Sandøyetal.BMCPublicHealth2012,12:10 http://www.biomedcentral.com/1471-2458/12/10 RESEARCH ARTICLE Open Access Targeting condom distribution at high risk places increases condom utilization-evidence from an intervention study in Livingstone, Zambia Ingvild Fossgard Sandøy1*, Cosmas Zyaambo2, Charles Michelo2 and Knut Fylkesnes1 Abstract Background: ThePLACE-methodpresumesthattargetingHIVpreventiveactivitiesathighriskplacesiseffectivein settingswithmajorepidemics.Livingstone,Zambia,hasamajorHIVepidemicdespitemanypreventiveeffortsinthe city.Abaselinesurveyconductedin2005inplaceswherepeoplemeetnewsexualpartnersfoundhighpartner turnoverandunprotectedsextobecommonamongguests.Inaddition,thereweremajorgapsinon-sitecondom availability.Thisstudyaimedtoassesstheimpactofacondomdistributionandpeereducationinterventiontargeting placeswherepeoplemeetnewsexualpartnersoncondomuseandsexualrisktakingamongpeoplesocializingthere. Methods: The 2005 baseline survey assessed the presence of HIV preventive activities and sexual risk taking in places where people meet new sexual partners in Livingstone. One township was selected for a non-randomised intervention study on condom distribution and peer education in high risk venues in 2009. The presence of HIV preventive activities in the venues during the intervention was monitored by an external person. The intervention was evaluated after one year with a follow-up survey in the intervention township and a comparison township. In addition, qualitative interviews and focus group discussions were conducted. Results: Young people between 17-32 years of age were recruited as peer educators, and 40% were females. Out of 72 persons trained before the intervention, 38 quit, and another 11 had to be recruited. The percentage of venues where condoms were reported to always be available at least doubled in both townships, but was significantly higher in the intervention vs. the control venues in both surveys (84% vs. 33% in the follow-up). There was a reduction in reported sexual risk taking among guests socializing in the venues in both areas, but reporting of recent condom use increased more among people interviewed in the intervention (57% to 84%) than in the control community (55% to 68%). Conclusions: It is likely that the substantial increase in reported condom use in the intervention venues was partially due to the condom distribution and peer education intervention targeting these places. However, substantial changes were observed also in the comparison community over the five year period, and this indicates that major changes had occurred in overall risk taking among people socializing in venues where people meet new sexual partners in Livingstone. Trial registration: ClinicalTrials.gov NCT01423357. Background generalised or concentrated, major modes of transmis- In order to design and implement effective HIV preven- sion, groups with the highest incidence and sexual risk tion interventions, it is essential to have a broad under- taking in different subpopulations [1,2]. One tool in the standing of the epidemiological context in a community. assessment of specific epidemiological contexts is the This includes determining whether the HIV epidemic is “Priorities for Local AIDS Control Efforts” (PLACE)- method, an approach to rapidly identify places with a high risk of HIV transmission. The underlying assump- *Correspondence:[email protected] 1CentreforInternationalHealth,UniversityofBergen,Bergen,Norway tions of the method are that risky behaviours often take Fulllistofauthorinformationisavailableattheendofthearticle ©2011Sandoyetal;licenseeBioMedCentralLtd.ThisarticleispublishedunderlicensetoBioMedCentralLtd.ThisisanOpenAccess articledistributedunderthetermsoftheCreativeCommonsAttributionLicense(http://creativecommons.org/licenses/by/2.0),which permitsunrestricteduse,distribution,andreproductioninanymedium,providedtheoriginalworkisproperlycited. Sandøyetal.BMCPublicHealth2012,12:10 Page2of12 http://www.biomedcentral.com/1471-2458/12/10 place in venues which are publicly available, and that Baseline survey targeting HIV preventive activities at such places is A PLACE-survey was conducted in Livingstone in 2005 likely to be more effective than targeting interventions to identify high risk places [10]. During the first phase just at perceived high risk groups in settings with high of the survey the interviewers walked through the streets HIV prevalence. In countries where HIV is primarily of Maramba, Dambwa and the city centre and asked taxi heterosexually transmitted, high risk places are defined drivers, bar workers, shop staff, health personnel, and as places where people meet new sexual partners [3]. young people to name places where people met new There are sharp geographical differentials in HIV pre- sexual partners. The second phase consisted of locating valence trends in Zambia. Data from pregnant women all the mentioned venues. In the places that were found indicate overall national declines among urban and rural and in operation, the owner, a bar worker or a regular participants since the mid 1990s. However, in certain guest was interviewed about activities taking place in antenatal surveillance sites, HIV prevalence has been the venue and the availability of condoms, posters and stable. In Livingstone in Southern province the HIV