Randomized controlled trial to investigate impact of site-based safer sex programmes in Kingston, Jamaica: trial design, methods and baseline findings

Authors


Corresponding Author Sharon S. Weir, Carolina Population Center, University of North Carolina at Chapel Hill, 206 West Franklin Street Campus, Box 8120, Chapel Hill, NC 27516, USA. Tel.: 919 846 9474; Fax: 919 966 2391; E-mail: sharon_weir@unc.edu

Summary

Objective  To describe the design, methods and baseline findings of a multi-level prevention intervention to increase consistent condom use among persons at public social sites in Kingston, Jamaica, who have new or concurrent sexual partnerships.

Methods  A two-arm randomized controlled trial (RCT) of 147 sites where persons meet new sex partners. Sites were identified by community informants as places where people meet new sexual partners, which include bars, street locations, bus stops, malls and others. Sites were sorted into 50 clusters based on geographic proximity and type of site and randomized to receive a multi-level site-based intervention or not. Intervention components include on-site HIV testing, condom promotion and peer education. Effectiveness of the intervention will be measured by comparing the proportion of persons with new or multiple partners in the past year who report recent inconsistent condom use at intervention vs. control sites.

Results  Baseline surveys were conducted at 66 intervention (711 men, 845 women) and 65 control sites (654 men, 738 women). Characteristics of intervention and control sites as well as the characteristics of patrons at these sites were similar. The outcome variable was balanced with approximately 30% of men and 25% of women at intervention and control sites reporting a new partner or more than one partner in the past year and recent inconsistent condom use.

Conclusions  The baseline findings confirm that the population is an appropriate target group for HIV prevention and that randomization will provide the means to estimate programme effectiveness.

Essai randomisé contrôlé pour évaluer l’impact des programmes sur les rapports sexuels protégés, basés sur le site à Kingston en Jamaïque: conception de l’étude, méthodes et observations de base

Objectif  Décrire la conception, les méthodes et les observations de base d’une intervention de prévention à plusieurs niveaux dans le but d’accroître l’utilisation régulière du préservatif par les personnes dans les endroits publics et sociaux qui ont un nouveau partenaire sexuel ou plusieurs partenaires à Kingston, en Jamaïque.

Méthodes  Un essai contrôlé randomiséà deux bras comprenant 147 sites où les gens rencontrent de nouveaux partenaires sexuels. Les sites ont été identifiés par des informateurs de la communauté comme étant des lieux où les gens rencontrent de nouveaux partenaires sexuels et comprennent des bars, des endroits de rue, des arrêts de bus, des centres commerciaux et autres. Les sites ont été classés en 50 groupes basés sur leur proximité géographique et le type de site et ont été randomisés pour recevoir une intervention à plusieurs niveaux basée ou non sur le site. Les éléments d’intervention sur le site comprennent: le dépistage, la promotion du préservatif et l’éducation par les pairs. L’efficacité de l’intervention sera mesurée en comparant la proportion de personnes avec un nouveau ou de multiples partenaires au cours de l’année précédente qui rapportent une récente utilisation irrégulière du préservatif lors de l’intervention par rapport à la même proportion dans les sites de contrôle.

Résultats  Des enquêtes de base ont été menées sur 66 sites d’intervention (711 hommes et 845 femmes) et 65 sites de contrôle (654 hommes et 738 femmes). Les caractéristiques des sites d’intervention et de contrôle ainsi que les caractéristiques des usagers de ces sites étaient similaires. Les variables des résultats étaient équilibrées, avec environ 30% des hommes et 25% des femmes dans les sites d’intervention et de contrôle ayant rapporté un nouveau partenaire ou plus d’un partenaire au cours de l’année précédente et de récentes utilisations irrégulières du préservatif.

Conclusion  Les observations de base confirment que la population étudiée est un bon groupe cible pour la prévention du VIH et que la randomisation fournira les moyens d’évaluer l’efficacité du programme.

Ensayo aleatorizado y controlado, para evaluar el impacto de programas de sexo seguro basado en el lugar, en Kingston, Jamaica: diseño del ensayo, métodos y hallazgos de base

Objetivo  Describir el diseño, los métodos y hallazgos de base de una intervención de prevención multinivel, para aumentar el uso consistente de condones entre personas en lugares sociales públicos que tienen parejas sexuales nuevas o concurrentes en Kingston, Jamaica.

Métodos  Ensayo aleatorizado y controlado (EAC), con dos brazos, en 147 lugares en los que las personas encuentran nuevas parejas sexuales. Mediante informantes comunitarios se identificaron los lugares en los que las personas encuentran con nuevas parejas sexuales, incluyendo bares, lugares específicos de la calle, centros comerciales y otros. Los lugares se organizaron en 50 grupos basándose en proximidad geográfica y tipo de locación, y fueron aleatorizados para recibir o no una intervención multinivel. Los componentes de la intervención incluían una prueba para VIH, la promoción del uso de preservativos y la educación entre iguales. La efectividad de la intervención se medirá comparando la proporción de personas con parejas nuevas o múltiples en el último año que reportan el uso reciente de preservativo en lugares con intervención vs. lugares control.

