Psychological Medicine, Page 1 of 10. doi:10.1017/S0033291714003158
OR I G I N A L A R T I C L E
© Cambridge University Press 2015
The link between mental health-related discrimination and suicidality: service user perspectives S. Farrelly1*†, D. Jeffery1†, N. Rüsch2, P. Williams1, G. Thornicroft1 and S. Clement1 1 2
Health Service and Population Research Department, Kings College London, Institute of Psychiatry, London, UK Department of Psychiatry II, University of Ulm and BKH Günzburg, Germany
Background. Suicide is a major global public health issue. Mental illness is a risk factor for suicide, but as many individuals with a diagnosed mental health problem do not experience suicidal ideation or attempt suicide, other individual and societal factors must be considered. Mental illness-related discrimination is one potential risk factor. Method. Using mixed methods, the influence of discrimination on suicidality amongst 194 individuals diagnosed with depression, bipolar or schizophrenia spectrum disorders was investigated. Qualitative interviews with a sub-sample of 58 individuals who reported a link between experience of discrimination and suicidality were analysed using framework analysis. Quantitative methods were used to examine the model derived from qualitative analyses. Results. Results indicate that the experience of discrimination led 38% of the overall sample of 194 participants, to suicidal feelings and 20% reported that it contributed to making a suicide attempt. The qualitative model derived from interviews with a sub-sample of 58 participants suggested that the experience of discrimination is experienced as a stressor that exceeds coping resources, leading to a negative self-image and a perception of decreased supportive networks/ social structure. The anticipation of further negative events and treatment, and the perception of a lack of supportive networks led individuals in this study to feelings of hopelessness and suicidality. Quantitative analyses provided support for the model. Conclusions. These data suggest that both psychological therapies aimed at improving coping skills and populationlevel anti-stigma interventions that reduce the occurrence of discrimination may provide some protection against suicide amongst individuals with mental health problems. Received 19 December 2013; Revised 9 December 2014; Accepted 11 December 2014 Key words: Bipolar disorder, depression, discrimination, England, mental illness, mixed methods, psychosis, suicide.
Introduction Every year 800000 people die from suicide, with a rate of 11.4 per 100000 (WHO, 2014). The WHO estimates that suicide represents 1.8% of the total global burden of disease and as such, suicide is a major, worldwide public health issue. It is well established that mental health problems are one of the key risks for suicide (Harris & Barraclough, 1997). In a systematic review of psychological autopsies (i.e. retrospective interviews with relatives of the deceased and review of psychiatric/medical case notes), approximately 90% of people who had completed suicide had a pre-existing mental
* Address for correspondence: S. Farrelly, PhD, Section of Community Mental Health, PO29, Health Service and Population Research Department, King’s College London, Institute of Psychiatry, De Crespigny Park, London SE5 8AF, UK. (Email: [email protected]) † These authors contributed equally to this work.
health problem (Cavanagh et al. 2003). However, cohort studies of specific diagnostic clusters suggest that as less than half of those individuals with mental illness experience suicidal ideation or make a suicide attempt (Novick et al. 2010; Robinson et al. 2010; Hauser et al. 2013). In this context, focus on the presence of pre-existing diagnoses is important but provides only part of the picture on risk. Research has identified a number of other risk factors for suicidal ideation and eventual suicide including social isolation (Trout, 1980; Beautrais, 2002; Eagles et al. 2003), psychiatric hospitalization (Beautrais et al. 1998; Beautrais, 2002), social and economic disadvantage such as unemployment and low income (Beautrais, 2000), and psychological vulnerability (Beautrais, 2000) such as negative appraisal biases and proneness to hopelessness (Brown et al. 2000; Acosta et al. 2013). Such research on risk factors is informative, yet examination of risk factors in isolation is of limited practical value. For example, how does social isolation
