Abstract
Evidence shows that stigma negatively influences the quality of life of persons with severe mental illness. Nonetheless, stigma towards mental illness is lower among persons with a lived experience of mental illness compared to the rest of the population. Understanding the association between stigma of mental illness and the mental status of individuals living in urban India and whether this association is moderated by demographic factors opens a new avenue for prevention of social exclusion. Persons diagnosed with schizophrenia, bipolar disorder, or severe unipolar depression (cases, n = 647) were recruited from among hospital patients in New Delhi between November 2011 and June 2012 and matched with non-psychiatric urban dwellers by age, sex, and location of residence (controls, n = 649). Propensity score matching with multivariable linear regression was used to test whether stigma towards mental illness, measured by a 13-item Stigma Questionnaire, differed between cases and controls. Cases reported significantly lower stigma scores than controls (b = -0.50, p < 0.0001). The strength of the association between mental illness and stigma was not affected after controlling for age, caste, sex, education, and employment status, while wealth marginally reduced the strength of the association. These findings suggest individuals with a lived experience of mental illness, in New Delhi, India, may be more tolerant towards mental illness and support the need to involve persons with lived experience in the development and implementation of health promotional campaigns and programs aimed at reducing stigma towards mental illness.
Introduction
Severe mental illness (SMI)—schizophrenia, severe depression, and bipolar disorders—is a leading cause of years lived with disability (YLDs) (Vigo et al., 2016). Stigma affects all spheres of life of persons with severe mental illness (PWSMIs), making combatting stigma a public health priority. Stigma was initially conceptualized by Goffman (1963) as a process by which an attribute is perceived as undesirable, and persons with the attribute are negatively stereotyped and undergo social discrediting (Goffman, 1963). Authors have since introduced new concepts around negative labeling of PWSMIs out of ignorance, stereotyping, and discrimination in a given social context characterized by relations of unequal power (Link & Phelan, 2001; Thornicroft et al., 2007). Link and Phelan (2001) introduced the concept of structural stigma associated with the idea of macrosocial forms of stigma that induce oppression of certain social groups with identities or statuses considered as devalued in the normative system of dominant cultural groups (Hatzenbuehler, 2016). PWSMIs represent one such group.
Stigma associated with mental illness is complex and comes in many forms. Institutional stigma is legitimated and perpetuated by policies and laws promoting discrimination (Corrigan et al., 2004; Evans-Lacko et al., 2012; Pryor & Reeder, 2011). Public stigma is expressed by the general population (Corrigan, 2005). Public stigma of mental illness affects recovery (National Academies of Sciences & Medicine, 2016), healthcare service utilization linked to dissimulation of illness (Corrigan, 2004; Henderson et al., 2014; Rüsch et al., 2014; Thornicroft, 2008), adherence to treatment (Fung et al., 2010), social relationships (Ando et al., 2013), access to employment and working conditions (Corrigan et al., 2006, 2009; Harris et al., 2014; McGurk et al., 2009; Thornicroft et al., 2009), housing (Axer et al., 2015; Rosentraub, 2007), and educational opportunities (Boysen & Vogel, 2008). Health professionals are not immune to public stigma: they endorse stereotypes (Kingdon et al., 2004), and many fail to recognize recovery as a possible outcome for SMI (Magliano et al., 2004). Public stigma is also prevalent among caregivers and family members, compromising treatment access, referral, adherence, dropout, and the recovery process (Corrigan & Miller, 2004; Larson & Corrigan, 2008; Liberman et al., 2002). Individuals with mental illness learn about public stigma and internalized ideas associated with mental illness such as that it means one is “dangerous,” “weak,” or “useless”. This process, called “modified labelling theory,” results in self-stigma and the internalization of negative beliefs and social responses (Corrigan et al., 2005, 2013; Ritsher & Phelan, 2004). Self-stigma negatively erodes PWSMIs’ self-esteem, leading to social withdrawal, demoralization, secrecy, and lower quality of life, contributing to delay in illness detection and treatment and affecting coping mechanisms to fight stigma (Corrigan et al., 2004; Lien et al., 2015; Link et al., 1989, 2001; Rosenfield, 1997). In addition, stigma by association towards the families and caregivers of PWSMIs adds to the challenging process of recovery (Andrea & Darryl, 2015; Koschorke et al., 2017; Roe, 2001).
