Recruiting people who inject drugs (PWID) into research: benefits and challenges of respondent driven sampling (RDS) in a sample in KwaZulu-Natal, South Africa

Abstract

Background

People who inject drugs (PWID) are often excluded in research due to social circumstances and a lack of creativity when recruiting this community. Respondent driven sampling (RDS) is one method used to access marginalised and stigmatised communities such as PWID. This study piloted RDS with PWID in eThekwini District, South Africa, and also explored participants’ perspectives of the usefulness of this approach to recruit PWID.

Methods

In-depth interviews (IDIs) were conducted with 10 key stakeholders, followed by a cross-sectional survey among 45 PWID recruited via RDS. Further, 11 of these PWID participated in IDIs. Survey data were entered into REDCap and descriptively analysed. IDIs were transcribed, translated and thematically analysed using Dedoose software.

Results

Participants described barriers and facilitators accessing PWID for research. Access to PWID was viewed as challenging because of factors including stigma and concerns of criminalisation. However, making use of PWID networks and peers was felt to be important to facilitate recruitment. RDS was described as an acceptable recruitment strategy, and was rated as “not burdensome” by most (98%). RDS was successfully carried out in this pilot. Participation of PWID in research was viewed as important, but challenging. It was felt that participation of PWID could be impacted by concerns of opioid withdrawal, therefore timing (length and time) of participation was important to consider in study design. Reimbursement strategies for PWID participating in research were discussed. Incentives for participation, including cash reimbursement, were viewed favourably for PWID.

Conclusions

RDS is plausible for recruitment of PWID in eThekwini District, South Africa. It can be used to further understand PWID networks and to better serve them for future research and appropriate care.

Trial registration

N/A, not a clinical trial.

Supplementary Information

The online version contains supplementary material available at 10.1186/s13011-026-00712-x.

Background

People who inject drugs (PWID) are an oftentimes marginalised and stigmatised population that face various unique barriers and challenges, particularly around access to health services and intervention research [1, 2]. In South Africa, research with PWID has faced various challenges. There are limited data on the size of the PWID population [3, 4], due to factors such as negative relationships with law enforcement which impact on population mobility [1]. Further, conventional sampling and recruitment strategies are hindered by social stigmatisation, structural barriers, and criminalisation, making recruitment of PWID into research challenging [1, 5, 6]. These challenges contribute to the “hidden” characteristics of this population, requiring researchers to consider/pursue creative and alternative strategies for the identification and recruitment of PWID for the purposes of clinical and socio-behavioural studies.

Respondent driven sampling (RDS) is a unique recruitment approach that makes use of peer-driven sampling methods and which is used primarily in research involving such hidden populations [6–8]. To achieve this, RDS leverages people’s social networks and involves participants, themselves, in the recruitment process [9]. RDS begins with the use of “seeds”, who are the first participants recruited into the study itself. In the majority of RDS-based studies, these seeds are then provided with coupons and are sent to recruit others from within their own social networks [10]. Each seed is sent out with a maximum target of peers to recruit, with each person they recruit then bringing in the unique coupon given to the seed. Researchers can then track which seed recruited which next-round participant for documentation and reimbursement purposes. Oftentimes, RDS occurs across numerous stages, wherein seeds recruit participants who then, themselves, go out with coupons to recruit more in an ever expanding series of “waves” [10]. This process is done iteratively until either equilibrium is reached, which occurs when the composition of the sample stabilises and becomes independent of the seeds, or when the target sample number has been achieved.

Previously, RDS has primarily been used in the recruitment of groups of people who are underserved and underrepresented, “hidden populations”, ranging from people who use drugs [5, 8, 11], to sex workers [6, 8, 12], and men who have sex with men (MSM) [8, 13]. RDS, through its peer-driven approach, presents a unique avenue in which to approach and recruit these communities, that meets them on their own terms, where participants can choose to recruit people that they already know in the target population [14]. They are able to self-select who they recruit, and whether/what information they will share with others [14], while maintaining a greater deal of confidentiality than may be offered by other approaches [12], which could help to protect them from facing increased social stigmatisation as a result of study recruitment/participation. Although RDS is a powerful means of understanding hidden populations, it can also be infeasible and/or unacceptable because of stigma, fear of legal repercussions or small social networks of selected seeds [15, 16]. Previous literature also highlights, in some instances, lack of standardisation and transparency in how RDS has been used [10].

In addition, the potential for ethical concerns in studies using RDS, particularly in relation to monetary incentives, has been highlighted [14, 15, 17]. Concerns, including the potential for coercion, underground marketplace for selling coupons and study related services, and falsifying eligibility criteria have been described [15, 17]. The use of monetary versus in-kind reimbursement has also been debated, with arguments that in-kind payments can compromise participants autonomy [14, 15] versus concerns that monetary payments can subsidise further drug use [15]. The South African Health Products Regulatory Authority (SAHPRA) has developed a time (T), inconvenience (I) and expenses (E) (TIE model) to assist with calculation of appropriate reimbursement for clinical trial participants [18], which is recommended by South African ethics committees.