2 S. Farrelly et al. link with suicidal ideation? Empirically validated models or explanations for the pathways through which these risk factors lead to subsequent suicidal behaviours are required. One potential risk factor that has received comparatively little attention is the stigma associated with mental health problems and associated experience of discrimination (Rüsch et al. 2014). It has been suggested that the acceptance of stigmatizing beliefs about mental illness may influence levels of hopelessness (Livingston & Boyd, 2010; Sharaf et al. 2012). For example, if individuals accept that they have a mental health problem, and internalize the often negative stereotypes associated with mental illnesses (e.g. dangerousness, chronicity), they will feel more hopeless and despairing of their future. There is some evidence for this proposition, for example, people with mental illness who hold their in-group in ‘low group value’ (conceptually similar to stereotype agreement and internalized stigma) experience increased hopelessness, even after controlling for depression (Rüsch et al. 2009a). Further, the cognitive appraisal of stigma as a stressor has also been shown to contribute to hopelessness (Rüsch et al. 2009a, b). While there is emerging evidence that the stigma associated with mental health problems may contribute to individuals’ feelings of hopelessness, the experience of behavioural manifestations of stigma (i.e. discrimination) may be particularly defining in that it may be interpreted as confirmatory evidence of negative stereotypes and/or a reduction in status in the community. Perceived racial discrimination and experience of acculturation in minority ethnic communities has been associated with increased risk of suicidal ideation (Gomez et al. 2011; Wang et al. 2013) and there are similar findings in the Lesbian, Gay, Bisexual and Transgender communities (Clements-Nolle et al. 2006; Cleary, 2012). However, we were unable to find any research linking the experience of discrimination due to mental health problems and subsequent suicidal ideation or behaviours. In this context, this mixed methods study aimed to: (1) describe the proportion of service users who were at risk of suicide and the extent to which discrimination is linked to suicidality (a pragmatic term for the constellation of suicidal ideation and attempts, self-harm and hopelessness studied here); (2) explore the process through which discrimination is linked to suicidality from the perspectives of service users using qualitative data; (3) test the model of service users’ perspectives (produced in aim 2) of the link between discrimination and suicidality, using quantitative methods.
Method The MIRIAD (Mental Illness-Related Investigations on Discrimination) study is a cross-sectional study of 202 individuals using secondary mental health services in South London, funded by the National Institute for Health Research as a part of the SAPPHIRE programme (http://www.sapphire.iop.kcl.ac.uk). Data were collected between September 2011 and October 2012. The study was approved by the East of England/Essex 2 Research Ethics Committee (ref. 11/EE/0052). This paper presents an analysis of the association between experiences of discrimination and suicidal ideation amongst MIRIAD participants. Recruitment of the sample has been described elsewhere (Farrelly et al. 2014) but is briefly described below. Qualitative and quantitative data were collected at the same time. Qualitative data analysis was conducted before the quantitative data to lessen the risk of bias. In this way, qualitative data were prioritized. The two sets of data were integrated at the point of quantitative analysis by using the quantitative data to investigate proposed qualitative theoretical framework. Recruitment and sample Inclusion criteria for participants were: aged at least 18 years; a clinical diagnosis of either major depression, bipolar or schizophrenia spectrum disorders (ICD-10: F32, F21 and F20-F29, respectively); self-defined Black, White or Mixed (Black and White) ethnicity; current treatment with a community mental health team; sufficiently fluent in English to provide informed consent; and sufficiently well for participation to not pose a risk to their or others’ health or safety. We did not include people of Asian ethnicities due to low prevalence numbers in the target area. Service user participants were recruited from community mental health teams in South London. Staff in the community mental health teams were approached for their approval to participate in the study and provided with a list of eligible service users. Clinicians were asked if the service user was sufficiently well to participate, and if so, a letter was posted to the service user inviting them to contact the research team if they were interested in participating. This letter was followed by a reminder flyer if there had been no response within 1 month. As reported elsewhere (Farrelly et al. 2014), 4233 service users were screened for eligibility for inclusion in the MIRIAD study. A total of 1345 (31.7%) were eligible and were invited to participate; of these 207 (15.4%) service users provided written and informed consent. There were no differences between
