Measuring recurring stigma in the lives of individuals with mental illness

Abstract

We present an exploratory factor analysis of the 8-item Daily Indignities Of Mental Illness (DIMI) scale, created to measure the detection and perceptions of recurring stigma among individuals with recent psychiatric hospitalizations. Structured in-person interviews were conducted with individuals with recent psychiatric hospitalizations in metropolitan New York. The 8-item DIMI scale's internal consistency for the sample (n=65), measured by Cronbach's alpha, was 0.869. Statistically significantly higher DIMI scale scores were observed among individuals with more than 2 psychotic episodes and those reporting seeing relatives less often after hospitalization. The DIMI scale possesses good internal consistency for research contextualizing perceptions around the occurrence or recurrence of mental illness-related stigma among individuals with recent psychiatric hospitalizations.

INTRODUCTION

Stigma, which has been defined as the co-occurrence of labeling, stereotyping, separation, status loss, and discrimination (Link & Phelan, 2001), is widely recognized as a barrier to success in treatment and stabilization of symptoms among individuals with mental illness (Link et al., 1997; Barney et al., 2006; Link & Phelan, 2006). Public stigma of mental illness, or mental illness-related stigma expressed or enacted by individuals in the general population towards individuals with mental illness, may catalyze general antagonistic attitudes and acts of discrimination in various spheres of the recipient’s life, including housing, education and employment (Rüsch, Angermeyer & Corrigan, 2005; Thornicroft et al., 2009). Downstream effects of self-stigma among individuals with mental illness include reduced self-esteem, lowered self-efficacy, avoidance of social interactions, and increased isolation (Markowitz, 1998; Perlick et al., 2001; Wahl, 2012). Furthermore, self-stigma among individuals with mental illness has been demonstrated to curb this population’s adherence to medical treatment and lessen their overall engagement with the healthcare system (Sirey et al, 2001; Barney et al, 2006) thus heightening their likelihood of premature morbidity and mortality.

There are methodological complexities to measuring the phenomenon of stigmatizing beliefs/behaviors from the general public toward marginalized groups, in part because of the reluctance which perpetrators may have in understanding or discoursing openly about their actions (Manderscheid et al., 2010; Parcesepe & Cabassa, 2013). Cross-sectional data from the Centers for Disease Control (CDC) has demonstrated that adults with mental illness symptoms (e.g., serious psychological distress), as compared to adults without such symptoms, are more likely to strongly disagree with the statement that people are generally caring and sympathetic to individuals with mental illness (Manderscheid et al, 2010).

The micro-aggression paradigm—first popularized by Pierce and Colleagues (Pierce et al., 1977)—provides a useful template for contextualizing the conduit through which public stigma and self-stigma are manifested. Micro-aggressions represent the nexus of recurrent, daily indignities, both overt and non-overt, that are directed at historically marginalized populations—such as racial or ethnic minorities, women, individuals who are LGBT, and other groups—which insult, degrade or otherwise demoralize their recipients (Pierce et al., 1977). These recurrent (daily) indignities may be expressed verbally or projected through the offender’s body language, and are particularly worrisome due to their ubiquity in the lives of their recipients and their attendant capacity to cause and sustain chronic psychological disturbance over time (Solorzano, Ceja & Yosso, 2000).

Research on the genesis and recurrence of stigmatizing experiences among individuals with mental illness, particularly those with recent psychiatric hospitalizations—a population that may be at the height of risk for public stigma and self-stigma (Kaufmann, 1995; Gonzalez, et al. 2007)—would enable a more robust understanding of perceptions of antecedents to ongoing stigma and micro-aggression-related experiences in this unique group. Existing research, as it applies to interweaving experiences of stigma and micro-aggression among individuals with mental illness, is limited and largely not descriptive in nature, thus making further general scholarship in this area valuable.

We present analysis of a scale developed to assess detection and perception processes of mental illness-related public stigma and the recurrence of such stigma, using data from interviews conducted with a sample of predominantly African American and Hispanic patients with recent psychiatric hospitalizations. Exploratory factor analysis was utilized to explore the dimensionality of the scale. We also assessed relationships between measures of perceived recurring stigma, participant race/ethnicity and socioeconomic status, and participants’ responses to a broad swath of questions assessing stigma-related anger, frustration, and shame experienced in intra and inter-personal relationships, ostensibly due to their mental health statuses.

