AIDS and Behavior

, Volume 14, Issue 1, pp 132–140

Correlates of Anal Intercourse Vary by Partner Type Among Substance-Using Women: Baseline Data from the UNITY Study

Authors

    • Laboratory of Infectious Disease PreventionNew York Blood Center
  • Donald R. Hoover
    • Rutgers University
  • Guozhen Xu
    • Laboratory of Infectious Disease PreventionNew York Blood Center
  • Victoria Frye
    • The New York Academy of Medicine
  • Mary H. Latka
    • Aurum Institute for Health Research
  • Debbie Lucy
    • Laboratory of Infectious Disease PreventionNew York Blood Center
  • Sebastian Bonner
    • The New York Academy of Medicine
Original Paper

DOI: 10.1007/s10461-008-9440-y

Cite this article as:
Koblin, B.A., Hoover, D.R., Xu, G. et al. AIDS Behav (2010) 14: 132. doi:10.1007/s10461-008-9440-y

Abstract

Data are limited about anal intercourse among women at risk of HIV infection. HIV-negative non-injection drug using women at sexual risk (N = 404) were recruited. At baseline, 41.7% reported anal intercourse in the prior 3 months; of these, 88.2% reported unprotected anal intercourse (UAI). Factors associated with UAI varied by partner type: UAI with a steady partner was associated with younger age, depressive symptoms, and experience of battering; UAI with casual partners was associated with younger age, cocaine use and negative outcome expectancies for condom use; UAI with exchange partners was associated with cocaine use, negative outcome expectancies for condom use and depressive symptoms. Younger women were more likely to report unprotected anal intercourse if they did not use birth control. Specific counseling messages are needed to identify and address this risk and associated factors, including partner relationships, substance use, birth control, mental health issues and domestic violence.

Keywords

WomenAnal intercourseHIVSubstance use

Introduction

Heterosexual transmission of HIV is a growing public health problem in the United States and is currently the second highest mode of HIV acquisition (Centers for Disease Control and Prevention 2007). Among women, heterosexual contact is the leading cause of newly reported HIV infection (Centers for Disease Control and Prevention 2007). Although vaginal intercourse is more common among heterosexuals, receptive anal intercourse carries the highest risk of HIV acquisition per act (Leynaert et al. 1998). A number of studies about anal intercourse among women have been published, but data are limited among women at high risk of HIV infection. Heterosexual anal intercourse is not uncommon; lifetime occurrence of anal intercourse has been reported to be 32% among college women (Flannery et al. 2003), 30% in a national survey of 15–44 year olds (Mosher et al. 2005) and 16–22% among urban young adults (Lindberg 1999; Misegades et al. 2001; Ompad et al. 2006). Other studies found a prevalence of recent anal intercourse of 22% among women at clinics for sexually transmitted infections (Satterwhite et al. 2007), 21% among HIV serodiscordant stable heterosexual couples (Skurnick et al. 1998), 12–25% among HIV positive women (Chirgwin et al. 1999; Kalichman 1999; Weinhardt et al. 2004; Wilson et al. 1999), 10–14% among HIV negative women (Chirgwin et al. 1999; Wilson et al. 1999), and 32% among high-risk HIV-negative women (Gross et al. 2000). In the few studies reporting on condom use with anal intercourse among women, condom use was infrequent (Flannery et al. 2003; Kalichman 1999; Lorvick et al. 2006).

Only a limited number of studies have examined correlates of heterosexual anal intercourse and even fewer have examined correlates of unprotected anal intercourse. Factors found to be associated with heterosexual anal intercourse include younger age, lower educational level, younger age at first intercourse, higher number of sex partners, exchanging sex for money, drugs or other services, unprotected vaginal sex, crack use, sexually transmitted infections, having a sex partner with a history of injection drug use, childhood sexual abuse and interpersonal violence (Champion et al. 2001; El Bassel et al. 2000, 2005; Flannery et al. 2003; Foxman et al. 1998; Gross et al. 2000; Misegades et al. 2001; Wingood and DiClemente 1997).