The treatment gap for persons with mental illness is widest in low- and middle-income countries (LMICs), where 76–85% of such persons go untreated, compared to 35–50% in high-income countries (World Health Organization, 2017). For example, in China and India, the two largest LMICs in terms of population, an estimated 80% of persons with mental illness—230 million and 150 million persons, respectively—are in need of mental health care (Gururaj et al., 2016; Huang et al., 2016; Phillips et al., 2009). Stigma faced by PWSMIs is highly prevalent in LMICs and constitutes a considerable barrier to accessing care as it plays an important role in treatment avoidance (Mascayano et al., 2015). Stigma towards mental illness results in a multipronged negative influence on PWSMIs’ overall quality of life through negative influence on healthcare provision, education, employment and relationship opportunities, self-esteem, and overall social inclusion, making recovery elusive (Kallivayalil & Enara, 2016; Mascayano et al., 2015; Sarkar & Punnoose, 2016). In Ethiopia, for instance, Shibre et al. (2001) have shown that patients with schizophrenia prefer to hide their condition when interacting with health professionals (Shibre et al., 2001).
Given its role as an obstacle to treatment seeking and recovery, investigating stigma of SMI is of central importance in India for several reasons. First, persons with mental illness are perceived as dangerous and aggressive, triggering more social distance (Kermode et al., 2009). Second, largely because of stigma, people with mental illness are more likely to be poor and unemployed (Trani et al., 2015), and to face barriers to recovery (Grover et al., 2016) and higher poverty, as is the case elsewhere (Trani et al., 2015). Stigma—and poverty—associated with mental illness is even higher for women (Thara et al., 2003a, 2003b) and for persons from historically disadvantaged social groups such as scheduled tribes, scheduled castes, and other backward castes (ST/SC/OBC) (Jaspal, 2011). Third, due to limited mental healthcare services, informal unpaid care is widespread and assumed by family members (Seshadri et al., 2019; Thara et al., 1998), particularly women (Balaji et al., 2012; Chatterjee et al., 2014; Jagannathan et al., 2014). Yet this role is not socially valued, and caregivers, particularly women, lose opportunities for social connection, including marriage but also education and employment, do not benefit from any social support, and go through different phases of emotions and attitudes associated with their experience (Dijkxhoorn et al., 2019, 2023; Mathias et al., 2019). The caregiver's burden has been shown to be correlated with the severity of the pathology (Jagannathan et al., 2011, 2014). The literature has demonstrated that stigma towards caregivers may result in efforts to conceal the mental illness from outsiders, as well as in family members’ negative attitudes and discriminatory behavior towards the PWSMI (Grover et al., 2019; Loganathan & Murthy, 2011; Shrivastava et al., 2011).
Many studies—in India and elsewhere—have investigated the association between familiarity with mental illness and stigma. A majority of studies found an inverse association between familiarity and stigma: the more familiar someone is with mental illness, including caregivers and family members close to a PWSMI, the less they display stigmatizing attitudes (Adewuya & Makanjuola, 2008; Mathias et al., 2018). Yet, other studies have found greater familiarity to be associated with greater stigma, particularly among family members who have a very close relationship with a PWSMI and report a high burden of caregiving (van der Sanden et al., 2016). Yet, to the best of our knowledge, the extent to which PWSMIs themselves share the stereotypes (negative beliefs), prejudice (negative attitudes), and discrimination (negative behaviors) towards mental illness has never been investigated. Given the complex nature of stigma experienced by people with mental illness, its potentially negative consequences for PWSMI, and the high burden of untreated mental illness in India, our study aims to address the four following questions: (1) Do persons with SMI themselves share the same public stigma towards mental illness in the Indian context as members of the general population? (2) Does the stigma expressed by persons with and without mental illness differ by demographic and socioeconomic characteristics such as age, sex, caste, education, employment, and wealth, all factors that have been shown to influence stigma? (3) Does knowing a PWSMI influence the expressed stigma of SMI? And (4) Do cultural factors, such as beliefs that PWSMI have special powers or that the illness could be caused by a spirit or by someone ill-intended, also influence stigma?
The present study investigates differences in the perception of stigma of SMI, comparing persons from New Delhi with a clinical diagnosis of SMI and persons in the general Indian population from New Delhi without a clinical diagnosis of SMI. We examined if PWSMIs have stigmatizing attitudes towards SMI that differ from persons in the general population, controlling for sex, caste, education level, employment status, and wealth. A finding of lower levels of expressed stigma among PWSMIs would substantiate enhancing the quality of anti-discriminatory programs by involving PWSMIs themselves in their conception and implementation.