Discrimination as a risk factor for suicide 3 eligible consenting and eligible non-consenting service users in terms of diagnoses, age, gender and ethnicity. Five service users were excluded after interview due to incorrect diagnoses (n = 4) or incomplete data (n = 1), leaving 202 participants for the overall MIRIAD sample. Data collection Research assistants administered the interview to consenting service users. The quantitative interview schedule collected demographic clinical information and contained a battery of measures on stigma, discrimination and access to physical and mental healthcare; those relevant to this paper are detailed below. Clinical data, including diagnoses generated by clinicians, were also extracted from electronic patient records. At the end of the quantitative interview, participants were asked the following screening questions: ‘Have stigma or discrimination ever contributed to you having feelings of hopelessness, you self-harming yourself, you having thoughts about suicide, or to you attempting suicide?’ Those who responded affirmatively to any of these were asked for examples of this and prompted to elaborate. These responses were audiorecorded and transcribed. For analysis, a purposive sub-sample of 60 participants was selected for analysis based on gender, ethnicity and self-reported diagnosis. Two of these participants were subsequently excluded from the overall study as information later collected their clinical notes indicated they did not met the diagnostic eligibility criteria, making a total sub-sample of 58 for the qualitative analyses. Measures Suicide Behaviours Questionnaire (SBQ; Linehan, 1996). The SBQ is a 4-item self-report version of the widely used 34-item SBQ. The revised brief version was shown to have good psychometric properties with α = 0.87 and differentiated between those with clinically assessed suicide risk and those not at risk (Osman et al. 2001). Each of the four items is rated on a likert scale: lifetime suicide ideation and attempts (1 = ‘never’ to 4 = ‘I have attempted’); past 12 months frequency of ideation (1 = ‘never’ to 5 = ‘very often’); communication of suicide intent to others (1 = ‘no’ to 3 = ‘yes, more than once’); and self-assessed likelihood of suicide attempt in the future (0 = ‘never’ to 6 = ‘very likely’). A total score is obtained by summing all four items. Higher scores on the SBQ-R mean greater suicidality, with scores 58 indicating suicide risk. In addition to the SBQ, we also analysed participants’ responses to the screening questions for the qualitative component.
Discrimination and Stigma Scale (DISC; Brohan et al. 2013). The DISC is an interviewer-delivered measure of experiences of discrimination (‘unfair treatment’) in the last 12 months due to having a diagnosis of mental illness. Participants report experiences of discrimination across 21 items including employment, parenting, mental health treatment, dating or intimate relationships and the extent to which participants had been avoided or shunned by people on a 4-point Likert scale [not at all (0), a little (1), moderately (2), a lot (3)]. The DISC has good psychometric properties (Brohan et al. 2013). A ‘severity’ score (range 0–3) was calculated by adding each item score (0, 1, 2 or 3) and dividing by the number of applicable, non-missing items in the subscale. Stigma Stress Scale (SSS; Rüsch et al. 2009a). The SSS is an eight-item measure completed in interview. Four items assessed the extent to which stigma was appraised as personally harmful (e.g. ‘Prejudice against people with mental illness will have harmful or bad consequences for me’). Four additional items measured appraisal of perceived resources to cope with stigma (e.g. ‘I have the resources I need to handle problems posed by prejudice against people with mental illness’). Items are scored from 1 to 7 with higher scores equalling higher agreement. A ‘stress appraisal’ variable (SSA) was computed by subtracting perceived resources from perceived harmfulness. A higher difference score indicated the appraisal of stigma as stressful and as exceeding personal coping resources. Beck Hopelessness Scale (BHS; Beck & Steer, 1988). The BHS is a self-completed measure assessing hopelessness. The version used in this study had a 5-point Likert scale (ranging from 1 ‘strongly agree’ to 5 ‘strongly disagree’) on 20 items. A total score was calculated by summing the items (possible range 20–100). Internalized Stigma of Mental Illness Scale (ISMI; Ritsher et al. 2003). The ISMI is a 29-item measure that assesses mental health service users’ experience of internalized stigma. Items such as ‘I feel out of place in the world because I have a mental illness’, are self-rated on a 4-point Likert scale. Strong internal consistency and test–retest reliability have been reported (Ritsher et al. 2003). There are five subscales including a 5-item ‘Discrimination Experience’ subscale, which due to being conceptually similar to the DISC was excluded. A total score was generated by summing the remaining 24 items. Confidant. A binary variable regarding the presence/ absence of a confidant was calculated based on the participants’ responses to three questions about their social network: ‘Of the people you see regularly, is there someone you consider a friend?’, ‘Do you have
