AIDS and Behavior

, Volume 17, Issue 3, pp 879–888

Risk for Heterosexual HIV Transmission Among Non-Injecting Female Partners of Injection Drug Users in Estonia

Authors

    • Department of Public HealthUniversity of Tartu
  • J. M. McMahon
    • School of NursingUniversity of Rochester
  • M. Kals
    • Department of Public HealthUniversity of Tartu
  • A. Talu
    • Estonian Drug Monitoring CenterNational Institute for Health Development
  • K. Abel-Ollo
    • Estonian Drug Monitoring CenterNational Institute for Health Development
  • K. Rüütel
    • Department of Infectious Diseases Surveillance and Drug Abuse PreventionNational Institute for Health Development
  • D. C. Des Jarlais
    • Baron Edmond de Rothschild Chemical Dependency InstituteBeth Israel Medical Center
Original Paper

DOI: 10.1007/s10461-011-0078-9

Cite this article as:
Uusküla, A., McMahon, J.M., Kals, M. et al. AIDS Behav (2013) 17: 879. doi:10.1007/s10461-011-0078-9

Abstract

The HIV epidemic in Estonia, as with other eastern European countries, is currently concentrated among injection drug users (IDUs). Non-IDUs who have IDU sex partners could serve as a potential bridge in an expanding epidemic. We applied HIV transmission modelling to data collected from non-IDU/IDU heterosexual couples in Kohtla-Järve, Estonia to estimate HIV risk from IDUs to their sex partners based on self-reported sexual behaviors shared by the couple. IDUs and their current main non-injecting sex partners were recruited for an interviewer-administered survey and HIV testing. Bernoulli modelling techniques were applied to estimate the risk of HIV transmission (incidence) among HIV negative non-injecting female partners of male IDUs. The estimated HIV incidence in this population of non-injecting women with only main sexual partners in the last 6 months ranged from 3.24 to 4.94 HIV seroconversions per 100 person years depending on the value used in the models for the per act transmission rate during acute stage infection. Non-IDUs who have IDU sex partners are at high risk for HIV and could serve as a potential bridge to a more generalized epidemic. Whether this might lead to an expansion of the HIV epidemic beyond core groups in Estonia or other Eastern European countries warrants closer study.

Keywords

HIV incidenceInjecting drug usersMain sexual partnersEstonia

Resumen

La epidemia de VIH en Estonia, al igual que otros países de la Europa oriental, se concentra actualmente en usuarios de drogas inyectables (UDI). Las no-UDI que tienen parejas UDI podrían ser un puente potencial de expansión de la epidemia. Hemos aplicado modelos de transmisión del VIH a los datos recopilados de parejas heterosexuales no-UDI/UDI en Kohtla-Järve, Estonia, para estimar el riesgo de transmisión de VIH de UDI a sus parejas sexuales en base a las conductas sexuales reportadas voluntariamente por la pareja. Los UDI y sus parejas sexuales principales actuales no inyectadoras fueron reclutados para una encuesta por entrevistador y para prueba de VIH. Se aplicó la técnica del modelo de Bernoulli para estimar el riesgo de transmisión del VIH (incidencia) entre parejas femeninas VIH-negativas no inyectadoras de UDI masculinos. La incidencia del VIH estimada en esta población de mujeres no inyectadoras con sólo parejas sexuales principales en los últimos 6 meses osciló entre 3.24 y 4,94 seroconversiones de VIH por 100 años-persona (a-p), dependiendo del valor utilizado en los modelos de proporción de transmisión por acto durante la etapa de infección aguda. Las no-UDI con parejas sexuales UDI corren alto riesgo de infección de VIH y podrían servir como un puente potencial para una epidemia más generalizada. Se necesita un estudio más detallado para determinar si esto podría conducir a una expansión de la epidemia de VIH más allá de los grupos principales de Estonia u otros países de Europa del Este.

