The 2022 multicountry mpox outbreak spread rapidly via air travel and undetected local transmissions among gay, bisexual, and other men who have sex with men (GBMSM) driven by social stigma, marginalization, and discrimination [1-5]. These factors often compel GBMSM to seek anonymous and multiple same-sex connections in high-risk environments, facilitating “superspreader” events and exacerbating both local and global mpox transmission [2,5-8]. The mpox outbreak exemplifies the critical importance of a One Health approach by revealing the fundamental interconnectedness or existential disconnectedness among human, animal, and environmental health. Of concern, originally a zoonotic disease with animal-human transmission in parts of Africa, mpox has adapted to “human-to-human” transmission networks worldwide, particularly in urban environments and initially mostly affecting GBMSM. This evolution demonstrates how ecological shifts, human mobility, and social determinants—including stigma and health care inequities—collectively shape infectious disease dynamics [2]. In 2023, Asia experienced sustained imported mpox cases, with mainland China documenting 501 cases in August alone [9-11]. The region faces unique challenges due to persistent stigmatization of same-sex behaviors, limited mpox vaccination programs, and the easing of COVID-19 travel restrictions [12,13], raising concerns about a resurgence if vaccine uptake among GBMSM remains low [4,6,14].

While Western countries effectively curtailed the outbreak through culturally sensitive campaigns—such as leveraging gay-identifying physicians and influencers to boost vaccine acceptance [15,16]—Asian contexts have struggled [17]. Despite the US Centers for Disease Control and Prevention recommending the 2-dose JYNNEOS vaccine for high-risk GBMSM [18], costs and access remain significant barriers. For example, JYNNEOS costs up to US $300 per dose in higher-income settings, whereas Hong Kong has offered fully subsidized vaccination since October 2022. However, this program is largely inaccessible to mainland Chinese GBMSM, and mpox-specific vaccines are still under development [19-21]. Even in places with subsidized programs such as Hong Kong and Taiwan, vaccine uptake remains suboptimal (21.7% in Hong Kong [22] and 59.9% in Taiwan [23], much lower than in Western cities such as Paris at 78% [24]) and below the modeled 80% threshold needed to prevent further outbreaks [25].

A growing body of research has examined risk perception and media exposure among GBMSM [22,26-28], but few studies have explored how varying types and sources of mpox-related information shape illness perceptions and vaccination intentions—especially in Chinese contexts. Most notably, the psychological mechanisms connecting information exposure with preventive intentions and actions remain poorly understood in Asia, where negative media portrayals and lesbian, gay, bisexual, transgender, and queer (LGBTQ+) stigma are widespread [29-31]. This constitutes a critical research gap as understanding these mechanisms is essential for designing effective, culturally sensitive risk communication strategies.

The common sense model of illness self-regulation (CSM) provides a valuable framework for examining how different types of mpox-related information influence GBMSM’s preventive intentions and behaviors. The extended CSM framework proposes that emotional states such as concerns, depression, and distress are more strongly correlated with threat-related illness perceptions, including the identification of symptom severity, duration, and their impacts on individuals [32]. For example, exposure to stigmatizing news about deaths due to mpox and HIV or AIDS coinfection may evoke negative emotions and elevate GBMSM’s perceptions of mpox severity and impact, leading to maladaptive coping responses such as denial, avoidance, or delayed treatment seeking [32,33]. Conversely, accurate and timely information about mpox transmission, symptoms, and prevention methods may contribute to adaptive illness representations, such as illness coherence and controllability, which predict positive coping behaviors such as adherence to preventive measures and timely health care seeking [34,35]. Applying the extended CSM to mpox-related information exposure among GBMSM can provide insights into the psychological mechanisms driving preventive intentions and behaviors. This understanding can inform the development of targeted risk communication strategies that promote adaptive coping and vaccination. These relationships remain unexplored in the context of culturally sensitive mpox risk communication and vaccination campaigns for GBMSM in the Asia-Pacific region.