Methods
Study design
The study design has been described elsewhere (Trani et al., 2015). In brief, data were collected in New Delhi, India from November 2011 to June 2012. Outpatients diagnosed either with schizophrenia or severe affective disorders using International Classification of Diseases 10th revision (ICD-10) criteria were randomly recruited after due informed consent from the psychiatry outpatient department of a public free teaching hospital in New Delhi, India. A comparison group of non-psychiatrically ill control individuals matched one to one with cases on the basis of sex, age, and place of residence was randomly selected from all over Delhi. There were no exclusion criteria other than refusal to consent to participate.
Study personnel, assisted by caregivers, conducted face-to-face interviews with 647 PWSMIs during their hospital visits. All respondents with SMI were in a clinically stable enough state to be able to provide informed consent and actively participate in the interview. Matched community controls (n = 647) were interviewed in their homes. To identify controls, we started at the level of the house of the PWSMI and randomly selected a direction by spinning a pointer. We then selected the closest household in the direction of the pointer and interviewed a resident of that household who matched the PWSMI by sex and age (plus or minus five years). Full confidentiality of the index case was maintained. We did not mention to anyone the house number of the person with mental illness. We stopped in front of the address but we spun the pointer outside in the street and did not reveal to anyone the reason why we stopped at such a spot in the street. We introduced the study in the neighborhoods where PWSMIs were living as being a health and livelihood survey done by the hospital. In addition to the measure of stigma (see below), the interview assessed demographic and socioeconomic characteristics such as sex, age, caste, marital status, level of education, asset ownership, income, employment, as well as information about health behavior, healthcare services, and social participation.
Measures
Outcome: Stigma
Stigma was measured using the Stigma Questionnaire (SQ), which was developed through a large-scale, cross-cultural collaborative initiative for understanding stigma, translated in Hindi and tested and validated in India (Littlewood et al., 2007). The SQ is based on “the psychiatric emphasis on extrusion and the sociological emphasis on a devalued identity” (Littlewood et al., 2007, p. 180), but also attempted to bring in local sociocultural concepts to define and measure stigma. From a total of 123 initial questions related to stigma, Littlewood et al. (2007) retained a validated set of 24 items measuring stigma that we used for the interview in the present study.
To assess attitudes and beliefs, study participants responded to questions about a hypothetical vignette, narrated in lay terms, about a young man with schizophrenia who partially responds to treatment (see Appendix for the vignette). Vignettes are often used in mental health research to describe signs of a mental health condition without a diagnostic label (Link et al., 1987). Vignettes offer a standardized presentation of various facets of the condition to a large number of respondents (Schomerus et al., 2014). The vignette was followed by a set of 24 questions that assessed how the respondent would treat the hypothetical PWSMI described in the vignette, including their perception of dangerousness and desire for social distance (Jorm & Oh, 2009). For instance, as a measure of the respondent's comfort with a level of physical proximity to a PWSMI, the first question asks: “Would you be frightened if this man came to live next door to you?” The second question, “Would you be happy if he married your sister?”, measures the threat to family reputation. The 24 stigma items have a four-category rating scale: 1 – yes, very much (highest stigma), 2 – yes, a little, 3 – no, not much, 4 – no, not at all (lowest stigma).
Among the 24 items, 10 were removed based on the findings of Littlewood et al. (2007): eight items were removed because they measure “aetiological beliefs that might be related to stigma” (p. 186) which were not relevant to our study, and two items were removed due to low correlations between the item responses and the total scale score. We also removed the item “Should he stay in hospital his whole life?” because of its correlation with the rest of the other items (cor = 0.48, p < 0.001). We then summed the remaining 13 items, taking into account those that were reverse coded, to create a single scale score. Higher composite stigma scores indicate higher levels of public stigma of SMI. Also following the methods used by Littlewood et al. (2007), we deleted observations with missing responses on two-thirds or more of the items (Littlewood et al., 2007). We therefore discarded data from 86 PWSMIs (13.3%) and 21 controls (3.2%) due to incomplete data.
Exposure: Severe mental illness
The mental health status of respondents was the primary independent variable (SMI or not). New outpatients were diagnosed by well-trained psychiatrists based on ICD-10 criteria.
Cofounding factors
We adjusted for individual characteristics that have been shown to influence self-stigma of mental illness (Grover et al., 2017b). Demographic covariates included age (continuous, 11–85 years old), education (three categories: below primary or primary completed, middle school, high school or higher education), employment status (three categories: no employment, stable work (i.e., work as regular wage or salaried employee), unstable work (i.e., occasional work without contract)), sex (man, woman), caste (ST/SC/OBC (i.e., disadvantaged groups), other castes). A three-category variable for wealth (lowest 20%, middle 20–80%, highest 20%) was created based on a 15-indicator assets index, with scores calculated using polychoric principal component analysis (PCA) (Kolenikov and Angeles, 2009).