4 S. Farrelly et al. a best friend you can confide in?’, ‘Are you particularly close to someone in your family?’ Data analysis Qualitative interviews were audio-recorded and transcribed by an independent transcriber. Transcripts were read while listening to recordings to check for accuracy. Transcripts were imported into NVivo 9 (QSR International’s NVivo 9 software) together with respondent’s gender, diagnosis and ethnicity. A framework analysis was conducted (Ritchie & Spencer, 1994). This analysis method is a five-stage method for analysing qualitative data and is widely used in social policy and health services research. In the first stage – the Familiarization Process – D.J. and S.C. immersed themselves in the data by reading all the transcripts and making notes of the recurring and salient ideas and themes in the data. In stage two – Developing a Thematic Framework – D.J. developed a ‘thematic framework’ or list of recurring themes from the familiarization process. In the Indexing stage, D.J. applied the thematic framework to the data and coded each part of text, merging and creating new themes as necessary. In stage four – Charting, data were organized in thematic charts in an Excel spreadsheet. Data were lifted from the transcript and placed in charts organized according to theme. This process was undertaken independently by D.J. and S.C. who met to agree any disparities. Finally, team members (D.J., S.F., S.C.) synthesized the data by comparing themes and sub-themes, merging, splitting and re-naming these as required and noting any associations or links between the themes. A schematic diagram representing the process of participant’s experiences was produced and the final model agreed. Quantitative analyses were conducted using Stata v. 11 (StataCorp., USA) and Mplus (http://www. statmodel.com/). The proportions of participants at risk of suicide (SBQ score ≥8) and the proportion reporting discrimination as a contributory factor to the suicidality indicators (i.e. qualitative screening questions) were reported. A path analysis was fitted in Mplus to test the proposed model linking discrimination experiences to subsequent suicidality as described by participants in the qualitative interviews, using maximum likelihood (ML) estimation, boostrapped standard errors and confidence intervals, and excluding those with missing data on predictor variables. To assess the overall fit of the model, the following indices were used: the χ2 statistic where non-significant scores indicate acceptable fit; the comparative fit index (CFI) for which values between 0 and 1 and >0.95 indicate good fit; the Tucker–Lewis Index (TLI) where scores >0.90 are
considered acceptable fit; and the root mean square error of approximation (RMSEA) where scores 0.95 indicate good fit. The TLI was 0.922; scores >0.90 are considered acceptable. The RMSEA was 0.076 (95% CI 0.025–0.125); scores
OR I G I N A L A R T I C L E
© Cambridge University Press 2015
The link between mental health-related discrimination and suicidality: service user perspectives S. Farrelly1*†, D. Jeffery1†, N. Rüsch2, P. Williams1, G. Thornicroft1 and S. Clement1 1 2
Health Service and Population Research Department, Kings College London, Institute of Psychiatry, London, UK Department of Psychiatry II, University of Ulm and BKH Günzburg, Germany
Background. Suicide is a major global public health issue. Mental illness is a risk factor for suicide, but as many individuals with a diagnosed mental health problem do not experience suicidal ideation or attempt suicide, other individual and societal factors must be considered. Mental illness-related discrimination is one potential risk factor. Method. Using mixed methods, the influence of discrimination on suicidality amongst 194 individuals diagnosed with depression, bipolar or schizophrenia spectrum disorders was investigated. Qualitative interviews with a sub-sample of 58 individuals who reported a link between experience of discrimination and suicidality were analysed using framework analysis. Quantitative methods were used to examine the model derived from qualitative analyses. Results. Results indicate that the experience of discrimination led 38% of the overall sample of 194 participants, to suicidal feelings and 20% reported that it contributed to making a suicide attempt. The qualitative model derived from interviews with a sub-sample of 58 participants suggested that the experience of discrimination is experienced as a stressor that exceeds coping resources, leading to a negative self-image and a perception of decreased supportive networks/ social structure. The anticipation of further negative events and treatment, and the perception of a lack of supportive networks led individuals in this study to feelings of hopelessness and suicidality. Quantitative analyses provided support for the model. Conclusions. These data suggest that both psychological therapies aimed at improving coping skills and populationlevel anti-stigma interventions that reduce the occurrence of discrimination may provide some protection against suicide amongst individuals with mental health problems. Received 19 December 2013; Revised 9 December 2014; Accepted 11 December 2014 Key words: Bipolar disorder, depression, discrimination, England, mental illness, mixed methods, psychosis, suicide.
