Design and baseline findings of a large-scale rapid response to an HIV outbreak in people who inject drugs in Athens, Greece: the ARISTOTLE programme

Abstract

Aims

To (i) describe an intervention implemented in response to the HIV-1 outbreak among people who inject drugs (PWIDs) in Greece (ARISTOTLE programme), (ii) assess its success in identifying and testing this population and (iii) describe socio-demographic characteristics, risk behaviours and access to treatment/prevention, estimate HIV prevalence and identify risk factors, as assessed at the first participation of PWIDs.

Design

A ‘seek, test, treat, retain’ intervention employing five rounds of respondent-driven sampling.

Setting

Athens, Greece (2012–13).

Participants

A total of 3320 individuals who had injected drugs in the past 12 months.

Intervention

ARISTOTLE is an intervention that involves reaching out to high-risk, hard-to-reach PWIDs (‘seek’), engaging them in HIV testing and providing information and materials to prevent HIV (‘test’) and initiating and maintaining anti-retroviral and opioid substitution treatment for those testing positive (‘treat’ and ‘retain’).

Measurements

Blood samples were collected for HIV testing and personal interviews were conducted.

Findings

ARISTOTLE recruited 3320 PWIDs during the course of 13.5 months. More than half (54%) participated in multiple rounds, resulting in 7113 visits. HIV prevalence was 15.1%. At their first contact with the programme, 12.5% were on opioid substitution treatment programmes and the median number of free syringes they had received in the preceding month was 0. In the multivariable analysis, apart from injection-related variables, homelessness was a risk factor for HIV infection in male PWIDs [odds ratio (OR) yes versus no=1.89, 95% confidence interval (CI)=1.41, 2.52]while, in female PWIDS, the number of sexual partners (OR for >5 versus one partner in the past year=4.12, 95% CI=1.93, 8.77) and history of imprisonment (OR yes versus no=2.76, 95% CI=1.43, 5.31) were associated with HIV.

Conclusions

In Athens, Greece, the ARISTOTLE intervention for identifying HIV-positive people among people who inject drugs (PWID) facilitated rapid identification of a hidden population experiencing an outbreak and provided HIV testing, counselling and linkage to care. According to ARISTOTLE data, the 2011 HIV outbreak in Athens resulted in 15% HIV infection among PWID. Risk factors for HIV among PWID included homelessness in men and history of imprisonment and number of sexual partners in women.

INTRODUCTION

For many years, Greece has experienced a low-level, stable HIV epidemic concentrated mainly in men who have sex with men [1]. In 2011, a 16-fold increase in the number of reported HIV cases in people who inject drugs (PWIDs), compared to 2010, was observed [2]. Molecular analysis of HIV-1 sequences from PWIDs sampled in 2011 confirmed the recency of this outbreak and the presence of phylogenetic clusters, which indicated transmission through injecting drug use and sharing of injection equipment [3]. Greece was severely hit by the financial crisis that unfolded in mid-2008, and it has been proposed that economic recession was a distal or macro-level cause of this outbreak [4].

In 2011 it was estimated that there were approximately 8000 problem drug users in Athens, of whom approximately 2800 had injected drugs in the last month [5]. Prior to the outbreak, the coverage of prevention and harm-reduction services such as opioid substitution therapy (OST) and needle and syringe programmes (NSP) was very low. In early 2010, more than 5500 opioid-dependent individuals in Athens were waiting to receive OST for an average of 7.6 years [6]. In addition to that, NSP programmes distributed approximately seven syringes per estimated PWID per year [6] whereas, for the prevention of HIV, United Nations (UN) agencies judge the annual distribution of 100 syringes per injecting drug user as low [6,7]. Soon after the outbreak was recognized, a plan was developed for the implementation of an intervention, targeting the population of PWIDs in the Athens Metropolitan area. The main objective was to identify rapidly as many PWIDs residing in Athens as possible, to inform them about how they could prevent HIV infection and transmission, to test them for HIV infection and to link those found HIV-positive to OST and anti-retroviral treatment (ART). To achieve rapid high coverage of the target population, a chain referral process with monetary incentives was employed (respondent-driven sampling, RDS). The programme was named ARISTOTLE, and recruitment was initiated in August 2012.

ARISTOTLE involved five RDS recruitment rounds. The findings of the first round concerning the prevalence of HIV infection and related risk factors have been reported elsewhere [8]. This paper aims to: (i) provide a detailed description of all design aspects of the programme, (ii) assess the success of the programme in identifying and testing the hard-to-reach population of PWIDs, and (iii) describe socio-demographic characteristics, risk behaviours (injecting, sexual) and access to treatment and prevention, estimate HIV prevalence and identify factors associated with HIV infection based on data from the first participations of all PWIDs recruited in the five rounds.

METHODS

ARISTOTLE was designed as a ‘seek, test, treat, retain’ (STTR) intervention [9]. This type of intervention involves reaching out to high-risk, hard-to-reach PWIDs (‘seek’), engaging them in HIV testing and providing information and materials to prevent HIV (‘test’) and initiating and maintaining anti-retroviral and opioid substitution treatment for those testing positive (‘treat’ and ‘retain’).

