An intervention is described that uses the principles of social cognitive theory (SCT) to reduce sexual risk behaviours among adults from low-income communities. A two-group randomised control trial was designed using 580 participants at self-reported sexual risk. Relative to an educational control condition, the intervention is designed to affect behaviour change through enhancing self-efficacy to execute safe sex behaviours and improving outcome expectancies of condom use. Outcome variables (number of unprotected sexual encounters) and mediating variables (SCT components) will be assessed at baseline and six and 12 months later. Strategies for testing efficacy are presented and limitations discussed.
1. Introduction
Despite continuing preventative efforts, the global incidence of HIV/AIDS is escalating. Given that the pandemic embodies an international health crisis that claims 2.8m lives a year (WHO, 2008), it is unsurprising that research into HIV prevention is constantly developing in its significance and sophistication. Indeed, according to Albarracín et al. (2005:857) it "constitutes one of the most significant paradigms for the discovery of health behaviour change techniques and for the understanding of the theoretical processes underlying such change."
At present, societal efforts to curb HIV infection often focus on educational campaigns (Moseley et al., 2008), yet while heightened awareness is an important prerequisite for self-protective conduct, information alone is rarely sufficient to improve refractory health habits (Sales et al., 2007; Santelli et al., 2009). Conversely, HIV preclusion requires individuals to exert self-directed change and control over both their behaviour and social environment (Johnson et al., 2008). Such self-regulation requires not only appropriate self-guidance and motivation, but also the behavioural resources to alter high-risk habits, and to employ these skills efficiently and consistently under challenging conditions. Essentially, "risk reduction calls for enhancement of interpersonal efficacy rather than simply targeting a specific…behaviour for change" (Bandura, 1994:27).
Identifying factors influencing precarious sexual practice is an important aspect of preventative HIV efforts. Within this endeavour, social cognitive theory (SCT; Bandura, 1986, 1997) highlights the role of both self-efficacy (confidence and skills) and outcome expectancies (OEs; beliefs about potential consequences) for converting health knowledge into operative self-protective behaviour. In the province of HIV prevention, research has identified the following skills and OEs as appropriate targets for promoting behaviour change: negotiating condom use (Cohen, 2009); expectations about condom use (Kanekar & Sharma, 2009); anticipated partner responses (Lewis et al., 2009); sexual self-control when condoms are unavailable (Kiene, 2008); and frank communication about sexual history (Noar et al., 2006).
This intervention aims to use SCT principles to reduce HIV risk behaviour within a group of participants (Ps) vulnerable to infection; specifically adults from economically disadvantaged communities, who are traditionally at higher risk than more affluent counterparts (Rubin et al., 2010). Its content is guided by Albarracin and colleague's (2005) consideration that active approaches to prevention (e.g., interpersonal skills such as role-play, behavioural strategies to promote correct condom use and self-management, such as practicing decision-making for high-risk scenarios) are associated with greater improvement in condom use than passive interventions (e.g., normative/attitudinal arguments and fear appeals). Although numerous interventions have addressed HIV risk-reduction, the majority of trials (regardless of theoretical framework) have focused solely on evaluating behavioural outcomes (Johnson et al., 2008). In contrast our invention is designed to verify the precise mechanisms of behaviour change through mediation analysis (Baron & Kenny, 1986). Indeed, failure to detect mediation effects for any SCT dimensions will be considered latently useful information, as it may help guide and refine future interventions by highlighting superfluous strategies.
The primary outcome is sexual risk behaviour, established on the proportion of unprotected sexual encounters. The secondary outcome will be changes in SCT mediators. It is expected that the intervention group will exhibit significant reductions in self-reported sexual risk behaviour relative to controls. Furthermore, we anticipate that targeted SCT components will mediate effectiveness in terms of (a) being higher post-intervention in the experimental group than the control group and (b) attenuating the effect of treatment condition on behavioural risk outcomes.
2. Method
Design
The study will employ a two-group randomised trial design, with measures taken at baseline (T1), and at six (T2) and 12 (T3) month intervals. A priori power analysis indicates that two samples of 240 would yield .80 statistical power, to detect an effect size of .60 within a Type 1 error rate of .05. A sample of 576 will be recruited to accommodate an attrition rate of up to 20%. Randomisation will be conducted by number generating software (Microsoft RAND) and dispatched in coded envelopes. To confirm groups have comparable baseline awareness of HIV prevention, Ps will be blocked according to scores on the HIV Knowledge and High Sexual Risk Scores' Scale (DiClemente et al. 1993) before randomisation.
