A Digital Health Intervention on Healthy Behaviors Among Men Who Have Sex with Men in China Based on Information-Motivation-Behavioral Skills Model: An Exploratory Real-World Study

Abstract

Background

Men who have sex with men (MSM) remain significantly affected by HIV/AIDS due to high-risk sexual behaviors. The adoption rate of preventive measures like pre-exposure prophylaxis (PrEP) use and HIV testing among MSM is insufficient in China, which requires more effective prevention strategies.

Objective

This study aims to evaluate the feasibility and effectiveness of an Information-Motivation-Behavioral Skills (IMB) model-based digital health intervention in promoting healthy behaviors and reducing high-risk activities among MSM in China.

Method

A real-world exploratory study in a naturalistic setting without strict control over participants’ daily behaviors was conducted using the Blued platform (a popular online geosocial networking platform in China) to deliver the self-developed IMB-based intervention materials to MSM in 3 cities in China, with 3 additional cities as the controls. A questionnaire including items on healthy behavior adoption was administered pre- and post-intervention and finally generated an unmatched pseudo-panel dataset. Multiple logistic regression, linear regression, and structural equation modeling analyses were used to examine the association between the IMB model and outcomes. The χ² test, Fisher’s exact test, and two-sample independent t-test were utilized to identify potential confounders for further adjustment, and were applied in both the intention-to-treat (ITT) and as-treated (AT) analyses to explore the effectiveness of the intervention.

Results

Post-intervention, the intervention group showed a significant increase in HIV testing rates and PrEP awareness compared to the control group. The willingness to use oral PrEP was also improved. The intervention materials were favorably received, with most of the MSM perceiving the form and content of these materials as useful and attractive.

Conclusion

The IMB model-based digital health intervention effectively increased HIV testing rates and PrEP awareness among MSM. The high acceptance of the intervention materials suggests that such digital platforms can be a cost-effective strategy for HIV prevention in hard-to-reach populations.

1. Introduction

Sexual health is a multidimensional concept encompassing both the absence of sexually transmitted infections (STIs) and dysfunction and a state of physical, emotional, mental, and social well-being in relation to sexuality (Hansen & Nichols, 2024). HIV prevention is inseparable from a broader commitment to promoting the overall sexual health, rights, and well-being of men who have sex with men (MSM). MSM are a vulnerable group in the HIV/AIDS epidemic. Due to deficiencies in knowledge about Human Immunodeficiency Virus (HIV) and a lack of self-protection awareness, MSM may engage in high-risk sexual behaviors, such as unprotected intercourse, having multiple sexual partners, and participating in chemsex, which could potentially escalate their susceptibility to HIV infection (Ramos et al., 2021; Avallone & Hickson, 2024). However, HIV and sexual health risks among MSM are not solely attributable to individual-level knowledge and motivation. Additionally, substantial evidence has indicated that HIV risk among MSM is shaped by a complex interaction of psychosocial, structural, and cultural factors. These include barriers to consistent condom use and safer sex practices caused by stigma associated with homosexuality, challenges in negotiating condom use within sexual partnerships, desires for intimacy and sexual pleasure, relationship dynamics such as trust or infidelity, experiences of discrimination across social and institutional contexts, and underlying mental health vulnerabilities (Capodieci et al., 2025). These same factors not only elevate HIV risk, but also negatively impact broader aspects of sexual health for MSM, including sexual satisfaction, self-efficacy, and access to needed services (Babel et al., 2021). MSM are disproportionately affected by HIV. It has been estimated that among MSM, the incidence of HIV infection worldwide was 1.3% in 2023, and the median prevalence of HIV infection worldwide in 2022 was approximately 7.5%, which is higher than that of the general adult population (UNAIDS, 2023). In 2020, MSM accounted for the majority of new HIV infections in several regions: 75% in Western Europe, the USA, and Canada, 53% in Asia, 46% in Latin America, 21% in the Caribbean, 20% in the Middle East and northern Africa, 16% in eastern Europe and central Asia, 14% in western and central Africa, and 4% in eastern and southern Africa (Operario et al., 2022). In China, the incidence of HIV infection among MSM was 5.7% from 2001 to 2018, with a continuing upward trend and regional variations (Dong et al., 2019; Zhu et al., 2019; WHO, 2023). Southwest China had the highest prevalence at 10.7% with Chongqing reaching 20.5%, followed by east China at 6.3%, while north, south, and northwest China had lower prevalence rate at 5.2%, 5.1%, and 3.7%, respectively (Dong et al., 2019; Zhang et al., 2019). Additionally, HIV infection can subsequently lead to Acquired Immune Deficiency Syndrome (AIDS) when the immune system has been sufficiently damaged by HIV (Ramos et al., 2021; Avallone & Hickson, 2024). The mortality rate of AIDS reached 2.23 per 100,000 population in 2019, representing 71.34% of total fatalities from China’s Notifiable Diseases (Xu et al., 2021). Furthermore, AIDS was responsible for 139 Disability-Adjusted Life Years (DALYs) per 100,000 population, constituting 0.37% of the total DALYs in China in 2019 (Xu et al., 2021).

Sexual transmission is the predominant pathway for HIV infection, and nearly 28% of HIV infections in China can be attributed to high-risk sexual behaviors among MSM (CDC, 2016). Preventative measures such as pre-exposure prophylaxis (PrEP) and post-exposure prophylaxis (PEP) have been proven effective in controlling the spread of HIV. However, the utilization of PrEP and PEP remains low among MSM in China, with only 1.2% and 2.3% having ever used them, respectively (Wang et al., 2024; Guan et al., 2023). Moreover, nearly 40% of MSM in China reported no intention to use condoms during sexual activity, and only 23% reported no history of chemsex in China (Zhao et al., 2016; Cao et al., 2023). A cross-sectional study in China also indicated that 8.0% of MSM never use condom in sexual activity (Hong et al., 2022). Furthermore, from 2016 to 2020, half of MSM reported not having undergone HIV testing in the last 12 months (Xu et al., 2022). These statistics not only implied a high burden of HIV, but also reflect broader gaps in comprehensive sexual health for MSM.

Within the sociocultural context of China, MSM remain largely stigmatized and socially unaccepted, often resulting in familial violence and potential discrimination in schools or the workplace (Liu & Choi, 2006; Wang et al., 2019). Furthermore, the association between anal sex and HIV exacerbates the stigma among MSM. Thus, some MSM may choose to conceal their sexual orientation, avoid engagement with the MSM community, and even internalize homophobic attitudes, thereby reducing their likelihood of accessing HIV-related health services (Xu et al., 2017). In addition, disparities in preventive measures remain prevalent among diverse MSM populations. Previous studies have indicated that bisexual MSM face challenges in terms of integration and acceptance within the gay community (Hong et al., 2022). As a result, conventional peer-led and community-based interventions frequently fail to effectively target bisexual MSM, reducing their likelihood of receiving HIV prevention information compared to gay men. Furthermore, with the accelerated development of internet technology, users from different age groups are drawn to diverse types of content, influencing their sexual behaviors and potentially shaping their perceptions and decisions regarding sexual health and the algorithms of online platforms further exacerbate these differences (Lyu & Chai, 2024). A sociological study indicated that media consumption, particularly engagement with new media platforms such as the internet, has become an increasingly important factor in shaping sexual attitudes and knowledge in contemporary China (Lyu & Chai, 2024). These findings highlight the urgent need to develop broader and more effective AIDS prevention interventions for China’s diverse MSM population.