pre- leaflets relating to HIV prevention. During the third valence was stable among young women around 30% phase of the study, all the venues mentioned by more during the period 1994-2002 [4] despite high invest- than 10 informants in the first phase and 1/3 of the ments in HIV prevention since the early 1990s. In 2005 remaining sites were selected (with probability of selec- a PLACE-survey was conducted in Livingstone in order tion proportional to the estimated number of guests to assess the need for and presence of HIV preventive during busy hours). Individuals socializing in these interventions targeting high risk places. This survey venues and who were standing along two imaginary found that 43% of the places where people met new sex- diagonal lines connecting the four corners of the room ual partners never had condoms available. People socia- were approached for an interview about sexual beha- lizing in the venues reported multiple partners in the viour. One nurse, one counsellor and 1-2 peer educators previous month, but less than half reported using con- from health clinics in Maramba and Dambwa and staff doms consistently with new partners. Reported condom of NGOs involved in HIV prevention activities in use was higher in places where condoms were always Livingstone were interviewed about HIV preventive pro- available, and this is in line with findings from other grams which existed in the city. studies [5-9]. Many of the respondents expressed a wish to discuss HIV with health personnel or peer educators, Intervention but they were afraid to go to the clinic since people Youth peer educators who were working at the health might suspect that they were infected [10]. clinics in Livingstone or had been involved in other peer A number of peer educators had previously been education activities were invited to participate in the trained in Livingstone, but most of them had dropped intervention study. The only selection criteria were will- out due to lack of incentives [10]. Since the survey indi- ingness to visit venues where people meet sexual part- cated a high potential for improvement of condom avail- ners several times per month for a small compensation ability, we intended to test whether distributing and ability to communicate with strangers about HIV, condoms to places where people met sexual partners sex and condoms. A two-day training was held in Feb- could increase condom use among those socializing in ruary 2008 for 48 youths. A second training/retraining these places. We also aimed to assess the operational which included 24 new youths was conducted in August feasibility of engaging peer educators in condom distri- of the same year. The training was based on materials bution and behaviour change discussions. and teaching methods developed by the WHO, the Min- istry of Health and different NGOs, but was adopted to Methods the specific context, i.e. focusing on condom demonstra- Description of study sites tion and communication with drunk people. To com- Livingstone is situated close to the Zimbabwean bor- pensate for drop-outs, 11 young persons were recruited der, and had a population size of 142,000 in 2010. in the course of the intervention and given one-to-one Maramba is a high density township and many of its training from the two local supervisors who were adults 39,000 inhabitants fetch drinking water from commu- with extensive experience with coordinating youth peer nal taps and access communal toilets. More HIV pre- educator activities. Monthly meetings were held between ventive projects have been organized in this township the peer educators and the local supervisors. The super- than in other parts of Livingstone because it has been visors recorded the reasons given for withdrawing when perceived to be a particularly high risk area for HIV. peer educators dropped out. Dambwa is a medium density township with some Maramba was selected as the intervention community smaller high density parts and a total population size because most of the active peer educators came from of 50,000 in 2010 [11]. this township. The list of venues from 2005 was Sandøyetal.BMCPublicHealth2012,12:10 Page3of12 http://www.biomedcentral.com/1471-2458/12/10 continually updated during the intervention year as new the two townships it functioned as a geographical corri- places were established and others closed down. The dor separating the two communities. In Maramba the owners and staff of the listed venues were requested to list of venues included in the intervention was utilized, allow peer educators to bring condoms to the venues whereas in Dambwa, phase one was repeated to make and to talk to guests about HIV-related issues. The an updated list of venues. In the follow-up survey, all youth peer educators were given responsibility to distri- the listed venues were visited in both the second and bute condoms and to check the condom stock twice a third phases of the survey. The same questionnaires as week in 1-2 venues each. They were told that it was in the baseline were used in addition to some new ques- very important that there were always enough condoms tions on experiences relating to HIV prevention inter- available in the venues. Boxes of condoms were to be ventions in the venues. put in a suitable place, but keeping in mind that people In addition, eight peer educators, the