Social factors associated with trajectories of HIV-related stigma and everyday discrimination among women living with HIV in Vancouver, Canada: longitudinal cohort findings

Abstract

Introduction:

Women living with HIV (WLHIV) experience stigma rooted in social inequities. We examined associations between social factors (food insecurity, housing insecurity, violence, sexual minority identity, substance use) and HIV-related stigma and Everyday Discrimination trajectories among WLHIV.

Methods:

This community-based open longitudinal cohort study with WLHIV living in and/or accessing HIV care in Metro Vancouver, Canada, plotted semiannual averages (2015–2019) of recent (past 6-month) HIV-related stigma and Everyday Discrimination. We examined distinct trajectories of HIV-related stigma and Everyday Discrimination using Latent Class Growth Analysis (LCGA), and baseline correlates of each trajectory using multinomial logistic regression.

Findings:

Among participants (HIV-related stigma sample: n=197 participants with n=985 observations; Everyday Discrimination sample: n=203 participants with n=1096 observations), LCGA identified two distinct HIV-related stigma and Everyday Discrimination trajectories: sustained low and consistently high. In multivariable analysis, concurrent food and housing insecurity (adjusted odds ratio [AOR]: 2.15, 95% Confidence Interval [CI] 1.12–4.12) and physical/sexual violence (AOR: 2.57, 95% CI: 1.22–5.42) were associated with higher odds of the consistently high (vs. sustained low) HIV-related stigma trajectory. Sexual minority identity (AOR: 2.84, 95% CI: 1.49–5.45), concurrent food and housing insecurity (AOR: 2.65, 95% CI: 1.38–5.08), and non-injection substance use (less than daily vs. none) (AOR: 2.04, 95% CI: 1.03–4.07) were associated with higher odds of the consistently high (vs. sustained low) Everyday Discrimination trajectory.

Conclusions:

Social inequities were associated with consistently high HIV-related stigma and Everyday Discrimination among WLHIV. Multi-level strategies can address violence, economic insecurity, intersecting stigma, and discrimination to optimize health and rights among WLHIV.

Introduction

Stigma and discrimination experienced by women living with HIV (WLHIV) is intricately linked with other social inequities experienced in daily lives [1,2]. Intersectional approaches to stigma examine interlocking social categories linked with power imbalances, and how these social inequities shape wellbeing [1,3]. There is a growing evidence base regarding how HIV-related stigma, and discrimination based on other social identities, change over time and what drive such changes. As HIV-related stigma and discrimination are related to poorer mental health [4] and worse HIV clinical outcomes [5], better characterizing stigma and discrimination drivers and trajectories can inform practice, research and policy.

Among WLHIV, HIV-related stigma is a well-established stressor linked with poorer HIV outcomes [6,7]. There is emerging literature on other forms of discrimination among WLHIV. Cross-sectional studies, for instance, have shown among WLHIV in Canada that HIV-related stigma, racial discrimination, and gender discrimination are associated with one another [6], and linked with poorer health-related quality of life [8]. Other cross-sectional analyses revealed higher racial and gender discrimination among sexual minority WLHIV compared with heterosexual counterparts [9]. Experiencing violence may be linked with increased HIV-related stigma exposure [10,11] through violence-related shame, blame, and mistreatment [10,12]. Few studies with WLHIV have examined violence victimization at large as a predictor of HIV-related stigma or other forms of discrimination.

People living with HIV are disproportionately affected by food and housing insecurity [13–15] which themselves may drive HIV-related stigma. For instance, among WLHIV in Canada, concurrent food and housing insecurity was linked with higher HIV-related stigma and racial discrimination in cross-sectional studies [16], and food and housing insecurity were individually associated with higher HIV-related stigma trajectories [17]. Poverty stigma was linked with poorer HIV care outcomes among WLWHIV in the United States [18], as was experiencing combinations of poverty, sex and racial stigma [19]. In sum, these studies signal that HIV-related stigma drivers, and other forms of day-to-day mistreatment—referred to as ‘everyday’ discrimination [20]—warrant further investigation among WLHIV who disproportionately experience poverty [13–15], violence [10,12], and discrimination based on race [21–23], gender [6,24], sexual orientation [9,25,26], among other socially devalued identities [1,3].

To address knowledge gaps regarding drivers of stigma and discrimination, this study aimed to estimate the associations between food and housing insecurity, violence, and socio-demographic characteristics with trajectories of HIV-related stigma and Everyday Discrimination among a cohort of WLHIV in Vancouver, Canada.

METHODS

Study setting and population

Data for these analyses came from the community-based open longitudinal cohort ‘Sexual Health and HIV/AIDS: Longitudinal Women’s Needs Assessment’ (SHAWNA) study with WLHIV living in and/or accessing HIV care in Metro Vancouver, Canada. Methods are described elsewhere [24]. Eight waves of data were collected between 2015–2019 at baseline and at 6-month intervals with participants who were aged 14 and older, self-identified as women (transgender and cisgender inclusive), living with HIV, and able to provide informed consent.