Palabras Clave

Incidencia del VIHUsuarios de drogas inyectablesParejas sexuales principalesEstonia

Introduction

Injection drug use accounts for over 60% of all human immunodeficiency virus (HIV) infections in the WHO East European region [1]. In Estonia, the HIV epidemic has been driven by injection drug use that has led to the highest incidence of HIV infection in Europe—with the number of new cases increasing from 0.7 per 100,000 in 1998 to 51 per 100,000 in 2004 and 31 per 100,000 in 2009 [1] and the highest prevalence of HIV infection among adults aged 15–49 in Europe (1.2% (1–1.5%) in 2009) [2]. The estimated prevalence of injection drug use is 2.4% among 15–44 year-olds in Estonia [3]. Recent studies among injection drug users (IDUs) conducted in the capital city of Tallinn have revealed high prevalence of HIV (~50%) [46].

Concurrently, in WHO East European region countries, the proportion of new HIV infections acquired through heterosexual contact has also been growing steadily, from 4.5% [7] in 2000 to 37% in 2006 [8] and 45% in 2008 [1]. Following this trend, while injection drug use remains the predominant mode of HIV transmission in Estonia, heterosexual contact has increased as a risk factor for new cases, especially among women [9]. Although the contribution of cases acquired by sexual contact with high-risk partners such as IDUs is not known, characteristics of the sexual networks of IDUs may be important in determining the evolution of the HIV/AIDS epidemics in Estonia and other Eastern European countries [10].

Studies from Eastern Europe (Russia, Ukraine, Estonia, Latvia) have found that the majority (~80%) of IDUs are male, that they are sexually active, having both injecting and non-injecting female partners, and that condom use is infrequent [1113]. Thus, given the high prevalence of HIV among IDUs [46] and their elevated sexual risk behavior [1113], non-injecting female partners of male IDUs may be at substantial risk for heterosexual transmission of HIV. IDUs may serve as a primary conduit for the introduction of infection into the heterosexual non-drug using population. There is limited behavioral and epidemiological data on these bridge populations from recent epidemics in Eastern Europe and interventions targeting these risk groups are generally absent [14]. Understanding the epidemiology and risk factors associated with sexual transmission of HIV among IDUs and their non-injecting heterosexual partners is an important research need that will help inform the development of effective prevention efforts.

In the present study, we determined the prevalence of HIV in a sample of IDUs and their non-injecting main sex partners and we applied HIV transmission modelling to data collected from these heterosexual couples in Kohtla-Järve, Estonia to estimate HIV risk from IDUs to their sex partners based on self-reported sexual behaviors shared by the couple.

Methods

In 2007, a cross sectional study designed to assess HIV prevalence and risk behaviors among current IDUs was conducted using respondent driven sampling (RDS) and complemented by the co-recruitment of main sexual partners of index IDUs. The study was conducted in Kohtla-Järve, a city in the North-east county of Estonia characterized by the highest numbers of reported new HIV cases in the country—182 per 100,000 population in 2007 [15].

RDS recruitment was executed through a coupon system. We used coupons of different color for IDUs and their main sexual partners. Each IDU participating in the study was asked to recruit up to three other IDUs and those reporting a main sexual partner were asked to recruit that partner. Six seed participants from diverse IDU categories (male, female, HIV infected/uninfected, using opiates/amphetamine as a main drug injected) were used to initiate recruitment. The use of coupons enabled recruiting networks (IDU-IDU and IDU-sexual partner) to be traced. RDS generally requires 4–6 waves of recruitment by seed participants to reach sample equilibrium, although this may have been achieved in our sample with a smaller number of waves (2–3) due to the homogeneity of the IDU population in the study location.

Study recruitment was executed in two sites. Site 1 involved recruitment and data collection for index IDUs. A syringe exchange program (SEP) was selected for this site based on the client population, research experience of staff, and location (in one of the residential areas of the city). Site 2 was used for recruitment and data collection of non-injecting main sex partners of index IDUs. Facilities for this site were rented from an out-patient clinic that provides regular health care (not focusing on IDU or HIV) and is centrally located in the city.

