This is the largest and most comprehensive investigation of HBV, HCV and HIV among KPs in South Africa to date. The study confirms earlier research highlighting the high HIV burden among these populations . Most importantly, it documents the alarmingly high seroprevalence of HCV infection among PWUD/ID (46%).
This study follows South African census distinctions of race. These remain relevant because education, socio-economic status and socio-cultural norms continue to map onto apartheid-era categories to some extent. In 2015, 81% of the population self-defined as Black, 9% as Coloured, 8% as White and 3% as Indian/Asian. The Western Cape has the highest proportion of people classified as Coloured (50% of the population) , partially explaining the proportionally higher number of Coloured people in Cape Town in each demographic category included in this study.
MSM were recruited from existing, specialised fixed site sexual health clinics. These served people from a wide range of socio-economic circumstances, reflected in the comparatively higher number of white MSM (up to 44% in Pretoria) included in the study and the lower rates (1 to 6%) of people who reported homelessness.
SWs were recruited through a combination of fixed sites and mobile services. Nationally, 5% of SWs are estimated to be male, a figure relatively stable across provinces  but this was not reflected in the study (<1% SWs male). Low numbers of SWs reporting homelessness may be because income generated from SW provides a means of paying for accommodation. Another explanation may be that current SW services are not reaching homeless SWs.
PWUD/ID were largely recruited at locations where mobile needle and syringe distribution and collection services where provided. The higher numbers of men who use drugs included in this study reflects programmatic data that indicates higher numbers of men (between 87% and 90%) to be using harm reduction services . The low proportion of females who use drugs in this and other local studies [17,44] is likely also reflective of the gender distribution of PWUD/ID. Female-specific barriers to services – including a lack of tailored services and higher rates of stigma and discrimination experienced by women who use drugs  – likely also affected the number of female PWUD/ID in this study. The need for gender-appropriate and specific HIV and HCV services for women who use drugs is increasingly recognised locally , regionally  and globally . The high levels of homelessness among PWUD/ID likely reflects the socio-economic circumstances of PWUD/ID who used the services and/or who were accessible by the study team. PWUD/ID with financial means could access prevention commodities and supplies from pharmacies and/or through the private healthcare system.
The overall HBsAg prevalence of 4% is similar to the general population [45, 9] with a marked difference between those born before and after the introduction of HBV childhood immunisation (4% versus 1%). This finding provides additional evidence supporting the benefit of the introduction of childhood HBV vaccination in 1995. However, some people born after this date have acquired HBV infection. The overall coverage of the 3-dose HBV vaccine schedule in South Africa was estimated to be 74% in 2016 , and the birth dose vaccine to prevent mother-to-child transmission is yet to be implemented.
The HCV findings are anticipated yet alarming, despite the study bias towards KPs who access HIV prevention, treatment and related services. Globally, HCV is a particular concern among PWID , with high prevalence being documented in Africa, ranging from 30% among PWID accessing opioid substitution therapy in Tanzania  to 97% of PWID in Mauritius . Our findings show PWUD/ID carried the highest HCV burden with some marked geographic variations. The extremely high rates of HCV seroprevalence in PWUD/ID in Pretoria (72%) is possibly due to the relatively longstanding nature of this injecting community, with sub-optimal access to prevention interventions . Viraemic rates at the upper limit (75-80%) of what would be anticipated may point towards repeated infections . Equally, HCV seroprevalence in MSM, with high HIV-HCV co-infection prevalence, is expected given the risks of HCV infection among MSM, especially in HIV positive MSM and MSM with a history of injecting drug use . The almost non-existent HCV seroprevalence among SW is supported by the low rate of injecting drug use and high condom use in this sample.
The circulating prevalence of HCV genotypes 1a and 3a is similar to other countries where HCV infection is networked and predominantly spread through the sharing of contaminated injecting equipment among PWID [54,55]. However genotype 5a (a unique and prevalent genotype) was not found. A study of blood samples from patients attending specialist clinics (n=941) and blood donors (n=294 ) between 2008 and 2012 identified genotype 5a to be common among specialist clinic patients (36%) and particularly common (60%) among people over 50 years old in both groups, and among Black African people (54%). The lack of genotype 5a in this study suggests that the modes of transmission of identified genotypes among PWUD/ID and other KPs differs from older patients attending specialist clinics, some of whom contracted HCV through blood transfusions and unsafe medical injections , traditional practices such as tribal scarification  or unknown risks.
Lower levels of self-reported HIV status compared with measured prevalence (25% and 37% respectively) has been documented previously in South African KPs . The fact that 25% of participants indicated that this was their first health-screening (including HIV testing) may indicate that a large number of people were genuinely unaware of their status due to fears of testing or competing priorities. This discrepancy may also relate to reporting bias and internalised stigma, fear of knowing one’s status and fears of acknowledging positive status while not on ART .
