The Self-Stigma of Suicide Attempt Survivors

Although suicide attempt survivors report feelings of shame and stigmatization, no published scale measures self-stigma experienced by attempt survivors. This article describes the creation and validation the Self-Stigma of Suicide Attempt Scale (SSSAS). In this study, the SSSAS was validated in an online sample (n = 292) of suicide attempt survivors. Results supported the progressive model of self-stigma, wherein a substantial proportion of suicide attempt survivors were aware of stigma, but fewer applied that stigma to themselves or felt harmed by it. Reliabilities of SSSAS subscales were high. Harm subscale scores were correlated with depression, self-esteem, recovery, empowerment, and stigma stress in the expected directions. Future research can seek to further validate the scale and explore relationship between self-stigma and other constructs.

. Higher levels of self-stigma endorsement have been consistently connected to reductions in selfesteem (Link, Struening, Neese-Todd, Asmussen, & Phelan, 2001;Livingston & Boyd, 2010;Picco et al., 2016) and increases in depression (Brohan et al., 2011;Livingston & Boyd, 2010;Yen et al., 2005). Individuals impacted by mental illness may also experience varying degrees of distress related to public or selfstigma. This "stigma-stress" has been connected with feelings of shame, reduced self-esteem, and hopelessness (R€ usch et al., 2009, 2014a). Yanos, Roe, and Lysaker (2010) propose a model whereby internalization of stigma could lead to decrements in self-esteem, increased symptoms, and ultimately to heightened suicide risk. Cross-sectional research on individuals with schizophrenia found that self-stigma predicted risk of suicide (Sharaf, Ossman, & Lachine, 2012) and that those with a suicide attempt report greater levels of selfstigma (Assefa, Shibre, Asher, & Fekadu, 2012). In further support of the selfstigma-suicide link, in a longitudinal study of individuals with mental illness (n ¼ 222), self-stigma scores at baseline were associated with prior-month suicide ideation and also predicted suicide ideation at one-year follow-up, while controlling for depression, baseline suicidality, age, and gender (Oexle, M€ uller, & Kawohl, 2018).

Protective Factors
Some individuals seem less susceptible to harmful effects of self-stigma. In the literature on mental illness, empowerment is one proposed antithesis to self-stigma (Yanos, Roe, Markus, & Lysaker, 2008). That is, a sense of empowerment might help people cope with stigma through righteous anger, power, control, and activism (Corrigan, 2002). Self-stigma of mental illness is negatively related to empowerment in the literature (Corrigan et al., 2015;Livingston & Boyd, 2010). Individuals with mental illness living in countries where public attitudes towards people with mental illness are more positive have higher levels of empowerment (Evans-Lacko, Brohan, Mojtabai, & Thornicroft, 2012). Similarly, self-stigma endorsement is connected with lower perceived recovery from mental illness (Oexle et al., 2018).

Suicide Stigma
While a substantial body of research has explored the stigma of mental illness, more recent work has focused on stigmatizing attitudes towards individuals who have attempted suicide (Calear, Batterham, & Christensen, 2014;Witte, Smith, & Joiner, 2010). Further, there is a growing recognition that mental illness stigma, although similar, may differ from suicide stigma in important ways (Sheehan, Dubke, & Corrigan, 2017). Attempt survivors may be stereotyped as irresponsible, cowardly, attention-seeking, or weak (Batterham, Calear, & Christensen, 2013;. Several scales have recently been validated to measure the public stigma of suicide, suicide attempt, and suicide loss survival (Batterham et al., 2013;Corrigan et al., 2016;Scocco, Castriotta, Toffol, & Preti, 2012); however, no scale has been published to measure self-stigma specific to suicide attempt survivors. Thus, the purpose of the current study was to develop and evaluate a measure of self-stigma of suicide attempt, based on the progressive model of self-stigma. Consistent with research on the progressive model, we expected that suicide attempt survivors would indicate the highest level of awareness, followed by progressively less agreement, application, and resulting harm. Secondly, we expected to demonstrate convergent and discriminant validity through positive relationships between self-stigma endorsement (application and harm subscales) and harmful outcomes (i.e., depression, stigma-related stress), along with negative relationships between self-stigma endorsement and positive outcomes (i.e., recovery, self-esteem, and empowerment). A secondary aim was to explore whether time since most recent suicide attempt and number of past attempts were related to self-stigma endorsement.