In addition to demographic covariates, we included as covariates other factors that have been shown to be associated with both SMI and public stigma (Jorm & Oh, 2009). Exposure to individuals with mental illness was assessed by the question in reference to the vignette: “Has any person you know personally ever had a similar illness?” (response options “yes” or “no”) (Adewuya & Makanjuola, 2008; Mathias et al., 2018). Familiarity with a PWSMI has been identified as reducing negative attitudes and discrimination (Dietrich et al., 2004). To assess cultural and spiritual beliefs in mental illness as an expression of supernatural powers, participants were asked regarding the young man described in the vignette: “Might this young man have any special powers (to heal, to predict future events, to cause illness)?” A previous study in urban India using the same vignette has shown that such beliefs were associated with higher expressed stigma (Jadhav et al., 2007). To investigate if study participants make some specific moral attribution about the etiology of mental illness—whether mental illness is considered a disease or a moral failure (Corrigan et al., 2003; Krendl & Freeman, 2019; Pescosolido & Martin, 2015)—we asked “could this illness be caused by some spirits or an enemy harming him [the young man in the vignette]?”; in other words, does he bear some responsibility and could he have brought the disease on himself? The extent to which such prejudice is prevalent potentially relates to discrimination and stigma of mental illness (Jadhav et al., 2007).
Statistical analyses
Descriptive statistics were computed and chi-squared tests and t-tests were used to compare the demographic and socioeconomic characteristics and stigma scores between persons with SMI and matched controls. Confirmatory factor analysis (CFA) was used to check for the unidimensionality of the composite stigma score. The root means square error of approximation (RMSEA), the comparative fit index (CFI), and the standardized root mean square residuals (SRMSR) were examined to confirm the acceptable fit of a unidimensional model. The internal reliability of the unidimensional composite stigma score was evaluated using Cronbach's alpha (Terwee et al., 2007).
Despite having matched PWSMIs to controls without SMI on sex, age, and residence, to further ensure that comparisons were made on similar individuals, we evenly balanced the distributions of observable confounding factors across these two groups using the nearest neighbor propensity score matching (PSM) method in a 1:1 ratio, and a 0.25 caliper (Stuart et al., 2009). In addition, robustness check analyses were conducted using an OLS model and a coerced exact matching model, both without PSM. The three models were recalculated using the CFA first factor as the dependent variable. Results were similar regardless of method used. Diagnostic assumptions for the models were met: normal distribution of stigma score residuals, absence of multicollinearity, and confirmation of assumptions of independence of observations. The balancing tests show that PSM using the nearest neighbor matching estimator removes most of the bias between the treatment and non-treatment groups: in all analyses, Rubin's B is below 25%, Rubin's R is between 0.5 and 2, and the percentage bias is below 10% for almost all covariates (with the exception of employment) (Figure 1) (Rubin, 2001). We interpreted any remaining difference in the outcomes as the average treatment effect on the treated, the group of PWSMIs. We reported the effect size by showing the partial version of eta-squared (η^2) with a one-sided confidence interval, which means the upper bound is fixed at 1.00. For a partial eta-squared, 0.14 is considered a large effect size (Lakens, 2013). All data analysis was conducted using R version 4.2.1.

Standardized % bias across covariates, nearest neighbor matching estimator.
We tested models with interaction terms that accounted for differing effects of SMI with sex, caste, education level, employment status, assets index, exposure to SMI, and beliefs in special powers and in supernatural causes of the illness, before introducing all terms together in the same model (Jaccard & Turrisi, 2003). The literature has shown that, in India, the level of public, associated, or self-stigma may vary by sex (Boge et al., 2018), age (Grover et al., 2017b; Thara & Srinivasan, 2000), education (Zieger et al., 2016), employment, caste (Trani et al., 2015), income, or socioeconomic status (Grover et al., 2019; Pal et al., 2017). We only retained significant interaction terms that improved the fit of the model (Jaccard & Turrisi, 2003).
Ethical clearance
The present study received ethical approval from the University College London Research Ethics Committee and the Dr Ram Manohar Lohia Hospital Institutional Ethics Committee.