The UNITY study was a trial designed to test the efficacy of an enhanced HIV risk reduction intervention and an enhanced educational intervention to increase knowledge about HIV vaccine trials among non-injection drug using women at sexual risk of HIV infection. At baseline and follow-up visits, women completed questionnaires regarding sexual risk behaviors, including anal intercourse, and potential correlates of risk behaviors. These analyses utilized the baseline data from the UNITY study to provide a better understanding of the prevalence and correlates of anal intercourse, and specifically unprotected anal intercourse, among high-risk women.

Methods

Study Population

From March 2005 to June 2006, women in New York City were recruited for the UNITY study. The intent was to enroll women at sexual risk of HIV infection and who would be eligible for HIV vaccine efficacy trials. A variety of recruitment strategies were used to recruit this convenience sample of high risk women including street outreach throughout the South Bronx area, flyers placed in the local neighborhoods, and through referrals from previous studies, current study participants (snowball referral) or friends, community agencies and clinics (Koblin et al. 2007). The following eligibility criteria were used: tested HIV antibody negative; 18 years of age or older; reported non-injecting use of heroin, cocaine or crack cocaine in the previous 6 months; had at least one instance of vaginal sex without a condom in the previous 3 months; had no reported injection drug use in the previous 3 years; not currently pregnant; and had no intent to become pregnant in the next 12 months. Eligibility was determined by interviewer-administered questions.

Data Collection

The data for these analyses were those collected at baseline during the conduct of the UNITY study which was designed to test the efficacy of an enhanced HIV risk reduction intervention and an educational intervention to increase knowledge about HIV vaccine trials. After informed consent for the baseline visit was obtained, audio computer-assisted self-interviewing (ACASI) was used to collect data, including demographic characteristics, reproductive history, sexual behaviors, attitudes about safer sex, depressive symptoms, alcohol and drug use, childhood abuse and experience of battering in adulthood. For these analyses, a subset of measures was utilized as described below.

Measures

Demographic Characteristics

These variables included age, race/ethnicity and years of education. For the categorical analysis, age groups were constructed by subdividing the sample approximately into thirds (18–40, 41–45 and 46+ years). For the multivariate models, age was entered as a continuous variable in 5 year increments.

Sexual Risk Behavior Measures

Participants were asked about three types of male partners with whom they had vaginal or anal intercourse in the previous 3 months: steady partner (a man you had sex with that you feel closest to in your heart), exchange partners (men you had sex with who gave you money or drugs or other services for having sex with them) and casual partners (non-steady and non-exchange partners). For each type of partner, participants were asked about numbers of times they had had vaginal and anal intercourse and number of times a condom was used for each activity. Unprotected vaginal or anal intercourse was defined as not using a condom at least once for that activity.

Drug use Measures

Participants were asked the frequency of non-injection use of specific drugs in the previous 3 months on a seven-point scale (never, once a month or less, 2–3 times a month, once a week, 2–3 times a week, 4–6 times a week, every day). The drugs included in these analyses were crack cocaine, cocaine, and non-injection heroin. The frequency of use was categorized as never, occasional (once a week or less) or frequent (2–3 times a week or more). Alcohol use was defined as light use (3 or less drinks/day on no more than 1–2 days per week), moderate use (4–5 drinks per day on no more than 1–2 days per week or 1–5 drinks per day on 3–6 days per week or 1–3 drinks per day on a daily basis) or heavy use (4 or more drinks every day or 6 or more drinks on a typical day when drinking) (Colfax et al. 2004).