Formative research prior to using RDS is recommended [15, 19] to facilitate understanding of the ethical concerns in the population of interest. In addition, the varied and lack of standardised data demonstrates the need for further, more detailed and structured studies that use, test and outline RDS strategies.

PWID globally experience unique and significant barriers when it comes to healthcare access. In South Africa, these barriers are exacerbated as there are high numbers of people living with HIV in the general population, creating additional burden on HIV prevention and treatment services [20]. Further research into this population and their experiences and needs is therefore necessary, which in turn can facilitate future targeted interventions. Therefore, RDS was piloted within the eThekwini District (Durban), KwaZulu-Natal Province, in South Africa, in a study which also explored access to and acceptability of sexual and reproductive health (SRH) and HIV services for PWID (which has been published elsewhere [11, 21]). We evaluated this recruitment strategy using an exploratory qualitative approach together with quantitative survey questions. In addition, questions about access to PWID as study participants and motivators to their participation in research, were explored. PWID who had been recruited via RDS and key informants working with PWID were specifically asked their opinions about methods of recruitment. The data were used to determine the usefulness of RDS as a strategy to reach PWID in this context, in order to begin addressing gaps in research and to inform future studies in this population. This manuscript describes barriers and facilitators to the recruitment of PWID into research, and the acceptability and feasibility of RDS as a method of recruitment.

Methods

This was a mixed methods study, and participants were recruited via RDS to participate in a survey (if eligible) and then also invited to participate in an in-depth interview (IDI). The study had various aims, including (1) To pilot the use of RDS to access PWID in KwaZulu-Natal and to assess the feasibility and acceptability of this sampling approach; and (2) Using qualitative methods to assess access to and acceptability of HIV prevention and treatment services among PWID in KwaZulu-Natal. This manuscript reports specifically on the first aim. The second aim has been reported elsewhere [11, 21].

We piloted RDS with 45 PWID in the eThekwini District, central Durban, KwaZulu-Natal, South Africa. All eligible participants recruited via RDS were invited to participate in a survey. In-depth interviews (IDIs) were also conducted with 11 PWID and 10 key informants (including healthcare providers, people who help PWID access healthcare services, and a legal representative with experience working with people who use drugs).

The survey asked participants to rate the acceptability and feasibility of using RDS to access and enrol PWID into research. The IDIs included questions which explored, in detail, their opinions about using RDS and/or other suitable methods for recruiting PWID into research.

We adhered to the COREQ (consolidated criteria for reporting qualitative research) guidelines for the reporting of this study [22].

Study population and data collection processes

Key informants

Ten key informants participated in IDIs during the period October 2020 to March 2021. These IDIs were conducted prior to any contact with PWID. Key informants were purposively selected and invited due to their experience with PWID – including representatives from non-governmental organizations (NGOs) providing targeted healthcare and/or harm reduction services, representatives from advocacy organisations, including a legal representative and relevant government department representatives. The key informants were selected based on their expertise and contributions to discussions regarding facilitators and barriers to accessing healthcare services, as well as for their perceptions regarding research recruitment and participation of PWID in research. The legal environment has potential to impact how PWID interact with healthcare services and research participation, therefore perspectives of legal representatives working with PWID were also deemed important. There were no refusals to participate, but some stakeholders recommended more suitable expert participants. All informants provided written informed consent. IDIs explored strategies to recruit PWID in research (including RDS), as well as information on access to SRH and HIV related services for PWID.

Interviews were conducted by female researchers with social science degrees, who were trained in conducting sensitive research. CM, TC and one other researcher, fluent in isiZulu (who is not a co-author), conducted the IDIs in English or isiZulu, depending on participant preference. Although one participant (an advocate for sex workers) was known to the researchers, there was no other prior relationship with informants. IDIs were conducted remotely (via telephone/Zoom) or in-person, and lasted between 1 and 1.5 hours. At the time of the IDIs, there were various regulations in place to minimise the spread of COVID-19, and the use of online platforms was encouraged. In-person IDIs were conducted with social-distancing, masking, and screens in place.