Programme site and personnel, training and formative research

The programme was implemented in a building of the Greek Organization Against Drugs (OKANA) located in the centre of Athens. The staff included a physician, interviewers with prior experience with the target population and a psychologist and two social workers. A flow manager was responsible for maintaining the flow of participants and ensuring that participants completed each of the steps of the process on a first-come, first-served basis. As the phylogenetic analysis of HIV sequences of the outbreak identified that the potential founder of the largest phylogenetic cluster was from Iran/Afghanistan [3], two cultural mediators speaking Farsi were selected for the programme. Apart from Greek and Farsi, additional languages included English, French, Arab and Kurdish. The personnel received training for 2 months (Supporting information, Appendix S1).

Before the initiation of the programme, formative research was conducted; interviews with a small number of PWIDs as well as discussions with key informants were performed in order to collect information that was deemed necessary for ensuring appropriate implementation (Supporting information, Appendix S2).

‘Seek’

The ‘seek’ part of the programme, i.e. the recruitment of participants, was performed using RDS. RDS is a type of chain-referral sampling where individuals receive coupons and are asked to draw from their existing social networks to identify up to three potential recruits, who then present themselves to the programme site [10,11]. Recruitment begins with a limited number of initial recruits (‘seeds’), and a dual incentive system is used in which participants receive incentives for participating in the programme (primary incentives) as well as for recruiting others (secondary incentives). Unique numbers are recorded on the coupons and allow the recruiter to be linked to his/her recruits.

ARISTOTLE was designed to run in five consecutive RDS rounds with a short break between rounds, and a recruitment target of approximately 1400 PWIDs per round. PWIDs could participate in multiple rounds, but only once in each round. Primary incentives were set to €5 and secondary incentives to €3 per each successfully recruited PWID (up to three recruits).

Respondents were eligible to participate to the programme if they presented a valid coupon, had injected drugs without a prescription in the past 12 months, were ≥ 18 years of age, lived in the Athens Metropolitan area and had not participated previously in the current sampling round. Seeds were recruited by the staff of OKANA. The criteria for selecting seeds are reported in more detail in the Supporting information (Appendix S3).

People arriving at the site with a valid coupon were screened for eligibility before participation. To verify self-reported PWID status, interviewers asked respondents to show injection marks and to provide answers to a short questionnaire (eligibility screener). Eligible people were then informed about the programme and were asked whether they agreed to participate. If informed consent was obtained, they were enrolled. From round B, biometric measurements were also performed in both arms (arm length, wrist circumference) to prevent repeated enrolment in the same round [12,13].

‘Test’, collection of biobehavioural data and information on how to prevent HIV

After enrolment, PWIDs proceeded to the interview with the core questionnaire. Then, a blood sample for HIV testing was collected by the programme physician. At the end of this process, they received three recruitment coupons, their primary incentive as well as syringes, condoms and leaflets with information about how to prevent blood-borne infections (information materials were also translated into multiple languages). Participants were asked to return in a few days to collect their secondary incentives and their HIV test result.

Computer-assisted personal interviewing was used for the interviews. The questionnaire from the National HIV Behavioral Surveillance System (NHBS) for PWIDs—that is implemented throughout the United States—was used as the basis for the questionnaire of ARISTOTLE [14]. This was modified appropriately in order to be administered to PWIDs in Greece, while ensuring that key infection-related behavioural indicators were covered [15]. It included an eligibility screener, as well as a core questionnaire with sections on injecting network size, demographics, injecting and non-injecting drug use history, sexual behaviour, alcohol use, alcohol and drug treatment, HIV testing experience, health conditions, imprisonments and assessment of prevention activities. The size of the participants’ injecting network was assessed by asking them to recall how many people they knew in Athens who inject and they have seen in the past 30 days. The questionnaire was enriched with additional sections in subsequent rounds (modules on knowledge and attitude for recent HIV infection, food insecurity, competitiveness, altruism and solidarity scales).

Collected blood samples were transported on a daily basis to the National Retrovirus Reference Centre for testing. HIV tests were performed with a microparticle EIA anti-HIV-1/2 (AxSYM HIV-1/2 gO; Abbott Laboratories, Abbott Park, IL) and HIV-1 and HIV-2 confirmation by Western blot (MP Diagnostics, Singapore). Additional testing was also performed, such as HIV-1 LAg-Avidity EIA (SEDIA Biosciences Corporation, Portland, OR, USA) and phylogenetic analysis of HIV-1 sequences; these results will be reported elsewhere.

‘Treat’/‘retain’: linkage and retention to care and treatment

The ‘treat’ part of the programme consisted of linking seropositive PWIDs to infectious diseases units for medical care, ART and to prioritized opioid substitution treatment (OST) programmes, if they so wished. HIV test results were available within 3 days from blood sample collection. To maximize the number of participants returning to collect their HIV test result, a tertiary incentive of €3 was provided from round B. The programme physician informed the participants about their HIV status and provided a brief counselling session. The programme psychologist had a counselling session with seropositive participants. An experienced volunteer from the NGO ‘Positive Voice’ was located in the same building to assist the counselling. Recommendation and prioritization of ART and OST was provided to all HIV seropositive participants. If the participant expressed an interest in HIV treatment or OST, dedicated linkage staff arranged the appointments with the doctors of the infectious diseases units (for HIV treatment) and the Organization Against Drugs (for OST). Seropositive migrants without documents were referred to NGO Praxis. Data on linkage and retention to ART and OST treatment are the focus of another report.