Resultados  Se llevaron a cabo encuestas en 66 lugares con intervención (711 hombre, 845 mujeres) y 65 lugares control (654 hombres, 738 mujeres). Las características de los lugares de intervención y control, al igual que las características de los patrones en estos lugares eran similares. La variable de resultado estaba balanceada con aproximadamente 30% de los hombres y 25% de las mujeres en los lugares de intervención y control, reportando una nueva pareja o más de una pareja en el último año, y un uso reciente e inconsistente de preservativos.

Conclusión  Los hallazgos de base confirman que la población es un grupo diana apropiado para la prevención del VIH y que la aleatorización proveerá los medios para estimar la efectividad del programa.

Introduction

The challenges of implementing adequate antiretroviral treatment programmes should spark a renewal of efforts to develop effective human immunodeficiency virus (HIV) prevention programmes. Although many community-based efforts to prevent the transmission of HIV have been implemented by local public health programmes, few have been rigorously evaluated and fewer still proved to be both effective at reducing HIV transmission and feasible for scale-up. Two programmes with evidence of success – the Thailand brothel (Rojanapithayakorn 2006) and the Nicaragua condom distribution programme in hotels (Egger et al. 2000) – were site-based interventions that increased condom use at the time and location of actual use. Although research suggests that condom use during commercial sex increases if condoms are available at the bedside, little attention has been paid to the development of environmental interventions at other public sites where people meet new sexual partners such as bars, clubs or malls that are a ‘step away’ from where sex occurs and where sexual partnerships are not overtly commercial. Although the rate of new sexual partnerships in the larger group of people who socialize at these places is lower than the rate among sex workers, the absolute number of new sexual partnerships formed by the larger group socializing at these sites is potentially much higher than the number formed by sex workers. Consequently, the role of this larger group on transmission of HIV and other sexually transmitted infections (STIs) may be of considerable importance as the epidemic shifts from a concentrated phase to a generalized phase in countries like Jamaica where there are no clear distinctions between core, bridge and general populations.

This paper describes the design, methods and baseline data for the evaluation of effectiveness of a multi-level prevention intervention developed in Kingston, Jamaica, in 2005. The primary aim of the evaluation was to determine whether the Kingston Priorities for Local AIDS Control Efforts (PLACE) intervention increased consistent condom use among persons at public social venues who have new or concurrent sexual partnerships. The approach used for evaluating this programme may be useful for other public health settings where rigorous programme evaluation is desired, but there are practical constraints on the research design, which preclude the implementation of a randomized study design with HIV incidence as the outcome.

Methods

Study population

Jamaica is the third largest island in the Caribbean with a population of 2.6 million and an adult HIV prevalence of 1.5% (Jamaica Ministry of Health 2005a). HIV transmission is primarily heterosexual with a male-to-female ratio among AIDS cases of 1.4:1.0. Adolescent females are three times more likely to be at risk than boys. HIV is the second leading cause of death of both men and women of age 30–34 years in Jamaica. Among all reported AIDS cases, >80% reported multiple sexual partners, 50% a history of STI, 25% had exposure to commercial sex and nearly 20% had no obvious risk factor. Approximately 10% of AIDS cases comprise men who have sex with men. Injection drug use is associated with less than 1% of cases. Although all parishes were affected by the epidemic, the two most urbanized parishes (St James and Kingston/St Andrew) reported the most cases of AIDS (Jamaica Ministry of Health 2005a), the highest cumulative case rate (830 and 595 cases per 100 000 population, respectively) and the highest prevalence among attendees of sexually transmitted disease (STD) clinics (5.4% and 5.8%) (Jamaica Ministry of Health 2005a). Jamaica has low-level generalized epidemic with incidence apparently highest in the urbanized areas and among those with multiple sex partners.

PLACE intervention model

The PLACE intervention was developed by the National HIV/STI Control Program within the Ministry of Health (MOH) based on the available information characterizing the HIV epidemic in Jamaica and long experience with HIV and STD prevention programmes. HIV prevention programmes in Jamaica have focussed on increasing condom use. Condom sales rose from approximately 2 million per year in 1985 to over 10 million by 1994, but have not increased since then. Reported condom use according to national population-based household surveys increased in the early 1990s but remained unchanged for the past 10 years. Since 1997, 24% of men and 34% of women reported not using condoms with non-regular sexual partners. The MOH determined that new approaches were needed to limit HIV transmission given the increasing number of infected persons and the apparent plateau in safe sex behaviour.

Assessments using the PLACE method (Weir et al. 2005) in Montego Bay and Kingston in 2002 and 2003 revealed a large diversity of public sites where people meet new sexual partners, including but not limited to commercial sex sites (Figueroa et al. 2006). Compared with the general population, site patrons were younger, more likely to have had two or more partners in the past 4 weeks and 12 months, and more likely to have had a new partner in the past year.