METHODS
The Self-Stigma of Suicide Attempt Scale (SSSAS) was modeled after the Self-Stigma of Mental Illness Scale (SSMIS; Corrigan et al., 2006) and included the four components of the progressive self-stigma model: awareness, agreement, application, and harm. The 14 items for each subscale were selected based on inclusion as stereotypes on a scale measuring the public stigma of suicide, the Suicide Stigma Assessment Scale (SSAS; Corrigan et al., 2016). Stereotypes from the SSAS were originally generated through community-based participatory research with suicide stakeholders (see . For the SSSAS, each stereotype was presented at each level of the progressive model, beginning with stereotype awareness. The institutional review board at the Illinois Institute of Technology approved the study protocol and all data were collected in August 2016. Participants responded to a solicitation on Amazon Mechanical Turk (MTurk), a web-based platform where individuals receive compensation for completing tasks, such as surveys. The solicitation described the research as an "academic survey for people who have attempted suicide." Interested participants were eligible if they were at least 18 years old with at minimum one past suicide attempt. To guard against those who might inaccurately self-identify as a suicide attempt survivor in order to qualify, we asked participants a second time if they ever tried to end their life (i.e., attempted suicide) immediately following the informed consent, prefaced by the statement "we appreciate your honesty." If participants again responded in the affirmative, the survey continued to ask further questions about their history of suicide attempt (i.e., number of attempts, time since attempt, help-seeking behaviors following the attempt). Participants took about 10 minutes to complete the survey and were paid $1.50 through Mturk. Four quality control questions were embedded within the survey to screen for quality responses. These questions (e.g., "please choose the answer choice 'strongly agree'") were designed to identify participants who did not thoroughly read the survey questions.

Participants
Of the U.S. MTurk participants (n ¼ 383) who responded to the survey, 18 were disqualified during screening because they had not attempted suicide, 10 had incomplete data, 3 participants did not enter a completion code for the payment, and 7 failed one or more of the quality control questions. Data from these participants were excluded from analyses, as were The Self-Stigma of Suicide Attempt Survivors data from 54 participants who completed the survey in 5 minutes or less, which, based on pilot testing, was an insufficient amount of time to thoroughly read and respond to all items. A total of 292 participant responses were analyzed.
We summarize the demographic composition of the sample in Table 1 and provide the highest percentage of each demographic characteristic in the text below. Included participants were roughly 52% female and on average 33.5 years old (SD ¼ 9.42). Most participants reported being heterosexual (82%) and self-identified as Caucasian (80%). About 44% had completed some college, nearly 60% of participants were employed full time, and 36% had household incomes between $25,001 and $49,999. Most respondents had attempted suicide once (68%) or twice (28%), but a few (6%) had attempted three or more times. Participants reported most recent suicide attempts occurring within the last year (14%), between 1 and 3 years (22%), or more than 3 years ago (63%). Some participants (33%) had been hospitalized because of suicidal thoughts or had been hospitalized after an attempt (48%).

