Healthy Choices intervention is associated with reductions in stigma among youth living with HIV in the United States (ATN 129)

Abstract

Considering the lack of validated stigma reduction interventions for youth living with HIV (YLWH), we evaluated effects of the Healthy Choices intervention on HIV-related stigma among YLWH. We analyzed data from the Adolescent Medicine Trials Network protocol 129, multi-site randomized controlled trial, applying latent growth curve modeling with two linear slopes estimating changes in Berger’s Stigma Scale pre-intervention, 16, 28, and 52 weeks post-intervention, as well as the trajectory of stigma scores over the follow-up period (N=183). Expected value for the growth intercept was statistically significant (B_intercept=2.53; 95% CI: 2.32, 2.73; p<.001), as were differences in the change from baseline to 16-week follow-up (B_{intercept_slope1}=−0.02; 95% CI: −0.04, 0.01; p=.034). Expected value of the slope factor measuring growth over the follow-up period was non-significant suggesting that stigma scores were stable from 28 to 52 weeks. Our findings warrant replication and additional research comparing effects of this intervention to counterfactual controls.

INTRODUCTION

Whereas notable strides have been made in the development and dissemination of HIV-focused interventions to ensure linkage to care upon diagnosis and promotion of healthcare engagement,[1–4] there are few interventions that have been shown to be effective in reducing stigma among people living with HIV (PLWH),[5–7] and -- to our knowledge -- no validated interventions exist that reduce HIV-related stigma among youth living with HIV (YLWH). Prior studies have illustrated the deleterious health effects of HIV-related stigma on PLWH, such as perceived and internalized stigma being associated with higher rates of missed clinical visits and lower rates of viral suppression,[8,9] which negatively influence the HIV continuum of care, making HIV stigma reduction is a high-priority public health target. Considering the lack of evidence-based stigma reduction interventions, we collected measures of HIV stigma, during the Healthy Choices intervention randomized controlled trial, conducted under the auspices of the Adolescent Medicine Trials Network for HIV/AIDS Interventions (ATN, protocol 129),[10] to assess if this intervention could reduce HIV-related stigma among YLWH. We applied latent growth curve modeling to assess intervention effects on perceived HIV stigma among YLWH immediately after receipt of the Healthy Choices intervention, and thereafter to 52-weeks post-intervention. Research with other stigmatized populations and other health conditions have offered some evidence that Motivational Interviewing may reduce perceived stigma among patients and may reduce stigmatizing language from providers;[11–13] thus, we hypothesize that receipt of the Healthy Choices intervention will be associated with reductions in perceived HIV stigma among YLWH.

Stigma and health

Stigma is a damaging structural phenomenon that is created and perpetuated by society.[14,15] Stigma refers to the process by which a group of individuals, in the case of this study YLWH, is labeled as socially undesirable and devalued due to attributes or behaviors that are societally deemed as “deeply discrediting.”[14,15] Berger and colleagues, suggest that perceived stigma can have negative effects on the sense of self-worth in PLWH, and that HIV stigma may become internalized, which in turn, may lead to negative affective, cognitive, and mental health outcomes.[14,16,17] Internalized HIV stigma is the acceptance of negative societal perceptions, characterizations, and labels about PLWH.[16,18] Internalized stigma may be associated with self-deprecating feelings such as shame, self-blame, embarrassment, and low self-worth.[19–21]

Perceiving and internalizing HIV stigma can increase the likelihood of missing clinical visits and reduce antiretroviral medication adherence, leading to increases in viral loads -- all of which are associated with reductions in well-being and lower quality of life.[8,9,16,22–25] Earnshaw et al. suggest that experiences of enacted HIV stigma in healthcare settings lead to chronic stress, and this chronic stress can lead to worse health outcomes among PLWH.[26–28] Chronic stress harms physical health through activating a stress-response, namely the hypothalamic-pituitary-adrenal axis, within the sympathetic nervous system.[29,30] Because HIV stigma is harmful to YLWH, and ideally, providers who support YLWH should be youth-friendly and free of stigma, identifying interventions that reduce HIV-related stigma is urgently warranted and has the potential to improve outcomes across the HIV continuum of care.