People who inject drugs

Recruitment of PWID took place from November 2021 to February 2022, and we aimed to recruit up to 45 PWID from the community. Contact was initiated with the PWID community via a harm reduction centre in Durban, South Africa. Three known members of the PWID community were recruited through contacts from the harm reduction centre, to function as seeds to recruit PWID from their individual social networks to enrol into the study. Once they agreed to participate, the seeds provided written informed consent to participate in a survey. Similar to other research and RDS methods [23], these three seeds were each allocated up to three “recruitment coupons” to recruit up to three other PWID to participate in the survey. Physical coupons were created – each with the study name, basic information about study eligibility and contact details for the study site. Coupons had unique codes for participant identification, as well as activation and expiration dates to encourage prompt enrolment [24]. Coupons had no value. Although physical coupons were created, most PWID opted not to take their coupons off-site with them, citing concerns of losing them or having them stolen. Instead, their details were recorded in logs on site, and coupons were stored in their participant files. When the seeds returned to the research centre with their “recruits”, their identity and coupon information were confirmed. Once coupon information was validated, and new participants identified, the coupons were marked as invalid/used. Seed participants were reimbursed for identifying potential “recruits” and were then able to leave the site. Following this, if eligible, the new “recruits” were given information about the study, invited to participate, and if interested provided independent and voluntary written informed consent to participate in the study. After enrolling in the study and completing the survey, new participants were also allocated up to three coupons each, to recruit additional PWID from their own social networks. This was done until we had reached a sample of 45 PWID participants (Fig. 1). The sample size was limited by funding for this pilot study.

Fig. 1.

RDS coupon allocation

All PWID provided written informed consent to participate in the study. Participants were 18 years or older, and self-reported injecting drugs in the last 6 months. There were no refusals, since people who were not willing simply did not come to the research centre to participate. The survey included a section on RDS – exploring relationships between participants and recruiters as well as acceptability of the sampling/recruitment method. Surveys were conducted by female interviewers trained in quantitative and qualitative data collection, with training in collecting sensitive data from vulnerable participants. Interviewers were not known by the participants. The surveys were conducted in English or isiZulu, in a private interview room at the research centre. Data were directly entered into REDCap [25] by the interviewers, and the process took approximately one hour.

During the period December 2021 to February 2022, eleven PWID were sampled from the survey participants and invited to participate in an IDI. They were purposively selected to ensure representation of male and female participants, in various stages of the RDS survey network. We had aimed to conduct approximately ten IDIs with PWID, or to conduct IDIs until saturation had been met. During the process of inviting PWID to participate in an IDI we identified eleven PWID who were willing to participate, and therefore included an additional participant who had valuable insights to share. The single session IDIs were conducted at a time convenient to the participants and on a different day from the survey. They were conducted one-on-one, in-person, in a confidential setting at the research centre. They were conducted by the same female interviewers who conducted the survey research, and were all conducted in isiZulu. The interviews lasted between 1 and 2 hours and explored factors including acceptability of RDS as a method of recruitment, as well as access to and use of SRH and HIV prevention and treatment services (published separately) [11, 21].

Each PWID participant was reimbursed R30 per eligible participant recruited via RDS (a maximum of R90, ~US$5). Further reimbursement was provided for participation in the survey (R250, ~US$15) and the IDI (R100, ~US$6) (where applicable). Reimbursement amount was approved by the ethics committees, and was standard for similar research activities conducted at the research site at that time. In addition, since PWID surveys and IDIs were conducted in-person during the COVID-19 pandemic, participants were provided with mini bottles of hand sanitiser and re-usable masks.

Data analysis

Interviewers entered survey data directly into a REDCap database as questions were answered [25]. Questions on the acceptability of RDS as a sampling method and participant relationship to the person who recruited them, were descriptively analysed.

All IDIs were audio recorded, transcribed, and translated (where necessary) into English. Transcripts from IDIs conducted with both key informants and PWID were reviewed and a single code list was developed using both inductive and deductive strategies. The codes were informed by the structure of the questions asked, and thematic areas arising from the data. Through repetition, identification of similarities and differences, and interpretation of data, two researchers (CM and TC) drafted a code list, which was reviewed and discussed by CM, TC and BZ. Once agreement was reached, CM and TC each coded a subset of the transcripts, until all data were coded. Dedoose software (version 9.0.62, LA, California), was used to facilitate the analysis of the data.

Quantitative data from the surveys are presented first, followed by qualitative data presented according to thematic areas specific to this manuscript, in the results section below.

Ethical considerations

This study was approved by the Human Research Ethics Committee (HREC) of the University of the Witwatersrand (Reference: 200202) and the Emory University Institutional Review Board (IRB) (ID: STUDY00000134). All participants (key informants and PWID) were 18 years or older, and provided written informed consent to participate in the survey and/or IDI, with separate consent for audio recording of their IDIs. Consenting was conducted in IsiZulu or English, depending on participant preference.

Results

Participant demographics

Participant demographics are outlined in Table 1. Key informants were slightly older than PWID participants. PWID reported high levels of unemployment (38%), and many were currently homeless (71%).

Table 1.

