Psychosocial Correlates of HIV-related Sexual Risk Factors among Male Clients in Southern India

Psychosocial theories suggest that individuals’ behavior is a reflect ion of their intention and ability to carry out a typical behavior. Th is study proposes to examine the psychosocial correlates of HIV-related sexual risk factor among male clients of female sex workers (FSWs). Data were used from a cross-sectional survey, collected using two-stage sampling, conducted among 2382 clients of FSWs in four states of India in November 2008. Clients were males who had engaged in paid sex with a FSW in the 12 months preceding the survey. Multiple logistic regression models were fitted to assess the effect of d ifferent psychosocial measure on HIV-related sexual risk factors: multiple sexual partners, inconsistent condom use and self reported sexually transmitted infections (STIs). The odds of inconsistent condom use with FSWs was more among clients with low self-efficacy (Adjusted Odds Ratio (AOR): 2.2, 95% Confidence Interval (CI): 1.7-3.0), low perceived social support (AOR: 1.8, 95% CI: 1.3-2.6), low perceived personal norms (AOR: 1.7, 95% CI: 1.2-2.3) and low perceived access to condoms (AOR: 1.5, 95% CI: 1.1-2.0) than others. Similarly, experience of STI-related symptoms in the last 12 months was associated with low self-efficacy, low perceived social support and low perceived vulnerability. Findings highlight strong influence of psychosocial attitudes on HIV-related sexual risk factors among male clients of FSWs, suggesting the need for designing HIV prevention strategies to address psychosocial issues like self-efficacy, vulnerability and social support.


Introduction
Socio-demographic and behavioral characteristics of a population are considered to be the key parameter for designing HIV prevention programs. Over the year program implementers have consistently ignored the role of psychosocial perspectives of program beneficiaries [1][2][3]. A growing body of literature suggests that translating psychological theories into practice can help in implementing more effective HIV intervention programs [1,3,4]. These studies suggest that individuals' behavior is a reflection of their intention and ability to carry out the behavior. Although the relationship between behavioral intention and actual behavior cannot be accurately measured, the intention to act can be considered as a proximate measure of behavior. Therefore, behavioral intentions can be measured in terms of attitudes, perceived social norms, perceived self-efficacy and perceived severity [5,6].
A rev iew o f psychosocial theo ries related to health seeking behavior suggests that psychosocial factors like perceived vulnerability, self-efficacy, social support, accessibility and personal norms are central to most of these theories [7,8]. Self-efficacy refers to beliefs about the ability and effort required to perform a pro moted health behavior effectively [9]. According to Bandura [10], perceived self-efficacy denotes people's belief that they can exercise control over their motivation as well as behavior. Perceived vulnerab ility is another factor which has received attention from researchers and program planners as it can influence health seeking behavior of ind ividuals. According to Rogers [11], vulnerab ility is the risk perception of being infected with HIV if the reco mmended behavior is not adopted. Further, along with self-efficacy and perceived vulnerability, indiv idual's perception about things matters a lot wh ile adopting a behavior [12,13]. Emp irical research shows that personal norms have strong anticipated affective outcomes to perform a behavior [7]. Personal norm is an individual's self regulated influence on own functioning with respect to intention to perform a behavior [10]. Personal norms are primarily used to identify internalized cognitive processes that are based on an individual's perception of the ethical correctness of performing a behavior [7]. Individuals' effort to perform a behavior cannot be Factors among M ale Clients in Southern India attributed to only their own attitudes and perception, but also can be due to social support and availability of services supportive to perform such behavior. In addit ion to personal motivation and attitudes, an individual's social support network can act as a source of information, source of emotional and practical support, approving institution to perform a behavior [14,15]. Personal network norms and normative beliefs about what others in the social and personal networks are actually doing are crit ical in adopting a new behavior. Non-availability and poor access to services can hamper the indiv iduals' utilization of services. Regardless of self motivation and positive social support, the perceived accessibility of a service can be detrimental in the utilization of that service. Several studies have documented a positive effect of these psychosocial factors on HIV risk behavior among general populations [16][17][18] including female sex workers (FSWs) [19] and their clients [20]. Emp irical research investigating psychosocial correlates of condom use have found that positive attitude towards condom use [5,21], self-efficacy [4,5,16,17,21] and perceived susceptibility to sexually transmitted infections (STIs) [22,23] are positively correlated to condom use practice. A few studies reveal that perceived barriers to condom use and peer norms can hamper condom use practices [18,24].
In India, the national AIDS control program has identified FSWs and their clients as two priority groups for HIV prevention interventions. The focus of such interventions are mainly provid ing HIV p revention education through peer educators combined with free condom distribution and treatment for STIs [25]. Recent research report suggests that in 2009 about 6% of clients of FSWs were HIV seropositives and nearly 60% reported using condom consistently with occasional FSWs [26]. A number of research studies have been undertaken in diverse settings to understand the correlates of condom use among clients of FSWs [4,21,[27][28][29]. On ly a few studies have been undertaken in India in this area of research; these studies have focused on understanding the socio-demographic and behavioral determinants of condom use. The role of psychosocial attitude in determin ing an indiv iduals' HIV risk behavior is yet to be explored in the Indian context. The current research is an attempt to document the relationship between different psychosocial measures and HIV risk behaviors among male clients of FSWs.