Other Potential Correlates

Questions about sterilization (ever had a hysterectomy or an operation to prevent pregnancy) were asked and women who were not sterilized were asked about use of any birth control methods in the previous 3 months (male or female condoms, spermicides, barrier methods, hormonal contraceptives, IUD, and/or withdrawl). Negative outcome expectancy for condom use was measured using four items (e.g., condoms ruin the mood) answered with a 5-point scale (strongly disagree, disagree, neither disagree or agree, agree, strongly agree) (alpha = 0.89) (Latka et al. 2006). A mean score was calculated for each participant and the scores were categorized into low (1.0–2.9), medium (3.0–3.9) or high (4.0–5.0) with a higher score indicating a higher level of negative outcome expectancies for condom use. Relationship power was measured by the Relationship Control sub-scale (12 items) (alpha = 0.86) (Pulerwitz et al. 2002). We used the standard 4-point response scale (strongly agree, agree, disagree, strongly disagree). A mean score was calculated for each participant and the scores were categorized into low (1.0–2.43), medium (≥2.431–2.82) or high (≥2.821–4.0) with a low score indicating poor relationship power. Depressive symptoms were measured using 18 of 20 items of the modified CES-D (Hann et al. 1999; Radloff 1977). Two items were inadvertently eliminated: “I felt that people disliked me” and “I could not get going”. The reliability of the 18 item scale was high (alpha = 0.85). The standard 4-point response scale was used (rarely or none of the time, some or little of the time, occasionally or moderate amount of time, most or all of the time). A sum score was calculated for each participant and the scores were categorized by subdividing the study sample approximately into thirds with a high score indicating more depressive symptoms. Women’s experience of battering was measured by 8 of 10 items in Bonomi’s scale (Bonomi et al. 2006). Two items were inadvertently eliminated: “My partner could scare me without laying a hand on me” and “My partner had a look that went straight through me and terrified me”. The reliability of the 8 item scale was high (alpha = 0.95). The standard 6-point response scale was used (strongly agree, somewhat agree, agree a little, disagree a little, somewhat disagree, strongly disagree). A sum score was calculated for each participant and the scores were categorized by subdividing the study sample approximately into thirds with a high score indicating more experience of battering. The entire questionnaire was piloted in the target population prior to use.

Statistical Analysis

We determined the proportion of women who reported any anal intercourse, any unprotected anal intercourse and unprotected anal intercourse with each partner type (steady, casual and exchange) at baseline. Contingency tables and exact tests compared the proportions of women reporting unprotected anal intercourse with each partner type by baseline demographics, sexual risk behaviors, use of birth control, drug use, and other potential correlates (outcome expectancies for condom use, depressive symptoms, relationship power, and women’s experience of battering). Three multivariate logistic regression models were constructed using forward selection to identify independently significant variables associated with unprotected anal intercourse with each partner type.

Results

From March 2005 through June 2006, 458 women came to the research site of whom 404 (88.2%) tested HIV antibody negative, were behaviorally eligible and had complete screening data. The mean age of the 404 participants was 42.2 years with only 9.2% of the women being 30 years of age or younger; most of the women were either African American (64.9%) or Latina (25.5%), and most (66.1%) had less than a high school education. Over half (54.1%) of the women were frequent users of crack cocaine, 29.3% were heavy alcohol users and 28.6% used cocaine frequently. With regard to sexual risk behaviors in the 3 months before the visit, 33.3% of women reported having ten or more male partners and 84.9% reported having at least one male partner with whom they received money or drugs for sex (exchange partner). Although one of the eligibility criteria during screening was to have unprotected vaginal intercourse, 10.0% of women did not report doing this via the ACASI interview at baseline.

Overall, 41.7% of women reported anal intercourse in the prior 3 months and of those women, 88.2% reported unprotected anal intercourse. Of the 159 women reporting both vaginal and anal intercourse, the median ratio of the frequency of vaginal to anal intercourse acts was 3 (25%:1.6; 75%:8.3); 11.2% of women had a ratio ≤1.0, indicating that they had at least as much or more anal than vaginal intercourse. Also, of the women reporting both vaginal and anal intercourse, 85.9% reported that both were unprotected, 9.5% reported that only vaginal sex was unprotected, 3.1% reported both were protected and 1.6% reported that only anal sex was unprotected.