Demographic details of PWID (Survey and IDI) and key informant participants

	Survey: PWID n = 45 (%)	IDI: PWID n = 11 (%)	IDI: Key informants n = 10 (%)
Age (range) (years)	21–43	21–36	29–65
Median (years)	29	28	37.5
Gender
Male  Female  Gender non-conforming	26 (58) 19 (42) 0 (0)	5 (45) 6 (55) 0 (0)	2 (20) 7 (70) 1 (10)
Employment
Unemployed  Part-time*  Full-time	17 (38) 27 (60) 1 (2)	4 (36) 6 (55) 1 (9)	N/A
Housing
Currently homeless	32 (71)	7 (64)	N/A
HIV status
Known living with HIV	15 (33)	1 (9)	N/A

*Part-time work includes self-reported hustling such as street vending, sex work and selling scrap metal; N/A: Not applicable and/or not asked

Survey: RDS acceptability data

All PWID participants who completed the survey (excluding the three seeds) (n = 42) were asked about the nature of their relationship to the person who recruited them. They were also asked how long they had known that person (Table 2). The majority reported that the person who recruited them was someone they took drugs with (n = 36, 85%). In addition, the majority had known the person who recruited them for more than one year (n = 24, 57%).

Table 2.

Survey: PWID reports of RDS acceptability and relationship with recruiter (n = 42)

	n (%)
Nature of relationship with recruiter:
Friend I take drugs with	27 (64)
Someone I take drugs with not a friend	9 (21)
Casual acquaintance not a friend, sexual partner or not someone I take drugs with	4 (10)
Friend	1 (2)
Don’t know	1 (2)
Length of relationship with recruiter:
Less than 6 months	7 (17)
6 months to 1 year	10 (24)
More than 1 year	24 (57)
Acceptability of process of RDS recruitment:
Perfectly acceptable	31 (74)
Acceptable	11 (26)
Slightly acceptable	0 (0)
Slightly unacceptable	0 (0)
Unacceptable	0 (0)
Totally unacceptable	0 (0)
Burden of process of RDS recruitment:
Not burdensome	41 (98)
Somewhat burdensome	1 (2)
Moderately burdensome	0 (0)
Extremely burdensome	0 (0)

All survey participants (excluding seeds) were also asked “How acceptable did you find the process of recruitment to be a participant in this survey?”. They were provided with a 6-point rating scale, ranging from “Totally unacceptable” to “Perfectly acceptable”. All participants felt that the method of recruitment was either acceptable (n = 11, 26,2%) or perfectly acceptable (n = 31,73,8%). In contrast, all participants in the survey (excluding seeds) were also asked “How burdensome did you find the process of recruitment to be a participant in this study?” using a 4-point rating scale, from “Not burdensome” to “Extremely burdensome”. Almost all (n = 41, 97,6%) responded that it was not burdensome, with only 1 responding that it was a somewhat burdensome process.

Thematic findings

Results are presented according to four thematic areas: (1) Access to PWID for research - which explores barriers to recruiting PWID as well as suggested strategies for recruitment of PWID into research; (2) RDS as a recruitment strategy –attitudes to/acceptability of RDS for recruiting PWID, concerns with using RDS as a recruitment strategy, as well as the relationship of the study participants to their recruiter in this study; (3) PWID participation in research – including barriers to participation and motivators/incentives to participate in research; and (4) Reimbursement strategies for PWID who are enrolled in research studies – the merits of reimbursement via vouchers versus cash reimbursement are also discussed.

To note, there were no observable differences in responses between in-person or remote/online IDIs.

Access to PWID for research

Barriers to recruiting PWID into research

PWID and key informants described various barriers to accessing/recruiting PWID, especially street based PWID, into research. These included concerns of safety (of recruiters), fear of judgement/stigmatisation experienced by PWID, and difficulties contacting/locating PWID.

It was noted by PWID that they often live in unsafe areas, and that some also reportedly engage in criminal activities that could put recruiters at risk.

[T]hese people who uses drugs are not to be trusted, you can get mugged while you still walking. (PWID, Female, 28 years, PID311)

Stigma and fear of being criminalised and/or incarcerated were also perceived as barriers to recruitment of PWID. Some key informants suggested that it would be important for recruiters to be sensitised and trained to facilitate recruitment:

[I]t depends on your staff who are doing it, so if they are not sensitised, or if they are, if their approach […], if it’s in any way stigmatising, […] it’s going to make it harder to find people. (Key informant, Female, 40 years, PIDKI09)

Furthermore, the lack of contact details, with no set address or mobile numbers was viewed as a challenge for locating and recruiting PWID.

Suggested strategies for recruiting PWID into research

Most key informants described the importance of using existing organisations who have relationships with PWID to facilitate their recruitment into research. They underlined the importance of having a trusting relationship, and mentioned the value of understanding and normalising the community of PWID to enable recruitment processes.