Motivational Interviewing communication framework

Motivational Interviewing provides a highly specified framework for improving patient-provider communication and promoting behavior change in PLWH leading to improved HIV continuum of care outcomes.[31–33] YLWH-provider communication is a significant intervention target for stigma reduction; health communication has long been identified as a critical facilitator of engagement in HIV care,[34] but has been underutilized in the development of stigma reducing interventions. A review of HIV stigma reduction interventions revealed the publication of only a few controlled trials of provider educational interventions on stigma, even fewer trials focused on changing HIV providers’ stigmatizing behaviors, with no trials found that target YLWH-provider communication to reduce stigma.[35] Since, “communication shapes stigma… and is critical to creation of, spread of, coping with, and elimination of stigmas,”[36] training providers in communication skills that demonstrate acceptance and support autonomy can reduce the manifestation of HIV stigma. The Institute of Medicine reports that communication is a key clinical skill, but few guidelines exist to assist clinicians and interventionists communicate effectively with patients. To deliver Motivational Interviewing, providers are trained in communication techniques that significantly reduce judgement (and therefore reduce stigmatizing language) by mandating reflective and question-based interactions. Motivational Interviewing trained providers actively engage their clients, guide their clients to focus on feasible behavior change, evoke “change-talk” with guided questions, and promote autonomy-centered planning.[31,37,38] The Healthy Choices intervention was built with Motivational Interviewing, because: 1) Motivational Interviewing promotes behavioral change across multiple health and health-related outcomes; 2) Due to its success, Motivational Interviewing is embedded in the clinical guidelines for HIV care in the United States; 3) Motivational Interviewing’s emphasis on non-judgmental, non-stigmatizing, autonomy-supportive communication behaviors is particularly relevant to the adolescent and emerging adult developmental period, and 4) A meta-analysis found that Motivational Interviewing was the only effective youth intervention to demonstrate impact across the entire youth HIV continuum of care.[39–42]

Hatzenbuehler and Pachankis stigma model

We leveraged the Hatzenbuehler and Pachankis stigma model for sexual and gender minority (SGM) youth to inform this study;[30] this model provides both a theoretical framework and intervention targets that map directly to components of Motivational Interviewing, and by extension the Healthy Choices intervention. The Hatzenbuehler and Pachankis model is particularly fitting for stigma reduction studies that are focused on YLWH, because while not all YLWH identify as SGMs, epidemiological studies have shown that SGM populations are disproportionately represented among YLWH.[43] Hatzenbuehler and Pachankis state that, “…prejudice among health care providers, as well as structural aspects of the medical institution itself, may compromise the quality of care that LGBT youth receive,”[30] providing support for the need of multi-level, stigma reducing interventions, such as Healthy Choices.

The Hatzenbuehler and Pachankis model asserts that SGM youth are affected by stigma at multiple levels (e.g., structural, interpersonal, and individual), and that stigma disrupts cognitive, affective, interpersonal, and physiologic processes through four stigma-inducing mechanisms: vigilance, rumination, loneliness and the stress response.[30] Since anti-SGM stigma is multifaceted, interventions that are multifaceted are most likely to be effective in reducing stigma and the negative health consequences related to stigma exposure among SGM youth. Motivational Interviewing trained providers (structural level) have expertise in how to actively engage youth clients, evoke “change-talk” with guided questions (interpersonal level), and promote autonomy-centered planning (individual level). Motivational Interviewing’s emphasis on acceptance has the potential to mitigate vigilance and stress response mechanisms, while its promotion of autonomy and support of autonomous decision making could buffer the rumination mechanism. These components map to the science of stigma with acceptance being a key component of stigma reduction and autonomy support being a major facilitator of ownership of stigmatized identities.[44] The development of autonomy is a universal adolescent development task, and autonomy support has been associated with many positive adolescent outcomes.[45] This alignment between the components of Motivational Interviewing and the domains of the Hatzenbuehler and Pachankis stigma model inform our hypothesis in which we assert that participation in the Healthy Choices intervention will be associated with reductions in HIV-related stigma among YLWH.