Sample size—statistical methods

The required sample size per round was based on what would be needed to estimate HIV prevalence with a given precision. In RDS, sample size estimation takes into account the design effect which is a measure of RDS efficiency compared to a simple random sample [16]. Using the formula $n=\mathit{design}\mathit{effect}\ast \frac{p^{\ast }(1-p)}{se{(p)}^2}$ and assuming that HIV prevalence (p) in PWIDs would be approximately 10%, allowing for a design effect of 2.0 and a precision of 2.2%, the required sample size per round was 1400 participants [16].

The data presented in this paper cover all PWIDs who participated in the programme. Descriptive analysis assessed participants’ characteristics and HIV prevalence in the sample; crude proportions are reported. The analysis assessing the association of participants’ characteristics with HIV serostatus accounted for RDS sampling methodology. More specifically, bivariate and multivariable analyses accounted for correlation between recruiter and recruits using generalized estimating equations (GEE) logistic regression using the STATA version 11.0 software package [17].We created a variable indicating who the recruiter of each subject was and used this as a cluster variable in the GEE algorithm. An exchangeable correlation structure within each cluster was assumed. Multivariable analysis also accounted for differential recruitment effectiveness by HIV status and for the differing sample inclusion probabilities. We calculated inverse probability weights based on individualized recruitment weights and included them as a covariate in the logistic regression models. These weights combine the individualized degree component (the inverse of the network size of the participant) with an adjustment for differential recruitment [18], and were derived via RDS Analytical Tool (RDSAT) [19].We also adjusted for the different rounds of the programme by including an indicator variable in the model. In assessing HIV prevalence and associated risk factors, seeds were excluded.

Variables considered in the multivariable models were all those with P<0.25 in the bivariate analyses. Variables retained in the final models were age, sex, education, homelessness, main substance of use, duration of injecting drug use, frequency of injecting drug use, sharing syringes, RDS round and RDS weights as well as all those variables with likelihood ratio test P-values of less than 0.1.

Ethical issues

The survey protocol and informed consent form were approved by the Institutional Review Board of Athens University Medical School. People who were eligible to participate based on the eligibility screening process were asked to provide written informed consent. The informed consent form included information about the programme, explained that confidentiality would be protected and that participants were free to withdraw their consent at any point of the process. It also allowed for verbal informed consent, in which case the interviewer would sign the form. Participants were informed that the provision of their name would help to run the programme more efficiently; if they did not wish to provide their name, the two first initials of their name and surname were recorded. In that case, these initials along with the full birth date and information on gender were used to identify the participants uniquely.

The questionnaire and the blood sample were linked through the coupon number. The name of the participant was not recorded in the questionnaire database, on the blood sample or the HIV test result. The correspondence of the name and coupon number was recorded in a separate file that could only be accessed by authorized members of the research team.

RESULTS

Programme flow—number of participants

A graphical depiction of the implementation of the programme is shown in Fig. 1. Apart from the monetary incentives there were additional incentives and services provided to the participants, listed in detail in Supporting information, Table S1. The time needed to reach the target of 1400 PWIDs in each round ranged from 10 to 12.6 weeks (average: 11.7 weeks) with a median (25th, 75th) number of 28 [20,21] participants per day (Table 1). In total, 7113 questionnaires and blood samples were collected from 3320 unique participants within a period of approximately 13.5 months (excluding breaks between rounds). The reported place of residence of the recruits covered the entire Athens Metropolitan area (Fig. 2). Fifty-four per cent of the PWIDs participated in multiple rounds; 20.5, 14.1, 11.3 and 7.9% participated in two, three, four and all five rounds, respectively. The number of first-time and repeat participants as well as demographic characteristics and HIV prevalence per round are shown in Table 1.

Figure 1.

Graphical depiction of the ARISTOTLE programme¹

¹ created using PaT Plot tool

Table 1.

The five rounds of the ARISTOTLE programme: duration, number of participants and demographics.

	Round A	Round B	Round C	Round D	Round E
Time period of recruitment	20 August 2012–29 October 2012	4 December 2012–1 March 2013	19 March 2013–7 June 2013	17 June 2013–11 September 2013	19 September 2013–16 December 2013
Duration of recruitment (weeks)	10.0	12.4	11.4	12.3	12.6
Maximum number of recruitment waves achieved	9a	23	19	18	20
Number of seeds	11	6	5	7	6
Number of participants
All participants	1415	1444	1434	1413	1407
First-time participants, n (%)	1415 (100.0)	766 (53.1)	458 (31.9)	370 (26.2)	311 (22.1)
Repeat participants, n (%)	0 (0.0)	678 (46.9)	976 (68.1)	1043 (73.8)	1096 (77.9)
Age, mean (SD)
All participants	35.5 (7.8)	35.9 (8.2)	36.2 (8.1)	36.3 (8.0)	36.3 (7.9)
First-time participants	35.5 (7.8)	36.0 (8.8)	36.3 (8.8)	36.1 (8.7)	35.5 (8.2)
Repeat participants	–	35.7 (7.5)	36.2 (7.8)	36.5 (7.7)	36.6 (7.8)
Male, n (%)
All participants	1206 (85.2)	1198 (83.0)	1185 (82.6)	1153 (81.6)	1153 (82.0)
First-time participants	1206 (85.2)	631 (82.4)	392 (85.6)	316 (85.4)	261 (83.9)
Repeat participants	–	567 (83.6)	793 (81.3)	837 (80.3)	892 (81.4)