Introduction Every year 800000 people die from suicide, with a rate of 11.4 per 100000 (WHO, 2014). The WHO estimates that suicide represents 1.8% of the total global burden of disease and as such, suicide is a major, worldwide public health issue. It is well established that mental health problems are one of the key risks for suicide (Harris & Barraclough, 1997). In a systematic review of psychological autopsies (i.e. retrospective interviews with relatives of the deceased and review of psychiatric/medical case notes), approximately 90% of people who had completed suicide had a pre-existing mental
* Address for correspondence: S. Farrelly, PhD, Section of Community Mental Health, PO29, Health Service and Population Research Department, King’s College London, Institute of Psychiatry, De Crespigny Park, London SE5 8AF, UK. (Email: [email protected]) † These authors contributed equally to this work.
health problem (Cavanagh et al. 2003). However, cohort studies of specific diagnostic clusters suggest that as less than half of those individuals with mental illness experience suicidal ideation or make a suicide attempt (Novick et al. 2010; Robinson et al. 2010; Hauser et al. 2013). In this context, focus on the presence of pre-existing diagnoses is important but provides only part of the picture on risk. Research has identified a number of other risk factors for suicidal ideation and eventual suicide including social isolation (Trout, 1980; Beautrais, 2002; Eagles et al. 2003), psychiatric hospitalization (Beautrais et al. 1998; Beautrais, 2002), social and economic disadvantage such as unemployment and low income (Beautrais, 2000), and psychological vulnerability (Beautrais, 2000) such as negative appraisal biases and proneness to hopelessness (Brown et al. 2000; Acosta et al. 2013). Such research on risk factors is informative, yet examination of risk factors in isolation is of limited practical value. For example, how does social isolation
2 S. Farrelly et al. link with suicidal ideation? Empirically validated models or explanations for the pathways through which these risk factors lead to subsequent suicidal behaviours are required. One potential risk factor that has received comparatively little attention is the stigma associated with mental health problems and associated experience of discrimination (Rüsch et al. 2014). It has been suggested that the acceptance of stigmatizing beliefs about mental illness may influence levels of hopelessness (Livingston & Boyd, 2010; Sharaf et al. 2012). For example, if individuals accept that they have a mental health problem, and internalize the often negative stereotypes associated with mental illnesses (e.g. dangerousness, chronicity), they will feel more hopeless and despairing of their future. There is some evidence for this proposition, for example, people with mental illness who hold their in-group in ‘low group value’ (conceptually similar to stereotype agreement and internalized stigma) experience increased hopelessness, even after controlling for depression (Rüsch et al. 2009a). Further, the cognitive appraisal of stigma as a stressor has also been shown to contribute to hopelessness (Rüsch et al. 2009a, b). While there is emerging evidence that the stigma associated with mental health problems may contribute to individuals’ feelings of hopelessness, the experience of behavioural manifestations of stigma (i.e. discrimination) may be particularly defining in that it may be interpreted as confirmatory evidence of negative stereotypes and/or a reduction in status in the community. Perceived racial discrimination and experience of acculturation in minority ethnic communities has been associated with increased risk of suicidal ideation (Gomez et al. 2011; Wang et al. 2013) and there are similar findings in the Lesbian, Gay, Bisexual and Transgender communities (Clements-Nolle et al. 2006; Cleary, 2012). However, we were unable to find any research linking the experience of discrimination due to mental health problems and subsequent suicidal ideation or behaviours. In this context, this mixed methods study aimed to: (1) describe the proportion of service users who were at risk of suicide and the extent to which discrimination is linked to suicidality (a pragmatic term for the constellation of suicidal ideation and attempts, self-harm and hopelessness studied here); (2) explore the process through which discrimination is linked to suicidality from the perspectives of service users using qualitative data; (3) test the model of service users’ perspectives (produced in aim 2) of the link between discrimination and suicidality, using quantitative methods.