Subjects and Procedures
A cross-sectional survey among clients of FSWs was conducted simultaneously in the Indian states of Andhra Pradesh, Karnataka, Tamil Nadu and Maharashtra in November 2008. These states had been identified as high epidemic states by the Indian National AIDS Control Organization (NA CO) prior to the year 2005 [25]. The study protocol was approved by an ad-hoc ethical co mmittee chaired by the Indian Council of Medical Research (ICM R). Data were co llected using a stratified t wo-stage cluster sampling approach; in the first stage hotspots were selected while part icipants were selected in the second stage. States were considered as strata except for Maharashtra, where Mumbai and rest of Maharashtra were t reated as separate strata. The sampling frame was developed based on a list of hotspots of sex work act ivity (a place where FSWs gather to solicit clients), prepared with the help of local non-governmental organizations (NGOs) imp lementing the HIV intervention among clients of FSWs in the survey states. The target sample size for the survey was 2400 individuals (480 per stratum). This desired sample size was arrived at after assuming a 15% change in the behavioral indicators over time with a 95% significance level and 80% power.
The primary samp ling units (PSU) were the hotspots of sex work act ivity wh ich were selected by probability proportional to size samp ling approach. A total of 30 hotspots per strata were selected. For each PSU, the number of interviews to be conducted was fixed and interviews were conducted between 10 a.m. and 7 p.m. on all days of the week. Multip le rounds of visits were made to the same PSU in case the target number of interviews could not be completed in one visit. Trained interviewers were stationed at selected points in the hotspots and instructed to approach every fifth man passing by during a specified timeframe to avoid introducing selection bias based on the interviewer's judgment. Interviewers were male graduates in a social science subject or statistics and had prior experience in data collection among most at risk population. Males who were 18 years or o lder and were involved in co mmercial sex (paid to have sex with an FSW) in the 12 months preceding the survey were eligib le to participate in the study. Continuous monitoring mechanisms were in place to ensure smooth data collection and quality control. Each interview took approximately 45 minutes to complete.
By the end of the survey, 20,850 individuals were approached; all of them were informed about the study and asked about their willingness to participate in the survey; of these 14,413 indiv iduals were not interested in part icipating in the survey due to lack of t ime or indifference. Individuals who volunteered to participate in the survey were asked for informed consent. These individuals were further screened for their eligib ility to participate in the survey. In all 3,810 individuals were found to be ineligib le for the survey. Further, 245 indiv iduals did not co mplete the interviews resulting in a total sample of 2,382 individuals.

Psychosocial Measures
The survey instrument was designed to collect extensive informat ion on different psychosocial dimensions of condom use. The items measuring psychosocial attitudes in the quantitative survey were selected based on the informat ion gathered in another qualitative research study conducted earlier in the study areas [30]. Measures of perception related to condom use and HIV/AIDS were collected using five-point Likert type scales (ranging from 'strongly agree' to 'strongly disagree'). So me statements were worded 'negatively' in order to avoid art ificially high consistency within the responses. Most of these statements were used in previous research on health behavior informat ion.
Information was collected for the fo llowing domains: perceived self-efficacy, perceived vulnerability, perceived social support, perceived accessibility of condoms, and personal norms. Co mposite scores were co mputed for each dimension by using the mean value of the items included in the scale. During the analysis stage, items worded 'negatively' were reverse coded to reflect the positive response in those items. Cronbach's alpha values were computed for each of the scales to demonstrate the reliability of the score (Table 1). In order to examine the effect of psychosocial variables on the outcomes, these scores were further d ivided into three equal parts to generate a scale with the categories low, moderate and high, where low represents the lowest scores and high represents the highest scores.