two local super- were likely to prefer picking up condoms in a private visors and the external monitor were interviewed in- spot. No specific guide was given concerning what time depth (IDI) using a semi-structured interview guide. of the day the visits should be conducted. During visits Two focus group discussions (FGDs) with peer educa- to the venues the peer educators were supposed to tors (one with 8 females and one with 9 males) were approach the guests about HIV-related issues and to also conducted based on a semi-structured discussion demonstrate condom use. According to the initial plan guide. In this paper we focus on the sections of the they would also put up HIV-related posters and distri- interviews/discussions relating to experiences with the bute leaflets, but this ended up being done irregularly intervention itself. The peer educators motivation and due to a limited budget to produce such materials. Since reasons for dropping out will be the focus of a separate some of the peer educators did not live within Mar- paper. Brief IDIs were also conducted with nine bar amba, a transport allowance equivalent to 4 USD was attendants and nine male patrons in selected interven- given for each of the two weekly visits. The peer educa- tion venues touching upon HIV prevention available in tors reported their weekly activities, including number the venue. All the interviews and discussions were tran- of visits and time spent in the venue, number of con- scribed verbatim. doms, posters and leaflets distributed, and number of persons talked to, to the local supervisors in a form Analyses called “Peer educator diary”. The quantitative data was entered in Epidata. The statis- The intervention ran from the 1st February 2009 to tical analyses were conducted primarily using SPSS ver- the 31st March 2010. Condoms were obtained from the sion 15, but StataIC 10 was also utilized. The data Livingstone District Health Management Team collected in the city centre in the baseline survey was (DHMT), Livingstone General Hospital and Planned not included in the analysis of this paper. The analyses Parenthood Association Zambia during the first 5 of data from individuals socializing in the venues were months of the intervention, and from the Medical Stores adjusted for the effects of clustering (venues were Limited during the remaining period. Excluding the regarded as clusters) and weighted to compensate for costs of the condoms, the intervention cost 63,400 USD, differential probability of being selected in popular ver- of which 18,796 USD covered travel and accommoda- sus less popular venues. Although the age of respon- tion expenses of the researchers. dents and number of partners were not normally An independent person was engaged as an external distributed, we used the mean and its SE as the main monitor to assess the presence of condoms, HIV-related measure of average since we could not compare med- posters and leaflets and to find out from staff whether ians while including frequency weights. However, the peer educators had engaged in behaviour change discus- independent samples t-test used to compare differences sions. She visited all the venues that were targeted by in means between subgroups could not be adjusted for the intervention on average three times during the inter- the effect of clustering. Differences in percentages vention year. The identity of the monitor was kept between Maramba and Dambwa and between the base- secret to the peer educators and her schedule was line and follow-up surveys were assessed with the Pear- unknown to the peer educators, the local supervisors son chi-square test of independence. The significance and the research team. The local supervisors discussed level was set at 5%. To test whether observed differences shortcomings that were highlighted in her reports with were due to differences in the types of venues included the peer educators at the following monthly meeting. in phase 2 or to differences in types of venues, and in age and gender of respondents included in phase 3, Evaluation adjusted logistic (for binary outcomes) and ordinal (for In March 2010, a new survey was conducted in Mar- ordinal dependent variables with more than two amba and Dambwa. Since the city centre is in-between response-categories) regression analyses were conducted Sandøyetal.BMCPublicHealth2012,12:10 Page4of12 http://www.biomedcentral.com/1471-2458/12/10 with township or year as the main independent variable. In the baseline survey in 2005, 434 persons were asked The age-adjustment was done with a categorical vari- to name places where people met new sexual partners. able. We comment on the results of the adjusted regres- Among those mentioned, 55 were located in the inter- sion analyses only when the adjustments changed the vention community and 71 in the control community. association between the dependent and the main inde- The majority of venues were sherbeens (informal drink- pendent variable from significant to non-significant or ing places serving alcohol without a license). The rest the other way around. The analyses were done both were night clubs, hotels and guest houses. During the stratified by gender and pooled. The Pearson correlation follow-up survey in 2010, 130 people in the control coefficient was used to assess the relationship between