In Cape Town and Pretoria, HIV prevalence was notably higher among the few MSM reporting recent injecting compared to their male PWUD/ID counterparts. The different HIV prevalence may be linked to the recruitment of MSM from well-established clinics that have been providing HIV treatment for MSM for several years, and the elevated risk among MSM who engage in multiple high-risk practices.
Overall reported use of ART, at 84%, was below the 90% target for people living with HIV, but higher than the national average of 57% , likely a consequence of the fact that participants were accessed through HIV service delivery platforms. The wide range across cities and populations (from 11% in PWUD/ID in Cape town to 96% MSM in Pretoria) also likely reflects which services are being primarily accessed at the included organisations by the target populations. The relatively lower levels of ART coverage reported among PWUD/ID (51%) further likely reflects the numerous barriers affecting ART uptake  in this population.
Blood borne virus co-infections
A global systematic review and meta-analysis of the prevalence and burden of HCV co-infection in people living with HIV reported a 6% coinfection prevalence in MSM and 82% in PWID compared to 2% within the general population . In comparison, our findings indicate 3% HCV co-infection prevalence in MSM living with HIV and 65% in PWUD/ID living with HIV. A South African study among antenatal attendees living with HIV found HCV prevalence of 0.1% .
Overall 2% were HIV-HBV co-infected with similar prevalence in all 3 KPs.
The number of participants who were co-infected with HIV/HBV or HIV/HCV, or HCV-HBV-HIV triple-infected is cause for concern given the extent to which co-infections can accelerate and amplify disease . There is therefore the need for screening for co-infections in KPs especially HIV-infected PWUD/ID.
Viral hepatitis awareness programmes that emphasise the risk of co-infections and reinfection and provide prevention strategies need to be up-scaled across all cities in South Africa for all KPs. The importance of preventative HBV vaccination needs to be stressed and routine access to vaccination at all levels of care is essential.
Risks for these blood-borne infections are exacerbated by a host of social and structural factors including limited access to opioid substitution therapy and needle and syringe services for PWID; criminalisation of sex work and drug use and corresponding marginalisation and difficulty accessing healthcare (for PWUD/ID and SWs) and homelessness (largely PWUD/ID).
In addition to alcohol, the high reported use of heroin is of concern. Though the majority of heroin used in South Africa is smoked , it is also the most commonly injected drug among PWID . The HIV and viral hepatitis risks related to heroin are largely through injecting, with heroin dependant people injecting on average four times per day . Methamphetamine use is more common in Cape Town than in the other cities. HIV and viral hepatitis risks relating to methamphetamine use are either indirect, through sexual risk as a result of increased sexual risk taking, or direct, through injecting , which has been documented in Cape Town and is not uncommon .
A small proportion of MSM across the three cities reported injecting drug use, mostly methamphetamine. The prevalence and patterns of heroin and/or methamphetamine injecting among MSM has not been well documented in South Africa . The injection of methamphetamine in the context of sexual encounters (Chemsex) has been documented among sub-groups of MSM . Chemsex among MSM can be associated with unprotected anal intercourse, and in some instances with multiple sex partners. This can be continued for long durations (up to days), and has been identified as a risk factor for HIV and viral hepatitis in other countries [66,67].
No SWs recruited reported injecting, however, networks of SWs who inject drugs have been reached through more recent harm reduction and sex worker programmes operating in Johannesburg. Risks involved in substance use are exacerbated by the criminalisation of drug use, limited access to harm reduction services, and discriminatory treatment within healthcare service settings .
Sexual practices, unsurprisingly, differed considerably between KP sub-groups. SWs were more likely to report recent sexual activity, multiple partners, transactional sex and substance use at last sex as well as having the highest reported condom use, in general. Local research among female SWs in South Africa has identified high reported levels of condom use in major metropolitan areas (89% in Cape Town and 84% in Durban) .
SWs in this study were least likely to report lubricant use in relation to receptive anal sex. Little data exists on the use of lubricant during anal sex among SWs. Data from a three city bio-behavioural survey among female SWs in Cape Town, Johannesburg and Durban showed that only a quarter of female SWs were aware of lubricant, with low levels of reported use .
The greatest regional variation in sexual practices was also seen among SWs. Those in the Mthatha region were more likely to engage in high-risk sex, with lower condom use and a greater propensity to engage in sex in exchange for goods and commodities. In contrast, those working in KwaZulu-Natal, reported much higher condom use and almost exclusively received money in exchange for sex. This may suggest lower negotiating power of sex workers in Mthatha and could explain the markedly lower rates of condom use at last penile vaginal sex reported there. Elsewhere, penile-vaginal condom usage in the general population has been shown to correlate to education and exposure to behaviour change interventions . Mthatha is in a more rural setting compared to the other included metropoles and SWs there may have relatively less access to education and information. There is however, little data on SW sexual practices outside of large metropolitan areas.
