Secondary Logo

Journal Logo


Stigma Reduction Among African American Women With HIV: UNITY Health Study

Rao, Deepa PhD, MA*; Kemp, Christopher G. PhDc, MPH; Huh, David PhD; Nevin, Paul E. MPH; Turan, Janet PhD, MPH§; Cohn, Susan E. MD, MPH; Simoni, Jane M. PhD; Andrasik, Michele PhD**; Molina, Yamile PhD††; Mugavero, Michael J. MD‡‡; French, Audrey L. MD§§

Author Information
JAIDS Journal of Acquired Immune Deficiency Syndromes: July 1, 2018 - Volume 78 - Issue 3 - p 269-275
doi: 10.1097/QAI.0000000000001673



Recent U.S. Centers for Disease Control and Prevention reports indicate that AIDS is among the top 10 causes of death for African American women between the ages of 20 and 54.1 This is likely due to the higher HIV prevalence among African Americans, but also their lower medication adherence rates and lower utilization of treatment resources compared with other racial and ethnic groups in the United States.2 A growing body of research has shown that internalized HIV-related stigma is one important factor that undermines individuals' adherence to lifesaving antiretroviral medications.3–5 Given the clear link between stigma and medication adherence, there is a critical need for interventions that address HIV-related stigma in hope of improving health outcomes, attenuating disparities, and achieving health equity for African Americas living with HIV, a principal tenet of the US National HIV/AIDS Strategy.6

HIV-related stigma has been heavily studied in observational and epidemiological studies.5,7 Still, little research exists on the effectiveness of HIV-stigma reduction interventions for People living with HIV (PLWH). In 2003, Brown et al reviewed interventions seeking to reduce HIV-related stigma, primarily from studies available at the time that examined changes in attitudes held by the public. Intervention mechanisms included information/education, participatory skill–building training, counseling/support groups, and contact (between PLWH and others).8 Sengupta et al9 conducted a review in 2011 of HIV prevention interventions measuring stigma and noted that effective interventions should: focus specifically on stigma reduction and use more than one of Brown identified strategies to address issues affecting a specific targeted population; choose highly validated instruments for measuring change in stigma; use Randomized Controlled Trial (RCT) study designs; and measure health outcomes. A 2013 systematic review of the literature on HIV-stigma reduction interventions found no randomized clinical trials aimed at testing the effectiveness of stigma reduction interventions among PLWH,10 and since then, 2 pilot studies demonstrated stigma reduction among people with HIV and one trial reported that a multilevel HIV-stigma intervention reduced 2-year mortality among people who inject drugs.11–13

Since 2008, our research group adapted and pilot tested a well-utilized intervention, the International Center for Research on Women (ICRW) HIV Stigma Toolkit.14 The primary objective of the intervention, called the “UNITY Workshop,” was to provide African American women living with HIV the tools to handle intersectional stigmas associated with HIV, race, and sex.15 This study examined the effect of the UNITY workshop on HIV-related stigma among African American women living with HIV, compared with a breast cancer education workshop.


Trial Design

This was a randomized controlled trial using a repeated-measures design. We examined changes in HIV-related stigma among women who were individually randomized using a 1:1 allocation ratio to receive either the UNITY workshop (intervention arm) or a breast cancer education workshop (comparison arm). These arms were matched on time and attention: both were two 4-hour group workshops over 2 days, followed by a one time 2-hour booster session 6 months later. No changes to methods were made after trial commencement. This study is reported according to the CONsolidated Standards Of Reporting Trials (CONSORT) 2010 guidelines.16


Women were eligible for inclusion in the study if they (1) self-identified as having an African American racial/ethnic background, (2) were 18 years of age or older, and (3) had documentation of living with HIV. We excluded women who were African born, of Afro-Caribbean descent, or identified as Black Latino in anticipation of culturally specific issues that would warrant additional adjustments to our stigma-reduction interventions.

We selected urban locations where African Americans make up a large segment of the population. Participants were recruited from Chicago, Illinois and Birmingham, Alabama at 3 HIV-specialized clinics: a state university-based clinic in Birmingham (public clinic), a private university-based clinic in Chicago (private clinic), and a safety-net hospital-affiliated clinic (public clinic) in Chicago. The public Chicago site was chosen after another community-based clinic identified for recruitment closed; during this transition, there were delays in recruitment for one Chicago cohort.