METHODS

Study Protocol

Explained in greater detail elsewhere (Link & Phelan, 2014), recruitment for the study was carried out between 2007 and 2009 at three psychiatric hospitals in New York City and at a fourth psychiatric hospital in New Jersey. Recruitment was restricted to patients who had a primary diagnosis of one of the following conditions: (1) schizophrenia, (2) schizoaffective disorder, (3) schizphreniform, and (4) delusional or (5) psychotic disorder not otherwise specified. Additional eligibility criteria included patients having experienced fewer than six prior hospitalizations in their lifetime; this caveat was introduced to ensure that participants’ experiences of stigma were not unduly influenced by the severity of the illness (Link & Phelan, 2006; Cohen & Cohen, 1984). Identification of eligible patients was achieved through the analysis of hospital medical records and recommendations from staff. All study procedures were authorized by the appropriate Institutional Review Boards at Columbia University and the participating psychiatric hospitals. Informed consent was acquired from all individuals participating in the study.

Development of the Daily Indignities Of Mental Illness (DIMI) scale

The overarching purpose of the research project and main clinical interview was to develop a sense of the participants’ perspectives and potential experiences around stigma related to [their] mental illness. Winnowing this focus, the DIMI scale was curated to assess participants’ perceptions of any recurring instances where they were stigmatized, either directly or indirectly, due to their possessing a mental illness, with the goal of characterizing participants’ interpretations of the offender’s general disposition and the offender’s intent upon delivery of the cue(s). Interviews, administered by a trained research team member, covered a range of topics including participants’ self-reported demographic traits, and addressed participants’ sense of agency and multitude of experiences, perceptions, and feelings related to public stigma, self-stigma, self-esteem, and quality of life. Items in the DIMI scale were selected from an existing 53-item scale in the main clinical interview manual called “Stigma in The Past 3 Months,” considering experiences that persons with mental illness may be likely to experience in everyday interactions with individuals in the general population.

Each of the eight DIMI scale items had a possible five-point score range, from 0 to 4, corresponding with the five following Likert scale responses surmising general recurrence of the exposure: Never (0), Almost Never (1), Sometimes (2), Fairly Often (3), and Very Often (4).

Individual questions for the DIMI scale were selected based on their potential to measure recurring mental illness-related stigmatizing processes and related constructs – namely, the disconcerting feelings that an individual with a history of psychiatric hospitalizations might experience in daily interactions with the general public that they deem attributable to their having a mental illness.¹ These individual questions broached themes such as the public’s expectations, doubts, and fears of individuals possessing a mental illness diagnosis (Markowitz, 1998; Solorzano, Ceja & Yosso, 2000; Link & Phelan, 2001), examining how the confluence of interactions influences the slighted individual’s feelings of self-worth, thereby potentially impacting self-esteem, self-efficacy, and overall well-being.

Analysis

Initial psychometric analyses examined whether the 8 items making up the DIMI scale exhibited unidimensionality to justify the creation of a scale score. Eigenvalues of the polychoric correlation matrix of the 8 items were ascertained and factor analyses were conducted. Unidimensionality was supported when only one eigenvalue was greater than one and a factor analysis model with one factor showed adequate model fit by the Comparative Fit Index and the Tucker-Lewis Index ≥0.90 (Hu & Bentler, 1999). In case two eigenvalues were greater than one, an exploratory factor analysis model with 2 factors using geomin rotation was performed to determine whether the potential additional factor provided a coherent and interpretable construct. For interpretation of the factors, loadings >0.40 indicated that the item and factor were related. Exploratory factor analysis, rather than confirmatory factor analysis, was used since the structure of the proposed criterion set was unknown, and we sought to allow for the possibility that there might be multiple dimensions (i.e., constructs) underlying the set of items (Bandalos et al, 2008).

After assessing unidimensionality, Cronbach’s alpha was used to indicate internal consistency (Cronbach, 1951), and a composite DIMI score was computed using a simple mean of the 8 items. Descriptive mean and standard deviation summaries and tests for departures from normality were performed on the composite DIMI score. Associations between the DIMI score and selected demographic variables (e.g., age, race, marital status, education, household income), clinical indicators (e.g., psychiatric diagnosis, alcohol and drug use, number of psychotic episodes, number of psychiatric hospitalizations), and psychological variables (e.g., Quality of Life Assessment, Self-Esteem scale) were examined using Wilcoxon and Kruskal-Wallis test for dichotomous and categorical variables, and Pearson’s correlations for continuous variables. Responses to survey items related to stigmatization were also examined. Eigenvalue and factor analyses were performed using MPlus 7.1 (www.statmodel.com); all other analyses were conducted in SAS V9.2 (Release 2, Cary, NC, USA). An alpha of less than 0.05 was considered statistically significant.