[I]f you partner with an organisation that’s got, you know, trust already […] You would be able to recruit very quickly. (Key informant, Male, 40 years, PIDKI03)

[W]e need to normalize that even a person who uses drugs must be known as a person in the community, […] that these people are drug users. Because if something [like research] is happening in South Africa they [PWID] are also entitled to participate. There should be a way for drug users to be known as a person. (PWID, Male, 36 years, PID3111)

Both PWID and key informants also described the usefulness of using PWID, peers or community mobilisers to access and recruit other PWID into research, stressing that they are more likely to know who they are and where they are located, and to be able to do it safely.

[I]t is good to send us [use PWID to recruit PWID] so that you know you are safe because if you come by yourself to those people and recruit by yourself to people who are injecting you can lose your things because they will think they can get something to smoke maybe […]. (PWID, Male, 36 years PID3111)

[W]e have found from work with people who inject drugs, it’s important to consider having a community mobiliser or somebody who helps to recruit. (Key informant, Male, 34 years, PIDKI02)

In addition, key informants suggested location/venue-based strategies for recruitment.

[PWID] then will be able to identify and do the mapping of where to actually find other users, because I mean you can’t ask someone who doesn’t know anything about, about drugs to actually [help recruit] […], they help a lot because they know each and every corner where you can find injecting drug users. (Key informant, Gender non-conforming, 35 years, PIDKI10)

One key informant noted different strategies of recruitment could be used for different types of research, highlighting respondent driven sampling as a suitable strategy for recruitment in quantitative research.

But if it’s something quantitative, you know, the use of respondent driven sampling is the best I think and it works really well for things that will provide generalisable […] data. What we often do is we use more opportunistic sampling so that it’s not just research for the sake of research (Key informant, Male, 40 years, PIDKI03).

RDS as a recruitment strategy

Although RDS was viewed as acceptable, participants also had concerns surrounding RDS as a strategy for recruitment of PWID into research.

Acceptability of RDS as a recruitment strategy

Key informants felt that RDS was an acceptable form of recruitment, with some describing it as “similar to snowballing” (Key informant, Female, 31 years, PIDKI04).

I think it [RDS] will actually bring more numbers in terms of like, you know, not seeing only faces that you are familiar with, you know, because somebody will be like you know, interested to actually “you know I am going to find somebody that you guys have never seen” […] [A]lso, they have enthusiasm to like say “you know, I also know somebody” because I, trust me, there’s always a connection of people who use drugs, you know. (Key informant, Gender non-conforming, 35 years, PIDKI10)

PWID (including seeds) were asked during their IDIs about their perceptions of the RDS recruitment process used for this study, and the survey results were verified. They described it as a “good” recruitment methodology, which they felt was beneficial, especially for researcher safety. They also felt that they would be better placed than researchers to identify other PWID through RDS.

I see it [RDS] as the right way because you [researcher] can’t just come on your own and just come to these places like [place name] to find people inject drugs, you see that thing. Where else for me I could see someone like me, I could tell that person like there is a thing going like this and like this and this is what they do [referring to the research], but for you [referring to interviewer], you can’t talk to a person like because firstly you will be scared of me, you see. Thinking that I am a criminal, so for me it will be someone I have lived with before that I know and see. (PWID, Male, 32 years, PID312)

Some described how there were many interested PWID when they told them about the study:

“[A]nd that we tell each other like this [about the research study], it is also easy […] like when I gave out coupons as I told them there were so many of them who wanted to come here [to the research centre] also. (PWID Seed, Female, 22 years, PID003)

Concerns with RDS as a recruitment strategy

Some key informants felt that location-based sampling, where researchers go to where participants are based, may be preferable to RDS. Key informants had concerns that PWID could have various challenges which could cause them to miss their scheduled research appointments.

[I]t would be better if you can go where you will find them [referring to PWID], like I explained earlier that they will not leave where they are, first they need a fix, secondly, they will even have to get transport from there to here and they do not have the money for transport. (Key informant, Female, 53 years, PIDKI08)

Both PWID and key informants expressed concern that PWID could lose the coupons developed for the study recruitment process which could impact on reliable recruitment via RDS.

Like when you give them tickets [coupon], and you ask them “do you want to take your ticket with you”, one would say, “can I leave it here” and another would say “I want to take it with me”. You see the one that took the coupon…. it is possible for them not to come back because it is obvious the people that he lives with…anything can happen, you see the police that chases us away, the coupon may get lost during that time while you are running away. Or maybe the people that you live with may steal it. (PWID, Male, 23 years, PID031)

When asked if digital coupons could be used, key informants confirmed that most street based PWID do not have phones/devices, so this would not be a feasible alternative to paper based coupons.