Healthy behavior adoption is a complicated process, and researchers have proposed several theoretical models for promoting behavior transition, such as the Theory of Reasoned Action (TRA), Theory of Planned Behavior (TPB), Protection Motivation Theory (PMT), Health Belief Model (HBM), and Information-Motivation-Behavioral Skills (IMB) model (Manstead et al., 1983; Pan et al., 2023; Wu et al., 2022; Rosenstock, 1974; Fisher & Fisher, 1992). Among these models, the IMB model was initially introduced to understand the mechanism behind HIV-risk behavior change among various populations (Fisher et al., 2006; Fisher et al., 1996). The IMB model systematically categorizes the behavioral determinants into three distinct dimensions: information, motivation, and behavioral skills. It aims to clarify the direct or indirect associations between each dimension and behavior, as well as the direction and strength of these associations, to predict preventive behavior and explore the best solution for behavioral intervention. According to the IMB model, facilitating behavioral transition requires a multi-step approach. Initially, interventionists should disseminate pertinent information or knowledge concerning the desired behavioral change. Concurrently, an environment conducive to contemplating health behavior transition should be cultivated. This environment should enhance motivation toward behavior transition. Subsequently, the requisite skills should be equipped to actualize the behavior transition. Finally, behavior change occurs when the levels of information, motivation, and behavioral skills reach a certain threshold. The model framework used in this study on HIV-related healthy behaviors is displayed in Figure 1.

Figure 1.

Information-motivation-behavioral skills model.

With the accelerated development of internet technology, digital health interventions have garnered increased attention among hard-to-reach populations such as MSM (WHO, 2019). According to the China Internet Network Information Center, by December 2024, the number of internet users in China had reached 1.108 billion, with 99.7% accessing the internet via mobile phones (CNNIC, 2025). This widespread mobile connectivity provides a solid foundation for digital health interventions in China. Digital health is defined as the application of information and communication technology in promoting population health (WHO, 2019). Identifying a cost-effective, efficient, practical, and sustainable strategy to encourage MSM to adopt HIV prevention measures is crucial for current HIV prevention efforts in China. Digital health platforms play a pivotal role in HIV prevention among the MSM community. These platforms not only transcend geographical barriers to extend the reach of interventions and promote health equity but also allow multimedia content to effectively engage the target population, facilitating improvements in their health outcomes. Furthermore, digital health platforms provide targeted and personalized online interventions for MSM, which diminish the risks associated with discrimination and stigmatization, thereby enhancing the effectiveness and acceptability of HIV prevention efforts.

Numerous digital health interventions designed to promote sexual health have been implemented among MSM populations, but the effectiveness of these interventions remains various (Du et al., 2025). An intervention study on PrEP uptake among MSM in China utilized a WeChat mini-program for content delivery, but failed to achieve statistically significant outcomes. The authors attributed this limitation to insufficient content updates and restricted diversity in presentation formats, which undermined sustained user engagement (Li et al., 2023). This highlights the critical importance of designing intervention materials with dynamic, multifaceted formats to enhance user involvement. Additionally, a notable gap in most existing studies is the absence of robust behavioral theory frameworks, which may inherently limit the interventions’ capacity to foster sustained health behaviors. Previous studies have suggested that the IMB model is a robust predictor of health behaviors in MSM, particularly when integrated with digital health interventions. A randomized controlled trial conducted in India found that an internet-based messaging intervention based on the IMB model increased the HIV testing rate over 6 months by 12.3% in MSM (Patel et al., 2020). Similarly, a randomized controlled trial in Thailand showed that a mobile phone app-based online intervention designed within the IMB framework increased adherence to daily PrEP engagement in MSM (Kawichai et al., 2022). Further evidence from a randomized controlled trial in China and a qualitative study in the U.S. indicated that internet-based IMB interventions could potentially increase medication adherence among HIV-positive MSM (Fan et al., 2023; Dworkin et al., 2018). Therefore, digital health interventions targeting the promotion of healthy behaviors among MSM based on the IMB model may potentially facilitate the adoption of healthier behaviors and decrease engagement in high-risk activities.

Currently, there are few studies examining the effect of IMB-based online interventions on the adoption of HIV-related healthy behaviors among MSM in China, and effective measures to systematically promote healthy behaviors for HIV prevention within this population are scarce. A research team in China has established an online digital health intervention and corresponding materials based on the IMB model to improve Human Papillomavirus (HPV) vaccination willingness among female college students. The effectiveness of the intervention materials has been proven through a randomized controlled trial and subsequent follow-up (Si et al., 2019; Si et al., 2021; Si et al., 2022; Si et al., 2022). To address this gap, our research team developed a set of intervention materials based on the IMB model targeting MSM, consisting of 12 articles/videos. We conducted a real-world digital health intervention, defined as research conducted in a naturalistic setting where participants’ daily behaviors and engagement with the intervention were not strictly controlled (i.e., reflecting real-life conditions rather than a highly regulated research environment), by delivering these materials via the Blued platform, a popular online geosocial networking (GSN) platform with over 40 million active MSM users. Blued also serves as a health education platform promoting healthy lifestyle practices, with numerous interventional and observational studies targeting MSM conducted on this platform (Wang et al., 2019; Dai et al., 2024). This exploratory study aims to assess the effectiveness of an IMB-based digital health intervention in promoting healthy behaviors for HIV prevention among MSM in a real-world setting (i.e., without artificial restrictions on participants’ daily activities or mandatory adherence to the intervention) using the Blued platform. Additionally, this study will also evaluate how the intervention influences MSM’s knowledge, motivation, and behavioral skills regarding HIV prevention, as well as their attitudes toward the form and content of the intervention. This study is expected to provide references for the prevention of HIV and interventions to promote the adoption of healthy behaviors in MSM.

2. Methods

2.1. Material development

The intervention materials were developed based on the IMB model framework, which contained three modules: “Information”, “Motivation”, and “Behavioral skills” (see Figure 1). The intervention materials were developed following a rigorous and comprehensive process to ensure their quality and effectiveness. The development was grounded in an extensive review of the existing literature and the experience from previous similar interventions conducted by our teams. Additionally, an iterative process involving several rounds of revisions was employed, and each round was informed by the feedback and insights from a panel of specialists in epidemiology, psychology, behavioral science, and venereology. We also conducted a series of focus group discussions with MSM from different backgrounds to understand their information needs, preferred learning styles, and concerns regarding HIV prevention. These could ensure that the materials were both theoretically sound and practically applicable in real-world settings.