Perceived Self-Efficacy
Twelve items were used to measure perceived self-efficacy for condom use (see Appendix 1 for a complete list of items included in the scale). Typ ical items included: "I find it difficu lt to dispose off a condom after using it", "I would hesitate to buy condoms fro m a shop if the shopkeeper is a wo man", "Even if I am drunk, I can remember to use a condom with an FSW" and "Even in situations when I get excited by an FSW, I can remember to use a condom". The Cronbach's alpha coefficient for the composite scale was 0.76 (Table 1).

Perceived Vulnerab ility
Items that were meant to assess respondents' self perception of the likelihood of contracting STI/HIV and their perception about HIV/AIDS in society were considered here. Typical items were: "If a man knows an FSW well, it is safe not to use a condom with her", "If the pimp reco mmends an FSW who does not have any disease, it is safe to have sex without a condom", "Very few people get HIV/AIDS" and "HIV/AIDS is happening only in big cities." (Cronbach's alpha: 0.67).

Perceived Social Support
Social support has been assessed in terms of the kind of assistance an individual has received fro m friends, family members, FSWs and other stakeholders. Though eight items were presented to respondents, we considered only five items to increase the reliability of the scale (Cronbach's alpha= 0.84). Typical items included were: "Most of my friends encourage me to use a condom"; "Most of my friends often tell me about diseases I can get by not using condoms" and "An FSW once explained the importance of using condoms to me".

Perceived Accessibility
Eight items were used to measure perceived barriers to condom use (Cronbach's alpha= 0.60). Of these eight items, four were related to condom availability and the remaining four reflected barriers to condom use. So me of the items included in this measure are: "Generally, condoms are always availab le near the p laces where one can find FSWs", "If needed, I can get a condom within the next 5-10 minutes walking distance fro m this place" and "I want to use a condom but the FSW does not want me to use a condom".

Personal Norms
Twelve items were used to assess an individual's person norms towards condom use. So me of the items included in the scale were: "I don't trust an FSW even if I have known her for somet ime", "I have developed a relationship with the FSW and I don't want to ru in it by introducing condoms", "I enjoy sex more if I use a condom" and "sex is for fun, so why bother about condoms" (Cronbach's alpha= 0.73).

Outcome Measures
The following measures of HIV-related sexual risk factors were examined in this study: inconsistent condom use with FSWs, inconsistent condom use with regular non-paying partners, mult iple sexual partners and self reported STI sympto ms.

Inconsistent Condom Use with FSWs
Inconsistent condom use with FSWs was derived by taking three separate questions together that assessed "condom use in last sex (no, yes)", "frequency of condom use in last 12 months (every time, most of the time, sometimes, very few t imes)" and "any occasion of not using a condom with an FSW in the last 12 months (no, yes)". An individual was defined as an inconsistent condom user (coded as 1) with FSWs in the 12 months prior to the survey if he d id not use a condom at last sex with an FSW, or d id not use a condom "every time" with FSWs in the last 12 months or reported at least one instance of not using a condom with an FSW in the last 12 months; otherwise he was identified as a consistent condom user (coded as 0).

Inconsistent Condom Use with Regular Partners
Single item questions about the number of sexual acts with regular partners (wife, fiancée and girlfriend) in the last 30 days and condom use in those sexual acts were asked. Respondents who reported condom use in all sex acts were consistent condom users with regular partners (coded as 0), else considered inconsistent condom users with regular partners (coded as 1). Factors among M ale Clients in Southern India

Mu ltiple Sexual Partners in the last 12 Months
Information on the nu mber of sexual partners in the last 12 months was collected from each survey participant. The total number of sexual partners was regrouped into two groups based on the median split: < 10 (coded as 0), 10+ (coded as 1).

Self Reported Sexually Transmitted Infections
Information about experience of the following five STI symptoms in the last 12 months was collected: (a) burning sensation/pain during urination, (b) thick discharge fro m penis, (c) ulcers/sores in the groin area, (d) scrotal swelling and pain, and (e) enlarged and/or painful inguinal ly mph nodes. Experience of any of these STI symptoms in the last 12 months was considered as experience of STIs and coded as 1, else coded as 0.

