Background
Antiretroviral therapy (ART) has greatly reduced HIV-related mortality since 1996. While human immunodeficiency virus (HIV) infection/acquired immune deficiency syndrome (AIDS) remains a leading cause of death and a disease burden in many countries and regions [
1]. Men who have sex with men (MSM) continue to have disproportionately high burdens of HIV infection in countries of low-, middle-, and high-income [
2]. Notification of HIV cases among MSM in China rose from 2.5% in 2006 to 25.8% in 2014 [
3]. The results of a multisite cross-sectional study reported high HIV prevalence of 9.9% among MSM in China [
4]. Unprotected anal intercourse (UAI) is a common high-risk behavior and reason for HIV infection among MSM [
5]. Evidence from a recent meta-analysis showed that the proportion of UAI with any male partner among MSM in China was 53% (95%
confidence interval [CI]: 51–56%) [
6]. One serial cross-sectional study presented that the proportion of UAI was 58.4% during the last 6 months among MSM in Guangzhou [
7]. Correct and consistent condom use, is one of the most effective strategies for preventing the spread of HIV among general and high-risk populations [
8]. Strengthening this behavior among MSM in China is thus an important issue for research.
Evidence shows that new public health and/or health-promotion interventions based on social and behavioral science theories are more effective than those lacking a theoretical framework [
9], this is because these interventions are tailored towards addressing the identified predictors of the health behavior of interest. Specifically, the Information-Motivation-Behavioral Skills (IMB) model has been found to significantly predict condom use among MSM [
10], female sex workers [
11], people attending sexually transmitted infection (STI) clinics [
12], and students [
13]. Consequently, interventions designed along IMB constructs have been found to increase condom use among MSM [
14], HIV-infected patients [
15], and students [
16]. The IMB model proposes that information, motivation, and behavioral skills are fundamental determinants of HIV-preventive behaviors, e.g. condom use [
17]. According to the model, possession of adequate information coupled with a strong motivation to act on the information propels the desired behavioral skills, which in turn initiate and sustain condom use. The behavioral skills that directly influence condom use also partially mediate the associations between information/motivation and condom use [
17‐
19].
Although the components of the IMB model influence one another, studying them in isolation from other social influences of behavior yields less predictive power [
20]. The IMB model in its non-extended form relies only on psychological or individual-level influences, whereas social factors have a significant influence on behaviors [
20,
21]. Meta-analyses of behavioral intervention show that individual-level theories cannot explain the heterogeneity among study outcomes [
22]. Thus, researchers have suggested extending the model by incorporating other social factors that influence preventive behaviors [
20,
21]. Modified IMB models with extended constructs have been found to explain more variance in HIV-preventive behavior than the original non-extended model [
20,
21,
23,
24]. However, existing literature has insufficiently integrated these social factors and explored their interactions [
25]. Recent efforts have been made to expand from individual-level factors to multilevel factors at different ecological levels. The multiple domain model (MDM) suggests that factors outside of the individual are modeled as factors shaping one’s behavior; these factors including structural factors, personality, and social environment situational/contextual variables [
26]. The network-individual-resource model (NIRM) proposes that individual behavior changes for HIV prevention can be supported and sustained by the tangible and mental resources possessed by individuals and networks and that the outcomes may be resilient over time and disseminated more broadly [
27]. The dynamic social systems model (DSSM) focuses on the structures that people face through interactions at the micro-, meso-, and macro-levels [
28]. The Transmission Reduction Intervention Project (TRIP) suggests that HIV prevention can be greatly improved by using social science as an integrative tool in transdisciplinary research and practice [
29]. Another review recommended a multilevel approach to HIV-related behavior changes, which suggested that the HIV risk and AIDS care involve complex behaviors beyond individual-level factors that are influenced by multilevel factors [
30]. These multilevel factors include individual-level factors, interpersonal/network-level factors, and structural-level factors. Among the individual-level factors, income, education and depression were found to be associated with condom use [
23,
31,
32], and these factors were integrated into a modified IMB model [
20,
23]. The interpersonal/network-level factors such as child sexual abuse (CSA) and intimate partner violence (IPV) tended to co-occur and increased the risk of UAI together with depression [
32‐
34]. With use of Internet and mobile applications, an increasing number of MSM seek sexual partners online. Sexual partner seeking behavior, which is another interpersonal/network-level factor, may also impact condom use behavior [
35]. Moreover, structural-level factors, such as access to HIV prevention and care services, may also have a positive effect on condom use [
36]. However, the original IMB model has no role for interpersonal/network-level or structural-level factors. Thus, an extension of the IMB model with interpersonal/network-level, structural-level, and more individual-level factors may be meaningful for improvement of the predictive power for the condom use behavior.