Behavioral intentions have been theorized (eg, theory of planned behavior) and shown to predict actual health-seeking behaviors [36], such as COVID-19 vaccination among UK adults aged 50 to 64 years [37], human papillomavirus vaccination among Dutch girls [38], and influenza vaccination for patients with diabetes in China [39]. However, Jongen et al [40] found that an increased level of perceived risk for mpox more consistently predicted and explained the difference between high intention for mpox vaccination and actual vaccination uptake among Dutch GBMSM attending sexual health services in a later survey wave. Similarly, in a cross-sectional survey of Canadian GBMSM, those with a lower perceived susceptibility to mpox were more likely to remain unvaccinated despite being offered vaccination after attending a sexually transmitted infection (STI) clinic [41]. Therefore, mpox illness perceptions may play a stronger role in the relationship between behavioral intent and subsequent uptake for GBMSM if it increases the perceived risk of contracting mpox—this hypothesized relationship is yet to be explored [42].

Mainland China and Hong Kong play pivotal roles in the global resurgence of mpox, largely due to their extensive interconnectivity and prominence as international air travel hubs. These 2 regions present distinct policy and social environments—Beijing, as the capital of mainland China, has limited vaccine availability and a higher level of societal stigma [43], whereas Hong Kong has implemented a fully subsidized vaccine program and has generally more open societal attitudes toward sexual minority groups [44]. Both cities experience significant cross-border interaction and are major population centers yet differ markedly in their public health policies, media environments, and levels of LGBTQ+ acceptance [44]. This contrast allows for a nuanced comparison of how structural factors—such as vaccine access, stigma, and risk communication—interact to shape health behaviors among GBMSM.

In 2023, China reported 1712 confirmed mpox cases, with the Beijing municipality (258 cases) and Guangdong province (342 cases) accounting for more than one-third of the country’s total cases [9]. Hong Kong, despite having only 68 confirmed cases, is surrounded by the Guangdong province with millions of daily cross-border travelers and observed steady local confirmed cases from January 2024 to August 2024 [45]. Using the extended CSM framework applied through multigroup structural equation modeling (SEM), this study aimed to examine how differential exposure to mpox information and perceptual processes may influence perceived susceptibility and vaccination intentions among GBMSM in these 2 highly interconnected municipal areas.

Given the multidimensional nature of information exposure, illness perceptions, and behavioral intent, this study used SEM because it enables the simultaneous estimation of complex, interrelated pathways and mediating effects, which are essential for understanding how latent constructs—such as threat perception and personal control—link information exposure to risk perception and vaccination intent. Unlike standard regression, SEM allows for the modeling of latent variables using multiple observed indicators, providing a more accurate representation of psychosocial processes. In addition, we used multigroup SEM to directly compare these pathways between GBMSM in Hong Kong and Beijing, allowing us to test for contextual differences in the relationships among key variables.

We hypothesized that regional differences in case numbers, risk communication, vaccine program availability, and cultural values lead to different levels of mpox information exposure, illness perceptions, and subsequent perceived risk and vaccination intentions. Specifically, we propose that the content of mpox information—factual, positive, or negative—may affect vaccination intentions through distinct perceptual processes by shaping threat or control perceptions. We further anticipate that these information attributes will differentially impact perceived risk and vaccination intent and that geographical and social differences between Hong Kong and Beijing may moderate these effects based on local case burdens, information exposure, perceived susceptibility, and vaccine access.