IDUs were eligible for the study if they reported injecting drugs in the past 2 months, spoke Estonian or Russian, were 18 years of age or older, and capable of providing informed consent. Participants’ IDU status was verified by examining the skin for injection marks and/or asking respondents to describe the process of preparing drugs for injection. RDS methods were employed to help IDU participants recruit their peers (current IDU) as well as their main sexual partners into the study. Non-injecting sexual partners were eligible for study participation if they spoke Estonian or Russian, were 18 years of age or older, a main sexual partner for an index IDU (“someone who is your most important regular sex partner, defined as husband/wife, common-law husband/wife, or steady boy/girlfriend of at least 6 months”), never injected drugs, and capable of providing informed consent. Self report and examination of skin for the absence of injection marks were used to verify the never-injecting status of sexual partners.

Since couples (index IDUs and their main sex partners) in our study were not required to be legally married, legal documentation, such as a marriage license, was not used to verify primary couple status. A sexual partner verification screening (SPVS) instrument was used to confirm the main heterosexual partnership status of IDUs and their non-injecting partners [16]. The SPVS tool assessed main relationship status by comparing the responses given by each member of the dyad on a series of questions of an intimate or sexual nature. Ten different versions or sets of six questions were used. All 10 SPVS question sets included the question on the timing of the last three acts of sexual intercourse, as well as an additional five questions drawn from a pool of 19 total questions. IDU and sexual partner responses to these questions were then compared for consistency. SPVS was considered successfully passed if a minimum of five out of six responses exactly matched.

IDU Sex Partner Recruitment

Four different strategies or options for recruiting sex partners, adapted from Witte and colleagues [17], were described to each index IDU. These included:
  1. 1.

    Brokering approach: the IDU described the study to their sex partner and attempted to enlist the partner into the study on his/her own.

     
  2. 2.

    Brokering plus invitation: the IDU hand-delivered a formal invitation letter on institutional letterhead addressed to his/her partner. The letter introduced the study, described its purpose, and provided a telephone number to call for more information if desired. Also, the letter requested that the partner call or visit the study site as soon as possible.

     
  3. 3.

    Phone co-recruitment: the IDU called their partner from the study office to extend an invitation to participate, with a staff member standing by to answer questions or provide details to the partner.

     
  4. 4.

    Face-to-face co-recruitment: the IDU asked the partner to come to the recruitment site (with or without him/her) to discuss the project in greater detail with a study staff member.

     

IDUs who completed the study received a primary incentive for study participation (gift coupon for grocery store, value $10) and a secondary incentive for each eligible person they successfully recruited into the study (gift coupon for a grocery store, value $5 for each peer and sexual partner). Sexual partners received a grocery store gift coupon worth $20.

Pre- and post- HIV counselling and testing was provided for study participants as a part of the study protocol.

The Ethics Review Board at the University of Tartu approved the study procedures.

Data Collection and Measures

Structured face-to-face private interviews were administered in Estonian or Russian language covering demographics, drug use history, sexual behavior (frequency of condom-protected and unprotected vaginal and anal intercourse within the last 4 weeks, by type of partner: main and casual partners), injection-related risk behavior, HIV testing, and access and utilization of harm reduction and health care services. World Health Organization (WHO) Drug Injecting Study Phase II Survey version 2b (rev. 2) [18] was used for surveying both IDUs and their main sexual partners. For the latter, we retained the screening questions on drug use, but omitted IDU-related questions. Venous blood was collected from all participants and screened for anti-HIV antibodies using commercially available test kits (Abbott Laboratories, Abbott Park, Illinois, USA).