No, they don’t have any kind of electronic devices. Or any […] ways of getting things digitally, so I don’t know how else you would do it. (Key informant, Female, 40 years, PIDKI09)

Relationship of PWID to recruiter in this study

Similar to the survey data, when describing the nature of their relationship to the person who recruited them, all PWID IDI participants described that person as someone they took drugs with:

They are people I live with that I inject with, whom I smoke with. (PWID, Female, 27 years, PID131)

Furthermore, all three of the seed participants participated in an IDI and noted that they recruited people that they took drugs with:

We know each other because we live together here where we live and these are people we know each other because we inject ourselves together. (PWID Seed, Female, 22 years, PID003)

IDI participants also reported having known the people who recruited them for varying lengths of time, with the majority knowing each other for a year or more (n = 7; 64%).

PWID participation in research

Barriers to participation of PWID in research

PWID and key informants described various challenges that could be barriers to the participation of PWID in research, including the effects of consuming drugs, and their impact on ability to remember, to concentrate and focus, as well as withdrawal and fear of withdrawal.

[T]he problem will be the cravings because if one is craving they don’t have time to talking to other people, they don’t feel like talking to anyone, they want to be alone […] they become short tempered, they feel like not being in front of people not interacting with people, feel like you can be alone, you feel sick you know, because the way you feel might makes you do anything in order for one to get the fix. It is something that is unbearable. (PWID Seed, Female, 22 years, PID003)

[I]t’s very, very difficult to get them concentrating, or, because most of the time the drug that they use makes them a bit numb-y and you know they sleep all the time […] they […] slow and they, they have a sleep when they talking to you, and you know, so really, I mean getting them before they take something. (Key informant, Gender non-conforming, 35 years, PIDKI10)

Other PWID participants and key informants described that it would be best for PWID to have injected prior to participation in order to be able to focus on their research participation.

Without your fix, nothing you will be able to do. […] Nothing. Because your mind is focussed on getting what you need. (Key informant, Female, 53 years, PIDKI08)

[O]ne thing that could make us not to get them [PWID] is that when you get there and ask the person to come in the office, one will say the time you go there he is having aroster [withdrawals] he cannot stand for aroster without injecting. That is why he cannot do anything before injecting. (PWID, Male, 36 years, PID3111)

One key informant also raised the importance of time – both the length of an interview/research appointment and the time of day (i.e., not too long, not too late in the day) – as influencing participation and quality of interaction in research.

Think about length of time of interview, close to where they hustle, time of day [….] I think that it was, honestly about managing the time and being very aware that after forty-five minutes to an hour like, people were starting to get pretty antsy [agitated]. (Key informant, Male, 34 years, PIDKI02)

Motivators/incentives for PWID to participate in research

Key informants felt that there should be engagement with PWID, to ask them what would motivate them to participate in research.

So, involve people, and […] let them engage, and let them maybe not decide, but like brainstorm into what do they think will work best for them. […] You know, you might not necessarily agree with that, but like it will maybe take you somewhere. (Key informant, Gender non-conforming, 35 years, PIDKI10)

PWID and key informants also discussed various factors which could motivate/provide incentives for PWID to participate in research – ranging from food, masks and sanitisers for COVID-19, and hygiene packs to sterile injecting equipment.

[M]aybe buy bread when they come in, buy them drinks, food so the person relaxes and feels alright. (PWID, Female, 21 years, PID032)

Just general hygiene packs actually are, are very well received by people, so soap, you know, deodorant […] tampons, pads, etcetera for women and whether or not the people need to use them for their personal hygiene or they need to sell them, that’s, you know, an independent thing, toothpaste, but that I think is an additional, value add, is good. (Key informant, Male, 40 years, PIDKI03)

A fundamental thing should be, is that, whatever you do you should be providing with […] sterile injecting packs as like a baseline minimum package, and if you engaging with research funders that don’t do that I think it’s unethical. (Key informant, Male, 40 years, PIDKI03)

Reimbursement in the form of cash was viewed as a strong motivator for participation in research.

So, if they can maybe get R30 [South African Rands, ~US$1.60] to get something to smoke so then they can be inspired that I already know that when I go there, I won’t be wasting my time because some they say they can’t because their time has been wasted while they can be hustling. (PWID, Female, 27 years, PID131)

Reimbursement strategies

There was strong opinion from key informants that participants should be reimbursed for time spent in a study as it was impacting on their time to hustle and make money to survive.

“[W]e wanna interview people who inject drugs”, that’s great but […] how are you going to make it worth someone’s time? We’ve also gotta […] take into account, that some people may be homeless and unemployed therefore them taking a day out of their time to hustle and to make money, because, you know, some of the ways that people, who are homeless that I’ve worked with have made money is they either collect recycle, they recycle or they may beg or they may do piece jobs or small once-off jobs and so forth, so to take someone’s time, an entire day away from them without reimbursing them, without ensuring that they don’t go back to their sleeping space with, with no means of eating and also using [drugs], we’ve gotta be straightforward. (Key informant, Female, 29 years, PIDKI07)

In terms of RDS recruitment and reimbursement for each person successfully recruited, PWID and key informants also felt that this was “a great idea, then, that’ll work well” (Female, 44 years, PIDKI01).