The UNITY workshop was first piloted in Seattle15 with an African American woman living with HIV (ie, a peer) serving as the primary facilitator. A masters-level social worker assisted the facilitator and led break-out group sessions. The same set of facilitators led groups in Chicago, and a new set of local facilitators led the Birmingham groups. The UNITY intervention included all 4 of Brown et al's categories of intervention mechanisms and Corrigan's best practices for “Strategic Stigma Change,”17 with a primary objective to help foster contact and social support. The differences between the UNITY workshop and breast cancer education workshop were in content rather than format. A side-by-side description of differences in the content of the 2 interventions is given in Supplemental Digital Content 1, Both workshops used the videos and a mix of activities delivered in the group format, including facilitated large and small group discussions, brainstorms, dyadic presentation, and role plays.

The UNITY workshops began with discussions of group expectations and what the term “stigma” meant to them. After watching a “trigger” video (3–4 minute videos produced to elicit discussion), the facilitator lead a guided discussion. Participants explored reactions to the video and personal experiences with stigma in a large group and in dyads. The next exercise involved brainstorming coping methods that group members have used to deal with stigma. In a segment about self-soothing, the facilitator led a guided visualization. A subsequent segment included modeling and practicing assertiveness (vs. passive or aggressive) in response to stigma. A self-esteem exercise and social support exercise finished out the first day. Day 2 focused on disclosure with case studies and role play. The final exercise of UNITY engaged the group regarding how to “live positively” with HIV.

The breast cancer education workshop was designed to match the UNITY workshop in terms of time and attention, and as such, was also held across two 4-hour sessions using similar group formats. The breast cancer education workshop contained elements of education, skill building, counseling/support, and contact, but the content covered related to breast cancer rather than HIV-related stigma. Instead of trigger videos, the coordinator showed a segmented video about an African American woman's journey through screening, mammograms, detection, and treatment of breast cancer. In addition, the workshop included a PowerPoint presentation of breast cancer disparities and how to both seek and provide support regarding breast cancer screening and treatment. The facilitator for this arm was one research coordinator per site; no peer was involved.

The breast cancer education workshop began with a discussion about experiences with screening, presentations on biology and treatment, and a discussion of challenges faced by African American women. Discussions focused on self-care and preventive practices such as clinical breast examinations and mammograms. Participants generated strategies for dealing with hesitations they might have about preventive practices. An assertiveness exercise is used in this arm to teach this coping and self-advocacy skills, again using role play. On day 2, discussions covered the topics of potential barriers to care, such as family and financial barriers. The final segment finished with discussions of self-efficacy, care-seeking behaviors, and the importance of social support.


All participants provided sociodemographic and clinical information (ie, age, education, time since diagnosis) at their baseline visit. The research coordinator obtained information on most recent CD4+ T-cell count and viral load from the participants' medical records at each study visit. HIV-related stigma was measured at all time points using the 14-item Stigma Scale for Chronic Illness (SSCI) to assess internalized and enacted stigma. Cognitive interviews and psychometric data support the use of this scale among African Americans living with HIV, and our previous work demonstrated that the scale was best used as a unidimensional stigma construct.18–20 Perceived social support was measured at all time points using the 19-item Medical Outcomes Study-Social Support Survey (MOS-SSS), a multidimensional measure that has been used internationally to assess perceived social support in chronic disease contexts.21 Individual summary scores for the SSCI and MOS-SSS scales were created as sum totals of scale responses at each time point. Missing responses were mean-imputed.

Sample Size

We used the results of our pilot feasibility study to determine necessary sample size. Those results suggested a trend but did not estimate statistically significant differences in mean total stigma scores. Therefore, we chose to calculate sample size for a conservative effect size of 0.2. With power set at 85%, alpha set at 0.05, 3 follow-up assessments, repeated measures correlated at r = 0.5, and effect size (for mean differences) set at 0.2, we estimated a necessary sample size of 94 to detect a clinically important difference between participants' scores before and after intervention within one location. We therefore sought to recruit at least 94 participants at each location. Anticipating 15%–20% dropout, we aimed to enroll 224 participants in the study (112 in Chicago and 112 in Birmingham), providing approximately 83%–85% power to detect differences in outcomes.