Measures
Self-Stigma of Suicide Attempt Scale (SSSAS). The 56-item SSSAS measured four components of the progressive selfstigma model: stereotype awareness, stereotype agreement, stereotype self-concurrence, and harm (Corrigan et al., 2006). Each of the 14 stereotype items were presented at each level of the progressive model, such that participants responded to four statements for each stereotype. These reflected the concepts of awareness (e.g., "I think the public believes most people who have attempted suicide are selfish"), The total percentage of some categories is greater than 100 because participants were allowed to choose more than one option in that category.
agreement (e.g., "I think most people who have attempted suicide are selfish"), application (e.g., "Because I attempted suicide, I am selfish"), and harm (e.g., "I currently respect myself less because I am selfish"). Participants responded using a 9-point scale (9 ¼ strongly agree). Scores were calculated as a sum of items, with higher scores indicating increased self-stigma.
The 5-item RAS-S measured self-reported recovery from suicide and was adapted from the mental illness Recovery Assessment Scale-Revised (RAS-R; Corrigan, Giffort, Rashid, Leary, & Okeke, 1999). The RAS-S was adapted by replacing "mental illness" with "suicide attempt" to reflect recovery from a suicide attempt. Participants respond to statement (e.g., "I have goals in life that I want to reach.") on a 5-point agreement scale (5 ¼ strongly agree). Scores were calculated as a sum of items, with higher scores indicating more positive recovery beliefs.
Rosenberg Self-Esteem Scale (RSE). The 10-item RSE (Rosenberg, 1965) measured self-esteem. Participants indicated their agreement with statements (e.g., I feel that I have a number of good qualities) on a 4point scale (4 ¼ strongly agree). Scores were calculated as a sum of items, with higher scores indicating greater self-esteem.
Empowerment Scale (ES). The 5-item ES measured empowerment, or the degree to which participants believe in their ability to exert control over aspects of their life (Rogers, Ralph, & Salzer, 2010). Participants responded to statements (e.g., "I generally accomplish what I set out to do") on a 4-point scale (4 ¼ strongly disagree). Scores were calculated as a sum of items, with higher scores indicating less empowerment.

Center for Epidemiologic Studies Depression
Scale (CESD-10). The 10-item CESD-10 measured depressive symptoms during the past week (Andresen, Malmgren, Carter, & Patrick, 1994). Respondents indicated their agreement with statements (e.g., "I felt hopeful about the future") on a 4point scale (3 ¼ most or all of the time). Scores were calculated as a sum of items, with higher scores indicating increased depression.
Stigma Stress Scale. The stigma stress scale was used to measure the degree to which stigma is a source of stress (R€ usch et al., 2009). Of the 8-item measure, four items assess whether stigma is perceived as a personal threat to the individual ("Prejudice against people who have attempted suicide will have a negative impact on my future") and an additional four items assess the perceived resources available to deal with stigma ("I am prepared to deal with prejudice against people who have attempted suicide"). All items are scored on a 7-point agreement scale (7 ¼ strongly agree), and a stress appraisal score is calculated by subtracting the perceived resources scores from the perceived harmfulness scores. Higher scores indicate more stigma stress.

Analytic Procedure
All analyses were performed in SPSS 22. Pearson product moment correlations were computed between study variables and Cronbach's alpha was used to measure reliability of study instruments. One-way analysis of variance procedures calculated differences between means on SSSAS subscales, and between SSSAS subscales, depression, time since most recent suicide The Self-Stigma of Suicide Attempt Survivors attempt (less than versus more than one year), and number of past suicide attempts (one versus multiple attempts). Pearson correlations were computed between participant age and study variables, while independent sample t-tests were performed to explore the relationship between gender and study variables. mean differences between all four subscales (p < .001). While participants were aware of stigmatizing stereotypes about suicide attempt survivors, they were less apt to agree with those stereotypes, apply those to themselves, or to feel harmed by the selfstigma. Table 2 shows that Cronbach's alphas for the four SSSAS subscales were in the good or excellent range (.842À.951).