METHODS

Healthy Choices intervention

The Healthy Choices intervention addresses alcohol use, promotes antiretroviral medication adherence, and seeks to improve related health outcomes among YLWH.[46] Healthy Choices is delivered by community health workers at the youth participant’s HIV clinic, YLWH’s home, or at any mutually agreed upon community venue. The intervention consists of four face-to-face sessions conducted over 10-weeks and employs Motivational Enhancement Therapy, an adaptation of Motivational Interviewing, throughout all sessions.[47] The four-session Healthy Choices intervention has been described in greater detail elsewhere.[46]

Randomization and intervention and trial procedures

After study enrollment, participants were randomized [1:1] to receive the intervention in their clinic or at a community-based location of the YLWH participant’s choice. The first intervention session was scheduled immediately after informed consent was obtained and the baseline assessment was completed. During the first session, study participants identified a behavioral target to address (supporting their personal autonomy); the community health worker used Motivational Interviewing to evoke “change talk” and guided the development of the participant’s individualized change plan. At this session, the community health worker provided personalized feedback on risk behaviors and health status while addressing participant’s knowledge gaps. In the second session, the YLWH participant and the community health worker focused on a second target behavior. In the third and fourth sessions, the community health worker reviewed and encouraged progress, supported the participant to problem-solve challenges related to their individualized plans, and discussed ways to sustain behavioral change.

Ethical approval

All study procedures were approved by the Institutional Review Boards (IRB) of Children’s Hospital of Los Angeles, Children’s Hospital of Philadelphia, Cook County-Health and Hospitals System, St. Jude Children’s Research Hospital, and Wayne State University. We employed a single IRB (sIRB) with participating academic centers relying on Wayne State University (IRB#097013B3E).

Participants

Participants were recruited from five ATN sites in Los Angeles, CA, Philadelphia, PA, Chicago, IL, Memphis, TN, and Detroit, MI. Inclusion criteria included living with HIV, being between the ages of 16 and 24 years, currently prescribed antiretroviral medications, having detectable viral load, able to understand and speak English, and reported use of alcohol in the past three months. Participants completed a computer-assisted self-interview at baseline and at follow-ups occurring at 16, 28, and 52 weeks post-intervention (N=183).

Statistical analyses

Descriptive statistics and bivariate analyses assessed simple associations. T-tests for continuous variables and chi-square for categorical variables were conducted to examine baseline differences between conditions (α=.05).

A latent growth modeling approach was used to assess between-condition differences in stigma across post-intervention time points. We utilized a stepwise model testing procedure. First, an initial model was estimated which included only an intercept. Second, we introduced a latent slope factor which modeled pre-versus-post intervention change in stigma scores. Finally, we introduced a second slope that modeled change in stigma over the post-intervention follow-up periods. We subsequently identified the best fitting growth (one slope vs. two slopes) by comparing changes in model χ² values and overall model fit. Good model fit was assumed when the χ²/df ratio was 3 or less, root-mean square error of approximation (RMSEA)≤0.05, Tucker-Lewis fit index (TLI)>0.95, and comparative fit index (CFI)>0.95. In the final model, we added a fixed effect of site and treatment condition as predictors of each latent growth factor. Missing data were handled within the context of full-information maximum likelihood estimation.

Measures

Sociodemographic measures were age, ethnicity, race, gender, sexual orientation, education, and clinic location. Stigma was assessed using the 10-item version of Berger’s (2001) Stigma Scale.[18,48] This scale consisted of four subscales. The first subscale on disclosure, included items, “you are very careful who you tell that you have HIV” and “you worry that people who know you have HIV will tell others.” The negative self-image sub-scale included three items, “you feel you are not as good a person as others because you have HIV,” “having HIV makes you feel unclean,” and “having HIV makes you feel that you are a bad person.” The public attitudes sub-scale included “most people think that a person with HIV is disgusting” and “most people with HIV are rejected when others find out.” The personalized sub-scale contained items, “you have been hurt by how people reacted to learning you have HIV,” “you have stopped socializing with some people because of their reactions to your having HIV,” and “you have lost friends by telling them you have HIV.” Response categories ranged from “strongly disagree” to “strongly agree” on a 4-point Likert scale.