community were asked to name venues where people the reported frequencies of visits by peer educators ver- met new sexual partners during phase one, and 53 sus venue representatives. places were mentioned. All of these were visited in The qualitative analysis was guided by the framework phase two. Twelve of the intervention venues were approach [12]. This approach included reading through closed permanently or temporarily at the time of the fol- all the interviews and discussions and labelling the sec- low-up survey. The majority of the venues in both com- tions relating to experiences with the PLACE interven- munities in 2010 were bars (Table 1). In the third tion according to predetermined subthemes: frequency phase, 190 and 339 individuals socializing in 25 and 34 and timing of visits, description of typical visits, experi- selected venues in the intervention and the control com- ences with condom distribution, experiences with poster munity were approached for an interview in the baseline distribution, and interactions with guests. The data was survey, whereas the corresponding numbers were 264 grouped and sorted according to the subthemes in a and 273 individuals in 43 and 50 venues, respectively, in chart, where each case or FGD formed a separate row. the follow-up survey. A few refusals were recorded in The chart was used to map the range of experiences the baseline (unweighted 3.8%), whereas no refusals with the intervention. were recorded in the follow-up survey (Table 2). Ethical aspects The intervention process Only adults aged 18 years and above were interviewed. On average, each venue was visited 1.6 times per week Oral consent was required of all participants. The inter- by peer educators, and they reported speaking to a med- views were anonymous, and the informants were ian of two persons about health and HIV-related issues assured that the information given would not be linked during a visit. Most visits were conducted on weekdays to the site or to them. The interviews with individuals between 0900 and 1100 hours. Only 9% of the visits socializing in the venues were conducted in a private were conducted after 1700 and less than 5% on Satur- corner of the venue or outside, depending on what the days or Sundays. However, according to the venue respondents preferred. Both the baseline and the inter- representatives, the busiest hours in the venues were vention protocols were approved by the Research Ethics typically between 1700 and 2200 hours, and Friday, Committee of the University of Zambia. Saturday and Sunday were the busiest days. A typical reason given by the peer educators in the in-depth inter- Results views for the visiting hours was that they had other Participation engagements (e.g. work) later in the day, and for those Fifty eight venues in the intervention community were who lived outside the intervention community evening includedatsomepointduringtheintervention,butsince visits were more difficult due to security concerns and some closed and new places opened, the number of less transport being available. In the interviews most of venuesincludedconcurrently variedbetween45and49. the peer educators reported placing the condom boxes One of the venues was dropped from the intervention on the counter. Some also mentioned that condoms during the first month of the intervention because the were placed in the toilets. Bar attendants in two of the ownerdecidednottoparticipateduetoreligiousreasons. guest houses reported that they provided them in the The age of the peer educators ranged between 17-32, rooms or gave them directly to women who were selling with 22 being the average. Forty percent of the peer sex. Both the peer educators and the local supervisors educators were females. Out of the 72 peer educators reported that some bar attendants did not display the trained before the intervention, 15 withdrew before it condoms, possibly because they were selling socially started and 23 dropped out at some point during the marketed condoms. intervention. The most common reasons given were The peer educators reported that many of the guests employment (seven), attending school (five), moving out were eager to learn about condoms and that the guests of town (nine), and being discouraged by late or low usually took the initiative to the discussions. Although allowances (nine). some people in the beginning accused the peer Sandøyetal.BMCPublicHealth2012,12:10 Page5of12 http://www.biomedcentral.com/1471-2458/12/10 Table 1Characteristics ofandactivities in venues where peoplemeetnew sexual partners Baseline Follow-up Intervention Control Intervention Control % N % N p-value % N % N p-value Placeverification Found 74 55 75 71 0.388 74 58 100 53 0.002 Found,refusedinterview 2 0 0 0 Notfound 14 11 0 0 Closedtemporarily 0 0 7 0 Closedpermanently 2 0 14 0 Notvisited 7 14 5 0 Typeofplace Sherbeen 55 42 60 53 0.439 26 43 15 52 0.042 Bar/restaurant 36 36 60 83 Nightclub 5 4 5 2 Hotel/guesthouse 5 0 9 0 Genderrespondent Male 74 42 62 53 0.233 37 43 57 53 0.059 Female 26 38 63 43 Positionofrespondent Staff 57 42 54 52 0.749 74 43 87 53 0.122 Patron 43 46 26 13 Refusedinterview 2 42 0 53 0.259 0 43 0 53 - Beerdrinking 95 41 100 53 0.104 98 43 100 53 0.264 Spiritsdrinking 71 41 62 53 0.305 58 38 76 50 0.071 Dancing 56 41 45 53 0.298 93 42 94 48 0.865 Menmeetnewfemalesexualpartnershere 78 41 79 53 0.888 100 43 92 52 0.063 Womenmeetnewmalesexualpartnershere 78 40 79 53 0.839 98 43 90 52 0.146 Menmeetnewmalesexualpartnershere 0 41 0 52 - 21 43 15 52 0.483 Womencometosellsex 78 41 76 53 0.770 93 43 73 52 0.012 educators of promoting promiscuous behaviour, there it’s nearer and he’ll like boost their business."