RESULTS
Correlations between study variables followed the expected patterns and were in the expected directions ( Table 2). The aware and agree subscales were not significantly correlated with other variables (recovery, self-esteem, empowerment, depression, or stigma stress), whereas, the apply and harm subscales were significantly correlated with both positive and negative self-stigma related variables. Participants who scored highly on apply and harm subscales scored lower on recovery, self-esteem, and empowerment, and higher on depression and stigma stress. Most correlations between the subscales and study variables were in the moderate range (Cohen, 1988), with the largest correlations between harm and self-esteem (r ¼ À.62), depression (r ¼ .57), and stigma stress (r ¼ .51). Self-esteem and L. Sheehan et al.
recovery were highly correlated (r ¼ À.74 and À.63 respectively) with depression, and recovery was also highly correlated with self-esteem (r ¼ .75). One-way ANOVAs were conducted to examine relationships between depression, self-stigma, time since most recent suicide attempt, and lifetime incidence of suicide attempt. Table 3 displays results, which indicate that participants who reported an attempted suicide within the past year scored significantly higher on the depression measure than those whose suicide attempt was a year ago or more (F(1, 289) ¼ 17.87, p < .001). Participants who had attempted suicide within the past year were more likely to apply stereotypes to themselves (F(1, 289) ¼ 8.03, p ¼ .005) and experience harm as a result (F(1, 289) ¼ 20.13, p < .001), than were those whose attempt was further in the past. Similarly, participants with multiple suicide attempts were more likely to apply stereotypes (F(1, 290) ¼ 10.80, p ¼ .001) and experience harm (F(1, 290) ¼ 8.28, p ¼ .004) than were those who had experienced only one attempt. Table 4 shows the stereotype items (N ¼ 14) that participants responded to for each subscale of the SSSAS. Stereotype endorsement on the aware subscales ranged from m ¼ 5.96 (not following treatment recommendations) to m ¼ 8.14 (depressed), whereas agree subscale item means ranged from 2.01 (looked down on) to 8.10 (in need of help). The ranges for the apply and harm subscale item means were more restricted, with apply The Self-Stigma of Suicide Attempt Survivors In Table 5 we report relationships between age, gender, and study variables. Apply and harm subscales were related to age, with older participants being less likely to apply stigma to themselves and experience harm. Overall, younger participants scored slightly lower on affirming attitudes (recovery, self-esteem, empowerment) and higher on harmful variables (stigma stress). Gender was related to aware and agree subscales, with females more likely to be aware of suicide stigma and males more likely to agree with stigma. When correlations between age and subscales were converted to z-scores and compared, only the correlation between the agree and apply subscales was significantly different (p < .05).

DISCUSSION
This study evaluated a new measure to capture the internalized stigma of attempting suicide. The current results support the application of the progressive model of self-stigma for suicide attempt survivors. Participants were most often aware of stereotypes, and less often agreed with the stereotypes, applied the stereotypes to themselves, and were harmed by the stereotypes, respectively. The correlations between the SSSAS Harm scale suggest that self-stigma of suicide attempt is related to other psychological constructs, in similar fashion to the self-stigma of mental illness (Corrigan et al., 2006;Corrigan, Rafacz, & R€ usch, 2011). That is, apply and harm subscales were associated with increased depression and stigma stress, and with decreased recovery, self- The Self-Stigma of Suicide Attempt Survivors esteem, and empowerment, whereas aware and agree subscales were not significantly correlated with any of those variables. In addition, high Cronbach alphas on SSSAS subscales support the reliability of the measure. Those items on the SSSAS that described the suicide attempt survivor as depressed or distressed tended to be highly endorsed across levels of the progressive model. While some might view depression as synonymous with a suicide attempt and thus a truth rather than a stereotype, a past suicide attempt does not necessarily mean that an individual is perpetually overwhelmed, suffering, or in need of help. Internalization of these public stereotypes ("I've attempted suicide, so I'll always be overwhelmed with problems") is where the harm of self-stigma can arise. However, it is important to note that many participants in the study were currently experiencing depressive symptoms (a CESD-10 score of 10 is indicative for risk of clinical depression and the sample mean for our study was 11.13). Depressed participants could have interpreted the harm items ("Because I attempted suicide, I am depressed") as a reality of their situation, rather than as stigma per se.
While older participants did not differ from younger participants in their awareness or agreement with suicide stereotypes, they were less likely to endorse self-stigma, to experience stigma stress, and to have symptoms of depression than younger participants. Older participants also endorsed higher levels of recovery, self-esteem, and empowerment, suggesting that, with age, individuals may become better able to manage the impact of self-stigma (Werner, Aviv, & Barak, 2008). Women were more likely to indicate awareness of stereotypes, whereas men were more likely to agree with stereotypes about suicide. However, no gender differences were found in application or resulting harm from self-stigma in this sample. In contrast, past research has found women more likely to report self-stigma (Girma et al., 2013). While these correlations are significant, we note that effect sizes were relatively small. This is consistent with past research, which finds minimal relationships between demographic variables and the self-stigma of mental illness (see Livingston & Boyd, 2010).