Procedures and Randomization

Clinic staff posted signs within clinic rooms encouraging eligible participants to call research coordinators for information on the study. In addition, a research coordinator verbally described study procedures to eligible and interested participants during their clinic visits. During the baseline visit, participants learned about the study and signed the informed consent, completed baseline questionnaires, and were randomized within each site to either UNITY or the Breast cancer education group. The principal investigator generated the random allocation sequence using an online randomization calculator in blocks of 2 participants. Study coordinators used the random allocation to assign participants to a study arm after completing the baseline assessment. Participants were not informed of their arm assignment until arriving at the first day of workshop. We began the first intervention and comparison arm groups when we had adequate women enrolled, with the goal of 28 women (14 for each group). We began recruiting the next set of groups with the aim of recruiting half the women from Chicago and half from Birmingham. A booster session (2-hour session with abbreviated exercises) was conducted 6 months after initial workshop participation in both arms. Participants in both arms completed tablet-based study questionnaires (ACASI: audio computerized assistant self-interviews) at baseline visit, immediately after their workshops, 4 months after their workshops, immediately after the 6-month booster session, 8 months after the initial workshop, and 12 months after the initial workshop.

Statistical Methods

Primary Outcome Analysis

We first compared baseline demographic characteristics and measures across arms to determine adequacy of randomization. All participants randomized at baseline were included in the primary outcome analysis (ie, an intention-to-treat approach). To evaluate the effect of the UNITY workshop (intervention condition) on changes in HIV-related stigma over time, compared with the breast cancer education workshop (comparison condition), we used 2-sample t tests, assuming unequal variance, and conducted longitudinal regression using generalized estimating equations (GEE).22 Cluster robust standard errors were estimated to account for the correlation of repeated measures within participants.23 The GEE approach permits an analysis of all the available outcome data, allowing for participants with incomplete follow-up data to be retained. The primary outcome was relatively normally distributed and modeled with a Gaussian link function.

In the GEE analysis, stigma score was regressed on treatment (UNITY vs. Breast Cancer), time, and the treatment by time interaction. The statistical test of the intervention effect was the magnitude and statistical significance of the UNITY × time interaction. We parameterized time in a linear fashion as the number of months since baseline [0, 2 (after workshop), 6, 8, 10, and 14 months], because we hypothesized that rates of change in stigma scores would be constant.

To evaluate whether the effect of the intervention differed across the 3 study sites, the primary analyses were extended to include indicator variables for study site and the interaction of study site with the intervention effect (ie, UNITY × time × site interaction). A Wald test was used to evaluate whether the site-specific intervention effects were statistically significant.

Post hoc Analysis of Social Support

We hypothesized that participant-level decreases in HIV-related stigma would be explained, in part, by increases in perceived social support resulting from meeting with peers in a supportive group. In a post hoc analysis, we assessed whether changes in perceived social support predicted changes in HIV-related stigma over time. This model also assessed whether the effect of social support on stigma differed by study arm. Our predictor of interest was change in perceived social support from a preceding time point to a current time point, standardized by dividing by the number of months elapsed. This analysis was therefore restricted to the last 5 time points of the study. As participants were not randomized to receive social support, we adjusted for variables that we hypothesized might confound the relationship between social support and HIV-related stigma. We adjusted for age in years and years with HIV as continuous variables, and marital status, education, number of children, occupation, and site as dummy variables. We chose these variables to align with previous analyses of the relationship between psychosocial factors and HIV-related stigma.24,25 To illustrate the association between perceived social support and stigma, counterfactual predictions of the effect of 3 hypothetical changes in social support (1 SD decrease, no change, 1 SD increase) on stigma scores were estimated using the simulation-based approach recommended by King, Tomz, and Wittenberg.26

Missing Data

We coded each participant as having (1) complete data from 6 study assessments, (2) missed a single assessment, (d) missed 2 to 3 assessments, or (e) 4 to 5 assessments. To assess for differences between participants with complete assessment data and those who were missing one or multiple assessments, Pearson χ2 tests and 1-way analyses of variance were conducted, respectively, on categorical and continuous sociodemographic characteristics and baseline levels of stigma and social support.