A cross-sectional survey was conducted with GBMSM in Beijing and Hong Kong from November 2023 to March 2024. Participants had to meet the following inclusion criteria: (1) being Chinese male individuals aged ≥18 years and (2) having engaged in anal intercourse with at least one man in the previous 6 months (geographic information and location of the study sites are presented in Figure 1). Recruitment and data collection methods were similar between Beijing and Hong Kong. In both cities, upon approval of the owners, trained and experienced field-workers approached potential participants in gay bars at different time slots on weekdays and weekends. On sites, the field-workers briefed potential participants about the study and gave them an information sheet. Online recruitment was also conducted in both cities by posting study advertisements on popular gay websites (in Hong Kong only), social media platforms such as Facebook (in Hong Kong only) or Weibo and WeChat (in Beijing only), and partner-seeking mobile apps frequently used by local GBMSM (Blued in Beijing and Grindr in Hong Kong). In Hong Kong, Facebook is the most popular social media platform, and Grindr is the most widely used gay dating app. However, these 2 online platforms are not available in mainland China. Weibo and WeChat are the 2 most commonly used social media platforms in mainland China, with functions very similar to those of Facebook. Blued is the most widely used gay dating app in mainland China, with features and functions similar to those of Grindr. Therefore, Weibo, WeChat, and Blued were chosen to facilitate online recruitment in Beijing. Potential participants who were interested in the study could contact the research team via private messaging, telephone, WhatsApp or WeChat, or email. Recruitment was supplemented through referrals made by peers and community-based organizations. Participants were screened for study eligibility. Participant anonymity was strictly maintained, and verbal informed consent was obtained. Various contact methods were used to arrange telephone interviews, which were conducted in Mandarin in Beijing and in Cantonese in Hong Kong.

This study was reviewed and approved by the Survey and Behavioural Research Ethics Committee of the Chinese University of Hong Kong (reference SBRE-22-0488). The research adhered to established ethical guidelines for human participant research. Verbal informed consent was obtained from all participants before taking part in the study. Participants were provided with a detailed information sheet outlining the purpose of the study, procedures, risks, and benefits before consenting. Consent was specifically obtained for the primary data collection, and no additional consent was required for the analysis as all data were anonymized.

Participant anonymity was strictly maintained throughout the study. Data collected during the phone interviews were deidentified at the point of collection, ensuring that no personally identifiable information was stored or linked to the responses. All data were securely stored in encrypted files accessible only to authorized researchers. Participants were compensated for their time upon completing the 30-minute phone interviews. In Hong Kong, participants received an HK $50 (US $6.40) supermarket coupon, and in Beijing, they were provided with a ¥20 (US $3) e-coupon. No identifiable images or personal data of individual participants were included in the manuscript or supplementary materials. As such, there is no risk of participant identification.

Recruitment and data collection procedures were standardized for both Beijing and Hong Kong. Trained field-workers, with permission from venue owners, approached potential participants at various times during the week and on weekends in gay bars. The participants were given a brief description of the study and an information sheet. Online recruitment included posting advertisements on popular gay websites (in Hong Kong), social media platforms (Facebook in Hong Kong and Weibo and WeChat in Beijing), and partner-seeking apps (Blued in Beijing and Grindr in Hong Kong). Interested individuals could contact the research team via private messages, phone, WhatsApp or WeChat, or email. Peer referrals and community-based organizations also supported recruitment efforts. Participant anonymity was strictly maintained, and verbal informed consent was obtained. Various contact methods were used to arrange telephone interviews, which were conducted in Mandarin in Beijing and in Cantonese in Hong Kong. To ensure content consistency, the study questionnaire was available in simplified Chinese for Beijing and traditional Chinese for Hong Kong.

The questionnaire was developed by a team of public health researchers, epidemiologists, health psychologists, physicians, and representatives of nongovernmental organizations. It was then pilot-tested with 5 GBMSM, and their feedback was used to refine the final version.

Sample characteristics were described using total row proportions, means, and SDs and by location (Beijing and Hong Kong), with differences between locations evaluated using the Pearson chi-square test or Wilcoxon rank sum test (analytic sample size: N=989; Hong Kong: n=470; Beijing: n=519) [49]. Zero-order (Pearson) correlations were computed to evaluate correlations between key variables in this study. Missing data were imputed using the nonparametric random forest algorithm [50]. Linear regression was conducted to examine the bivariate relationship between background variables and mpox vaccination intention in the following 6 months. All significant background variables in univariate regression were adjusted for in the following analyses aided by directed acyclic graphs to avoid omitted confounding effects and conditioning on a collider [49,51].