Socio-Demographic Variables
Information on socio-demographic variables like age, marital status and education were assessed using single item questions. Similarly, single item questions were used to collect informat ion on occupation (recoded as unskilled worker, skilled worker, self-emp loyed, salaried and unemployed) and alcohol consumption (never, occasional consumption, regular consumption). We used age (continuous), marital status (unmarried, married living with wife, married living without wife), education (no formal education, Class 1-9, Class 10-12, graduation or more), alcohol consumption and occupation as covariates when predicting a particu lar outcome with d ifferent psychosocial measures.

Statistical Anal yses
Univariate, b ivariate and mult ivariate analyses were performed. Pearson's Chi-square test was used to measure the strength of association in the bivariate analyses between psychosocial measures and outcome variables. Multip le logistic regression models were used to predict different HIV risk behavior variables for different psychosocial measures. Two sets of regression analyses were performed; in the first case, the effect of one psychosocial measure was measured after controlling fo r respondents' socio-demograp hic covariates; in the second case, the effect of all psychosocial measures was measured simu ltaneously after adjusting for socio-demographic covariates. The significance level for all statistical tests was 5% or otherwise specified. Results were presented in the form of percentages, adjusted odds ratios (AOR) and 95% confidence interval (CI) of point estimates. Data were analyzed using STATA 11.1.

Results
Clients of sex workers were, on average, 30 years old (Standard deviation (SD): 7.3 years) and around one-quarter (26%) were 25 years or less (Table 2). Most respondents (88%) had some education and more than half (55%) were educated up to class 10 th or more. Nearly two-th irds (67%) were married and about half (52%) were residing with their wife. About two-thirds of respondents were either unskilled workers (34%) o r skilled workers (33%) and only 6% were unemployed. Twenty-eight percent did not consume alcohol and 46% consumed alcohol regularly.  Around 16% of clients reported inconsistent condom use with FSWs in the last 12 months and 82% reported inconsistent condom use with their regular non-paying partners in the last 30 days (Table 3). Inconsistent condom use with FSWs was reportedly higher among individuals with low self-efficacy (21%), lo w perceived social support (24%), low perceived accessibility of condoms (19%) and low perceived personal norms as compared to those identified as high in these psychosocial measures. Furthermore, 24% of ind ividuals with h igh perceived vulnerability reported inconsistent condom use with FSWs as compared to 10% among those with low perceived vulnerability (P<0.001). The pattern of differentials in inconsistent condom use with regular partners was similar to that of inconsistent condom use with FSWs. For examp le, 88% of respondents with lo w self-efficacy reported inconsistent condom use as compared to 78% with high self-efficacy (P<0.001). About two-fifths (44%) reported experiencing an STI-related symptom in last 12 months. Higher levels of STIs were reported by respondents with low self-efficacy (46%), h igh perceived vulnerab ility (49%) and low perceived personal norms (48%) than their counterparts.
The mu ltivariate analyses confirmed the findings fro m bivariate analyses (Table 4). Logistic regression model adjusted for socio-demographics revealed that respondents with low self-efficacy were two times more likely to use condom inconsistently with FSWs than those with high self-efficacy (AOR: 2.2, 95% CI: 1.7-3.0). The association between self-efficacy and inconsistent condom use hold true even after controlling for other psychosocial measures (AOR: 1.9, 95% CI: 1.4-2.5). The other psychosocial factors associated positively with inconsistent condom use with FSWs were lo w perceived social support (AOR: 1.8, 95% CI: 1.3-2.6), lo w perceived accessibility (AOR: 1.5, 95% CI: 1.1-2.0) and low perceived personal norms (AOR: 1.7, 95% CI: 1.2-2.3). Associations of these psychosocial measures with inconsistent condom use with regular partners and experience of STI-related symptoms were statistically significant and in d irections similar to inconsistent condom use with FSWs. Perceived vulnerability was negatively associated with mult iple sexual partners, but positively associated with inconsistent condom with FSWs and regular partners and experience of STIs.