Statistical Analysis

Descriptive statistics were used to describe the sample of IDUs and their non-injecting main sexual partners, both at the individual and dyadic level. Point estimates for HIV prevalence based on biological test results were calculated separately for IDUs and their non-injecting sex partners, and 95% confidence intervals for the point estimates were determined using the Clopper–Pearson (Exact) method. In addition, Bernoulli modelling techniques [19, 20] were applied to estimate the risk of HIV transmission (incidence) among HIV negative non-injecting female partners of male IDUs. This technique estimates each woman’s risk of becoming HIV infected over a one-year period based on self-reported HIV sexual risk behaviors weighted by the transmission risk of the sexual behavior and the probability that an IDU partner is HIV infected. The Bernoulli model takes the form:
$$ {\text{HIVRISK}} = {{\uppi}}\left[ {1 - \left( {1 - \alpha_{v} } \right)^{vu} \left( {1 - \varepsilon \alpha_{v} } \right)^{vp} \left( {1 - \alpha_{a} } \right)^{au} \left( {1 - \varepsilon \alpha_{a} } \right)^{ap} } \right] $$
where π is the probability that the male IDU partner is HIV infected, α is the per-contact probability of HIV transmission (infectivity) from an HIV infected male to an uninfected female though vaginal (αv) or anal (αa) intercourse (both condom protected and unprotected), and ε is the estimated condom failure rate. The probability that a male IDU partner is infected was determined by his HIV test result and by observed annual HIV incidence rates among IDUs: IDU participants who tested positive for HIV were assigned a probability of infection of 1.0; IDUs that tested negative for HIV were assigned a probability of becoming infected over the next year of 0.265 for those injecting <3 years or 0.107 for those injecting <3 years. Per-contact infectivity rates for unprotected vaginal and anal intercourse varied according to stage of infection. IDU partners who tested HIV positive were assumed to be in the chronic stage of infection. Per-contact chronic stage infectivity rates were taken from the literature as follows: αv = 0.0008 (vaginal sex) [21], αa = 0.014 (anal sex) [22]. Male IDUs who tested negative for HIV were assigned HIV seroconversion probabilities (see above) with seroconversion occurring at mid-year followed by a 3 month period of acute stage infection. Sensitivity analysis was performed to determine how three different published estimates of per-contact vaginal sex acute stage infectivity (Wawer et al.: 0.008; Hollingsworth et al.: 0.021; and Pinkerton: 0.036) [2325] influenced the results. The per-contact anal sex acute stage infectivity rate of 0.18 reported by Leynaert et al. [26] was applied to all models. Other estimates in the formula are based on self-report and include the frequency of unprotected vaginal sex (vu), protected vaginal sex (vp), unprotected anal sex (au) and protected anal sex (ap), within main couples extrapolated over a 1 year period. A condom failure rate of ε = 0.10 was assumed [27]. The Bernoulli model generated a new composite HIV risk variable (HIVRISK) that quantified the probability of each HIV negative female in the sample becoming infected by a main male IDU partner over a 1 year period. The robust sample mean (using 10 percentile Windsorizing) of these individual HIV risk estimates thus provides a point estimate of HIV incidence (number of HIV seroconversions per 100 person years (py)) in this population of non-injecting women with primary male IDU partners. Bootstrapped sample distributions were used to estimate 95% confidence intervals around the point estimates. SAS ver. 9.2 was used to perform all statistical analyses.

Results

Socio-Demographic Characteristics of the Samples (Table 1)

The majority of IDUs were men (86%). Most of the male IDUs were Russian-speaking (97%), and had stable living conditions (living in a house or flat, 90%). The mean age was 27.1 years (SD 4.4, range 17–48 years), and over half had less than 9 years of education (56%). Half (48%) of male IDUs had a full or part-time job. The mean age at IDU initiation in the group was 18.3 years (SD 2.2, range 13–30) and the mean duration of the injection career was 8.8 years (SD 3.8, range 0–30). Daily (or more frequent) injection was reported by 34% of the male IDUs. Most were sexually active within the last 12 months (78%).
Table 1

Sample characteristics of male IDUs, male IDUs recruiting a non-injecting main sexual partner and female sexual partners, Kohtla-Järve, Estonia, 2007

