Well-Being Among Persons at Risk of Psychosis: The Role of Self-Labeling, Shame, and Stigma Stress

Early intervention programs for schizophrenia and bipolar disorders have gained traction in recent years. The aim of these programs is to improve clinical and social outcomes of these diseases by intervening while the individual is at risk of or in the initial stages of the illness (1). Young people at risk of psychosis may be labeled as having a mental illness because of early signs of the disorder or as an unintended consequence of early intervention (2–4). Qualitative studies found that stigma is a common concern of individuals experiencing early psychosis (5). However, quantitative data on labeling and stigma among young people at risk of psychosis appear to be lacking (6).

Mental illness stigma mainly comes in two forms. Public stigma refers to endorsement of negative stereotypes of individuals with mental illness by members of the general public (7,8). Self-stigma occurs when persons with mental illness agree with and internalize negative stereotypes (9) and is typically associated with shame about having a mental illness (10). Two models help explain how public stigma and self-stigma may affect young people at risk. First, according to Link and others’ (11) modified labeling theory, individuals are aware of societal negative attitudes toward persons with mental illness (perceived public stigma). Once a person has been labeled as having a mental illness, these public attitudes become self-relevant, potentially threatening and negatively affecting feelings of well-being (12,13). Second, stress-coping models conceptualize stigma as a stressor for people with mental illness (14–17). Stigma stress occurs when stigmatized individuals perceive that the harm due to stigma (primary appraisal) exceeds their resources to cope with this threat (secondary appraisal), and stigma stress can undermine well-being (Figure 1).

The overall aims of this study were twofold—first, to collect data on the relationship between self-labeling, stigma, and well-being among people at risk, and second, to examine how the underlying mechanisms of the stress-coping model may explain how stigma affects this group. Both questions have policy and research implications. For instance, if stigma is unrelated to well-being, it would need less future attention. If the two are related, a better understanding of the mechanisms involved can inform interventions to reduce public stigma and self-stigma associated with at-risk status.

We tested the following hypotheses among young people at risk of psychosis or bipolar disorder who did not yet fulfill diagnostic criteria of either disorder (Figure 1). First, we expected that higher levels of perceived public stigma, shame about one’s mental illness (a proxy for self-stigma), and self-labeling as having a mental illness would predict more stigma stress. Second, we anticipated that increased stigma stress would predict reduced well-being after the analyses controlled for symptoms, comorbid psychiatric illnesses, and sociodemographic variables. Third, we expected that stigma stress would mediate the effect of public stigma, shame, and self-labeling on well-being.

Methods

Participants

Participants were recruited from the Zürich region of Switzerland in the context of a larger study on early recognition of psychosis (www.zinep.ch); data on personal experience of stigma and attitudes toward help seeking among this group are reported elsewhere (18). Participants learned about the study from a study Web site, flyers, and newspaper ads or were referred to our staff by general practitioners, school psychologists, counseling services, psychiatrists, or psychologists. After participants were provided a complete description of the study, written informed consent was obtained. In cases in which the participants were minors, the written informed consent of their parents was also obtained. The study was approved by the regional ethics committee of Zürich. This study is based on the larger study’s baseline data.

Data were available from 172 German-speaking persons between 13 and 35 years of age (mean±SD=21.37±5.8) of whom 70 (41%) were female. Participants were required to fulfill at least one of the following three inclusion criteria: high risk of psychosis, ultra-high risk of psychosis, or risk of bipolar disorder. High risk of psychosis was assessed by the adult (19) or child and youth (20) version of the Schizophrenia Proneness Interview and was indicated by having at least one cognitive-perceptive basic symptom or at least two cognitive disturbances. Ultra-high risk of psychosis was rated by the Structured Interview for Prodromal Syndromes (21) and was indicated by having at least one attenuated psychotic symptom or brief, limited intermittent psychotic symptom or by meeting state-trait criteria (>30% reduction in global assessment of functioning in the past year plus either schizotypal personality disorder or a first-degree relative with psychosis). Risk of bipolar disorder was defined by a score of ≥14 on the Hypomania Checklist, a self-report measure of lifetime hypomanic symptoms (22).