Participants
Recruitment will take place between June 2010 and February 2011 within NHS sexual health clinics serving low-income populations in the West Yorkshire area. To be eligible for inclusion, Ps must (a) be aged 18 years or over (b) have had unprotected sex within the last three months and (c) meet one of the following HIV risk criteria: having a sexually transmitted infection; having multiple sexual partners; having sexual partners who are HIV positive and/or an intravenous drug user. Exclusion criteria will be an inability to read English, incarceration, and indication of cognitive incapacity. Baseline data will be collected to determine various P characteristics, including: sexual behaviour; sexual health status; HIV risk status; and sociodemographic information. Figure 1 displays anticipated P flow throughout the study.
Excluded n=220
Did not meet criteria n=150
HIV risk criteria = 110
Age = 20
Cognitive/English = 10
Incarceration = 2
Other = 8
Declined to participate n=70
ENROLMENT
Assessed for eligibility n=800
Randomised n=580
ALLOCATION
Intervention n=290
Control n=290
Did not receive intervention n=25
Time commitments = 12
Loss of interest = 7
Moved = 2
Death = 1
Unknown = 4
Did not receive intervention n=25
Time commitments = 12
Loss of interest = 7
Moved = 2
Death = 1
Unknown = 4
Lost to follow-up n=25
Lost to follow-up n=25
FOLLOW-UP
Completed baseline and at least one follow-up measure n=240
T1, T2 & T3 n=120
T1 & T2 n=60
T1 &T3 n=60
Completed baseline and at least one follow-up measure n=240
T1, T2 & T3 n=120
T1 & T2 n=60
T1 &T3 n=60
ANALYSIS
Figure 1. Flow diagram of (projected) progress through the intervention.
Experimental Protocols
Multi-Component Intervention. Ps will be divided into groups of 20, and receive four three-hour HIV risk-reduction session designed to induce behaviour change using the principles of SCT. Sessions are scheduled weekly and feature components designed to enhance confidence and self-efficacy around condom use, negotiating condom use with partners, and sexual self-control. OEs will be targeted by providing information about potential outcomes in terms of condom use/non-use, modeling strategies, and goal setting with feedback.
Small-Group Educational Programme. This intervention serves as a control for sexual health awareness and knowledge of HIV prevention. Groups of 20 Ps will each receive a two-hour session containing a presentation from an HIV nurse-counselor, followed by a question-and-answer session. Free condoms will be provided in both conditions.
Assessment Instruments
The following measures will be used to operationalise SCT constructs (see Figure 2).
Outcome Expectancies
Evaluating consequences
Potential outcomes of condom use/non-use
Modeling strategies
Goal setting with feedback
Enhancing confidence and mastery
Negotiating condom use with partners
Overcoming hedonic aversion to condoms
Refusing to have unsafe sex
Behavioural skills
Correct and appropriate condom use
Self-Efficacy
Knowledge
Raising awareness of HIV transmission and strategies for prevention
Figure 2. How behaviour change strategies are linked to SCT constructs within the intervention.
Self-Efficacy for Condom Use. Associations between behaviour and self-efficacy are greatest when measures are tailored to both specific risk behaviours and the situational dimensions connected with them (Bandura, 1994). The Condom Use Self-Efficacy Scale (Dilorio et al., 1997) is a Likert-type measure (not confident-extremely confident) that determines self-assurance for condom use according to various situational factors related with unsafe sexual practice (e.g., initiating condom use with an unwilling partner; overcoming hedonistic aversion to condoms; refusing to have unsafe sex).
Skills Measure of Condom Use. Because prophylactics are a crucial constituent of HIV risk-reduction, researchers will conduct a behavioural assessment to determine Ps awareness of correct and safe components of condom use. An eight-item scale developed by The National Institute of Mental Health HIV Prevention Trial Group (2001) will be used to rate Ps between 0-5 if they demonstrate awareness of particular skills.
