Effects of HIV and alcohol stigma on biomarker-confirmed alcohol use following a peer-delivered intervention in South Africa

Highlights

• •HIV and alcohol use (AU) stigmas and biomarkers of AU were evaluated over six months.
• •The study tested whether HIV and AU stigma predicted later AU.
• •Enacted alcohol and internalized HIV stigma predicted AU at three-month follow-up.
• •Internalized alcohol stigma predicted AU at six-month follow-up.
• •Stigma was not a mediator of intervention outcomes.

Abstract

Background

Despite documented relationships between multiple forms of stigma and health outcomes, limited research has examined the effects of both HIV and alcohol stigmas on objectively measured alcohol use. Research is needed to better understand relationships between stigma and health outcomes to inform intervention efforts that reduce stigma.

Methods

Participants (N = 60) who met criteria for unhealthy alcohol use and suboptimal HIV medication adherence were recruited from two public HIV care sites in South Africa. Internalized alcohol stigma, enacted alcohol stigma, internalized HIV stigma and phosphatidylethanol [PEth] levels, an alcohol use biomarker, were assessed at baseline, and three- and six-months post-baseline. Participants were randomized to a peer-delivered behavioral intervention or enhanced treatment as usual (facilitated referral to a co-located substance use treatment program). A cross-lagged panel structural equation model with three mediators was used to test lagged effects of stigma on PEth outcomes and whether stigma mediated the effects of the peer intervention.

Results

Significant lagged effects were identified such that higher levels of enacted alcohol stigma and internalized HIV stigma at baseline separately predicted higher PEth levels at three-month follow-up. Higher levels of internalized alcohol stigma predicted higher PEth levels at six-month follow-up. No significant intervention effects were found on stigma (ps > 0.05). Stigma did not mediate the effect of the intervention.

Conclusions

Findings suggest that higher HIV and alcohol stigma predict greater alcohol use. Future research should explore how stigma reduction strategies can be incorporated into peer-delivered interventions and evaluate the effects of reducing stigma on health outcomes.

Trial Registration: ClinicalTrials.gov NCT03529409.

1. Introduction

HIV stigma is a well-documented barrier to HIV care engagement (Katz et al., 2013), including in South Africa (SA), the country with the highest HIV burden globally (UNAIDS, 2024). While SA has the world’s largest public antiretroviral therapy (ART) program, nearly 20 % of people with HIV (PWH) are not taking ART (UNAIDS, 2024). Of those on ART, approximately 500,000 people are not virally suppressed (UNAIDS, 2024), suggesting suboptimal ART adherence. Although HIV is viewed as a common chronic disease in high-burden HIV settings such as SA, HIV-related stigma is still common (Hargreaves et al., 2018).

Concurrently, problematic alcohol use (AU), defined as AU that will likely lead to health or other problems (i.e., social, financial, legal; Humeniuk et al., 2010), is common among PWH in SA. Nearly 30 % of PWH in SA are estimated to have an AU disorder (Necho et al., 2020). PWH in SA who are using alcohol also experience multiple levels of stigma—defined as discrediting that causes one to be viewed as an inferior person (Goffman, 1963) due to AU. Alcohol stigma is prevalent amongst PWH in SA (Kalichman et al., 2020, Magidson et al., 2019, Regenauer et al., 2020, 2022), and there is evidence that PWH in SA use alcohol and other drugs to cope with HIV stigma (Regenauer et al., 2020). This may be due to stigmatization leading to high levels of shame, which then may lead to higher engagement in avoidance behaviors, (e.g., drinking alcohol to avoid emotionally experiencing shame) (Drapalski et al., 2013). Higher alcohol stigma has been associated with worse HIV outcomes (Kalichman et al., 2020, Regenauer et al., 2022) and worse alcohol or other drug problem severity (Crapanzano et al., 2018, Luoma et al., 2017). However, this research has largely taken place in high-income countries.