All study procedures were approved by the University of Washington (study coordinating site) Institutional Review Board, and in addition, the institutional review boards of each institution where we recruited participants approved study procedures. This study was also registered under number NCT01893112.


Two hundred thirty-nine women were recruited into the study. One hundred twenty-four were randomized to the UNITY workshop arm, and 115 were randomized to the breast cancer education arm. The median time between BL and postworkshop assessment was 34 days [interquartile range = (19, 59)]. Seventy-seven women attended the UNITY booster session and 72 women attended the breast cancer education booster session. There was a 1 to 1 correspondence between women who attended the booster session and the 6-month follow-up assessment because these measures were administered immediately after the booster session. Figure 1 presents a flowchart of recruitment, refusal, dropout, and assessment attendance over the course of the study. Recruitment started in May 2013 and ended in October 2015. The last follow-up assessment was completed in December 2016.

Study flow diagram.

Descriptive Analyses

Baseline sociodemographic characteristics are provided in Table 1. There were no statistically significant differences in observed sociodemographic characteristics across treatment arms, suggesting that randomization was successful. Mean participant age was 47.0 years in the UNITY group, and 46.5 years in the breast cancer education comparison group. Almost 40% of participants had not completed high school, fewer than half (45.7% in comparison, 44.0% in intervention) were employed, and most (62.0% in comparison, 55.4% in intervention) had no children. Participants had been living with HIV for a mean of 14.6 years in the intervention group, and 13.7 years in the comparison group. Across treatment groups, participants from Birmingham were older (mean age 49.5 years vs. 44.6 years in Chicago) and more likely to be separated from marriage (53.3% separated vs. 30.5%) than participants from Chicago.

Baseline Participant Sociodemographic and Clinical Characteristics*

Missing Data

With respect to the primary outcome data, approximately half of participants (51%, n = 122) had complete data from all 6 assessment points, 19% (n = 46) missed a single assessment, 10% (n = 23) missed 2 to 3 assessments, and 20% (n = 48) missed 4 to 5 assessments. There were no statistically significantly associations between degree of missing data and age, marital status, education, employment, number of children, years with HIV, study site, or baseline levels of stigma or social support. Furthermore, treatment assignment was not associated with number of missed assessments. Given that missingness was not associated with treatment assignment, we used all available assessment data in our primary and post hoc analyses, including partial data from those who missed one or more assessments.

Primary Analyses

An evaluation of site-specific intervention effects did not indicate that there was variation in the effect of the intervention across the 3 study sites. Therefore, the primary analyses focused on the overall intervention effect combining participants across all sites. Table 2 presents mean stigma scores over time for each treatment group. Both groups' stigma scores at each time point decreased below that of baseline, but these differences were not larger in the UNITY group as compared to the breast cancer education group. In both groups, after the immediate decrease in mean stigma scores, mean stigma scores increased slightly at each time point, but never increasing beyond baseline stigma scores. Table 3 presents the results of the primary GEE analysis of the association between treatment assignment and stigma score at each month. Our model estimates that allocation to UNITY was associated with an additional 0.096 stigma points per month; however, this difference was not statistically significant [95% confidence interval (CI): −0.099 to 0.292]. These results suggest that the rate of change in mean stigma scores over the year of follow-up did not differ for subjects randomized to UNITY compared with subjects randomized to the breast cancer education arm.