Among the 172 participants, 138 (80%) fulfilled high-risk criteria for psychosis, 85 (49%) fulfilled ultra-high-risk criteria for psychosis, and 135 (79%) fulfilled risk criteria for bipolar disorder. A total of 150 (87%) participants met criteria for either high or ultra-high risk of psychosis, and 73 (42%) fulfilled criteria for both. Twenty-two (13%) participants fulfilled risk criteria only for bipolar disorder. Because the participants who were at risk only for bipolar disorder did not differ significantly from participants who met criteria for high or ultra-high risk of psychosis on any variable of interest (predictors of stigma stress, stigma stress, or well-being; all p values >.25), further analyses were based on the entire sample.

Exclusion criteria for study participation were manifest schizophrenic, substance-induced, or organic psychosis, bipolar disorder, or current substance or alcohol use dependence; age below 13 or above 35 years; or low intellectual abilities (IQ <80). Current axis I comorbidity, assessed by the Mini International Neuropsychiatric Interview (23), was common: 95 (55%) had a major depressive disorder diagnosis; 87 (42%), an anxiety disorder diagnosis; 55 (32%), both a depressive and an anxiety disorder diagnosis; and 45 (26%), neither diagnosis.

Measures

Predictors of stigma stress.

Perceived public stigma was measured by Link’s (24) 12-item Perceived Devaluation-Discrimination Questionnaire. Shame about one’s mental illness was assessed by the item “I would feel ashamed to have a mental illness.” This phrasing renders the item suitable for all participants, whether or not they considered themselves as having a mental illness. Labeling oneself as “mentally ill” (self-labeling) was measured by asking participants how they perceive their mental health. Possible responses range from 1, “I am perfectly mentally healthy,” to 9, “I am severely mentally ill.”

Stigma stress.

The cognitive appraisal of mental illness stigma as a stressor was assessed by a previously validated 8-item measure (17,25,26) that is based on Lazarus and Folkman's (14) conceptualization of stress appraisal processes. Four items relate to the primary appraisal of stigma as harmful, and four relate to the secondary appraisal of perceived coping resources. As in previous studies (14,17,25–27), a single stress appraisal score was computed by subtracting perceived coping resources from perceived harm, with higher scores equaling more stigma stress.

Well-being and psychopathology.

Subjective quality of life was assessed by the 12 subjective items from the Manchester Short Assessment of Quality of Life (28). General self-esteem was measured by Rosenberg’s (29) ten-item self-esteem scale. General self-efficacy was examined by using Schwarzer and Jerusalem’s (30) ten-item self-efficacy scale. Because of a strong positive association between scores for quality of life and self-esteem (r=.60, p<.001) and self-efficacy (r=.51, p<.001) and between scores for self-esteem and self-efficacy (r=.69, p<.001), for more parsimonious analyses, we averaged these three z-standardized scores into one well-being mean score. Positive and negative symptoms were assessed by the Positive and Negative Syndrome Scale (31).

Statistical analyses

Bivariate associations were examined by Pearson correlations. Magnitudes of dependent correlations between predictors of stigma stress on the one hand and the primary versus the secondary stress appraisal on the other hand were compared by Williams’ tests (32). Multiple linear regressions were used to assess whether perceived public stigma, shame, and self-labeling significantly and independently predicted stigma stress and whether stigma stress predicted well-being. Age, gender, symptoms, and comorbid psychiatric disorders were controlled for in a second step of each regression.

For regressions on well-being, we used an incremental F test to test whether the variance explained by the unconstrained model was significantly larger than the variance (R²) explained by the constrained model. In the unconstrained model, the primary and secondary appraisals were both independent variables and, therefore, were not constrained to be equal. In the constrained model, the stigma stress difference score was one independent variable, and the primary and secondary appraisals were, therefore, forced to have equal regression coefficients (33,34). A significant difference would suggest that both the primary and the secondary appraisals should be entered as separate independent variables.

