A systematic review of multi-level stigma interventions: state of the science and future directions

Stigma can aggravate disease processes and add numerous socioeconomic, psychosocial, and health burdens on people who hold marginalized identities or statuses, including reduced educational attainment, exposure to psychosocial stressors, and challenges in accessing healthcare [1]. Behavioral scientists have studied the severe negative consequences of stigma for individuals coping with various health conditions and have learned that stigma can deter individuals from optimally engaging in treatment for their condition, which has serious impacts on morbidity and mortality [2]. Strikingly, when disease morbidity and mortality are low but the condition is highly stigmatized, the burden of stigma may exceed the burden of the disease in its impact on social, emotional, and work functioning, thus negatively affecting the overall quality of life [3]. Researchers have long recognized that stigma operates on intrapersonal, interpersonal, organizational, and structural levels, and as such, stigma is conceptualized as an inherently multi-level phenomenon [1]. The multi-level nature of stigma renders the development of stigma interventions particularly challenging, in part because addressing multiple levels through research is more complex, requires more resources, and may be more burdensome to participants than single-level interventions. However, for research teams willing to take on the task of addressing multiple levels, the impacts on stigma reduction efforts can be farther reaching, more synergistic, and more holistic than single-level interventions [4].

Two previous papers have reviewed the literature on stigma reduction interventions. In Heijnders and Van Der Meij’s 2006 review [5], consistent with the multi-level approach to stigma [1], the authors identified five levels of examination and mapped strategies and target populations directly onto each level. First, at the intrapersonal level, the focus of interventions is on characteristics of the individuals living with a stigmatized condition, and strategies involve self-help, counseling, and treatment. Second, at the interpersonal level, the intervention is focused on the enhancement of care and support in the stigmatized persons’ local environment. Third, at the community level, the focus is on reducing stigmatizing attitudes and behaviors in (non-stigmatized) community groups using strategies such as education, contact, and advocacy. Heijnders and Van Der Meij define contact as any interactions between the public and the affected person for the purpose of reducing stigma [5]. Fourth, at the organizational/institutional level, interventions focus on reducing stigma in an organization or institution, and strategies include training programs and institutional policies. Fifth, at the governmental/structural level, interventions focus on establishing and enforcing legal, policy, and rights-based structures.

In 2014, Cook and colleagues [6] conducted a narrative review that similarly considered multiple levels in which stigma interventions can operate as part of an ecological system [7], but focused on only three levels: intrapersonal, interpersonal, and structural. Cook et al.’s definitions of these levels differed slightly from those of Heijnders and Van Der Meij’s and were more flexible, in that one strategy, such as education, could operate on multiple levels. The authors’ primary purpose was to describe how each strategy operates on multiple levels, while targeting both stigmatized and non-stigmatized populations.

Although neither review was systematic, both challenged investigators to build and evaluate multi-level stigma reduction interventions. In Heijnders and Van Der Meij’s review [5], while all of the strategies reviewed had the potential to operate on multiple levels, the authors reviewed studies that evaluated stigma reduction strategies at a single level of analysis. In their conclusion, they called for researchers to combine multiple strategies to target multiple levels. Cook and colleagues [6] conducted an updated narrative review of stigma interventions and analyzed these studies for cascading impacts across multiple levels. The authors determined that studies examining cascading effects across levels were rare, concluding that stigma reduction interventions that examine effects across levels were urgently needed. While these two prior reviews pointed out important lacuna in the literature on stigma interventions, our study addresses another knowledge gap by conducting the first systematic review of multi-level stigma interventions. We describe the country of origin of research studies, depict the design and participants of each multi-level stigma intervention, discuss the strategies and outcomes used by these interventions, and highlight opportunities for new research and program development.

We conducted this review in accordance with the Preferred Reporting Items for Systematic Reviews (PRISMA) guidelines [8]. We included studies that focused on stigma reduction interventions operating on multiple levels, both within and outside of the USA. The overall purpose of our review was descriptive, rather than evaluative. Thus, for each study, we provide a basic indication of effectiveness in reducing stigma, but a detailed evaluation of study effectiveness was beyond the scope of this review.

We used Heijnders and Van Der Meij’s categories for the levels of the ecological system (i.e., intrapersonal, interpersonal, community, organizational/institutional, governmental/structural) [5]. However, we expanded our organization of these predefined strategies (e.g., education, contact) such that they could map onto multiple levels, which Heijnders and Van Der Meij did not do in their analysis. For example, if one target of an intervention was to improve attitudes held, whether by the stigmatized or the non-stigmatized, we categorized this focus at the intrapersonal level. If an intervention’s target was to improve interactions between people with stigmatized conditions and other stakeholders (e.g., caregivers, healthcare workers), we categorized this focus at the interpersonal level. If the (non-stigmatized) public was targeted, we identified the community level as the focus. If an organization was targeted, we identified the organizational/institutional level as the focus. If a policy or administrative structure was targeted, we identified governmental/structural level as the focus.

