Background
Feeling stigmatized due to living with HIV has been shown to be related to poor quality of life in different cultures [
1‐
4]. Since the beginning of the HIV pandemic, experiencing stigma related to HIV has also been shown to be a barrier to treatment and prevention [
5,
6]. Several theoretical frameworks have been presented for how HIV-related stigma manifests and operates [
7‐
12]. One of the most cited frameworks, the HIV stigma framework by Earnshaw and Chaudoir [
10], aims to capture the effects of HIV-related stigma on an individual level. The HIV stigma framework [
10] covers mechanisms and outcomes of HIV-related stigma for individuals living with HIV and individuals who are (known or assumed to be) uninfected. The present study focuses on the part of the framework that addresses individuals living with HIV. According to the HIV stigma framework, the social phenomenon of HIV stigma impacts persons living with HIV through three different mechanisms: (1) Enacted stigma involving experiences of prejudice, marginalization and negative treatment by others due to one’s HIV, (2) Anticipated stigma involving expectations of enacted stigma and (3) Internalized stigma referring to when the stereotypes, labels and beliefs that constitute the stigma are endorsed and applied to oneself by the stigmatized person. The HIV stigma framework stresses the importance of differentiating between stigma mechanisms during measurement, since they may relate differently to health-related outcomes for persons living with HIV. When the framework was first described, the three stigma mechanisms were hypothesized to be related differently to psychological, behavioral and health outcomes for persons living with HIV [
10]. In Earnshaw et al.’s [
13] empirical evaluations of the HIV stigma framework for persons living with HIV, these outcomes of HIV-related stigma were specified as physical, behavioral and affective health and wellbeing. It has been suggested that enacted and anticipated stigma affects the individual’s physical health and wellbeing, since enacted and anticipated stigma can be a stressful experience, a stress considered so severe that physical health may be affected [
13‐
15]. Anticipated stigma was thought to have consequences for behavioral health and wellbeing, since the individual may, for example, avoid medical care visits and skip medicine doses for fear of disclosing their HIV-status [
5]. Internalized stigma was also proposed to influence behavioral health in the form of adherence to therapy due to feelings of not deserving treatment for their HIV infection or not deserving to feel well [
5]. Internalized stigma could result in the individual having negative feelings about him or herself, e.g. feeling “less than” others, and has been associated with affective consequences such as mental health problems, for example depression [
16]. It was therefore hypothesized that internalized stigma impacts affective health and wellbeing [
13].
Isolated parts of the framework have been further explored [
6,
17‐
22]. Anticipated stigma has been shown to be related to physical health in the form of more HIV symptoms among people living with HIV in the US, and social support has been found to buffer this relationship [
17]. An extensive review of the literature has linked HIV-related stigma to behavioral health in the form of low antiretroviral adherence [
6]. More specifically, internalized stigma has been shown to be related to lower antiretroviral adherence among people living with HIV in a US context [
18] and also to be related to affective health in the form of depression among men who have sex with men in China [
19]. Furthermore, the relationship between internalized stigma and behavioral health (low antiretroviral adherence) has been shown to be mediated by social support and depressive symptoms [
21]. Internalized stigma has also been found to be related to engagement in care (mediated by HIV disclosure) and to higher virus levels for persons living with HIV in Italy [
22]. The HIV stigma framework has also been expanded into a model where enacted and internalized stigma is related to perceived community stigma [
20]. Perceived community stigma was found to be related to lower self-esteem (affective health) and lower antiretroviral adherence (behavioral health), relationships that were mediated by internalized stigma [
20]. However, the original HIV stigma framework for persons living with HIV has, to our knowledge, only been evaluated empirically once, where empirical support for all the hypothesized relationships were found, except for the relationship between anticipated stigma and adherence [
13].
According to Deacon [
7] theories should constantly be reassessed in relation to empirical evidence. Since isolated parts of the framework have only been tested in a limited amount of studies [
6,
17‐
22] and the entire original framework has only been tested once, and then only in a US context with a high rate of persons with HIV symptoms [
13] (indicating suboptimal access to efficient treatment), we conclude that there is a lack of knowledge regarding the validity of the framework for different populations as well as in contexts with high access to contemporary antiretroviral treatment. Furthermore, since the stigma mechanisms are correlated [
13], we argue that it is of significance to test all hypothesized relationships in the framework simultaneously. The present work is therefore an attempt to contribute to the theory building regarding factors that correlate with HIV-related stigma for individuals living with HIV. Our hypothesis was that the framework is valid for persons living with HIV in contexts other than those previously tested, including a context with very high access to antiretroviral therapy. This hypothesis was tested empirically with data collected from persons living with HIV in Sweden, where antiretroviral treatment is available to all, free of charge and where all UNAIDS/WHO 90-90-90 goals are met [
23]. The aim of the present work was to test the tenets of the HIV stigma framework and its potential covariates for persons living with HIV in Sweden.