Results
Table 1 presents information on the demographic characteristics of the PWSMIs and of the controls in the general population. There were no significant differences between cases and controls in terms of age or sex, with a mean age of 36 years and women comprising slightly over one-third of each group. The mean stigma score differed significantly between groups, with PWSMIs having a lower mean score of 22.74 (SD = 5.3) compared to that of 29.44 (SD = 6.2) (p < 0.001) for controls. PWSMIs were significantly more likely to be in the lowest quintile of the wealth index, to be unemployed, to be of a lower caste, and to have a lower education level than control participants. In addition, PWSMIs were much more likely to personally know another person with mental illness. Conversely, there was no significant difference in cultural beliefs about power or moral attribution between PWSMIs and controls.
Demographic and perception of stigma of PWSMIs and controls.
Table 2 compares mean and median scores for each of the 13 stigma items between PWSMIs and controls in the general population. It shows that the stigma score was higher for controls on each item. The highest mean difference was observed for the question “Would you avoid talking to him if possible?” (mean score 2.73 among controls compared to 1 among PWSMIs). The smallest difference of 0.27 in both cases was observed for “Do you think he will get ill again even if he takes the doctor's medicine?” and “Would it be wise for this man to inherit his parent's property?”
Stigma score per stigma item for persons with SMI and controls in the general population.
Figure 2 shows the distribution of the stigma score between PWSMIs and controls. The overall distribution is more compact and towards the origin of the x axis for PWSMIs compared to controls.

Distribution of cases with SMI and controls according to stigma score.
The Cronbach's Alpha for inter-factor correlation of stigma items was 0.72, indicating acceptable reliability (Nunnally & Bernstein, 1994). A CFA showed that one factor solution provided an excellent fit, with all factor loadings between 0.51 and 1.88 (see Table 3) (Nunnally & Bernstein, 1994). Similarly, the RMSEA of 0.035 (good fit at < 0.05), TLI of 0.94, and CFI of 0.96 demonstrate an excellent fit (Bentler & Bonett, 1980; Bollen, 1986). These findings confirm that it is relevant to treat the 13-item stigma questionnaire as a unidimensional score scale (Littlewood et al., 2007).
Standardized loading estimates on factor 1 from the confirmatory factorial analysis.
The crude regression model showed that the average stigma scores of PWSMIs were five points lower than those of controls. The effect size of mental health on stigma is considered large (0.26), even after adjusting for sex, age, caste, education level, employment status, asset index, familiarity with mental illness, beliefs in supernatural power, and moral attribution about the etiology of mental illness (see Table 4). It shows that PWSMIs exhibited less stigma towards others with similar illness, compared to controls. Beliefs in supernatural powers was marginally associated with less stigma (−0.05, 95CI[-0.11–0.00], p < 0.001), with a negligible effect size (5.05e-03). Conversely, moral attribution of mental illness to a spirit or an enemy translated into a 0.10 point higher stigma (95CI[0.05–016], p < 0.001) but the effect size was small (0.01).
Linear regression models following propensity score matching for stigma score.
Note: ***p < .001; **p < .01; *p < .05.
We reported the effect size by showing the partial version of eta-squared (η^2) with one sided confidence interval, which means upper bound fixed at 1.00. The treatment variable explains most of the variance in the 13-item stigma index.
We examined several interaction effects between having a mental illness and each of the covariates and found no significant differences between models with and without the interactions. The adjusted R2 also did not increase with the addition of any of the interaction terms, indicating that the interaction effect had minimal influence on the main effect of SMI on stigma scores (results not shown). Fitness of the model did not increase with interaction terms between mental illness and sex, caste, education level, assets index, beliefs in supernatural power, moral attribution, and familiarity with mental illness, indicating that the association between mental illness status and stigma score did not vary by levels of these variables. In addition to testing interactions, we also fit mediation models to examine whether any demographic attributes or beliefs in supernatural power, moral attribution, or familiarity with mental illness mediated the association between mental illness status and stigma (results not shown). These models produced some statistical evidence of a mediating effect of moral attribution on the association between mental illness status and stigma; however, the magnitudes of these effects were negligible, accounting for only ∼1.5% of the total effect.
Discussion
This study investigated whether public stigma related to mental illness differed according to one's mental health status—testing whether one's own mental health status influences desire for social distance from PWSMIs—in New Delhi, India, where the high level of public stigma towards PWSMIs and its consequences on the internalization of stigma has been investigated previously (Koschorke et al., 2014). We controlled for demographic and socioeconomic characteristics, cultural beliefs in supernatural faculties, moral attribution, and familiarity with the disease.
