Efficacy of a Woman-Focused Intervention to Reduce HIV Risk and Increase Self-Sufficiency Among African American Crack Abusers

Abstract

Objectives. This study compares 3- and 6-month outcomes of a woman-focused HIV intervention for crack abusers, a revised National Institute on Drug Abuse standard intervention, and a control group.

Methods. Out-of-drug-treatment African American women (n = 620) who use crack participated in a randomized field experiment. Risk behavior, employment, and housing status were assessed with linear and logistic regression.

Results. All groups significantly reduced crack use and high-risk sex at each follow-up, but only woman-focused intervention participants consistently improved employment and housing status. Compared with control subjects at 6 months, woman-focused intervention participants were least likely to engage in unprotected sex; revised standard intervention women reported greatest reductions in crack use.

Conclusions. A woman-focused intervention can successfully reduce risk and facilitate employment and housing and may effectively reduce the frequency of unprotected sex in the longer term.

One of the devastating consequences of crack cocaine use among African American women is HIV prevalence rates, which range from approximately 1.7% among noninjecting drug users with no sex-trading history1 to 54% among homeless women who are more likely to trade sex for drugs.2–5 Although the Centers for Disease Control and Prevention does not track the relationship between crack use and HIV transmission, research indicates that crack use is associated with increased sexual activity,6,7 sex trading, multiple partners, polysubstance use, and unprotected sex.8–17

African American women who use crack are especially at risk because many engage in high-risk behaviors and live in social contexts that increase their vulnerability.18 Few of these women are self-sufficient, and they report high levels of public assistance, homelessness, and unemployment; low levels of education; and poverty-level incomes.19–24 The social contexts of substance-using women are also characterized by violence and crime, with high rates of childhood and current victimization.3,25–31 As a result, they report low self-image; symptoms of depression, anxiety,32,33 and posttraumatic stress disorder34; significant medical problems; frequent emergency room visits27; and needs for other resources.35

HIV risk-reduction interventions for women can be effective in reducing sex-risk behaviors and increasing HIV/AIDS knowledge and self-efficacy.36–45 Such interventions should be grounded in social psychological theory; incorporate multiple, women-only sessions led by peers; address gender-related influences; and be culturally sensitive.36,38

Recently, researchers have recognized the importance of the multiple social contexts in which women live and have targeted personal empowerment in economic resources and relationships46 as a vehicle to help women understand and change HIV risk behaviors.46–58 HIV research grounded in women’s empowerment theory recognizes that a woman’s ability and willingness to protect herself from HIV is influenced by her sense of empowerment developed through daily interactions and experiences in her social contexts.52,55–58 This approach moves beyond behavioral skill-building methods emphasized in social cognitively based studies to address power differentials in society and partner influences in relationships that may affect a woman’s sexual behavior and negotiation practices.53,54 These empowerment-based studies acknowledge that women’s sexual behaviors may occur within a social context that condones passivity and inequality in sexual matters and oppression by socioeconomic class and race. Interventions targeting African American women need to acknowledge the unique multiple influences on these women and identify strategies to effectively change risk behaviors and life situations.

The health benefits of facilitating empowerment for minority women have only recently received attention in the HIV prevention literature.50,53,56–59 Incorporating ways for women to increase their personal power within relationship and economic contexts has been critical to the success of HIV risk-reduction interventions.46,48,51,53,54,58,60 For example, a culturally sensitive HIV prevention intervention designed specifically to facilitate relationship power with inner-city women effectively increased safer-sex intentions and behaviors through follow-up at 6 months.53 This intervention used skill-building, cognitive rehearsal, and guided imagery to promote women’s ownership of their bodies and positive health choices in the social contexts in which their behaviors occurred. A study of HIV risk behavior among Latinas who did not use drugs found that economic resource power, as indicated by employment, had both direct and mediating effects on psychological symptoms and unprotected vaginal sex.46,60 Becoming employed may reduce women’s HIV risk by increasing a woman’s resource power and providing behavioral alternatives that foster a healthier sense of self-worth. Similarly, overcoming homelessness, an established risk factor among African American women who use crack,3,5 may also help women to reduce drug use, to become independent of a drug-using lifestyle, and to instead make healthier life choices.