Stigma can occur through multiple mechanisms: enacted (i.e., experienced discrimination), anticipated (i.e., expectation of future discrimination), and internalized (i.e., application of stigma to oneself) (Earnshaw and Chaudoir, 2009, Smith et al., 2022). Although all levels of stigma exist concurrently and may intersect, few studies have examined the relationships between multiple forms (i.e., HIV and alcohol) and levels of stigma (i.e., anticipated, internalized, enacted) and AU outcomes, particularly in low- and middle-income countries with high HIV burden such as SA.

Peer interventionists, individuals with lived substance use and recovery experience, may reduce stigma by sharing their lived experience and normalizing experiences with problematic substance use (ASTHO, 2020), as well as through building connection to a community of like individuals (Brener et al., 2021). While sharing lived experiences and instilling hope of recovery may uniquely shift internalized stigma (Corrigan and Rao, 2012, Kleinman et al., 2024, Thornicroft et al., 2016), to our knowledge, few studies have examined if contact with a peer decreases internalized stigma for alcohol and/or HIV. Thus, the current study aimed to examine longitudinal associations between both internalized and enacted HIV and alcohol stigmas and biomarker-verified AU. Further, we tested whether a peer-delivered behavioral intervention reduced these stigmas, and whether stigma mediated the effects of the intervention on alcohol outcomes.

2. Methods

2.1. Participants and procedures

This study conducted secondary data analysis using clinical trial data from a pilot Type 1 hybrid effectiveness-implementation randomized controlled trial (Curran et al., 2012) that evaluated “Khanya,” a peer-delivered intervention designed to improve ART adherence and reduce alcohol and other substance use among PWH (Magidson et al., 2020, Magidson et al., 2021). All participants were PWH accessing HIV services in Khayelitsha, a low-resource, peri-urban community near Cape Town, SA who exhibited at least moderate AU and past three-month ART nonadherence. Participants completed assessments of HIV and alcohol stigma and dried blood spots for alcohol biomarker testing at three-month intervals over six months. This study was approved by the University of Cape Town’s human research ethics committee and the City of Cape Town’s health department. See (Magidson et al., 2020, Magidson et al., 2021) for more details about the parent study.

Khanya intervention. Khanya is a six-session, peer-delivered behavioral intervention that includes a combination of evidence-based approaches for medication adherence and AU, including Life Steps, behavioral activation, mindfulness, and relapse prevention (Belus et al., 2020). Sessions occurred weekly, each lasting ∼ 60 min. Formative qualitative research was conducted to design the intervention was guided by the ADAPT-ITT framework and included theatre testing methods (Wingood & DiClemente, 2008). This phase included adaptation to ensure the feasibility and acceptability of all components individually as well as the combination of the components (Magidson et al., 2020, Magidson et al., 2021, Magidson et al., 2020, Rose et al., 2022). During this process, key stakeholders suggested that a person with lived substance use experience should deliver the intervention because this person may be less judgmental towards substance use than other provider types (Magidson, Fatch, et al., 2019). As stigma reduction was not a primary aim of the Khanya study, intervention content did not explicitly focus on stigma. However, the peer interventionist was trained in using nonjudgmental communication and sharing their lived experience of AU when they believed it would be beneficial for the participant as opposed to the interventionist (e.g., improve the client’s understanding of skills, normalize treatment challenges). The peer had lived AU and recovery experience, and did not have personal lived experience of HIV, although they were educated on HIV and had extensive experience working with PWH. Weekly supervision with a US-based clinical psychologist encouraged the appropriate use of self-disclosures (i.e., beneficial to the participant, related to the intervention content and goals) and providing examples of opportunities for appropriate self-disclosures. Further details on the study procedures, control arm, methods, and main outcomes can be found elsewhere (Magidson et al., 2020, Magidson et al., 2021).