However, less is known about the application of modified IMB models for predicting condom use among MSM in China. Based on the high proportion of UAI [
6,
7], increasing HIV prevalence [
3,
4], and lack of evidence for the influence of the modified IMB model among MSM in China, we conducted this cross-sectional study to integrate the multilevel factors into the conventional IMB model and built a modified IMB model for predicting the condom use among MSM in China. The aim of this study is to provide a theoretical framework for safe sex behavioral intervention.
Discussion
The current study applied the conventional IMB model and developed a modified IMB model to predict HIV preventive behavior among Chinese MSM. The modified IMB model fitted the data better than the conventional model. The modified IMB model demonstrated that the multilevel factors, such as education, income, depression, CSA, IPV, sexual partner seeking behavior, and access to HIV prevention services, impacted condom use in addition to information, motivation, and behavioral skills.
Correct and consistent condom use is one of the most effective strategies for preventing the spread of HIV among MSM [
8]. The proportion of consistent condom use among MSM was 52.05%, which was slightly higher than that (47%) reported in one previous meta-analysis in China [
6]. When compared to the result of a serial cross-sectional study conducted among MSM in Guangzhou from 2008 to 2013 [
7], the proportion of consistent condom use among MSM increased from 41.6 to 52.05% in 2017, which revealed that the basic HIV control and prevention measures were making progress [
44]. However, condom promotion remains a challenge due to the large MSM population and the increasing HIV prevalence among MSM [
44,
45]. Moreover, one previous study presented that condom use showed no significant protection when comparing occasional use to no use among MSM having any anal sex with an HIV-positive male partner [
46]. Therefore, it is important to develop an HIV prevention program for consistent condom use under the guidance of health promotion theoretical frameworks. The IMB model is one theory that effectively predicts the condom use [
10‐
13] and guides condom promotion interventions in various populations, including MSM [
14‐
16].
According to the original IMB model proposed by Fisher [
17,
18], condom use was affected primarily by information, motivation and behavioral skills. Consistent with the original IBM model, behavioral skills were important component in the current study and had a direct effect on condom use. Information and motivation did not appear to contribute to condom use directly as Fisher proposed but instead indirectly mediated behavioral skills. Nevertheless, some other studies also demonstrated that the effect of information and motivation on condom use was not significant [
23,
47].
In the current study, 81.25% of the participants had adequate HIV-related knowledge. Only the questions “Can correct condom use reduce the risk of HIV infection and transmission” and “Should HIV consultation and testing be actively sought after high risk behaviors (needle-sharing, drug abuse, unsafe sex, etc.)” were correctly answered by 98.05 and 96.21% of the participants, respectively, which achieved the target (90%) of having adequate HIV-related knowledge among MSM [
48]. Thus, more efforts should be made to improve the gaps in HIV-related knowledge among MSM. Our results added to existing literature showing that information does not always have a direct effect on sexual risk behavior. This lack of effect could be due to a ceiling effect, since the overall level of HIV-related knowledge was high and did not yield any additional explanatory power [
23]. Moreover, previous studies suggested that adequate HIV-related knowledge appeared to be an insufficient determinant to predict condom use and change risk behavior [
23,
49]. Although information did not have a direct influence on condom use, information indirectly contributed to condom use mediated by behavioral skills, and remained a necessary component of HIV prevention interventions, which was consistent with previous studies [
11,
20].
Motivation also indirectly contributed to condom use mediated by behavioral skills. Individuals with a positive personal attitude towards condom use and higher subjective norms and intentions regarding condom use were more likely to show more positive motivation to use condoms [
12,
13,
20]. It is easy to understand the relationship between a positive attitude and condom use. The subjective norm represented the perceptions of social referents and motivation to comply, which could affect HIV prevention intentions and behaviors [
11,
12,
17,
20]. However, previous studies suggested that motivation had a direct effect on condom use among male street laborers [
20] and STI clinic patients [
12]. Behavioral skills contributed to condom use directly, which revealed that individuals who displayed better skills in preparation, negotiation and practice were more likely to engage in HIV prevention behavior [
20]. The current study placed behavioral skills in the role as a mediating factor for all components of the model, which was consistent with some previous studies [
12,
20] but inconsistent with other studies [
11,
24].