Like giving them the R50 [South African Rands, ~US$3] when they bring a person, that motivates them, because they know that they are going to get something. (PWID, Male, 23 years, PID031)

Vouchers versus cash as a reimbursement strategy

Key informants were asked about the most appropriate type of reimbursement to provide PWID, to compensate them for their time and costs getting to the research centre. Most felt that reimbursement should be in the form of cash rather than vouchers (for groceries or airtime). Many felt that PWID would sell vouchers for cash, as they would not have anywhere to store/cook groceries and that they did not own mobile phones. A few felt strongly that it was paternalistic (and unethical) to choose for participants and that participants should (and could) make their own decisions how to spend their money.

I am quite sure the grocery vouchers would be traded for cash [laughing] […] It’s just that they’ve got nowhere to, so even if they get groceries, they can’t store them anywhere, […] and cooking is also a challenge. (Key informant, Female, 30 years, PIDKI06)

[I]t’s almost saying “oh, we don’t want you to spend it on drugs” or “we don’t want you to spend it on certain things so you have to buy groceries with it” and a lot of the time I know that those [vouchers] are usually swopped out for cash anyway, […] I personally am very much against using vouchers, I much prefer, and I would highly suggest as well, looking at cash reimbursement. (Key informant, Female, 29 years, PIDKI07)

Although, one key informant felt “ethically it will be better to provide shopping vouchers [than cash]” to PWID (Key informant, Female, 31 years, PIDKI04).

Discussion

PWID are an understudied population who are often discriminated against, thereby impacting their ability and willingness to access lifesaving prevention and treatment services [26, 27]. As a result it is important to include PWID in public health research, in order to be able to understand their needs and facilitate and improve their access to care. Participants in this study highlighted several reasons that can make it challenging to recruit PWID into research, from unsensitised researchers being afraid to physically enter areas that PWID reside in because of perceptions of lack of safety, to PWID fear of criminalisation and stigmatisation [1, 5, 6], making them reluctant to engage with researchers/people that they do not know. Furthermore, street-based PWID lack fixed addresses, mobile phones and/or reliable contact details for follow-up in research. Consequently, recruitment strategies need to take these factors into consideration to faciliate recruitment that is flexible and non-threatening to the population being researched. Other studies utilising RDS in South Africa have demonstrated that it is a suitable method for recruiting similar populations, such as homeless youth and MSM, whose voices would otherwise be missed by conventional and traditional sampling and recruitment methods [23, 28].

Participants in this study highlighted that different recruitment strategies could be followed for different research aims, but that building on existing relationships that have been established with PWID with people known to them – either through organisations working with PWID or their peers – is critical for enabling recruitment. The importance of trust between peers in recruitment has also been demonstrated elsewhere [29]. Similar to other research [9], this study highlighted that RDS is a strategy that makes use of these trusting peer relationships, whereby participants themselves facilitate the process of recruitment. This peer recruitment strategey was viewed positively by both key informants and PWID.

In this pilot study, we were able to successfully implement RDS as a recruitment strategy for PWID in the eThekwini District, South Africa. All PWID participants in the survey (excluding seeds who were not asked) described it as an acceptable strategy of recruitment, and almost all felt that it had not been a burdensome recruitment process. In addition, qualitative data demonstrated that it was an acceptable method of recruitment in this context. In this study, most participants had an extended (more than 6 months) relationship with the person that recruited them, similar to research in Cape Town, South Africa, where the majority had known their recruiters for more than 2 years [5]. The extended length of relationships demonstrates that there are existing networks within the community, further highlighted by PWID during IDIs, where they described inside knowledge of their community, and an ability to identify other eligible PWID, facilitating the process of RDS recruitment.

At a practical level, RDS makes use of coupons in the recruitment process. These are useful to monitor recruitment chains and to know who was recruited by whom – as each coupon has a unique code that links participants to those that have introduced them to the study [30]. The number of coupons distributed in this pilot study was limited by the sample size. Similar to other research using RDS in Sub-Saharan Africa and South Africa, acceptability of this recruitment strategy was high [5, 29] – evident by the high coupon return rate/ return rate of participants. In our study, almost all participants requested that their physical recruitment coupons be stored at the study site, for fear that they could be stolen or that they could lose them due to not having access to adequate storage facilities, and this request was accomodated by the research team. In spite of not having the physical coupon as a reminder, they still returned to the research centre with their recruits.

Contrary to key informant concerns that PWID would forget their survey or IDI appointments, not have transport, or be looking for their “fix”, almost all PWID returned without receiving any study reminders. Furthermore, once the planned number of participants had been recruited via RDS for the survey, PWID still expressed willingess to continue with recruitment, demonstrating that more waves of recruitment would have been possible.