2.2. Modules of the intervention materials

2.2.1. Information

The “Information” module was designed to improve knowledge related to HIV prevention among MSM and contains three thematic educations: knowledge on HIV infection and its related diseases, high-risk factors and protective factors of HIV infection, and prevention and treatment of HIV. The details are summarized in Figure 2.

Figure 2.

Intervention materials summary.

2.2.2. Motivation

The “Motivation” module was designed to improve MSM’s positive attitude toward healthy behavior (i.e., personal motivation, including personal attitudes and behavioral intentions) and related subjective norms (social motivation). This module contains four thematic educations: susceptibility and severity of HIV infection (to arouse MSM’s psychological resistance to high-risk behaviors), the benefit of adopting HIV prevention behaviors, and social norms on healthy sexual behaviors. The details are summarized in Figure 2.

2.2.3. Behavioral Skills

The “Behavioral Skills” was designed to improve MSM’s decision-making when faced with potential high-risk behaviors, self-efficacy, and objective skills in HIV prevention. This module contains 3 thematic educations: techniques to make decisions with conviction and obtain understanding and support from partners through effective communications, the strategies to facilitate the conversion of motivation into tangible action, and the skills for self-protection from HIV infection. The details are summarized in Figure 2.

2.3. Study design and participants

This study is an internet-based, multi-center, non-randomized, real-world exploratory trial. Participants were stratified and sampled from the Blued platform in China based on geographical divisions from December 16th, 2020 to March 1st, 2021. We selected MSM from six cities based on the number of basic users and active status in the platform, the prevalence of HIV infection, the city’s cost of living (to ensure that the research indicators are not influenced by economics and drug accessibility), and the city population size. These cities were divided into two groups: the intervention group, which includes Nanning, Wenzhou, and Tangshan; and the control group, which includes Nanchang, Shantou, and Dalian. During the recruitment process, researchers disseminated study posters containing a description of this study to potential participants via the Blued messaging interface. Interested MSM could then access an online questionnaire by clicking on the link provided in the study posters. Before proceeding with the questionnaires, participants were required to read the informed consent presented on the initial page and express their willingness to participate by clicking a button labeled “I am willing to participate in this study.” To screen the participants, we first presented a set of prescreening questions on the initial page of the online questionnaire. These questions were in line with our inclusion and exclusion criteria. The inclusion criteria were: (1) being biologically male; (2) having had oral/anal sex with men in the last year; (3) never having had an HIV testing or the testing result was negative. The exclusion criteria were: (1) being below 18 years old; (2) self-reporting as HIV positive. The Blued user identifier (uid) was used to exclude duplicated records automatically by the software. A few logical questions were embedded in the questionnaire for quality control. The criteria for an invalid questionnaire were as follows: (1) the question “In the past 6 months, have you engaged in anal sex with a male?” was answered as “No”, but the response to the question “In the past 6 months, how many males have you engaged in anal sex with?” is greater than 0; or the question “In the past 6 months, have you engaged in anal sex with a male?” was answered as “Yes”, but the response to the question “In the past 6 months, how many males have you engaged in anal sex with?” is equal to or greater than 300. (2) The options exhibit a suspicious pattern, such as all items having identical choices. Considering the exploratory nature of this study, we did not follow a rigorous sample size calculation. Instead, we aimed to attain a minimum sample size of at least 600 participants in each group. Ultimately, our study included 810 samples in the pre-intervention phase and 620 samples in the post-intervention phase. However, due to the privacy policy of the Blued platform, we did not have access to the uid of the participants. Therefore, it was not feasible to match the participants’ data pre- and post-intervention, and we finally generated a pseudo-panel dataset. No incentives were offered to the participants in this study to avoid potential bias in participant responses. Additionally, all items in the questionnaire were designed as mandatory responses. Participants were unable to proceed to the next section without completing each question. As a result, the final dataset contains no missing values. Ethical approval for this study was obtained from the Ethics Committee of Danlan Beijing Media Limited on May 20, 2020, and March 10, 2023 (Number: DLIRB202005-01, DL202303-3), and the Ethics Committee of Chinese Academy of Medical Science on May 12, 2023 (Number: CAMS&PUMC-IEC-2023-021).

2.4. Intervention and implementation

Participants in the cities assigned to the intervention group received digital IMB-based intervention materials via the Blued platform from August 23, 2021, to July 28, 2022. The materials were delivered in the form of splash-screen advertisements and private messages. MSM who clicked on the materials were directed to content related to the materials. The directed platform included Youzan (Youzan Article), WeChat (WeChat Article or WeChat video), and Blued (Blued Dynamics or Blued Happy Testing). The specific delivery dates, forms, and directed platform of the materials are summarized in Supplementary Table 1. The number of the materials delivered ranged from 13,616 to 57,547, and the specific delivery and dynamics of the intervention materials are displayed in Supplementary Table 2, including directed location, number of deliveries, private message click number, private message click rate, link click number, and link click rate. Participants in the cities assigned to the control group received regular health education related to HIV prevention. The follow-up questionnaires were delivered from March 17, 2023, to May 28, 2023, via the Blued platform based on the cities.

2.5. Measures

2.5.1. Primary outcomes

Awareness of PrEP: Participants’ awareness of PrEP was measured by one item “Have you ever heard of PrEP?” with a binary response of “No” and “Yes”.

HIV testing in the last 6 months: Participants’ HIV testing in the last 6 months was measured by one item “Have you ever tested for HIV infection status in the last 6 months?” with a binary response of “No” and “Yes”.

Chemsex in the last 6 months: Participants’ chemsex in the last 6 months was measured by one item “Have you ever used substance (such as rush popper, 5-MeO-Dipt, Amphetamines, and hallucinogen) in the last 6 months during sexual intercourse?” with a binary response of “No” and “Yes”.

Information level: Participants’ information level was measured by the HIV Knowledge Questionnaire (HIV-KQ-18) developed by Carey in 2002 (Carey & Schroder, 2002). This questionnaire comprises 18 items, each requiring a response of “Yes”, “No”, or “Unclear”. Participants received one point for each correct response, and these points were aggregated to calculate a total score for each participant. Higher total scores indicated a higher level of HIV knowledge. The Kuder-Richardson reliability of the questionnaire in this study was 0.753 at baseline and 0.711 at post-intervention assessment.

Motivation level: Participants’ motivation level on HIV prevention was measured by a motivation scale developed by our teams based on previously validated scales (Si et al., 2021; Dai et al., 2023). This scale contains 18 items, and each item was 5-point Likert scaled from 1 (completely disagree) to 5 (completely agree). Higher total scores indicate a higher level of motivation for HIV prevention. The Cronbach’s α of the scale in this study was 0.792 at baseline and 0.821 at post-intervention assessment.