To the best of our knowledge, this is the first study from a middle-income country to scrutinize public stigma using an ethnographically derived stigmatization scale among a large urban sample of PWSMIs and matched controls in the community. Existing studies focus on other dimensions of stigma and are exclusively hospital-based studies. One study investigated experienced stigma of PWSMIs and/or their caregivers (Grover et al., 2017; Singh et al., 2016). Both caregivers and patients experienced high levels of stigma, with persons with schizophrenia perceiving the most stigma. Those who had more modern perceptions of mental illness reported less stigma (Mukherjee & Mukhopadhyay, 2018). Similar results have been found in other LMICs. For instance, family members of persons with mental illness—both from an outpatients’ department and in a community survey—reported high levels of ‘family stigma’ in two studies carried out in Ethiopia (Girma et al., 2014a, 2014b).
The results indicate that people in the general population exhibited significantly higher stigma scores compared to PWSMIs. Our findings suggest that acceptance of social stereotypes about mental illness among PWSMIs is less intense than among members of the general public. PWSMIs might have identified with the person in the vignette due to their shared experience and demonstrated a more tolerant attitude towards someone else with mental illness. If PWSMIs identify with the person with schizophrenia described in the vignette, but express less stigma than individuals in the community, it may indicate that they agree to a lesser degree with a perception of mental illness characterized by prejudice and discrimination (Corrigan & Watson, 2002). A meta-analysis of anti-stigma programs towards mental illness reported that in-person contact with members of the stigmatized group is an effective strategy to fight public stigma—with long-lasting effects on attitudes (Corrigan et al., 2012, 2015). By analogy, we argue that PWSMIs develop knowledge from their lived experience with mental illness and can better relate to other PWSMIs.
Findings were consistent across age, sex, caste, level of education, wealth, and employment status, and the size of the difference in stigma between the two groups did not significantly vary with sociodemographic predictors. This is at odds with existing literature that indicates some disparity in stigma according to socioeconomic characteristics. Yet, comparison must be carried out with caution because of the diversity of tools used to measure various dimensions of stigma, which may explain why different factors are associated in a variety of ways with stigma measures (Griffiths et al., 2008; Wolff et al., 1996). This variety might also suggest the pervasiveness of stigma across individual demographic and socioeconomic characteristics in India (Boge et al., 2018; Loganathan & Murthy, 2011). Some socioeconomic characteristics—such as being an older adult, poor, from an SC/ST/OBC, and being unemployed or uneducated, especially for women—are associated with prejudice and exclusion in India, particularly when these groups also show signs of mental illness (Grover et al., 2016; Kijima, 2006; Loganathan & Murthy, 2011) (Trani et al., 2015).
Studies looking at expressed stigma or desire for social distance towards mental illness have reported sex and age differences in different directions. Multiple studies in the general population showed no significant sex difference (Angermeyer et al., 2003, 2004; Martin et al., 2000), while some showed greater expressed stigma among women (Lauber et al., 2004) and others showed greater expressed stigma among men (Jorm & Griffiths, 2008; Martin et al., 2007). In India specifically, women with schizophrenia scored higher on stereotype endorsement (Singh et al., 2016). Another study conducted in the general population of five Indian urban centers reported higher rates of perceived stigma among women (Boge et al., 2018). Nevertheless, in our study, our mixed sample of women did not show a higher mean score of negative stereotypes than men.
Literature also reports that lower education, particularly illiteracy, increases perceived and experienced stigma (Girma et al., 2013; Lincoln et al., 2015; Zieger et al., 2016). Yet, higher negative perception of mental illness has been observed among caregivers with higher education in West Bengal (Mukherjee & Mukhopadhyay, 2018). Other studies, including ours, have shown a small or no significant association (Jorm & Griffiths, 2008; Taskin et al., 2003). A possible explanation lies in the relatively high average level of education of our study population.
Surprisingly, employment status and wealth background did not influence stigma scores. In India, a study showed that patients with schizophrenia who were employed were more likely to exhibit stigma resistance (Singh et al., 2016). Studies in other cultural contexts examining the association between socioeconomic background and stigma have found that low socioeconomic class affects social distance and discrimination (Sağduyu et al., 2001; Taskin et al., 2003). A study in China has shown that levels of internalized stigma among PWSMIs from families with lower income levels were higher than for those living in families with higher incomes: the authors argued that PWSMIs with lower wealth might face social stigma more often and have lower levels of self-esteem (Ran et al., 2018).
Beliefs in supernatural causes and powers of mental illness are widespread in India (Srinivasan & Thara, 2001), and therefore the cure can only come from God and those in direct relation with Them such as pandit (priests) and mali (oracles) (Rawat et al., 2021). We found that such beliefs were marginally associated with lower stigma scores. A study comparing rural and urban India using the same measure found that such beliefs were associated with lower stigma of mental illness, particularly in rural areas (Jadhav et al., 2007). Cultural beliefs about supernatural powers tend to move away from negative labeling and translate to lower rejection and discrimination of PWSMIs. A possible explanation is that PWSMIs are believed to interact with divinity and intercede with God on behalf of lay people (Sinha & Ranganathan, 2020).