Based on the proximate determinants model (Boerma & Weir 2005), the intervention objective was to decrease HIV transmission by increasing condom use among persons with high rates of new and concurrent sexual partnerships. The strategy adopted to reach this target population was to transform the places where people meet new sexual partners into locations where condom use is the normative choice. The strategy recognizes the limitations of individual-level interventions (Diez-Roux 1998; Susser 1998) and the potential for influencing behaviour change by changing the social and physical environment of sexual negotiation. It draws from theoretical models that include social models for changing health behaviour (Friedman et al. 1994), including social learning theories (Valdiserri et al. 1992), diffusion theory (Rogers 1982), social cognitive theory (Bandura 1994) and the stages of change theory (Prochaska et al. 1994). Specific factors that may influence behaviour addressed by intervention are:

  • • The role of physical environment in shaping behaviour (Sanders & Campbell 2007);
  • • The importance of characteristics of the events occurring prior to sex including the consumption of alcohol or drugs and venue activities (Kalichman et al. 2007; Lewis et al. 2005; Weir et al. 2004);
  • • The role of peer pressure to initiate early sex and have multiple partners and the positive role of social and peer support in maintaining safer sex (Valdiserri et al. 1992);
  • • Development of skills to increase self-esteem and self-efficacy, including the ability to weigh costs and benefits, recognize symptoms of STI and negotiate safer sex (Murray-Johnson et al. 2005);
  • • Condom availability and environmental ‘cues to action’ to promote condom use; and
  • • The role of poor information, stigma, apathy and fear in limiting condom use and health-seeking behaviours (Ramirez-Valles 2002; Murray-Johnson et al. 2005).

These factors are addressed in the Kingston PLACE intervention at three levels: the environment or site level, the group level among people socializing at these sites and the individual level through one-on-one outreach efforts (Table 1). The full strategy (Jamaica Ministry of Health 2005b) includes the following programme objectives implemented onsite as feasible by trained outreach workers:

Table 1.   Objectives and components of the multi-level PLACE intervention at sites in Kingston, Jamaica, 2005
ObjectivesEnvironmental levelGroup levelIndividual level
Motivate gatekeepers and influential persons to implement and maintain safe sex at siteGatekeeper support Focus on persons who have new or multiple partners in past year
Influential support
Staff development
DJ support
Establish peer education programmesTrain and supervise patrons, sex workers and dancers as peer educatorsPeer educators conducting group educationOne-to-one sessions by trained peer educators to increase self-efficacy in sexual negotiation and condom use
Establish consistent condom use supply and promotionCondom use policy for sex workers and dancersEstablish norm to carry condom 
Brochures, condom posters on-siteCondom demonstrations
Sustainable supply of condoms on-siteCondom negotiation skills
Condom use cues
Regular condom surveys of patrons
Provide cues and activities to empower persons to adopt safer sexWeekly PLACE staff visitsInteractive games, comedians, dramatic skits, role play 
Risk cards and posters displayed
Promote improved health-seeking behaviour, testing and treatmentClinic referral lists HIV/STI status
‘Know Your HIV Status’ posters displayedPre/post-test counselling
  • • Mobilize site and obtain site gatekeeper support for PLACE intervention;
  • • Establish sustainable condom supply using vending machines or site vendors;
  • • Place posters and ‘condom cues’ to promote condom use;
  • • Strategically place other ‘cues for action’ to reduce partners and get tested;
  • • Establish site programmes with trained influential persons to promote safe sex;
  • • Recruit and encourage disc jockeys to promote safe sex;
  • • Regularly interact with staff, patrons and sex workers/dancers and conduct condom demonstrations and risk reduction counselling;
  • • Conduct condom-carrying surveys to encourage condom-carrying habit;
  • • Recruit and train sex workers/dancers and patrons as peer educators;
  • • Conduct group and individual activities to increase practice of safer sex;
  • • Conduct on-site HIV testing periodically.

Specific intervention activities are described in Table 2. Intervention implementation will proceed in five phases.