RESULTS

A total of 65 individuals participated in the study and had interview response data available to validate the DIMI scale. Briefly, the mean age (+ SD) of participants was 31.4 years (+ 8.2 years). The majority of participants were male (n=46, 73.0%), and identified as either African American (n=30, 47.6%) or Hispanic (n=17, 27.0%). In addition, most participants were never married (n=55/63, 87.3%) and had an annual household income of under $35,000 (n=54/61, 91.5%). The most common DSM Axis I diagnoses among participants was psychosis not otherwise specified (n=25 or 40.0%), followed by schizophrenia – undifferentiated (n=16, 25.4%), schizophrenia – paranoid type (n=8, 12.7%), and schizoaffective disorder (n=6, 9.5%).

Psychometric analysis of the DIMI scale

The first three eigenvalues of the 8-item polychoric correlation matrix were 5.2, 1.0, and 0.6, indicating that 65% of the variability was explained by just one dimension, and the fit indices for the one factor model, CFI = 0.96 and TLI = 0.95 supported a 1-factor model, as did factor loadings ranging from 0.70 to 0.90. Because the second eigenvalue was right at the cut-off of 1.0, we explored the possibility of two factors, but did not find any interpretable structure with most items cross-loading on both factors. Hence, it was concluded that the 8 items exhibited strong evidence for unidimensionality. The overall Cronbach’s alpha of the 8-item DIMI scale was 0.869 and did not improve with removal of any of the items.

Relationships between DIMI scale score and covariates

The DIMI scale scores were not normally distributed; therefore, non-parametric tests were used to examine associations with the covariates. There were no statistically significant differences in mean DIMI scale scores for any of the participant demographic variables of interest, including gender, race/ethnicity, household income, or Axis I diagnosis. Cronbach’s alphas, within race/ethnicity, were 0.904 for African Americans; 0.882 for Hispanics; 0.758 for Caucasians; and 0.654 for individuals categorized as “Other” (data not shown).

Overall, individuals with more than two psychotic episodes (n=20, 32.8%), as compared to individuals with less than two psychotic episodes (n=41, 67.2%), had statistically significantly higher mean DIMI scale scores (p=0.04), and there was a trend toward statistical significance in the relationship between mean DIMI scale score and number of psychiatric hospitalizations (p=0.06).

Relationships between DIMI scale score and participant responses to selected contextual questions

There was a statistically significant association between higher DIMI scale scores and affirmative responses to the following questions: “[Have you] been given off a sense of stigmatization even if something is not explicitly said to you?” (p=0.002); “Have you felt that people judged you based on your mental illness?” (p=0.027). Further, significantly more participants answered “less” to the question “After hospitalization, did you see relatives more, less often or the same?” (p=0.010) [with this question framed to the respondent in the context of the relative initiating the contact (or not)]. Finally, higher mean DIMI scores were observed among participants who reported having been in a situation where others said something unfavorable or offensive about people with mental illness (n=33/60 or 55%) as opposed to those who did not (27/60 or 45%); this relationship approached statistical significance (p=0.084).

DISCUSSION

Data from our exploratory factor analysis of the 8-item DIMI scale demonstrated that the original 8-item had strong evidence for unidimensionality; subsequent removal of items did not improve the solid internal consistency originally present in the preliminary factor analyses (κ=0.869). Accordingly, the DIMI scale possess sufficient psychometric rigor to be utilized in research settings to assess processes of perception and detection of stigma attributable to mental illness among individuals with mental illness, and specifically individuals with recent psychiatric hospitalizations. Our analysis is one of only several to examine stigma among this particularly vulnerable subgroup of individuals: individuals who have recently been hospitalized due to a psychiatric episode and may thus be at an elevated risk of public stigma and self-stigma and their negative health consequences (Corrigan, Larson & Sells, 2007; West et al., 2011; Link & Phelan, 2014; Gonzales et al., 2015). Furthermore, this is the first peer-reviewed research to develop and validate a stigma tool for use in this group and to also examine associations with various socioeconomic and contextual correlates.