Various factors were considered motivators for PWID to participate in research, and could have contributed to the success of this pilot. In line with recommendations [8, 15, 19] appropriate reimbursement and incentives for participation were discussed with key informants prior to recruiting PWID via RDS. Key informants strongly recommended that PWID participants should be reimbursed for time (which could otherwise have been spent hustling) and costs of getting to the research centre. Reimbursement in the form of cash was considered most appropriate, similar to previous research demonstrating that payment to economically vulnerable populations is ethically viable [31] and important for supporting individual autonomy [14, 15]. It was also viewed by key informants and PWID as a strong motivator to participate in research. Provision of masks and sanitisers for COVID-19 could also have contributed to motivation to participate. The provision of reimbursement is in line with recommendations that all participants should be appropriately compensated for their time (T), inconvenience (I) and expenses (E) (TIE model) [18, 32, 33], and in order to respect their contributions to respective research studies/science.

As with any study, there are some limitations for consideration. This was a pilot study, so the sample of PWID recruited via RDS was small and limited due to funding. However, the enthusiasm of participants to continue with recruitment once the sample size was reached highlighted that this method of recruitment could be used further in this population in the eThekwini District in South Africa. RDS can further be tested in other areas, to see if it works similarly well with a bigger sample. Furthermore, it could be extended to more rural areas to determine if networking in these areas is also possible. Another potential limitation is that the three seeds were recruited from a methadone treatment centre, therefore participants’ ability to remember to recruit others and to come to the research centre on an agreed date may have been facilitated by the treatment they were accessing, as studies have shown that neurocognitive functioning of people on opioid substitution therapy (e.g. methadone) is better than those using illicit drugs [34]. However, many participants were recruited from other environments in the city and came to the research centre on the agreed dates without reminders. It is also important to note that the participants recruited in this study were all from the city centre. Although there is limited information on the size of the PWID population in this area, it is likely that there is a much wider network of PWID in the District, Province and more rural areas, whose needs are much greater [21]. If possible, a larger sample could facilitate access to PWID from a wider network, and therefore enable researchers to gain a better understanding of their needs, to facilitate access to treatment, support and care.

It must also be noted that this study took place when many COVID-19 restrictions were in place, and social distancing, barriers and masks were recognised as the norm. Therefore, all consenting, survey discussions and interviews took place in a spacious interview room, where participants and interviewers had masks, and there was a screen divider for personal protection. In spite of this, interview discussions were rich. Many of the PWID were living in conditions with greater health risk, and reported having experienced stigma and discrimination in their lives. They were happy to be able to respond to questions and share their stories with the interviewers in these settings, however the possibility that this could have impacted on quality of discussions should not be overlooked.

Conclusions

This study demonstrates that RDS is plausible and is an acceptable method to recruit PWID, one “hidden population”, in the eThekwini District in South Africa. There is limited data on PWID in South Africa, and in KwaZulu-Natal more specifically [11]. Future research can and should use recruitment methods like RDS when seeking to better understand PWID networks, their locations, and their needs in an attempt to better deliver appropriate and person-centred healthcare services and care to this community.

Supplementary Information

Below is the link to the electronic supplementary material.

Abbreviations

HREC

Human Research Ethics Committee

IDI

In-depth interview

IRB

Institutional Review Board

MSM

Men who have sex with men

PWID

People who inject drugs

RDS

Respondent driven sampling

SRH

Sexual and reproductive health

TIE

Time, inconvenience, expense

Author contributions

CM, SB, JS and BCZ contributed to the study conception and data collection processes. Data analysis was performed by CM and TC. The first draft of the manuscript was written by CM, and TSH. All authors (CM, TSH, TC, SB, MW, JS and BCZ) provided input to the draft, and read and approved the final manuscript.

Funding

This work was funded by the Centre for AIDS Research at Emory University (P30AI050409).

Data availability

Access to the data from this study may be requested through submission of a research concept to: cecilia.milford@gmail.com. The concept must include the research question, data requested, analytic methods, and steps taken to ensure ethical use of the data. Access will be granted if the concept is evaluated to have scientific merit and if sufficient data protections are in place.

Declarations

Ethics approval and consent to participate

This study was approved by the Human Research Ethics Committee (HREC) of the University of the Witwatersrand (Reference: 200202) and the Emory University Institutional Review Board (IRB) (ID: STUDY00000134). All participants (key stakeholders and PWID) were 18 years or older, and provided written informed consent to participate in the survey and/or IDI, with separate consent for audio recording of their IDIs. Consenting was conducted in IsiZulu or English, depending on participant preference.

Consent for publication

Not applicable, individual data/images not used. Consent covered the use of anonymous quotes in written articles and professional meetings.

Competing interests

The authors declare no competing interests.