Table 2.   Intervention activities
Creating the PLACE environment
PLACE environmentRationaleAction
General HIV prevention postersProvides positive HIV prevention messages and skills with referrals in some instancesPlace in prominent and visible location
‘How to Use Condom’ posterReinforces condom use skillsUse with condom demonstration
Risk assessment poster that is a large version of risk assessment cardEncourage individuals to assess their risk to contracting HIVGo through cards with groups (remember, some patrons may be illiterate)
Risk wheelInteractive self-assessment of risk to contracting HIVLeave in prominent and not so prominent area
‘Condom Sold Here’ signPoint of sale sign for condomsSubmit names to commercial condom distributor
‘101 Ways to Say No to Sex’ posterAbstinence: secondary focus to PLACE interventionGo through chart with individuals (remember, some patrons may be illiterate)
Condom machinesProvide a consistent supply of affordable condomsEnsure condom machines are prominently displayed and stocked
Condom-carrying surveyEncourage men and women visiting site to carry condomsCount 10 consecutive men and women carrying condoms and present giveaways
Condom use policyMaking condom use mandatory at site (if applicable)Solicit agreement/partnership with gatekeeper
Condom demonstrationProvide skills for correct and consistent condom useDemonstrate skills to patrons and encourage patrons to practise condom skills
Incentives and giveawaysEncourage full participation of patrons in PLACE interventionElicit giveaways from neighbouring private entities for sustainable public/private partnership
Looped Video MOH Mass Media CampaignsProvide messages on HIV prevention educationEnsure cassettes and equipment are available to show videos at appropriate times
Trained DJ if applicableProvide HIV/STI prevention messages through musicTrain DJ and provide with appropriate music and sound-bites for events
Rapid HIV Onsite TestingTo promote patrons at site to know their own statusConduct 2 testing sessions at each site during phase 3 and 4 of the intervention
Risk reduction
Group-level interaction
PLACE dialogueRationaleAction
Dramatic skits addressing myths, condom negotiation, other with discussion afterwardsUsing entertainment as a cultural medium for HIV prevention educationPrepare short entertaining skits to promote condom use, partner reduction and in some cases abstinence and addressing condom negotiation and discussion
Risk assessment and negotiation skillsEncourage risk assessment and negotiation skills dialogue at the group level – via skits and discussionEmploy creative strategies to incorporate risk assessment in skits and discussions
Establish self-efficacy plans (SE) for regular patronsRecognize that behaviour change is a process and most people need a plan to assist changeUse the skills taught from SE training to develop SE plans
Use of cultural musicProvides a current and appealing medium to carry HIV prevention messagesUse popular music with safe sex messages
Anatomical modelsDemonstrates insertion of female condoms in anatomically similar vagina and pulling on male condom over anatomically similar penisDemonstrate skills to patrons and encourage patrons to practise condom skills
Training influential personsTo sustain intervention outside of MOHIdentify influential persons for training and frequent consultations. (see Appendix 4)
Training peer educatorsTo sustain intervention outside of MOHIdentify peer educators for more in-depth training. (see Appendix 4)
Referral cardsEffective referral to nearby clinicsLearn names of local clinics to make effective referrals
 Family planning
 AIDS/STDs
 Youth-friendly services
Interactive games for youthInteractive resource for HIV prevention educationDevelop interactive material to engage small groups of patrons
Individual-level interaction
PLACE one-on-oneRationaleAction
Individual counsellingPromote personal engagement on self-assessmentConduct HIV pre- and post-test counselling (see Appendices 3 and 4)
Condom demonstrationTo transfer skills for correct and consistent condom useConduct condom skills for both males and females
Self-efficacy plan (SE)Recognize that behaviour change is a process and most people need a plan to assist changeUse the skills taught from SE training to develop SE plans
Risk assessment cardEncourage individual to assess their risk to contracting HIVGo through cards with groups (remember, some patrons may be illiterate)
Risk wheelInteractive self-assessment of risk to contracting HIVLeave in prominent and not-so-prominent area
Referral cardsEffective referral to nearby clinicsLearn names of local clinics to make effective referrals
 Family planning
 AIDS/STDs
 Youth-friendly services
HIV testingTo know one’s statusFollow HIV testing outreach protocol

Phase 1: Observation

The objective of this phase is to update the profile of each site from the baseline survey, verify the site’s existence, ascertain the busiest times at site, identify gatekeepers and onsite influential persons, and generally observe and note important characteristics of the site. Are condoms visible and/or accessible? Are HIV prevention education materials visible? Are condom distribution agents at or nearby the site? What is the ambiance of the site, does it vary throughout the week? Who are the site gatekeepers and influential persons? Which is the best place to install the condom-vending machine? Output from phase I includes the Field Coordinator’s Preliminary Assessment Report, which describes the strengths and weaknesses of the site, its most popular days and hours, identifies the gatekeepers and influential persons, and proposes a schedule for site visits and activities.

Phase 2: Dialogue and decisions

The objective of phase 2 is to develop strategic intervention partnerships at the site. The MOH staff work to create the PLACE environment; the Field Coordinator seeks to solicit the full partnership of a gatekeeper at each site and begin ongoing sensitization and consultations with 2–3 influential persons at each site. Outreach Officers begin to build the PLACE environment and, where applicable, assist in the installation of condom-vending machines. PLACE staff approach patrons to assess their levels of risk of contracting HIV/STIs. Activities include a presentation to the gatekeeper, identification of gatekeeper concerns, identification of back-up gatekeeper and influential persons at site who can be trained to participate in the intervention, use of reflective listening with gatekeeper and encouragement of ‘change talk’ from gatekeeper about making his site safe. The output of phase 2 is an amended Field Coordinator’s Assessment Report for each site.

Phase 3: Implementation

The objective of phase 3 is to execute a sustainable and strategic intervention for each site and monitor implementation. Outreach Officers and Field Coordinators submit a weekly report on each PLACE site that specifies the activities at environment, group and individual levels and monitors implementation using reports based on process indicator data collected during every site visit. Outreach Officers also establish self-efficacy plans with regular patrons at the site and document achievements in condom distribution, training and HIV testing. The output of phase 3 comprises a defined schedule for visits, onsite condom distribution/demonstration and monitoring of activities at site using process indicators generated from site visit cards.