Mean (SD) Stigma Scores Stratified by Period and Treatment Group
GEE Estimates of Association Between Treatment and Stigma Scores Over 12 Months

Post hoc Analysis of Social Support

Table 4 presents the results of the post hoc GEE analysis of the association between changes in perceived social support over time and stigma score at each month, adjusting for treatment status, age, marital status, education, employment, number of children, years with HIV, and site. This adjusted model estimated that each point increase in social support over time was associated with a decrease in stigma score of 0.039 (95% CI: −0.077 to −0.0002) among breast cancer education participants. This difference was statistically significant (P = 0.049). Among UNITY participants, this estimate was not statistically significant: β = −0.048 (95% CI: −0.138 to 0.042). Supplemental Digital Content 2, presents a figure (Model-Based Counterfactual Estimates of Stigma by Change in Social Support) illustrating the predicted stigma score over time among participants whose perceived social support scores (1) decreased by 1 SD, (2) had no change, or (3) increased by 1 SD. These results suggest that preceding increases in perceived social support are associated with decreased HIV-related stigma in this population.

GEE Estimates of Association Between Changes in Social Support and Stigma Score


Although stigma was reduced in both treatment and control groups over time, the UNITY workshop was not more effective at reducing stigma than the breast cancer education workshop. This finding suggests that stigma-specific content may not be necessary to reduce stigma. Many participants in both arms experienced increases in social support through the peer group experience, which was associated with subsequent decreases in stigma. Peer social support may be an essential ingredient that makes such interventions work, and has potential to reduce stigma among African American women with HIV. These findings are consistent with other longitudinal studies that have found peer support to be instrumental for stigma reduction outside the HIV context.27,28

We examined stigma reduction every 3 months after participation in initial workshops up to 12 months after initial workshop participation. Although stigma was reduced in the UNITY group immediately after workshop participation, this reduction was not statistically significant. The uptick of stigma among our participants, even after participation in a booster session at 6 months, may be the consequence of ongoing stigma experienced in families, communities, and places of work. It suggests the need for ongoing peer support groups—even in a clinical setting—to give women an option for additional social support.


Our study had some limitations, one of which was our choice of using a time and attention comparison group instead of a comparison group that did not have active elements of contact or social support. The comparison group provided a way for all participants to obtain therapeutic attention in a group setting, but ultimately, we were not able to determine whether a peer support group worked better at reducing stigma than an intervention without contact or social support mechanisms. Including a true control may have required a third arm and more resources.29 In addition, our findings cannot be generalized to rural populations because we recruited our participants from urban clinics. As such, caution should be taken in interpreting findings from this study.


In all, our participants gained some important benefits from participation in 2 workshops that were both tailored specifically for African American women living with HIV. In addition to gaining support and experiencing reductions in stigma, the women learned about coping with consequences of living with certain health conditions such as HIV and breast cancer. Future stigma-reduction interventions should consider including a peer support group component, and taking a holistic approach to women's health, including concerns related to both HIV and women's other health concerns such as breast cancer, while attending to issues at the intersection of health, race, and sex.