Behavioral skills level: Participants’ behavioral skills level on HIV prevention was measured by a behavioral skills scale developed by our teams based on previously validated scales (Si et al., 2021; Dai et al., 2023). This scale contains 10 items, and each item was 5-point Likert scaled from 1 (completely disagree) to 5 (completely agree). Higher total scores indicate a higher level of behavioral skills in HIV prevention. The Cronbach’s α of the scale in this study was 0.907 at baseline and 0.934 at post-intervention assessment.

2.5.2. Secondary outcomes

Condom use in the last 6 months: Participants’ condom use in the last 6 months was measured by one item “Have you always used condom when having sex with male in the last 6 months?” with a binary response of “No” and “Yes”. Only those who had had sex with male in the last 6 months would answer this question.

Commercial sex in the last 6 months: Participants’ commercial sex in the last 6 months was measured by one item “Have you had commercial sex with male in the last 6 months?” with a binary response of “No” and “Yes”. Only those who had had sex with male in the last 6 months would answer this question.

Use of oral PrEP: Participants’ use of PrEP was measured by one item “Have you ever used oral PrEP?” with a binary response of “No” and “Yes”. Only those who had ever heard of PrEP would answer this question.

Willingness to use oral PrEP: Participants’ willingness to use PrEP was measured by one item “If taking PrEP daily or continuously before and after each sexual encounter can effectively prevent HIV infection, would you consider using it in the future?” with a 5-Likert scaled response from 1 (completely unwilling) to 5 (completely willing). Only those who had not used PrEP would answer this question.

Willingness to use long-acting injectable PrEP: Participants’ willingness to use long-acting injectable PrEP was measured by one item “If taking a long-acting PrEP injection every two months can effectively prevent HIV infection, would you consider using it in the future?” with a 5-Likert scaled response from 1 (completely unwilling) to 5 (completely willing).

Use of oral PEP: Participants’ use of PEP was measured by one item “Have you ever used oral PEP?” with a binary response of “No” and “Yes”. Only those who had ever heard of PEP would answer this question.

2.5.3. Potential confounders

The potential confounders in this study were the demographic information of the participants, including age, ethnicity, marriage status, work status, education level, income, sexual orientation, whether had sex with male in the last 6 months, and whether had sex with female in the last 6 months.

2.5.4. Acceptance of the intervention

Participants’ acceptance of the intervention was measured by the Acceptability of Intervention Measure (AIM) developed by Weiner in 2017 (Weiner et al., 2017). This measure contains 4 items, and each item was measured on a 5-point Likert scale from 1 (completely disagree) to 5 (completely agree). Higher total scores indicate greater acceptability of the invention form and materials. The Cronbach’s α of the measure in this study was 0.953 at post-intervention assessment.

2.6. Statistical analysis

Descriptive analyses were employed to describe the data. Categorical variables were described by frequency and percentage, and continuous variables following a normal distribution were described by mean and standard deviation. The χ² test and Fisher’s exact test were utilized to examine the distribution of demographic characteristics between the control and intervention groups, both pre- and post-intervention. Statistically significant variables were identified as potential confounders and were adjusted in subsequent analysis. To compare the baseline outcomes between the control and intervention groups, the χ² test, Fisher’s exact test, and two-sample independent t-test were used, with statistics and p-values adjusted for confounders. Multiple logistic regression and linear regression analyses were used to examine the association between the IMB model and outcomes using the baseline data. The applicability of the IMB model to the outcomes was evaluated using structural equation modeling (SEM) based on the IMB theory, considering various model fit indices such as χ², degrees of freedom (df), root mean square error of approximation (RMSEA), comparative fit index (CFI), Tucker-Lewis index (TLI), and Standardized Root Mean Squared Residual (SRMR). The criteria for good model fit are: non-significant χ² test, RMSEA < 0.08, CFI > 0.9, TLI > 0.9, and SRMR < 0.08. The χ² test, Fisher’s exact test, and two-sample independent t-test, adjusted for confounders, were applied in both the intention-to-treat (ITT) and as-treated (AT) analyses to explore the effectiveness of the intervention using pre- and post-intervention data. The significance level of this study was set at α = 0.05. All statistical analyses were conducted with SAS 9.4 and Mplus 8.3.

3. Results

3.1. Baseline characteristics of the participants

In the baseline of this study, participants ethnicity (χ²=43.200, P < 0.001), work status (χ²=12.028, P < 0.001), education level (χ²=3.867, P = 0.049), and whether having sex with a male in the last 6 months (χ²=4.707, P = 0.030) were potential confounders between control and intervention groups, see Table 1.

Table 1.

Baseline demographic characteristics between control and intervention groups.

	Control (N = 391)	Intervention (n = 419)	χ2	P
Age			0.056	0.813
< =30	216 (55.24%)	228 (54.42%)
>30	175 (44.76%)	191 (45.58%)
Ethnicity			43.200	<0.001
Han	371 (94.88%)	332 (79.24%)
Other	20 (5.12%)	87 (20.76%)
Marriage			1.999	0.157
Not married	305 (78.01%)	309 (73.75%)
married	86 (21.99%)	110 (26.25%)
Work status			12.028	<0.001
No job or part‐time job	146 (37.34%)	109 (26.01%)
Full‐time job	245 (62.66%)	310 (73.99%)
Education level			3.867	0.049
Below bachelor	182 (46.55%)	224 (53.46%)
Bachelor or above	209 (53.45%)	195 (46.54%)
Income (yuan/month)			2.660	0.103
<7000	272 (69.57%)	313 (74.70%)
> =7000	119 (30.43%)	106 (25.30%)
Sexual orientation			<0.001	0.436
Homosexual	290 (74.17%)	298 (71.12%)
bisexual	83 (21.23%)	105 (25.06%)
Heterosexual	2 (0.51%)	0 (0.00%)
Uncertain	14 (3.58%)	15 (3.58%)
Other	2 (0.51%)	1 (0.24%)
Had sex with male in the last 6 months			1.760	0.185
No	122 (31.20%)	113 (26.97%)
Yes	269 (68.80%)	306 (73.03%)
Had sex with female in the last 6 months			4.707	0.030
No	352 (90.03%)	356 (84.96%)
Yes	39 (9.97%)	63 (15.04%)

After adjusting for potential confounders, the comparison of baseline outcomes between the control and intervention groups is shown in Table 2. The information level of the intervention group was statistically higher than that of the control group (14.00 ± 2.97 vs 13.63 ± 3.33, t = 2.202, P = 0.028). The condom use when having sex with male in the last 6 months in the intervention group was also higher than that of the control group (61.11% vs 52.79%, χ²=4.917, P = 0.027).

Table 2.

Comparison of baseline outcomes between control and intervention groups.