Data collection

Outcomes:

Both outcomes were time updated to reflect events in the last six months at each semiannual study visit. Participants completed a validated HIV-related stigma scale, using a modified version of Wright’s 10-item HIV-related stigma scale [27] rated on a scale from 1 to 5 (internal consistency in this study: 0.86) [11].

We used the Everyday Discrimination Scale [20] (internal consistency in this study: 0.91) to assess frequency of daily experiences of mistreatment, micro-aggressions, and disrespect with a scale of 1 to 5. [20]. These items assessed discrimination experiences without requiring an attribution; this can be more valid for persons holding multiply marginalized identities as it can include any reason for discrimination [28].

Exposures:

Baseline demographic and social-structural variables included: age (in years) race/ethnicity (self-identified Indigenous, Black and other racialized identity vs. White), sexual identity (lesbian, gay, bisexual, asexual, Two-Spirit, queer, other vs. heterosexual), gender identity (trans [transgender, transsexual, other transfeminine identity], non-binary, genderqueer, and/or Two-Spirit vs. cisgender). We assessed education (completed high school or further vs did not complete high school).

Other explanatory variables captured recent (past 6-months) events at baseline: age (continuous); physical and/or sexual violence experiences (any); injection drug use (daily, less than daily, or none); non-injection drug use (daily, less than daily, or none); food insecurity (modified Cornell-Radimer Hunger Scale [29], as previously described [30]); and housing insecurity (Canadian Observatory of Homelessness definition of unsheltered/unstably housed [31], as previously described [32]. We created a composite variable for food and/or housing secure (having both or just one of food security or housing security) vs food and housing insecure (having both food and housing insecurity concurrently).

Statistical analyses

Using data from 2015–2019, semiannual averages in recent (past 6-month) HIV-related stigma [27] and Everyday Discrimination [20], were plotted. Latent class growth analysis (LCGA) was then used to identify distinct trajectories of HIV-related stigma and Everyday Discrimination, and baseline correlates of each trajectory were examined using multinomial logistic regression.

We conducted LCGA to characterize longitudinal trajectories of HIV-related stigma and Everyday Discrimination separately. LCGA assumes populations comprise a combination of distinct trajectory patterns, as defined by an outcome of interest.[33] We applied LCGA to ascertain the number of ‘clusters’ of participants with similar patterns in outcomes across time [34]. The sample was restricted to participants who completed at least 3 study visits to consider up to a quadratic polynomial function for each group. Our analyses examined for 1–3 possible latent trajectory groups, due to concerns about small sample size [35,36]. While these restrictions result in our sample having sufficient power [35,36], the best-fitting model with respect to the final number of latent trajectory groups was selected based on Bayesian Information Criterion (BIC) as it outperforms other model selection tools [37], particularly with small sample sizes [38]. Individuals were assigned to the trajectory group in which they had the highest probability of group membership, and we assessed the average posterior probabilities of group membership as an internal reliability indicator (0.80 indicates participants with similar trajectory patterns are grouped together, and distinct from individuals in other trajectories) [34].

Once trajectory groups were determined, we calculated descriptive statistics for exposure variables and stratified by trajectory group. These were summarized as frequencies and proportions for categorical variables and medians and first to third quartile (Q1-Q3) for continuous variables. Differences were assessed using Wilcoxon rank-sum test or Kruskal-Wallis test by ranks for categorical variables with two groups and more than two groups, respectively, and Spearman’s rank correlation coefficient for continuous variables. Generalized logits models for nominal response data were used to assess bivariate associations between explanatory variables as measured at first interview and outcome trajectory groups, considering variables significant at p<0.10 for inclusion in the multivariable model. We calculated odds ratios (OR) and adjusted odds ratios (AORs) with 95% Confidence Intervals (95%CI) and two-sided p-values, with the final multivariable model adjusting for all of the explanatory variables identified in the best-fitting model for each outcome. Beginning with the full model, we used a manual backward elimination process to identify the best fitting model, as indicated by the lowest quasi-likelihood under the independence model criterion (QIC).We used SAS version 9.4 for analyses (SAS Institute Inc., Cary, North Carolina, USA) and SAS and Excel to generate graphics.

Ethical considerations

Participants provided informed consent prior to completing surveys. Research ethics board approval was obtained from Providence Health/University of British Columbia Research Ethics Board.

RESULTS

The sample included 203 participants with Everyday Discrimination scores (n=1096 observations), and 197 participants with HIV-related stigma scores (n=985 observations) over 4 years of follow-up. Socio-demographic data is presented in Table 1 for the larger sample of 203 participants. Semiannual time trend plots showed little variation over time for HIV-related stigma or Everyday Discrimination.