We examine the effectiveness (at 3- and 6-month follow-up) of a personalized HIV intervention tailored to gender and culture, compared with a standard intervention and a delayed treatment control group, to reduce sex-risk behaviors and drug use and increase employment and housing status among African American women who use crack.

METHODS

Outreach and Recruitment

The study was marketed on the streets by means of a brochure through indigenous community outreach workers. Out-of-treatment African American women who use crack were recruited according to a prespecified sampling plan through standardized street-outreach techniques61 and peer-advocate chain-referral procedures that have been used in numerous community studies.61–64 Outreach occurred in targeted inner-city neighborhood segments to ensure that the sample comprised multiple communities. Outreach workers were trained to approach and engage African American women to market the study and screen them for eligibility. Peer advocates who had participated in previous studies also were trained to work in concert with outreach workers to find prospective participants.

Eligibility Criteria

Women who met preliminary eligibility criteria were referred to field sites for final determination of eligibility. Eligible participants were at least 18 years of age, reported engaging in unprotected sex during the previous 90 days, and admitted using crack on at least 13 of the past 90 days. Women who reported being enrolled in substance abuse treatment within the past 30 days were excluded.

Data Collection

All study participants were assessed by self-report at a 2-part intake occurring 1 to 2 weeks apart and at 3- and 6-month follow-up interviews. Participants received $20 compensation at each intake and $25 and $40 for the 3- and 6-month follow-up interviews, respectively. Urine drug screens for cocaine and opiates were performed on site at intake and at 6-month follow-up to help validate self-report data; HIV antibody testing was conducted at intervention session 1 and at 6-month follow-up. Follow-up interviews also captured data on health services received. Process measures helped determine exposure to interventions (including competing programs), compliance, and costs associated with exposure. They also provided quality checks on protocol fidelity. For additional quality assurance, staff audiotaped every eighth interview session for review by the field supervisor.

Recruitment began in January 1999 and ended in August 2001. A total of 938 women completed intake 1. Seven hundred sixty two women (81%) completed intake 2 and were randomly assigned to study conditions; we controlled for injection drug use history, age (< 30 or ≥ 30years), and previous intervention experience. Although the 176 women who did not return for intake 2 were comparable to study participants on most demographic characteristics, they were more likely to report daily crack use (33% vs 22%, P < .01), sex trading (56% vs 44%, P < .01), and multiple sexual partners in the past 30 days (61% vs 47%, P < .001).

Intervention

After intake 2, women were randomly assigned to 1 of 3 study conditions: (1) the woman-focused intervention, (2) a revised intervention modeled on the National Institute on Drug Abuse (NIDA) standard HIV prevention intervention (the standard-R group),65 and (3) a delayed-treatment control group. African American women indigenous to the community were trained to deliver both interventions to minimize interventionist effects. The field supervisor conducted biweekly supervision meetings with interventionists, observed staff conducting individual sessions, and videotaped group sessions to monitor the fidelity of intervention implementation. The woman-focused and standard-R interventions each comprised 4 modules that included 2 individual and 2 group sessions. Session 1 immediately followed intake 2, and each subsequent session occurred within 2 weeks of the prior session. All women in both the standard-R and woman-focused groups completed at least 1 intervention session, with 67% of women in the standard-R group and 65% of women in the woman-focused group completing all 4 intervention sessions. The womanfocused and standard-R interventions are summarized in the following sections. Table 1 ▶ outlines each intervention session in detail.