(Local were less critical voices as time went on and the peer supervisor, female nurse aged 39.) educators explained the purpose of the intervention. However, seven out of the eight peer educators who “This time they’ve accepted the study. So, we’re wel- were interviewed stated that it was difficult to talk to come to each and everyone. Any person you find in older people about condom use. the venue say, “Uh, madam after you - when you fin- ish talking to the bar-man, you come here. Me, it’s “In some cultures it is not allowed for a youth like me. You’re going to start with me. There’s some- me to be talking to someone older about sex and thing I want to find out more from you”. So it’s like condoms. Not just - it’s like you are disrespecting they - they accepted us.” (Female peer educator aged them. (...)” (Female peer educator aged 23, IDI 3) 32, IDI 2) In some bars there were complaints from the bar The peer educators reported meeting a lot of people attendants that the peer educators only left the condoms who expressed appreciation for the free condom distri- and did not make an effort to start behaviour change bution. The local supervisor also perceived that the bar discussions. The external monitor and the peer educa- owners were happy with the intervention. tors were of the impression that this occurred more often with peer educators who joined the project late “Some of the bar owners - those ones whereby now and had not received the initial training. At the same we’ve felt that relationship when they meet you- time the high turnover of staff in the venues was a chal- they would talk about it to say it is really helping lenge mentioned by the peer educators since it meant them. Because you’d find that - that time maybe that new bar attendants had to be sensitized to under- when the customer needs to run out from the bar to stand the idea behind the intervention and why con- go and look for a condom, he wouldn’t go because doms should be clearly displayed. Sandøyetal.BMCPublicHealth2012,12:10 Page6of12 http://www.biomedcentral.com/1471-2458/12/10 Table 2Sociodemographic characteristics ofindividuals socializingin venues where people meetnew sexual partners Men Baseline Follow-up Intervention Control Intervention Control % N % N p-value % N % N p-value Refusals 3.3 155 1.3 261 0.172 0 199 0 222 - Single 30 198 45 221 0.003 Married/cohabiting 61 50 Divorced 8 2 Widowed 1 3 Fromtownship 82 148 63 254 0.019 88 199 91 220 0.392 Anothertownship 16 36 8 8 Fromoutoftown 1 2 4 1 Est 95%CI Est 95%CI p-value Est 95%CI Est 95%CI p-value Medianage 29 30 31 30 Meanage 28.6 27.8-29.3 30.8 29.3-32.3 <0.001 32.4 30.8-34.1 31.0 29.7-32.2 <0.001 Mediannoschoolyears 12 12 11 12 Meannoschoolyears 11.2 10.9-11.6 11.0 10.2-11.7 0.025 10.4 9.9-11.0 11.6 11.1-12.1 <0.001 Women Baseline Follow-up Intervention Control Intervention Control % N % N p-value % N % N p-value Refusals 2.9 33 3.1 77 0.953 0 65 0 51 - Single 41 65 56 51 0.264 Married/cohabiting 16 24 Divorced 22 17 Widowed 20 4 Fromtownship 83 31 76 70 0.537 96 65 94 51 0.796 Anothertownship 17 24 3 5 Fromoutoftown 0 0 1 1 Est. 95%CI Est. 95%CI p-value Est. 95%CI Est. 95%CI p-value Medianage 24 24 30 28 Meanage 24.0 22.5-25.5 25.6 23.7-27.6 0.002 29.8 27.1-32.5 27.5 25.2-29.8 <0.001 Mediannoschoolyears 10 9 9 9 Meannoschoolyears 10.4 9.7-11.0 9.1 8.4-9.8 <0.001 9.1 8.5-9.6 10.1 8.7-11.6 <0.001 Monitoring increased sharply in the intervention community The external monitor carried out the first monitoring between 2005 and 2010 (22% to 91%, p < 0.001), but round in May 2009 and found that 26% (10 out of 39) there was only a slight increase in the control commu- of the visited venues did not have condoms available. In nity (21% to 30%, p = 0.280). There was a small, but sig- the next two monitoring rounds this was down to 8% (3 nificant, increase in the proportion of venues in the out of 36 visited venues in June and 2 out of 24 in July), control community where peer education had taken and then increased to 11% (2 out of 18 visited venues) place (0 to 8%; p = 0.036), and there was a clear in August and October and 12% (2 out of 16 visited increase in condom distribution and peer education in venues) in the December round. Posters were observed the intervention community. However, the frequency of in 23% of the venues in May, 53% in June, 62% in July, visits reported by peer educators themselves did only 28% in August, 6% in October and 12% in December. weakly correlate with the frequency reported by venue representatives in the intervention community (Pearson Impact correlation 0.240; p = 0.131). In both the intervention The percentage of venue representatives who reported and control communities there was an increase in the that HIV prevention activities had ever taken place proportion of venues where condoms were reported to Sandøyetal.BMCPublicHealth2012,12:10 Page7of12 http://www.biomedcentral.com/1471-2458/12/10 be always available, but the increase in the control com- partner from that