Table 1.

Socio-demographic characteristics and HIV-related Stigma and everyday discrimination Scores among ‘Sexual Health and HIV/AIDS: Longitudinal Women’s Needs Assessment’ (SHAWNA) study participants (n=203)

	Median (quartile) or N (%) (n=203)
Age at baseline1	46.00 (lower quartile: 38.00, upper quartile: 53.00)
Gender identity2
Trans	24 (11.82)
Cisgender	179 (88.18)
Sexual orientation
LGBQ2S	94 (46.31)
Heterosexual	109 (53.69)
Educational attainment
Less than high school	110 (54.19)
High school or greater	93 (45.81)
Race/ethnicity
Indigenous 3	125 (61.58)
White	64 (31.53)
Other racialized race/ethnicity	14 (6.90)
Racialized vs. white
Racialized	139 (68.47)
White	64 (31.53)
Food security1 (n=202)
Food insecure	143 (70.79)
Food secure	59 (29.21)
Housing security1
Housing insecure	117 (57.64)
Housing secure	86 (42.36)
Concurrent food and housing insecurity (n=202)
Yes concurrent food/housing insecurity	83 (41.09)
No concurrent food/housing insecurity	119 (58.91)
Sexual and/or physical violence1 (n=187)
No	149 (79.68)
Yes	38 (20.32)
Frequency of injection drug use1 (n=200)
None	110 (55.00)
Daily	36 (18.00)
Less than daily	54 (27.00)
Frequency of non-injection substance use1 (n=201)
None	114 (56.72)
Daily	22 (10.95)
Less than daily	65 (32.33)

LGCA identified two distinct trajectories of HIV-related stigma and Everyday Discrimination: sustained low (HIV-related stigma: n=121; Everyday Discrimination: n=112) and consistently high (HIV-related stigma: n=61; Everyday Discrimination: n=71). In multivariable analysis (Table 2), baseline concurrent food and housing insecurity (Adjusted Odds Ratio (AOR): 2.15, 95% Confidence Interval: 1.12–4.12) and baseline physical/sexual violence (AOR: 2.57, 95% CI: 1.22–5.42) were associated with higher odds of the consistently high (vs. sustained low) HIV-related stigma trajectory. Identifying as a sexual minority (AOR=2.84, 95% CI=1.49–5.44), baseline concurrent food and housing insecurity (AOR=2.65, 95% CI=1.38–5.01), and non-injection drug use (less than daily vs. none) (AOR: 2.04, 95% CI: 1.03–4.07) were associated with the consistently high (vs. sustained low) Everyday Discrimination trajectory. In sum, key findings were a) baseline concurrent food and housing insecurity and physical/sexual violence were associated with higher HIV-related stigma trajectories, and b) baseline concurrent food and housing insecurity, non-injection drug use, and sexual minority identity were linked with higher Everyday Discrimination trajectories.

Table 2.

Associations between baseline social factors and HIV-related stigma and Everyday Discrimination trajectories among ‘Sexual Health and HIV/AIDS: Longitudinal Women’s Needs Assessment’ (SHAWNA) study participants

Effect	Everyday Discrimination trajectory group (n=203) (reference: sustained low Everyday Discrimination trajectory group)	Adjusted Odds ratio1	95% Confidence Interval
Sexual minority orientation (Lesbian, gay, bisexual, queer or Two Spirit (LGBQ2S) vs. heterosexual)	Consistently high Everyday Discrimination	2.84*	1.49	5.44
Baseline concurrent food and housing insecurity (yes vs. no)	Consistently high Everyday Discrimination	2.65*	1.38	5.08
Non-injection drug use (less than daily vs. none)	Consistently high Everyday Discrimination	2.04*	1.03	4.07
Non-injection drug use (daily vs. none)	Consistently high Everyday Discrimination	1.11	0.34	3.51
Education (graduated high school vs. not)	Consistently high Everyday Discrimination	0.57	0.29	1.11
Effect	HIV-related stigma trajectory group (n=197) (reference: sustained low HIV-related stigma trajectory group)	Adjusted Odds Ratio2	95% Confidence Interval
Baseline concurrent food and housing insecurity (yes vs. no)	Consistently high HIV-related stigma	2.15*	1.12	4.12
Baseline sexual or physical violence (yes vs. no)	Consistently high HIV-related stigma	2.57*	1.22	5.42

^*p<0.05

^1.The AORs are adjusted for education level.

^2.This best fitting model includes two predictor variables (food and housing insecurity, sexual or physical violence). The AORs are adjusted for each other.