Phase 4: Maintenance to ensure sustainability of intervention

The objective of phase 4 is to maintain activities at environment, group and individual levels, identify problems, provide training as necessary and improve the site. Outputs of phase 4 include peer educator training, sustained partnership between influential and condom distributor, saturation of condom distribution and successful onsite HIV testing.

Phase 5: Evaluation by Outreach Officers and Field Coordinators

Field Coordinators submit their final report, which includes their own assessment based on discussions with Field Coordinators and review of the site data. The report seeks to assess the successes and challenges of the PLACE intervention model and identify strategies to manage ongoing MOH support to the site.

Formative research and ethical approval

Pilot interventions were conducted for 6 months in four settings to develop intervention approaches and monitoring tools and train outreach workers. No formal evaluation of the pilot interventions was conducted. The University of North Carolina School of Public Health Institutional Review Board and the Jamaican MOH Ethics Committee granted ethical approval for the study.

Principal design of the trial

The study was designed and powered as a cluster randomized trial with two arms. The primary outcome indicator was self-reported recent inconsistent condom use among persons with new and/or multiple sexual partners in the past year. The inclusion criteria for patrons were age 18 years and older, or 15–17 years and not with family or on a family errand. The intervention was implemented for a period of 9 months at each site randomized to the intervention arm. Interviewers and data analysts were blinded at baseline and follow-up about the intervention vs. control status of the site. Outreach workers conducting the intervention were given the names of intervention sites but not of control sites. The Principal Investigator knew the names of the intervention sites and control sites, but was blinded to intervention vs. control status during analysis. There was no placebo intervention. However, an active multi-sectoral national response to HIV including mass media campaigns was ongoing. The extent to which the intervention was implemented has been assessed using monthly summaries of process indicators collected during site visits, by monthly assessments for each site of the extent to which 14 objectives of programme implementation were achieved (e.g. condom availability), by responses at follow-up from site informants about the extent to which intervention activities were implemented at the site, and by responses from individuals socializing at sites at follow-up about their exposure to programme activities. In order to assess patron mixing, patrons were asked to name other sites where they regularly socialize.

Randomization

Geographic clusters of sites identified as public places in Kingston and St Andrew where people meet new sexual partners were the elements of randomization. Identification of 421 study sites considered for study inclusion was based on information obtained from 800 community informants in 2003 (Figure 1). Of these, 287 sites were excluded from the intervention study based on characteristics of the site recorded in 2003 or characteristics of the site apparent during revisits to the site. Sites outside the study area or no longer operational, sites expected to have fewer than 25 persons at peak times, sites with significant political, religious, logistical or safety constraints, including churches, street dances, sport venues, and sites in high crime areas were excluded . Thirteen new sites, named by at least two respondents during baseline assessment, were added. These 13 sites did not have baseline data.

Figure 1.

 Flowchart showing randomization of site clusters to intervention and control arms.

Each of the 147 study sites was categorized into one of three types: (1) bars, nightclubs, hotels; (2) street sites and (3) other public areas where people socialize. Sites were then visited by MOH staff experienced with outreach prevention in Kingston and grouped into 50 clusters based on geographic proximity and the pattern of patron visits. For example, three sites (a bar, a fast food restaurant and a taxi stand) located on the same street corner that shared patrons could comprise one cluster of sites. Clusters varied in size from 1 site to 10 sites and were created to limit mixing across clusters. Each cluster was then categorized based on the type of sites in the cluster. A cluster that contained a bar, a club or a hotel was considered a ‘bar, club, hotel’ cluster; of the remaining clusters, a cluster that contained a street site was considered a ‘street’ cluster; and the remaining sites were considered ‘other’ clusters. Clusters were sorted in SAS first by type of cluster as defined above and within type of cluster by the size of cluster. The first two cluster pairs on the list were randomized – one to the intervention arm and one to the control arm – using a ‘random’ function in SAS; the next two clusters were then randomized, etc., until the last two clusters were randomized. This approach was taken to assure that each arm would contain the same number of clusters and approximately the same number and type of sites. The pairing of clusters was not carried over into analysis. In this set-up, we made a standard assumption that individuals across cluster pairs are not correlated, but outcome measures within a cluster in a pair are correlated. The proposed use of robust variance in the multi-level models was appropriate even if violations in this assumption occurred.

Primary and secondary behavioural outcomes

The primary outcome indicator is a comparison of proportion of patrons in intervention vs. control clusters who report new or concurrent partnerships and recent inconsistent condom use at follow-up. The indicator was constructed using data from interviews with a cross-sectional sample of patrons at intervention and control sites one year after the initiation of the intervention in Kingston. Each person interviewed was categorized into one of seven categories (Table 3) ranging from persons who have never had sex (category 2) to persons with new or multiple partners and recent inconsistent condom use (category 7). Categorization was based on responses to face-to-face interview (FFI) questions asked and recorded by the interviewer and questions asked by the interviewer but answered by the respondent privately using a self-completed (SC) answer sheet. These answer sheets were placed in an envelope at the close of the interview. In case of inconsistency, persons reporting any or new sexual partnerships on SC questionnaires but not on FFI were categorized based on SC questionnaire responses.