Characteristics 

Male IDUs

IDUs recruiting female sexual partner (N = 69)

Non-injecting female sexual partners (N = 69)

(N = 298)

N

%

N

%

N

%

HIV status and testing history

 HIV-positive (%)

208

69.8

41

59.4

24

34.8

 Ever been tested for HIV

  No

71

23.8

18

26.1

38

55.1

  Yes

227

76.2

51

73.9

31

44.9

 Ever tested + for HIV

  No

88

29.5

26

37.7

22

31.9

  Yes

136

45.6

25

36.2

9

13.0

Socio-demographics

 Age

  <20

11

3.7

3

4.3

12

17.4

  20–24

71

23.8

12

17.4

26

37.7

  ≥25

216

72.5

54

78.3

31

44.9

 Marital status

  Married/cohabiting

54

18.1

21

30.4

45

65.2

  Single

216

72.5

41

59.4

18

26.1

  Else

20

6.7

3

4.3

6

8.7

 Education

  ≤9 years

168

56.4

32

46.4

12

17.4

  10–12 years

123

41.3

36

52.2

54

78.3

  >12 years

1

0.3

0

0.0

3

4.3

 Main source of income (last 6 months)

  Regular or temporary job

144

48.3

43

62.3

48

69.6

  Theft, robbing, or stealing

52

17.4

6

8.7

0

0.0

  Other

101

33.9

20

29.0

21

30.4

 Ethnicity

  Russian+

290

97.3

68

98.6

61

88.4

  Estonian

8

2.7

1

1.4

8

11.6

Drug use behavior

 Duration of IDU

  1–5 years

48

16.1

9

13.0

NA

 

  >5 years

250

83.9

60

87.0

  

 Injecting daily (last 4 weeks)

  <Daily

195

65.4

35

50.7

NA

 

  ≥Daily

102

34.2

33

47.8

  

 Shared needles/syringes (last 4 weeks)

  No

277

93.0

67

97.1

NA

 

  Yes

21

7.0

2

2.9

 

 

Sexual Partner Recruitment by Male IDU

Among male IDUs, 67% (199/298) reported having a main sexual partner in last 6 months; of these, 62% (124/199) reported exclusively non-injecting partners (two were not sure of their partner’s IDU status), 29% (57/199) exclusively injecting, and 8% (16/199) both injecting and non-injecting main sexual partners. Thus, 70% (140/199) of male IDUs with main partners reported an eligible non-injecting heterosexual partner.

A current non-injecting main sexual partner was successfully recruited into the study by 69 male-IDUs. Of the brokering approaches offered for sexual partner recruitment, most (78%) IDUs preferred approach 1 (engaging sexual partner participation on his/her own + written invitation), followed by the approach 2 (phone co-recruitment with the study staff) (12%), and approach 3 (engaging sexual partner participation on his/her own) (6%).

Sexual Partner Enrolment

Altogether 82 women were screened at the study Site 2 for participation, 69 (84%) were enrolled. The reasons for declining study participation were failing on the SPVS (N = 11) and failing non-IDU status screen (N = 2).

All of the enrolled sexual partners (N = 69) were women, most of whom were Russian speaking (88%) and lived in either a house or flat (100%). The mean age was 24.2 years (SD 4.7, range 18–37 years); a minority (17%) had less than 9 years of education; 70% had a permanent job; 18% of the main sexual partners of IDUs reported having both casual and main partners within the last 6 months(Table 2); 60% of the casual partners were reportedly also IDUs.
Table 2

Sexual behavior characteristics of male IDUs, IDUs recruiting non-injecting main sexual partner and female sexual partners, Kohtla-Järve, Estonia, 2007

Sexual behavior

Male IDUs

IDUs recruiting non-injecting main partner (N = 69) 

Non-injecting female sexual partners (N = 69)

(N = 298)

HIV + (N = 41)

HIV − (N = 28)

HIV + (N = 24)

HIV − (N = 45)

Number of sexual partners (last 12 months)

