Patterns of HIV testing, drug use, and sexual behaviors in people who use drugs: findings from a community-based outreach program in Phnom Penh, Cambodia

This was a mixed-methods study that was composed of a large quantitative cross sectional survey, with a smaller qualitative component comprising focus group discussions (FGDs).

The research was conducted among people who use drugs in Phnom Penh, who were in contact with a community-based program known as Sustainable Action against HIV and AIDS in Communities (SAHACOM). SAHACOM was implemented over a 5-year period ending in December 2014, focusing on key populations [24, 25]. The project was implemented by KHANA, a leading HIV non-governmental organization working through numerous community-based organizations (CBOs).

The project utilized a community-based approach to empower and create ownership of the project among key populations, including people who use drugs. The model was composed of itinerant peer-based outreach, and static drop-in centers. Through this model, key population peers were hired by CBOs to provide HIV and harm reduction services to people who use drugs, including HIV education, condoms, clean needles, and syringes. Through peer involvement, the model transformed people who currently or formerly used drugs from passive service recipients to active providers of services [26].

Besides providing the above services in their own localities, peers referred people who use drugs to the static drop-in centers, which were part of community-level infrastructure, where methadone, harm reduction counselling, peer discussions, personal care, needles, syringes, and condoms, among other commodities, were provided. In addition, people who use drugs in contact with the outreach teams were encouraged to form and attend peer discussions at drop-in centers for education purposes.

At the end of the implementation period, data were collected to assist in evaluation of the project. The purpose of the evaluation was to understand the HIV testing and prevention needs of the participants, and their perception of the SAHACOM services. Mixed methods were used to allow contextualization of quantitative survey data. This is a widely used approach of linking quantitative and qualitative data within implementation research [27, 28].

By the time the SAHACOM was implemented, the estimated total number of people who use drugs in Phnom Penh was 4188; of these 1068 injected drugs. In Phnom Penh, SAHACOM was implemented in 33 communes with high concentration of people who use drugs out of 96 communes in the city. Although accurate general population size for these communes was not available, strategic information related to HIV suggested that these 33 communes covered approximately 1969 people who use drugs, of whom 353 injected drugs. A two-stage cluster sampling method was used to select participants for the quantitative component. At the first stage, the research team identified five communes with at least 20 people who use drugs to be included in the study. A list of ‘hotspots’ where people who use drugs and congregated was made based on geographical clustering of high risk behaviors as is practiced elsewhere [29]. At the second stage, all people who use drugs in the selected hotspots were approached and invited to participate in the study, and were screened for eligibility.

Potential participants were included in the study if they: (1) were at least 18 years of age, (2) had used any form of illicit drugs in the past 3 months, (3) were able to present themselves on the day of the interview, and (4) were able to provide informed consent to participate in the study. Of the 192 people screened, 22 were excluded, leaving 170 eligible participants. For the qualitative component, a convenience-based subsample was selected from the primary participants in the quantitative phase. At the end of quantitative data collection, participants were invited to take part in focus group discussions based on interest, convenience, availability, and willingness to consent. The participant selection process is illustrated in Fig. 1.

The quantitative questionnaire was developed in English, translated into Khmer, back-translated to English, and pretested to ensure that it was culturally suitable and clearly understandable. Standardized questions were adapted from previous studies and Demographic and Health Surveys [14, 30, 31] to document participants’ socio-demographic characteristics (e.g. age, sex, marital status, education, occupation, income, residence, and mobility), drug use characteristics (e.g. age of initiation, frequency, method and types and circumstances of drug use, sharing of needles or syringes, as well as history of drug-related arrests and voluntary rehabilitation or incarceration/imprisonment), HIV-related characteristics (e.g. prior HIV testing, venue of HIV testing and counselling, and sources of HIV education), sexual behaviors (e.g. sexual activity, number of sex partners, consistency of condom use, history of STI symptoms, whether STI treatment was sought, and nature of facility from which treatment for STI was sought), and self-perceived risk of HIV. To get a sense of participants’ self-perceived risk of HIV, participants were asked to indicate whether they thought they were at lower, same, or higher risk, relative the general population.

Similarly, a topic guide for qualitative focus groups was developed in English, translated into Khmer, back-translated to English, pretested, and adjusted accordingly following piloting. The objective of the qualitative component was to contextualize and provide insights to the quantitative findings. The focus group topic guide questions were open ended and explored participants’ drug use behaviors, HIV testing experiences, sexual behaviors, perceived risk of HIV, and perception of services they were receiving through the SAHACOM.

Data were collected by researchers in private rooms in drop-in centers. Before data collection, researchers were provided with an overview of the study objectives and trained on the questionnaire filling techniques including error checking, focus group facilitation, privacy assurance, confidentiality, and data protection. Questionnaires were administered face-to-face to 170 participants, each lasting 30–45 min. At the end of survey questionnaires, participants were invited to participate in the focus groups.