Women were asked about the types of male sex partners (steady, casual and exchange) they had had in the previous 3 months. Women could report more than one type of partner; 288 of the women reported that they had a steady partner, 313 had casual partners and 343 had exchange partners. Complete data on anal intercourse were available for 256 (88.9%) of the women who reported having a steady partner, 294 (93.9%) of women who reported having casual partners and 320 (93.3%) of the women who reported having exchange partners. Among women who reported having a steady partner, 27.0% (95% CI: 21.6–32.8%) practiced unprotected anal intercourse with this partner; among women who reported having casual partners, 21.4% (95% CI: 16.9–26.6%) practiced unprotected anal intercourse with their casual partners; and among women who reported having exchange partners, 19.1% (95% CI: 14.9–23.8%) practiced unprotected anal intercourse with their exchange partners.

Correlates of Unprotected Anal Intercourse by Partner Type

Among women with a steady partner, those who were younger and Latina were more likely to report unprotected anal intercourse with their steady partner in bivariate analysis (Table 1). Unprotected anal intercourse with a steady partner was also significantly associated with frequent cocaine use, a high level of negative outcome expectancies for condom use, depressive symptoms, poorer relationship power and a high level of experience of battering (Table 2). Among women with casual partners, those not using birth control were more likely to report unprotected anal intercourse with casual partners. Unprotected anal intercourse with casual partners also was significantly associated with frequent cocaine use and a high level of negative outcome expectancies for condom use (Tables 1 and 2). Among women with exchange partners, those who were Latina, had a higher number of partners and not using birth control were more likely to report unprotected anal intercourse with exchange partners. Unprotected anal intercourse was also significantly associated with cocaine use, high level of negative outcome expectancies for condom use, and a high level of depressive symptoms (Tables 1 and 2).
Table 1

Unprotected anal intercourse (UAI) by demographics, number of partners and use of birth control, UNITY Study

Variable

UAI with…

Steady partner (N = 256)

Casual partners (N = 294)

Exchange partners (N = 320)

N

%

OR

95% CI

N

%

OR

95% CI

N

%

OR

95% CI

Total

69

27.0

  

63

21.4

  

61

19.1

  

Age

    18–40

37

40.2

3.5

1.7, 7.3

31

26.7

1.9

1.0, 3.8

28

23.3

1.5

0.8, 2.8

    41–45

19

22.9

1.6

0.7, 3.4

17

20.2

1.3

0.6, 2.9

14

15.6

0.9

0.4, 1.9

    46+

13

16.1

1.0

 

15

16.0

1.0

 

19

17.3

1.0

 

Race/ethnicity

    Black

36

21.4

1.0

 

36

18.8

1.0

 

32

15.5

1.0

 

    Latina

25

42.4

2.7

1.4, 5.1

21

26.9

1.6

0.9, 3.0

23

27.7

2.1

1.1, 3.8

    Mix/white/other

8

27.6

1.4

0.6, 3.4

6

25.0

1.4

0.5, 3.9

6

19.4

1.3

0.5, 3.4

Years of education

    <High school graduate

49

30.6

1.7

0.9, 3.1

48

23.9

1.7

0.9, 3.2

45

21.1

1.4

0.8, 2.7

    ≥High school graduate

19

20.7

1.0

 

14

15.9

1.0

 

16

15.7

1.0

 

In last 3 months…

No. of male partners

    0–9

45

24.3

1.0

 

32

18.3

1.0

 

29

14.8

1.0

 

    10+

24

35.3

1.7

0.9, 3.1

31

26.3

1.6

0.9, 2.8

32

26.2

2.0

1.2, 3.6

Use of birth control in last 3 months

    Sterilized

17

23.0

1.0

 

10

14.5

1.0

 

10

12.5

1.0

 

    Other methods

24

30.8

1.5

0.7, 3.1

16

15.7

1.1

0.5, 2.6

15

13.4

1.1

0.5, 2.6

    No birth control

27

26.7

1.2

0.6, 2.5

34

28.6

2.4

1.1, 5.1

35

28.2

2.8

1.3, 5.9

Table 2

Unprotected anal intercourse (UAI) by drug use and psychosocial variables, UNITY Study