	Control	Intervention	χ 2 /t*	P*
Main outcomes
Awareness of PrEP (N = 810; Nc = 391, Ni = 419)			0.009	0.925
No	137 (35.04%)	151 (36.04%)
Yes	254 (64.96%)	268 (63.96%)
Awareness of PEP (N = 810; Nc = 391, Ni = 419)			0.100	0.752
No	136 (34.78%)	162 (38.66%)
Yes	255 (65.22%)	257 (61.34%)
HIV testing in the last 6 months (N = 810; Nc = 391, Ni = 419)			0.078	0.78
No	200 (51.15%)	213 (50.84%)
Yes	191 (48.85%)	206 (49.16%)
Substance use in the last 6 months (N = 810; Nc = 391, Ni = 419)			0.221	0.638
No	248 (63.43%)	271 (64.68%)
Yes	143 (36.57%)	148 (35.32%)
Information	13.63 ± 3.33	14.00 ± 2.97	2.202	0.028
Motivation	70.36 ± 8.68	70.85 ± 8.27	1.031	0.303
Behavioral skills	43.21 ± 6.00	43.52 ± 5.56	0.587	0.558
Secondary outcomes
Condom use when having sex with male in the last 6 months (N = 575; Nc = 269, Ni = 306)			4.917	0.027
No	127 (47.21%)	119 (38.89%)
Yes	142 (52.79%)	187 (61.11%)
Had commercial sex with male in the last 6 months (N = 575; Nc = 269, Ni = 306)			2.774	0.096
No	247 (91.82%)	291 (95.10%)
Yes	22 (8.18%)	15 (4.90%)
Ever used oral PrEP (N = 522; Nc = 254, Ni = 268)			0.414	0.784
No	242 (95.28%)	253 (94.40%)
Yes	12 (4.72%)	15 (5.60%)
Ever used oral PEP (N = 512; Nc = 255, Ni = 257)			1.704	0.192
No	233 (91.37%)	226 (87.94%)
Yes	22 (8.63%)	31 (12.06%)
Willingness to use oral PrEP (N = 575; Nc = 379, Ni = 404)	3.85 ± 1.17	3.94 ± 1.15	1.341	0.18
Willingness to use long-acting injectable PrEP (N = 810; Nc = 391, Ni = 419)	3.83 ± 1.19	3.83 ± 1.19	0.089	0.929

Note: *Adjusted for ethnicity, work status, education level, and had sex with female in the last 6 months; N: Total sample size, Nc: Sample size of control group, Ni: Sample size of intervention group.

3.2. The association and applicability between the IMB model and outcomes

In the baseline data of this study, the results of multiple logistic and linear regressions showed that the IMB model dimensions of information, motivation, and behavioral skills were associated with most outcomes, as shown in Supplementary Table 3-4. The assumptions of linear, independence, normality, homoscedasticity, and non-multicollinearity have all been satisfied for the corresponding regression models. To evaluate the applicability of the IMB model to the outcomes, we specified information and motivation as independent variables, behavioral skills as a mediator, and the outcomes as dependent variables to build structural equation models and examine the model fit using the baseline data. Despite the majority of χ² values being significant, the RMSEAs ranged between 0.040 and 0.074, the CFIs were between 0.945 and 0.978, the TLIs were between 0.829 and 0.899, and the SRMRs were between 0.030 and 0.033. The results indicated that the overall model fit for all outcomes was satisfactory (Doll et al., 1994). The detailed model fits are shown in Supplementary Table 5.

3.3. The intervention effects of the IMB-based online materials on outcomes

In the post-intervention data, we initially identified ethnicity as a potential confounder in both the ITT (χ²=21.233, P < 0.001) and AT (χ²=16.121, P < 0.001) analyses between the control and intervention groups. This variable was subsequently adjusted for in further analyses, as shown in Table 3.

Table 3.

Post-intervention demographic characteristics between control and intervention groups.

	Control (N = 287)	Intervention (n = 333)	χ2/t*	P
	(N = 436) #	(n = 184) #
Age			0.33	0.566
< =30	195 (67.94%)	219 (65.77%)	0.520#	0.471#
	295 (67.66%) #	119 (64.67%) #
>30	92 (32.06%)	114 (34.23%)
	141 (32.34%) #	65 (35.33%) #
Ethnicity			21.233	<0.001
Han	272 (94.77%)	276 (82.88%)	16.121#	<0.001#
	400 (91.74%) #	148 (80.43%) #
Other	15 (5.23%)	57 (17.12%)
	36 (8.26%) #	36 (19.57%) #
Marriage			0.708	0.400
Not married	245 (85.37%)	276 (82.88%)	0.395#	0.530#
	369 (84.63%) #	152 (82.61%) #
married	42 (14.63%)	57 (17.12%)
	67 (15.37%) #	32 (17.39%) #
Work status			0.996	0.318
No job or part‐time job	119 (41.46%)	125 (37.54%)	0.065#	0.799#
	173 (39.68%) #	71 (38.59%) #
Full‐time job	168 (58.54%)	208 (62.46%)
	263 (60.32%) #	113 (61.41%) #
Education level			0.646	0.422
Below bachelor	129 (44.95%)	139 (41.74%)	1.788#	0.181#
	196 (44.95%) #	72 (39.13%) #
Bachelor or above	158 (55.05%)	194 (58.26%)
	240 (55.05%) #	112 (60.87%) #
Income (yuan/month)			0.641	0.423
<7000	205 (71.43%)	228 (68.47%)	0.450#	0.502#
	308 (70.64%) #	125 (67.93%) #
> =7000	82 (28.57%)	105 (31.53%)
	128 (29.36%) #	59 (32.07%) #
Sexual orientation			<0.001	0.172
Homosexual	212 (73.87%)	222 (66.67%)	0.001#	0.882#
	308 (70.64%) #	126 (68.48%) #
bisexual	60 (20.91%)	90 (27.03%)
	104 (23.85%) #	46 (25.00%) #
Heterosexual	2 (0.70%)	2 (0.60%)
	2 (0.46%) #	2 (1.09%) #
Uncertain	10 (3.48%)	18 (5.41%)
	19 (4.36%) #	9 (4.89%) #
Other	3 (1.05%)	1 (0.30%)
	3 (0.69%) #	1 (0.54%) #
Had sex with male in the last 6 months			0.037	0.847
No	85 (29.62%)		101 (30.33%)	2.475#	0.116#
	139 (31.88%) #	47 (25.54%) #
Yes	202 (70.38%)	232 (69.67%)
	297 (68.12%) #	137 (74.46%) #
Had sex with female in the last 6 months			0.002	0.966
No	264 (91.99%)	306 (91.89%)	0.141#	0.708#
	402 (92.20%) #	168 (91.30%) #
Yes	23 (8.01%)	27 (8.11%)
	34 (7.80%) #	16 (8.70%) #

Note: #Results of the as-treated analysis.

For the main outcomes and in the ITT analysis, the HIV testing rate in the last 6 months was significantly higher in the intervention group compared to the control group (62.76% vs 54.36%, P = 0.034). In the AT analysis, not only was the HIV testing rate in the last six months higher in the intervention group (69.57% vs 54.36%, P = 0.001), but also the awareness rate of PrEP (81.52% vs 71.79%, P = 0.005) and information level (14.42 ± 2.68 vs 13.62 ± 2.99, P = 0.005) compared with the control group. The results are presented. For the secondary outcomes, we only employed ITT analysis to examine the effectiveness of the intervention. The results indicated that the willingness to use oral PrEP was significantly higher in the intervention group compared with the control group (4.08 ± 1.05 vs 3.87 ± 1.20, P = 0.005). The findings are detailed in Table 4.