DISCUSSION

Among WLHIV in this Vancouver study, we found consistent levels of HIV-related stigma and Everyday Discrimination over 4-years of follow-up. This signals the importance of better understanding social factors associated with the chronicity of stigma [39] to inform stigma/discrimination reduction interventions. We identified violence, concurrent food and housing insecurity, sexual minority identity, and drug use as social factors linked with high levels of stigma and discrimination exposure over time.

Our findings build on prior work that demonstrates that food and housing insecurity may result in higher HIV-related stigma exposure [17,40]. Findings extend beyond HIV-related stigma drivers to reveal that these resource insecurities are also associated with higher trajectories of Everyday Discrimination—itself a chronic stressor associated with a range of poorer health outcomes [41,42], increased health care utilization [41,42], and chronic disease [42]. Stigma and discrimination linked with food/housing insecurity reflect the ways in which poverty is stigmatized and perceived as “failure” and/or a disruption of public spaces [43–46]. WLHIV may experience the “pathologization of poverty” (p. 78) [43] which refers to moral judgment toward persons receiving social assistance. Our finding that substance use is associated with Everyday Discrimination corroborates research on associations between substance dependence and HIV-related stigma [47], and reflects the ways in which shaming and blaming people who use drugs as ‘immoral’ are entrenched in social norms and HIV discourse [48].

We found that recent violence was associated with consistently higher HIV-related stigma over time, corroborating research on how violence among WLHIV may lead to shame and mistreatment—central aspects of stigma processes [10,12]. To illustrate, prior research with WLHIV in Vancouver documented that verbal/physical violence attributed to HIV positive serostatus was associated with higher HIV-related stigma [11,24]. Our findings expand on this finding as we found any sexual/physical violence—not necessarily related to HIV serostatus—was linked with higher HIV-related stigma over time. This suggests the importance of conceptualizing violence as rooted in structural, community, interpersonal and intrapersonal dimensions [49] that exacerbate the devaluation of WLHIV.

Finally, our finding that sexual minority WLHIV reported higher trajectories of Everyday Discrimination builds on prior cross-sectional research that shows increased racial and gender discrimination [9], and increased HIV-related stigma [17], among sexual minority WLHIV compared with heterosexual counterparts. Heteronormativity, the assumption that all persons are or should be heterosexual, results in the experiences of sexual minority WLHIV being overlooked [50–52].

Notable is the chronicity of HIV-related stigma and everyday discrimination—with little change over four years—identified among this sample of WLHIV. Earnshaw et al. describe that a stigmatized status associated with chronic illnesses such as HIV may not change over time [53]. Everyday discrimination is a chronic stressor understudied with WLHIV and we identified persistence of daily mistreatment over time [54].

Strengths and limitations

Study limitations included the small sample size, which may have limited power to detect all associations with trajectory outcomes. All data are self-reported, which may result in recall and social desirability bias. The non-random sample limits generalizability of findings. We were unable to stratify analyses by reasons/attributions identified for Everyday Discrimination, as the high number of different reasons resulted in insufficient power to include in analyses. Finally, LCGA uses fixed effects approaches that do not account for potential changes over time in the predictors. However there may be bidirectionality in associations between HIV-related stigma and social marginalization indicators such as food insecurity [55] and violence [56] that we did not capture. Future research would benefit from using a statistical procedure that would allow for time-varying covariates at each timepoint over follow-ups, to better understand associations between covariates and outcomes in trajectory groups over time [57,58].

Despite these limitations, this approach of assessing attribution-free outcomes of Everyday Discrimination aligns with research that shows that discrimination’s negative impacts on wellbeing span attributions of reasons for discrimination [59,60], and in fact, attribution-free measures may advance analysis of intersectional discrimination [28,61]. We identified multiple social factors associated with trajectories of higher Everyday Discrimination, enhancing knowledge of how discrimination among WLHIV may be exacerbated among persons who are sexual minorities, food and housing insecure, and/or using substances. Other study strengths include capturing stigma and discrimination multiple times over 4 years—providing novel insight into recent experiences linked with trajectories of stigma/discrimination as well as its chronicity. This granular assessment of stigma and discrimination every 6-months may increase the accuracy of assessing “the chronic accumulation of discrimination experiences” (p. 10) [54].

Findings have public health implications for screening WLHIV for food/housing insecurity and violence victimization in health and social service provision to provide linkages to to resources, and to investigate and address facets of HIV-related stigma that are chronic and amplified by these experiences. Service providers and researchers can also tailor services for WLHIV who use substances and sexual minority WLWH to identify factors that contribute to the persistence of discrimination and strategies for reducing chronic exposure to discrimination. Findings also suggest the utility of a syndemics approach that conceptualizes the clustering of and interaction between social and health disparities [62,63]. Multi-level strategies that address intersecting stigma and food/housing insecurity, and women-centred, trauma and violence aware clinic-based approaches [64–66], hold potential to optimize health and rights with WLHIV.