Variable

UAI with…

Steady partner (N = 256)

Casual partners (N = 294)

Exchange partners (N = 320)

N

%

OR

95% CI

N

%

OR

95% CI

N

%

OR

95% CI

Total

69

27.0

  

63

21.4

  

61

19.1

  

In last 3 months…

Alcohol use

    None or Light

32

25.6

1.0

 

24

19.8

1.0

 

20

14.5

1.0

 

    Moderate

18

28.6

1.2

0.6, 2.3

15

21.7

1.1

0.5, 2.3

15

20.0

1.5

0.7, 3.1

    Heavy

19

29.7

1.2

0.6, 2.4

24

24.2

1.3

0.7, 2.5

25

24.3

1.9

1.0, 3.6

Crack use

    Never

15

36.6

1.0

 

13

30.2

1.0

 

11

27.5

1.0

 

    Occasional

21

26.9

0.6

0.3, 1.4

16

19.1

0.5

0.2, 1.3

17

18.1

0.6

0.2, 1.4

    Frequent

33

25.2

0.6

0.3, 1.2

34

21.0

0.6

0.3, 1.3

33

18.3

0.6

0.3, 1.3

Cocaine use

    Never

19

19.6

1.0

 

10

10.9

1.0

 

7

7.4

1.0

 

    Occasional

30

30.3

1.8

0.9, 3.5

21

19.4

2.0

0.9, 4.5

22

18.0

2.8

1.1, 6.8

    Frequent

20

36.4

2.3

1.1, 4.9

31

34.4

4.3

2.0, 9.5

31

31.3

5.7

2.4, 13.8

Heroin use

            

    Never

37

22.7

1.0

 

36

19.6

1.0

 

34

16.9

1.0

 

    Occasional

19

35.9

1.9

1.0, 3.7

14

26.4

1.5

0.7, 3.0

13

21.3

1.3

0.7, 2.7

    Frequent

13

34.2

1.8

0.8, 3.8

13

24.1

1.3

0.6, 2.7

14

25.5

1.7

0.8, 3.4

Outcome expectancies for condom use (high score = negative expectancies)

    Low

18

19.2

1.0

 

14

13.7

1.0

 

10

9.1

1.0

 

    Medium

25

26.0

1.5

0.7, 3.0

19

18.5

1.4

0.7, 3.0

20

17.0

2.0

0.9, 4.6

    High

26

40.6

2.9

1.4, 5.9

29

34.5

3.3

1.6, 6.8

28

31.8

4.7

2.1, 10.3

Depressive symptoms

    Low

12

12.9

1.0

 

12

14.3

1.0

 

9

9.9

1.0

 

    Medium

26

32.9

3.3

1.5, 7.1

21

21.7

1.7

0.8, 3.6

18

17.1

1.9

0.8, 4.4

    High

31

37.8

4.1

1.9, 8.7

30

26.8

2.2

1.0, 4.6

34

27.6

3.5

1.6, 7.7

Relationship power (low score = poor relationship power)

    Low

29

44.6

4.3

2.1, 8.6

22

26.8

1.7

0.9, 3.3

23

26.7

1.9

1.0, 3.7

    Medium

21

31.3

2.4

1.2, 4.9

18

23.7

1.5

0.7, 2.9

14

16.7

1.1

0.5, 2.2

    High

19

15.8

1.0

 

23

17.6

1.0

 

23

16.0

1.0

 

Experience of battering

    Low

17

15.7

1.0

 

15

14.7

1.0

 

18

15.9

1.0

 

    Medium

18

25.4

1.8

0.9, 3.8

23

24.7

1.9

0.9, 3.9

13

13.7

0.8

0.4, 1.8

    High

34

44.2

4.2

2.1, 8.4

25

25.3

2.0

1.0, 4.0

30

26.8

1.9

1.0, 3.7

Multivariate analysis for unprotected anal intercourse with each partner type is presented in Table 3. Women who were younger, had depressive symptoms and a high level of experience of battering were independently more likely to report unprotected anal intercourse with a steady partner. Women who were younger, used cocaine frequently and had a high level of negative outcome expectancies for condom use were independently more likely to report unprotected anal intercourse with casual partners. Women who used cocaine, had a high level of negative outcome expectancies for condom use and had a high level of depressive symptoms were independently more likely to report unprotected anal intercourse with exchange partners.
Table 3