TABLE 1—

Woman-Focused and Standard-R Intervention Sessions: Duration, Format, and Content

		Content
Duration	Format	Woman-Focused	Standard-R
Session 1, 30–40 min	Individual	Cue cards	Cue cards
		• AIDS and African-American Women—Wake and Durham counties	• What is AIDS?
		• Usual course of HIV infection and AIDS	• Usual course of HIV infection and AIDS
		• How do women get infected with HIV?	• How does someone get infected?
		• Routes of indirect sharing	• Routes of indirect sharing
		• Crack or snorting risk among African-American women	• What behaviors put you at risk
		• Other hurts that cause women to use drugs and be at risk for HIV	• Risks associated with crack or cocaine
		• Reduce your risk of HIV/AIDS: Always use protection	• Reasons for condom use
		• A condom you can control: the female condom	• What about female condoms?
		• Why clean needles and syringes?	• How to talk with your partner about safe sex
		• The injecting gold standard: bleach, bleach, water, water	• Needle and syringe cleaning and disinfection guidelines and practice
		• “Walk the talk” (how to talk to your partner about safer sex)
		• Issues of concern we have learned from women in your community	• Stopping unsafe drug use
		• Benefits of recovery: sister to sister	• Benefits of drug treatment
		• Ways you can become independent in your community	• HIV antibody test
		• The HIV antibody test	• If you are infected
		• If you are infected	Behavioral skills training
		Behavioral skills training	• Male and female condoms, dental dam/plastic wrap
		• Male and female condoms, dental dam/plastic wrap	• Syringe cleaning
		• Syringe cleaning	Materials distributed
		Materials distributed	• HIV/AIDS, condom use literature
		• HIV/AIDS, condom use literature	• Risk-reduction kit
		• Risk-reduction kit	• HIV and non-HIV referrals for services
		• HIV and non-HIV referrals for services	• Local services information
		• Local services information
Session 2, 30–40 min	Individual	Test results	Test results
		Cue cards	Cue cards
		• NIDA posttest counseling for HIV-positive or HIV-negative	• NIDA posttest counseling for HIV-positive or HIV-negative
		• Review of session 1 cue cards	• Review of session 1 cue cards
		Individualized risk assessment plan	Behavioral skills training
		• Personalized risk and lifestyle review	• Male and female condoms, dental dam/plastic wrap
		• Short- and long-term goal setting	• Syringe cleaning
		Behavioral skills training	Materials distributed
		• Male and female condoms, dental dam/plastic wrap	• HIV/AIDS, condom use literature
		• Syringe cleaning	• Male and female condoms
		Materials distributed	• HIV and non-HIV referrals for services
		• HIV/AIDS, condom use literature	• Local services information
		• Male and female condoms
		• HIV and non-HIV referrals for services
		• Local services information
Session 3, 60–90 min	Group (2–5 women)	Support-based group session: “Sisters Uplifting Sisters Out of Drug-Related Bondage”	Didactic general health education: “Learning About Drugs and HIV and How I Can Take Better Care of Myself”
		• Introductions, group rules, session objectives, overview	• Video: “Drugs and How They Affect Your Body Chemistry”
		• Identifying the things that keep you in bondage	• Lecture: Nutrition and hygiene
		• What’s different for women and African American women? Why is it harder?	Review of the Food Guide Pyramid
		• Learning how to break free from the various forms of bondage that African American women today find themselves in	Basics of hygiene
			Good hygiene, better self-image leading to better self-esteem
		• Learning how to take the next step
		• Free yourself! How to make good decisions and take care of yourself
		• Knowing when and how to change playthings, playmates, and playgrounds (triggers)
		• Accepting responsibility for and consequences of our decisions
		• Effective communication and problem solving: gaining freedom in high-risk situations
		• Solving problems using effective communication and ideal problem-solving method
Session 4, 60–90 min	Group (2–5 women)	Support-based group session: “Sisters Uplifting Sisters Out of Drug-Related Bondage”	Didactic general health education: “You Choose—Good Cocaine or Good Physical Health”
		• Introductions, session objectives, overview	• Lecture/video: “What is Cocaine/Crack Doing to My Body—Cocaine and Human Physiology”
		• First things first: take charge of your physical health
		• Family issues—succeeding as a parent or partner	• Lecture/handouts: how you can take responsibility for your physical health
		• Coping with stress in the “hood”: being free in our environment
		• The next step toward freedom: What are you going to do to get and stay free?	Breast self-examination
			Other examinations for ensuring good health
		• Community resources to help you get and stay free	• Lecture/handouts: making quality decisions for reducing your risk of unwanted pregnancy
		• Closure: goal setting for 3 and 6 mo
		• Reaffirmation of women	Types of birth control

