). Additional exclusions were due to poor relevance of outcomes (e.g., cost effectiveness, medication adherence; n = 174), failure to segregate results by population (n = 16), and reviews or non-peer reviewed articles (n = 63).

Overall, 20 eligible studies published between 2007 and 2021 were included in the meta-analysis, involving 12,579 MSM. Supplementary Table 5 in the Supplementary Materials presents the key characteristics of the studies included in the review. Of the included studies, 19 were quantitative, and one was a mixed-method study. The two-phase mixed-method study (30) included a cross-sectional survey with a self-administered questionnaire and a prospective descriptive study using an in-depth interview, but data needed were collected in the quantitative survey phase. Another study used a longitudinal cohort study to evaluate the impact of HIV leadership programs, and its baseline survey provided the data we needed (31). All of the remaining quantitative studies were cross-sectional surveys.

The included studies used consistent definitions of PEP for MSM with different terms, such as non-occupational PEP (nPEP) or PEP after sexual exposure (PEPSE). Given that many persons did not know or knew little about PEP, an introductory explanation of PEP may have been provided to participants during the survey.

All 20 studies reporting awareness of PEP employed a simple binary question asking participants whether they were aware (or had heard) of PEP, except for one study (32) that asked whether a medical treatment existed that could reduce the chance of becoming HIV-positive.

Meta-analysis of the 20 studies found that the pooled estimate of the proportions of MSM who were aware of PEP was 59.9% (95% CI: 50.5~68.7) (Figure 2). Awareness of PEP in these studies was inconsistent, ranging from 22.09% among MSM in Guangxi, China (33), to 88.27% among MSM in London who use geosocial networking smartphone applications (11). Seven studies reported higher awareness than 70% among gay men in Israel (15), young MSM of color in the United States (34), MSM users of geosocial networking applications in England (11), non-HIV-positive MSM in Italy (32), sexually active MSM living with HIV in France (35), MSM in 22 Sub-Saharan African countries (36), and among MSM college students in three cities of China (37). Four studies reported much lower levels of awareness (<40%) among MSM in China (33); MSM tested in community centers in Portugal (38) MSM receiving rapid HIV testing in Spain (39), and among black MSM in the United States (13). Nine studies reported modest PEP awareness: 41.23% among men engaging in condomless anal sex with men in the United States (8), 42.5% among MSM in Beijing, China (16), 46.67% among gay and bisexual men in California (40), 50% among MSM in Brazil (41), 56.70% among sexually active MSM in Vancouver, Canada (31), 57.06% among MSM in Vancouver, Canada (42), 60.22% among Thai MSM (30), 60.6% among MSM in four cities of China (43), and 65.77% among HIV-positive MSM in England (44). There was significant heterogeneity between studies (Q statistic = 1990.122, I² = 99.045; p < 0.001).

Two studies equated awareness with knowledge of PEP (32, 40, 45). However, self-reported awareness did not necessarily reflect a precise understanding of PEP. Several studies checked whether self-reported knowledge about PEP was accurate or evaluated PEP knowledge in MSM. Those MSM who were aware of PEP had adequate knowledge about PEP; however, less than half of them correctly answered that PEP needs to be taken for 28 days (30, 34). Among 94 MSM who had heard of PEP, 11% were concerned about side effects; most (68%) felt they would know how to get PEP, and nearly two-thirds anticipated that they would not be able to afford it (8). Only 40% of those who reported being aware of PEP (24% of total participants) were aware that the time window for an effective PEP is 72 h (15).

Bivariate meta-regression showed that temporal changes could not explain PEP awareness (slope = −0.031, 95% CI:-0.038~-0.024; p < 0.001): the minimum value of PEP awareness was collected in 2016 (33), and the maximum value was also gleaned in 2016 (11). Subgroup analysis revealed that developed countries PEP awareness had no differences compared to that in developing countries (Q statistic = 0.865; p = 0.352): fourteen studies from developed countries found that the point estimate of proportions of MSM who were aware of PEP was 51.9% (95% CI: 50.9~53), and six studies from developing countries found the point estimate was 52.9% (95% CI: 51.1~54.7). In addition, the studies had no significant publication bias in the awareness subgroup analysis (intercept = 8.544, 95% CI:−2.2~19.29, p = 0.112).

In addition to overall awareness, seven studies explored other factors that were correlated with awareness of PEP among MSM. Table 1 illustrates the range of factors affecting MSM's awareness of PEP documented in the included studies. These factors, which could potentially impede or facilitate participant's awareness of PEP, were conceptually divided into different categories within the individual, social (including partners, families, and communities), and structural domains (health systems and legal factors). In the individual domain, low educational attainment, unemployment, having casual partners, and closeted bisexual were negative factors; high-level education, white race/ethnicity, gay sexual identity, knowing the HIV status of one's self, higher personal sexual altruism, metropolitan resident, higher annual income, using the internet as the main way of meeting partners, having sex with an HIV-positive male partner, and having unprotected anal sex were positive factors. How age, drug use, and number of partners exert influence on awareness of PEP is uncertain.