2.2. Measures

Stigma. All stigmas were measured using the enacted (alcohol; e.g. “family members have thought that I cannot be trusted because of my AU history”) and internalized (alcohol and HIV; e.g. “I feel ashamed of having HIV/used alcohol”) subscales of the Stigma Mechanism Scales (Earnshaw et al., 2013, Smith et al., 2016). The Substance Use Stigma Mechanisms Scale was adapted to ask about alcohol specifically given evidence that alcohol and drug use are conceptualized differently in this context (Regenauer et al., 2020). This measure has previously been used in this context and was translated and back-translated for isiXhosa by bilingual translators (Anvari et al., 2022, Regenauer et al., 2022). All stigma items were scored on a 1 (strongly disagree) to 5 (strongly agree) scale, with higher scores indicating greater stigma. Subscale items were then averaged to create a final subscale score.

Biomarker-confirmed alcohol consumption. Phosphatidylethanol (PEth) was used as a biomarker of AU, obtained via dried blood spots (DBS), and tested using liquid chromatography-tandem mass spectrometry after methanol was extracted (Jones et al., 2011). PEth was treated as a continuous variable as scores were normally distributed. Detectable levels of alcohol are ≥ 8 ng/mL and unhealthy levels of AU are considered ≥ 50 ng/mL (Magidson, Fatch, et al., 2019). The primary analyses did not find an effect of the intervention on biomarker-verified AU, as both groups displayed significant decreases over time.

2.3. Data analytic plan

We first conducted exploratory data analysis for the variables of interest. Time trends were observed for the main outcomes; for the hypothesized mediators (i.e., stigmas), we found the correlations between the mediators were non-collinear (Pearson rs < 0.49). The most parsimonious way to test all three study aims was to conduct a single cross-lagged panel model (estimated using maximum likelihood) with three mediators using structural equation modeling. We tested (1) the longitudinal associations between stigmas and PEth; (2) whether the peer-delivered intervention shifted stigma; and (3) whether changes in stigma mediated the effects of the intervention. We assumed a lagged effect of both intervention on mediator, and mediator on outcome (Cole and Maxwell, 2003, MacKinnon, 2008, Selig and Preacher, 2009). This model takes the intervention arm into account, and assumes that outcomes are affected by mediators from the last time point, while the intervention effect influences outcome and mediator variables at both post-intervention and follow-up (see Fig. 1). All variables were standardized before model fitting. Analyses were conducted using the “lavaan” package in R (Rosseel, 2012). We followed the package’s recommendation and applied pairwise deletion to handle missing values; few variables had missing values, all of which were less than 10 %. Confidence intervals (CIs) for the indirect effects were calculated using bootstrapping to get bias-corrected CIs (Preacher and Hayes, 2004, Preacher and Hayes, 2008).

Fig. 1.

Cross-lagged panel model testing the hypotheses of interest. All three stigma mediators are included in a single model.

3. Results

One participant from the parent study was not included because they did not present with AU (only drug use), resulting in a sample size for this analysis of N = 60. Participants were 98 % isiXhosa-speaking Black African. Average age of participants was 37.3 (SD = 9.5) years old, 53 % (n = 32) were women, and 22 % (n = 13) had high school or above education. The majority of participants (83 %, n = 50) demonstrated unhealthy levels of drinking using PEth. Of the participants randomized to the intervention, 70 % completed all six sessions. Four participants did not complete the three- and six-month follow-up assessments. Additional demographic details can be found in the primary outcomes paper (Magidson et al., 2021).

Lagged effects of stigmas on PEth. Significant lagged effects of the stigmas were found on PEth (see Table 1). After controlling for baseline PEth and the cross-sectional association between stigma and PEth, higher internalized alcohol stigma scores were significantly associated with PEth at six-months (B = 0.211; p = 0.017). Higher internalized HIV stigma at baseline was significantly associated with higher PEth values at three-months (B = 0.205, p = 0.004). Higher enacted alcohol stigma was significantly associated with higher PEth values at three-months (B = 0.152, p = 0.034).