Note. NIDA = National Institute on Drug Abuse. The complete manual can be obtained from the first author.

The woman-focused intervention.

The woman-focused intervention development was informed by focus groups with African American women crack users to identify relevant language, issues, and risk patterns. The intervention includes culturally enriched content that is grounded in empowerment theory57,58 and African American feminism.30 It acknowledges specific barriers facing African American women and how these barriers affect daily experiences and choices. On the basis of focus group findings, the woman-focused intervention addressed drug dependence as a form of “bondage” and was designed to facilitate greater independence and increase personal power and control over behavior choices as well as life circumstances. The intervention contained psychoeducational information and skills training on reducing HIV risk and drug use, presented within the context of African American women’s lives in the inner city, where pervasive poverty and violence limit women’s options and increase the likelihood of poor (i.e., high-risk) behavior choices.3

The 2 individual sessions focused on pre- and posttest counseling that addressed risk issues specific to the study population and offered personalized feedback about risk that allowed women to develop individualized behavior change plans according to their unique life situations. Group sessions used a support-based format to help women understand how they are affected by the multiple contextual influences in their lives and to teach portable skills to reduce risk and increase a sense of power. Women also received information for developing support networks and linkages to social services. Goal-setting focused not only on drug use and sex-risk behaviors, but also on life issues such as education, employment, housing, and parenting. The goals for women were to reduce risk and to take concrete empowering steps toward independence (e.g., through full-time employment and stable housing).

The standard-R group.

The standard-R group—modeled on the NIDA standard HIV prevention intervention as revised by a cohort of 6 cooperative agreement sites6,65 —was similar in educational content to the woman-focused intervention but did not incorporate the gender- or culture-specific empowerment approach to develop one’s life and change social contexts. After 2 individual sessions focusing on HIV pre- and posttest counseling, 2 group sessions offered general health education lectures and videos in a didactic format.

The control group.

The control group received no intervention during the first 6 months of study enrollment. At 6-month follow-up, women in the control group were invited to participate in the standard-R intervention; 35% of these women (n = 87) attended at least 1 intervention session.

Study Sample

The present sample comprises 620 (81%) randomly assigned participants who completed either the 3-month or 6-month follow-up interview. HIV prevalence at baseline among those tested was 5.5%, with no seroconversions. Table 2 ▶ presents background characteristics of the sample.

TABLE 2—

Background Characteristics of Study Sample at Baseline

Characteristic	All Participants (n = 620)
Sociodemographic data
Age, mean (SD)	36.7 (6.9)
Married or living with partner	25.4%
High school graduate	53.8%
Have children under age 18 years	27.0%
Have any type of health insurance	36.9%
Employed full time	11.0%
Currently homeless	29.9%
Receiving public assistance benefits	45.0%
Drug use
Age of first alcohol use, mean (SD)a	15.4 (4.3)
Age of first crack use, mean (SD)	25.3 (7.0)
No. of days drank alcohol in the past 30 days, mean (SD)	14.8 (12.1)
No. of days smoked crack in the past 30 days, mean (SD)	17.1 (10.0)
Average continuous crack use longer than 24 h	42.5%
Ever injected	10.7%
Sexual behavior
Less than 1 sexual partner in the past 30 days	45.5%
Engaged in unprotected sex in the past 30 days	88.5%
Ever traded sex for money or drugs	66.7%
Traded sex for money or drugs in the past 30 days	42.8%
Childhood trauma and current psychological distress
Sexually abused before age 18	30.2%
Physically abused before age 18	29.7%
Depression scale score, mean (SD)b	12.9 (4.9)
Anxiety scale score, mean (SD)b	11.9 (6.2)
Traumatic stress scale score, mean (SD)c	18.2 (10.4)
Tested HIV-positived	5.5%