^aIn round A, a 10-digit number was allowed for in the coupons where the first one or two digits identified the seed and the remaining digits identified the waves. Thus, up to nine waves were allowed for in the recruitment. Since round B, this limit in the coupon numbering was removed. SD = standard deviation.

Figure 2.

Map of Athens marked with the areas where people who inject drugs (PWIDs) reported that they live at their first participation in the programme (green circles). The size of the circles is proportional to the number of participants reporting that area. The blue symbol indicates the location of the ARISTOTLE site, where questionnaires and blood samples were collected and coupons were distributed to participants

Socio-demographic and network characteristics of participants (first participation)

There were 2806 men and 513 women participating in the programme. Participants’ mean [standard deviation (SD)] age was 35.8 (8.3) years (range=18–67 years); 84.1% were Greek, 5.4% were from countries in the Balkan region and eastern Europe (mainly Albania, followed by Georgia, Bulgaria, Poland, Russia and Romania), 3.9% from Afghanistan/Iran, 3.0% from other countries in the Middle East and 2.9% from African countries (Table 2).

Table 2.

Socio-demographic, network, injecting drug use and sexual behaviour characteristics as well as access to prevention and treatment of 3320 PWIDs participating in ARISTOTLE, overall and according to gender (based on their first visit to the programme).

	Total (n = 3320)	Male (n = 2806)	Female (n = 513)	P
A. Socio-demographic characteristics
Age (years), mean (SD)	35.8 (8.3)	36.1 (8.3)	34.3 (8.3)	< 0.001
Country of origin, n (%)				< 0.001
Greece	2793 (84.1)	2327 (82.9)	465 (90.6)
Afghanistan/Iran	130 (3.9)	130 (4.6)	0 (0.0)
Middle East (other)	100 (3.0)	100 (3.6)	0 (0.0)
Balkans and eastern Europe	178 (5.4)	140 (5.0)	38 (7.4)
Africa	97 (2.9)	93 (3.3)	4 (0.8)
Other	22 (0.7)	16 (0.6)	6 (1.2)
Highest completed level of education, n (%)				< 0.001
Primary school or below	907 (27.6)	811 (29.2)	96 (18.8)
Middle/secondary school	1004 (30.5)	871 (31.4)	133 (26.0)
High school	972 (29.6)	799 (28.8)	173 (33.8)
University or equivalent	405 (12.3)	294 (10.6)	110 (21.5)
Homeless,a n (%)				0.073
No	2183 (65.9)	1862 (66.6)	320 (62.4)
Yes (not currently)	364 (11.0)	294 (10.5)	70 (13.7)
Yes (currently)	765 (23.1)	642 (22.9)	123 (24.0)
Currently health insurance, n (%) yes	1185 (35.9)	979 (35.1)	206 (40.4)	0.021
Employment status, n (%)				0.003
Employed full-time, part-time	403 (12.2)	362 (13.0)	41 (8.0)
Unable to work for health reasons, unemployed	2860 (86.6)	2397 (85.9)	463 (90.3)
Student, retired, other	39 (1.2)	30 (1.1)	9 (1.8)
History of imprisonment (ever), n (%) yes	1613 (48.8)	1413 (50.6)	200 (39.1)	< 0.001
B. Network
Size of participant’s network, n (%)				0.338
1–10	898 (27.3)	750 (26.9)	148 (29.3)
11–30	1104 (33.6)	930 (33.4)	174 (34.4)
>30	1289 (39.2)	1105 (39.7)	184 (36.4)
C. Injecting drug use behaviour
Main substance of use, n (%)				< 0.001
Heroin/Thai	2700 (81.9)	2318 (83.2)	381 (74.7)
Cocaine	462 (14.0)	361 (13.0)	101 (19.8)
Buprenorphine (non-prescribed)	25 (0.8)	23 (0.8)	2 (0.4)
Sisha, methamphetamine	17 (0.5)	16 (0.6)	1 (0.2)
Speedball	92 (2.8)	67 (2.4)	25 (4.9)
Injecting drug use in the past month, n (%) yes	2689 (81.2)	2281 (81.6)	407 (79.3)	0.232
Duration of injecting drug use (years), median (25th, 75th)	12 (6, 18)	13 (6, 19)	11 (6, 17)	0.005
Frequency of injecting drug use,a n (%)				0.005
More than once per day	1013 (30.6)	877 (31.3)	136 (26.5)
Once per day	247 (7.5)	218 (7.8)	29 (5.7)
At least once per week	763 (23.0)	641 (22.9)	121 (23.6)
At least once per month	489 (14.8)	416 (14.9)	73 (14.2)
Less than once per month	801 (24.2)	647 (23.1)	154 (30.0)
Sharing syringes,a n (%)				0.356
Never	1805 (59.7)	1536 (60.2)	268 (56.9)
Rarely	968 (32.0)	815 (31.9)	153 (32.5)
About half the time	138 (4.6)	110 (4.3)	28 (5.9)
Most of the time	87 (2.9)	70 (2.7)	17 (3.6)
Always	26 (0.8)	21 (0.8)	5 (1.1)
Use drugs divided with a syringe that someone else had already injected with,a n (%)				0.166
Never	2192 (66.6)	1867 (67.1)	324 (63.7)
Rarely	838 (25.4)	691 (24.8)	147 (28.9)
About half the time	155 (4.7)	136 (4.9)	19 (3.7)
Most of the time, always	109 (3.3)	90 (3.2)	19 (3.7)
D. Sexual behaviour
Same-sex contacts,a n (%) yes		54 (2.0)	37 (7.4)	< 0.001
Received money/drugs in exchange for sex (any partner),a,b n (%) yes		118 (5.3)	120 (27.1)	< 0.001
Last partner PWID (any partner),a,b n (%) definitely/probably not PWID		1191 (55.5)	131 (29.6)	< 0.001
E. Access to testing, treatment and prevention
Ever participated in a drug treatment programme, n (%) yes	2190 (66.2)	1808 (64.7)	381 (74.6)	< 0.001
Participated in an opioid substitution programme, n (%)				< 0.001
Yes, currently	409 (12.5)	325 (11.8)	84 (16.7)
Yes, in the past	440 (13.5)	332 (12.0)	108 (21.4)
Never	2422 (74.0)	2110 (87.1)	312 (61.9)
Received syringes through prevention activities,a n (%) yes	1675 (50.6)	1384 (49.5)	291 (56.7)	0.003
Ever tested for HIV, n (%) yes	2154 (65.4)	1758 (63.2)	395 (77.5)	< 0.001