Table 1.

Mean and standard deviation of stigmas by time and intervention group and results of cross-lagged panel model with HIV and alcohol stigmas as mediators between intervention and outcomes.

Stigma type	Khanya (n = 30)			ETAU (n = 30)
	Baseline (T1)	Post-intervention (T2)	Follow-up (T3)	Baseline (T1)	Post-intervention (T2)	Follow-up (T3)
Internalized alcohol Mean(SD)	2.82 (1.36)	3.07 (1.23)	3.06 (1.16)	2.47 (1.22)	2.72 (1.05)	2.66 (1.05)
Internalized HIV Mean(SD)	2.01 (1.18)	2.28 (0.85)	2.32 (0.96)	1.94 (1.15)	2.40 (1.03)	2.48 (1.12)
Enacted alcohol Mean(SD)	1.58 (0.59)	1.40 (0.59)	1.46 (0.76)	1.45 (0.58)	1.29 (0.56)	1.39 (0.56)
PEth levels, M (SD), % unhealthy use	90.00 %	92.31 %	92.00 %	76.67 %	75.00 %	78.57 %
Effects			Alcohol: PEth
			B	SE	p
Aim 1
Int alc stigma T1 −> PEth T2			−0.123	0.082	0.131
Int alc stigma T2 −> PEth T3			0.211	0.088	0.017
Enact alc stigma T1 −> PEth T2			0.152	0.071	0.034
Enact alc stigma T2 −> PEth T3			−0.043	0.086	0.622
Int HIV stigma T1 −> PEth T2			0.205	0.072	0.004
Int HIV stigma T2 −> PEth T3			0.028	0.084	0.740
Aim 2
Ix −> Int alc stigma T2			−0.180	0.226	0.424
Ix −> Int alc stigma T3			−0.153	0.167	0.359
Ix −> Enact alcohol stigma T2			−0.118	0.240	0.622
Ix −> Enact alcohol stigma T3			−0.017	0.222	0.940
Ix −> Int HIV stigma T2			0.169	0.209	0.418
Ix −> Int HIV stigma T3			0.090	0.213	0.674
Aim 3
Ix −> PEth T2			0.167	0.164	0.310
Ix −> PEth T3			−0.131	0.176	0.459
Model indirect effects, estimate [95 % CI]			−0.028 [−0.634, 0.088]

Note. Bolded values are significant at p < 0.05. Ix = Intervention group (effect for enhanced treatment as usual group [ETAU] compared to Khanya). Int = internalized. Alc = alcohol. Enact = enacted. T1 = baseline. T2 = post-treatment. T3 = follow-up.

Changes in stigma. There were no intervention effects on any stigma outcomes (ps > 0.05) (see Table 1). Further, none of the model indirect effects were significant (see Table 1), indicating that stigma did not mediate the effects of the intervention on PEth outcomes.

4. Discussion

This study contributes to the literature by evaluating the effects of both internalized and enacted HIV and alcohol stigma on objectively assessed AU. Results indicated that higher levels of enacted alcohol stigma and internalized HIV stigma at baseline separately predicted greater AU at three-months. Higher levels of internalized alcohol stigma at three-months predicted higher levels of AU at six-months.

One theory that may explain the significant lagged effects of HIV stigma at the three-month follow-up and higher AU stigma at six-month follow-up is the social identity threat conceptual model (Major et al., 2017). This model highlights that situational cues can heighten awareness of stigma, impacting health outcomes. Participants with poor ART adherence at screening (an inclusion criteria) may have encountered more HIV-related cues early in the intervention, increasing awareness of HIV stigma and affecting PEth at three-months. A significant intervention effect on ART adherence was found at three-month follow-up in the primary analyses (Magidson et al., 2021), suggesting that improved ART adherence by three months may have shifted focus to AU-related cues, increasing AU stigma and affecting PEth at six months. Thus, changes in ART adherence over time may moderate these findings and should be explored in future research. Similarly, enacted alcohol stigma by providers at baseline may have led to later increased internalized alcohol stigma, subsequently leading to increased PEth at six months. It is also possible that changes in stigma are what predicts AU levels, which was not the focus of the current analyses.