MSM were most likely to have engaged in receptive anal sex in the past week, had relatively low reported use of condoms but high lubricant use and comparatively moderate substance use at last sex. High-risk sexual practices, particularly low levels of condom use have been identified among MSM in South Africa . HIV prevention services for MSM have been operational in South Africa for almost a decade in the cities where MSM were recruited from, where condoms and lubricant are distributed . Pre-exposure prophylaxis for MSM is available from the clinics where MSM were recruited, however, experience of PrEP was not assessed, which may have influenced condom use.
PWUD/ID were least likely to have engaged in sexual activity in the last month and reported relatively high condom use at last sex. Surprisingly, substance use at last sex was moderate and comparable to MSM, however, due to the frequency of substance use among PWUD/ID with a substance use disorder, under reporting may have taken place. While data on sexual practices among PWUD/ID in South Africa is limited, international research reflects that sexual practices are influenced by a range of factors, including relationship status  and mental health.
This paper is limited to a description of the findings, without in-depth analysis. Additional analysis around HBV, HCV and HIV infection and risks factors among PWID has been published elsewhere .
Several of the limitations and biases described here are inherent in research that is implemented in the context of programmatic service delivery in a way that minimizes impact on clients and staff.
This study drew on individuals already accessing available HIV and related prevention and treatment services and was also limited to a few of the sites and cities with existing targeted health services for KPs.
The selection bias associated with convenience sampling limits extrapolation to other members of the included populations. The findings may not be applicable to KPs with less access to health services, KPs who access services in the private healthcare sector, KPs with different risk profiles, or to KPs from other cities or regions in the country. Another important KP group that was not addressed by this study were prisoners in whom HBV, HCV and HIV prevalence in other global settings is known to be higher than in the general population [68,72,73].
Categorisation bias may have taken place in relation to same sex practices among male SWs and injecting drug use among MSM. However, no notable differences were identified among male SWs reporting recent anal sex and the prevalence of HIV and HBV (no SWs had HCV infection) compared to MSM. The higher HIV prevalence among the few MSM with a recent history of injecting is potentially linked to selection bias from MSM HIV and sexual health clinics, and the likelihood of them engaging in different sexual practices. The likely lower injection frequency among MSM who inject may have contributed to the lower prevalence of HBV and HCV identified among MSM with recent injection history compared to male PWUD/ID. Differences in type of drug injected and injecting frequency suggest that viewing these groups separately is useful.
The study did not comprehensively assess all the risk factors for blood borne infections, and did not include an assessment of knowledge or attitudes related to HIV and viral hepatitis.
The use of English for the questionnaire in a context of multiple different first languages (including Afrikaans, Xhosa and Zulu), and potential ad hoc translating by researchers, may have undermined accuracy of answers in the questionnaire.
Long-standing relationships between participants and research implementation organisations is likely to have impacted on reporting accuracy. However, reporting bias may have resulted in under- or mis-reporting of measures assessing substance use and sexual activity. This may have been especially marked in relation to activities generally associated with the KP the participant was not identifying as (for example, SWs may have been less comfortable reporting substance use than people identifying as PWUD/ID). Self-protective behaviours encouraged by implementing organizations (such as the use of sterile injecting equipment among PWUD/ID or condoms across all KPs) may also have been over-reported, resulting in an under-representation of risk practices, and an over-representation of harm reduction practices. The questionnaire did not assess frequency of use of the various substances enquired about thus limiting the insights into substance use within these populations.
Additional research is needed on the epidemiology of these infections among hard to reach PWUD/ID not routinely accessing mobile health services including those who may belong to other KPs or have additional risk factors, such as within the correctional service system. Further understanding of whether, and to what degree, knowledge of these infections and their associated risk factors among KP affects risk practices and prevalence would be valuable in the design of prevention and care services. Greater insights are also needed into factors associated with on-going parenteral risk in the presence of harm reduction services.
Pilot projects that explore innovative ways to provide integrated HBV-HCV-HIV and other holistic services to KPs in the South African context are needed to inform the granular detail and implementation experience that will be necessary for an effective integrated HBV/HCV/HIV response. Future research is needed to understand health care providers’ knowledge around viral hepatitis and HIV among at risk KPs to complement the existing epidemiological data.
 Personal communication with Leora Casey, key population programme manager at NACOSA, 19 June 2019.
 At the time of study implementation HIV prevention, sexual health and harm reduction services for KPs existed in several other locations that were not part of this study. Notably, SW services existed in major urban areas like Johannesburg, Ekurhuleni and Bloemfontein. MSM services existed in Durban, Pietermaritzburg, Bloemfontein, Kimberly and Ehlanzeni. PWUD/ID harm reduction services were started in Port Elizabeth and Johannesburg after the study started.