venue had obtained it in the same munity was not significant when adjusted for the type of venue, whereas this was the case for less than a quarter sites included. In the follow-up survey, HIV-related leaf- in the control community (p < 0.001). This condom was lets and posters and condom distribution were all more more likely to have been free in the intervention than in likely to be observed by the interviewers in the interven- the control venues (reported by 72% vs. 25%). Respon- tion than the control community. All the venues in the dents in the intervention community were also more intervention community provided condoms free of likely to report condom use with the most recent new charge to the guests in 2010 whereas only a quarter in partner in 2010 compared to the control community the control community did the same. The free condoms (Figure 1) and compared to the baseline (p < 0.001). were reported to be provided by the district or NGOs in However, there was a considerable increase in reported both townships (Table 3). condom use with the previous new partner in the con- Just over half the respondents in the baseline survey trol community too (p = 0.054). The percentage stating reported having used a condom with the previous part- that they did not use a condom because they did not ner from the venue where they were interviewed, and have one at hand dropped in the intervention commu- this increased significantly to 82% in the intervention nity (p = 0.041), whereas there was no significant change community (p = 0.001) and non-significantly to 68% in in the control community. The proportion of respon- the control community (p = 0.118). In 2010, 80% of the dents who believed that condoms were very or some- respondents in the intervention community who what effective in preventing STIs and HIV increased in reported having used a condom with the previous both communities (p = 0.001 in the intervention and p Table 3HIVprevention activities invenues where people meetnew sexual partners Baseline Follow-up Intervention Control Intervention Control % N % N p- % N % N p- value value EverHIVpreventionactivitiesinvenue 22 41 21 42 0.888 91 43 30 50 < 0.001 EverHIV-relatedlectures/seminarsinvenue 2 41 2 53 0.854 14 43 6 50 0.196 EverHIV-relatedpamphlets/leafletsinvenue 2 41 0 53 0.253 19 43 2 50 0.007 EverHIV-relatedpostersinvenue 15 41 15 53 0.950 26 43 4 50 0.003 Evercondomdistributioninvenue 10 41 21 53 0.149 84 43 20 50 < 0.001 Everpeereducationinvenue 0 41 0 53 - 46 43 8 50 < 0.001 Howoftencondomsavailable Always 32 41 17 52 0.036 84 43 33 51 < 0.001 Sometimes 32 19 16 35 Never 37 64 0 31 Wheredoyouobtaincondomsthatavailabletopeoplewho Buythem 0 43 58 43 < cometothisvenue? 0.001 Obtainfrom 95 14 NGO Obtainfrom 2 26 district Other 2 2 Condomfreeofchargetoguests 100 43 24 42 < 0.001 Condomsattimeofvisit 49 41 24 53 0.015 86 43 56 52 0.001 Ifyes,CanIseeone? 100 19 100 13 - 97 37 100 28 0.381 Postersobservedbyinterviewers 29 41 21 53 0.341 15 41 2 53 0.020 Leafletsobservedbyinterviewers 0 41 0 53 - 2 41 0 53 0.253 Willingnessamongstafftodistributefreecondoms 100 31 98 43 0.393 Willingnessamongstafftosellcondoms? 88 24 70 27 0.138 Sandøyetal.BMCPublicHealth2012,12:10 Page8of12 http://www.biomedcentral.com/1471-2458/12/10 Figure1Proportionofrespondentsreportinghavingusedacondomwiththepreviousnewpartner. < 0.001 in the control area). In the follow-up survey the intervention community was partially due to the respondents in the intervention community were more condom distribution and peer education intervention in likely to report having experienced HIV preventive these places. activities in the venue where they were interviewed and Based on previous research there is mixed evidence to having discussed HIV prevention with a peer educator back up the hypothesis that targeting HIV preventive in the previous 6 months than respondents in the con- activities at high risk places has an impact on the beha- trol community (Table 4). However, there was no asso- viour of people socializing there. A randomised con- ciation between reported condom use and having talked trolled trial in Nicaragua, found that providing condoms to a peer educator (results not shown). in motels lead to increased condom utilisation among In both communities respondents reported less new guests, but the presence of leaflets and posters promot- partners and less partners overall in the previous month ing condoms did not [5]. PLACE-surveys in East Lon- in the follow-up than in the baseline survey (Table 5), don, South Africa, found increased reporting of condom and there was a decline in the proportion of men and use and a reduction in multiple partnerships among women engaging in transactional sex in the previous 3 guests socializing in venues where people met new sex- months (85% to 65% of women (p = 0.050) and 87% to ual partners over a three year period, and this could 43% of men (p < 0.001) in the intervention community). possibly be attributed to a behaviour-change interven- Women who admitted exchanging sex for money, were tion targeting these venues [13]. However, a randomised more likely to report having used a condom the pre- controlled trial conducted in Kingston, Jamaica, where vious time in the follow-up than in the baseline survey venues where people met new sexual partners were ran- (the increase