Social factors associated with greater awareness among MSM included more interaction with gay culture (39), lower levels of HIV stigma (32), and HIV leadership programming attendance (31). One study conducted logistic regressions predicting awareness of PEP and found that location was the only significant predictor: compared to Puerto Rican participants, those in Pittsburgh were 5.7 times more likely to have heard of PEP, and those in Boston were 10.1 times more likely to have heard of PEP (8).

Regarding structural factors, greater access to condoms (42), more frequent contact with HIV/AIDS organizations (32), previous sexually transmitted infection (STI) diagnosis (42), HIV testing (32, 34, 35, 43), and disclosure of one's sexual orientation to their general practitioner (11) were associated with having heard of PEP.

Thirteen studies assessed quantitative proportions of participants who had used PEP. All of them used a simple binary question, asking participants whether they had used or taken PEP in the past. Although the question was asked of all MSM participants in some studies (11, 13, 30, 39), we calculated the proportion of MSM who had used PEP among subsamples who were aware of PEP. Meta-analysis of the thirteen studies found that 4.9% (95% CI: 2.4~9.8) of MSM had used PEP (Figure 3). Previous use of PEP in these studies was low and inconsistent, ranging from 1.19% among HIV-negative MSM in Vancouver, Canada (42) to 40.74% among MSM in 22 Sub-Saharan African countries. Most studies reported much <10% of respondents that indicated previous use, with the exception of three studies that reported previous use of 40.74% among 22 Sub-Saharan African countries (36), 11.86% among young MSM of color in the United States (34), and 27.37% in London (11). As expected, there was a high level of heterogeneity between these studies (Q statistic = 656.703, I² = 98.173, p < 0.001).

One study reported a qualitative assessment of willingness to use PEP, asking participants “If you have risk behaviors for HIV infection, will you seek PEP?,” and 416 participants (92.4%) answered yes (30). Another study (8) showed that participants were especially likely to say they would use PEP in the future, scoring an average of 9.1 (on a 10-point scale, with 10 being extremely likely), but did not report the proportion of MSM participants who were very willing to use PEP. One study used a single item with a 5-point Likert-type scale to assess willingness to use PEP and reported 461 (73.06%) MSM participants who were very-to-extremely likely to use PEP (45).

Bivariate meta-regression showed that years of data collection could predict PEP use (slope = −0.031, 95% CI:−0.038~-0.024; p < 0.001): PEP use among MSM significantly increased from 4.18% in 2006 to 30.01% in 2016 (11). Subgroup analysis revealed that PEP use in developing countries was significantly higher than that in developed countries (Q statistic = 96.77; p < 0.001): eight studies from developed countries found that the point estimate of proportions of MSM who previously used PEP was 5.4% (95% CI: 4.8~6.1), and five studies from developing countries found the point estimate was 13.2% (95% CI: 11.5~15). The studies had no significant publication bias in the use subgroup analysis (intercept = −7.260, 95% CI:−18.18~3.66, p = 0.171).

In addition to the overall previous use of PEP or willingness to use PEP, our review provides comprehensive information regarding barriers and facilitating factors of its use. Table 2 illustrates the range of factors influencing MSM's previous use or potential use of PEP mentioned in the eligible studies. Several studies explored factors that were associated with previous use of PEP among MSM. In the individual domain, concern about side effects (8) and lack of PEP information (11, 34) were barriers, whereas having adequate knowledge about PEP (30), having been involved in high-risk sexual intercourse (39), being circumcised (30), and using methamphetamine (11) were facilitating factors. Associations between past PEP use and club drug use have been tested for HIV in the past year, and recent sexually transmitted infection (STI) diagnoses have not persisted (11). In the social domain, being in a relationship (11) was a facilitating factor. In the structural domain, cost emerged as an important barrier to the use of PEP in some cities in the United States (8) or sometimes in Thailand, where PEP is not available free of charge (30).

Factors associated with an intention to take PEP included awareness of PEP, HIV knowledge with > 80% correct answers, absence of penetrative anal sex in the past 3 months, and circumcision (30). Partnered men's willingness to use PEP was positively associated with having an individual income < $30,000 USD and serosorting within the relationship (sexual partner selection on the basis of HIV status), whereas it was not clear why higher investment in couple relationships and greater age differences between primary partners were barriers (45).