Discussion
In this article, Earnshaw and Chaudoir’s [
10] HIV stigma framework for people living with HIV has been tested using data collected from persons living with HIV in Sweden, a context where all UNAIDS/WHO 90-90-90 goals are met [
23]. The HIV stigma framework was only partly confirmed in this empirical test, indicating that the HIV stigma framework needs to be revised for contexts where a very high proportion of people living with HIV are diagnosed and under efficient treatment. As hypothesized in the model, anticipated stigma (measured by
Concerns about public attitudes) was related to lower physical health and wellbeing.
Disclosure concerns, also used as a measure of anticipated stigma, was not related to physical health and wellbeing, indicating that it is not
Disclosure concerns themselves that are related to lower physical wellbeing, but
Concerns about other people’s negative opinions about HIV. As also hypothesized, internalized stigma (measured by
Negative self-
image) was associated with lower affective health and wellbeing. Low behavioral health and wellbeing was only found to be inversely related to age and there was no significant relationship between enacted stigma (measured by
Personalized stigma) and low physical health and wellbeing. There was a lower rate of persons with CD4 counts below 200 × 10
6 cells/ml, i.e. there were more individuals whose HIV infection was in a better state, in the sample used in the present study than the sample used in the previous evaluation of the framework [
13]. The HIV stigma framework, however, proposes to capture health-related effects of stigma for all persons living with HIV [
10] and our hypothesis was therefore that the framework would also be valid for persons who are virally suppressed. This hypothesis could not be confirmed and a new hypothesis, drawn from our results, is that HIV-related stigma may have fewer or alternate outcomes for persons with good physical health and full access to antiretroviral treatment, such as the sample included in the present study. The impact that HIV-related stigma can have on the lives of persons living with HIV who have full access to antiretroviral treatment and high treatment adherence needs to be further explored in future research, which could lead to a revision of the HIV stigma framework for persons living with HIV.
The analysis of the covariates showed that women born in countries other than Sweden experienced anticipated stigma to a higher extent, and that younger age was related to higher levels of anticipated and internalized stigma. However, the effect size of these relationships was small. Earlier research shows contradictory results regarding whether the levels of HIV-related stigma differ with gender, origin and age. In a meta-analysis of health and demographic correlates to HIV-related stigma in a north-American context [
2], only two of the articles included explored differences related to ethnicity, with contradictory results. In the same meta-analysis, differences related to gender were explored in three studies, with no statistically significant difference found. However, younger age was found to be related to higher levels of HIV-related stigma [
2], in line with the results of the present study. Regarding the results of the present study, with a majority of respondents being under successful treatment, we want to emphasize that the effect of the covariates were small, indicating that persons experiencing high levels of stigma mechanisms may be found in all groups regardless of gender, origin and age.
Physical health and wellbeing (measured by
Physical functioning) had, in contrast to earlier research, no relation to enacted stigma. To experience enacted stigma related to one’s HIV has earlier been shown to be a stressful experience that via neuroendocrine and sympathetic nervous system pathways impacts physical health [
14,
15]. In the sample used in the present study, very few persons had CD4 levels under 200 × 10
6 cells/ml or viral loads > 150 copies/ml. We therefore chose
Physical functioning as a measure of physical health and wellbeing, since this variable had a larger variance, but this measure was not related to enacted stigma. Social support and community support have earlier been shown to buffer the association between anticipated stigma, stress and HIV symptoms, but not the association between enacted stigma, stress and HIV symptoms [
17]. Further research is needed to explore if such factors have functioned as mediators and buffered the relationship between enacted stigma and physical health and wellbeing in a Swedish context. We did, however, find a statically significant relationship between physical health and wellbeing and anticipated stigma, which could possibly be explained by anticipated stigma being a stressful experience. Earlier research has also shown that people living with chronic illnesses who anticipate stigma were less likely to access care [
35], which may also be a possible explanation to lower physical health and wellbeing for persons who anticipate stigma.
Behavioral health and wellbeing (as measured by VL as a marker of antiretroviral adherence) had no relation to HIV stigma mechanisms, which stands in contrast to earlier research where substantial empirical evidence has linked stigma to adherence difficulties [
6,
18,
21] and less access to care [
35], also in a Swedish context [
36]. In Sweden, all persons diagnosed with HIV are obliged by the Swedish Communicable Diseases Act to avoid lost contacts with care and each individual is linked to a specialized HIV care center with quality assured care and treatment [
23]; this may counteract negative effects of HIV stigma on behavioral health. There is also a possibility that respondents in the sample experience dimensions of stigma, e.g. layered stigma [
37] not covered by the HIV stigma scale. Similarly, the hypothesized path between anticipated stigma and behavioral health and wellbeing was not confirmed when the HIV stigma framework was tested in an American context, and the authors related this to the cross-sectional design of the study [
13]. The authors reasoned that anticipated stigma would have the strongest effect on future behavior, and therefore it would be preferable to measure this using a longitudinal design. We disagree with this reasoning and propose that anticipated stigma is something that can be a part of and affect everyday life. Therefore, a relationship between stigma and behavior should show even when investigated in a cross-sectional design, although causal relationships need to be investigated in longitudinal designs. One study has earlier shown that the relationship between HIV-related stigma and medical adherence was partially mediated by depression [
38], something that is not addressed in the HIV stigma framework. However, if this mediated relationship was accurate in a Swedish context, we would have expected a stronger correlation between
Emotional wellbeing and
Antiretroviral adherence.