^an = 602; 18 participants had never used alcohol.

^bGriffith J et al.66 Symptoms of depression and anxiety in the past 90 days were assessed using the Drug Abuse Treatment Assessment and Research (DATAR) depression and anxiety scales developed at Texas Christian University, which have demonstrated adequate reliability. The DATAR depression scale and the DATAR anxiety scale each consist of 7 items measured on a 5-point Likert scale ranging from never (0) to almost always (4). Possible scores for each scale ranged from 0 to 28. In this sample, coefficient α was 0.76 for the depression scale and 0.84 for the anxiety scale.

^cDennis M.67 Symptoms of traumatic stress were summarized with items from the Global Appraisal of Individual Needs. In this study, the traumatic stress scale consisted of 12 items measured on a 5-point Likert scale ranging from never (0) to always (4). Possible scores ranged from 0 to 48. In the present sample, coefficient alpha for the scale was 0.90.

^dn = 398, 22 HIV-positive; only women in the standard-R and woman-focused interventions were tested for HIV.

The only significant difference among the groups at baseline was that fewer women in the woman-focused group (36%) reported that they had received public assistance benefits in the past year than women in the standard-R (48%) and control (51%) groups (P = .005). Women who completed at least 1 follow-up interview were similar to those who did not, except that noncompleters were younger (35 years vs 37 years, P = .02). Attrition rates were similar for both 3- and 6-month follow-up interviews, with no meaningful or statistical differences in attrition across the 3 study conditions at either assessment period.

Measures

African American women from the community were trained as interviewers to administer the Revised Risk Behavior Assessment (RRBA) and other supplemental assessments at intake 1 and intake 2 with a computerassisted personal interview. Each session lasted approximately 45 minutes. The RRBA is based on the NIDA-developed Risk Behavior Assessment (RBA) and has demonstrated acceptable levels of reliability with the present study population.3 The RBA, which has yielded acceptable levels of reliability (r ≥ .7) and concurrence with urine tests,68–70 was revised for this project to focus on women’s risk issues.71 The RRBA was the core instrument for assessing the following 10 domains: demographics, drug use (ever, past month), drug injecting, drug use (past 48 hours), drug user treatment, sexual activity, sex trading, health, arrests, and work and income. Outcome measures used in the present study are described in the following paragraphs.

Sex risk.

Sex risk was measured as any unprotected sexual acts (vaginal or anal sex, fellatio) and any sex trading in the past 30 days.

Crack use.

Crack use was measured as the number of days crack was used in the past 30 days.

Homelessness.

Homelessness was assessed by the question, “Do you consider yourself to be homeless?” This question was included in the RBA and has been established as a measure of homelessness in numerous studies.3,4,72

Full-time employment.

Employment status was assessed with the question, “Which of the following best describes your current work situation?” Response options included unemployed and looking for work; unemployed and not looking; working full time, 35 hours or more per week; working part time, less than 35 hours per week; employed but on leave; full-time homemaker; retired; student; or disabled. Study participants who selected the response, “working full-time, 35 hours or more per week,” were coded as working full time; other responses were coded as not working full time.

Analysis

The initial analysis examined differential attrition, comparing baseline characteristics of women who did not complete either 3- or 6-month assessment by conducting cross-tabulation and statistical testing (t test for continuous and χ² test for binary data).