Multivariate analysis of unprotected anal intercourse (UAI) by partner type, UNITY Study

Variable

UAI with steady partner

UAI with casual partners

UAI with exchange partners

OR

95% CI

OR

95% CI

OR

95% CI

Age (in 5 years increments)

0.66

0.54, 0.81

0.80

0.67, 0.96

 

Cocaine use

    Never

a

 

1.0

 

1.0

 

    Occasional

  

2.3

1.0, 5.4

3.0

1.1, 8.1

    Frequent

  

3.8

1.6, 9.0

5.0

1.9, 13.0

Outcome expectancies for condom use (high score = negative expectancies

    Low

 

1.0

 

1.0

 

    Medium

  

1.7

0.8, 3.8

2.0

0.8, 4.6

    High

  

3.2

1.5, 7.1

3.4

1.5, 7.8

Depressive symptoms

    Low

1.0

 

 

1.0

 

    Medium

2.5

1.1, 5.7

  

1.8

0.7, 4.6

    High

3.2

1.4, 7.3

  

2.9

1.2, 7.0

Experience of battering

    Low

1.0

 

 

 

    Medium

2.3

1.0, 5.2

    

    High

3.7

1.7, 8.0

    

Also included but eliminated: race/ethnicity, education, number of partners, birth control, alcohol use, relationship power

aVariable was also eliminated

We considered that younger women may be using anal intercourse as a means of birth control, a phenomenon observed among adolescents (Houston et al. 2007). Thus, we explored the association of unprotected anal intercourse with birth control use stratified by age (Table 4). Among the youngest women, the percent reporting unprotected anal intercourse was highest among those not using birth control, reaching statistical significance for those with casual or exchange partners. This pattern was not seen among older women.
Table 4

Relationship of age and anal intercourse by birth control use, UNITY Study

 

Percent reporting UAI with steady partners (N = 256)

Age group (years)

18–40

41–45

46+

Birth control method

    Sterilized

23.8

27.3

19.4

    Other methods in last 3 months

38.2

31.0

13.3

    No birth control in last 3 months

51.4

10.3

14.3

Exact test P-value

0.123

0.133

0.923

 

Percent reporting UAI with casual partners (N = 294)

Age group (years)

18–40

41–45

46+

Birth control method

    Sterilized

13.6

20.0

11.1

    Other methods in last 3 months

17.4

19.4

8.0

    No birth control in last 3 months

41.7

16.7

22.0

Exact test P-value

0.011

>0.999

0.279

 

Percent reporting UAI with exchange partners (N = 320)

Age group (years)

18–40

41–45

46+

Birth control method

    Sterilized

8.0

18.2

12.1

    Other methods in last 3 months

14.6

6.1

19.4

    No birth control in last 3 months

40.4

21.9

20.0

Exact test P-value

0.002

0.153

0.659

Discussion

Although a number of studies have examined the prevalence of anal intercourse among a variety of populations of women, this study is one of the few which addresses the prevalence of unprotected anal intercourse, considers partner type and focuses on substance-using women. A high proportion of women in this study reported recent anal intercourse, a proportion higher than that reported in most other studies of women from urban areas or at risk of HIV infection (Chirgwin et al. 1999; Gross et al. 2000; Lindberg 1999; Misegades et al. 2001; Ompad et al. 2006; Satterwhite et al. 2007; Wilson et al. 1999). This study supports previous findings that most of the time, condoms are not used for anal intercourse among women (Flannery et al. 2003; Misegades et al. 2001; Wilson et al. 2003). These findings may be related to the eligibility criteria used for this study (non-injecting use of heroin, cocaine or crack cocaine and unprotected vaginal sex) but emphasize the need to inquire about anal intercourse among substance using women and engage women in counseling about the associated risks.