Implications
Significant research has explored how the self-stigma of mental illness is related to empowerment, self-esteem, hope, recovery, and well-being (Livingston & Boyd, 2010;Oexle et al., 2018). The present study suggests that suicide self-stigma is an additional barrier to recovery after a suicide attempt, and a target for suicide prevention efforts. Suicide attempt survivors are at substantially increased risk for reattempt (Gibb, Beautrais, & Fergusson, 2005). Recent research suggests that stigma can be a risk factor for suicide via several paths (Carpiniello & Pinna, 2017). Prejudice and discrimination towards suicide attempt survivors (e.g., social exclusion) can directly contribute to suicidality, but awareness and fear of potentially stigmatizing reactions (i.e., perceived stigma) may result in self-imposed isolation and reductions in help-seeking (Corrigan, Druss, & Perlick, 2014). Self-stigma introduces an additional layer, whereby suicide attempt survivors internalize public perceptions, feeling shame, and decrements in self-worth. Self-stigma related to suicide and mental illness could be further exacerbated by membership in other stigmatized groups (e.g., ethnic minority or LGBTQþ; Carpiniello & Pinna, 2017).
Clinicians who serve suicide attempt survivors should be aware of the specific stereotypes that can be internalized, so that they can work with clients to explicitly address any self-stigma that may be present. Given that suicide prevention depends on people disclosing current suicidal thoughts or past attempts, mental health providers should acknowledge how self-stigma can prevent suicide attempt survivors from talking about suicidality, even with healthcare providers (Fialko et al., 2006;Fulginiti, Pahwa, Frey, Rice, & Brekke, 2016). Younger individuals, those with more recent suicide attempts, and those with a more extensive history of suicidality, can be targeted for self-stigma interventions. Formal programs for combatting self-stigma of a suicide attempt, such as through peer education (Conner, McKinnon, Ward, Reynolds, & Brown, 2015) or strategic disclosure programs (Corrigan et al., 2015) should be pursued for attempt survivors. One peer-led group intervention, which includes lessons on identifying and challenging self-stigma, considering pros and cons of disclosing mental health challenges, and developing scripts for talking about mental health, has resulted in reductions in self-stigma, stigma stress, and depression for people with mental illness (Corrigan et al., 2015;R€ usch et al., 2014b). Adaptations to existing mental health self-stigma interventions could make these more relevant for suicide attempt survivors and the SSSAS allows for evaluation of adapted interventions.
Longitudinal research could help clarify relationships between self-stigma of a suicide attempt and recovery, including how variables from the present study interact with each other over time. A particularly important question is the relationship between self-stigma and depression. Although one possibility is that self-stigma contributes directly to depression, another option is that people with depression, because of their negative depression lens, are more susceptible to perceptions of self-stigma.

Limitations
Development and evaluation of the SSSAS in a large sample of suicide attempt survivors is noteworthy progress; however further validation is needed to examine test-retest reliability, sensitivity to change, and generalizability to other populations. Although we employed screening and data management procedures to enhance the reliability of the data, we cannot be sure that results presented here will generalize to other settings. In particular, we could not verify for certain that all participants had a history of suicide attempt. Similar efforts with documented suicide attempt survivors should seek to substantiate our findings. The empowerment, recovery, and stigma-stress scales used for this study were adapted from existing mental illness measures due to lack of available measures specific to suicide. Exploration of recovery, empowerment, and stigma-stress from the perspective of suicide attempt survivors might reveal important differences from those of mental illness, which are not reflected in the present study. In addition, the empowerment scale had low reliability, calling into question the results pertaining to empowerment.
A further challenge of studying suicide stigma is the considerable overlap between suicide and mental illness stigma. Although research demonstrates distinction between public perceptions of SSSAS stereotypes for depressed and suicide attempt survivor targets (Sheehan, Dubke et al., 2017), similar research has not shown that suicide attempt survivors themselves clearly The Self-Stigma of Suicide Attempt Survivors differentiate between stigma of mental illness and that of suicide. In this study, endorsement for the apply and harm scales were relatively low. Future research should explore how and at what point the selfstigma of a suicide attempt becomes damaging to overall quality of life.