Analyses of change compared the differences in the proportions and means of the woman-focused and standard-R intervention groups with controls separately for 3- and 6-month follow-up interviews, accounting for baseline values and within-subject correlation. Statistical significance of changes in crack use was assessed using paired t tests; changes in homelessness, full-time employment, and any unprotected sex were assessed using the McNemar test.

We estimated and tested the effects of intervention assignment on major study outcomes at 3 and 6 months. Crack use at each follow-up was assessed using linear regression analysis that included baseline use as a covariate. Intervention effects on binary outcome variables of homelessness, employment, and any unprotected and trading sex were assessed using multiple logistic regression analysis that controlled for baseline status. Group comparisons were performed using an intent-to-treat analysis according to group assignment.

RESULTS

Table 3 ▶ presents the baseline, 3-month, and 6-month follow-up data for each of the primary study outcomes by study condition.

TABLE 3—

Group Means at Baseline, 3-Month, and 6-Month Follow-Up

Variable	Baseline	3 -Mo Follow-Up	6-Mo Follow-Up
No. at each time point
Woman-focused	213	186	200
Standard	199	186	180
Control	207	190	189
No. of days used crack during the past 30 days, mean (SE)
Woman-focused	16.9 (0.05)	9.9 (0.05) ***	11.5 (0.06) ***
Standard	16.8 (0.05)	11.4 (0.06) ***	10.8 (0.06) ***
Control	17.7 (0.05)	14.1 (0.06) ***	13.4 (0.06)***
Reporting any unprotected sex in the past 30 days, %
Woman-focused	88	57***	56***
Standard	90	67***	60***
Control	87	67***	66***
Reporting any sex trading in the past 30 days, %
Woman-focused	41	26***	26***
Standard	42	23***	23***
Control	43	36*	28***
Homeless, %
Woman-focused	34	18***	19***
Standard	26	20	20
Control	29	31	23*
Employed full time, %
Woman-focused	12	25***	24**
Standard	12	17	23**
Control	9	12	17*

Note. P values are calculated using paired t tests or the McNemar test for comparison to baseline.

*P < .05, **P < .01, ***P < .001.

Crack Use

All 3 groups reported significant decreases in the number of days of crack use between baseline and 3- and 6-month follow-up (Table 3 ▶). In the regression model adjusting for crack use at baseline, days of crack use in both the woman-focused and standard-R groups were significantly lower than in the control group, with womanfocused participants reporting the greatest reduction. At 6 months relative to controls, standard-R group reductions in crack use remained significant, whereas reductions reported by womanfocused intervention participants reached marginal significance. Table 4 ▶ presents the results of the regression model.

TABLE 4—

Regression Models for Crack Use, Unprotected Sex, Trading Sex, Homelessness, and Employment at 3- and 6-Month Follow-Up

	3 Mo Follow-up		6 Mo Follow-up
Outcomes	Estimate (95% CI)	P Value	Estimate (95% CI)	P Value
Mean no. of days used crack during the past 30 days,a coefficient
Woman-focused vs control	−3.69 (−5.65, −1.74)	.0002**	−1.70 (−3.68, 0.28)	.0918
Standard vs control	−2.18 (−4.13, −0.23)	.0288*	−2.36 (−4.39, −0.33)	.0227*
Woman-focused vs standard	−1.52 (−3.48, 0.44)	.1284	0.66 (−1.34, 2.66)	.5185
Reporting any unprotected sex in the past 30 days, odds ratio b
Woman-focused vs control	0.69 (0.45, 1.09)	.1153	0.62 (0.41, 0.96)	.0335*
Standard vs control	0.95 (0.61, 1.50)	.8319	0.72 (0.47, 1.13)	.1511
Woman-focused vs standard	0.73 (0.47, 1.14)	.1729	0.86 (0.57, 1.33)	.5081
Reporting any sex trading in the past 30 days, odds ratio b
Woman-focused vs control	0.58 (0.34, 1.00)	.0456*	0.94 (0.55, 1.59)	.8144
Standard vs control	0.48 (0.28, 0.82)	.0070**	0.80 (0.46, 1.39)	.4284
Woman-focused vs standard	1.21 (0.70, 2.08)	.4842	1.18 (0.68, 2.04)	.5698
Homeless, odds ratio b
Woman-focused vs control	0.35 (0.20, 0.61)	.0002**	0.69 (0.40, 1.20)	.1931
Standard vs control	0.56 (0.33, 0.96)	.0366*	1.01 (0.57, 1.76)	.9859
Woman-focused vs standard	0.61 (0.34, 1.11)	.1043	0.69 (0.39, 1.22)	.2013
Employed full time, odds ratio b
Woman-focused vs control	2.53 (1.38, 4.66)	.0027**	1.47 (0.87, 2.48)	.1503
Standard vs control	1.25 (0.66, 2.41)	.4871	1.42 (0.83, 2.43)	.2039
Woman-focused vs standard	2.016 (1.13, 3.60)	.0175*	1.04 (0.63, 1.71)	.8892