 1

119

39.9%

22

53.7%

15

53.6%

16

66.7%

27

60.0%

 >1

113

37.9%

14

34.1%

10

35.7%

8

33.3%

18

40.0%

Number of main heterosexual partners (last 6 months)

 1

173

58.1%

31

75.6%

22

78.6%

14

58.3%

37

82.2%

 >1

26

8.7%

4

9.8%

2

7.1%

8

33.3%

8

17.8%

Reporting both main and casual partner(s) (last 6 months)

 Yes

59

19.8%

4

9.8%

9

32.1%

7

29.2%

8

17.8%

Condom use, main partner (last 6 months)

 Always

139

46.6%

25

61.0%

8

28.6%

1

4.2%

19

42.2%

 Sometimes/else

42

14.1%

3

7.3%

6

21.4%

9

37.5%

13

28.9%

 Never

60

20.1%

7

17.1%

9

32.1%

12

50.0%

13

28.9%

Numbers of vaginal sex episodes, median (min, max) (last 4 weeks)

14 (0–30)

 

14 (0–28)

 

15 (4–28)

 

12 (0–25)

 

12 (4–25)

 

Proportion (%) protected vaginal sex episodes (last 4 weeks)

59.8%

 

71.4%

 

44.0%

 

9.1%

 

44.4%

 

Numbers of anal sex episodes, median (min, max) (last 4 weeks)

0 (0–15)

 

0 (0–2)

 

0 (0–15)

 

0 (0–10)

 

0 (0–20)

 

Proportion (%) of protected anal sex episodes (last 4 weeks)

37.0%

 

0.0%

 

28.6%

 

0.0%

 

33.3%

 

The point estimate for HIV prevalence among male IDUs is 69.8% (95% CI: 64.2, 75.0%; N = 298). The prevalence estimate for HIV infection among non-injecting female partners is 34.8% (95% CI: 23.7–47.2%; N = 69). Of the 69 couples studied, 52% were concordant by HIV serostatus (16 HIV+; 20 HIV− couples), and 48% discordant (25 HIV + male-IDU/HIV– female; 8 HIV– male-IDU/HIV + female couples).

Bernoulli modelling of HIV transmission risk was used to estimate each non-injecting female’s risk of becoming HIV infected over a one-year period, which was calculated for currently HIV negative women. The robust sample means of these individual HIV risk estimates provide point estimates of HIV incidence in this population of non-injecting women with only main IDU sexual partners in the last 6 months (N = 37 in this analysis). Table 3 presents three model-estimated HIV incidence rates based on three different approximations of per-contact acute-stage heterosexual transmission during vaginal sex: 3.24 HIV seroconversions per 100 py (95% CI: 2.21, 4.27) assuming an acute stage infectivity of 0.008 [23]; 4.29 HIV infections per 100 py (95% CI: 3.12, 5.46) assuming acute stage infectivity of 0.021 [24]; and 4.94 HIV infections per 100 py (95% CI: 3.79, 6.06) based on an acute stage transmission rate of 0.036 [25].
Table 3

Model-estimated HIV incidence per 100 person years among HIV negative non-injecting women with primary male IDU partners, by pattern of condom use, Kohtla-Järve, Estonia, 2007

 

Model estimated HIV incidence (95% CI)

Male-to-female per-act vaginal sex acute stage infectivity estimates

Overall (N = 37)

Condom use, always (N = 17)

Condom use, less than always (N = 20)

0.008 [23]

3.24 (2.21, 4.27)

1.06 (0.66, 1.71)

5.09 (3.73, 6.63)

0.021 [24]

4.29 (3.12, 5.46)

1.30 (0.88, 1.93)

6.85 (5.50, 8.14)

0.036 [25]

4.94 (3.79, 6.06)

1.61 (1.13, 2.22)

7.76 (6.73, 8.85)

Parameter estimates used in all models

 Chronic stage male-to-female per-act vaginal transmission rate: 0.0008

 Chronic stage male-to-female per-act anal transmission rate: 0.014

 Acute stage male-to-female per act anal transmission rate: 0.18

Further, there were substantial differences in estimated HIV incidence by condom use practices. HIV incidence was approximately five times greater among women reporting no or inconsistent condom use with main IDU partners compared to women reporting consistent condom use.