Table 4.

Effects of IMB-based intervention on outcomes.

	Control (N = 287)	Intervention (n = 333)	χ2/t*	P*	Power
	(N = 436)#	(n = 184)#
Main outcome
Awareness of PrEP			0.558	0.455	0.126
No	77 (26.83%)	80 (24.02%)	6.278#	0.005#	0.775#
	123 (28.21%) #	34 (18.48%) #
Yes	210 (73.17%)	253 (75.98%)
	313 (71.79%) #	150 (81.52%) #
Awareness of PEP			0.019	0.890	0.051
No	81 (28.22%)	93 (27.93%)	1.269#	0.260#	0.266#
	129 (29.59%) #	45 (24.46%) #
Yes	206 (71.78%)	240 (72.07%)
	307 (70.41%) #	139 (75.54%) #
HIV testing in the last 6 months			4.500	0.034	0.565
No	131 (45.64%)	124 (37.24%)	10.483#	0.001#	0.956#
	199 (45.64%) #	56 (30.43%) #
Yes	156 (54.36%)	209 (62.76%)
	237 (54.36%) #	128 (69.57%) #
Chemsex in the last 6 months			1.232	0.267	0.140
No	203 (70.03%)	246 (73.87%)	0.875#	0.350#	0.117#
	312 (71.56%) #	137 (74.46%) #
Yes	84 (29.27%)	87 (26.13%)
	124 (28.44%) #	47 (25.54%) #
Information	13.71 ± 2.91	13.98 ± 2.93	0.768	0.443
	13.62 ± 2.99 #	14.42 ± 2.68 #	2.808#	0.005#
Motivation	71.41 ± 9.05	71.69 ± 9.19	0.274	0.784
	71.38 ± 9.04 #	71.99 ± 9.32 #	0.676#	0.499#
Behavioral skills	43.23 ± 6.36	43.46 ± 6.82	0.275	0.784
	43.16 ± 6.45 #	43.83 ± 6.96 #	1.021#	0.307#
secondary outcomes
Always using condom when having sex with male in the last 6 months (N = 434; Nc = 202, Ni = 232)			2.317	0.128
No	100 (49.50%)	98 (42.24%)
Yes	102 (50.50%)	134 (57.76%)
Ever used oral PEP (N = 446; Nc = 206, Ni = 240)			3.164	0.075
No	180 (87.38%)	198 (82.50%)
Yes	26 (12.62%)	42 (17.50%)
Had commercial sex with male in the last 6 months (N = 434; Nc = 202, Ni = 232)			2.066	0.151
No	195 (96.53%)	215 (92.67%)
Yes	7 (3.47%)	17 (7.33%)
Ever used oral PrEP (N = 463; Nc = 210, Ni = 253)			1.470	0.225
No	187 (89.05%)	217 (85.77%)
Yes	23 (10.95%)	36 (14.23%)
Willingness to use oral PrEP (N = 561; Nc = 264, Ni = 297)	3.87 ± 1.20	4.08 ± 1.05	2.790	0.005
Willingness to use long-acting injectable PrEP (N = 620; Nc = 287, Ni = 333)	3.87 ± 1.20	3.79 ± 1.25	1.223	0.222

Note: *Adjusted for ethnicity; #Results of the as-treated analysis; N: Total sample size, Nc: Sample size of control group, Ni: Sample size of intervention group.

We conducted post-hoc analyses to determine the statistical power of the main HIV-related outcomes. A parallel two-group design was used to test whether the intervention group proportion (P1) is different from the control group proportion (P2) (H0: P1 - P2 = 0 versus H1: P1 - P2 ≠ 0). The comparison was made using a two-sided, two-sample Z-Test with un-pooled variance, with a Type I error rate (α) of 0.05. For the ITT analyses, the powers ranged from 0.051 to 0.565. For the AT analyses, the powers ranged from 0.117 to 0.956. The detailed powers are summarized in Table 4. The powers were computed using PASS 2023, version 23.0.2.

3.4. The attitude to the intervention materials

Among the participants who engaged with the intervention materials, approximately 70% found the materials useful, 65% found them attractive, 68% preferred these materials, and 66% welcomed them. These details are summarized in Supplementary Table 6.

4. Discussion

While the application of digital technology in the field of HIV prevention and control is becoming increasingly widespread, comprehensive evaluations of the effectiveness and acceptance of these digital interventions are still insufficient. Based on the IMB model, we have developed a series of digital HIV prevention intervention materials specifically tailored to the characteristics of MSM. Subsequently, we conducted an exploratory real-world study to evaluate the effectiveness of the online intervention in promoting healthy behaviors in HIV prevention and alleviating high-risk behaviors among MSM in China. Findings from the baseline data suggest that the IMB model has satisfactory applicability to the outcomes. The intervention demonstrated potential in increasing the HIV testing rate over 6 months, improving willingness to use oral PrEP, and possibly augmenting the awareness of PrEP and the level of information among MSM. A majority of the MSM who underwent the intervention expressed high levels of perceived usefulness, attractiveness, preference, and acceptance of the intervention materials. While we did not evaluate the effectiveness of each IMB component separately, the intervention’s design intentionally combined Information, Motivation, and Skills-building strategies to reflect real-world implementation.

HIV testing is a critical component in the prevention and control of HIV, especially among MSM (Palich et al., 2024). A study conducted in 2016 utilizing a dynamic transmission model of HIV among adults in the United States indicated that improvements in HIV testing could yield modest epidemiologic benefits and could also lead to a 21% reduction in new HIV infections at an associated cost of 65,700 dollars per Quality-Adjusted Life-Year (QALY) gained (Shah et al., 2016). Furthermore, an economic assessment of HIV testing for MSM across six European cities also suggests that community-based voluntary counseling and HIV testing could be a cost-effective strategy for alleviating HIV infection in this population (Perelman et al., 2017). In this study, it was observed that MSM in the intervention group exhibited an HIV testing rate that was 8.40% higher over the past six months compared to the control group, and the rate increased to 15.21% in the AT analysis, which is similar to the results of a randomized controlled trial in India in 2016 (Patel et al., 2020). This finding initially revealed the effectiveness of the online intervention in improving the HIV testing rate based on the IMB model. Currently, most interventions targeting HIV testing among MSM concentrate on proximal strategies, such as the distribution of complimentary HIV self-testing kits and enhancements in the accessibility of health services, while distal interventions such as the promotion of knowledge and motivation levels were rarely recorded (Rodger et al., 2022; Rocha et al., 2023). In this case, the IMB model framework could offer a systematic cascade for the intervention of HIV testing (Stannah et al., 2023). Relevant departments may consider establishing structural online interventions to augment knowledge, foster positive attitudes and subjective norms, and improve behavioral skills associated with HIV testing among MSM. It is anticipated that the HIV testing rate within this population will increase, thereby reducing HIV infection rates and promoting HIV prevention efforts.