As expected, higher stigma was demonstrated by participants who attribute the etiology of mental illness to moral factors. Sixty percent of study participants believe that PWSMIs might be responsible for their condition, by making enemies or deserving the wrath of spirits, in accordance with other studies in different cultural contexts (Bignall et al., 2015; Ventevogel, 2016). The role of malevolent forces, an evil spirit or an ancestor, or someone using black magic or witchcraft in causing mental disorders is recognized as crucial in various non-Western contexts, and justifies the role of traditional methods for treating mental illness (Crawford & Lipsedge, 2004; Galvin et al., 2022; Joel et al., 2003). A large majority of subjects believing in moral attribution also showed negative attitudes and discrimination towards mental illness. Predisposition to reject mental illness is particularly apparent in the situation of individual proximity, such as being afraid to have a neighbor with mental illness (85.9% of controls and 74.8% of PWSMIs) or being unhappy to work with a PWSMI (respectively 76.2% and 68.7%) or if a PWSMI were to marry one’s sister (77.4% and 65.0%), as found in other studies (Taskin et al., 2003). A study among patients and their relatives in Vellore, south India, reported that belief in karma and evil spirits as causes of the illness among relatives was associated with higher expressed stigma (Charles et al., 2007). Such beliefs have been shown to influence the way people present a potential mental health problem and to have implications for the quality of care (delay in seeking care, poor patient–health professional interaction, and limited medication compliance), rehabilitation, and social inclusion of individuals with mental illness in India (Rawat et al., 2021). Community beliefs are largely ignored by mental health services in India and a previous study examining a “Programme for Improving Mental Health Care” (PRIME), a multiplatform multicomponent mental healthcare intervention in Madhya Pradesh for people with depression and alcohol use, suggested to increase contextually defined community-based care for better impact (Shidhaye et al., 2019). Going a step forward, psychosocial interventions could be built and implemented in partnership with the local community through an empowering approach (Mathias et al., 2020).
Surprisingly, familiarity with PWSMIs did not significantly predict less desire for social distance, in contrast with findings from studies in other cultural contexts that have found that familiarity reduced desire for social distance (Lyndon et al., 2019), particularly among caregivers (Aromaa et al., 2011). It may be that familiarity is counterproductive in India, where discrimination has been shown to primarily come from community members and even from family and friends of PWSMIs (Grover et al., 2020).
Limitations
Our study is not without its limitations. First, the cross-sectional nature of the study does not allow us to draw any causal inference. Second, patients were diagnosed by a psychiatrist but we did not equally assess each control; therefore, it is possible that some control participants may themselves be PWSMIs. Third, PWSMIs were selected from among people seeking care at a hospital and therefore might not be representative of all PWSMIs in urban India. Persons from lower socio-economic groups who cannot afford private mental healthcare may be overrepresented, while those with the financial resources to obtain private mental healthcare and those from the most marginalized groups who do not even seek care may be underrepresented. It can be argued that some PWSMIs do not seek care because of stigma—among other reasons such as cost of transportation, unavailability of a caregiver to come with them, and disbelief in the capacity of the healthcare system to address the issue. Yet the bias due to interviewing treated individuals might be minimal here because several studies have found that stigma delays care more than suppresses it (Dockery et al., 2015; Lauber & Rossler, 2007). Fourth, the study took place in New Delhi and findings cannot be generalized to rural India. Fifth, there could be a difference in understanding of the stigma questionnaire between PWSMIs and controls. However, we would argue that any such difference would likely be minimal because the SQ has previously been validated (Jadhav et al., 2007), and we tested it for content validity. Sixth, the SQ directly measured the manifestation of public stigma only, not of internalized stigma. By exhibiting less public stigma, PWSMIs may be indicating their disapproval of the social stereotype, which may be interpreted as an indirect way to cope with self-stigma. Seventh, one could consider the possibility of social desirability explaining the difference in score. We discard this possibility because we tested the questionnaire thoroughly. The use of the vignette and the questions without mentioning any mental illness tends to protect against social desirability. Besides, enumerators were trained to never mention mental illness and did not anticipate any specific result as this was uncharted ground. In focus group discussions, we found that some patients were adamant to fight what they called mockery and discrimination. Finally, we did not collect information about the specific diagnosis of our participants, making it impossible to compare differences in stigma scores between different conditions. Overall, our findings indicate that personal experience of discrimination or exposure to mental illness may be effective in reducing stigma towards other mentally ill people. Further research is needed to explore to what extent the stigma expressed by PWSMIs towards others with mental illness is indicative of self-stigma or whether it is due to other life circumstances or personal attributes.