Overall, six of the 24 studies were randomized controlled trials (two used individual randomization [11, 12] and four used cluster randomization [13‐16]) (Table 1). Eighteen studies did not randomize or use a control group and thus were not considered randomized controlled trials (RCTs). Most studies reported on pilot trials of interventions. Of the studies reviewed that did use RCT designs, four used cluster randomization. All studies used convenience samples rather than population-based sampling.

We found approximately equal numbers of studies originating from low- and middle-income country (LMIC) and high-income country (HIC) settings, with 13 studies conducted in HIC and 11 studies conducted in LMIC. Five studies were based in the US, three in the UK, two in Canada, two in Indonesia, two in South Africa, and one study spanned five African countries (Lesotho, Malawi, South Africa, Swaziland, and Tanzania). One study was conducted in each of the following countries: Kenya, Zambia, China, India, Vietnam, Israel, Haiti, Australia, and Japan.

Twelve articles examined stigma associated with mental illness, six HIV, two leprosy, one Moebius syndrome, and one each of epilepsy, orphans and vulnerable children, and substance use. Eighteen articles described studies targeting stigmatized participants, 12 included participants who were community members (e.g., students, police), six articles included healthcare workers as participants, eight articles addressed stigma among caregivers/family members, and two articles examined stigma among youth at risk for HIV. Of the articles targeting stigmatized populations, six studies targeted both stigmatized and community populations, eight studies targeted both stigmatized and caregiver populations, and six studies targeted both stigmatized and healthcare worker populations.

Five of the six articles examining HIV-related stigma originated from LMICs. Conversely, articles examining mental illness-related stigma predominantly came from a HIC (e.g., UK, US, Canada), with only one out of 12 articles from a LMIC (India). Five studies were published prior to 2010, whereas 19 were published between 2010 and 2017. Five of the six studies of HIV-related stigma were published after 2010, and nine of 12 studies of mental health-related stigma were published in 2010 or after.

The most common levels examined together were intrapersonal and interpersonal, with 15 articles including these two levels together. Of the 24 articles, four examined the community level alongside both the intrapersonal and interpersonal levels. Three articles examined intrapersonal and community levels together, and another three examined interpersonal and community levels together. Thus, a total of 10 articles examined community levels alongside either interpersonal, intrapersonal, or both levels. Only one article examined the institutional level (along with intrapersonal), and no articles targeted the structural level. Figure 2 depicts these findings, separated by LMIC and HIC study location.

The most common stigma reduction strategy studied was education, with 16 studies using this strategy. Ten studies examined contact, five counseling or coping skills acquisition, three social support, three drama, and two problem solving. Individual studies also examined communication skills, voluntary counseling and testing, psychiatric treatment, and outdoor adventure as stigma reduction techniques. Of the 12 articles that examined the community level alongside at least one other level, six used contact as a primary strategy. Eight studies used education and contact strategies together, and six of these eight studies originated from HIC.

Seven of the 24 studies examined one stigma reduction strategy across more than one level. For example, Patalay and colleagues [17] trained university medical students in the UK to lead workshops with secondary school students on mental health conditions and services (educational strategy). Investigators measured medical students’ levels of stigma and impact on the interpersonal level to examine the workshop leadership’s potential impact on medical students’ future practice behaviors. The researchers also measured attitudes towards mental illness of the recipients of the intervention, (the secondary school students), thereby using the same educational strategy and then assessing community-level stigma. Of note, one study used an educational strategy with specialized areas of content: Brown and colleagues [18] had nursing students deliver a stigma reduction program in the US to community members by providing information on mental health issues as well as on a faith-based framework for dealing with mental health issues. In other words, this intervention provided education on a condition as well as education on using faith to cope with the condition (two educational strategies), without using faith-based counseling techniques as the intervention itself.

The majority (16 of 24) of studies reviewed used stigma scales that were validated and used in multiple countries. The remaining eight studies used scales that were adapted from validated scales, used in a neighboring country, or validated by the study team for use in the country where the study had taken place. Beyond use of validated and adapted measures, the articles provided little information on how well the instruments performed across cultures and contexts.

In terms of effectiveness, 17 studies reported that their intervention reduced stigma scores (p < 0.05) on at least one measure of stigma and seven studies reported non-significant results. Of these seven studies that found non-significance, five were conducted in high-income countries and two were conducted in middle-income countries (China and South Africa). Only two of the 24 articles provided information on confidence intervals. In terms of practical significance, only 11 out of the 24 studies provided information to calculate effect sizes or the effect sizes themselves. Cohen’s d values that were reported ranged from 0.4 to 2.51, Eta squared and R-squared values ranged from 0.02 to 0.32, indicating small-to-moderate effects across studies.

We set out to review intervention studies that targeted multiple levels of stigma reduction and identified 24 studies. Notably, the majority of studies identified and reviewed were published after 2010, demonstrating an increasing urgency and movement in the research community towards developing and validating stigma reduction interventions. Articles that originated from HICs tended to examine mental illness-related stigma, whereas those from LMICs tended toward the examination of HIV-related stigma. This may be due to availability of funds, as global health spending in LMICs has decreased over time except for HIV-related work [19].