In addition to HIV testing, oral PrEP is also a critical tool in the prevention of HIV among MSM. A multinational Phase III clinical trial conducted in 2010 demonstrated that consistent utilization of oral PrEP could result in a 73% decrease in HIV transmission among MSM (Grant et al., 2010). Moreover, two European trials conducted in 2015 and 2016 in the UK and France proved a high protective rate of 86% for oral PrEP, thereby catalyzing PrEP-related initiatives in Europe (McCormack et al., 2016; Molina et al., 2015). In 2022, a non-randomized controlled trial in China further affirmed the effectiveness of oral PrEP in reducing the incidence of HIV among MSM (Wang et al., 2022). These findings emphasized the importance of PrEP as a preventive measure against HIV transmission among MSM. Furthermore, the utilization of PrEP also plays a significant role in enhancing sexual pleasure and promoting a positive sexual attitude, which is an aspect that has been substantially overlooked in the health promotion endeavors targeted at MSM in China. Empirical studies have shown that PrEP use could both strengthen intimate relationships and reduce anxiety and fear associated with sexual activity among MSM, thereby enhancing their sexual confidence and pleasure (Curley et al., 2022). These findings not only emphasize the importance of PrEP as a preventive measure against HIV transmission but also highlight its role in promoting sexual health (Ford et al., 2021). Therefore, designing interventions to further increase PrEP uptake and adherence among this population remains an important priority.

In this study, we found that the IMB-based online intervention has the potential to improve the awareness rate of PrEP among MSM. Furthermore, it effectively facilitates the willingness to use oral PrEP among this population who have heard of PrEP. Notably, awareness of PrEP saw an increase of 9.73% in the intervention group compared with the control group in the AT analysis, while the effectiveness of this intervention in real-world settings, as suggested by the ITT analysis, remains to be determined. Despite these promising results, the promotion of PrEP continues to present challenges globally. Research has indicated that even with increasing awareness and normalization of PrEP, ensuring easy and equitable access remains a significant hurdle (Tieosapjaroen et al., 2023). The global uptake of PrEP has been insufficient and concentrated in a limited number of developed countries (Bavinton & Grulich, 2021). In China, despite the demonstrated efficacy of PrEP, its widespread adoption among key populations like MSM remains suboptimal. This highlights a substantial gap between clinical efficacy and real-world implementation (Huang et al., 2025). Several critical issues are likely to persist or emerge as main concerns for PrEP use in China, such as the need to sustain adherence and retention over time, address the distinct needs of specific subgroups, and effectively integrate PrEP delivery within existing health systems (Hong et al., 2022; Smith et al., 2024; Irungu et al., 2021). A cross-sectional study conducted in Beijing, China emphasized the relatively low awareness of PrEP among MSM and highlighted the PrEP inequities faced by MSM experiencing homosexuality stigma (Sun et al., 2023). This is not conducive to promoting the use of PrEP among MSM, with studies indicating that the current usage rate of PrEP among MSM in China is only between 1% and 2.5%, which is much lower than the global level (11.23%) (Zhang et al., 2019; Xu et al., 2020; Huang et al., 2023). A trajectory model targeting Chinese MSM also revealed that mental disorders such as depression and anxiety could negatively affect PrEP adherence among this population (Chen et al., 2024). These findings highlight the requirement for relevant departments to establish more ambitious global objectives for the uptake of PrEP and the development of health systems that facilitate its use. To sum up, addressing issues such as stigma, mental disorders, limited access to care, and financial barriers is not only crucial for improving PrEP awareness and willingness to use among MSM, but it can also improve adherence to PrEP usage. Moreover, recognizing the role of PrEP in enhancing sexual pleasure and sexual health could further increase its long-term adherence within this population to better protect them from HIV infection (Calabrese & Underhill, 2015). Therefore, future research could explore the feasibility of incorporating mental health interventions aimed at alleviating stigma, depression, and anxiety into IMB-based online interventions, and investigate the role of sexual pleasure in enhancing PrEP adherence. These approaches may serve as beneficial strategies to improve MSM’s awareness, willingness, and adherence to use PrEP. On the other hand, Accessibility to HIV clinics in China remains a pivotal determinant in the uptake of PrEP. While specialized sexually transmitted infection STI clinics have played a significant role in facilitating HIV testing, it is imperative to broaden their remit to include comprehensive PrEP services. Historically, STI clinics in China have primarily focused on HIV and syphilis testing. Although syphilis testing is widely accepted, HIV testing has often been hindered by societal stigma, which can contribute to patient attrition from PrEP programs (Tucker et al., 2012). Data from a Seattle clinic indicated that 17% of enrolled patients did not fill their PrEP prescription, and 40% discontinued PrEP within 12 months (Dombrowski et al., 2018). To address these challenges, healthcare providers should consider strengthening human resources to improve PrEP delivery (e.g., reducing pill burden, managing side effects), implementing youth-friendly services, increasing awareness campaigns, engaging key populations, and minimizing financial barriers to PrEP initiation and maintenance (Skovdal et al., 2022). In addition, at-home HIV testing represents a promising solution to several access barriers in China, particularly those associated with convenience, privacy, and stigma. By enabling individuals to bypass geographical obstacles and reduce concerns about potential judgment during clinic visits, at-home testing is especially relevant in settings where HIV-related stigma persists (Ballard et al., 2021; Tucker et al., 2011). Evidence from South Africa indicates that preferences for HIV testing modalities among PrEP users are different, with self-testing preferred by some due to its increased convenience and privacy (Martin et al., 2025). Therefore, integrating self-testing into the PrEP care continuum from initial screening to ongoing monitoring may offer a feasible strategy to enhance engagement. Furthermore, combining at-home testing with digital platforms, such as mobile applications and telemedicine, may further expand the reach and support available to PrEP users in China (Martin et al., 2025; Wang et al., 2025). Indeed, synergistic integration of digital and physical approaches is already underway in HIV PrEP promotion initiatives across the country (Wang et al., 2025). In summary, while PrEP offers substantial promise for HIV prevention in China, its translation from demonstrated efficacy to equitable, widespread access requires a multifaceted strategy that addresses financial, systemic, and social barriers. Leveraging innovations such as at-home testing and diversifying service delivery models will be crucial to expanding PrEP accessibility and maximizing public health impact nationally.