Conclusion
This is the first large-scale study to examine public stigma of SMI assessed using a locally validated stigma scale comparing PWSMIs to a matched group of controls randomly selected in the general population. Overall, we conclude that having personal experience or knowledge of mental illness may lead to less stigmatizing attitudes towards PWSMIs in New Delhi, India, particularly among families with higher levels of wealth, reflecting a lower level of internalization of negative societal views about mental illness or self-stigma than could have been expected (Corrigan et al., 2013; Grover et al., 2019; Pal et al., 2017). Having a significantly lower level of expressed stigma towards SMI among cases with mental illness compared to controls and understanding of the forces that promote stigma could have important implications for public health intervention in the sociocultural context of India.
National education campaigns and local interventions by increasing mental health literacy have been shown to reduce stigma in LMICs (Deimling Johns et al., 2018; Kadri & Sartorius, 2005; Sartorius & Schulze, 2005). Our main finding highlights the importance of relying on the involvement of PWSMIs themselves in such initiatives to foster changes in discriminatory attitudes and behaviors resulting from stereotypes amidst the general population (Mascayano et al., 2020) and to allow PWSMIs overall social participation, which has been shown to increase self-esteem, lower self-stigma, enhance feelings of self-efficacy, and improve quality of life (Dunn et al., 2008; Evans-Lacko et al., 2012; Priebe et al., 1998). Engaging PWSMIs in educational programs to fight stigma in the general public can be more effective if they do not themselves endorse the public stigma of mental illness: PWSMIs become an asset in changing the perception of the general public as well as mental health professionals if they can deconstruct with them how public stigma plays out.
In LMICs such as India, studies have shown that discrimination primarily comes from community members and even from family and friends (Grover et al., 2020); therefore, such initiatives should also include family members, who are traditionally the primary caregivers and often involved in the support of PWSMIs (Seshadri et al., 2019; Shrivastava et al., 2011). Too often, PWSMIs tend to hide their mental health status due to the substantiated fear of stigmatization by others, which translates to discrimination and social exclusion (Loganathan & Murthy, 2008). Although they have higher exposure to mental illness through clinical interaction, such interaction does not take place on equal ground, making contact less effective (Peris et al., 2008). To tackle stigma, PWSMIs and their family members need to share their experiences, clarify misconceptions, and create awareness about mental illness in their own local communities. This is of central importance as we know that PWSMIs already have a lower quality of life than the rest of the population (Srivastava et al., 2010) and public stigma only contributes to reinforce this association.
Policy recommendations
Mental health professionals are not exempt from stigmatizing behaviors towards PWSMIs, which results in the latter avoiding seeking, or dropping out prematurely of treatment and the former not being able to deliver impartial care particularly because of implicit negative attitudes (Kopera et al., 2015; Peris et al., 2008). To address such stigma through contact and education (Aberson et al., 2004; Ashburn-Nardo & Johnson, 2008), we suggest new strategies engaging PWSMIs in the development of anti-stigma programs. They could share their experience of navigating the health system, and describe difficult experiences and the ensuing feelings, while challenging the negative views and paternalistic attitudes of those professionals.
Mental health policy in India should consider that public stigma among the general population towards mentally ill people is higher than acceptance of social stereotypes among those who are themselves ill. Hence, mental health policy could prioritize public stigma reduction in its health promotion campaigns and fight public beliefs and promote the rights of PWSMIs (Deshpande et al., 2013). The Mental Health Care Act 2017 stated reducing stigma as an important goal (Duffy et al., 2017). But greater progress is still needed, particularly in terms of increasing available resources and getting states involved (Chadda, 2020). Overall, we hope our findings will help improve the lives of PWSMIs in India through the strengthening of public health efforts towards greater social visibility, contact, and inclusion to make the public, including mental health professionals, more aware of what mental illness really means in the hope of changing their behavior. Without a change in the public stigma, the path to recovery for PWSMIs in India is at risk (Hatzenbuehler, 2017; Link et al., 2017).
Footnotes
Declaration of Conflicting Interests
The author(s) declared no potential conflicts of interest with respect to the research, authorship, and/or publication of this article.
Funding
This fieldwork was funded by the DFID Cross-Cutting Disability Research Programme, at the UCL International Disability Research Centre (GB-1–200474)