^aIn the past 12 months.

^bFor those reporting at least one sexual partner within the past 12 months. PWID = people who inject drugs; SD = standard deviation.

Almost half the participants had a history of imprisonment (48.8%); this was more pronounced among men. Twenty-three per cent were currently homeless. Those reporting currently living in a household were asked to provide information on the constitution of the household (number of people, presence of children, etc.) as well as the average monthly income of the household; 979 of 2547 PWIDs living in households provided these data. Using the poverty threshold of €5708 per person annually (Hellenic Statistical Authority with 2011 as the reference income period), 99.4% of PWIDs living in households were at risk of poverty, i.e. their equivalized disposable income was below that threshold.

The median (25th, 75th) size of their injection network— i.e. of PWIDs they knew and had met in the past month—was 30 [10,22] people.

Compared to male PWIDs, women were younger, more often of Greek origin, had a higher educational level, were more likely to be unemployed and less likely to have been imprisoned (Table 2). There was no difference in the income of male and female PWIDs.

Injecting drug use and sexual behaviour (first participation)

The main substance injected was heroin (82% of participants) (Table 2). The median duration of injecting drug use was 12 years, with 10.6% of participants being ‘new’ injectors (defined as duration of injecting drug use ≤ 2 years). A substantial proportion of participants (30.6%) injected drugs more than once per day, with a median of three injections in a usual day. Eight per cent of the participants reported that, in the past 12 months, they had used syringes with which someone else had already injected ‘about half the time or more’ and an additional 32% reported sharing a used syringe ‘rarely’.

Sex with a partner of the opposite sex in the past 12 months was reported by 79.1% of males and 84.6% of females, whereas same-sex contacts were reported by 1.9% of males and 7.2% of females. In opposite-sex contacts, use of condoms (‘usually yes’ or ‘always’) was reported more frequently by men than women (56.7 and 37.7%, respectively). Five per cent (5%) of male and 27% of female PWIDs reported having received money or drugs in exchange for sex in the past 12 months (P<0.001) (Table 2). A large proportion of men (55.5%) and 29.6% of women reported that their last sexual partner was most probably not a PWID.

Access to drug treatment, OST, NSP and HIV testing

Two-thirds (66.2%) had participated in a drug treatment programme in the past and 50% within the past 12 months. A large proportion (74.1%) had never entered OST (Table 2).

Fifty per cent of participants had received free syringes through health prevention activities carried out in Athens within the past year. The median (25th, 75th) number of syringes they had received during the month preceding their participation to Aristotle was 0 (0, 15) syringes (among those who received syringes: 20 [10,23] syringes).

Two-thirds (65.4%) of the participants reported having been tested for HIV in the past. Of 3320 participants, 129 (3.9%) were aware of a previous positive test result and 40 of 129 (31.0%) reported being currently on anti-retroviral treatment.

Compared to men, women were more likely to have participated in a drug treatment or in an OST programme, to have received free syringes within the past year and to have been tested for HIV, even after adjusting for differences in age, country of origin and HIV status (data not shown).