Note. CI = confidence interval.

^aRegression analysis with baseline days of crack use as covariate.

^bMultiple logistic regression models controlling for baseline status.

*P < .05, **P < .01.

Unprotected Sex

All 3 groups reported significant reductions in the proportion of women having any unprotected sex in the past 30 days between baseline and 3- and 6-month follow-up (Table 3 ▶). Although the woman-focused group demonstrated greater reductions in unprotected sex than the standard-R and control groups at 3 months, these results were not statistically significant at the .05 level. However, at 6 months this trend was statistically significant relative to controls, with fewer woman-focused group participants reporting any unprotected sex in the past 30 days (odds ratio [OR] = 0.62, P = .03). The results of the logistic regression model adjusted for baseline involvement in unprotected sex are presented in Table 4 ▶.

Trading Sex

All study conditions demonstrated significant reductions in the proportion of women reporting trading sex for money or drugs in the past 30 days between baseline and 3- and 6-month follow-up (Table 3 ▶). Both intervention groups showed significant reductions in the percentage of women who traded sex compared with control subjects, with the standard-R group (OR = 0.48, P = .007) having slightly stronger effects than the woman-focused group (OR = 0.58, P = .046) at 3-month follow-up. At 6 months, these trends in reduction continued, although they were not statistically significant. The results of the logistic regression model adjusted for trading sex in the past 30 days at baseline are presented in Table 4 ▶.

Homelessness

Between baseline and 3-month follow-up, only woman-focused participants demonstrated a significant reduction in homelessness. The standard-R group showed nonsignificant decreases in homelessness, whereas the percentage of control group participants reporting homelessness increased slightly. At 6 months, the woman-focused and control groups showed significant decreases in homelessness, whereas the standard-R group showed nonsignificant improvements in housing status from baseline (Table 3 ▶). In multiple logistic regression analysis controlling for baseline homelessness, woman-focused and standard-R groups reported statistically significant decreases in homelessness compared with the control group. At 3 months, the odds of being homeless were the lowest in the woman-focused group (OR = 0.35, P = .0002). At 6-month follow-up, the woman-focused group continued to report the lowest percentage of homeless women compared with the standard-R and control groups, although the effects were not statistically significant. Results of the logistic regression model for homelessness are presented in Table 4 ▶.

Employment

Between baseline and 3-month follow-up, the percentage of participants employed full time increased in all 3 groups, with significant improvements experienced only within the woman-focused group. Between baseline and 6 months, all 3 groups reported significant increases in full-time employment (Table 3 ▶). In multiple logistic regression analysis controlling for full-time employment at baseline, the odds of being employed full time at 3 months were significantly higher in the woman-focused group relative to both controls (OR = 2.53; P = .0027) and the standard-R group (OR = 2.02, P = .0175). The standard-R group also showed greater increases in employment than controls at 3 months, although this effect was not statistically significant (OR = 1.25, P = .4871). At 6 months, both the woman-focused and standard-R groups showed relatively greater improvements in full-time employment, although these effects were not statistically significant. Logistic regression model results for full-time employment are presented in Table 4 ▶.