Interested and consenting participants were provided with appointment schedules to attend the discussions. Invitation to take part in the focus groups was not based on types of drugs used, as long as participants had participated in the quantitative survey. In total, two focus group sessions, each with six participants, were conducted on the premises of CBOs. Recruitment for the focus groups was guided by saturation of themes, rather than by pre-specified sample size as recommended in qualitative studies [32]. At the end of the second focus group session, it was clear that additional discussions would not yield further insights and recruitment was halted. The focus groups lasted 45–60 min, were tape-recorded, and were conducted in the local Khmer language.

Quantitative data were coded and entered into a computer using Epi Data version 3 (Odense, Denmark). Double data entry was performed to limit errors. Logic checks were performed, and 19 participants who reported that they had not used any illicit drugs in the past 3 months were excluded from the analyses. Summary of frequencies for nominal and ordinal variables was documented. Descriptive analyses were used to compute means for continuous variables. SPSS version 20.0 (IBM Corporation, New York, USA) was used for all data analyses.

Qualitative audio data were transcribed and transcripts translated to English from Khmer and imported into NVivo (QSR International). Thematic data analysis [33] was then conducted inductively within the original aims of the study, in order to provide context for the quantitative findings regarding sexual and injecting behaviors, nature of services received by participants, and participants’ perception of those services. Nodes were created, and populated with appropriately labelled codes. Codes were populated with textual segments refined using constant comparison approach [34], and then categorized based on similarities to generate overarching themes [32, 33]. Verbatim quotes were extracted from participants’ transcripts and displayed in the text to aid transparency of interpretation as recommended [35]. Qualitative text and quotes were presented alongside quantitative data in order to contextualize or explain survey findings in an integrated fashion [36].

This study was approved by the National Ethics Committee for Health Research (No. 082NECHR). Verbal consent was obtained from each participant. Before consenting, participants were informed of their right to discontinue their participation at any time, without any consequences. Names and contact information of the researchers were provided to participants for use in case they had any questions. Identifiable data were not collected on questionnaires; instead, unique codes were utilized in order to protect confidentiality of participants. Participant were compensated approximately US$5 for their time and transport to participate in the study.

In total, 151 people who use drugs participated in the survey, whose mean age was 31.2 (SD = 6.5) years; a subsample of whom (n = 12) participated in the focus groups. Of the total participants, 77.5% were male; 39.1% were married, 40.4% were single, and 20.5% were either divorced or widowed. In addition, 14.6% were unemployed, and the sample was generally non-migratory, having lived in their current locations for a mean of 19 years (Table 1).

As shown in Table 2, 38.0% of the total participants had injected drugs in the past 3 months. On average, participants started using drugs at the age of 21.3 (SD = 6.5) years, and had used drugs for the past 8.4 years. The most common drugs used were amphetamines (72.8%) and heroin (39.7%). Although most participants used drugs either with friends (54.4%) or alone (31.1%), a notable proportion (14.5%) used drugs with their sexual partners. More than half (55.3%) of the overall sample had been referred to a rehabilitation center in their lifetime, and 21.9% had been sent to the center in the past 12 months. A third (31.3%) had ever been incarcerated/imprisoned due to drug possession and other offences.

Less than a third (27.5%) of participants who injected drugs in the past 3 months reported that they had used a needle that had been used by someone else the last time they injected. Contextualizing this finding a participant explained that, “we stopped sharing syringe…we try to prevent it” (respondent # 1, focus group 1), while another asserted that, “we take a methadone regularly, stopped injecting and stopped sharing syringes” (respondent # 4, focus group 1). Participants mentioned that peer educators were an important source of syringes for the participants. Besides commodity supplies, participants relied on peer educators for “transportation service to get methadone” (respondent # 4, focus group 1).

However, participants raised concerns regarding a lack of supplies on weekends or other days when national holidays and ceremonies were being held. One participant asserted that, “on Saturday and Sunday, if we don’t have, we need to buy” (respondent # 6, focus group 1), because as explained by another participant, “they just don’t work at weekends but Monday to Friday they come to meet people at local” (respondent # 5, focus group 1). Participants also explained that re-using or sharing needles was common at night when getting syringes or needles was limited:

As shown in Table 3, 96.0% had ever been tested for HIV in lifetime, and 83.3% had received an HIV test in the past 6 months. In addition, 97.2% had received counselling during their most recent HIV test. A majority (62.5%) of the HIV testing was delivered through community- or peer-initiated counseling and testing, with a minority (9.0%) having been provided through voluntary counseling and testing centers. In addition, 86.6% had received HIV education in the past 12 months, and although most participants received this information from several different sources, the most common source of HIV education was peer educators or outreach workers (89.2%). These qualitative data suggested that community-based outreach program was an important source of HIV education, counselling and testing which was relatively acceptable to participants.