HIV prevalence and risk factors

Of 3308 participants (excluding seeds), 499 tested anti-HIV+ the first time they participated in the programme. The crude estimate of HIV prevalence was 15.1% [95% confidence interval (CI)=13.9%, 16.4%)]. HIV prevalence was 14.8 and 16.7% among men and women, respectively (P=0.279). ‘New’ injectors had an equally high prevalence compared to injectors with more than 2 years of injecting drug use (13.3 versus 15.0% in men, 17.0 and 16.7% in women, respectively) (Table 3).

Table 3.

Anti-HIV prevalence according to selected characteristics for male and female PWID at their first participation to ARISTOTLE along with crude and adjusted odds ratios for the risk of HIV infection (excluding seeds).

	Male PWIDs (n = 2797)			Female PWIDs (n = 510)
	Anti-HIV (+), n (%)	Bivariate analysis Crude ORa (95% CI)	Multivariable analysis Adjusted ORb (95% CI)	Anti-HIV (+), n (%)	Bivariate analysis Crude ORa (95% CI)	Multivariable analysis Adjusted ORb (95% CI)
A. Socio-demographic characteristics
Age		(per 10 year increase)	(per 10-year increase)		(per 10-year increase)	(per 10-year increase)
18–30	118 (18.2)	0.97 (0.96, 0.99)	0.98 (0.96, 0.99)	38 (22.9)	0.96 (0.93, 0.98)	0.97 (0.92, 1.02)
31–40	210 (15.2)			35 (15.2)
>40	86 (11.2)			12 (10.6)
Country of origin
Greece	324 (14.0)	1.00	1.00	73 (15.8)	1.00	–
Afghanistan/Iran	30 (23.1)	1.87 (1.22, 2.87)	0.77 (0.44, 1.34)	–	–
Middle East (other)	19 (19.2)	1.51 (0.91, 2.52)	0.81 (0.44, 1.50)	–	–
Balkans and eastern Europe	29 (20.9)	1.62 (1.05, 2.49)	1.16 (0.68, 1.98)	10 (26.3)	1.83 (0.85, 3.95)
Africa	10 (10.9)	0.78 (0.40, 1.51)	0.39 (0.17, 0.86)	1 (25.0)	–
Education
High school or higher	123 (11.3)	1.00	1.00	39 (13.8)	1.00	1.00
Primary school, middle/secondary school	288 (17.2)	1.63 (1.30, 2.05)	1.45 (1.12, 1.88)	46 (20.3)	1.54 (0.96, 2.48)	1.04 (0.56, 1.96)
Employment status
Employed full-time/part-time	38 (10.5)	1.00	–	5 (12.2)	1.00	–
Unable to work for health reasons, unemployed	374 (15.7)	1.58 (1.11, 2.25)		79 (17.2))	1.44 (0.55, 3.77)
Student, retired, other	2 (6.7)	0.70 (0.14, 2.65)		1 (11.1)	0.89 (0.09, 8.66)
Homeless in the past 12 months
No	210 (11.3)	1.00	1.00	37 (11.6)	1.00	1.00
Yes (not currently)	44 (15.0)	1.40 (0.98, 1.98)	1.21 (0.82, 1.78)	12 (17.7)	1.53 (0.73, 3.18)	1.09 (0.45, 2.62)
Yes (currently)	160 (25.1)	2.63 (2.09, 3.31)	1.89 (1.41, 2.52)	36 (29.3)	3.16 (1.88, 5.32)	1.50 (0.73, 3.07)
History of imprisonment (ever)
No	160 (11.6)	1.00	1.00	32 (10.4)	1.00	1.00
Yes	254 (18.1)	1.66 (1.34, 2.06)	1.27 (0.99, 1.64)	52 (26.2)	3.09 (1.90, 5.03)	2.76 (1.43, 5.31)
B. Injecting drug use behaviour
Duration of injecting drug use		(per 10-year increase)	(per 10-year increase)		(per 10-year increase)	(per 10 year increase)
≤ 2 years	39 (13.3)			9 (17.0)
>2 years	373 (15.0)	0.88 (0.78, 0.99)	1.00 (0.98, 1.02)	76 (16.7)	0.72 (0.53, 0.97)	0.97 (0.93, 1.01)
Main substance of use
Heroin/Thai	322 (13.9)	1.00	1.00	46 (12.1)	1.00	1.00
Cocaine	63 (17.5)	1.30 (0.97, 1.75)	1.74 (1.24, 2.45)	24 (24.0)	2.40 (1.37, 4.18)	2.25 (1.12, 4.49)
Buprenorphine (non-prescribed)	2 (8.7)	0.58 (0.13, 2.58)	0.60 (0.13, 2.79)	–	–	–
Sisha, methamphetamine	2 (12.5)	0.86 (0.19, 3.88)	0.92 (0.19, 4.35)	–	–	–
Speedball	23 (34.3)	3.28 (1.95, 5.51)	4.54 (2.48, 8.30)	12 (50.0)	7.16 (3.05, 16.8)	6.07 (2.06, 17.9)
Injecting drug use in the past month
No	44 (8.6)	1.00	–	11 (10.4)	1.00	–
Yes	369 (16.2)	2.07 (1.49, 2.88)		74 (18.3)	1.97 (0.99, 3.90)
Frequency of injecting drug use (past 12 months)
Once per day or less	159 (9.4)	1.00	1.00	43 (12.4)	1.00	1.00
More than once per day	255 (23.4)	2.97 (2.39, 3.68)	2.07 (1.60, 2.67)	42 (25.9)	2.48 (1.54, 3.99)	1.79 (0.95, 3.38)
Sharing syringes (past 12 months)
Never	175 (11.4)	1.00	1.00	42 (15.9)	1.00	1.00
Rarely	164 (20.2)	1.96 (1.55, 2.47)	1.62 (1.21, 2.16)	24 (15.7)	0.99 (0.57, 1.71)	0.54 (0.23, 1.25)
About half the time or more	50 (25.0)	2.58 (1.81, 3.69)	1.91 (1.26, 2.90)	17 (34.0)	2.74 (1.40, 5.35)	1.19 (0.45, 3.18)
Use drugs divided with a syringe that someone else had already injected with (past 12 months)
Never	208 (11.2)	1.00	1.00	43 (13.4)	1.00	1.00
Rarely	141 (20.5)	2.05 (1.62, 2.59)	1.30 (0.96, 1.77)	28 (19.2)	1.54 (0.91, 2.60)	1.90 (0.86, 4.20)
About half the time or more	61 (27.4)	2.96 (2.13, 4.11)	1.91 (0.98, 2.19)	13 (34.2)	3.38 (1.60, 7.09)	2.90 (1.01, 8.29)
C. Sexual behaviour
Number of sex partners (any partner, past 12 months)
0	128 (23.4)	1.94 (1.47, 2.57)	–	9 (15.8)	1.99 (0.85, 4.68)	2.66 (0.94, 7.50)
1	113 (13.8) 110	1.00		19 (9.0)	1.00	1.00
2–5	(11.6)	0.83 (0.63, 1.10)		18 (13.9)	1.65 (0.83, 3.29)	1.54 (0.70, 3.41)
>5	58 (13.0)	0.92 (0.65, 1.29)		37 (38.5)	6.36 (3.39, 12.0)	4.12 (1.93, 8.77)
Among those with ≥ 1 partner: received money/drugs in exchange for sex (past 12 months)
No	262 (12.5)	1.00	–	36 (11.2)	1.00	–
Yes	18 (15.4)	1.20 (0.71, 2.04)		40 (33.9)	4.00 (2.38, 6.72)
Among those with ≥ 1 partner: whether last partner was an PWID (past 12 months)
Definitely/probably PWID	135 (14.2)	1.00	–	38 (12.2)	1.00	–
Definitely/probably not PWID	139 (11.7)	0.80 (0.62, 1.04)		37 (28.9)	3.02 (1.81, 5.05)
Same-sex contacts (past 12 months)
No	402 (14.8)	1.00	–	72 (15.6)	1.00	–
Yes	9 (16.7)	1.20 (0.59, 2.46)		11 (29.7)	2.43 (1.15, 5.10)
D. Access to treatment
Ever participated in a drug treatment programme
Yes	240 (13.3)	1.00	1.00	59 (17.6)	1.00	–
No	173 (17.6)	1.39 (1.12, 1.72)	1.28 (0.99, 1.66)	25 (19.4)	1.22 (0.72, 2.06)