SEM was conducted to examine the hypothesized model using a 2-step approach. First, a confirmatory factor analysis was conducted to assess the goodness of fit of the measurement model. The construct reliability of the latent variables was evaluated using both the Cronbach α and McDonald ω [52]. SEM was used to test the proposed structural model. The mean and variance–adjusted weighted least squares estimator was used to estimate modeled parameters accounting for ordinally distributed endogenous outcomes [49,53]. Indirect effects were estimated using Bollen-Stine bootstrapping with 5000 repetitions, which is robust against deviation from multivariate normality corresponding to mean and variance–adjusted weighted least squares estimator and ordinal endogenous outcomes [54]. Overall model fit was determined using (1) the scaled chi-square badness-of-fit index (P>.05), (2) a scaled root mean square error of approximation (S-RMSEA) of <0.06, (3) the scaled comparative fit index (S-CFI), and (4) a scaled Tucker-Lewis index (S-TLI) of >0.95 [51,55,56].

Multigroup SEM was further used to examine potential differences in mpox information exposure and their associations with mpox perceptions across cities (moderation by city). We tested for partial scalar invariance in the measurement model and partial path coefficient invariance in the structural model. Invariance was determined by a difference in the S-CFI of <0.01 and a significant log-likelihood ratio test (P<.05) between the baseline models (ie, configural fits) and models with equality constraints on each of the latent factor loadings (weak or metric invariance) and thresholds or intercepts (strong or scalar invariance) [57,58]. The moderation of parameter estimates from the final multigroup structural equation model was examined using a 2-sided Wald test for heterogeneity using the appropriate unstandardized coefficients (B) and variance parameters [59], with P<.05 indicating a significant difference. We conducted sensitivity analyses of the direct and indirect effects of multigroup SEM by replicating the same model restricted to participants from Hong Kong and Beijing separately (Multimedia Appendix 1). Psychometric analyses were conducted using the psych and lavaan packages in R (R Foundation for Statistical Computing) [60,61].

Of the eligible participants approached through outreach in gay venues, online recruitment, and peer referral (n=713 from Hong Kong and n=602 from Beijing), 86% (613/713) from Hong Kong and 87% (524/602) from Beijing provided verbal consent and completed telephone interviews. Of the original total sample (N=1137), 0.44% (5/1137) of participants from Beijing and 12.58% (143/1137) of participants from Hong Kong were vaccinated against mpox and were removed from the analytic sample. Table 1 presents the unvaccinated participant characteristics by total sample (N=989) and group (Beijing: 519/989, 52.5%; Hong Kong: 470/989, 47.5%). Among the participants from both groups, most were aged ≥25 years (859/989, 86.9%), tertiary educated (836/989, 84.5%), single (752/989, 76%), and employed full time (774/989, 78.3%); identified as homosexual (823/989, 83.2%); and earned above the median city income (612/989, 61.9%). Mpox vaccination rates were significantly higher in Hong Kong than in Beijing (116/519, 22.4% vs 4/470, 0.9%; P<.001); however, the average levels of vaccination intention (mean score 2.31, SD 1.24 in Beijing vs mean score 2.32, SD 1.22 in Hong Kong; P=.60) and perceived risk of contracting mpox (mean score 1.84, SD 0.88 in Beijing vs mean score 1.86, SD 0.82 in Hong Kong; P=.90) among those who were unvaccinated were not significantly different between Beijing and Hong Kong. GBMSM in Hong Kong were less likely to be married, cohabit with a woman, or identify as bisexual or other. Participants in Beijing reported a higher uptake of sexual health services, including HIV testing (440/517, 85.1% vs 185/470, 39.4%; P<.001) and PrEP use (100/517, 19.3% vs 39/470, 8.3%; P=.03), in the previous 6 months.