In this study, we also evaluated participants’ attitudes toward the online IMB-based intervention materials. The results suggested that most of the participants perceived the form and content of these materials as useful and attractive. Furthermore, these materials were generally preferred and well-accepted by the participants, and this positive response might be attributed to the development of targeted content and its digital delivery method. The positive attitudes indicated the potential for a broader application of online IMB-based interventions on healthy behaviors in HIV prevention among various populations, not solely limited to MSM. However, in this study, only about 55% of the participants assigned to the intervention group engaged with the intervention materials at least once, and the click rates of the materials were relatively low, with the highest rate being 23.86%. Consequently, it is imperative to consider not only the specific needs and characteristics of the target population but also the strategies to encourage the target population to engage with the materials during their design phase. Further research could explore how these interventions can be customized to optimize their effectiveness for different populations. On the other hand, the focus of this study on holistic sexual health, which integrates information, motivation, and behavioral skills, is consistent with emerging evidence indicating that addressing sexual pleasure alongside health and rights can strengthen HIV prevention efforts (Ford et al., 2021). The IMB model emphasizes health behaviors such as partner communication about safer sex and, in doing so, implicitly acknowledges that pleasure and intimacy are core motivations for sexual activity. By normalizing discussions of pleasure within HIV prevention messages, our intervention may have enhanced participant engagement, as reflected by the high levels of material acceptance. This positive reception likely stems from the materials’ relevance to real-life sexual experiences, including the desire for connection and satisfaction. Moreover, these findings are in line with research demonstrating that pleasure-inclusive interventions improve adherence to preventive behaviors, such as PrEP use, by framing such practices as facilitators of positive sexual experiences rather than as restrictive measures (Ford et al., 2021).

To be noted, although we utilized unmatched pre- and post-intervention data to construct a pseudo-panel dataset, several methodological measures were implemented to ensure that the observed changes could be attributed to the intervention rather than external factors. First, we have carefully selected the study cities for the intervention and control groups. The cities were chosen based on multiple factors such as the number of basic users and active status on the Blued platform, the prevalence of HIV infection, the city’s cost of living, and the city population size, which could minimize the influence of external factors related to the cities such as economic and demographic differences between the samples from the 2 waves of the data. Second, potential confounders were identified and adjusted in the statistical analysis, thereby isolating the intervention’s effect from the potential impact of confounding variables. Finally, the intervention materials were developed based on a well-established theoretical framework, the IMB model. The model can provide a systematic approach to promoting behavior change, and the materials were designed to target specific aspects of HIV-related healthy behaviors. The consistency between the theoretical basis of the intervention and the observed changes in the associated outcomes further provides strong evidence for the effectiveness of the intervention.

This study examined the effect of the IMB-based online intervention on the adoption of healthy behaviors among MSM. However, it has certain limitations. First, this is an exploratory real-world study without rigorous randomization and sampling, which may cause selection and confounder bias, indicating that the participants in our study are not fully representative of the entire MSM population. This might have led to an over- or underestimation of the intervention’s effectiveness. Further studies with randomized sampling approach, such as using a multi-stage random sampling method across different regions in China, could be employed to improve the representativeness of the sample. Second, the study design did not incorporate a matching process for participants pre- and post-intervention during data collection, which may lower the statistical power of the results. This may have made it more difficult to detect smaller but still meaningful effects of the intervention, which has also been reflected in the power values of the HIV-related outcomes. Future studies could consider using unique identifiers (if privacy policies allow) or more sophisticated statistical methods for handling unmatched data to better detect the intervention effect. The concerns about the comparability of the groups over time caused by the pseudo-panel dataset have been discussed before. Third, this exploratory study focused on the intervention of overall healthy behaviors and may ignore the specific characteristics of certain behaviors. Future research could consider developing behavior-specific intervention materials based on in-depth qualitative research on each high-risk behavior among MSM. Fourth, this study found that the IMB-based online intervention showed no obvious effectiveness in enhancing awareness of PEP, reducing chemsex, decreasing the rate of commercial sex, or increasing the usage of oral PrEP and PEP, and the willingness to use long-acting injectable PrEP among MSM. This lack of observable effectiveness may be attributed to the exploratory nature of the real-world design, which cannot ensure complete adherence to the delivered materials, and the extended interval between the delivery of the materials and the follow-up questionnaire. Future studies could incorporate strategies to increase participant engagement, such as gamification elements or personalized nudges in the digital intervention. Fifth, the selection of cities based on various factors such as the number of Blued users, and the rigorous eligibility criteria may limit the generalizability of our results. Future research could expand the inclusion criteria to include a more diverse range of MSM, such as those with a positive HIV diagnosis, to improve the generalizability of the results. Sixth, a critical limitation is the inability to perform dose-response analyses or link intervention engagement with individual outcomes, as the digital platform lacked functionality to track participant-level interaction data. Although aggregate engagement metrics for each material are reported, the absence of unique participant-engagement linkages precluded stratified analyses, representing an important avenue for future research with platforms enabling granular data tracking. Seventh, the absence of rural-urban residential data impeded the analyses of digital divide implications. Furthermore, the paucity of older users (n = 7 at baseline, n = 2 post-intervention) likely reflects engagement barriers associated with the digital divide (e.g., technological access, digital literacy), precluding meaningful subgroup analysis. Future research should prioritize inclusive sampling strategies to address these gaps and explicitly investigate digital equity dynamics. Finally, we did not implement specific measures to control for the potential influence of internet bots, which could potentially distort our data and affect the validity of our results. Future research in this area could consider implementing more robust methods to identify and exclude bot-generated responses, such as using CAPTCHA verification at the start of the questionnaire and implementing machine-learning-based algorithms to identify patterns typical of bot behavior.

5. Conclusion

This exploratory real-world study highlighted the effectiveness of an IMB-based digital intervention in promoting healthy behaviors among MSM, especially in increasing HIV testing, awareness and willingness to use PrEP, which could potentially decrease high-risk behaviors and subsequent risk of HIV infection. The IMB model, to some extent, effectively addresses the challenges faced by MSM in China, such as knowledge gaps, stigma, and the need for safer sex practices by providing relevant information, boosting motivation, and building behavioral skills to facilitate behavior change in a comfortable online environment. Relative departments may consider implementing this online behavioral intervention under the framework of the IMB model across a broader MSM community, thus contributing to the control of the HIV/AIDS epidemic.

Funding Statement

This study is supported by the CAMS Innovation Fund for Medical Sciences (ClFMS) (Grant No.2023-I2M-2-001), the Non-profit Central Research Institute Fund of Chinese Academy of Medical Sciences (Grant No. 2022-ZHCH330-01), and the Sanming Project of Medicine in Shenzhen (Grant No. SZSM202211032).

Author contributions

ZD, XS, and ST conceived and designed the study. ZD, XinL, XiaoL, and SJ conducted the literature review. ZD, FY, and GM collected data and performed the statistical analysis. ZD and XinL interpreted the findings. ZD, XinL, and XiaoL drafted the article. JF prepared the materials and revised the manuscript. XS and ST critically revised the manuscript for important intellectual content.

Disclosure statement

No potential conflict of interest was reported by the author(s).

Ethics approval and consent to participate

Ethical approval for this study was obtained from the Ethics Committee of Danlan Beijing Media Limited on May 20, 2020, and March 10, 2023 (Number: DLIRB202005-01, DL202303-3), and the Ethics Committee of Chinese Academy of Medical Science on May 12, 2023 (Number: CAMS&PUMC-IEC-2023-021).

Data availability statement

Data can be provided upon reasonable request.