was only significant in the intervention domised to a site-based intervention (including on-site community: 54% to 80%; p = 0.043). HIV testing, condom promotion, and peer education), found no significant differences in reporting of number Discussion of partners or consistent condom use between guests in The comparison of the two PLACE-surveys conducted intervention and control venues. However, there were in two townships in Livingstone in 2005 and 2010 indi- several factors that possibly could explain the lack of cated clear improvements over the five year period in impact: implementation difficulties (condoms and edu- condom availability and outreach of peer education cational materials not always being available in the activities in venues where people meet new sexual part- intervention venues), spill-over effects due to patrons ners, particularly in the intervention community. In visiting both intervention and control venues, national addition, interviews with people socializing in the venues HIV prevention campaigns, time-gap between the inter- indicated marked changes in their sexual behaviour. vention and the post-intervention survey, and other There were decreases in the reported number of sexual interventions being run in some of the control venues partners and the proportion reporting engaging in trans- [14]. actional sex and increases in reported condom use in The overall objective of this targeted condom distribu- both communities. However, the increase in reported tion and peer education intervention was to reduce the condom use with the previous partner met in the venue incidence of HIV and other STIs, but we did not mea- was particularly sharp in the intervention community sure any biological outcomes of the intervention. Mea- where most of the respondents reported obtaining the suring effects on incidence of HIV in a population condom in the same venue. It thus seems likely that the requires a much bigger sample size and investment. increase in reported condom use in high risk places in Very few other intervention studies targeting high risk Sandøyetal.BMCPublicHealth2012,12:10 Page9of12 http://www.biomedcentral.com/1471-2458/12/10 Table 4Behaviour, perceptions andprevention-related experiences ofguests invenues where people meetnew sexual partners Baseline Follow-up Intervention Control Intervention Control % N % N p-value % N % N p-value Cometomeetsexualpartner 53 150 58 267 0.649 29 224 32 211 0.720 Evermetsexualpartnerhere 70 180 66 326 0.453 55 260 54 264 0.792 Condomlasttimepartnerfromhere 57 116 55 217 0.745 82 126 68 123 0.075 Howoftencondomw/newpartnerlastmonth Always 52 135 37 249 0.050 66 118 52 147 0.102 Sometimes 41 46 21 20 Never 7 17 13 29 Paidforlastcondom 22 170 72 170 <0.001 Condomwithyou? 39 178 32 325 0.277 11 243 7 257 0.289 Condomshownifclaimedtohavebrought 94 79 90 114 0.350 100 21 96 14 0.189 Howeffectivearecondoms Very 51 178 38 325 0.054 67 264 57 272 0.0961 Somewhat 17 22 17 23 Notvery 25 26 6 12 Notatall 7 7 5 6 Don’tknow 1 8 5 1 AtriskofHIV No 32 179 36 325 0.211 41 264 36 271 0.736 Moderate 40 33 34 40 High 21 18 18 16 Veryhigh 6 14 7 8 Discussedwithanyonehowtopreventinfection 81 178 75 324 0.391 71 263 66 271 0.372 Ifyes,withwhom Parents 4 148 0.4 249 0.012 1 197 2 184 0.823 Spouse 41 148 39 249 0.783 19 197 29 184 0.105 Friends 85 148 81 249 0.458 72 197 65 184 0.328 Peereducators 11 148 21 249 0.0571 55 197 6 184 <0.001 Healthpersonnel 24 148 31 249 0.361 22 197 28 184 0.299 EverexperiencedanyHIVpreventiveactivitiesinthisvenue 62 242 16 257 <0.001 1Significantdifferencewhenadjustedfortypeofsiteandageandgenderofrespondents places have attempted to do so. However, reported STI from 2008-2010) (clinic registries of the public clinics cases indicate a decreasing prevalence of STIs overall in serving the intervention and control communities). The both communities during the intervention period, and sharper decline in Maramba may be partially due to our the relative decline in reported cases was sharper in the intervention, but any attribution must be done with cau- intervention than the control community (38% vs. 16% tion. A study in Zimbabwe with a much stronger design, Table 5Differencesin reported number ofpartners by guestssocializing in venues where peoplemeetnew sexual partners Baseline Follow-up Intervention Control Intervention Control Est. N SD1 Est. N SD1 p-value Est. N SD1 Est. N SD1 p-value Mediannumbersexpartnerslast4weeks 3 177 3 323 1 260 1 273 Meannumbersexpartnerslast4weeks 3.04 177 1.85 3.16 323 2.19 0.200 2.30 260 19.1 2.04 273 5.22 0.443 Mediannumbernewsexpartnerslast4weeks 2 177 2 323 0 254 1 268 Meannumbernewsexpartnerslast4weeks 1.75 177 1.46 2.09 323 1.83 <0.001 1.20 254 4.19 1.78 268 5.03 <0.001 Mediannumbersexpartnerslast12months 2 256 2 262 Meannumbersexpartnerslast12months - - - - - 5.26 256 21.3 7.61 262 16.9 <0.001 1Median,meanandSDofmeanarenotadjustedforclustering Sandøyetal.BMCPublicHealth2012,12:10 Page10of12 http://www.biomedcentral.com/1471-2458/12/10 a cluster-randomised trial, which included a peer educa- first monitoring round. Receiving feedback from the tion and condom distribution component, did not find monitor probably motivated the peer educators to any impact on HIV and STI incidence [15]. ensure that they distributed sufficient condoms after Condom distribution to high risk places