^aGenerated using generalized estimated equations (GEE) clustered on recruiter.

^bGenerated using GEE clustered on recruiter and adjusting for RDS individualized recruitment weights and programme round. PWID = people who inject drugs; OR = odds ratio; CI = confidence interval.

From multivariable analysis, factors associated independently with increased risk of HIV infection among men were younger age, lower level of education, being currently homeless, using cocaine or speedball as main substance of injecting drug use, injecting more than once per day in the past 12 months, sharing syringes in the past 12 months and using drugs divided with a syringe with which someone else had already injected in the past 12 months (Table 3). History of imprisonment and not having participated in a drug treatment programme in the past were associated marginally statistically significantly with increased risk of HIV. Factors associated independently with increased risk of HIV infection among women were history of imprisonment, using cocaine or speedball as main substance of injecting drug use, using drugs divided with a syringe with which someone else had already injected in the past 12 months and higher number of sexual partners (Table 3). Women with more than five partners in the past year had a 4.1 times higher risk of being HIV infected compared to those reporting one partner (95% CI=1.9, 8.8, P<0.001).

DISCUSSION

ARISTOTLE recruited 3320 PWIDs over the course of 16 months in the context of an HIV outbreak in Athens. To our knowledge, this is the largest RDS programme ever performed in one city [11,20,24–27]. It is unique, as it allowed rapid identification of a hidden population experiencing an outbreak and provide HIV testing, counselling and linkage to care. Based on the number of ARISTOTLE participants who reported having injected drugs in the previous month and the corresponding capture–recapture population estimate in the Athens Metropolitan area in 2011 [28], the programme coverage was 96%. However, this may be an overestimate, and a study aiming to reassess the number of drug injectors in Athens is in progress (C. Richardson, personal communication). ARISTOTLE is also unique in the number of repeated contacts with many subjects within the time-period of the study. The high number of recruitment waves in each round suggests that the programme reached deep into the target population [29].