Most investigators used validated or adapted measures of stigma in their studies, but provided little information on how well the measures performed in diverse settings. Contextual psychometric information and sensitivity/specificity of measures are useful pieces of information to determine accurate interpretation of intervention effectiveness. This is particularly relevant for studies that used adapted measures or measures validated in languages or contexts that differed from where the studies were conducted. More detailed examination of measures used to evaluate intervention effectiveness will be an important direction for future research on multi-level interventions.

Similarly, we found relatively few studies that used randomized controlled trial (RCT) designs. The lack of RCT designs may be due to the challenges of conducting RCTs across multiple levels. Investigators in future studies of multi-level stigma interventions may consider use of non-traditional hybrid trial designs, quasi-experimental designs, or other types of pragmatic designs used in complex real-world settings. Similarly, we also noted that just under half of the reviewed articles provided effect sizes, and those that were reported varied widely in magnitude. Adding rigor to these designs may help to narrow information on the potential benefits of interventions that operate on multiple levels.

The intrapersonal and interpersonal levels were most often targeted by the multi-level stigma interventions studied, which may be due to several factors. The broader stigma literature has focused almost exclusively on these two levels of analysis [1]; thus, multi-level interventions have a larger evidence base from which to draw at these levels. Relatedly, research has accumulated a wealth of stigma measures at the individual/interpersonal levels of analysis. In contrast, until recently, fewer measures of stigma existed at community, organizational, and structural levels, which likely hindered the assessment of multi-level stigma interventions that incorporated communities and social structures [20]. This focus on the individual/interpersonal levels in multi-level stigma interventions may also be due to convenience—intervention studies are often easier to implement in clinical settings where people with health-related stigmatizing conditions seek care and where their family members (who are needed for research at the interpersonal level) are more easily identified and assessed. More research is needed to incorporate community-, organizational-, and structural-level influences into multi-level stigma interventions.

Approximately half the studies reviewed examined community-level stigma reduction, with intrapersonal and/or interpersonal levels. Studies that targeted community levels of stigma predominantly used methods of interaction, or contact, across populations studied, examining the impact of exchanging information and making use of bi-directional learning and including people living with stigmatized conditions in the process (e.g., teaching, drama). In addition, these studies tended to incorporate exchanges of support, particularly when family members and health care workers were involved.

Despite accumulating research indicating that structural forms of stigma contribute to adverse health outcomes among members of stigmatized groups [21, 22], only one study combined an institutional-level approach, and no studies combined the structural-level approach, alongside another level. Researchers may consider institutional- and structural-level interventions challenging, since they require time and financial resources to examine stigma in large samples. Despite these challenges, single-level studies are beginning to emerge that examine stigma reduction as a result of policy changes at the structural level [23]. Thus, one important direction for future development of multi-level interventions is greater attention to, and incorporation of, policy-level interventions to address stigma at the institutional and structural levels.

With respect to stigma-reduction strategies used by these multi-level stigma interventions, most focused on education, either alone or in combination with other strategies, such as contact. Corrigan and colleagues found over years of research that stand-alone educational programs can lead to stereotype suppression, in which members of the public suppress—rather than reject—stereotyped beliefs upon learning that such beliefs are socially undesirable [24, 25]. Thus, educational programs alone are often ineffective in reducing stigmatizing attitudes in members of the public, and the little resulting stigma reduction that occurs may be short-lived and superficial [26]. Future research on multi-level stigma interventions is therefore needed to explore a wider range of stigma-reduction strategies and to utilize evidence-based strategies that prior research has shown to be effective in reducing stigma.

This review has several limitations. First, although we introduced independent secondary article reviewers and coders, our process of article selection, non-inclusion of gray literature, inclusion of studies reported in English only, and content analysis may have introduced selection biases that restrict the generalizability of our findings to all multi-level stigma interventions. Second, the scope of our study did not include detailed comments regarding a methodological appraisal of studies and we included limited information on intervention effectiveness. The lack of rigor in these studies may have led to sampling bias and non-generalizable conclusions. Thus, additional research will need to be done before recommendations on effectiveness can be made.

Stigma is inherently a cross-sectoral phenomenon [1] and thus efforts to reduce stigma and its pernicious effects require a multi-level approach. Despite progress over the past decade in the development of multi-level stigma interventions, much work remains to strengthen and broaden this approach. In Table 2, we highlight several opportunities for new research and program development in multi-level stigma interventions, organized around several key domains (e.g., measurement, mechanisms of change, implementation). This list is not exhaustive, but rather is meant to underscore some of the most important areas of inquiry that are needed to advance the knowledge base in this incipient field. For instance, multi-level stigma interventions may not always be appropriate; future research is therefore needed to systematically compare the efficacy of single-level vs. multi-level stigma interventions in order to determine the conditions under which multi-level stigma interventions may be preferable to single-level interventions. Future research is also needed to evaluate how changes at one level of stigma (e.g., intrapersonal) impact other levels of stigma (e.g., community) in order to guide the development of more effective multi-level interventions, to identify mechanisms of change in multi-level stigma interventions, and to explore the barriers and facilitators to the dissemination of multi-level stigma interventions across diverse contexts. Only after answering these questions will it be possible to fully evaluate whether multi-level stigma interventions are effective in addressing the predicament of stigma in the lives of the stigmatized.