may obviously this. However, some peer educators seemed to continue be organized in different ways, and using peer educators to underestimate the demand for condoms, possibly to do this is possibly not the cheapest (at least if the because they did not visit the venue frequently enough, peer educators are provided financial compensation) and for example in relation to busy weekends. quickest way. Nonetheless, using peer educators pro- The improved condom availability in the control com- vides an additional opportunity to engage people in dis- munity could indicate a trend towards improved con- cussions about HIV prevention. Studies indicate that dom distribution, i.e. that it was easier for venue owners peers of the same sex are an important source of infor- to obtain free or subsidised condoms that could be mation about sex-related issues among young people given or sold to customers. It is also possible that the [16]. It is likely that a higher number of individuals increased availability was partly a spill-over effect from socializing in the venues would have been reached with the intervention as some of the peer educators lived in HIV-related information in this study if the peer educa- the control community and reported that they had dis- tors had been available when the venues were busier. tributed condoms in bars and night clubs in their own However, it is not possible to distinguish whether the neighbourhood too since they had not been aware that increased reporting of condom use among guests was a it would serve as a comparison during the evaluation. result of the combination of condom distribution, con- Since we found that there was an increase in the pro- dom demonstrations and behaviour change discussions portion who believed that condoms were effective as conducted by peer educators or of improved condom HIV prevention, it is also possible that venue staff and availability alone. People who had been in contact with owners may have experienced an increased demand for a youth peer educator were not more likely to report condoms from guests, and this may have motivated using condoms. On the other hand it seems likely that them to make efforts to offer condoms. The reduction the peer educators would have had a bigger impact if in high risk behaviours reported both among respon- persons of different ages had been recruited although dents in the intervention and control communities may the evidence for peer educator effectiveness from other indicate a general trend. In addition to influence from studies again is mixed. A randomised controlled peer national campaigns, local prevention efforts carried out education intervention study among male beer hall by different NGOs in partnership with the DHMT may patrons in Zimbabwe, which included condom informa- have had an impact. These have included drama, train- tion and demonstrations and recruitment of men of dif- ing of youths, improved VCT and PMTCT services, free ferent ages, did not find any impact of the intervention provision of condoms and STI treatment services for on unprotected sex with non-marital partners [17], and female sex workers, and promotion of subsidized con- a review of youth peer education intervention studies doms in high risk places (personal communication with conducted between 1998 and 2005 found no impact on former District Director of Health, Dr. Chinyonga). A condom use [18]. Nonetheless, a review of studies on behaviour change in the general population would youth peer education interventions for HIV prevention explain the decline in HIV and syphilis prevalence in low- and middle-income countries conducted observed among young pregnant women in Livingstone between 1994 and 2008 found that such programs often during the period 2002-2008 [4,20] and also be in line resulted in increased HIV-related knowledge and with behaviour changes reported in other studies in the increased reporting of condom use, but that there was region [21-24]. less evidence for an effect on sexual abstinence and The assignment of the intervention was not rando- number of partners [19]. mised. Thus we cannot rule out that there were other It is likely that lack of knowledge among new staff important differences between venues and respondents explained why not all representatives interviewed in the in the intervention and the control community, which intervention venues in the follow-up survey reported were not related to the intervention, but which could that condom distribution had taken place there. At the explain some of the observed changes. Since there was a same time, the external monitor revealed that continu- rather long period of five years between the baseline ous availability of condoms was not fully achieved and follow-up surveys, it is possible that other programs although this was one of the most important objectives and changes had taken place in both the intervention of the intervention. It was expected that the peer educa- and the control community. Fortunately, the adjusted tors would need some time to sort out the demand for logistic regression analyses indicated that differences in condoms in the venues, and this may explain why as types of venues included in the surveys did not influence many as a quarter of the venues lacked condoms in the the main findings. The low number of refusals in the