RESULTS

Descriptive and bivariate statistics

The mean age of participants was 21.38 years (SD=1.86); 79.2% self-identified as male; 82.5% self-identified as Black or African American. Over half of the participants identified as gay or lesbian; 8.2% reported being infected with HIV perinatally. Bivariate analyses found no differences in gender identity (χ²=2.61, p=.271), race (χ²=3.78, p=.151), Hispanic ethnicity (χ²=0.02, p=.879), sexual orientation (χ²=0.24, p=.886), education (χ²=0.61, p=.739), contracting HIV prenatally (χ²=1.64, p=.200), clinic site location (χ²=1.67, p=.795), age (t=1.24, p=.215), and baseline stigma scores (t=−0.04, p=.965) between study arms. See Table I.

Table I.

Baseline Characteristics

	Total	Home	Clinic	Test Statistic
	n(%)	n(%)	n(%)
N	183	90 (49.2)	93(50.8)
Gender Identity				χ2 (2) = 2.6l, p = .271
Male	145(79.2)	73(81.1)	72(77.4)
Female	25(13.7)	9(10.0)	16(17.2)
Transgender or Gender Non-Conforming	13(7.1)	8(8.9)	5(5.4)
Hispanic	21(11.5)	10(11.1)	11(11.8)	χ2 (1) = 0.02, p = .879
Race				χ2 (2) = 3.7S, p = .151
Black	151(82.5)	79(87.8)	72(77.4)
White	8(4.4)	2(2.2)	6(6.5)
Other	24(13.1)	9(10.0)	15(16.1)
Sexual Orientation				χ2 (2) = 0.24, p = .886
Heterosexual	38(21.0)	18(20.2)	20(21.7)
Gay or Lesbian	103(56.9)	50(56.2)	53(57.6)
Bisexual or questioning	40(22.1)	21(23.6)	19(20.7)
Education				χ2 (2) = 0.61, p = .739
Less than high school	52(28.4)	27(30.0)	25(26.9)
High school/GED	67(36.6)	34(37.8)	33(35.5)
Some college	64(35.0)	29(32.2)	35(37.6)
Contracted HIV prenatally	15(8.2)	5(5.6)	10(10.8)	χ2 (2) =1.64, p = .200
Sites				χ2 (4) =1.67, p = .795
Los Angeles	24(13.1)	12(13.3)	12(12.9)
Philadelphia	41(22.4)	19(21.1)	22(23.7)
Chicago	45(24.6)	23(25.6)	22(23.7)
Memphis	12(6.6)	4(4.4)	8(8.6)
Detroit	61(33.3)	32(35.6)	29(31.2)
	Mean (SD)	Mean (SD)	Mean (SD)	T-Test
Age	21.38(1.86)	21.56(1.88)	21.22(1.82)	t(181) = 1.24, p = .215
Stigma Total	2.60(0.69)	2.60(0.70)	2.60(0.69)	t(181) = −0.04, p = .965

Latent Growth Modeling

The initial model, which included only the latent growth intercept had poor fit to the data (RMSEA=.151, CFI=.67, TLI=.82). The introduction of the first slope component, modeling pre-versus-post intervention change resulted in significant model improvement (χ²=266.41, df=13, p=<.001), overall fit for this model also improved (RMSEA=.04, CFI=.98, TLI=.97). Finally, the inclusion of the second slope, modeling post-intervention stigma level changes, did not result in a significant improvement in model fit (χ²=3.52, df =6, p=.31)] as well as reduction in overall model fit indices (RMSEA=.07, CFI=.94, TLI=.87). Thus, the one slope model was accepted as the best fitting model.