Table 3.   Categorization of risk for PLACE intervention
CategoryDescription
1 UncategorizedPersons with incomplete data
2 VirginsPersons who reported never having had sex
3 Not sexually activeNon-virgins who reported no sex in past year
4 Long-term monogamousSexually active persons who reported only one partner in the past 12 months and the relationship began over 12 months ago
5 No sex in past 3 monthsPersons who reported a new or more than one partner in the past 12 months but no sex at all in the past 3 months
6 New or multiple partners and recent consistent condom useAmong persons who reported a new partner or more than one sexual partner in the past 12 months, those who had sex in the past 3 months and reported condom use for each of the most recent three coital acts with each of the most recent two partners and, if the person engaged in commercial sex in the past three months, condom use during last commercial sex
7 Target population: new or multiple partners and recent inconsistent condom useAmong persons who reported a new or more than one sexual partner in the past 12 months, those who had sex in the past 3 months and reported not using a condom at least one time during the most recent three coital acts with each of the most recent two partners or, if the person engaged in commercial sex, no condom use during last commercial sex

Information about whether the person had had sex in the past 3 months and questions to assess consistency in condom use were gathered using SC questionnaires. Consistent condom use was assessed by asking about condom use during the three most recent coital acts with each of the two most recent partners as well as condom use during the most recent sexual contact in the past 3 months with a partner with whom a gift or money was exchanged for sex. During the pilot study, asking about the three most recent partners instead of two most recent partners did not change how people were categorized.

The aim of intervention was to reduce the proportion of persons in category 7 either by reducing the overall proportion of persons with new or multiple partnerships or by increasing consistent condom use among persons with new or multiple partnerships.

The secondary behavioural outcomes include a comparison of the proportions of:

  • • Interviewed patrons (in intervention vs. control clusters) who show a condom when requested by an interviewer;
  • • Interviewed patrons (in intervention vs. control clusters) who report being tested for HIV in the past 6 months and receiving the result; and
  • • Interviewed patrons who are inconsistent condom users (category 7 above) according to their exposure to intervention regardless of whether they were interviewed in an intervention or control cluster (per protocol analysis).

Statistical analysis

The data are hierarchically structured. Individuals are interviewed in sites that have been grouped into clusters. A multi-level model for binary outcome data will be used to assess the effectiveness of intervention. The model to assess the treatment effect, controlling for baseline characteristics if necessary, will be: logit (pjk2) = β0 + β1Gk + β2Skβ3Xjk2 + uk, where Gk designates where the site is in a cluster assigned to the intervention or control arm, Sk is a site-specific covariate, Xjk2 is an individual-specific covariate and uk is the random effect at the site level, which is assumed to be N (0,σuk2).

The regression coefficient β1 will be used to determine the intervention effect. It is possible that many in the control group will visit intervention clusters during the intervention period and contaminated our sample; however, our initial analysis will adopt an ‘intent to treat assumption’.

Sample size and power

The study was designed to be sufficiently powered to detect a programmatic difference in condom use, which can be attributed to the intervention. A sample size of 1300 men and 1600 women was determined to be adequate to estimate a 10% change in recent inconsistent condom use among persons with new or multiple partnerships in the past year (alpha < 0.05 in each of two tails, beta = 0.8, intraclass correlation coefficient among clusters = 0.05). This estimate assumed that each of the two study arms would comprise 25 clusters of sites, that 25% of men and 20% of women have new or concurrent sexual partnerships and are inconsistent condom users at baseline, and that the intervention would have been proved effective if the proportion of inconsistent users dropped to 15% or lower among men and 10% or lower among women. Allocation of the required sample across sites was carried out by distributing the number of interviews proportional to the expected number of persons at the site during busy time. At baseline, the expected number of persons was obtained from the Kingston 2003 survey data and confirmed by visits by two consultants to the sites in 2005. At follow-up, the target number of interviews per site was based on the median number of patrons counted at the site at busy times by interviewers during baseline interviews. This approach was considered more accurate than estimates based on Kingston 2003 site informant data. For analyses of baseline data, the actual counts of persons socializing at the sites at the time of interview were used to weight the data. These counts were also used to set interview targets for follow-up.

Results

Randomization of sites

Characteristics of intervention and control sites were remarkably similar at baseline (Table 4). The only characteristics significantly unbalanced were the geographic area where the site was located and the proportion of sites where female staff meet new sex partners at site (52%vs. 20%). The proportion of sites where sex workers reportedly solicit on-site, however, was balanced (25%vs. 36%). Differences were expected in the geographic distribution of intervention and control sites because of the design that tried to separate intervention from control sites geographically in order to limit mixing between intervention and control sites.