An amalgam of factors, spanning from cultural taboos on mental illness and help-seeking (Wahl, 1999; West et al., 2011), to negative portrayals of mental illness in popular media (e.g., television, movies, music, and video games), contribute to the unfavorable disposition that some in the general population may possess in relation to individuals with mental illness (Gary, 2005; Manderscheid et al, 2010). In this vacuum, public stigma against persons with mental illness is galvanized by a broad spectrum of mischaracterizations and stereotypes, including those which posit that persons with mental illness, and especially those with the most severe forms (e.g., schizophrenia, bipolar disorders), are intellectually weak or inferior, or more prone to unrefined, slavish or aggressive behavior (Link et al., 1999; Schulz & Angermeyer, 2003). In this analysis, we determined that daily indignities were significantly higher among participants with more than two psychotic episodes, providing evidence that individuals with the most severe or persistent symptoms may be the most likely to perceive, experience—or be impacted by—public stigma (Gaebel, Zäske & Baumann, 2006).

Through our analysis, we also observed that DIMI scale scores were higher among those who reported that their relatives visited less frequently following their psychiatric hospitalizations. This aligns with findings from a Gonzalez’ qualitative analysis (Gonzales et al., 2015) consisting of focus groups with adult mental health consumers and college students with mental health diagnoses. In this study (Gonzales et al., 2015), it was determined that mental illness-related micro-aggressions experienced by mental health consumers most commonly emanated from members of the respondents’ personal and professional circles–chiefly, friends, family members, and authority figures—indicating that some the ripest opportunities for addressing stigma may rest in the most intimate realms of individuals’ social networks. However, because we did not ask respondents about other sources of stigma, it is not possible to determine if, relatively speaking, any particular social group or type of acquaintance was more commonly observed or “deeply felt” as the source of stigma.

Importantly, objective interpretations of stigma and micro-aggression illustrate that certain attitudes and behaviors may be deemed to be especially prescriptive, overzealous or otherwise patronizing, irrespective of the offender’s conscious intent (Solorzano, Ceja & Yosso, 2000; Sue et al, 2007; Gonzales et al., 2014). Indeed, diametrically-opposed relationships can exist in intent and outcome, where behaviors or comments may be interpreted negatively by the recipient though not intentionally formulated by the speaker to be taken with offense (Gonzales et al., 2014). Such complexities give weight to calls for mental health clinicians and public health officials to more incisively educate the general public and family and friends of individuals with mental illness on the nuances and sensitivities ensconced in the experience of having a mental illness.

While there were no statistically significant associations between DIMI scale score and participant demographics, we observed higher scores among African Americans and Hispanics. Prior research in this growing body of investigation has illustrated that stigmatizing attitudes related to mental illness and treatment may be higher among African Americans and other racial/ethnic minority populations, as compared to Caucasians groups (Wahl, 1999; Cooper et al., 2003; Gary, 2005; Parcesepe & Cabassa, 2013). Further interdisciplinary examinations of this potential for a “double burden” (Doty & Fund, 2003; Carliner et al., 2014)—that is, having a stigmatized mental illness condition and being a member of a historically marginalized or disadvantaged racial or ethnic group—may prove valuable in understanding nuances in this subset of individuals whose rates of self-stigma may be even more pronounced (Wahl, 1999; Harris, Edlund & Larson, 2005; Parks et al., 2006).

Limitations to this analysis exist. First, the study sample was small and consisted primarily of males and African Americans and Hispanics with low SES; thus, broad generalizations and inferences to other demographic and socioeconomic groups not proportionally represented in this sample cannot be consummately made. Furthermore, participation in the study was limited to individuals with a clinician-diagnosed serious mental illness, and focused on recently hospitalized individuals, a group which is thought to be at higher risk for stigma and other negatives outcomes associated with their condition (Kessler et al., 2003); therefore, the tool’s validated properties may not have equivalent utility in assessments conducted with individuals with other forms or manifestations of mental illness than those delineated in this analysis. Moreover, we did not explicitly assess construct or discriminant validity; however, our factor analysis results point to a substantial degree of fidelity in the resulting tool, and our approach has been supported in the literature (Baghaei & Yazdi, 2017). Additionally, this tool does not directly address self-stigma, but rather illuminates how individuals with mental illness may detect and perceive mental illness-related stigmatization, framing trends in recurrence. Future research in this area should incorporate larger cohorts and attempt to be inclusive of other subtypes of individuals experiencing mental illness and its broad sequelae of psychosocial effects.

In summary, results from our analysis suggest that the DIMI scale possesses good internal consistency and is appropriate for use in research settings, with the potential to better characterize how and to what extent mental illness-related stigmatizing attitudes and behaviors are perceived as being directed at individuals with mental illness or a history of psychiatric hospitalizations, and how these modalities in stigma detection may continuously manifest in the recipient.