Results of a model including the fixed effects of site and treatment as a predictor of latent growth components suggested that the study conditions did not differ with respect to initial stigma scores at baseline (B_{clinic intercept}=0.01; 95% CI: −0.17, 0.21; p = .877). Similarly, intervention condition (mode of delivery) was not significantly associated with the amount of change from pre-to-post intervention (B_{clinic slope1}=−0.01; 95% CI: −0.02, 0.01; p=.589).

While there were no between-condition differences in latent growth factors, the expected value for the latent slope indicated that stigma decreased on average over the course of the study. The average pre-to-post intervention change (the intercept of slope 1) was statistically significant (B_{intercept slope1}=−0.02; 95% CI: −0.04, 0.01; p=.034). The overall pattern therefore suggests that, regardless of delivery modality, stigma scores significantly decreased from pre-to-post intervention and that the reduced levels of stigma post-intervention were sustained throughout the remaining follow-ups.

DISCUSSION

By utilizing two latent growth factors modeling stigma across time -- the latent intercept (this is the expected value for each YLWH study participant at baseline) and pre-post intervention slope -- we found that not only did YLWH who received the Healthy Choices intervention experience reductions in self-reported levels of stigma from pre- to post-intervention, but that these reductions were sustained over time. Each of our latent factors had its own intercept; that intercept was the expected value for that factor. Since the slope expected value was statistically significant, the trajectory of self-reported stigma over time differed from zero, zero being the slope of a line reflecting stability over time. Thus, participating in the intervention was associated with participants reporting reduced levels of stigma.

We also found that at no point did level of stigma return to its original pre-intervention level, and at the end of the data collection cycle, 52-weeks post-intervention, regardless of where the intervention was delivered (modality), levels of stigma converged. This finding provides evidence that the intervention produced similar reductions regardless of setting, suggesting that Motivational Interviewing is flexible and can influence outcomes across different environments. Prior non-HIV studies using Motivational Interviewing have produced similar positive outcomes. Specifically, Motivational Interviewing was shown to reduce structural and social stigma related to substance use disorders,[49] was associated with increased quality of life among people living with epilepsy,[50] and enhanced sense of control and acceptance among adolescents with type 1 diabetes.[51] Our findings add to the growing body of literature indicating that Motivational Interviewing may be able to reduce stigma across diverse conditions, populations, and settings.

Considering the negative health consequences associated with stigma among PLWH,[19–21] reducing HIV-related stigma is a high priority public health goal. However, targeting stigma reduction is challenging since stigma is structural, is embedded in systems, and may underscore YLWH-provider interactions.[14,34,35] While interventionists may seek to fortify the individual against the negative consequences of perceiving and experiencing HIV-related stigma, doing so does not address the point of origination, stigma produced by providers who hold greater social capital as compared to their YLWH patients. The Healthy Choices, Motivational Interviewing, intervention, addresses this imbalance in power by adapting the communication style of providers, potentially reducing their stigmatizing behaviors and attitudes.

Contexts and considerations

About 96% of our YLWH sample identified as Black (African American) or non-White, and nearly 80% identified as gay, bisexual, or questioning, so while our findings -- in the most direct sense -- speak to an intervention’s ability to reduce HIV-related stigma among YLWH, the demographics of our sample behoove us to consider the role of intersectionality;[52] and, therefore if the Healthy Choices intervention may also reduce intersectional stigmas related to race and identification as SGM. Although HIV stigma reduction is a notable priority, considering that people of color and SGMs (particularly, young Black men who have sex with men) are disproportionately affected by HIV,[53] there is an opportunity to expand this focus from being on HIV stigma reduction to targeting reductions in intersectional stigmas by addressing YLWHs’ experiences related to living with HIV, being young, being a person of color, and identifying as a SGM. Doing so could reduce compounded harm and negative health consequences experienced by YLWH who hold intersectional identities and may perceive intersectional stigmas.