Table 4.   Baseline characteristics of 147 randomized sites included in study at Kingston, Jamaica, 2005
CharacteristicsIntervention groupControl group
  1. *P < 0.001.

Number of clustersn = 25n = 25
 %%
Number of sites in cluster
 136.036.0
 220.024.0
 3+44.040.0
Type and number of randomized sitesn = 75n = 72
 Clubs/bars/motels20.018.0
 Street sites16.022.2
 Socializing64.059.8
 n = 75n = 72
Location*%%
 Downtown Kingston12.04.2
 New Kingston13.334.7
 ‘Half Way Tree’ area30.716.7
 Cross Roads14.78.3
 UWI/UTECH5.315.3
 Liguanea13.35.6
 Greater Portmore0.011.1
 Hellshire Beach1.31.4
 Port Henderson9.32.8
Activities at the siten = 68n = 66
 In operation more than 2 years 88.293.9
 Patrons include students89.789.4
 Beer or hard alcohol consumed92.693.9
 Ganja smoked57.459.1
 TV or video watched on-site39.739.4
 Explicit sex videos on-site7.43.0
 Individual/group or Go-Go dancing23.516.7
 Music on-site67.648.5
 People waiting for bus or taxi60.359.1
 People coming to buy fast food47.148.5
 People buying from street vendor57.451.5
 People meeting new sex partners on-site77.981.8
 Facilitator on-site for meeting partners13.212.1
 Female sex workers solicit on-site25.036.4
 Female staff meet new sex partners 51.5*19.7*
 Gay men meet partners44.139.4
 Condoms seen on-site day of interview36.830.3
 Condoms never available past year26.530.3
 Condoms always available past year50.048.5
 Interviewer saw condoms on-site79.487.9

Characteristics of patrons at sites

Statistical comparison of the characteristics of persons at intervention vs. control sites used weighted data to adjust for differences between the expected number of patrons at a site and the actual number counted at a site during the interview period and took into account clustering by site. Characteristics of patrons were similar at intervention and control sites (Table 5). All seven levels of the outcome variable were balanced. Approximately 30% of men and 25% of women at intervention and control sites reported a new partner or more than one partner in the past year as well as recent inconsistent condom use. Approximately 5% of persons reported conflicting information about having ever had sex or having a recent new partnership in FFI as compared with SC questionnaires.

Table 5.   Baseline characteristics of patrons interviewed at intervention and control sites at Kingston, Jamaica 2005
Socio-demographic characteristics of patrons of sitesMaleFemale
Intervention groupControl groupIntervention groupControl group
  1. †Weighted N. The other n shown is the actual n.

  2. *Difference between intervention and control groups P < 0.05.

Number of respondentsn = 711, N = 1243†n = 654, N = 1127†n = 845, N = 1314†n = 738, N = 983†
Mean age (years)28.128.425.525.5
Age 15–1915.215.022.822.8
Age 20–2424.024.131.630.2
Age 25–2920.118.518.620.9
Age 30–3924.423.620.419.1
Age 40+16.218.86.67.0
Less than 10 years education9.49.46.94.8
Currently a student27.929.839.544.2
Unemployed26.730.040.042.8
Resident of Kingston93.395.393.594.8
Site visiting behaviour
 Attend site weekly or more67.973.371.6*58.9*
 Most frequently socializes at another site in Kingston49.248.143.646.8
 Came to site to drink alcohol18.222.47.2*12.3*
 Ever met a new partner at this site22.120.48.16.6
 Has or intends to visit more sites tonight46.142.738.040.3
 Came to site alone64.059.048.945.6
Showed a condom to interviewer25.921.75.76.2
Sexual behavior and condom use level
 1. Uncategorized0.21.00.31.0
 2. Virgins2.32.611.012.1
 3. Not sexually active in past 12 months5.95.87.56.5
 4. Long-term monogamous20.122.135.132.5
 5. No sex in past 3 months5.86.82.43.5
 6. New/multiple partners in past 12 months but consistent condom use31.431.217.219.5
 7. New/multiple partners in past 12 months but inconsistent condom use34.430.526.624.8
 Total100.0100.0100.0100.0
STD symptoms and HIV testing
 STD symptoms in past 3 months2.32.86.55.8
 Been tested for HIV41.940.244.443.1
Sexual partnerships
 Currently has a live-in sexual partner31.929.732.932.0
 In past 3 months has given or received money from partner for expenses30.4*21.6*31.0*20.8*
 In past 3 months has given or received cash money from partner for sex11.09.55.35.0
 In past 12 months has had sex with someone not living with or married to64.059.846.244.2
 In past 12 months has had sex with both men and women1.60.84.33.5
 In past 12 months has had gay/lesbian sex4.23.67.78.2
General exposure to AIDS prevention in past 3 months
 Attended HIV/AIDS educational session21.020.329.029.3
 Seen an HIV/AIDS film or video39.240.246.045.9
 Heard HIV/AIDS programme on the radio80.583.583.182.8
 Seen an HIV/AIDS prevention poster80.283.882.786.8
 Seen an HIV/AIDS programme on TV83.882.688.185.0
 Been counselled on HIV by health worker23.821.533.831.9
Exposure to prevention programmes at site
 Talked with someone about HIV/AIDS19.0*12.9*16.118.4
 Attended AIDS education session5.2*8.0*5.7*8.9*
 Obtained a condom at the site19.920.712.413.8
 Saw poster at site17.417.017.219.3
 Had an HIV test at site2.02.72.94.5