Method The MIRIAD (Mental Illness-Related Investigations on Discrimination) study is a cross-sectional study of 202 individuals using secondary mental health services in South London, funded by the National Institute for Health Research as a part of the SAPPHIRE programme (http://www.sapphire.iop.kcl.ac.uk). Data were collected between September 2011 and October 2012. The study was approved by the East of England/Essex 2 Research Ethics Committee (ref. 11/EE/0052). This paper presents an analysis of the association between experiences of discrimination and suicidal ideation amongst MIRIAD participants. Recruitment of the sample has been described elsewhere (Farrelly et al. 2014) but is briefly described below. Qualitative and quantitative data were collected at the same time. Qualitative data analysis was conducted before the quantitative data to lessen the risk of bias. In this way, qualitative data were prioritized. The two sets of data were integrated at the point of quantitative analysis by using the quantitative data to investigate proposed qualitative theoretical framework. Recruitment and sample Inclusion criteria for participants were: aged at least 18 years; a clinical diagnosis of either major depression, bipolar or schizophrenia spectrum disorders (ICD-10: F32, F21 and F20-F29, respectively); self-defined Black, White or Mixed (Black and White) ethnicity; current treatment with a community mental health team; sufficiently fluent in English to provide informed consent; and sufficiently well for participation to not pose a risk to their or others’ health or safety. We did not include people of Asian ethnicities due to low prevalence numbers in the target area. Service user participants were recruited from community mental health teams in South London. Staff in the community mental health teams were approached for their approval to participate in the study and provided with a list of eligible service users. Clinicians were asked if the service user was sufficiently well to participate, and if so, a letter was posted to the service user inviting them to contact the research team if they were interested in participating. This letter was followed by a reminder flyer if there had been no response within 1 month. As reported elsewhere (Farrelly et al. 2014), 4233 service users were screened for eligibility for inclusion in the MIRIAD study. A total of 1345 (31.7%) were eligible and were invited to participate; of these 207 (15.4%) service users provided written and informed consent. There were no differences between
Discrimination as a risk factor for suicide 3 eligible consenting and eligible non-consenting service users in terms of diagnoses, age, gender and ethnicity. Five service users were excluded after interview due to incorrect diagnoses (n = 4) or incomplete data (n = 1), leaving 202 participants for the overall MIRIAD sample. Data collection Research assistants administered the interview to consenting service users. The quantitative interview schedule collected demographic clinical information and contained a battery of measures on stigma, discrimination and access to physical and mental healthcare; those relevant to this paper are detailed below. Clinical data, including diagnoses generated by clinicians, were also extracted from electronic patient records. At the end of the quantitative interview, participants were asked the following screening questions: ‘Have stigma or discrimination ever contributed to you having feelings of hopelessness, you self-harming yourself, you having thoughts about suicide, or to you attempting suicide?’ Those who responded affirmatively to any of these were asked for examples of this and prompted to elaborate. These responses were audiorecorded and transcribed. For analysis, a purposive sub-sample of 60 participants was selected for analysis based on gender, ethnicity and self-reported diagnosis. Two of these participants were subsequently excluded from the overall study as information later collected their clinical notes indicated they did not met the diagnostic eligibility criteria, making a total sub-sample of 58 for the qualitative analyses. Measures Suicide Behaviours Questionnaire (SBQ; Linehan, 1996). The SBQ is a 4-item self-report version of the widely used 34-item SBQ. The revised brief version was shown to have good psychometric properties with α = 0.87 and differentiated between those with clinically assessed suicide risk and those not at risk (Osman et al. 2001). Each of the four items is rated on a likert scale: lifetime suicide ideation and attempts (1 = ‘never’ to 4 = ‘I have attempted’); past 12 months frequency of ideation (1 = ‘never’ to 5 = ‘very often’); communication of suicide intent to others (1 = ‘no’ to 3 = ‘yes, more than once’); and self-assessed likelihood of suicide attempt in the future (0 = ‘never’ to 6 = ‘very likely’). A total score is obtained by summing all four items. Higher scores on the SBQ-R mean greater suicidality, with scores 58 indicating suicide risk. In addition to the SBQ, we also analysed participants’ responses to the screening questions for the qualitative component.