Despite literature that suggests that social contact with peers can shift AU stigma (Livingston et al., 2012), this peer-delivered intervention did not significantly reduce stigma, nor did stigma mediate intervention effects. Although we hypothesized that working with a person with lived alcohol experience could decrease agreement with stigma and improve views of oneself, it is possible that contact with one person was not enough to counteract internalized stigma (Corrigan & Rao, 2012b). Further, although a significant intervention effect is not necessary to test for mediation (O’Rourke & MacKinnon, 2018), the pilot trial main outcomes results demonstrated that both groups (Khanya and ETAU) improved with regarding to AU, and Khanya only demonstrated improved AU outcomes vs. ETAU in a subsample of individuals who used both AU and other substances (Magidson et al., 2021). These results suggest Khanya may be uniquely suited for individuals with more severe substance use problems, which is being tested in a larger ongoing clinical trial (NCT05933226). Results must be interpreted in the context of study strengths and limitations. Strengths included the use of a randomized, longitudinal design to bolster study power (Hertzog et al., 2008), a difficult-to-reach population of PWH with unhealthy AU, and the use of alcohol biomarkers. Regarding limitations, this was a secondary data analysis from a pilot trial with a relatively small sample size, which prevented us from analyzing the intersection or layering of HIV and alcohol stigma, which has been described as prevalent among PWH and substance use disorder (Stutterheim et al., 2016) or controlling for possible confounders. Additionally, the sample had relatively low baseline levels of HIV and AU stigmas, potentially leading to a floor effect in stigma reduction over time. This may be due to underreporting, given the sensitive nature of the topic and that all assessments were administered verbally with research staff. Little is known about how representative these stigma levels are of the broader population of PWH who actively use alcohol in this or similar communities, limiting the generalizability of these results.

5. Conclusion

Overall, study findings suggest that the presence of either alcohol or HIV stigmas have the potential to negatively impact alcohol outcomes over time. To our knowledge, this is the first study to examine if working with a peer interventionist shifts internalized alcohol and HIV stigma in a low- to middle-income country. Results suggest that reduction of stigma may not simply occur as a by-product of engaging with a peer interventionist, rather may need to be directly targeted by the intervention. Future work should consider how peer-delivered, structured behavioral interventions can intentionally incorporate stigma reduction, while also remaining efficacious for reducing unhealthy AU.

CRediT authorship contribution statement

Jennifer M. Belus: Writing – original draft, Visualization, Methodology, Conceptualization. Morgan S. Anvari: Writing – review & editing. Hongjie Ke: Formal analysis. Kristen S. Regenauer: Writing – review & editing, Project administration, Methodology, Data curation. Tianzhou Ma: Formal analysis. Bronwyn Myers: Writing – review & editing, Supervision. Lena S. Andersen: Writing – review & editing, Supervision. John A. Joska: Writing – review & editing, Supervision, Funding acquisition. Jessica F. Magidson: Writing – review & editing, Supervision, Methodology, Funding acquisition, Conceptualization.

Declaration of competing interest

The authors declare the following financial interests/personal relationships which may be considered as potential competing interests: Morgan S. Anvari reports financial support was provided by National Institute on Drug Abuse. Jennifer Belus’ time on the project was supported through the Swiss National Science Foundation (Grant # PZ00P1_201690). Kristen S. Regenauer reports financial support was provided by National Institute on Drug Abuse. Jessica F. Magidson reports financial support was provided by National Institute on Drug Abuse. If there are other authors, they declare that they have no known competing financial interests or personal relationships that could have appeared to influence the work reported in this paper.

Data availability

Data will be made available on request.