Table 2 shows the means and SDs of the observed items of the latent factors from the measurement model. The average self-rating of most observed items of the latent factors was higher among participants in Beijing than among those in Hong Kong except for some mpox illness perceptions (Brief Illness Perception Questionnaire) measures. Reliability coefficients indicated adequate internal consistency for most latent factors except for the control perception domain of mpox illness perception. Thus, only 1 personal control item from mpox illness perceptions was used in subsequent structural equation models. Zero-order correlations showed high correlations among factual, positive, and negative mpox information content (Table 3). The threat and personal control perception domains of mpox illness perceptions had low to moderate correlations with other latent variables, indicating discriminant validity that was consistent across the Beijing and Hong Kong subsamples.

Table S1 in Multimedia Appendix 2 shows that the bivariate relationships of having tertiary education or higher, having an income level above the city’s median income, having been tested for HIV and STIs and using PrEP, having higher numbers of regular and nonregular sex partners in the previous 6 months, and having a higher perceived risk for contracting mpox in the following 12 months were associated with higher levels of intent to vaccinate against mpox in the following 6 months. Figure S1 in Multimedia Appendix 2 shows the directed acyclic graphs of the hypothesized causal relationships to derive a minimal sufficient adjustment set for confounding control consisting of age, educational level, HIV testing, PrEP use, and number of sexual partners [62,63]. Multimedia Appendix 3 shows the Strengthening the Reporting of Observational Studies in Epidemiology statement checklist.

The indirect effect of positive mpox information exposure on vaccination intent through personal control and mpox risk was significant and positive in Hong Kong (B=0.01, 95% CI 0.002-0.03) but not in Beijing (B=−0.01, 95% CI –0.01 to 0.003), with a significant difference between cities (Wald test: Z=2.52 and P=.01). The indirect effect through personal control was significant and positive in Hong Kong (B=0.05, 95% CI –0.002 to 0.01) but not in Beijing (B=−0.01, 95% CI –0.007 to 0.01), with a significant difference between cities (Wald test: Z=2.17 and P=.03).

Since the 2022 multicountry mpox outbreak, mpox has continued to spread in places with little access to vaccines, whereas its resurgence across countries is imminent [67]. China is home to >12 million GBMSM and has the highest mpox case burden in the Western Pacific region [4,68]. However, only 1% (5/524) of participants from China’s mpox epicenter, Beijing, had been vaccinated against mpox. To anticipate the regulatory approval of mpox vaccines in mainland China, this study examined the novel pathways through which exposure to different attributes of mpox information and perceptual processes could influence GBMSM’s vaccination decisions, specifically by raising GBMSM’s level of perceived risk of contracting mpox. Our findings provide timely implications for GBMSM-tailored risk communication and community partnerships to raise awareness of mpox among GBMSM across borders.

Our study highlights the critical roles of mpox-related information content and disease-related perceptions in shaping perceived risk and subsequent vaccine-related decision-making among GBMSM, extending the findings of Jongen et al [40]. In Hong Kong, increased exposure to positive mpox information was associated with higher personal control and vaccination intent, mediated by increased perceived risk. However, this relationship was not observed in Beijing. Factual mpox information was associated with increased vaccination intent in both cities, as demonstrated by the multigroup structural equation model. Correlation analyses also revealed that perceived risk was positively correlated with personal control perceptions in Hong Kong but not in Beijing. Vaccination intent had a stronger positive correlation with personal control in Hong Kong than in Beijing. These results suggest that, in places where fully government-sponsored mpox vaccines are in place, promoting factual and positive mpox information (eg, positive portrayal of patients with mpox, recovery stories, and health advice) could potentially increase vaccination intent by raising the awareness of their level of perceived risk of contracting mpox, which may also promote GBMSM’s personal control of mpox, which may include having accessible vaccination programs in place at GBMSM’s disposal and their related promotional campaigns enacting effective coping strategies. This finding is consistent with behavior change theories, including the extended CSM framework underlying the role of control perceptions in motivating vaccine uptake behaviors [32].

