Background
Transgender women (TGW) are those whose gender identity (woman) is different from their birth-assigned sex (male) [
1]. Scarce as this population is (0.5–0.9% of the whole population are identified as transgender worldwide [
2‐
4]), their HIV prevalence is remarkable. The overall HIV prevalence (self-reported and laboratory confirmed) for TGW worldwide is 18.8–28% [
1,
5‐
8]; this group has a 49-fold higher odds of infection compared with other reproductive age adults [
6], and twice the odds compared with men who have sex with men (MSM) [
5]. In China, TGW are not covered by routine HIV surveillance [
9], but several small-sample studies suggest TGW’s high HIV prevalence rate. The number might even be underestimated because of low rates of HIV testing among this particular population in China [
9]. For example, in a cross-sectional study among 220 TGW sex workers conducted in 2014, 16.8% of the participants self-reported as HIV positive and another 9.1% were detected to be HIV positive through free HIV testing [
10].
Among all risky sexual behaviors, unprotected anal intercourse (UAI; anal intercourse without condoms) has the highest risk for HIV acquisition and transmission among TGW [
11,
12]. Pinkerton et al. found that inconsistent or non-condom users were 10–20 times more likely to acquire HIV than consistent condom users [
13]. Promoting consistent condom use among key population has been verified to be a promising risk reduction strategy. Smith et al. estimated that condom effectiveness of reducing HIV incidence was 72.3% (95% confidence intervals [CI]: 60.7–80.5%) for receptive anal sex, 62.9% (95% CI: 46.3–74.3%) for insertive anal sex, and 70.5% (95% CI: 58.2–79.2%) for any receptive or insertive anal sex [
12]. A meta-analysis by the World Health Organization of five cohort studies [
14] assessed condom effectiveness among MSM and transgender people in developed countries and found that the relative effect on HIV transmission of condom use was a relative risk of 0.36 (95% CI: 0.20–0.67) and that consistent condom use (CCU) reduced HIV transmission by 64% (95% CI: 32–80%).
However, none effective interventions have been carried out among TGW so far. UAI prevalence remains high among TGW, even higher than other sex/gender minorities. A cross-sectional study conducted by Shan et al. in two Chinese provinces found that 81.5% TGW reported not always using condoms with stable partners, and 70.9% not with casual partners (Shan et al., 2018). According to Poteat et al., TGW had more than twice the odds as MSM of reporting condomless receptive anal sex with male partners (Poteat et al., 2017). The high prevalence of UAI highlights the urgent need for targeted and effective prevention interventions targeting this population.
Theory-based intervention is widely believed to be more likely to effectively change behavior and maintain such behavior change than non-theory-based intervention. Among extant theory and models, the Information-Motivation-Behavioral Skills (IMB) model has received consistent empirical support in the context of HIV prevention among various key populations, including heterosexual people living with HIV [
15], sexually transmitted infection (STI) clinic patients [
16], female sex workers [
17], senior high school students [
18,
19], college students [
20], male street laborers [
21], unmarried female migrants [
22], and MSM [
23].
The model posits that HIV-preventive behavior is a function of individuals’ knowledge (information) of HIV transmission and prevention, their motivation to carry out HIV prevention (attitudes toward and subjective norms regarding HIV-preventive acts), and their skills in conducting specific preventive behaviors (possession of the requisite skills and a sense of self-efficacy) [
24,
25]. The model further assumes that information and motivation may have direct (when complicated or novel behavioral skills are unnecessary to behavior) or indirect effects (through behavioral skills) on behavior [
24,
25].
Fisher et al. pointed out that the IMB model constructs were highly generalizable determinants of HIV-preventive behavior in diverse populations of interest, but these constructs should have content that is specific to particular target populations as well as to particular HIV-preventive behaviors [
24]. To the best of our knowledge, there have been no tests of the utility of the IMB model of condom use among TGW. It is therefore important to conduct a test before using it to guide the development of interventions targeting this population.
Based on the above-mentioned evidence, we intended to examine the application of the IMB model to clarifying the association between condom use correlates and condom use frequency among TGW in China. Theoretical evidence was expected to be provided to guide the development of interventions targeting increasing condom use frequency in this population. We hypothesized that information and motivation would affect condom use directly and indirectly through behavioral skills and would be independent from each other.
Discussion
The results of this study demonstrated the utility of the IMB model for understanding condom use among TGW in China. The IMB model fit the data well and accounted for 38.7% of the variance in condom use frequency. The findings supported some, but not all, of the hypothesized associations between information, motivation, behavioral skills and condom use.