1. Centers for Disease Control and Prevention. Leading Causes of Death (LCOD) by Age Group, Black Females-United States, 2014. Available at: Accessed May 25, 2017.
2. Kahn JG, Zhang X, Cross LT, et al. Access to and use of HIV antiretroviral therapy: variation by race/ethnicity in two public insurance programs in the U.S. Public Health Rep. 2002;117:252–262; discussion 231–252.
3. Rao D, Kekwaletswe TC, Hosek S, et al. To medication adherence with urban youth living with HIV. AIDS Care. 2007;19:28–33.
4. Halkitis P, Parsons J, Wolitski P, et al. Characteristics of HIV antiretroviral treatments, access and adherence in an ethnically diverse sample of men who have sex with men. AIDS Care. 2003;15:89–102.
5. Katz IT, Ryu AE, Onuegbu AG, et al. Impact of HIV-related stigma on treatment adherence: systematic review and meta-synthesis. J Int AIDS Soc. 2013;16(3 suppl 2):18640.
6. White House Office of National AIDS Policy. National HIV/AIDS Strategy for the United States. Washington, DC: White House Office of National AIDS Policy; 2010.
7. Earnshaw V, Chaudoir S. From conceptualizing to measuring HIV stigma: a review of HIV stigma mechanism measures. AIDS Behav. 2009;13:1160–1177.
8. Brown L, Macintyre K, Trujillo L. Interventions to reduce HIV/AIDS stigma: what have we learned? AIDS Education Prev. 2003;15:49–69.
9. Sengupta S, Banks B, Jonas D, et al. HIV interventions to reduce HIV/AIDS stigma: a systematic review. AIDS Behav. 2011;15:1075–1087.
10. Stangl AL, Lloyd JK, Brady LM, et al. A systematic review of interventions to reduce HIV-related stigma and discrimination from 2002 to 2013: how far have we come? J Int AIDS Soc. 2013;16(3 suppl 2):18734.
11. Go VF, Frangakis C, Le Minh N, et al. Increased survival among HIV-infected PWID receiving a multi-level HIV risk and stigma reduction intervention: results from a randomized controlled trial. J Acquir Immune Defic Syndr. 2017;74:166–174.
12. Barroso J, Relf MV, Williams MS, et al. A randomized controlled trial of the efficacy of a stigma reduction intervention for HIV-infected women in the Deep South. AIDS Patient Care STDS. 2014;28:489–498.
13. Nyamathi A, Ekstrand M, Salem BE, et al. Impact of Asha intervention on stigma among rural Indian women with AIDS. West J Nurs Res. 2013;35:867–883.
14. Kidd R, Clay S, Chiiya C. Understanding and challenging HIV/AIDS stigma: a Toolkit for action. International Center for Research on Women, Academy for Educational Development (AED). Washington. DC: International AIDS Alliance; 2007.
15. Rao D, Desmond M, Andrasik M, et al. Feasibility, acceptability, and preliminary efficacy of the “unity workshop”: an internalized stigma reduction intervention for African American women living with HIV. AIDS Patient Care STDs. 2012;26:614–620.
16. Moher D, Hopewell S, Schulz KF, et al. CONSORT 2010 Explanation and Elaboration: updated guidelines for reporting parallel group randomised trials. J Clin Epidemiol. 2010;63:e1–e37.
17. Corrigan P. Best practices: strategic stigma change (SSC): five principles for social marketing campaigns to reduce stigma. Psychiatr Serv. 2011;62:824–826.
18. Rao D, Andrasik M, Acharya X, et al. Internalized Stigma Among African Americans Living with HIV: Preliminary Scale Development Based on Qualitative Data. Stigma, Discrimination and Living with HIV/AIDS: A Cross-cultural Perspective. Dordrecht, the Netherlands: Springer; 2013:155–168.
19. Rao D, Molina Y, Lambert N, et al. Assessing stigma among african Americans living with HIV. Stigma and health. 2016;1:146–155.
20. Rao D, Choi S, Victorson D, et al. Measuring stigma across neurological conditions: the development of the stigma scale for chronic illness (SSCI). Qual Life Res. 2009;18:585–595.
21. Sherbourne CD, Stewart AL. The MOS social support survey. Soc Sci Med. 1991;32:705–714.
22. Liang KY, Zeger SL. Longitudinal data analysis using generalized linear models. Biometrika. 1986;73:13–22.
23. Chandler RE, Bate S. Inference for clustered data using the independence loglikelihood. Biometrika. 2007;94:167–183.
24. Earnshaw VA, Smith LR, Chaudoir SR, et al. HIV stigma mechanisms and well-being among PLWH: a test of the HIV stigma framework. AIDS Behav. 2013;17:1785–1795.
25. Turan B, Hatcher AM, Weiser SD, et al. Framing mechanisms linking HIV-related stigma, adherence to treatment, and health outcomes. Am J Public Health 2017;107:863–869.
26. King G, Tomz M, Wittenberg J. Making the most of statistical analyses: improving interpretation and presentation. Am J Polit Sci. 2000;44:341–355.
27. Whitley R, Campbell RD. Stigma, agency and recovery amongst people with severe mental illness. Soc Sci Med. 2014;107:1–8.
28. Corrigan PW, Sokol KA, Rusch N. The impact of self-stigma and mutual help programs on the quality of life of people with serious mental illnesses. Community Ment Health J. 2013;49:1–6.
29. Freedland KE, Mohr DC, Davidson KW, et al. Usual and unusual care: existing practice control groups in randomized controlled trials of behavioral interventions. Psychosom Med. 2011;73:323–335.

stigma; psychosocial; social support

Supplemental Digital Content

Copyright © 2018 Wolters Kluwer Health, Inc. All rights reserved.