Unprotected anal intercourse with steady partners was associated with depressive symptoms, experience of battering and was more likely among younger women. Other studies have reported an association of interpersonal violence and anal intercourse among women (Champion et al. 2001; El Bassel et al. 2005; Raj et al. 2006). Women in abusive relationships with a steady partner may be less able to negotiate condom use, as it is often perceived as a sign of infidelity or suspicion of infidelity, both of which may result in further abuse (El Bassel et al. 2000). This may be particularly true around anal sex, which carries no pregnancy risk, thus condom use is inherently aimed at disease prevention.

In contrast to steady partners, anal intercourse with casual partners and exchange partners was associated with drug use, in particular, cocaine. Other studies have found associations of heterosexual anal intercourse with crack use (Gross et al. 2000) and methamphetamine use by both partners (Zule et al. 2007). This study suggests that women may be engaging in sexual risk for drugs, as with exchange partners, or the use of a stimulant may facilitate risky sexual activities with new or one-time partners (Zule et al. 2007).

Negative beliefs about condoms were associated with unprotected anal intercourse with casual and exchange partners, as has been seen with condom use for vaginal and/or anal sex among women (Erickson et al. 1995; Latka et al. 2006), illustrating the importance of developing effective counseling messages to overcome such beliefs.

Exploratory analyses indicated that younger women were much more likely to be engaging in unprotected anal intercourse if they were not using one of the birth control methods asked about in the questionnaire. This finding suggests that women in the child-bearing age range may be using this method as a means of birth control as has been observed in adolescents (Houston et al. 2007), in particular with non-steady partners. This finding would need to be explored further in other studies, since, first, birth control did not stay in the multivariate model, second, we did not explicitly ask about the use of anal intercourse as a means of birth control and thus were unable to corroborate this finding with other questions in the questionnaire, and, lastly, other variables, such as drug use, were more important with regards to anal intercourse with non-steady partners. Furthermore, these findings may be subject to bias since this study excluded women who explicitly stated that they intended to become pregnant within the year. However, this bias may be minimal since 50–60% of births among women with a high school degree or less are reported to be unintended (Chandra et al. 2005).

There are several limitations to this study. The women in this study were not necessarily representative of non-injection drug using women in New York City. The eligibility criteria were established to enroll HIV negative women who were at high risk of HIV infection, primarily through sexual contact. Since risky behaviors tend to cluster, our eligibility criteria for women at high risk through their substance use and vaginal practices may have selected for women with a higher likelihood of engaging in other risky behaviors, such as anal sex. Thus, our selection criteria may overestimate the extent of anal sex among this group of women. We also did not measure situation-specific variables, and thus cannot determine if a correlate such as drug use was coincident with anal intercourse acts. Finally, these data are cross-sectional and thus represent correlations, but not causality. In particular, these analyses do not determine whether anal sex is antecedent or consequent to abuse or depression. However, other studies have found similar associations between depression, abuse, self-efficacy and HIV risk (Wagner et al. 2007). In particular, depression and self-esteem, but not self-efficacy for risk reduction, mediated the link between partner abuse and HIV risk through injection drug use, which suggests that counseling women who are depressed and abused may also mitigate HIV risk associated with unprotected anal sex.

In summary, these data illustrate a high prevalence of heterosexual anal intercourse among non-injection drug using women, and highlight the specific counseling messages needed to address this risk, including relationships to partners, substance use, birth control, attitudes about condoms, mental health issues and the potential for domestic violence.

Acknowledgments

The authors would like to thank Nikki Englert, Sean Lawrence, and Evelyn Rivera for their work and devotion in conducting this study, the Project ACHIEVE Community Advisory Board for their advice and contributions, and the study participants who gave their time and effort. This work was supported by a grant to the New York Blood Center from the National Institute on Drug Abuse, NIH (R01 DA017482).

Copyright information

© Springer Science+Business Media, LLC 2008