Discussion

Previous research has revealed high risk for HIV transmission by examining risk among non-IDUs who have IDU sexual partners [14] or the sexual behaviors of drug users [11, 12]. The present study combines behavioral and biological data from both IDUs and their main sex partners. This research focused on main partners due to the higher transmission potential derived from the repeated exposure and low condom use in those partnerships in comparison to other types of partnerships [2830]. We estimate a risk of between 3.24–4.94 HIV seroconversions per 100 py for non-IDU females with IDU-male partners recruited from north east Estonia.

The range of estimated incidence underscores the high potential impact that acute-stage HIV infection can have on sexual transmission and the spread of HIV among non-IDU female partners of male IDUs in Estonia. This is in part due to the high rates of HIV infection observed among IDUs, especially new IDUs in this region. Our findings thus support prior work indicating that acute-stage infection plays a disproportionate role in the sexual transmission of HIV [31, 32]. These data also show that consistent condom use can substantially reduce the risk of transmission even in the face of increased infectivity during the acute-stage.

These estimates suggest high HIV risk within this group of young heterosexual women, and are consistent with the observed high HIV prevalence among main non-injecting sexual partners of IDUs. The estimates are comparable with the estimated HIV incidences among women in the general population in some of the Sub-Saharan countries hardest hit by the HIV pandemic (HIV incidence 2.6 per 100 py among women in the general population in Zambia [33]; HIV incidence 4.7 per 100 py among women of reproductive age in Malawi and Zimbabwe [34]; HIV incidence 2.2 per 100 py among women aged 15–24 in South Africa [35]); with the incidences measured in placebo arms of the HIV prevention trials conducted in South Africa (3.8 per 100 person-years [36]; 9.1 per 100 person-years [37]); and with the HIV incidences reported among female-IDUs from developed countries (4.2 [38] and 3.7 [39] per 100 py in Vancouver and 1.1 per 100 py in Montreal, Canada [40]; 9.8 per 100 py in Valencia, Spain [41]). The model-based rates reported here, however, are significantly higher than those measured among heterosexual (non-injecting) women (accessing voluntary HIV counselling and testing services) from developed countries (0.62 per 100 py in Valencia, Spain [41]).

The analysis presented here is based on data from couples in which the female partner (non-IDU) reported only having a main sexual IDU partner in the last 6 months (N = 37). In the larger sample of 45 HIV negative women in the study, 18% had casual partners concomitant to their main partner within the specified time period, and the additional risk from these casual partners was not included in our risk estimate. In addition, concomitant partners are also important for potential transmission from the female partner to other male partners.

This study has several important limitations. Our HIV incidence estimates were based on self-reported risk behaviors and published transmission rates for these risk behaviors, both of which might incorporate measurement error. Further, the model did not take into account the role of sexually transmitted infections or other potential cofactors. This might have resulted in an underestimation of HIV incidence. In addition, due to the small sample size the point estimates for the model-derived incidence rates must be viewed with caution. This is partially addressed by the use of robust estimation methods and presentation of 95% confidence intervals around the point estimates. Moreover, the high model-based HIV incidence in this non-injecting group is consistent with high prevalence in the group and with observed HIV incidence in other groups of high-risk women.

Less than half of the male IDUs having an eligible partner successfully recruited these partners. Our analysis revealed a trend involving a difference in HIV prevalence between male IDUs who successfully recruited a main partner compared to overall male IDUs (61 vs. 70%, P = 0.07). Importantly, there were no significant differences in reporting always using condoms with the main partner between male IDUs recruiting a main sexual partner and male-IDU overall (47 vs. 45%, P = 0.86). It is unlikely that female partner sampling error substantially biased the risk estimates. Experiences of women with casual and sex trading partners were not included in this analysis. For those subgroups of women HIV incidence might differ from the estimate provided in this study.