Our study shows that exposure to negative mpox information could raise GBMSM’s threat perceptions of mpox without necessarily increasing their perceived risk of contracting mpox and subsequent vaccination intent but that it is associated with increased behavioral intention for mpox vaccination. These results suggest that the association between negative mpox information and behavioral intention for mpox vaccinations through threat perceptions may be less predictive of actual vaccination behaviors as compared to positive content through personal control, which in turn increases both risk perceptions and behavioral intention to receive the mpox vaccine. Highly threatening or stigmatizing content may trigger maladaptive coping behaviors such as denial, avoidance, or disengagement [33]. Rather than motivating action, overwhelming threat can result in fatalism or psychological withdrawal, especially in communities with limited resources or structural barriers to prevention (eg, lack of vaccine access in mainland China). This finding may correspond to the high prevalence of sensationalized headlines and exaggerated reporting of severe mpox cases, mortality, and HIV or AIDS coinfection across regions [69]. This negatively charged media coverage that appeals to the public’s consumption of polarizing views and affective processes without providing risk information relevant to GBMSM, including unmoderated homophobic user comments on media outlets (homophobic slurs remain unmoderated in one of Hong Kong’s online news outlets [70]), could counter mpox control efforts [4]. Among Chinese adults exposed to disgusting graphic mpox information, those with higher disgust endorsed more avoidant preventive measures and perceived higher mpox severity but not susceptibility or perceived risk of contracting mpox [71]. In addition, higher levels of mpox-related stigma attenuate the association between perceived severity and preventive intent [71]. Emotionally provoked viewers, including GBMSM, may dissociate from their own mpox risk and view those at risk as “distant others” (eg, people living with HIV) who should be receiving the vaccine rather than themselves [72]. In addition, exposure to stigmatizing mpox information could galvanize GBMSM’s self-stigma and identity concealment, which negatively predicted mpox vaccination intent and uptake among Canadian GBMSM in the study by Le Forestier et al [8].

Despite higher levels of mpox information exposure and illness perceptions, GBMSM from Beijing were different from those from Hong Kong. This suggests that GBMSM from Beijing may not be able to identify relevant mpox risk information even with more exposure to local cases and case information. In Hong Kong, health authorities collaborate with community partners and media channels to disseminate tailored mpox risk communication and vaccination promotion for GBMSM and remove hate speech toward LGBTQ+ individuals in online news outlets [73,74]. Moreover, our finding that personal control perception more strongly predicted vaccination intention among GBMSM in Hong Kong than in Beijing may be explained by several contextual and theoretical factors. First, Hong Kong’s public health infrastructure and government-sponsored vaccination programs are widely regarded as transparent and accessible, which can enhance individuals’ trust in health authorities and the perceived efficacy of their own preventive actions. This aligns with the CSM, which posits that, when individuals believe that their actions (such as seeking vaccination) will be supported and effective within their health care environment, perceptions of personal control are more likely to translate to concrete health behaviors [32,33]. In contrast, in Beijing and mainland China, limited access to mpox vaccines, lower visibility of LGBTQ+-inclusive health campaigns, and persistent stigma may undermine individuals’ confidence that their actions will be effective or socially supported. In addition, Hong Kong’s relatively open media environment and consistent dissemination of affirmative health messages may reinforce the notion that vaccination is both accessible and beneficial, thus strengthening the link between personal control and intent. Conversely, in Beijing, mixed or negative messaging, combined with structural and social barriers, may weaken this relationship [75,76]. Thus, the interplay among health care system trust, policy transparency, and supportive media narratives likely explains why personal control perceptions are more predictive of vaccination intention in Hong Kong.

Conversely, GBMSM from mainland China may lack adequate exposure to relevant environmental risk communication. Leiwen et al [77] found that only one-fourth of mainland Chinese GBMSM correctly identified GBMSM as being at risk of contracting mpox, whereas half believed that only people living with HIV are at risk. Similarly, Zheng et al [78] found that more than half of GBMSM incorrectly identified that there were effective medications for mpox and less than half could identify >3 ways to prevent mpox infections. These findings suggest that tailored and comprehensive risk communication for GBMSM in mainland China could improve their mpox risk self-assessment and promote preventive measures even when vaccines are not within the range of options, for example, reducing the number of sexual partners and using condoms to reduce the risk of mpox transmission through direct genital contact [79]. Such communication could be synergized with positive portrayals of stories of patients with mpox, such as gay-identifying physicians and social media influencers with mpox [15,16], in raising GBMSM’s cognitive perceptions of mpox and reducing their emotional distance on perceived susceptibility while empowering them to enact preventive behaviors.