Results showed that the CCU rate (47.0%) was much higher than that reported in previous Chinese research (18.5% with stable partners, 29.1% with casual partners, 16.7% when buying sex, and 14.0 when selling sex) [
40]. We attributed it to the higher proportions of subjects reporting transactional sexual intercourse in the current sample compared to the previous study; 170/198 (85.9%) have had transactional sexual intercourse in the previous 6 months. Cambou et al. discovered that unprotected anal intercourse was more prevalent in partnerships with a greater degree of interpersonal commitment like stable partners [
41]. So condom use with transactional sexual partners was normally more frequent than stable and casual partners, which buffered the overall rate. Variable CCU rates with different types of sexual partners detected in the current study proved this speculation; CCU rates with regular sexual partners, casual sexual partners, and transactional sexual partners were 26.3, 28.8, and 59.1%, respectively. Anyway, the overall CCU rate in China was much lower than that reported in previous systematic research among US transgender population (61.8%) [
1], indicating the poor control of high risk sexual behavior among Chinese TGW and appealing to urgent and effective intervention.
Furtherly, a SEM was conducted to demonstrate the utility of the IMB model in clarifying the association among information, motivation, behavioral skills and condom use. Inconsistent with our hypothesis, motivation was not directly associated with condom use. However, in accord with some previous studies [
19,
42], the effect of motivation on condom use was completely mediated by behavioral skills. More specifically, TGW who are highly motivated (who have a positive attitude and strong subjective norms about condom use) are likely to acquire the requisite skills (condom use skills and self-efficacy) for condom use. The results highlighted the crucial role of behavioral skills in promoting condom use.
While condom use is the most reliable method to reduce HIV risk, it is also a complex sexual behavior. Previous studies found that using a condom conveys potentially undesirable implicit messages such as lacking trust, lacking commitment or being HIV/STI-infected [
43]. As such, complicated skills, such as condom use negotiation, were needed to convert one’s safer-sex intentions into practical safer-sex behaviors (e.g. condom use). The important role of behavioral skills on condom use has been proved among other populations in previous studies. A meta-analysis of psychosocial correlates associated with heterosexual condom use found that partner communication about condom use was the most highly correlated factor in association with condom use [
44]. The intervention effect of behavioral skills has also been verified among other populations. Gause et al. concluded through their meta-analysis that participants who were exposed to HIV prevention interventions targeting improved partner communication used condoms more frequently [
43]. It’s reasonable to speculate based on our findings and present studies that intervention targeting improving TGW’s condom use behavioral skills can significantly elevate frequency of condom use among this population.
Consistent with the IMB model, correlation between motivation and information was insignificant, identifying motivation and information as independent constructs [
11]. The path from motivation to condom use by way of behavioral skills can exist independently. It indicated that well-informed was an unnecessary prerequisite for an individual to be well-motivated.
Contrary to the IMB model, information showed no association with behavioral skills and condom use. Previous tests of the IMB model have found inconsistencies in the relationship between information and behavioral skills [
19,
21,
22] and between information and condom use [
16‐
18,
20,
23,
24]. Some researchers ascribed it to the speculation that information might impact initial behavioral change rather than maintain such behavior over time [
22,
45]. Others speculated that one’s information might have limited impact on dyadic behavioral change such as condom use, which involves communication and negotiation with his partner [
22,
46]. In the current study, the lack of association between information and behavioral skills may be further ascribed to these two constructs not being measured at the same level of specificity [
16]; information comprised knowledge about HIV transmission and self-protective behaviors, while behavioral skills consisted of condom use negotiation skills and unprotected sex refusal skills. Nevertheless, results of information provided some hints for future intervention. Subjects showed obvious deficits and misconceptions about HIV risks and self-protective behaviors (11.0/18) in the current study, indicating that the extant health propaganda and education on HIV prevention in China remains insufficient among TGW population.
The present study has some limitations that should be acknowledged. Firstly, cross-sectional study limits our ability to make causal inference. Longitudinal studies are needed to further verify the causal mechanism. Secondly, for lack of sampling frame, the participants were recruited by non-probability sampling approaches, namely snowball sampling. Sampling bias might exist. Another method for recruitment of hard to reach population, which is widely believed to be better than snowball sampling, is respondent-driven sampling (RDS). RDS lends statistical rigor to conventional snowball sampling through longer recruitment chains, recruitment limits, and the collection of data used to statistically adjust for the biases inherent in how persons of similar characteristics are networked and likely to recruit each other. Nevertheless, RDS also needs much higher funds than snowball sampling because with each qualified new participant recruited by the seeds in RDS, extra financial reimbursements are needed. Limited to research grant, we were only able to provide such a high reimbursement (200 CNY, about 30 USD) to the participants for completing the questionnaires alone (i.e. no more money for referral), in order to guarantee the quality of the answers provided by the participants. In future research, we will figure out a better method to solve this problem. Furthermore, as the subjects were all from Shenyang, Liaoning Province, it may not be possible to generalize the findings to TGW from other locations. Lastly, some IMB model studies among other populations have used it to compare different groups (such as patients reporting prior STI treatment versus those not reporting such treatment (Scott-Sheldon et al., 2010)) to define a more specific target subgroup. The CCU rate showed substantial variation across different types of sexual partners in this study, indicating possible disparity of the modeling results. Limited to sample size, we failed to make further explorations. Future research is recommended to make clear the mechanisms underlying how information, motivation, and behavioral skills affect condom use with different types of sexual partners among TGW.
Publisher’s Note
Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.