Affective health and wellbeing was, in line with earlier research, related to internalized stigma. Experiencing internalized stigma was associated both with having felt more negative feelings during the past week (negative affect) and having felt fewer positive feelings during the past week (positive affect). This corresponds to recent research where positive and negative affect were shown to mediate a relation between self-stigma and depression among Chinese men who have sex with men [
19]. Future research should examine the relationship between internalized stigma, negative and positive affect, and depression in a population with high access to antiretroviral treatment.
Although the present study used a cross-sectional design and therefore does not provide information about causality among hypothesized relationships, we would like to address the question about causality in the HIV stigma framework. Earnshaw et al. [
13] hypothesize causal relationships in the HIV stigma framework, where it is implied that higher stigma causes lower wellbeing. As shown, persons who anticipate stigma (measured by
Concerns about public attitudes) to a higher extent rated lower
Physical functioning. According to Earnshaw et al. [
13], this may be explained as a causal relationship where low
Physical functioning is caused by anticipated stigma among individuals. We propose that alternate explanations can be equally valid. A person with high
Physical functioning could conceal their HIV status, which would prevent the individual from experiencing stigma [
39], and low
Emotional wellbeing could make a person more vulnerable to internalized stigma. We therefore propose a shift of focus from causal relationships to intertwined relationships between HIV-related stigma and measures of health and wellbeing.
Implications for Care
When designing care for persons living with HIV it is valuable to know if individuals with certain background characteristics risk experiencing more stigma than others or are at higher risk for certain consequences of HIV-related stigma. Even if the patterns were not strong, the results of the present study imply that persons of younger age and women born in other countries than Sweden may be more exposed to HIV-related stigma. Therefore, in addition to broadly focusing stigma-reducing interventions, special resources targeting persons of lower age and women born in other countries than Sweden may be warranted. Furthermore, since the present study, along with earlier research, shows an inverse relationship between internalized stigma and emotional wellbeing, identifying and paying special attention to persons with internalized stigma may be warranted. Both the internalized stigma and the low emotional wellbeing could be targeted within a healthcare setting with support from a wide spectrum of healthcare professions. Existing cognitive and behavioral interventions that target internalized stigma have mainly been developed for women [
40,
41], which is important, however there also seems to be a need for interventions targeting internalized stigma among men.
Methodological Considerations
We did not use variables that were identical to those used when the framework was previously tested [
13] and this may have affected the results. The areas in the HIV stigma framework that are hypothesized as being affected by HIV stigma (physical, behavioral and affective health and wellbeing) are, however, broad concepts and measures of health and wellbeing used in the present study match these concepts. In Earnshaw and Chaudoir’s [
10] review of HIV stigma mechanism measurements, the HIV stigma scale [
25] was considered to measure enacted, anticipated and internalized stigma. However, in their empirical test of the HIV stigma framework [
13], an alternative measure for anticipated stigma was used with items more explicitly phrased about anticipation of enacted stigma in the future. In the present study, measures of disclosure concerns and concerns about public attitudes about HIV were used to measure anticipated stigma. These two scales include items both about what respondents anticipate happening in the future if their HIV status becomes known and what the respondent thinks the attitudes of people in general are regarding those with HIV, which are indicators of what would happen to the respondent if his or her HIV status became known. The inclusion of sociodemographic correlates into the path model rendered a large number of tested relationships, thus limiting the number of variables that could be included. Dropping the correlates could have given the opportunity to turn physical, behavioral and affective health and wellbeing into latent variables, measured by several manifest variables, but we prioritized the inclusion of sociodemographic correlates considering the lack of research concerning how HIV-related stigma varies across different sociodemographic backgrounds. Future research could examine if HIV-related stigma varies across persons with different paths of transmission. We decided against this since the number of persons whose route of transmission was intravenous drug use was low in this sample. Furthermore, the cross-sectional design precludes assumptions of causality among hypothesized relationships. The use of PLS-SEM supports reliability of the results, since PLS-SEM is preferred over covariance based SEM (CB-SEM) when data is non-normal and the model is complex [
42]. In the model analyzed in the present study comprising 37 free parameters to estimate, a maximum of paths directed towards a construct was eight (including correlations as paths), for Personalized stigma. A minimum sample size would then preferably be ten times eight [
42], i.e. a minimum sample size of 80, which is more than doubled in the present study. A diverse sample of respondents was included in this study, which we consider a strength, since studies that include self-reported data from persons living with HIV often use samples that do not reflect the population of people living with HIV. Although the rate of eligible respondents that declined to participate was high (45%), the sample was found to reflect the population characteristics of people living with HIV in Sweden, despite an underrepresentation of persons born in countries other than Sweden [
24]. The persons attending the clinic from which the respondents were recruited live predominantly in a metropolitan area, which may have had an impact on the results. It is unknown if persons living with HIV experience stigma differently depending on where in the country they live. The prevalence of persons with problems with antiretroviral adherence is low in Sweden and in the analysis this subgroup is small. The results regarding antiretroviral nonadherence should therefore be interpreted with caution.