In our analyses, we did not examine differences in growth intercept or slope by sociodemographic characteristics such as race or age, because there was no indication during the validation of randomization, that such adjustments were warranted to evaluate the association between treatment condition and growth trajectory. The randomization protocol was successful with respect to the sociodemographic composition of participants and stigma outcomes across study arms. This said, sociodemographics could predict trajectories of stigma over time, but their inclusion would not have altered the intervention effect, which was our primary outcome of interest.

Limitations

In this quasi-experimental trial, we randomized participants to test whether intervention delivery modality affected HIV-related stigma among YLWH. In assessing stigma outcomes across all participants, we found that the intervention was associated with reductions, regardless of delivery location. However, since we did not randomize participants to the intervention and a counterfactual control condition, we must be cautious about extending findings, and therefore suggest further testing to validate our evidence that the Healthy Choices, Motivational Interviewing intervention reduces HIV-related stigma. Second, study sites were all located in major metropolitan areas. YLWH in rural communities, receiving care at smaller clinics, may respond differently to this intervention. We did not examine differences in growth intercept or slope by sociodemographic characteristics, because doing so was outside of the scope of this specific project. We used the short version of Berger’s Stigma Scale. Although, its psychometric properties are comparable to the original scale, this version is slightly less sensitive.[54] Finally, longitudinal sustained reductions in stigma may have been attributable to the sequential, multiple administration this this scale (assessment effects).

Future research

While, we found evidence that participation in the Healthy Choices intervention was associated with reductions in HIV-related stigma among YLWH, within the scope of this study, we were unable to elucidate the stigma mechanisms that are proposed in the Hatzenbuehler and Pachankis model. While the Hatzenbuehler and Pachankis stigma model aligns with Motivational Interviewing in that trained providers (structural level) develop skills to evoke “change-talk” through guiding (interpersonal level) and can promote autonomy (individual level), we did not directly test these domains, and therefore, doing so remains an opportunity for further examination. Because we did not explore differences in growth intercept and slope by race and age, both of which could be associated with differential trajectories of stigma reduction, this is also an opportunity for future research that could inform scholarship on intersectionality. While our findings are promising and have potential to improve public health impact, they should be validated through the conduct of effectiveness trials in which outcomes from participants randomized to the intervention are compared to outcomes from participants randomized to counterfactual controls.

Conclusions

To date, few interventions have been tested for their effects on stigma reduction, and -- to our knowledge -- no intervention has been shown to be associated with sustained reductions in HIV stigma among YLWH, a diverse and often underserved population. Considering our finding that the Healthy Choices intervention was associated with sustained stigma reductions, and that HIV stigma reduction is a high priority public health target, intervention researchers who seek to improve the health of YLWH should further evaluate, through replication and validation, if embedding Motivational Interviewing in behavior change interventions does indeed reduce HIV-related stigma and sustain those reductions over time.

Figure I.

Longitudinal Stigma Scores from Baseline to 1-Year Post-Intervention

Table II.

Results of Latent Growth Curve Models Predicting Stigma Post-Intervention

	Stigma
	B	95% CI	P value
Intercept
Threshold/Intercept	2.53	(2.32, 2.73)	<.001
Clinic	0.01	(−0.17, 0.21)	.877
Sitef†
Los Angeles, CA	−0.15	(−0.41, 0.12)	.288
Chicago, IL	−0.01	(−0.27, 0.26)	.954
Memphis, TN	0.19	(−0.15, 0.53)	.263
Detroit, MI	0.07	(−0.18, 0.32)	.695
Slope 1
Intercept	−0.02	(−0.04, 0.01)	.034
Clinic	−0.01	(−0.02, 0.01)	.589
Sites
Los Angeles, CA	0.01	(−0.01, 0.04)	.409
Chicago, IL	−0.01	(−0.04, 0.02)	.524
Memphis, TN	−0.02	(−0.06, 0.03)	.512
Detroit, MI	0.01	(−0.02, 0.03)	.695

^†Model was analyzed with Philadelphia as the referent category.