The results of this study should be interpreted within their limitations. First, the cross-sectional design cannot ascertain causality, specifically the temporal direction of how previous information affects subsequent intent. It is causally plausible that behavioral intent assessed at 1 point in time is a function of previous information and psychological processes. Longitudinal studies are preferred to clarify causality and the potential for reverse or bidirectional effects. Second, our recruitment strategy—using social media, gay bars, and peer referrals—may have introduced selection bias. This approach likely overrepresented younger, more digitally connected, and socially active GBMSM who are comfortable accessing online platforms or community venues while potentially underrepresenting older, less connected, or more marginalized individuals who may not participate in such spaces. As a result, the findings may not fully reflect the experiences or perspectives of the entire GBMSM population in Beijing and Hong Kong, particularly those less engaged with the LGBTQ+ community or less reachable through digital or venue-based outreach. Third, there is a possibility that unmeasured confounders influenced our study results. Factors such as previous vaccination hesitancy, underlying attitudes toward vaccination, previous experiences with health care systems, or structural barriers to health care access were not directly measured but could have independently affected both information exposure and vaccination intentions. Fourth, although our study included participants from 2 major urban centers—Beijing and Hong Kong—the generalizability of our findings to GBMSM living in smaller cities, rural areas, or other regions of China may be limited. GBMSM in less urbanized areas may have different patterns of information exposure, levels of stigma, health care access, or vaccine availability, which could influence both risk perceptions and preventive behaviors. Finally, our sample of participants from Beijing had a higher level of education, included less individuals who identified as bisexual, and had been more recently HIV tested than participants in another respondent-driven sample of GBMSM from Beijing [80]. This indicates that we were more likely to have overestimated the prevalence of behavioral intent for mpox vaccination. However, our population parameter estimate is still more conservative than those in other studies (>80%) with larger sample sizes [78,81]. On the other hand, participants from Hong Kong had similar characteristics to those of a larger, more representative sample [82]. All data were self-reported and, therefore, were subject to recall and social desirability biases. The use of trained interviewers with structured questionnaires and interview designs in this study may have reduced some reporting bias. However, questions related to sexual behavior could be stigmatized when asked by interviewers, which could have been underreported.

This study highlights the critical role of mpox-related information exposure in influencing behavioral intentions toward vaccination among GBMSM in Beijing and Hong Kong. Positive information exposure, particularly regarding recovery stories and prevention, significantly increased vaccination intent by enhancing personal control perceptions and perceived risk of contracting mpox among GBMSM in Hong Kong, where a fully government-subsidized mpox vaccination program is in place. Conversely, while negative information related to mpox increased threat perceptions and vaccination intentions, it did not reliably translate to a higher perceived risk, especially in Beijing. These findings suggest several practical implications for public health messaging aimed at increasing mpox vaccine uptake among GBMSM in China. Health agencies should prioritize accurate, positively framed, and stigma-free information—such as recovery stories and practical prevention advice—delivered through channels or key opinion leaders trusted by the target community. Proactive efforts are needed to counter misinformation and stigma, for example, through myth-busting campaigns and the promotion of supportive narratives on both social and traditional media platforms. In addition to the careful moderation of stigmatizing messaging, news media channels should implement stricter standards when framing their reporting on mpox and moderate user comments on their platforms to prevent the spread of misinformation that could harm public health. Partnerships with LGBTQ+ organizations and community leaders can enhance the credibility and reach of messaging, whereas targeted outreach via social media and mobile apps can ensure that information is accessible to those most at risk. Finally, communications should provide clear guidance on accessing vaccination and related services and be culturally sensitive and inclusive to empower GBMSM to assess their risk realistically and take preventive action.