We examined the mediation effects of stigma stress on the association between the three independent variables (predictors of stigma stress) and the dependent variable (well-being). Mediation occurs when the following four conditions are met (35,36): the independent variable significantly affects the mediator; the independent variable significantly affects the dependent variable in the absence of the mediator; the mediator has a significant, unique effect on the dependent variable; and the effect of the independent variable on the dependent variable is reduced after the mediator is added to the model. Partial mediation occurs if the impact of the independent variable on the dependent variable decreases but remains significant. We used the Sobel test to examine the significance of mediation effects (35–37). Except for the Sobel and Williams’ tests, analyses were conducted by using SPSS, version 20. Findings were considered significant at p<.05.

Results

Correlations between predictors of stigma stress and stress appraisals

Perceived public stigma was positively correlated with shame about one’s mental illness (r=.25, p<.001) and self-labeling (r=.33, p<.001), whereas shame was unrelated to self-labeling (r=.03, p=.69). Because the measure of stigma stress results from a difference score, we examined how the three predictor variables were associated both with stigma stress and with the primary and secondary appraisals (Table 1). Perceived public stigma was significantly more strongly associated with the primary appraisal (perceived harm) than with the secondary appraisal (perceived coping resources), whereas shame and self-labeling were similarly related to both.

Predictors of stigma stress

We ran multiple linear regressions to examine whether perceived public stigma, shame, and self-labeling independently predicted stigma stress (Table 2; Figure 1: paths a₁, a₂, and a₃). In a first step, all three variables significantly predicted increased levels of stigma stress, explaining nearly one-third of the variance in stigma stress. After the analyses controlled for symptoms, comorbid psychiatric disorders, age, and gender in a second step, perceived public stigma and shame remained predictive of stigma stress.

Stigma stress as a predictor of well-being

Mean scores for well-being were calculated by averaging z-standardized scores for three measures, including the Manchester Short Assessment of Quality of Life (28) (possible scores range from 1 to 7, with higher scores indicating better quality of life; mean±SD=4.1±1.1; Cronbach’s α=.82); Rosenberg’s (29) 10-item self-esteem scale (possible scores range from 0 to 3, with higher scores indicating greater self-esteem; 1.6±.7; Cronbach’s α=.92); and Schwarzer and Jerusalem’s (30) 10-item self-efficacy scale (possible scores range from 1 to 4, with higher scores indicating increased self-efficacy; 2.5±.6, Cronbach’s α=.91).

In a second set of regressions, we examined predictors of reduced well-being (Table 3; Figure 1, path b). Stigma stress significantly predicted poorer well-being, explaining a fifth of the variance in well-being. Stigma stress remained significantly associated with reduced well-being after the analyses controlled for symptoms, comorbid psychiatric disorders, age, and gender.

When both stress appraisal scores were included as independent variables, R² increased minimally (from .197 to .206). Because the incremental F test of equality constraints was nonsignificant, we used the stigma stress difference score as the predictor variable.

Stigma stress as a mediator

We tested whether stigma stress met the four criteria of a mediator. We found, first, positive associations between each independent variable and stigma stress (Figure 1: paths a₁, a₂, and a₃; Table 1), and, second, between stigma stress and well-being (Figure 1, path b; Table 3). Third, the independent variables significantly predicted well-being (Table 4; Figure 1: paths c₁, c₂, and c₃). Finally, as illustrated by Table 4, the standardized regression coefficients of the independent variables decreased in magnitude from the first regression of each pair to the second. We found partial mediation for perceived public stigma and self-labeling and full mediation for shame. The Sobel tests yielded significant results for all three mediation models (Table 4).

Discussion

Our findings suggest that stigma-related stress is associated with impaired well-being among young people at risk of psychosis, independent of symptom levels, comorbid psychiatric disorders, age, and gender. Regarding the overarching aims of this study, our results have two potential implications. First, labeling and stigma should be taken into account by early intervention and prevention programs. Second, a stress-coping model may be useful to understand some of the mechanisms involved in the damaging effects of stigma. That said, our results should be replicated, and longitudinal studies are needed to establish causality between, for example, stigma stress appraisals and reduced well-being among people at risk.