Regardless of the procedures that were in place (self verification, checking for injection marks on the skin) it is possible that some of the women recruited as non-injecting sexual partners were actually ex-injectors, and some of the high HIV prevalence in this group of women is attributable to past IDU. However, this potential misclassification would not have affected our model-based incidence estimates since these estimates were based on more recent self-reported risk behavior of HIV uninfected women and their IDU partners. Other potential sources of bias associated with the sensitive (private and illegal) behaviors under investigation are socially desirable responses and recall bias. However, we made all attempts to minimize these biases with the carefully planned study design (anonymous participation, order of the topics in the questionnaire, time-frames of questions, selecting setting for study procedures and competent non-judgmental interviewers).

The literature on interventions targeting sexual practices of IDUs and non-injecting drug users suggests that individual interventions are of limited effectiveness [42, 43]. Research suggests that couples interventions addressing both partners’ motivations for engaging (or not) in safer sexual behaviors may be needed to address the needs of heterosexuals in relationships, particularly for those in close primary relationships [44, 45].

The model presented here was based on data from a single city in Estonia. We expect that the basic biological components of the model would generally apply to HIV transmission from IDUs to main sexual partners (who do not inject drugs). However, specific variables such as the HIV prevalence among IDUs, the prevalence of acute HIV infections among the IDUs, the frequencies of unsafe sexual behaviour between IDUs and their main sexual partners, are likely to vary considerably across different locations. Russians are an ethnic minority in Estonia (though they are a majority in the city in which this study was conducted), and being a member of an ethnic minority group is often associated with higher HIV prevalence among IDUs [46].

The HIV epidemics in several countries in Eastern Europe, which are currently concentrated among IDUs, are at risk for penetrating into the general population. Non-injecting (main) sexual partners of IDUs are at substantial risk for HIV infection. This coupled with the high prevalence of IDU in the adult population reported from several counties in the region [3, 47, 48] may lead to a substantial escalation of the epidemic. Characteristics of the sexual networks of IDUs and their main sexual partners may play a critical role in determining the further generalization of the HIV epidemics [49, 50]. Blanchard [51] emphasized the importance of understanding the variability and determinants associated with the trajectory of STI epidemics, and in recognizing the “inflection points” in the epidemic, which are characterized by a shift in the relative importance of different subpopulations for the continued growth and maintenance of the epidemic. Blanchard describes several scenarios in which initial expansion of an STI in a high-risk core group can spread to low-risk populations. Whether the epidemic expands beyond the core risk group and becomes somewhat independent of the core group is largely dependent on the structure of the risk networks among high- and low-risk subpopulations. In our sample, 20% of the non-injecting women reported casual partners in addition to a main-IDU partner, and 43% of these casual partners were non-IDU (data not shown). Non-IDUs who have IDU sex partners are at high risk for HIV and could serve as a potential bridge in an expanding epidemic. Whether this will lead to an expansion of the HIV epidemic beyond core groups in Estonia or other Eastern European countries warrants closer study. The methodology for sampling presented in the paper would be of use for future research in similar populations.

Acknowledgments

This research was supported in part through grant R01 DA 03574 from the US National Institute on Drug Abuse correct number, by the Global Fund to Fight AIDS, Tuberculosis and Malaria who funded this study through the National Institute for Health Development in Estonia (EST-202-G01-H-00), Norwegian Financial Mechanism/EEA (grant EE0016), US Civilian Research Development Foundation grant (ESX0-2722-TA-06), and the Target Financing of Estonian Ministry of Education and Research (grant SF0180060s09). Authors express their gratitude to the teams at the NGOs Me aitame sind, and to all participants of the study.

Conflict of Interest

There are no conflicts of interest.

Copyright information

© Springer Science+Business Media, LLC 2011