Outcome Expectancies for Condom Use. SCT specifies three domains for anticipated OEs: self-evaluative, social and physical (Bandura, 1986). These will be evaluated using the Outcome Expectancy for Condom Use Questionnaire (Dilorio et al., 1997), a Likert-type scale which assesses (i) feelings about self as a result of condom use (ii) expected partner responses to condom use and (iii) the hedonic impact of condom use. Higher scores denote more positive beliefs.
Knowledge of HIV Prevention. An essential determinant of behaviour change within SCT is knowledge. The adequacy of HIV prevention awareness will be assessed using the HIV Knowledge and High Sexual Risk Scores' Scale (DiClemente et al., 1993). Response options are dichotomous (safe-not safe), with higher scores representing a greater understanding of HIV risk-reduction.
Procedure
After obtaining approval from relevant NHS Care Trusts, the study will be conducted in sexual health clinics at various sites across West Yorkshire. At each clinic, nurses will be asked to review patients with forthcoming appointments to determine possible eligibility. Upon arrival at the clinic, patients meeting relevant criteria will be provided with a leaflet describing the study. Names of interested persons will then be passed onto research nurses employed by the study. At this point, potential Ps will complete a screening interview to ascertain eligibility, then be contacted for follow-up whereby consent and T1 measures will be obtained. Randomisation envelopes will also be dispatched, with Ps assigned to either intervention or control condition. Participant screening, preparation of the random sequence, and group allocation will be undertaken by separate researchers. Group sessions will take place in private rooms within the clinics and administered by trained nurses specialising in sexual health. For purposes of quality control, sessions will be audio-taped and assessed for fidelity. T2 and T3 assessment material will be sent with stamped-addressed envelopes to Ps two weeks before the completion date. All Ps will receive £5 after completing each measure (T1, T2, T3), for a total reimbursement of £15.
Statistical Analysis
All statisticians will be blinded in terms of group assignment. Ps included in analysis must have at least one follow-up measure with non-missing values. Separate analyses
will be run to assess immediate and delayed intervention effects between T1/T2, and T1/T3. Baseline data (i.e., sexual behaviour; sexual health status) will also be collected at these points. Evaluation of SCT mediators will not take place pre-intervention (i.e., before randomisation), as this may prove reactive (O'Leary, 1992). However, because P characteristics will be roughly equivalent in both conditions, it is expected that SCT variables will be similar at baseline. Intention to treat analysis will also be performed to minimise latent artifacts, such as attrition effects.
It is expected that there will be a significant intervention effect on condom use, which will be assessed using a linear regression model unadjusted for SCT variables. Correlational analyses of SCT variables (OEs, skills, self-efficacy in these skills, relevant knowledge) will be conducted to determine whether they represent distinct constructs. Mediation analyses will then be used to ascertain whether these components are responsible for anticipated effects. Logistic regression models will be used to asses whether the experimental condition predicts scores on mediator variables, with repeated-measures models used to determine the effects of the intervention and the mediators on sexual health outcomes at various time-points. SCT variables will be considered individually in order to determine any differential contribution to intervention effect and reduction in sexual risk behaviour. Separate analyses will be run for both intervention and control groups.
In addition to group assignment, exploratory subgroup analyses will be performed to determine the differential impact (and potential moderating effect) of the intervention across: gender; age; educational level; ethnicity and; sexuality. Interactions with SCT mediators will also be assessed within these subgroups.
Conclusions
This intervention is designed to provide empirical support for the utility of SCT principles in enhancing the practice and personal application of HIV risk-reduction behaviour in an efficacious, precise and cost-effective manner. In its present form, potential design limitations include the small-group format, which may impair the skill-building components in that they cannot be sufficiently customized to individual circumstances (Bandura, 1982). Furthermore positive, yet unmeasured, effects may also facilitate behaviour change (e.g., influencing subjective norms, enhancing social support and/or self-worth). Such variance may affect definitive conclusions about the role of SCT constructs in intervention outcomes. Finally, stronger inferences about the utility of SCT may have been drawn through direct comparison to another theoretical model, such as the Health Belief Model (Rosenstock, 1974) or Theory of Planned Behaviour (Ajzen & Fishbein, 1980).
Notwithstanding these potential shortcomings, the current intervention offers a theoretically-strong approach to support, develop and refine important competencies for safeguarding against HIV infection in vulnerable groups.