Contextualizing these findings, qualitative focus groups supported the view that HIV testing and education provided by peers and outreach workers were acceptable to participants. Referring to these peers and outreach workers, a focus group participant affirmed that, “they help us to know that we are reactive HIV…. if we are HIV positive we go to get ART” (respondent # 1, focus group 2). This participant further commented that, “the organization always comes to educate us.” (respondent # 1 focus group 2), while others contrasted access to testing in health facilities by asserting that they found outreach services easily accessible “because they come two times per day” (respondent # 6, focus group 2).

As shown in Table 4, 37.3% of the participants reported having sex while intoxicated with drugs. In addition, 10.8% had been diagnosed with an STI in the past 6 months. At the same time, 41.1% reported having been involved in sexual intercourse in the past 3 months, and only 35.6% reported using a condom in their last sexual intercourse act.

Qualitative data provided additional insights regarding the inconsistent condom use. In particular, narratives from the focus groups suggested that participants did not believe that they needed to use condoms with their spouses because they trusted them. Asked why they were not using condoms, participants responded by saying that, “I trust my wife” (respondent # 6, focus group 1), “Yes, I believe my wife” (respondent # 5, focus group 1), among other similar responses.

As shown in Table 3, a majority of participants (56.5%) viewed themselves to be at the same or lower risk of HIV infection compared to the general population. Although there were other reasons for not using condoms, such as a “desire to have a child” (respondent # 3, focus group 1), the finding of low condom use among the sample is relevant as it corroborates the low self-perception of HIV risk among the participants. While responding to a question regarding whether he thought he was at risk of HIV, a participant stated that, “I have only my wife, I don’t have any other partners” (respondent # 4, focus group 1), showing that participants viewed HIV risk as external to themselves and their drug-using behaviors.

Further exploration of this issue in the focus groups suggested that participants believed that their increased knowledge was contributing to their protection from HIV, and they were no longer at high risk of HIV. When asked whether they thought their group was a high-risk group, participants contrasted the previous lack of knowledge with their current situation, stating that:

This participant, who belonged to a specific peer discussion group of drug users in one of the drop-in centers went on to state that:

These sentiments were supported by others who emphasized the changes in risks occurred due to the changes in sexual and injecting behaviors, primarily due to increased supply of harm reduction commodities. In a typical response regarding why they thought they were at low risk of HIV, a participant stated that, “previously, we didn’t have condoms to use and didn’t have any NGOs coming to provide them” (respondent # 2, focus group 2). Others also referred to condoms, stating that their HIV risk was reduced “because we have NGOs who taught about using condoms; now we know how to use it, especially to prevent STIs” (Respondent # 6, FGD 1).

Apart from availability of condoms, few participants also referred to availability of clean syringes as giving them a sense that they were at low risk. For instance, one participant mentioned that, “as for now, we do not have any disease because we use new syringes” (respondent # 1, focus group 1). As can be noted from the above excerpts, participants felt that, apart from knowledge per se, their use of preventive interventions was protecting them from HIV. Within these discussions, it emerged that the use of these comodities was enhanced by their free provision:

The study involved participants who were already in contact with peer outreach. It is therefore possible that this recruitment bias may have affected our results related to participants’ uptake of HIV testing and perception towards peer-led community-based outreach. A reason for the finding that participants held the perception that they had lower HIV risk compared to the general population may have related to their recent HIV testing. Moreover, the proportion of participants who were already HIV positive was not known as participants were not required to divulge their HIV status. Consequently, the relationship between recent test results and risk perception could not be explored in this study. All participants were recruited in Phnom Penh, limiting generalizability to other people who use drugs nationally. Participants in the focus groups were selected based on convenience, which may exacerbate this selection bias. Despite application of eligibility criteria designed to retain participants that were currently using drugs, a small minority reported they had not used any form of illicit drugs in the past 3 months. More broadly, the methodology used in this study largely depended on participants self-reporting, and some participants skipped some questions, which may have introduced further limitations.

Despite these limitations, our findings are consistent with research in Cambodia and elsewhere showing the effectiveness of community-based outreach in reaching people who use drugs [20, 50]. While this study reports on uptake of HIV testing, downstream data on linkage and retention on ART was unavailable, yet are central in the care continuum. Besides, the extent to which the program contributed to achievement of the first 90 in the 90–90–90 targets cannot be deduced because of lack of denominators and prior rates of testing. This is due to lack of local level strategic information regarding numbers of people at risk. Nationally, past efforts predominantly focused on increasing HIV testing, with significant limited focus on the continuum on local levels [23]. This was certainly the case in 2014 when this study was conducted [23].

However, the Health Sector Strategic Plan IV (2015–2020) includes strategies for improving the strategic information base as one of its cross-cutting strategy. Generation of localized data will enable programs monitor their contributions to 90–90–90 targets better, within their catchment areas. In addition, focus on the care continuum is central to the relatively recent approach—known as Boosted Continuum of Prevention to Care and Treatment—which aims to identify and reach new infections, ensure they are brought into care, and are retained in ART. This approach is targeted specifically to areas where new infections are occurring, among key populations, and should therefore be central to future community-based outreach programming and research.