Discrimination and Stigma Scale (DISC; Brohan et al. 2013). The DISC is an interviewer-delivered measure of experiences of discrimination (‘unfair treatment’) in the last 12 months due to having a diagnosis of mental illness. Participants report experiences of discrimination across 21 items including employment, parenting, mental health treatment, dating or intimate relationships and the extent to which participants had been avoided or shunned by people on a 4-point Likert scale [not at all (0), a little (1), moderately (2), a lot (3)]. The DISC has good psychometric properties (Brohan et al. 2013). A ‘severity’ score (range 0–3) was calculated by adding each item score (0, 1, 2 or 3) and dividing by the number of applicable, non-missing items in the subscale. Stigma Stress Scale (SSS; Rüsch et al. 2009a). The SSS is an eight-item measure completed in interview. Four items assessed the extent to which stigma was appraised as personally harmful (e.g. ‘Prejudice against people with mental illness will have harmful or bad consequences for me’). Four additional items measured appraisal of perceived resources to cope with stigma (e.g. ‘I have the resources I need to handle problems posed by prejudice against people with mental illness’). Items are scored from 1 to 7 with higher scores equalling higher agreement. A ‘stress appraisal’ variable (SSA) was computed by subtracting perceived resources from perceived harmfulness. A higher difference score indicated the appraisal of stigma as stressful and as exceeding personal coping resources. Beck Hopelessness Scale (BHS; Beck & Steer, 1988). The BHS is a self-completed measure assessing hopelessness. The version used in this study had a 5-point Likert scale (ranging from 1 ‘strongly agree’ to 5 ‘strongly disagree’) on 20 items. A total score was calculated by summing the items (possible range 20–100). Internalized Stigma of Mental Illness Scale (ISMI; Ritsher et al. 2003). The ISMI is a 29-item measure that assesses mental health service users’ experience of internalized stigma. Items such as ‘I feel out of place in the world because I have a mental illness’, are self-rated on a 4-point Likert scale. Strong internal consistency and test–retest reliability have been reported (Ritsher et al. 2003). There are five subscales including a 5-item ‘Discrimination Experience’ subscale, which due to being conceptually similar to the DISC was excluded. A total score was generated by summing the remaining 24 items. Confidant. A binary variable regarding the presence/ absence of a confidant was calculated based on the participants’ responses to three questions about their social network: ‘Of the people you see regularly, is there someone you consider a friend?’, ‘Do you have
4 S. Farrelly et al. a best friend you can confide in?’, ‘Are you particularly close to someone in your family?’ Data analysis Qualitative interviews were audio-recorded and transcribed by an independent transcriber. Transcripts were read while listening to recordings to check for accuracy. Transcripts were imported into NVivo 9 (QSR International’s NVivo 9 software) together with respondent’s gender, diagnosis and ethnicity. A framework analysis was conducted (Ritchie & Spencer, 1994). This analysis method is a five-stage method for analysing qualitative data and is widely used in social policy and health services research. In the first stage – the Familiarization Process – D.J. and S.C. immersed themselves in the data by reading all the transcripts and making notes of the recurring and salient ideas and themes in the data. In stage two – Developing a Thematic Framework – D.J. developed a ‘thematic framework’ or list of recurring themes from the familiarization process. In the Indexing stage, D.J. applied the thematic framework to the data and coded each part of text, merging and creating new themes as necessary. In stage four – Charting, data were organized in thematic charts in an Excel spreadsheet. Data were lifted from the transcript and placed in charts organized according to theme. This process was undertaken independently by D.J. and S.C. who met to agree any disparities. Finally, team members (D.J., S.F., S.C.) synthesized the data by comparing themes and sub-themes, merging, splitting and re-naming these as required and noting any associations or links between the themes. A schematic diagram representing the process of participant’s experiences was produced and the final model agreed. Quantitative analyses were conducted using Stata v. 11 (StataCorp., USA) and Mplus (http://www. statmodel.com/). The proportions of participants at risk of suicide (SBQ score ≥8) and the proportion reporting discrimination as a contributory factor to the suicidality indicators (i.e. qualitative screening questions) were reported. A path analysis was fitted in Mplus to test the proposed model linking discrimination experiences to subsequent suicidality as described by participants in the qualitative interviews, using maximum likelihood (ML) estimation, boostrapped standard errors and confidence intervals, and excluding those with missing data on predictor variables. To assess the overall fit of the model, the following indices were used: the χ2 statistic where non-significant scores indicate acceptable fit; the comparative fit index (CFI) for which values between 0 and 1 and >0.95 indicate good fit; the Tucker–Lewis Index (TLI) where scores >0.90 are
considered acceptable fit; and the root mean square error of approximation (RMSEA) where scores 0.95 indicate good fit. The TLI was 0.922; scores >0.90 are considered acceptable. The RMSEA was 0.076 (95% CI 0.025–0.125); scores