The data obtained from the first contact of PWIDs with the programme allowed us to assess HIV prevalence, access to prevention, their socio-demographic characteristics and risk behaviours. According to 2010 sentinel data on PWIDs entering drug treatment or accessing low-threshold services, HIV prevalence among PWIDs in Athens before the outbreak was 0.8% [4]. The prevalence reported in this paper based on data from the second half of 2012 and 2013 was 15.1%. The finding that ‘new’ injectors had an equally high HIV prevalence compared to injectors with more than 2 years of injecting drug use confirms the sudden spread of HIV [4,30].

Despite the reduction in the mean waiting time to enter OST in Athens (from 90 months in September 2011 to 46.5 months in August 2012) and the expansion of NSP since the start of the HIV outbreak [31], the coverage of these programmes, as reported by ARISTOTLE participants, was low.

As ARISTOTLE participants were, in the majority, infected within 1–2 years preceding their participation [3,4], evaluation of the association of their socio-demographic characteristics, injecting and sexual behaviour with the risk of HIV shows a clearer picture of the drivers of this epidemic. Homelessness was associated independently with increased risk of HIV infection among male PWIDs, as has been reported by others [21,32–38]. The fact that unstable housing is a risk factor independently of high-risk behaviour points to the network characteristics of homeless PWIDs; Friedman et al. have shown that homeless people were more likely to be in the core of large connected components and argued that this is the locus of HIV transmission, in case an epidemic occurs in the population [39]. HIV prevalence was much higher among male participants from Afghanistan/Iran and Balkans/eastern Europe compared to those from Greece. These differences disappeared in the multivariable analysis, as also reported by another study [40]. The variable whose adjustment in the model caused the largest change in the effect measurement of PWIDs from Afghanistan/Iran versus Greece was homelessness. It is noteworthy that 72% of PWIDs from Afghanistan/Iran were homeless at the time they participated in the programme. Male and female PWIDs differed in their socio-demographic characteristics, injection and sexual behaviours, as well as their access to treatment and prevention. In addition, there were gender differences in the identified risk factors for HIV infection. Among women, there was an association between the number of sexual partners and the risk of HIV infection, which corroborates the argument that the sexual behaviour of female PWIDs seems to be as important a component in HIV transmission [23,41]. History of imprisonment was also found to be associated more strongly with HIV risk among women compared to men.

ARISTOTLE was designed as an intervention programme to control the HIV-1 outbreak among PWIDs in the Athens Metropolitan area. The large size of this study in an area with very few HIV transmissions among PWIDs up to 2011 may mitigate some of the causal-order limitations observed in cross-sectional studies based on prevalence instead of incidence. In addition, the large size and the high recruitment rate we have reached in our study may reduce biases related to RDS design, such as homophily and differential recruitment. However, the study population cannot be considered strictly as representative of the whole PWID population of the Athens Metropolitan area as more affluent PWIDs are probably less likely to participate in programmes conducted in one of the less privileged areas in downtown Athens.

The outbreak of HIV infection in Athens presents some unique characteristics. It is one of the rare epidemics among PWIDS occurring since the mid-1990s in a European city outside the former Soviet Union. After the early epidemics of HIV among PWID in Europe in the 1980/90s, many western European cities rapidly introduced measures such as NSP and OST programmes that may have either averted PWID epidemics or brought them under control [42–44]. Furthermore, it is not related to increasing rates of injecting drug use, as was the case of outbreaks in eastern European cities in the mid to late 1990s [45–48]. The long median duration of injecting drug use and the small proportion of ‘new’ injectors in our participants suggest that the HIV outbreak occurred among an established population of PWIDs. It is also different from the outbreak occurring during the same period in Bucharest, Romania [49]. Romania experienced the appearance of new stimulant drugs on the market which were initially legally available; this resulted in more frequent injecting among traditional opiate users which, in the context of low prevention coverage, led to this outbreak [50]. In Athens, austerity and the economic crisis resulted in increased risk behaviours in the context of consistently low OST and NSP provision [4].

This analysis verifies and extends previous observations on the risk environment of HIV transmission in PWIDs and underline that behavioural and structural factors needed to be addressed. In the Athens outbreak, apart from increasing the coverage of NSP and OST, it is necessary to address cocaine and speedball use as well as homelessness, which affects mainly migrant PWIDs, and to tackle the increased sexual transmission risk of HIV among female PWIDs. Prevention programmes should also take into account non-injecting sexual partners of PWIDs, who could serve as a ‘bridging population’ for transmission to the wider heterosexual population [22,51–53]. The implementation of ARISTOTLE suggests that repeat RDS sampling may be an effective way to reach high coverage of interventions in the hard-to-reach population of PWIDs within a short period of time. This is of paramount importance for countries experiencing similar epidemics.

Supplementary Material

Supporting Information

Appendix S1 Training of the personnel

Appendix S2 Formative research

Appendix S3 Selection of seeds

Table S1 Services and incentives provided in ARISTOTLE programme.