The conventional IMB model focused only on individual-level factors that influenced condom use, and did not fit the data ideally in the current study. According to some multilevel models, the HIV risk and AIDS care behaviors are influenced by multilevel factors at different ecological levels [
26‐
30]. One comprehensive and practical review recommended that multilevel approaches for HIV-related behavior changes could serve as the basis for extending the conventional IMB model with multilevel factors [
30]. Beyond individual-level factors, the modified IMB model integrated interpersonal/network-level and structural-level factors to fit the data ideally. Moreover, the total effect of the modified IMB model (0.927) on condom use was much larger than that of the non-extended IMB model (0.599). Therefore, this modified IMB model with multilevel factors was suggested for use as a theoretical framework to guide the behavior interventions and improve the condom use among MSM in China.
The results of the final modified IMB model showed that more educated individuals had more adequate HIV prevention and transmission information [
50], which contributed to condom use [
32]. However, another study showed that in one personal resource construct, education did not have an impact on information or behavioral skills [
23]. Low-income individuals may be more likely to encounter stressors and/or limit their access to resources to buffer against these stressors [
51]. Individuals with socioeconomic disadvantages utilize their limited cognitive coping resources in dealing with these excess stressors, especially lack of social support, leading to engagement in condomless sexual behaviors [
31]. One modified IMB model suggested that depressed MSM were more likely to have sexual risk behaviors to mitigate distress, which might further compromise motivation for behavior changes [
23]. Another modified IMB model presented that the latent variable “psychosocial factors”, including depression as one of the five constructs, was negatively associated with behavioral skills [
20], which was consistent with the current results.
CSA and IPV had negative effects on condom use, with the former mediated by motivation and the latter by behavioral skills. Individuals with a CSA experience might foster problematic coping styles and relational instability, resulting in greater exposure to risk opportunities and their condom use decisions were more likely to be affected by their partner’s reaction [
52]. They were more likely to think condoms interfered with sexual pleasure and less likely to think condoms were important [
53], which was consistent with the negative association between CSA and motivation in the current study. Previous studies suggested that individuals who had experienced IPV were significantly less likely to report having felt able to negotiate condom use [
54,
55], which was consistent with the negative association between IPV and behavioral skills in the current study.
Due to technological advances, MSM have multiple platforms for online sex seeking, such as gay-specific forums, chat rooms, and dating websites [
56]. Previous studies presented that condom use self-efficacy, which referred to condom acquisition, proper condom use, and negotiation skills, played an important role in condom use behavior among MSM. Higher rates of condom use self-efficacy were associated with lower rates of risky sexual practices [
57‐
59]. The good behavioral skills among the MSM who sought sexual partners online in the current study might be attributed to high condom use self-efficacy among these participants [
53]. However, whether seeking sex partners online or offline increases condom use remains controversial. Some previous studies indicated that partners sought online could increase the risk of UAI [
35,
60], whereas other studies believed that no difference in UAI existed between gay app users and non-app users [
56,
61]. Nevertheless, the fact that most (89.96%) of the participants sought homosexual partners online and that 46.81% of the MSM reported UAI in the current study suggested that websites and gay apps were both risk environments and that Internet-based behavioral interventions were necessary. HIV prevention services provided adequate HIV prevention and transmission knowledge and thus had a positive effect on information that indirectly resulted in condom use. However, a gap existed between the target of more than 90% coverage of HIV prevention services [
48] and the fact that only approximately two-thirds (63.83%) of the MSM received HIV prevention services. Therefore, more efforts should be made to expand the coverage of HIV prevention services among MSM in China.
Despite these strengths, several limitations should be noted when interpreting and generalizing the results. First, causal inference remained ambiguous due to the cross-sectional nature of this study [
12]. Thus, prospective studies are called for to further confirm the effects of the modified IMB model components on HIV-preventive behavior. Second, the convenience sampling approach may limit the generalizability [
23]. Nevertheless, the six VCT clinics and the two popular community-based HIV service centers, which covered 8 of the 11 districts in Guangzhou, are believed to have provided a relatively adequate representative sample. The birthplaces of the participants in Guangzhou, other cities in Guangdong province, and other provinces each accounted for approximately one-third of the participants, which might extend the generalizability of the current study. Third, only Cronbach’s alpha was used to assess the reliability of the scale items. More forms of reliability and validity assessment, such as test-retest reliability, and discriminate validity should be conducted in future studies. Finally, the validity of self-reported data should be taken into consideration in any study of sexual behavior. However, an anonymous electronic questionnaire was administered to remove such barriers to participation in this survey [
23,
62].