DISCUSSION

Study Limitations

Intervention effects on risk behaviors, employment, and housing were supported statistically; however, the analyses cannot disentangle other unmeasured factors, such as staff attributes that may affect most intervention research. Analyses also cannot distinguish other possible effects of the assessment process on actual behavior change versus response bias (i.e., social desirability). To reduce this potentially confounding effect, for any given participant, no staff member was allowed to serve as both an interventionist and a follow-up interviewer.

Although street outreach targeted a range of neighborhoods, this method limits the sample to women approached by outreach workers and peer advocates and thus does not engender a fully representative sample of African American women who use crack cocaine. Nonetheless, this sample has been found to be similar to other African American, crack-using women in low-income inner-city communities who are at high risk for acquiring HIV.57,73

CONCLUSIONS

Results of this study are consistent with previous research findings42–45 that out-of-drugtreatment African American women who use crack and are at high risk for acquiring HIV who receive either a standard or tailored educational and skill-building intervention made significant reductions in crack use and sex-risk behaviors at 3 and 6 months. In addition, at 3-month follow-up, compared with the standard-R group, women who received the tailored intervention showed greater improvement in both employment and housing status, factors that are associated with improved health outcomes.74–78 Findings suggest that empowerment-based interventions tailored to develop concrete solutions within personal social contexts, more than standard interventions, can influence other life changes that facilitate independence for African American women. The personalized nature of the risk profile and change plan was likely a key element that increased the saliency of a participant’s behaviors and life choices that place her at risk. Discussion of these behaviors and life choices within a woman’s multiple contexts allowed for goal-setting and action plans that were concrete and relevant to each woman’s life.

Although the woman-focused participants reported significant improvements in risk behaviors, employment, and housing at both 3 and 6 months (Table 3 ▶), the gains made by standard-R and control group participants at 6-month follow-up reduced the relative effects of the woman-focused intervention when compared with the control group. The exception to this trend was the percentage of women reporting any unprotected sex. By 6 months, woman-focused participants were significantly more likely than women in the other groups to stop having unprotected sex. Overall, some intervention appears better than no intervention, and a gender-specific and culturally tailored intervention may be more effective than a standard intervention at reducing sex risk over time. Historically, sex-risk behaviors have been difficult for HIV interventions to change, which makes this finding encouraging.

Because, relative to the other groups, significantly fewer participants in the womanfocused intervention received public assistance benefits in the past year, we ran additional models that included this variable. However, post hoc logistic regression analysis with group assignment, baseline employment, and benefits status found no independent effects of benefit status on full-time employment at 3-month follow-up. Similar analysis for homelessness at 3-month follow-up found that benefits status, homelessness at baseline, and intervention group assignment each significantly predicted homelessness at follow-up. This analysis indicated that women who did not receive benefits at baseline were more likely to be homeless at follow-up, lending further support to the woman-focused intervention effects at helping women empower themselves to find housing.

This study lends credence to the utility of interventions that target women with high-risk behaviors who live in multiple-stressor environments to be individualized and tailored to their culture, gender, and unique life circumstances. Helping women to empower themselves to reduce specific risk behaviors and raise their self-sufficiency through full-time employment and stable housing, which are linked to improved health outcomes, may be one key to sustaining short-term and potentially longer term risk reduction and healthy behaviors. It is essential to conduct further studies to measure long-term durability of intervention effects that specifically address how risk behavior and contextual barriers change over time to affect women’s healthy behavior choices. Given the cyclical nature of addiction, additional interventions may be needed to help these women continue the process of reducing HIV risk and attaining independence despite contextual barriers.