Discussion
This study, based on a cross-sectional survey, examined the effect of five d ifferent psychosocial measures on inconsistent condom use, mu ltiple sexual partners and experience of STI-related symptoms among male clients of FSWs. This research adds to existing evidence and demonstrates a strong relationship between individuals' attitudes and perceptions with HIV-related sexual risk factors. Perceived self-efficacy and perceived vulnerability to HIV were associated with practices such as multip le sexual partners, inconsistent condom use with FSWs and regular partner as well as with experience of STI-related symptoms. The current study also demonstrated that inconsistent condom use with FSWs is more likely to occur among individuals perceiving lo w self-efficacy, low social support, low personal norms and high perceived vulnerability. Similar associations were documented for other HIV-related sexual risk factors.
Consistent with earlier studies among youths and IDUs [6,12], our study also observed positive effect high self-efficacy on consistent condom use with FSWs and regular partners. This indicates that clients of sex workers had the ability to understand the risk associated inconsistent condom use. This, further, is supported by the post-hoc analysis which suggests that individuals with low perceived self-efficacy have higher perceived vulnerability than those with high perceived self-efficacy. Furthermo re, inconsistent condom use with FSWs and regular partners was higher among indiv iduals with high perceived vulnerability than with low vulnerability. Th is can be a reflection of their actual behavior, that is, those not using condoms consistently were aware of engaging in some kind of activity carry ing a certain risk leading to STI/HIV in fection. A similar relat ionship between perceived vulnerability and condom use has been documented in a recent study among sex wo rkers [31]. Though individuals have accurately assessed their vulnerability to HIV in fection, such assessments have not been translated into behavior.
Behavior change commun ication interventions should make an additional effort to promote a positive behavior change.
The study results further suggest that high perceived social support can have a positive influence on consistent condom use and reduce exposure to STI-related symptoms. The importance of a supportive environment of peers and social networks for condom use has been highlighted in past research which corroborates our study finding [14,[32][33][34]. This indicates that not only the personal motivation level of an individual, but also the perception that condom use is encouraged and approved by the people in the surroundings is important for consistent condom use. Another important predictor of condom use is perceived personal norms about condoms. However, perceived personal norms have no significant effect on mult iple sexual relat ionships. This suggests that personal norms do not restrict individuals fro m selecting sexual partners but rather make them aware of the need to adopt safe sexual practices. The study findings also show that inconsistent condom use was higher with regular partners than with sex workers, suggesting that even with the same degree of psychosocial attitudes individuals do not use condoms with regular partners. This suggests that motivation to use condoms with sexual partners depends on whether the relationship is casual or steady in nature. There are potential limitat ions to our study and hence some caution is required wh ile interpreting and drawing conclusions. First, the find ings are based on a cross-sectional study and hence, the causality of measures cannot be established. For examp le, it is difficult to infer whether low condom use led to higher perceived vulnerability or otherwise. Second, as data collection in the survey was confined to day-time , that is, fro m 10a .m.-7p. m., there may be some kind of bias regarding the representativeness of the client population. A preliminary assessment before the init iation of the study suggested that profile of clients at any time of a day remain same. Third, responses on the frequency of condom use can be biased due to social desirability. In order to reduce such bias, interviews were conducted in a private location after ensuring confidentiality of the informat ion provided. Further, we asked different questions, such as condom use at last sex and any instances when the respondent did not use a condom, to validate the responses on frequency of condom use. While generating the variable on inconsistent condom use, these checks were considered, wh ich helped in reducing the effect of social desirability bias.
The study findings presented here have important implications for policy makers on HIV interventions. First, there is a need to imp rove realization of risk associated with commercial sex as the study findings suggest inconsistent condom use to be associated with high risk perception. Second, the role of an individual's social environment should not be ignored, as higher perceived acceptability of condoms in the informant's peer group is related to mo re consistent condom use. Hence, efforts should be made to promote condom acceptability among male groups, in order to increase the acceptability of condom use. One possible approach to developing "environments of approval" could be to design programs reaching locally known informal leaders who can help to develop positive attitudes about condom use among their networks of young men. In addition, prevention programs should explore mechanisms for enhancement of perceived self-efficacy of individuals. A first step towards this would be to develop skills among individuals that can help them to buy condoms fro m either shops or vending mach ines placed in public places. Also, programs should continue to increase the availability of condoms at suitable locations.

Conclusions
In summary, an indiv idual's psychosocial characteristics are part of the co mplex set of factors that influence unsafe sexual behaviors. This study documented positive association of perceived self-efficacy, perceived social support, perceived accessibility and perceived personal norms on use of condoms among clients of sex workers in India. Several of these psychosocial factors were also associated with the experience of STI-related sympto ms and sex with mu ltiple partners, highlighting the important role played by psychosocial factors in shaping sexual behaviors of individuals. Integrating psychosocial factors into the existing framework of HIV prevention programs can lead to better outcomes. Co mprehensive interventions for preventing HIV/AIDS should include components directed to self-efficacy, perceived social support, and other psychosocial characteristics.