The results support our first hypothesis that public stigma, shame (a proxy of self-stigma), and self-labeling as “mentally ill” independently contribute to the appraisal of stigma as a stressor that exceeds one’s coping resources. Therefore, interventions that aim to alleviate stigma’s impact in this population should address both public stigma and self-stigma. It is consistent with Lazarus and Folkman's (14) stress-coping model that perceived public stigma as an external threat was associated more strongly with perceiving stigma as harmful than with one’s coping resources. The negative effects of shame are probably greater than suggested by our findings because we were not able to assess other negative consequences—besides stigma stress—that are typically associated with shame, such as secrecy, social withdrawal, or other dysfunctional coping styles (10).

Self-labeling contributed to stigma stress, which highlights how labels for mental illness function as a two-edged sword. On the one hand, they might have positive effects, such as facilitating treatment (18); on the other hand and in accordance with Link and colleagues' (11) modified labeling theory outlined above, they increase the vulnerability of individuals to stigma and discrimination (38). Furthermore, by labeling oneself as “mentally ill,” one comes close to identifying with the group of people with mental illness. Consistent with the findings of this study, research has shown that group identification has mixed effects on stigmatized individuals, depending on the social environment and on how positively persons value their own group (15,39).

The link between perceived public stigma, shame, and self-labeling and reduced well-being was mediated in part by stigma stress, supporting our third hypothesis. Consistent with stress-coping models of stigma (15,26), this finding highlights the extent to which the cognitive appraisal of stigma as a stressor is a key determinant of reactions to stigma. Stigmatized individuals are not passive recipients of stigma, and their response can augment or reduce stigma’s negative impact. This observation, of course, is not meant to blame people with mental illness for experiencing stigma, which remains a societal injustice. But as long as stigma persists, positive coping processes could help stigmatized individuals and be encouraged in different settings, for example, by improving disclosure strategies or reducing shame associated with mental illness (40–42).

This study had several limitations. The role of family members, caregivers, peers, and health care professionals was not examined. Future studies should compare the impact of stigma variables among persons at risk of psychosis, persons with established early psychosis, and persons with a long history of psychotic disorders; that said, stigma stress levels in this study were slightly higher than among individuals from the same region with a long history of mental illness and recent involuntary psychiatric hospitalization (27). Because we recruited a convenience sample in a Swiss metropolitan area, our findings cannot be generalized to other settings. Psychiatric comorbidity was common, as is typical among samples of individuals with early psychosis (43); although we assessed self-labeling, the participant’s history of having been labeled—for example, due to at-risk status, psychiatric comorbidity, or previous contact with health services—was not assessed. However, controlling for psychiatric comorbidity did not affect our findings. Finally, except for symptom ratings, the data were based on self-report.

Conclusions

Early recognition and intervention programs for people at risk of psychosis will become more common, and recent findings highlight the potential benefits of such treatments (1,44). However, our results suggest that treatment alone may not be sufficient to overcome the deleterious effects of stigma. Interventions to help young people at risk of psychosis respond to stigma, shame, and social exclusion are needed as well. This effort is especially relevant because social networks appear to be impaired even before the first onset of psychosis (45).

Researchers, policy makers, clinicians, family members, and (potential) service users should critically discuss the costs of early intervention in terms of labeling and stigma: should they avoid using the “at risk” label, not only in psychiatric classifications (46) but also in clinical settings? Are labeling and stigma an acceptable price to pay for early intervention? The answer does not become easier when one considers that according to current estimates, a majority of individuals deemed to be at risk will not develop a psychosis (1). We need thorough longitudinal studies about the effects of labels and stigma in different settings on different groups of people at risk, including those who do not eventually develop psychosis. It seems likely that in order to minimize stigma’s negative impact, we will need programs that address both public stigma—for example, among key groups such as teachers and adolescents—and self-stigma, as well as early intervention services that minimize shame and stigma stress (47). Only a comprehensive approach, taking into account emerging symptoms as well as labeling, shame, and stigma, is likely to address the needs of young people at risk and to increase their chances to achieve full recovery and social inclusion.