Although there was a difference in the proportion giving or receiving money for expenses from a sexual partner (transactional sex) among male and female patrons, there were no significant differences in the proportions reporting receiving cash money for sex (commercial sex). Approximately 20% of men and 7% of women reported having ever met a partner at the site. Intervention and control groups both reported that they frequently socialize at other sites. Over 20% of men, as compared with approximately 5% of women, showed a condom to the interviewer.

Exposure to general prevention programmes was high. More than 80% of persons reported exposure in the past 3 months to HIV radio/TV programmes or posters. More than 40% reported having been tested for HIV. However, fewer than 20% reported exposure to any on-site AIDS prevention activity and fewer than 5% reported being tested for HIV at the site. Approximately 20% of men and 13% of women reported obtaining a condom at the site.

Discussion

Appropriate target population for intervention

A key component of any prevention strategy is defining the target population. Clinic-based approaches that target symptomatic persons seeking treatment may document success in reducing the incidence of STI among these persons; however, asymptomatic carriers of infection or persons disinclined to seek treatment are often missed. The advantage of taking prevention programmes to public places where people go to socialize and meet new sexual partners is that persons who are at risk of acquiring and transmitting infection but who may be asymptomatic can be reached with prevention messages. Baseline findings reveal that many patrons socializing at these sites (approximately 30% of men and 25% of women) have had recent new or multiple sexual partnerships and do not use condoms consistently. Thus, baseline findings confirm that the population is a reasonable group to target for HIV prevention.

Assessment of randomization

Randomization resulted in two similar groups. Two variables not balanced at baseline were the proportion of sites where female staff have sex with patrons as reported by the site informant and the proportion of women who report sex in exchange for money for expenses. These variables are related and suggest that randomization resulted in more transactional sex at intervention sites than control sites. These variables will be assessed at follow-up, and a sensitivity analysis would shed light on whether adjusting for this factor may affect the interpretation of the findings.

Strengths and limitations of outcome indicator

An ideal outcome indicator for HIV prevention intervention is HIV incidence. However, HIV incidence was not a viable outcome indicator because the incidence of HIV is low in Jamaica, and a prohibitively large number of persons would need to be recruited to the study to have sufficient power to detect a difference between intervention and control arms. Incidence studies generally require a prospective cohort, and it would be extremely difficult to identify, recruit, enroll and follow site-based cohorts of patrons. Other biomedical indicators were considered, including testing for other STIs, but cost and logistical difficulties were a challenge.

Self-reported condom use has some significant limitations as an outcome indicator (Cowan & Plummer 2003), including its suspected lack of validity due to self-presentation bias (Catania et al. 1990; Des Jarlais & Semaan 2002) and its lack of direct correlation with STD endpoints (Minnis & Padian 2001; Aral 2004; Warner et al. 2006). Because condom use has been actively promoted in Jamaica for years and condoms are readily available at health clinics, pharmacies and a variety of non-traditional outlets, the authors explored strategies for reducing over-reporting of condom use. Focus-group discussions were held with men and women on the preferred mode of providing responses to questions about condom use and three versions of a questionnaire using combinations of FFI and SC questionnaires were pre-tested. Focus-group discussions and pre-test results suggested that persons were more likely to over-report condom use in FFI than when using SC questionnaires. These results were consistent with other research, indicating that alternatives to FFIs often yield higher and more plausible estimates of sexual risk (Gregson et al. 2002; Cleland et al. 2004). Computer-assisted methods were gauged not feasible because of interviewer reluctance to carry the technology into interviewing areas. The selected outcome measure, based on SC questionnaires that asked about condom use during the three most recent acts with the two most recent partners, revealed the highest level of inconsistent condom use in the pilot study. There is some evidence that this measure still over-estimates condom use as few patrons who claimed to consistently use condoms could show the interviewer a condom at the time of their on-site interview.

The determination of the effectiveness of the intervention will be based on an intent-to-treat analysis of the extent to which inconsistent condom use was lower in intervention vs. control communities. However, interpretation of the effectiveness of the intervention will require examining (1) the extent to which patrons mix across intervention and control clusters; (2) whether any difference between intervention and control sites in the proportion of female staff having sex with patrons at follow-up can be attributed to baseline differences; and (3) the extent to which the intervention was implemented in intervention sites. Based on the findings of the intervention study, the MOH will determine whether and how to extend the Kingston PLACE intervention to other cities in Jamaica.

Ancillary